PI and PID Controller Tuning Rules For Time Delay Processes

PI and PID controller tuning rules for time delay processes: a summary
Technical Report AOD-00-01, Edition 1
A. O’Dwyer,
School of Control Systems and Electrical Engineering,
Dublin Institute of Technology, Kevin St., Dublin 8, Ireland.
15 May 2000
Phone: 353-1-4024875
Fax: 353-1-4024992
e-mail: aodwyer@dit.ie
Abstract: The ability of proportional integral (PI) and proportional integral derivative (PID) controllers to
compensate many practical industrial processes has led to their wide acceptance in industrial applications. The
requirement to choose either two or three controller parameters is perhaps most easily done using tuning rules. A
summary of tuning rules for the PI and PID control of single input, single output (SISO) processes with time
delay are provided in this report. Inevitably, this report is a work in progress and will be added to and extended
regularly.
Keywords: PI, PID, tuning rules, time delay.
1. Introduction
The ability of PI and PID controllers to compensate most practical industrial processes has led to their
wide acceptance in industrial applications. Koivo and Tanttu [1], for example, suggest that there are perhaps 5-
10% of control loops that cannot be controlled by SISO PI or PID controllers; in particular, these controllers
perform well for processes with benign dynamics and modest performance requirements [2, 3]. It has been stated
that 98% of control loops in the pulp and paper industries are controlled by SISO PI controllers [4] and that, in
process control applications, more than 95% of the controllers are of PID type [3]. The PI or PID controller
implementation has been recommended for the control of processes of low to medium order, with small time
delays, when parameter setting must be done using tuning rules and when controller synthesis is performed
either once or more often [5]. However, Ender [6] states that, in his testing of thousands of control loops in
hundreds of plants, it has been found that more than 30% of installed controllers are operating in manual mode
and 65% of loops operating in automatic mode produce less variance in manual than in automatic (i.e. the
automatic controllers are poorly tuned); this is rather sobering, considering the wealth of information available in
the literature for determining controller parameters automatically. It is true that this information is scattered
throughout papers and books; the purpose of this paper is to bring together in summary form the tuning rules for
PI and PID controllers that have been developed to compensate SISO processes with time delay. Tuning rules for
the variations that have been proposed in the ‘ideal’ PI and PID controller structure are included. Considerable
variations in the ideal PID controller structure, in particular, are encountered; these variations are explored in
more detail in Section 2.
2. PID controller structures
The ideal continuous time domain PID controller for a SISO process is expressed in the Laplace domain
as follows:
U( s) = G c (s) E (s) (1)
1
with G c (s) = Kc (1 + + Tds) (2)
Ts
i
and with Kc = proportional gain, Ti = integral time constant and Td = derivative time constant. If Ti = ∞
and Td = 0 (i.e. P control), then it is clear that the closed loop measured value, y, will always be less than the
desired value, r (for processes without an integrator term, as a positive error is necessary to keep the measured
value constant, and less than the desired value). The introduction of integral action facilitates the achievement of
equality between the measured value and the desired value, as a constant error produces an increasing controller
output. The introduction of derivative action means that changes in the desired value may be anticipated, and
thus an appropriate correction may be added prior to the actual change. Thus, in simplified terms, the PID
controller allows contributions from present controller inputs, past controller inputs and future controller inputs.
Many tuning rules have been defined for the ideal PI and PID structures. Tuning rules have also been
defined for other PI and PID structures, as detailed in Section 4.
3. Process modelling
Processes with time delay may be modelled in a variety of ways. The modelling strategy used will
influence the value of the model parameters, which will in turn affect the controller values determined from the
tuning rules. The modelling strategy used in association with each tuning rule, as described in the original
papers, is indicated in the tables. Of course, it is possible to use the tuning rules proposed by the authors with a
different modelling strategy than that proposed by the authors; applications where this occurs are not indicated
(to date). The modelling strategies are referenced as indicated. The full details of these modelling strategies are
provided in Appendix 2.
K e − sτ m
A. First order lag plus delay (FOLPD) model ( G m ( s) = m ):
1 + sTm
Method 1: Parameters obtained using the tangent and point method (Ziegler and Nichols [8], Hazebroek and
Van den Waerden [9]); Appendix 2.
Method 2: Km , τ m assumed known; Tm estimated from the open loop step response (Wolfe [12]); Appendix
2.
Method 3: Parameters obtained using an alternative tangent and point method (Murrill [13]); Appendix 2.
Method 4: Parameters obtained using the method of moments (Astrom and Hagglund [3]); Appendix 2.
Method 5: Parameters obtained from the closed loop transient response to a step input under proportional
control (Sain and Ozgen [94]); Appendix 2.
Method 6: Km , Tm , τ m assumed known.
Method 7: Parameters obtained using a least squares method in the time domain (Cheng and Hung [95]);
Appendix 2.
Method 8: Parameters obtained in the frequency domain from the ultimate gain, phase and frequency
determined using a relay in series with the closed loop system in a master feedback loop. The
model gain is obtained by the ratio of the integrals (over one period) of the process output to the
controller output. The delay and time constant are obtained from the frequency domain data
(Hwang [160]).
Method 9: Parameters obtained from the closed loop transient response to a step input under proportional
control (Hwang [2]); Appendix 2.
Method 10: Parameters obtained from two points estimated on process frequency response using a relay and
a relay in series with a delay (Tan et al. [39]); Appendix 2.
Method 11: Tm and τ m are determined from the ultimate gain and period estimated using a relay in series
with the process in closed loop; Km assumed known (Hang and Cao [112]); Appendix 2.
Method 12: Parameters are estimated using a tangent and point method (Davydov et al. [31]); Appendix 2.
Method 13: Parameters estimated from the open loop step response and its first time derivative (Tsang and
Rad [109]); Appendix 2.
Method 14: Tm and τ m estimated from Ku , Tu determined using Ziegler-Nichols ultimate cycle method;
Km estimated from the process step response (Hang et al. [35]); Appendix 2.
Method 15: Tm and τ m estimated from Ku , Tu determined using a relay autotuning method; Km estimated
from the process step response (Hang et al. [35]); Appendix 2.
Method 16: G p ( jω135 ) , ω135 and Km are determined from an experiment using a relay in series with the
0 0
process in closed loop; estimates for Tm and τ m are subsequently calculated. (Voda and
Landau [40]); Appendix 2.
Method 17: Parameter estimates back-calculated from discrete time identification method (Ferretti et al.
[161]); Appendix 2.
* Method 18: Parameter estimates calculated from process reaction curve using numerical integration
procedures (Nishikawa et al. [162]).
* Method 19: Parameter estimates determined graphically from a known higher order process (McMillan [58]
… also McMillan (1983), pp. 34-40.
* Method 20: Km estimated from the open loop step response. T90% and τ m estimated from the closed loop
step response under proportional control (Astrom and Hagglund [93]?)
Method 21: Parameters estimated from linear regression equations in the time domain (Bi et al. [46]);
Appendix 2.
Method 22: Tm and τ m estimated from relay autotuning method (Lee and Sung [163]); Km estimated
from the closed loop process step response under proportional control (Chun et al. [57]);
Appendix 2.
* Method 23: Parameters are estimated from a step response autotuning experiment – Honeywell UDC 6000
controller (Astrom et al. [30]).
Method 24: Parameters are estimated from the closed loop step response when process is in series with a
PID controller (Morilla et al. [104a]); Appendix 2.
Method 25: τm and Tm obtained from an open loop step test as follows: Tm = 1.4( t 67% − t 33% ) ,
τ m = t 67% − 1.1Tm . K m assumed known (Chen and Yang [23a]).
Method 26: τm and Tm obtained from an open loop step test as follows: Tm = 1.245 ( t 70% − t 33% ) ,
τ m = 1. 498 t 33% − 0. 498 t 70% . K m assumed known (Miluse et al. [27b]).
* Method 27: Data at the ultimate period is deduced from an open loop impulse response (Pi-Mira et al.
[97a]).
B. Non-model specific
Method 1: Parameters K u , K m , ωu are estimated from data obtained using a relay in series with the process
in closed loop and from the process step response (Kristiansson and Lennartsson [157]).
need to check how the other methods define these parameters –
K me− sτ m
C. Integral plus time delay (IPD) model ( G m ( s) = )
s
Method 1: τ m assumed known; Km determined from the slope at start of the open loop step response
(Ziegler and Nichols [8]); Appendix 2.
Method 2: Km , τ m assumed known.
Method 3: Parameters estimated from the ultimate gain and frequency values determined from an experiment
using a relay in series with the process in closed loop (Tyreus and Luyben [75]); Appendix 2.
Method 4: Parameters are estimated from the servo or regulator closed loop transient response, under PI
control (Rotach [77]); Appendix 2.
Method 5: Parameters are estimated from the servo closed loop transient response under proportional
control (Srividya and Chidambaram [80]); Appendix 2.
Method 6: K u and Tu are estimated from estimates of the ultimate and crossover frequencies. The ultimate
frequency estimate is obtained by placing an amplitude dependent gain in series with the
process in closed loop; the crossover frequency estimate is obtained by also using an amplitude
dependent gain (Pecharroman and Pagola [165]); Appendix 2.
Km e− sτ m
D. First order lag plus integral plus time delay (FOLIPD) model ( G m (s) = )
s(1 + sTm )
* Method 1: Method of moments (Astrom and Hagglund [3]).
Method 3: Parameters estimated from the open loop step response and its first and second time derivatives
(Tsang and Rad [109]); Appendix 2.
Method 4: K u and Tu are estimated from estimates of the ultimate and crossover frequencies (Pecharroman
and Pagola [165]) – as in Method 6, IPD model.
K m e− sτ m
Km e− sτm
E. Second order system plus time delay (SOSPD) model ( G m (s) = , )

s + 2ξ m Tm1s + 1 (1 + Tm1s)(1 + Tm2 s)
2 2
Tm1
Method 1: Km , Tm1 , Tm2 , τ m or Km , Tm1 , ξ m , τ m assumed known.
Method 2: Parameters estimated using a two-stage identification procedure involving (a) placing a relay in
series with the process in closed loop and (b) placing a proportional controller in series with the
process in closed loop (Sung et al. [139]); Appendix 2.
* Method 3: Parameters obtained in the frequency domain from the ultimate gain, phase and frequency
determined using a relay in series with the closed loop system in a master feedback loop. The
model gain is obtained by the ratio of the integrals (over one period) of the process output to the
controller output. The other parameters are obtained from the frequency domain data (Hwang
[160]).
Method 4: Tm and τ m estimated from Ku , Tu determined using a relay autotuning method; Km estimated
from the process step response (Hang et al. [35]); Appendix 2.
* Method 5: Parameter estimates back-calculated from discrete time identification method (Ferretti et al.
[161]).
Method 6: Parameteres estimated from the underdamped or overdamped transient response in open loop to a
step input (Jahanmiri and Fallahi [149]); Appendix 2.
* Method 7: Parameters estimated from a least squares time domain method (Lopez et al. [84]).
Method 8: Parameters estimated from data obtained when the process phase lag is − 900 and − 180 0 ,
respectively (Wang et al. [143]); Appendix 2.
* Method 9: Parameter estimates back-calculated from discrete time identification method (Wang and
Clements [147]).
Method 10: Km , Tm1 and τm are determined from the open loop time domain Ziegler-Nichols response
(Shinskey [16], page 151); Tm 2 assumed known.
Method 11: Parameters estimated from two points determined on process frequency response using a relay
and a relay in series with a delay (Tan et al. [39]); Appendix 2.
* Method 12: Parameter estimated back-calculated from discrete time identification method (Lopez et al. [84]).
* Method 13: Parameters estimated from a step response autotuning experiment – Honeywell UDC 6000
controller (Astrom et al. [30]).
Method 14: Tm1 = T m2 . τm and T m1 obtained from an open loop step test as follows:
Tm1 = 0. 794 ( t 70% − t 33% ) , τ m = 1.937 t 33% − 0.937 t 70% . K m assumed known (Miluse et al.
[27b]).
Method 15: K u and Tu are estimated from estimates of the ultimate and crossover frequencies (Pecharroman
and Pagola [165]) – as in Method 6, IPD model.
K me − sτ m
F. Integral squared plus time delay ( I 2PD ) model ( G m (s) = )
s2
Km e −s τ m
G. Second order system (repeated pole) plus integral plus time delay (SOSIPD) model ( G m (s) = )
s (1 + sTm )
2
Method 1: K u and Tu are estimated from estimates of the ultimate and crossover frequencies (Pecharroman and
Pagola [165]) – as in Method 6, IPD model.
Km e − sτ
m
H. Third order system plus time delay (TOLPD) model ( G m (s) = ).

(1 + sTm1 )(1 + sTm2 )(1 + sTm 3 )
Method 1: Km , Tm1 , Tm2 , Tm3 , τ m known.
K me − sτ
m
I. Unstable first order lag plus time delay model ( G m (s) = )

1 − sTm
Method 1: Km , Tm , τ m known.
Method 2: The model parameters are obtained by least squares fitting from the open loop frequency
response of the unstable process; this is done by determining the closed loop magnitude and
phase values of the (stable) closed loop system and using the Nichols chart to determine the
open loop response (Huang and Lin [154], Deshpande [164]).
K m e− s τm
J. Unstable second order system plus time delay model ( G m (s) = )

(1 − sTm1 )(1 + sTm 2 )
Method 1: Km , Tm1 , Tm2 , τ m known.
Method 2: The model parameters are obtained by least squares fitting from the open loop frequency
response of the unstable process; this is done by determining the closed loop magnitude and
phase values of the (stable) closed loop system and using the Nichols chart to determine the
open loop response (Huang and Lin [154], Deshpande [164]).
K m (1 − sTm3 ) e− s τ m
K. Second order system plus time delay model with a positive zero ( G m (s) = )
(1 + sTm1 )(1 + sTm2 )
K m (1 + sTm3 )e − sτ m
L. Second order system plus time delay model with a negative zero ( G m (s) = )
(1 + sTm1 )(1 + sTm 2 )
K m (1 + b1s + b2 s2 + b3s3 + b4 s4 + b ss5 ) e− s τ m
M. Fifth order system plus delay model ( G m ( s) = )

(1 + a s + a s
1 2
2
+ a3s 3 + a 4 s4 + a 5s5 )
Method 1: Km , b1 , b 2 , b 3 , b 4 , b5 , a1 , a2 , a3 , a4 , a5 , τ m known.
K m e − sτ
m
N. General model with a repeated pole ( G m ( s) = )

(1 + sTm ) n
* Method 1: Strejc’s method
O. General stable non-oscillating model with a time delay
P. Delay model ( G m ( s) = e − sτ m )
Note: * means that the procedure has not been fully described to date.
4. Organisation of the report
The tuning rules are organised in tabular form, as is indicated in the list of tables below. Within each table, the
tuning rules are classified further; the main subdivisions made are as follows:
(i) Tuning rules based on a measured step response (also called process reaction curve methods).
(ii) Tuning rules based on minimising an appropriate performance criterion, either for optimum regulator or
optimum servo action.
(iii) Tuning rules that gives a specified closed loop response (direct synthesis tuning rules). Such rules may be
defined by specifying the desired poles of the closed loop response, for instance, though more generally,
the desired closed loop transfer function may be specified. The definition may be expanded to cover
techniques that allow the achievement of a specified gain margin and/or phase margin.
(iv) Robust tuning rules, with an explicit robust stability and robust performance criterion built in to the design
process.
(v) Tuning rules based on recording appropriate parameters at the ultimate frequency (also called ultimate
cycling methods).
(vi) Other tuning rules, such as tuning rules that depend on the proportional gain required to achieve a quarter
decay ratio or magnitude and frequency information at a particular phase lag.
Some tuning rules could be considered to belong to more than one subdivision, so the subdivisions cannot be
considered to be mutually exclusive; nevertheless, they provide a convenient way to classify the rules. Tuning
rules for the variations that have been proposed in the ‘ideal’ PI and PID controller structure are included in the
appropriate table. In all cases, one column in the tables summarise the conditions under which the tuning rules
are designed to operate, if appropriate ( Y ( s) = closed loop system output, R( s) = closed loop system input).
K me − sτ
m
Tables 1-3: PI tuning rules – FOLPD model - G m ( s) =

1 + sTm
 1
Table 1: Ideal controller G c ( s) = Kc  1 +  . Eighty-five such tuning rules are defined; the references are
 i 
Ts
(a) Process reaction methods: Ziegler and Nichols [8], Hazebroek and Van der Waerden [9], Astrom
and Hagglund [3], Chien et al. [10], Cohen and Coon [11], Wolfe [12], Murrill [13] – page 356,
McMillan [14] – page 25, St. Clair [15] – page 22 and Shinskey [15a]. Twelve tuning rules are
defined.
(b) Performance index minimisation (regulator tuning): Minimum IAE - Murrill [13] – pages 358-363,
Shinskey [16] – page 123, ** Shinskey [17], Huang et al. [18], Yu [19]. Minimum ISE - Hazebroek
and Van der Waerden [9], Murrill [13]– pages 358-363, Zhuang and Atherton [20], Yu [19].
Minimum ITAE - Murrill [13] – pages 358-363, Yu [19]. Minimum ISTSE - Zhuang and Atherton
[20]. Minimum ISTES – Zhuang and Atherton [20]. Thirteen tuning rules are defined.
(c) Performance index minimisation (servo tuning): Minimum IAE – Rovira et al. [21], Huang et al. [18].
Minimum ISE - Zhuang and Atherton [20], han and Lehman [22]. Minimum ITAE - Rovira et al. [21].
Minimum ISTSE - Zhuang and Atherton [20]. Minimum ISTES – Zhuang and Atherton [20]. Seven
tuning rules are defined.
(d) Direct synthesis: Haalman [23], Chen and Yang [23a], Pemberton [24], Smith and Corripio [25], Smith
et al. [26], Hang et al. [27], Miluse et al. [27a], Gorecki et al. [28], Chiu et al. [29], Astrom et al. [30],
Davydov et al. [31], Schneider [32], McAnany [33], Leva et al. [34], Khan and Lehman [22], Hang et
al. [35, 36], Ho et al. [37], Ho et al [104], Tan et al. [39], Voda and Landau [40], Friman and Waller
[41], Smith [42], Cox et al. [43], Cluett and Wang [44], Abbas [45], Bi et al. [46], Wang and Shao
[47]. Thirty-one tuning rules are defined.
(e) Robust: Brambilla et al. [48], Rivera et al. [49], Chien [50], Thomasson [51], Fruehauf et al. [52],
Chen et al. [53], Ogawa [54], Lee et al. [55], Isaksson and Graebe [56], Chun et al. [57]. Ten tuning
rules are defined.
(f) Ultimate cycle: McMillan [58], Shinskey [59] – page 167, **Shinskey [17], Shinskey [16] – page
148, Hwang [60], Hang et al. [65], Zhuang and Atherton [20], Hwang and Fang [61]. Twelve tuning
rules are defined.
 1
Table 2: Controller G c ( s) = Kc  b +  Two direct synthesis tuning rules are defined by Astrom and Hagglund
 Ti s
[3] - page 205-208.

 1 
Table 3: Two degree of freedom controller: U(s) = K c 1 + E (s ) − α Kc R( s) . One performance index
 Tis 
minimisation tuning rule is defined by Taguchi and Araki [61a].
Tables 4-7: PI tuning rules - non-model specific

 1
Table 4: Controller G c ( s) = Kc  1 +  . Nineteen such tuning rules are defined; the references are:
 i 
Ts
(a) Ultimate cycle: Ziegler and Nichols [8], Hwang and Chang [62], ** Hang et al. [36], McMillan [14] –
page 90, Pessen [63], Astrom and Hagglund [3] – page 142, Parr [64] – page 191, Yu [122] – page
11. Seven tuning rules are defined.
(b) Other tuning rules: Parr [64] – page 191, McMillan [14] – pages 42-43, Parr [64] – page 192,
Hagglund and Astrom [66], Leva [67], Astrom [68], Calcev and Gorez [69], Cox et al. [70]. Eight
(c) Direct synthesis: Vrancic et al. [71], Vrancic [72], Friman and Waller [41], Kristiansson and
Lennartson [158a]. Four tuning rules are defined.
 1 
Table 5: Controller G c ( s) = K c  b +  . One direct synthesis tuning rule is defined by Astrom and Hagglund
 Ti s 
[3] – page 215.
 1 
Table 6: Controller U( s) = Kc  1 +  E( s) + Kc ( b − 1)R (s) . One direct synthesis tuning rule is defined by
 Tis
Vrancic [72].
Kc
Table 7: Controller U( s) = Kc Y(s) − E ( s) . One direct synthesis tuning rule is defined by Chien et al. [74].
Ti s
K me− sτ m
Tables 8-11: PI tuning rules – IPD model G m ( s) =
s
 1
Table 8: Controller G c ( s) = Kc  1 +  . Twenty such tuning rules are defined; the references are:
 i 
Ts
(a) Process reaction methods: Ziegler and Nichols [8], Wolfe [12], Tyreus and Luyben [75], Astrom and
Hagglund [3] – page 138. Four tuning rules are defined.
(b) Regulator tuning – performance index minimisation: Minimum ISE – Hazebroek and Van der
Waerden [9]. Minimum IAE - Shinskey [59] – page 74. Minimum ITAE - Poulin and Pomerleau [82].
Four tuning rules are defined.
(c) Ultimate cycle: Tyreus and Luyben [75], ** Shinskey [17]. Two tuning rules are defined.
(d) Robust: Fruehauf et al. [52], Chien [50], Ogawa [54]. Three tuning rules are defined.
(e) Direct synthesis: Wang and Cluett [76], Cluett and Wang [44], Rotach [77], Poulin and Pomerleau
[78], Kookos et al. [38]. Five tuning rules are defined.
(f) Other methods: Penner [79], Srividya and Chidambaram [80]. Two tuning rules are defined.
 1 1
Table 9: Controller G c ( s) = Kc  1 +  . One robust tuning rule is defined by Tan et al. [81].
 i  1 + Tf s
Ts
Kc
Table 10: Controller U( s) = Kc Y(s) − E ( s) . One direct synthesis tuning rule is defined by Chien et al. [74].
Ti s
 1 
Table 11: Two degree of freedom controller: U(s) = K c 1 + E (s ) − α Kc R( s) . Two performance index
 Tis 
minimisation - servo/regulator tuning rules are defined by Taguchi and Araki [61a] and Pecharroman
and Pagola [134b].
Km e− sτ m
Tables 12-14: PI tuning rules – FOLIPD model G m (s) =

s(1 + sTm )
 1
Table 12: Ideal controller G c ( s) = Kc  1 +  . Six such tuning rules are defined; the references are:
 i 
Ts
(a) Ultimate cycle: McMillan [58]. One tuning rule is defined.
(b) Regulator tuning – minimum performance index: Minimum IAE – Shinskey [59] – page 75. Shinskey
[59] – page 158. Minimum ITAE - Poulin and Pomerleau [82]. Four tuning rules are defined.
(c) Direct synthesis - Poulin and Pomerleau [78]. One tuning rule is defined.
 1 
Table 13: Controller G c ( s) = Kc  b +  . One direct synthesis tuning rule is defined by Astrom and Hagglund
 Ti s 
[3] – pages 210-212.
 1 
 Tis 
minimisation tuning rules are defined by Taguchi and Araki [61a] and Pecharroman and Pagola [134b].
K m e− sτ m
Km e− sτ
m
Tables 15-16: PI tuning rules – SOSPD model or

Tm1 s2 + 2ξ m Tm1s + 1
2
(1 + Tm1s)(1 + Tm2 s)
 1
Table 15: Ideal controller G c ( s) = Kc  1 +  . Ten tuning rules are defined; the references are:
 i 
Ts
(a) Robust: Brambilla et al. [48]. One tuning rule is defined.
(b) Direct synthesis: Tan et al. [39]. One tuning rule is defined.
(c) Regulator tuning – minimum performance index: Minimum IAE - Shinskey [59] – page 158, **
Shinskey [17], Huang et al. [18], Minimum ISE – McAvoy and Johnson [83], Minimum ITAE –
Lopez et al. [84]. Five tuning rules are defined.
(d) Servo tuning – minimum performance index: Minimum IAE - Huang et al. [18]. One tuning rule is
defined.
(e) Ultimate cycle: Hwang [60], ** Shinskey [17]. Two tuning rules are defined.
 1 
Table 16: Two degree of freedom controller: U(s) = K c 1 + E (s ) − α Kc R( s) . Three performance index

 Tis 
minimisation tuning rules are defined by Taguchi and Araki [61a] and Pecharroman and Pagola [134a],
[134b].
Km e −s τ m
Table 17: PI tuning rules – SOSIPD model (repeated pole) G m (s) =

s (1 + sTm )
2
 1 
 Tis 
minimisation tuning rules are defined by Taguchi and Araki [61a] and Pecharroman and Pagola [134b].
Km e − sτ
m
Tables 18-19: PI tuning rules – third order lag plus delay (TOLPD) model
(1 + sTm1 )(1 + sTm2 )(1 + sTm 3 )
 1
Table 18: Ideal controller G c ( s) = Kc  1 +  . One *** tuning rule is defined. The reference is Hougen [85].
 i 
Ts
 1 
Table 19: Two degree of freedom controller: U(s) = K c 1 + E (s ) − α Kc R( s) . One performance index
 Tis 
minimisation tuning rule is defined by Taguchi and Araki [61a].
K me − sτ
m
Table 20: PI tuning rules - unstable FOLPD model

1 − sTm
 1
Table 20: Ideal controller G c ( s) = Kc  1 +  . Six tuning rules are defined; the references are:
 i 
Ts
(a) Direct synthesis: De Paor and O’Malley [86], Venkatashankar and Chidambaram [87], Chidambaram
[88], Ho and Xu [90]. Four tuning rules are defined.
(b) Robust: Rotstein and Lewin [89]. One tuning rule is defined.
(c) Ultimate cycle: Luyben [91]. One tuning rule is defined.
K m e− s τ
m
Table 21: PI tuning rules - unstable SOSPD model

(1 − sTm1 )(1 + sTm 2 )
 1
Table 21: Ideal controller G c ( s) = Kc  1 +  . Three tuning rules are defined; the references are:
 i 
Ts
(a) Ultimate cycle: McMillan [58]. One tuning rule is defined.
(b) Minimum performance index – regulator tuning: Minimum ITAE – Poulin and Pomerleau [82]. Two
Table 22: PI tuning rules – delay model e −s τ m

 1
Table 22: Ideal controller G c ( s) = Kc  1 +  . Two tuning rules are defined; the references are:
 i 
Ts
(a) Direct synthesis: Hansen [91a].
(b) Minimum performance index – regulator tuning: Shinskey [57].
K m e− sτ
m
Tables 23-40: PID tuning rules - FOLPD model

1 + sTm
 1 
Table 23: Ideal controller G c ( s) = Kc  1 + + Td s . Fifty-seven tuning rules are defined; the references are:
 Ti s 
(a) Process reaction: Ziegler and Nichols [8], Astrom and Hagglund [3] – page 139, Parr [64] – page
194, Chien et al. [10], Murrill [13]- page 356, Cohen and Coon [11], Astrom and Hagglund [93]-
pages 120-126, Sain and Ozgen [94]. Eight tuning rules are defined.
(b) Minimum performance index – regulator tuning: Minimum IAE – Murrill [13] – pages 358-363,
Cheng and Hung [95]. Minimum ISE - Murrill [13] – pages 358-363, Zhuang and Atherton [20].
Minimum ITAE - Murrill [13] – pages 358-363. Minimum ISTSE - Zhuang and Atherton [20].
Minimum ISTES - Zhuang and Atherton [20]. Minimum error - step load change - Gerry [96]. Eight
(c) Minimum performance index – servo tuning: Minimum IAE - Rovira et al. [21], Wang et al. [97].
Minimum ISE - Wang et al. [97], Zhuang and Atherton [20]. Minimum ITAE - Rovira et al. [21],
Cheng and Hung [95], Wang et al. [97]. Minimum ISTSE – Zhuang and Atherton [20]. Minimum
ISTES – Zhuang and Atherton [20]. Nine tuning rules are defined.
(d) Ultimate cycle: Pessen [63], Zhuang and Atherton [20], Pi-Mira et al. [97a], Hwang [60], Hwang and
Fang [61], McMillan [58], Astrom and Hagglund [98], Li et al. [99], Tan et al. [39], Friman and
Waller [41]. Fourteen tuning rules are defined.
(e) Direct synthesis: Gorecki et al. [28], Smith and Corripio [25], Suyama [100], Juang and Wang [101],
Cluett and Wang [44], Zhuang and Atherton [20], Abbas [45], Camacho et al.[102, Ho et al. [103],
Ho et al [104], Morilla et al. [104a]. Fourteen tuning rules are defined.
(f) Robust: Brambilla et al. [48], Rivera et al. [49], Fruehauf et al. [52], Lee et al. [55]. Four tuning rules.
 1  1
Table 24: Ideal controller with first order filter G c ( s) = Kc  1 + + Td s . Three robust tuning rules are
 Ti s  Tf s + 1
defined by ** Morari and Zafiriou [105], Horn et al. [106] and Tan et al. [81].
 1  1 + b1s
Table 25: Ideal controller with second order filter G c ( s) = Kc  1 + + Td s . One robust tuning
  1 + a 1s + a 2s
2
Ti s
rule is defined by Horn et al. [106].
 1 
Table 26: Ideal controller with set-point weighting G c ( s) = Kc  b + + Td s . One direct synthesis tuning rule
 Ti s 
is defined by Astrom and Hagglund [3] – pages 208-210.
Table 27. Ideal controller with first order filter and set-point weighting:
 1  1  1 + 0.4Trs 
U(s) = K c 1 + + Td s  R (s ) − Y( s)  . One direct synthesis tuning rule is defined
 Tis  Tf s + 1  1 + sTr 
by Normey-Rico et al. [106a].
 1  1 + sTd
Table 28: Classical controller G c ( s) = K c 1 +  . Twenty tuning rules are defined; the references are:
 Ti s  T
1+ s d
N
(a) Process reaction: Hang et al. [36] – page 76, Witt and Waggoner [107], St. Clair [15] – page 21,
Shinskey [15a]. Five tuning rules are defined.
(b) Minimum performance index – regulator tuning: Minimum IAE - Kaya and Scheib [108], Witt and
Waggoner [107]. Minimum ISE - Kaya and Scheib [108]. Minimum ITAE - Kaya and Scheib [108],
Witt and Waggoner [107]. Five tuning rules are defined.
(c) Minimum performance index – servo tuning: Minimum IAE - Kaya and Scheib [108], Witt and
Waggoner [107]. Minimum ISE - Kaya and Scheib [108]. Minimum ITAE - Kaya and Scheib [108],
Witt and Waggoner [107]. Five tuning rules are defined.
(d) Direct synthesis: Tsang and Rad [109], Tsang et al. [111]. Two tuning rules are defined.
(e) Robust: Chien [50]. One tuning rule is defined.
(f) Ultimate cycle: Shinskey [59] – page 167, Shinskey [16] – page 143. Two tuning rules are defined.
 
  
1  Td s  . Two tuning rules are defined; the
Table 29: Non-interacting controller U (s) = K c 1 +  E( s) − Y (s )
 Tis  Td s 
 1+ 
 N 
references are:
(a) Minimum performance index – regulator tuning: Minimum IAE - Huang et al. [18].
(b) Minimum performance index – servo tuning: Minimum IAE - Huang et al. [18].
 1 Td s
Table 30: Non-interacting controller U( s) = Kc  1 +  E ( s) − Y(s) . Five tuning rules are defined; the
 Ti s
1+ d
sT
N
references are:
(a) Minimum performance index – servo tuning: Minimum ISE - Zhuang and Atherton [20], Minimum
ISTSE - Zhuang and Atherton [20]. Minimum ISTES - Zhuang and Atherton [20]. Three tuning
rules are defined.
(b) Ultimate cycle: Zhuang and Atherton [20], Shinskey [16] – page 148. Two tuning rules are defined.
 1  Td s
Table 31: Non-interacting controller U (s) =  K c + E (s) −
 Y ( s) . Six tuning rules are defined; the
 T i 
s sTd
1+
N
references are:
(a) Minimum performance index – regulator tuning: Minimum IAE – Kaya and Scheib [108]. Minimum
ISE – Kaya and Scheib [108]. Minimum ITAE – Kaya and Scheib [108].
(b) Minimum performance index – servo tuning: Minimum IAE – Kaya and Scheib [108]. Minimum ISE –
Kaya and Scheib [108]. Minimum ITAE – Kaya and Scheib [108].
Table 32: Non-interacting controller with setpoint weighting:
 1
Y( s) + Kc ( b − 1)Y( s) . Three ultimate cycle tuning rules are
Kc Td s
U( s) = Kc  b +  E (s) −
 Ti s  1 + Td s N
defined by Hang and Astrom [111], Hang et al. [65] and Hang and Cao [112].
 
 1  1 + Td s 
Table 33: Industrial controller U( s) = Kc 1 +   R( s) − Y(s)  . Six tuning rules are defined: the
 Ti s   1+ d
Ts 

 N 
reference are:
(a) Minimum performance index – regulator tuning: Minimum IAE - Kaya and Scheib [108]. Minimum
ISE - Kaya and Scheib [108]. Minimum ITAE - Kaya and Scheib [108]. Three tuning rules are
defined.
(b) Minimum performance index – servo tuning: Minimum IAE - Kaya and Scheib [108]. Minimum ISE -
Kaya and Scheib [108]. Minimum ITAE - Kaya and Scheib [108]. Three tuning rules are defined.
 1 
Table 34: Series controller Gc ( s) = Kc  1 +  (1 + sTd ) . Three tuning rules are defined; the references are:
 Ti s 
(a) ******: Astrom and Hagglund [3] – page 246.
(b) Ultimate cycle: Pessen [63].
(c) Direct synthesis: Tsang et al. [110].
 
 1  sTd 
Table 35: Series controller with filtered derivative Gc ( s) = Kc  1 +  1 +  . One robust tuning rule is
 Ti s  1 + d 
sT

 N 
defined by Chien [50].
 
 1 Td s 
Table 36: Controller with filtered derivative G c ( s) = Kc  1 + +  . Three tuning rules are defined; the
 Ti s 1 + s Td 

 N
references are:
(a) Robust: Chien [50], Gong et al. [113]. Two tuning rules are defined.
(b) Direct synthesis: Davydov et al. [31]. One tuning rule is defined.
 1 
Table 37: Alternative non-interacting controller 1 - U( s) = Kc 1 +  E( s) − Kc Td sY ( s) . Six ultimate cycle
 Ti s 
tuning rules are defined; the references are: Shinskey [59] – page 167, ** Shinskey [17], Shinskey [16] –
page 143, VanDoren [114].
 1   1 + 05
. τm s + 0.0833τ m s 2 
2
Table 38: Alternative filtered derivative controller - G c ( s) = Kc  1 +  . One direct
 Ti s   [1 + 01. τ ms]
2 

synthesis tuning rule is defined by Tsang et al. [110].
E (s) − K c (1 + Td s)Y(s) . Two direct synthesis tuning rules are defined by Chien
Kc
Table 39: I-PD controller U(s) =
Ti s
et al. [74] and Argelaguet et al. [114a].
   
   
1 T s β T s
Table 40: Two degree of freedom controller: U (s) = K c 1 + + d  E( s) − K c  α + d  R (s) . One
 Tis Td   Td 
 1+ s  1+ s
 N   N 
performance index minimisation tuning rule is defined by Taguchi and Araki [61a].
Tables 41-48: PID tuning rules - non-model specific

 1 
Table 41: Ideal controller G c ( s) = Kc  1 + + Td s . Twenty five tuning rules are defined; the references are
 Ti s 
(a) Ultimate cycle: Ziegler and Nichols [8], Blickley [115], Parr [64] – pages 190-191, De Paor [116],
Corripio [117] – page 27, Mantz andTacconi [118], Astrom and Hagglund [3] – page 142, Astrom
and Hagglund [93], Atkinson and Davey [119], ** Perry and Chilton [120], Yu [122] – page 11, Luo
et al. [121], McMillan [14] – page 90, McAvoy and Johnson [83], Karaboga and Kalinli [123], Hang
and Astrom [124], Astrom et al. [30], St. Clair [15] - page 17, Shin et al. [125]. Nineteen tuning rules
are defined.
(b) Other tuning: Harriott [126], Parr [64] – pages 191, 193, McMillan [14] - page 43, Calcev and Gorez
[69], Zhang et al. [127], Garcia and Castelo [127a]. Six tuning rules are defined.
 
 1 Td s 
Table 42: Controller with filtered derivative Gc ( s) = Kc 1 + +  . Eight tuning rules are defined; the
 Ti s 1 + Td s 

 N 
references are:
(a) Direct synthesis: Vrancic [72], Vrancic [73], Lennartson and Kristiansson [157], Kristiansson and
Lennartson [158], Kristiansson and Lennartson [158a]. Six tuning rules are defined.
(b) Other tuning: Leva [67], Astrom [68]. Two tuning rules are defined.
Table 43: Ideal controller with set-point weighting:
(
U( s) = Kc Fp R(s) − Y(s) + ) 1
Ts
( Fi R(s) − Y(s)) + Td s( Fd R(s) − Y(s)) . One ultimate cycle tuning rule is
i
defined by Mantz and Tacconi [118].

 1 
Table 42: Ideal controller with proportional weighting G c ( s) = Kc  b + + Td s . One direct synthesis tuning
 Ti s 
rule is defined by Astrom and Hagglund [3] – page 217.
 1  K Ts
Table 44: Non-interacting controller U( s) = Kc 1 +  E( s) − c d Y( s) . One ultimate cycle tuning rule is
 Ti s  sT
1+ d
N
defined by Fu et al. [128].
 1
Table 45: Series controller U( s) = Kc  1 +  (1 + sTd ) . Three ultimate cycle tuning rules are defined by Pessen
 Ti s 
[131], Pessen [129] and Grabbe et al. [130].
 
 1  sTd 
Table 46: Series controller with filtered derivative U( s) = Kc 1 + 1 +  . One ultimate cycle tuning
 Ti s   1 + d 
sT

 N 
rule is defined by Hang et al. [36] - page 58.
 1  1 + sTd
Table 47: Classical controller U( s) = Kc  1 +  . One ultimate cycle tuning rule is defined by Corripio
 Ti s  T
1+ s d
N
[117].
 1
Table 48: Non-interacting controller U( s) = Kc  1 +  E ( s) − Kc Td sY (s) . One ultimate cycle tuning rule is
 Ti s
defined by VanDoren [114].
K m e− sτ m
Tables 49-58: PID tuning rules - IPD model
s
 1 
Table 49: Ideal controller Gc ( s) = K c  1 + + Td s . Five tuning rules are defined; the references are:
 Ti s 
(a) Process reaction: Ford [132], Astrom and Hagglund [3] – page 139. Two tuning rules are defined.
(b) Direct synthesis: Wang and Cluett [76], Cluett and Wang [44], Rotach [77]. Three tuning rules are
defined.
Table 50: Ideal controller with first order filter, set-point weighting and output feedback:
 1  1  1 + 0.4Trs 
U(s ) = K c 1 + + Tds   R( s) − Y(s )  − K 0 Y(s) . One direct synthesis tuning rule
 Ti s  Tf s + 1  1 + sTr 
has been defined by Normey-Rico et al. [106a].
 
 1 Td s 
Table 51: Ideal controller with filtered derivative G c ( s) = Kc  1 + +  . One robust tuning rule has
 Tsi 1+
sTd 

 N 
been defined by Chien [50].
 
 1  Td s 
Table 52: Series controller with filtered derivative G c ( s) = Kc  1 +  1 +  . One robust tuning rule has
 i  1 + d 
Ts Ts
 N 
 
 1   1 + Td s 
Table 53: Classical controller G c ( s) = Kc  1 +   . Five tuning rules have been defined; the references
 Ti s   1 + Td s 
 
 N 
are:
(a) Ultimate cycle: Luyben [133], Belanger and Luyben [134]. Two tuning rules have been defined.
(b) Robust: Chien [50]. One tuning rule has been defined.
(c) Performance index minimisation – regulator tuning: ** Minimum IAE - Shinskey [17], Shinskey [59]
– page 74. Two tuning rules have been defined.
 1 
Table 54: Alternative non-interacting controller 1: U( s) = Kc 1 +  E( s) − Kc Td sY ( s) . Two performance index
 T is
minimisation rules – minimum IAE regulator tuning have been defined by Shinskey [59] – page 74 and
** Shinskey [17].
E (s) − K c (1 + Td s)Y(s) . One direct synthesis tuning rule has been defined by
Kc
Table 55: I-PD controller U(s) =
Ti s
Chien et al. [74].
1
Table 56: Controller U(s) = Kc (1 + ) E( s) + Kc ( b − 1) R (s) − Kc TdsY( s) . One direct synthesis tuning rule has
Tis
been defined by Hansen [91a].
   
   
1 T s β T s
Table 57: Two degree of freedom controller: U (s) = K c 1 + + d  E( s) − K c  α + d  R (s) . Two
 1 + s   1 + s 
 N   N 
minimum performance index – servo/regulator tuning have been defined by Taguchi and Araki [61a] and
Pecharroman and Pagola [134a].
Km e − sτ
m
Tables 58-67: PID tuning rules - FOLIPD model

s(1 + sTm )
 1 
Table 58: Ideal controller Gc ( s) = K c  1 + + Td s . One ultimate cycle tuning rule has been defined by Millan
 Ti s 
[58].
 1  1
Table 59: Ideal controller with filter G c ( s) = Kc  1 + + Td s . Three robust tuning rules have been
 Ti s  1 + Tf s
defined by Tan et al. [81], Zhang et al. [135] and Tan et al. [136].
 1 
Table 60: Ideal controller with set-point weighting G c ( s) = Kc  b + + Td s . One direct synthesis tuning rule
 Ti s 
has been defined by Astrom and Hagglund [3] - pages 212-213.
 
 1   1 + Td s 
Table 61: Classical controller G c ( s) = Kc  1 +   . Five tuning rules have been defined; the references
 Ti s   1 + Td s 
 
 N 
are as follows:
(a) Robust: Chien [50]. One tuning rule is defined.
(b) Minimum performance index – regulator tuning: Minimum IAE – Shinskey [59] – page 75, Shinskey
[59] – pages 158-159, Minimum ITAE - Poulin and Pomerleau [82], [92]. Four tuning rules are
defined.
 
 1  Td s 
Table 62: Series controller with derivative filtering G c ( s) = Kc  1 +  1 +  . One robust tuning rule has
 i  1 + d 
Ts Ts
 N 
 1 
Table 63: Alternative non-interacting controller 1: U(s) = Kc 1 +  E (s) − K c Td sY ( s) . Two minimum
 Ti s 
performance index (minimum IAE) – regulator tuning rules have been defined by Shinskey [59] – page
75, page 159.
 
 1 Td s 
Table 64: Ideal controller with filtered derivative: G c ( s) = Kc  1 + +  . One robust tuning rule has
 Ti s 1 + s Td 

 N

( )
U(s) = Kc Fp R ( s) − Y( s) +
Kc
Ti s
[ Fi R(s) − Y(s)] + Kc Td s[ Fd R (s) − Y(s)] . One ultimate cycle tuning rule
has been defined by Oubrahim and Leonard [138].
  
 1+ T s   1 + T s 
Table 66: Alternative classical controller: G c ( s) = Kc  i  d  . One direct synthesis tuning rule has
 1 + Td s   1 + Td s 
 N  N 
been defined by Tsang and Rad [109].
Table 67: Two degree of freedom controller:

   
   
1 T s β T s
U (s) = K c 1 + + d  E( s) − K c  α + d  R (s) . Two minimum performance index –
 1+ s  1+ s
 N   N 
servo/regulator tuning have been defined by Taguchi and Araki [61a] and Pecharroman and Pagola
[134a].
Km e− sτm
K m e− sτ m
Tables 68-79: PID tuning rules - SOSPD model or

(1 + sTm1 )(1 + sTm 2 ) Tm1 s + 2ξ m Tm1s + 1
2 2
 1 
Table 68: Ideal controller Gc ( s) = K c  1 + + Td s . Twenty seven tuning rules have been defined; the
 Ti s 
references are:
(a) Minimum performance index – servo tuning: Minimum ITAE – Sung et al. [139]. One tuning rule is
defined.
(b) Minimum performance index – regulator tuning: Minimum ITAE – Sung et al. [139], Lopez et al.
[84]. One tuning rule is defined.
(c) Ultimate cycle: Hwang [60], Shinskey [16] – page 151. Three tuning rules are defined.
(d) Direct synthesis: Hang et al. [35], Ho et al. [140], Ho et al. [141], Ho et al. [142], Wang et al. [143],
Leva et al. [34], Wang and Shao [144], Pemberton [145], Pemberton [24], Suyama [100], Smith et al.
[146], Chiu et al. [29], Wang and Clemens [147], Gorez and Klan [147a], Miluse et al. [27a], Miluse et
al. [27b], Seki et al. [147b], Landau and Voda [148]. Nineteen tuning rules are defined.
(e) Robust: Brambilla et al. [48], Chen et al. [53], Lee et al. [55]. Three tuning rules are defined.
 1  1
Table 69: Filtered controller G c ( s) = Kc  1 + + Td s . One robust tuning rule has been defined by Hang
 Ti s  Tf s + 1
et al. [35].
 1  b s+1
Table 70: Filtered controller G c ( s) = Kc  1 + + Td s 1 . One robust tuning rule has been defined by
 Ti s  a 1s + 1
Jahanmiri and Fallahi [149].
 
 1   1 + Td s 
Table 71: Classical controller G c ( s) = Kc  1 +   . Seven tuning rules have been defined; the
 Ti s   1 + Td s 
 
 N 
references are:
(a) Minimum performance index – regulator tuning: Minimum IAE - Shinskey [59] – page 159, **
Shinskey [59], ** Shinskey [17], ** Shinskey [17]. Minimum ISE – McAvoy and Johnson [83]. Five
(b) Direct synthesis: Astrom et al. [30], Smith et al. [26]. Two tuning rules are defined.
 1   1 + NTd s 
Table 72: Alternative classical controller G c ( s) = Kc  1 +   . One ***** tuning rule has been
 Ti s   1 + Td s 
defined by Hougen [85].
 1
Table 73: Alternative non-interacting controller 1: U( s) = K c  1 +  E ( s) − Kc Td sY ( s) . Three minimum
 Ti s
performance index (minimum IAE) – regulator tuning rules have been defined by Shinskey [59] – page
158, ** Shinskey [17], ** Shinskey [17].
 1 
Table 74: Series controller G c ( s) = K c 1 +  (1 + Td s) . One minimum performance index - regulator tuning rule
 Ti s 
has been defined by Haalman [23].
 
  
1  Td s  . Two tuning rules have been
Table 75: Non-interacting controller U (s) = K c 1 +  E( s) − Y (s )
 Tis  Td s 
 1+ 
 N 
defined. The references are:
(a) Minimum performance index – regulator tuning: Minimum IAE - Huang et al. [18].
(b) Minimum performance index – servo tuning: Minimum IAE - Huang et al. [18].
( )
U(s) = Kc Fp R ( s) − Y( s) + c [ Fi R( s) − Y(s) ] + Kc Td s[ Fd R ( s) − Y( s)] . One ultimate cycle tuning rule
K
Ti s
has been defined by Oubrahim and Leonard [138].
 1 
Table 77: Non-interacting controller U( s) = Kc  b + [ R (s) − Y(s) ] − ( c + Tds)Y( s) . One direct synthesis tuning
 i 
Ts
rule has been defined by Hansen [150].
 
 1 Td s 
Table 78: Ideal controller with filtered derivative G c ( s) = Kc  1 + +  . Two tuning rules are defined;
 Ts
i 1 + d 
sT
 N 
the references are:
(a) Direct synthesis: Hang et al. [151].
(b) Robust: Hang et al. [151].
   
   
1 T s β T s
Table 79: Two degree of freedom controller: U (s) = K c 1 + + d  E( s) − K c  α + d  R (s) . Three
 1 + s   1 + s 
 N   N 
minimum performance index – servo/regulator tuning rules have been defined by Taguchi and Araki
[61a] and Pecharroman and Pagola [134a], [134b].
K me − sτ m
Table 80: PID tuning rules - I 2PD model G m (s) =
s2
1
Table 80: Controller U(s) = Kc (1 + ) E( s) + Kc ( b − 1) R (s) − Kc TdsY( s) . One direct synthesis tuning rule has
Tis
been defined by Hansen [91a].
Km e −s τ m
Table 81: PID tuning rules – SOSIPD model (repeated pole) G m (s) =
s (1 + sTm )
2
   
   
1 T s β T s
Table 81: Two degree of freedom controller: U (s) = K c 1 + + d  E( s) − K c  α + d  R (s) . Two
 1+ s  1+ s
 N   N 
minimum performance index – servo/regulator tuning rules have been defined by Taguchi and Araki
[61a] and Pecharroman and Pagola [134a].
K m (1 − sTm3 ) e− s τ m
Tables 82-84: PID tuning rules - SOSPD model with a positive zero
(1 + sTm1 )(1 + sTm2 )
 
 1 Td s 
Table 82: Controller with filtered derivative G c ( s) = Kc  1 + +  . One robust tuning rule has been
 Ti s 1 + Td s 

 N 
 
 1   1 + Td s 
Table 83: Classical controller G c ( s) = Kc  1 +   . One robust tuning rule has been defined by Chien
 Ti s   1 + Td s 
 
 N 
[50].
 
 1  Td s 
 i  1 + d 
Ts Ts
 N 
K m (1 + sTm 3 ) e− sτ m
Tables 85-88: PID tuning rules - SOSPD model with a negative zero
(1 + sTm1 )(1 + sTm 2 )
 1 
Table 85: Ideal controller Gc ( s) = K c  1 + + Td s . One minimum performance index tuning rule has been
 Ti s 
defined by Wang et al. [97].
 
 1 Td s 
Table 86: Ideal controller with filtered derivative G c ( s) = Kc  1 + +  . One robust tuning rule has been
 Ti s 1 + Td s 

 N 
 
 1   1 + Td s 
Table 87: Classical controller G c ( s) = Kc  1 +   . One robust tuning rule has been defined by Chien
 Ti s   1 + Td s 
 
 N 
[50].
 
 1  Td s 
 i  1 + d 
Ts Ts
 N 
Km e− sτm
Table 89-90: PID tuning rules - TOLPD model

(1 + sTm1 )(1 + sTm 2 )(1 + sTm 3 )
 1 
Table 89: Ideal controller Gc ( s) = K c  1 + + Td s . Two minimum performance index tuning rules have been
 Ti s 
defined by Polonyi [153].
   
   
1 T s β T s
Table 90: Two degree of freedom controller: U (s) = K c 1 + + d  E( s) − K c  α + d  R (s) .One
 1+ s  1+ s
 N   N 
minimum performance index tuning rule has been defined by Taguchi and Araki [61a].
K m e− sτ
m
Tables 91-93: PID tuning rules - unstable FOLPD model

1 − sTm
 1 
Table 91: Ideal controller Gc ( s) = K c  1 + + Td s . Three direct synthesis tuning rules are defined by De Paor
 Ti s 
and O’Malley [86], Chidambaram [88] and Valentine and Chidambaram [154].
 1 KTs
Table 92: Non-interacting controller U( s) = Kc  1 +  E ( s) − c d Y(s) . Two tuning rules have been
 Ti s sT
1+ d
N
defined; the references are:
(a) Minimum performance index – servo tuning: Minimum IAE - Huang and Lin [155]
(b) Minimum performance index – servo tuning: Minimum IAE - Huang and Lin [155]
 
 1   1 + Td s 
Table 93: Classical controller G c ( s) = Kc  1 +   . One performance index minimisation – regulator
 Ti s   1 + Td s 
 
 N 
tuning rule has been defined by Shinskey [16]– page 381.
Km e − sτm
Tables 94-97: PID tuning rules - unstable SOSPD model G m ( s) =

(1 − sTm1 )(1 + sTm 2 )
 1 
Table 94: Ideal controller Gc ( s) = K c  1 + + Td s . Two tuning rules have been defined; the references are
 Ti s 
(a) Ultimate cycle: McMillan [58]
(b) Robust: Rotstein and Lewin [89].
 
 1   1 + Tds 
Table 95: Classical controller G c ( s) = Kc  1 +   . Two minimum performance index tuning rules
 Ti s   1 + Td s 
 
 N 
(regulator - minimum ITAE) have been defined by Poulin and Pomerleau [82], [92].
 1
Table 96: Series controller Gc ( s) = Kc 1 +  (1 + Td s) . One direct synthesis tuning rule has been defined by
 Ti s
Ho and Xu [90].
 1 KTs
Table 97: Non-interacting controller U( s) = Kc  1 +  E ( s) − c d Y(s) . Two tuning rules have been
 Ti s sT
1+ d
N
defined; the references are
(a) Minimum performance index – servo tuning: Minimum IAE - Huang and Lin [155]
(b) Minimum performance index – regulator tuning: Minimum IAE - Huang and Lin [155]
K m e − sτ
m
Table 98: PID tuning rules – general model with a repeated pole G m ( s) =
(1 + sTm ) n
 1 
Table 98: Ideal controller Gc ( s) = K c  1 + + Td s . One direct synthesis tuning rule has been defined by
 Ti s 
Skoczowski and Tarasiejski [156]
Table 99: PID tuning rules – general stable non-oscillating model with a time delay
 1 
Table 99: Ideal controller Gc ( s) = K c  1 + + Td s . One direct synthesis tuning rule has been defined by Gorez
 Ti s 
and Klan [147a].
Tables 100-101: PID tuning rules – fifth order model with delay
K (1 + b1s + b2 s2 + b3s3 + b4 s4 + b ss5 ) e− s τ m
G m ( s) = m
(
1 + a1s + a2 s2 + a3s 3 + a 4 s4 + a 5s5)
 1 
Table 100: Ideal controller Gc ( s) = K c  1 + + Td s . One direct synthesis tuning rule is defined by Vrancic et
 Ti s 
al. [159].
 
 1 Tds 
Table 101: Controller with filtered derivative G c ( s) = Kc  1 + +  . One direct synthesis tuning rule is
 Ts
i 1+
Td s 

 N 
defined by Vrancic et al. [159].
** some more information needed.
The number of tuning rules in each table is included in the data. Servo and regulator tuning rules are counted
separately; otherwise, rules in which different tuning parameters are provided for a number of variations in
process parameters or desired response parameters (such as desired gain margin, phase margin or closed loop
response time constant) are counted as one tuning rule. Tabular summaries are provided below.
Table A: Model structure and tuning rules – a summary for PI controllers
Process Minimise Direct Ultimate Robust

Model reaction Performanc Synthesis cycle tuning Other rules Total
e index
Stable 12 21 33 10 12 0 88 (53%)
FOLPD
Non-model 0 0 7 7 0 8 23 (14%)
specific
IPD 4 6 6 2 4 2 24 (14%)
FOLIPD 0 6 2 1 0 0 9 (5%)
SOSPD 0 9 1 2 1 0 13 (7%)
SOSIPD 0 2 0 0 0 0 2 (1%)
TOLPD 0 1 0 0 0 1 2 (1%)
Unstable 0 0 4 1 1 0 6 (4%)
FOLPD
Unstable 0 2 0 1 0 0 3 (2%)
SOSPD
Delay 0 1 1 0 0 0 2 (1%)
model
TOTAL 16 48 54 24 18 11 171
Table B: Model structure and tuning rules – a summary for PID controllers

e index
Stable 13 45 24 28 12 1 123 (44%)
FOLPD
Non-model 0 0 7 27 0 8 42 (15%)
specific
IPD 2 6 6 2 3 0 19 (7%)
FOLIPD 0 8 2 2 6 0 18 (6%)
SOSPD 0 16 23 4 6 1 50 (17%)
I 2PD 0 0 1 0 0 0 1 (0%)
SOSIPD 0 2 0 0 0 0 2 (1%)
SOSPD – 0 0 0 0 3 0 3 (1%)
pos. zero
SOSPD – 0 1 0 0 3 0 4 (1%)
neg. zero
TOLPD 0 3 0 0 0 0 3 (1%)
Unstable 0 3 3 0 0 0 6 (2%)
FOLPD
Unstable 0 4 1 1 1 0 7 (2%)
SOSPD
Higher 0 0 4 0 0 0 4 (1%)
order
TOTAL 15 88 71 64 34 10 282
Table C: Model structure and tuning rules – a summary for PI/PID controllers

e index
Stable 25 66 57 38 24 1 211 (47%)
FOLPD
Non-model 0 0 14 34 0 16 65 (15%)
specific
IPD 6 12 12 4 7 2 43 (9%)
FOLIPD 0 14 4 3 6 0 27 (6%)
SOSPD 0 25 24 6 7 1 63 (14%)
I 2PD 0 0 1 0 0 0 1 (0%)
SOSIPD 0 4 0 0 0 0 4 (1%)
SOSPD – 0 0 0 0 3 0 3 (1%)
pos. zero
SOSPD – 0 1 0 0 3 0 4 (1%)
neg. zero
TOLPD 0 4 0 0 0 1 5 (1%)
Unstable 0 3 7 0 1 0 12 (3%)
FOLPD
Unstable 0 6 1 2 1 0 10 (2%)
SOSPD
Delay 0 1 1 0 0 0 2 (0%)
model
Higher 0 0 4 0 0 0 4 (1%)
order
TOTAL 31 136 125 88 52 21 453
Table D: PI controller structure and tuning rules – a summary
Stable Non-
Controller structure FOLPD model IPD FOLIP SOSPD Other Total
specific D
 1
G c ( s) = Kc  1 +  85 19 20 6 10 12 152
 i 
Ts
(92%)
 1
G c ( s) = Kc  b +  2 1 0 1 2 0 6
 Ti s
(4%)
 1 
U(s) = K c 1 + E (s ) − α Kc R( s)
Tis 
1 1 2 2 1 3 10
 (4%)
Kc
U( s) = Kc Y(s) − E ( s) 0 1 1 0 0 0 2
Ti s
(1%)
 1 1
G c ( s) = Kc  1 +  0 0 1 0 0 0 1
 i  1 + Tf s
Ts
(0%)
Total 88 21 23 8 12 15 167
Table E: PID controller structure and tuning rules – a summary
Stable Non-
specific D
 1  126
G c ( s) = Kc  1 + + Td s 57 25 5 1 27 11 (45%)
 Ti s 
 
 1 Td s 
G c ( s) = Kc  1 + +  3 8 1 1 2 3 18
 Ti s 1 + s Td 
 N  (6%)
 1  1 7
G c ( s) = Kc  1 + + Td s 3 0 0 3 1 0 (3%)
 Ti s  Tf s + 1
 1  b s+1 1
G c ( s) = Kc  1 + + Td s 1 0 0 0 0 1 0 (0%)
 Ti s  a 1s + 1
 1  1 + b1s 3
G c ( s) = Kc  1 + + Td s 3 0 0 0 0 0 (1%)
 Ti s  1 + a 1s + a 2s 2
 1  3
G c ( s) = Kc  b + + Td s 1 1 0 1 0 0 (1%)
 Ti s 
Subtotal 67 34 6 6 31 14 158
(56%)
 1  1 + sTd
G c ( s) = K c 1 + 
Ti s  T 20 1 5 5 7 5 43
 1+ s d (15%)
N
  
 1+ T s   1 + T s  1
G c ( s) = Kc  i  d 
0 0 0 1 0 0 (0%)
 1 + Td s   1 + Td s 
 N  N 
 1   1 + NTd s  1
G c ( s) = Kc  1 +   0 0 0 0 1 0 (0%)
 Ti s   1 + Td s 
 1  8
Gc ( s) = Kc  1 +  (1 + sTd ) 3 3 0 0 1 1 (3%)
 Ti s 
 
 1  sTd  6
Gc ( s) = Kc  1 +   1 +  1 1 1 1 0 2 (2%)
 Ti s  sT
1+ d 

 N 
 1   1 + 05
. τ m s + 0.0833τ m s 2  1
2
G c (s) = K c  1 +  
  1 0 0 0 0 0 (0%)
 Ti s   [1 + 01. τ ms]
2

Subtotal 25 5 6 7 9 8 60
(22%)
 1 Td s
U( s) = Kc  1 +  E ( s) − Y(s) 5 0 0 0 0 0 5
 T i 
s
1+
sTd
(2%)
N
 
  
1  Td s 
U (s) = K c 1 +  E( s) − Y (s )
 Tis  Td s  2 0 0 0 4 0 6
 1+  (2%)
 N 
Stable Non-
specific D
 1  Td s
U (s) =  K c + E (s) −
 Y ( s) 6 0 0 0 0 0 6
 Ti 
s sTd
1+ (2%)
N
 1  K Ts
U( s) = Kc 1 +  E( s) − c d Y( s) 0 1 0 0 1 0 6
 Ti 
s sT
1+ d (2%)
N
 1 Kc Td s 3
U(s) = Kc  b +  E( s) − Y(s) + Kc ( b − 1) Y(s)
 Ti s 1 + Td s N 3 0 0 0 0 0 (1%)
 
 1  1 + Td s  6 0 0 0 0 0 6
U( s) = Kc 1 +   R( s) − Y(s)  (2%)
 Ti s   1+ d
Ts 

 N 
 1  14
U( s) = Kc 1 +  E( s) − Kc Td sY ( s) 6 1 2 2 3 0 (5%)
 Ti s 
 1 1
U(s) = Kc  b + [R (s) − Y(s) ] − ( c + Td s) Y(s) 0 0 0 0 1 0 (0%)
 i 
Ts
E (s) − K c (1 + Td s)Y(s)
Kc 3
U(s) =
Ti s 2 0 1 0 0 0 (1%)
   
   
U(s ) = Kc  1 +
1
+
Td s  E(s ) − K  α + βTds  R (s ) 1 0 1 1 3 3 9
 Ti s T  c
 T 
 1+ d s  1+ d s (3%)
 N   N 
(
U(s) = K c Fp R(s) − Y(s) + ) 1
Ti s
( Fi R(s) − Y(s)) + Td s( Fd R(s) − Y(s)) 0 1 0 1 1 0 3
(1%)
1 0 0 1 0 0 1 2
U( s) = K c (1 + ) E (s ) + K c ( b − 1)R ( s) − K cTd sY(s )
Tis (1%)
 1  1  1 + 0.4Trs  1 0 0 0 0 0 1
U( s) = K c 1 + + Td s − Y( s)
 T s +1
R (s)
 1 + sTr
 Ti s  f   (0%)
 1  1  1 + 0.4Tr s  0 0 1 0 0 0 1
U(s) = Kc 1 + + Td s  R (s ) − Y(s) − K0 Y (s )
 Ti s  Tf s + 1  1 + sTr  (0%)
Subtotal 32 3 6 4 8 8 61
(22%)
Total 124 42 18 17 49 30 280

3. Tuning rules for PI and PID controllers
K me − sτ m
 1
Table 1: PI tuning rules - FOLPD model - G m ( s) = – ideal controller G c ( s) = Kc  1 +  . 84 tuning
1 + sTm  i 
Ts
rules
Rule Kc Ti Comment
Process reaction
Ziegler and Nichols [8] 09
. Tm τm
3.33τm Quarter decay ratio. ≤1
K mτ m Tm
Model: Method 1.
α Tm βτ m
Hazebroek and Van der K mτ m
Waerden [9]
τm Tm α β τm Tm α β τm Tm α β
Model: Method 1 0.2 0.68 7.14 1.1 0.90 1.49 2.0 1.20 1.00
0.3 0.70 4.76 1.2 0.93 1.41 2.2 1.28 0.95
0.4 0.72 3.70 1.3 0.96 1.32 2.4 1.36 0.91
0.5 0.74 3.03 1.4 0.99 1.25 2.6 1.45 0.88
0.6 0.76 2.50 1.5 1.02 1.19 2.8 1.53 0.85
0.7 0.79 2.17 1.6 1.06 1.14 3.0 1.62 0.83
0.8 0.81 1.92 1.7 1.09 1.10 3.2 1.71 0.81
0.9 0.84 1.75 1.8 1.13 1.06 3.4 1.81 0.80
1.0 0.87 1.61 1.9 1.17 1.03
τ τm τm
α = 0.5 m + 01
. β= > 35
.
Tm . τ m − 1.2Tm
16 Tm
Astrom and Hagglund [3] – 063
. Tm Ultimate cycle Ziegler-
page 138 Km τ m 3.2τm Nichols equivalent
Model: Not relevant
0.6Tm 0% overshoot -
Chien et al. [10] - regulator K mτ m 4τ m τ
. < m < 10
011 .
Tm
Model: Method 1
07
. Tm τm 20% overshoot -
2.33
K mτ m Km τ
. < m < 10
011 .
Tm
. Tm
regulator – page 150 K mτ m 2.3τm 20% overshoot
Model: Method 1
035
. Tm 0% overshoot -
Km τ m 117
. Tm τ
Chien et al. [10] - servo . < m < 10
011 .
Tm
0.6Tm 20% overshoot-
Model: Method 1
K mτ m Tm τ
. < m < 10
011 .
Tm
 2

 3.33 τ m + 0.31 τm  
Cohen and Coon [11] 1  Tm 

process reaction  0.9 + 0083
.  Tm  Tm   Quarter decay ratio
K m  τm  Tm  
 τ 
1 + 2.22 m
Model: Method 1.  Tm 
 
Rule Kc Ti Comment
α Tm βτ m
Two constraints criterion - K mτ m Decay ratio = 0.4; minimum
Wolfe [12] error integral (regulator
τm Tm α β τm Tm α β
mode).
Model: Method 2. 0.2 4.4 3.23 1.0 0.78 1.28
0.5 1.8 2.27 5.0 0.30 0.53
0.946 0 .583 Quarter decay ratio;
0928
.  Tm  Tm  τ m 
Two constraints criterion -     minimum error integral
Murrill [13] – page 356 K m  τm  1078
.  Tm  (servo mode).
τ
Model: Method 3 . ≤ m ≤ 10
01 .
Tm
McMillan [14] – page 25 Km τm Time delay dominant
Model: Method 3 3 processes
St. Clair [15] – page 22 0333
. Tm Tm Tm
≤ 30
.
Model: Method 3 K mτ m τm
Shinskey [15a] 0. 667 Tm 3.78τ m Tm
= 0.167
Model: Method 1 K mτ m τm
Regulator tuning Performance index minimisation
Minimum IAE - Murrill [13] –  Tm 
0.986
Tm  τ m 
0. 707
τm
pages 358-363
0984
.
    . ≤
01 ≤ 10
.
Km  τm  Tm
Model: Method 3 0608
.  Tm 
. Tm Km τ m
100 . τm
30 τm Tm = 0.2
Minimum IAE - Shinskey . Tm K mτ m
104 2.25τ m τm Tm = 0.5
[16] – page 123
Model: Method 6 . Tm Km τ m
111 . τm
145 τm Tm = 1
. Tm K mτ m
139 τm τm Tm = 2
Minimum IAE - Shinskey 0.95Tm K mτ m . τm
34 τm Tm = 01
.
[17] – page 38 0.95Tm K mτ m 2.9τ m τm Tm = 0.2
Model: Method 6
Minimum IAE – Huang et τm
al. [18] ( 1) 1 ( 1) . ≤
01 ≤1
Kc Ti Tm
Model: Method 6
0.214 − 0 .346
τm
− 1.256
τm
− 055 TL τm
≤ 2 .641 + 016
1977
. . 1123
.
0.685  TL  Tm  τm  Tm  TL  Tm  τm  . ;
Minimum IAE – Yu [19]         Tm Tm
K m  Tm   Tm  0. 214  Tm   Tm 
K e− sτ L τm
≤ 0.35
(Load model = L )
1 + sTL Tm
− 0099 + 0159
τm
− 1041 −4 .515
τm
+ 0. 067 τm TL
+ 0.16 ≤ ≤1;
. . . 0 .876
0874
.  TL  Tm  τm  Tm  TL  Tm  τm  2. 641
        Tm Tm
Model: Method 6 Km  Tm   Tm  0.415  Tm   Tm 
τm
≤ 0.35
Tm
1  τ τ 
−0.9077
 τm 
−0.063
 τm 
0.5961 τ m

1
Kc ( 1)
= 6.4884 + 4.6198 m + 0.8196 m  − 52132
.   − 7.2712  − 0.7241e T m

Km  Tm  Tm   Tm   Tm  
 τ  τm 
2
 τm 
3
 τm 
4
 τm 
5
 τm  
6
Ti (1) = Tm 00064
. + 3.9574 m − 64789
.   + 9.4348  − 10.7619   + 7.5146  − 2.2236  
 Tm  Tm   Tm   Tm   Tm   Tm  
Rule Kc Ti Comment
Minimum IAE – Yu [19] − 0015
τ
. + 0 .384 m − 1055
τ
− 0. 217 m − 0. 213 TL τm
1< ≤3; ≤ 0.35
. 0867
.
0871
.  TL  Tm  τm  Tm  TL  Tm  τm 
K e− sτ L  
K m  Tm 
 
 Tm 
 
0. 444  Tm 
 
 Tm 
Tm Tm
(Load model = L )
1 + sTL τm τm τm
− 1451 − 0. 003 − 0. 084
> 035
0 .218 . 0 .56
(continued) 0513
.  TL  Tm  τm  Tm  TL  Tm  τm  .
        Tm
Model: Method 6 K m  Tm   Tm  0. 670  Tm   Tm 
Tm  τ 
Minimum ISE - Hazebroek . + 0.3 m 
 074 143
. Tm τm Tm < 0.2
K mτ m  Tm 
and Van der Waerden [9]
α Tm βτ m
Model: Method 1 K mτ m
τm Tm α β τm Tm α β τm Tm α β
0.2 0.80 7.14 0.7 0.96 2.44 2.0 1.46 1.18
0.3 0.83 5.00 1.0 1.07 1.85 3.0 1.89 0.95
0.5 0.89 3.23 1.5 1.26 1.41 5.0 2.75 0.81
Minimum ISE - Murrill [13]  Tm 
0.959
Tm  τ m 
0.739
τ
– pages 358-363
1305
.
    . ≤ m ≤ 10
01 .
K m  τm  Tm
.  Tm 
 Tm 
0.945
Tm  τ m 
0.586
τm
Minimum ISE – Zhuang and
1279
.
    . ≤
01 ≤ 10
.
K m  τm  Tm
Atherton [20] 0535
.  Tm 
 Tm 
0 .675
Tm  τ m 
0. 438
τm
Model: Method 6
1346
.
    . ≤
11 ≤ 2.0
Km  τm  Tm
0552
.  Tm 
0 .181− 0 .205
τm
−1. 214
τm
− 0 .49 TL τm
≤ 2.310 + 0. 077 ;
0. 954 0 .639
0. 921  TL  Tm  τm  Tm  TL  Tm  τm 
Minimum ISE – Yu [19]         Tm Tm
K m  Tm   Tm  0. 430  Tm   Tm 
K e− sτ L τm
≤ 0.35
(Load model = L )
1 + sTL Tm
− 0. 045 + 0. 344
τm
−1. 014 − 2. 532
τm
− 0 .292 τm TL
+ 0.077 ≤ ≤1;
0. 899
1157
.  TL  Tm  τm  Tm  TL  Tm  τm  2 .310
        Tm Tm
Model: Method 6 Km  Tm   Tm  0. 359  Tm   Tm 
τm
≤ 0.35
Tm
−0 .065+ 0 .234
τm
− 1047 − 1112
τm
− 0. 094 TL τm
1< ≤3; ≤ 0.35
. . 0 .898
.  TL 
107 Tm  τm  Tm  TL  Tm  τm 
        Tm Tm
K m  Tm   Tm  0.347  Tm   Tm 
0. 04 + 0. 067
τm
− 0889
τm
− 0. 44 τm
> 035
. 0 .372 0. 46
1289
.  TL  Tm  τm  Tm  TL  Tm  τm  .
        Tm
K m  Tm   Tm  0596
.  Tm   Tm 
Minimum ITAE - Murrill  Tm 
0. 977
Tm  τ m 
0.680
τm
[13] – pages 358-363
0859
.
    . ≤
01 ≤ 10
.
Km  τm  Tm
.  Tm 
0. 272 − 0. 254
τm
− 1341
τm
− 0. 112 TL τm
≤ 2.385 + 0.112 ;
. 0 .304 0. 196
0598
.  TL  Tm  τm  Tm  TL  Tm  τm 
Minimum ITAE – Yu [19]         Tm Tm
K m  Tm   Tm  0805
.  Tm   Tm 
K e− sτ L τm
≤ 0.35
(Load model = L )
1 + sTL Tm
−0 .011− 1945
τm
−1. 055 −5. 809
τm
+ 0 .241 τm TL
+ 0112 ≤ ≤1;
. 0 .901
0. 735  TL  Tm  τm  Tm  TL  Tm  τm  2 .385 .
        Tm Tm
Model: Method 6 K m  Tm   Tm  0. 425  Tm   Tm 
τm
≤ 0.35
Tm
0 .084 + 0154
τm
− 1. 042 − 0148
τm
− 0. 365 TL τm
1< ≤3; ≤ 0.35
. . 0. 901
0.787  TL  Tm  τm  Tm  TL  Tm  τm 
        Tm Tm
Km  Tm   Tm  0. 431  Tm   Tm 
Rule Kc Ti Comment
Minimum ITAE – Yu [19] − 0. 057

τm
−0 .909
τm
− 0257 τm
> 035
0172
. 0 .228 . 0. 489
0878
.  TL  Tm  τm  Tm  TL  Tm  τm  .
K e− sτ L  
K m  Tm 
 
 Tm 
 
0. 794  Tm 
 
 Tm 
Tm
(Load model = L )
1 + sTL
(continued)
Model: Method 6
 Tm 
0.957
Tm  τ m 
0.552
τm
Minimum ISTSE - Zhuang
1015
.
    . ≤
01 ≤ 10
.
K m  τm  Tm
and Atherton [20] 0667
.  Tm 
Model: Method 6  Tm 
0.673
Tm  τ m 
0.427
τm
1065
.
    . ≤
11 ≤ 2.0
K m  τm  Tm
0687
.  Tm 
 Tm 
0.953
Tm  τ m 
0.546
τm
Minimum ISTES – Zhuang
1021
.
    . ≤
01 ≤ 10
.
Km  τ m  Tm
and Atherton [20] 0629
.  Tm 
 Tm 
0 .648
Tm  τ m 
0. 442
τm
Model: Method 6
1076
.
    . ≤
11 ≤ 2.0
Km  τm  Tm
0650
.  Tm 
Servo tuning Performance index minimisation
Minimum IAE – Rovira et al.  Tm 
0.861
Tm τm
0758
.
  . ≤
01 ≤ 10
.
[21] τ Tm
Model: Method 3 K m  τm  1.020 − 0.323 m
Tm
Minimum IAE - Huang et al. τm
[18] . ≤
01 ≤1
Kc( 2) 2 Ti (2 ) Tm
Model: Method 6
 Tm 
0.892
Tm τm
0980
.
  . ≤
01 ≤ 10
.
Minimum ISE - Zhuang and τ Tm
Atherton [20] Km  τm  0.690 − 0.155 m
Tm
 Tm 
0.560
Tm τm
.
  . ≤
11 ≤ 2.0
τ Tm
K m  τm  0.648 − 0.114 m
Tm
 0.7388 03185
.  Tm Kc Km τm
Minimum ISE – Khan and  +  0.01 ≤ ≤ 0.2
 τm Tm  Km  0.5291 00003082
.  Tm
Lehman [22]  − 
 Tm τ m τm 
2
Model: Method 6  0.808 0.511 0.255  T Kc Km τm

  m 0.2 ≤ ≤ 20
 τm + −  
Tm τ m  K m  0.095 + 0.846 − 0.381
Tm
 Tm 
 τ 2 τ m Tm τ m Tm τ m 
 m 
Minimum ITAE – Rovira et  Tm 
0 .916
Tm τm
0586
.
  . ≤
01 ≤ 10
.
al. [21] τm Tm
Model: Method 3 Km  τ m  1.030 − 0165
.
Tm
1  
−1.0169 3.5959 3.6843 τ
τm  τm   τm   τm 
m
2
Kc = ( 2)
− 130454
. − 9.0916 + 0.3053  + 11075
.   − 2.2927  + 4.8259e T m
Km  Tm  Tm   Tm   Tm  

 τm  τm 
2
 τm 
3
 τm 
4
 τm 
5
 τm  
6
(2 )

Ti = Tm 0.9771 − 0.2492 + 3.4651  − 7.4538  + 8.2567   − 4.7536  + 11496 .   
 Tm  Tm   Tm   Tm   Tm   Tm  
 
Rule Kc Ti Comment
 Tm 
0.921
Tm τm
0712
.
  . ≤
01 ≤ 10
.
Minimum ISTSE - Zhuang τ Tm
and Atherton [20] Km  τm  0.968 − 0.247 m
Tm
 Tm 
0.559
Tm τm
.
  . ≤
11 ≤ 2.0
τ Tm
Km  τm  0.883 − 0.158 m
Tm
 Tm 
0.951
Tm τm
0569
.
  . ≤
01 ≤ 10
.
Minimum ISTES – Zhuang τ Tm
and Atherton [20] K m  τm  1.023 − 0.179 m
Tm
Model: Method 6
 Tm 
0.583
Tm τm
0628
.
  . ≤
11 ≤ 2.0
τ Tm
Km  τ m  1.007 − 0.167 m
Tm
Direct synthesis
Haalman [23] 2Tm Closed loop sensitivity
Model: Method 6 3K m τ m Tm M s = 19
. . (Astrom and
Hagglund [3])
Chen and Yang [23a] 0. 7T m Ms = 1. 26 ; A m = 2. 24 ;
Model: Method 25 Km τ m Tm
φ m = 500
Minimum IAE – regulator -
Pemberton [24], Smith and Tm Tm τm
Corripio [25] – page 343- Km τ m . ≤
01 ≤ 0.5
Tm
346. Model: Method 6
Minimum IAE – servo - 3Tm τm
. ≤
01 ≤ 05
.
Smith and Corripio [25] – 5Km τ m Tm Tm
page 343-346. Model:
Method 6
5% overshoot – servo – τm
Smith et al. [26] – deduced Tm 0.04 ≤ ≤ 14
.
052
. Tm Tm
from graph. K mτ m
Model: Method not stated
1% overshoot – servo – τm
Smith et al. [26] – deduced Tm 0.04 ≤ ≤ 14
.
044T
. m Tm
from graph K mτ m
Model: Method not stated
5% overshoot - servo - Tm
Smith and Corripio [25] – 2K m τ m Tm
page 343-346. Model:
Method 6
5% overshoot - servo - 13Tm
Hang et al. [27] 25K m τ m Tm
Model: Method 1
Closed loop response
Miluse et al. [27a] 0. 368Tm overshoot = 0%
K m τm Tm (Model: Method 26 – Miluse
Model: Method not stated et al. [27b])
0. 514Tm Closed loop response
K m τm Tm overshoot = 5%

Rule Kc Ti Comment
Miluse et al. [27a] - 0. 641Tm Closed loop response
continued K m τm Tm overshoot = 15%

Model: Method not stated 0. 748Tm Closed loop response



0. 957 Tm Closed loop response

K mτ m Tm overshoot = 45%
1.008 Tm Closed loop response

Km τ m Tm overshoot = 50%
Pole is real and has max.
Regulator – Gorecki et al. Kc ( 3)
Ti ( 3) attainable multiplicity
[28]
Low freq. part of magnitude
(considered as 2 rules)
Kc( 4) 3 Ti ( 4) Bode diagram is flat
Model: Method 6
Chiu et al. [29] λ Tm λ variable; suggested
Model: Method 6 K m (1 + λτm ) Tm values: 0.2, 0.4, 0.6, 1.0.
Astrom et al. [30] 3
 3τ m  Tm Honeywell UDC 6000
Model: Method 22 K m1 +  controller
 Tm 
Davydov et al. [31] 1  τm  Closed loop response
 0153
. + 0.362  Tm
 τ   Tm 
damping factor =
Model: Method 12 K m  1.905 m + 0.826 . ≤ τ m Tm ≤ 1 .
0.9; 02
 Tm 
 
2
2  τm  τm
2 Tm   τ  2 +  −2 −
2 +  m  − 1 e
 2 Tm 
=
3 ( 3) 2Tm
Kc
K m τm   2Tm  
 
2
 τ 
1+  m 
 2Tm 
= τm
( 3)
Ti
 τ  τ 
2
 τ 
3
  τ  
2
 τ 
2
3 +  m  +  m  +  m  − 2 +  m   2 +  m 
 2Tm   Tm   2Tm    2Tm    2Tm 
2 3 2 3
Tm T  T  Tm T  T 
1+3 + 6 m  + 6 m  1+ 3 + 6 m  + 6 m 
1 τm  τm   τm  τm  τm   τm 
Kc ( 4) = , Ti ( 4) = τ m
Km  Tm  Tm  
2
 Tm  Tm  
2

4 1+ 3 + 3   
3 1+ 2 + 2  
 τm  τm    τm  τm  
 
Rule Kc Ti Comment
Tm Closed loop response
0.368
Schneider [32] K mτ m Tm damping factor = 1
Tm Closed loop response

Model: Method 6 0.403
K mτ m Tm damping factor = 0.6
McAnany [33] (1.44Tm + 0.72Tm τ m − 0.43τ m − 2.14) (
. + 2τ m + τ m
Km 556
2
) Closed loop time constant =
(. τ m + 0.36 τ m 2 + 2
K m 12 ) 4Tm + 128
. τ m − 2.4
167
. Tm .
Model: Method 5
Leva et al. [34]  π   Tm  π 
ω cn Tm 1 + ω cn 2 Tm 2 tanφ m − + τ m ω cn + tan −1 (ω cn Tm )  2.82 − φ m − tan −1   − φm 
 2   τ m  2  
ω cn =
Model: Method 16 Km 1 + ω cn 2 Ti 2 ω cn τm
Tm  0. 3852 0.723 τ  KcKm τm

Khan and Lehman [22]
 + − 0.404 m2   0.4104 0.00024 0.525 0.01 ≤ ≤ 0.2
Km  τm Tm Tm   − −  Tm
 τ m Tm τ m2 Tm 2 
Model: Method 6
 0.404 0.256  KcKm τm
Tm  + −
01275
. 
 0.719 0.0808  0.2 ≤ ≤ 20
Km  τm T τ  0.324 Tm
 m m Tm   + − 
 τ m Tm τ
2
τ 
 m m τ mTm 
1048
. Tm Tm Gain Margin = 1.5
Hang et al. [35, 36] Km τ m Phase Margin = 30 0
07854
. Tm Tm Gain Margin = 2
Model: Method 11
Or K mτ m Phase Margin = 45 0
. Tm Tm Gain Margin = 3
K mτ m Phase Margin = 60 0
0.393Tm Tm Gain Margin = 4
K mτ m Phase Margin = 67 .5 0
0314T
. m Tm Gain Margin = 5
Km τ m Phase Margin = 72 0
Gain and phase margin – Ho ω p Tm 1 A m φ m + 0.5πA m (A m − 1)
ωp =
et al. [37] K m Am 4ω p τ m 2
1 (A m
2
)
− 1 τm
2ω p − +
(considered as 2 rules) π Tm
τm
Model: Method 6 Given A m , ISE is minimised when φ m = 688884
. − 34.3534 A m + 91606
. for servo
Tm
tuning (Ho et al [104]).
Given A m , ISE is minimised when φ m = 45.9848A m 0.2677 ( τ m Tm )
0. 2755
for regulator
tuning (Ho et al [104]).
τm
( )
2 1
β Ti ω φ 1 + βTm ω φ β = 08 < 05
Tan et al. [39] [
βω φ tan − tan − 1 βT mω φ − βτ m ω φ − φ ] . ,
Tm
. ;
( ) ωφ < ω u
2
A m 1 + βTi ω φ τm
Model: Method 10 β = 0.5, > 05
.
Tm
Symmetrical optimum 1 4.6 τm
≤ 01
.
principle - Voda and Landau . G p ( jω135 )
35 0
ω135 0 Tm
[40]
1 4 τm
. <
01 ≤ 015
.
Model: Method not relevant 2.828 G p ( jω135 ) 0
ω135 0 Tm
1 . G p ( jω 135 0 ) + 0.75K m
115
τm
4.6 G p ( jω135 ) − 0.6K m
0
[
ω 135 0 2.3 G p ( jω 1350 ) − 0.3K m ] 0.15 <
Tm
≤1
Rule Kc Ti Comment
Friman and Waller [41] 0.2333 1 τ m > 2Tm . Gain margin = 3;
Model: Method 6 G p ( jω135 )0
ω135 0 Phase margin = 45 0
Voda and Landau [40] Phase margin = 60 0 ;
Tm Tm τ
0.25 ≤ m ≤ 1
Model: Method 6 2K mτ m Tm
Smith [42] 0.35 6 dB gain margin - dominant
Model: Method not Km . τm
042 delay process
specified
Kc( 5) Ti (5) τm
Modulus optimum principle ≤1
Tm
- Cox et al. [43]4
0.5Tm τm
>1
Model: Method 17 Km τ m Tm Tm
0.019952τ m + 0.20042Tm 0.099508τ m + 0.99956Tm Closed loop time constant =
τm
K mτm 0.99747τ m − 8742510
. . − 5 Tm 4τm
Cluett and Wang [44] 0.05548 τ m + 0.33639Tm 016440
. τ m + 0.99558Tm Closed loop time constant =
τm
K mτm 0.98607τ m − 15032
. .10− 4 Tm 2τm
Model: Method 6
0.092654τ m + 0.43620Tm 0.20926τ m + 098518
. Tm Closed loop time constant =
τm
K mτm . −3 Tm
0.96515τ m + 4.255010 . τm
133
012786
. τ m + 051235
. Tm 0.24145τ m + 0.96751Tm Closed loop time constant =
τ
Km τ m 0.93566 τ m + 2.298810
. −2 Tm m τm
016051
. τ m + 0.57109Tm 0.26502τ m + 0.94291Tm Closed loop time constant =
τ
Km τ m 089868
. τ m + 6.935510
. −2 Tm m . τm
08
019067
. τ m + 0.61593Tm 0.28242τ m + 0.91231Tm Closed loop time constant =
τ
Km τ m 0.85491τ m + 015937
. Tm m . τm
067
−1.045 V = fractional overshoot,
τ 
Abbas [45] 0148
. + 0186
.  m Tm + 0.5τ m 0 ≤ V ≤ 0.2
 Tm  τ
Model: Method 6 Km (0.497 − 0.464 V0.590 ) . ≤ m ≤ 5.0
01
Tm
Bi et al. [46] 05064
. Tm
Model: Method 20 Km τ m Tm
0.5  Tm 3 + Tm2 τ m + 0.5Tm τ m 2 + 0.167τ m 3   Tm3 + Tm 2τ m + 0.5Tmτ m 2 + 0.167τ m 3 

4
Kc
( 5)
=   , Ti
( 5)
=  
 Tm τm + Tm τ m + 0.667τ m   Tm + Tm τ m + 0.5τ m 
2 2 3 2 2
Km
Rule Kc Ti Comment
λ = inverse of the maximum
Wang and Shao [47] ( 5a ) 5 (5 a )
Kc Ti of the absolute real part of
the open loop transfer
Model: Method 6 function; λ = [1.5, 2.5]
Robust
Tm + 0.5τ m
Brambilla et al. [48] - K m( λ + τ m ) Tm + 0.5τ m
Model: Method 6 Closed loop response has less than 5% overshoot with no model uncertainty:
τ τ τ
. ≤ m ≤ 1 ; λ = 1 − 05
λ = 1 , 01 . log 10 m , 1 < m ≤ 10
Tm Tm Tm
Tm
Rivera et al. [49] λKm Tm λ ≥ 17
. τ m , λ > 01
. Tm .
Model: Method 6 2Tm + τ m

2λK m Tm + 0.5τ m λ ≥ 17
. τ m , λ > 01
. Tm .
Chien [50] Tm λ = Tm [50];
K m (τ m + λ) Tm λ > Tm + τ m , τ m << Tm
Model: Method 6
(Thomasson [51])
Thomasson [51] τm τ m >> Tm ; λ = desired
2Km ( τ m + λ ) . τm
05 closed loop time constant
Model: Method not defined
5Tm τm
< 0.33
Fruehauf et al. [52] 9τ m Km 5τ m Tm
Tm τm
Model: Method 1
Tm ≥ 0.33
2τ m Km Tm
0. 50Tm A m = 3.14 , φ m = 61. 40 ,
Chen et al. [53] τm K m Tm
Ms = 1. 00
Model: Method 6 0. 61Tm A m = 2.58 , φ m = 55. 00 ,
τ m Km Tm
Ms = 1. 10
0. 67Tm A m = 2. 34 , φ m = 51.6 0 ,
τ mK m Tm
Ms = 1. 20
0. 70Tm A m = 2. 24 , φ m = 50.0 0 ,
τm K m Tm
M s = 1. 26
0. 72Tm A m = 2.18 , φ m = 48. 70 ,
τm K m Tm
Ms = 1. 30
1  1 
= f 2 ( ω900 ) − ,
5 ( 5a )
Kc
λf 1 (ω90 0 )  ω90 0 
f 1 (ω90 )=
Km
[
( Tm + {1 + ω90 Tm }τ m ) sin(−ω90 τm − tan −1 ω90 Tm ) − ω90 Tm cos(− ω90 τ m − tan− 1 ω90 Tm )
2 2 2
]
( )
0 0 0 0 0 0 0
2 1 .5
1 + ω90 Tm
2
0
f 2 ( ω90 ) = −0
1
1 + ω 90 Tm
2 2
[(T m + {1 + ω 90 Tm }τ m ) cot(−ω90 τm − tan −1 ω90 Tm ) + ω90 Tm
0
2 2
0 0 0
2
]
0
Ti
( 5a )
=
[ ]
ω900 Tm + (1 + ω90 0 2Tm 2 ) τ m cos(− ω90 0 τm − tan− 1 ω90 0 Tm ) + (1 + 2ω 900 2 Tm 2 ) sin(− ω90 0 τm − tan −1 ω900 Tm )
− ω900 Tm cos(−ω90 0 τm − tan −1 ω90 0 Tm ) + ω90 0 [Tm + (1 + ω900 Tm ) τm ] sin(− ω900 τ m − tan− 1 ω90 0 Tm )
3 2 2 2 2
Rule Kc Ti Comment
Chen et al. [53] 0. 76Tm A m = 2. 07 , φ m = 46. 50 ,
- continued τm K m Tm
Ms = 1. 40
Model: Method 6 0. 80Tm A m = 1.96 , φ m = 44.10 ,
τm K m Tm
Ms = 1. 50
α βTm
Ogawa [54] – deduced from Km
graphs
τm Tm α β τm Tm α β
Model: Method 6 0.5 0.9 1.3 2.0 0.45 2.0 20% uncertainty in the
1.0 0.6 1.6 10.0 0.4 7.0 process parameters
0.5 0.7 1.3 2.0 0.4 2.2 33% uncertainty in the
Desired closed loop
Lee et al. [55] Ti τm
2
e− τ m s
Tm +
K m ( λ + τm )
response = ,
Model: Method 6
2( λ + τ m ) ( λ s + 1)
λ = 0.333 τm
Isaksson and Graebe [56] Tm + 0.25τ m Tm + 025
. τm Tm > τ m
Model: Method 6 Km λ
Chun et al. [57] Tm ( τ m + 2λ ) − λ2 Tm ( τ m + 2λ ) − λ2
λ = 0.4Tm
Km (τ m + λ ) τ m + Tm
2
Model: Method 21
Ultimate cycle
 
    T  
McMillan [58] 0.65 Tuning rules developed
1881
. Tm  1 
. τ m 1 + 
 0.65 
166 m
  from Ku , Tu
Model: Method 1 or Km τ m   T     Tm + τ m  
1 +  T + τ 
m

Method 18   m m 
Regulator – minimum IAE – Ku  T T  
2
Shinskey [59] – page 167. Tu  0.87 − 0.855 u + 0.172  u  

T  τm  τ m  
Model: Method not 3.05 − 0.35 u 
τm
specified
Regulator – minimum IAE – 055
. Ku 0.78Tu τ m Tm = 0.2
Shinskey [17] – page 121. 0. 48Ku 0.47 Tu τ m Tm = 1
Model: Method not
specified
0.5848Ku 0.81Tu τ m Tm = 0.2
Regulator – minimum IAE – 0.5405K u 0.66Tu τ m Tm = 0.5
Shinskey [16] – page 148
Model: Method 6 0.4762K u 0.47Tu τ m Tm = 1
0.4608 K u 0.37Tu τ m Tm = 2
Regulator – nearly minimum ( 1 − µ1 ) K u , Kc µ 2 Ku ω u , Decay ratio = 0.15 -
IAE, ISE, ITAE – Hwang
( ) (
0.0694 −1 + 2.1ω u τ m − 0.367ω 2u τ m2 ) τ
[60] µ1 =
. 1 − 0.482ω u τ m + 0.068ω 2u τ 2m
114
µ2 = . ≤ m ≤ 2.0
01
Ku K m 1 + K u K m Ku Km 1 + Ku Km Tm
Model: Method 8
Rule Kc Ti Comment
( 1 − µ1 ) K u , Kc µ 2 Ku ω u , Decay ratio = 0.2 -
Regulator – nearly minimum
( ) ( −1 + 2.54ω u τm − 0.457ωu2 τm2 ) τ
IAE, ISE, ITAE – Hwang µ1 =
1.09 1 − 0.497ω u τ m + 0.0724ω 2u τ 2m
µ2 =
0.054
. ≤ m ≤ 2.0
01
K u K m 1 + K u Km K u Km 1 + K u Km Tm
[60]
(continued) ( 1 − µ1 ) K u , Kc µ 2 Ku ω u , Decay ratio = 0.25 -
( ) τ
Model: Method 8
µ1 =
(
1.03 1 − 0.51ωu τm + 0.0759ω u2τ m2 ) µ2 =
0.0386 −1 + 3.26ω uτ m − 0.6ω 2u τ m2
. ≤ m ≤ 2.0
01
Ku Km 1 + Ku Km K u Km 1 + K u Km Tm
( 1 − µ1 ) K u , Kc µ 2 Ku ω u , Decay ratio = 0.1, r = 0.5 -
Servo – small IAE – Hwang ( ) τ
[60] µ1 =
(
1.07 1 − 0.466ω uτ m + 0.0667ω u2τ m2 ) µ2 =
0.0328 −1 + 2.21ω u τm − 0.338ω 2u τ m2
. ≤ m ≤ 2.0
01
Ku Km 1 + Ku Km K u Km 1 + K u Km Tm
Model: Method 8 ( 1 − µ1 ) K u , Kc µ 2 Ku ω u , Decay ratio = 0.1, r = 0.75 -

( ) (
0.0477 − 1 + 2.07ω u τ m − 0.333ω 2u τ 2m ) τ
µ1 =
. 1 − 0.467ω u τ m +
111 0.0657ω 2u τ 2m
µ2 = 01. ≤ m ≤ 2.0
Ku Km 1 + K u K m Ku K m 1 + K u K m Tm
r = parameter related to the
position of the dominant real ( 1 − µ1 ) K u , Kc µ 2 Ku ω u , Decay ratio = 0.1, r = 1.0 -
pole.
( ) τ
µ1 =
(
. 1 − 0.466ωu τ m +
114 0.0647ω 2u τ m2 ) µ2 =
0.0609 −1 + 197
. ω uτ m − 0.323ωu2 τm2
. ≤ m ≤ 2.0
01
Ku Km 1 + Ku Km Ku Km 1 + Ku Km Tm
  τm 
  11 T + 13   
 12 + 2   τm  
Hang et al. [65] m
 11 T + 13   τ
 τ
 37 m − 4   4 . ≤ m < 096
 + 1Tu
016 . ;
    0.2  m
 37 τm − 4   Tm
Model: Method not 5  Tm  
15
6  τm 
Ku  T   Servo response: 10%
specified    
  11 T + 13 
m
overshoot, 3% undershoot
 15 + 14 m

  τm 
  37 − 4
  Tm 
Servo – minimum ISTSE – τm

Zhuang and Atherton [20] 0.361K u 0.083 (1.935 K m K u + 1)Tu . ≤
01 ≤ 2.0
Tm
Model: Method not relevant
Regulator – minimum ISTSE  1892 . K m K u + 0244 .   0.706Km K u − 0.227  τm
– Zhuang and Atherton [20]   Ku   Ku . ≤
01 ≤ 2.0
 3249
. K K
m u + 2 .097   0.7229Km K u + 12736 .  Tm
Model: Method not relevant
Servo – nearly minimum IAE  τm  τ m  
2 ( Kc / K uω u ) τm
and ITAE – Hwang and  0.438 − 0110 . + 0.0376  Ku 
τm  τ m  
2 . ≤
01 ≤ 2.0 ;
Tm  m  
T  0.0388 + 0108 − 0.0154  Tm

.  
Fang [61]  Tm  T m  
decay ratio = 0.03
Model: Method 9
Regulator – nearly minimum  τm  τ m  
2 ( Kc / K uω u ) τm
IAE and ITAE – Hwang and  0.515 − 0.0521 Tm + 0.0254  Tm   Ku  τm  τm  
2 . ≤
01 ≤ 2.0 ;
0.0877 + 0.0918 − 0.0141   Tm
Fang [61]  Tm  Tm  

decay ratio = 0.12
Model: Method 9
Simultaneous    
2
 0.46 − 0.0835 τ m + 0.0305 τ m   K

( Kc / K u ω u ) τm
     
2 . ≤
01 ≤ 2.0
Servo/regulator – Hwang   0.0644 + 0 .0759 τ m − 0.0111  τ m  
u
Tm  Tm   Tm
and Fang [61]  Tm  Tm  

Model: Method 9
K me − sτ
m
 1
Table 2: PI tuning rules - FOLPD model - G m ( s) = – Controller G c ( s) = Kc  b +  . 2 tuning rules
1 + sTm  Ti s
Rule Kc Ti Comment
Direct synthesis (Maximum sensitivity)
− 2.7 τ + 3.7τ 2
8.9τ m e −6.6 τ + 3.0 τ or b = 0.81e0.73 τ + 1.9 τ
2 2
029
. e Tm
Astrom and Hagglund [3] - ,
K mτ m τ
0.79Tm e −1.4 τ + 2.4 τ
2
dominant pole design – . ≤ m ≤ 55

Ms = 1.4 , 014 .
page 205-208 τ = τ m ( τ m + Tm ) Tm
. e− 4.1τ + 5.7 τ Tm 8.9τ m e −6.6 τ + 3.0 τ or b = 0.44e0.78 τ − 0.45τ ;

2 2 2
078
Model: Method 3 or 4
Km τ m τ
0.79Tm e −1.4 τ + 2.4 τ
2
. ≤ m ≤ 55
Ms = 2.0, 014 .
Tm
Astrom and Hagglund [3] -
modified Ziegler-Nichols – 04
. Tm 0.7Tm τm
. ≤
b = 0.5; 01 ≤2
page 208 K mτ m Tm
Model: Method 3 or 4
K me − sτ m
Table 3: PI tuning rules - FOLPD model - G m ( s) = – Two degree of freedom controller:

1 + sTm
 1 
U(s) = K c 1 + E (s ) − α Kc R( s) . 1 tuning rule
 Tis 
Rule Kc Ti Comment
Minimum servo/regulator Performance index minimisation




τm
1 
≤ 1.0 .
Taguchi and Araki [61a] 0.7382  Ti
( 5b ) 6
Tm
0. 1098+
K m  τm 
 − 0.002434 Overshoot (servo step)
Model: ideal process  Tm 
≤ 20% ; settling time
2
τ τ  ≤ settling time of tuning
α = 0. 6830 − 0.4242 m + 0.06568  m  rules of Chien et al. [10]
Tm  Tm 
 τ τ 
2
τ  
3
= Tm  0.06216 + 3.171 m − 3.058  m  + 1. 205  m  

6 ( 5b )
Ti
 

Tm  Tm   Tm  
 1
Table 4: PI tuning rules - non-model specific – controller G c ( s) = Kc  1 +  . 20 tuning rules
 i 
Ts
Rule Kc Ti Comment
Ultimate cycle
Ziegler and Nichols [8] 0.45Ku 0.83Tu Quarter decay ratio
p1 , T1 = decay rate, period
0.45Ku 1  5.22  measured under
Hwang and Chang [62] . −
 522  proportional control when
p1  T1 
Kc = 05
. Ku
** Hang et al. [36] 0.25Ku 0.2546Tu dominant time delay process
McMillan [14] – page 90 0.3571K u Tu
Pessen [63] 0.25Ku 0.042 K uTu dominant time delay process
0.4698 K u 0.4373Tu Gain margin = 2, phase
margin = 20 0
. Ku 0.0882Tu Gain margin = 2.44, phase
page 142 margin = 61 0
0.2015 Ku 01537
. Tu Gain margin = 3.45, phase
margin = 46 0
Parr [64] – page 191 05
. Ku 043
. Tu Quarter decay ratio
Yu [122] – page 11 0.33Ku 2Tu
Other tuning rules
Parr [64] – page 191 0667
. K25% T25% Quarter decay ratio
McMillan [14] – pages 42- 042
. K25% T25% ‘Fast’ tuning
43 0.33K25% T25% ‘Slow’ tuning
0333
.
Parr [64] – page 192 G p ( jω u ) 2Tu Bang-bang oscillation test
0.5 4 Alfa-Laval Automation

Hagglund and Astrom [66] G p ( jω 135 )
0
ω135 0
ECA400 controller
0.25 1.6 Alfa-Laval Automation

G p ( jω 135 )
0
ω135 0
ECA400 controller - process
has a long delay
Rule Kc Ti Comment
(
tan φ m − φ p ω − 0 .5 π ) (
tan φ m − φ pω − 05
. π ) φ m > φ pω + 0.5π ,
Leva [67]
(
G p ( jω ) 1 + tan φ m − φ p ω − 05
2
.π ) ω φ pω = process phase at
frequency ω
sin φ m tanφ m
Astrom [68] G p ( jω 90 ) 0
ω 90 0
1 1 φ m = 450 ,‘small’ τ m
Calcev and Gorez [69] 2 2 G p ( jω u ) ωu φ m = 15 0 , ‘large’ τ m
Cox et al. [70] . VTu sin φ m
020 . Tu tan φ m
016 V = relay amplitude, A = limit
A cycle amplitude.
Direct synthesis
05
. A3 1 A3 A m ≥ 2 , φ m ≥ 60 0
Vrancic et al. [71] A1A2 − Km A 3 A2
05
. A3 Modified Ziegler-Nichols
Vrancic [72] A1A2 − Km A 3 . τm
333 process reaction method
0.4830 3.7321 Gain margin = 2,

Friman and Waller [41] G p ( jω150 ) 0
ω150 0
Phase margin = 30 0
1.18 K u K m − 1. 72 1.18K u K m − 1.72 0.1 ≤ Ku Km ≤ 0.5

Kristiansson and K uK m
2
(0.33K u K m − 0.17) ωu
Lennartson [158a]
0. 50K u K m − 0. 36 0. 50K u K m − 0. 36 K u K m > 0.5
Ku Km
2
(0.33K u K m − 0.17) ωu
20K135 K m − 160
0 20K135 Km − 160 0 K u K m < 0. 1 ;
K135 K m 0
2
(0.315 K135 Km − 0.175 )ω u
0
K 135 0 K m ≤ 0 .1
5.4K135 K m − 13. 6
0 5. 4K135 K m − 13. 6
0 K u K m < 0. 1 ;
K135 K m 0
2
(1.32K135 K m − 3.2)ω u
0
K 135 K m > 0 .1
0
1
A1 = y1( ∞) , A 2 = y 2 ( ∞) , A 3 = y3 ( ∞)
1 + b1 s + b2 s2 + b 3s3 − s τ
Alternatively, if the process model is G m (s) = K m e , then m
1 + a1s + a 2 s3 + a 3s3
(
A1 = Km ( a1 − b1 + τ m ) , A2 = K m b 2 − a2 + A1a1 − b1 τm + 05
. τ m2 , )
(
A3 = Km a 3 − b3 + A 2 a1 − A1a 2 + b2 τ m − 05
. b1τ m 2 + 0167
. τ m3 )
 1 
Table 5: PI tuning rules - non-model specific – controller G c ( s) = K c  b +  . 1 tuning rule
 Ti s 
Rule Kc Ti Comment
2.9 κ − 2.6κ 2
. e −0.0061κ + 1.8 κ ;
b = 11
2
0.053 Ku e ,
Astrom and Hagglund [3] - 0.90Tu e− 4.4κ + 2. 7κ 0 < Km K u < ∞ .
2
κ = 1 Km Ku
Ms = 1.4 – page 215
maximum sensitivity Ms
=1.4
. e 0.40κ − 0.17 κ ;
b = 048
2
1.9κ − 1.3κ 2 − 4.4κ + 2. 7κ 2

013
. K ue 0.90Tu e 0 < Km K u < ∞ .
Ms = 2.0
 1 
Table 6: PI tuning rules - non-model specific – controller U( s) = Kc  1 +  E( s) + Kc ( b − 1)R (s) . 1 tuning rule
 Ti s 
Rule Kc Ti Comment
Direct synthesis
A1 b = [0.5,08
. ] - good servo
Vrancic [72] 2
Kc ( 6)
Km +
1
2 K c (6 )
+
K c( 6) K m 2
2
(1 − b )
2
and regulator response
A 1A 2 − K m A 3 − ( K mA 3 − A 1A 2 ) 2 − (1 − b2 )A 3 ( Km 2 A3 + A13 − 2K mA 1A2 )
2 ( 6)
= , Km A 3 − A1A 2 < 0
(1 − b )( K )
Kc
A 3 + A 1 − 2K mA 1A 2
2 2 3
m
A 1A 2 − K m A 3 + ( K m A3 − A 1A 2 ) 2 − (1 − b2 )A 3 ( Km 2 A3 + A13 − 2K m A1A2 )
( 6)
= , Km A 3 − A1A 2 > 0
(1 − b )( K )
Kc
A 3 + A 1 − 2K mA 1A 2
2 2 3
m
t t t
 y( τ ) 
y1( t) =  Km −
 ∫
0
∆u 
 dτ , y2 ( t) = ∫ (A 1 − y1 (τ))dτ ,
0
y3 ( t) = ∫(A
0
2 − y2 ( τ) )dτ
Kc
Table 7: PI tuning rules - non-model specific – controller U( s) = Kc Y(s) − E ( s) . 1 tuning rule
Ti s
Rule Kc Ti Comment
Direct synthesis
− TCL + 1414 TC LTm + τ mTm − T CL + 1414
. T C L Tm + Tm τ m
2 2
. Underdamped system
Chien et al. [74]
(
K m TC L2 + 1414
. TCL τ m + τ m 2
) Tm + τ m
response - ξ = 0.707 .
τm > 0.2Tm
K me− sτ m  
Table 8: PI tuning rules - IPD model G m ( s) = - controller G c ( s) = Kc  1 + 1  . 21 tuning rules
s  i 
Ts
Rule Kc Ti Comment
Process reaction
Ziegler and Nichols [8] 0.9 Quarter decay ratio
Model: Method 1. Km τ m . τm
333
0.6 Decay ratio = 0.4; minimum
Two constraints method – Km τ m 2.78τ m error integral (regulator
Wolfe [12] mode).
Decay ratio is as small as
Model: Method 1 087
. 4.35τm possible; minimum error
Km τ m integral (regulator mode).
Tyreus and Luyben [75] Maximum closed loop log
0.487 8.75τ m modulus = 2dB ; closed loop
Model: Method 2 or 3 Km τ m time constant = τ m 10
Astrom and Hagglund [3] – 0.63 Ultimate cycle Ziegler-
page 138 Km τ m 3.2τm Nichols equivalent
Model: Not relevant
Regulator tuning Performance index minimisation
Minimum ISE – Hazebroek 15.
and Van der Waerden [9] K mτ m . τm
556
Model: Method 1
Shinskey [59] – minimum
IAE regulator – page 74. 09259
. 4τ m
Model: Method not Km τ m
specified
Poulin and Pomerleau [82] –
minimum ITAE (process 05264
. 4.5804 τ m
output step load K mτ m
disturbance)
Model: Method 2
Poulin and Pomerleau [82] – 05327
.
minimum ITAE (process K mτ m 38853
. τm
input step load disturbance)
Model: Method 2
Ultimate cycle
Tyreus and Luyben [75] Maximum closed loop log
0.31Ku 2.2Tu modulus = 2dB ; closed loop
Model: Method 2 or 3. time constant = τ m 10
Regulator – minimum IAE –
Shinskey [17] – page 121. 061
. Ku Tu
Model: method not
specified
Robust
Fruehauf et al. [52] 05
.
Model: Method 5 Km τ m 5τ m
Rule Kc Ti Comment
 1 
λ= , τm  [50];
1  2λ + τ m 
Chien [50]
2λ + τm  Km 
Model: Method 2 Km  [λ + τ m ]2  λ > τ m + Tm (Thomasson
[51])
λ Overshoot TS PP RT Zhang et al. [135]

. τm
15 58% 6τ m 17
. Km 7Km λ = [15
. τ m ,4.5τ m ]
τm τm - values deduced from
2.5τ m 35% 11τ m 2.0K m 16K m graphs
τm τm
. τm
35 26% 16τ m 2.2K m 23Km
τm τm
. τm
45 22% 20τ m 25
. Km 30K m
τm τm
0.45 20% uncertainty in the
Ogawa [54] – deduced from K mτ m 11τ m process parameters
graph
Kmτm 12τ m process parameters
Model: Method 5
K mτ m 13τ m process parameters
Km τ m 14τ m process parameters
K mτ m 15τ m process parameters
Rule Kc Ti Comment
Direct synthesis
α
K mτ m βτ m
Wang and Cluett [76] –
Closed Damp. Gain Phase Closed Damp. Gain Phase
deduced from graph
loop Factor margin margin α β loop Factor margin margin α β
time ξ Am φm time ξ Am φm
Model: Method 2
const. [deg.] const. [deg.]
τm 0.707 1.3 11 0.9056 2.6096 τ m 1.0 1.3 14 0.8859 3.212
2τm 0.707 2.5 33 0.5501 4.0116 2τm 1.0 2.3 37 0.6109 5.2005
3τ m 0.707 3.6 42 0.3950 5.4136 3τ m 1.0 3.0 46 0.4662 7.1890
4τm 0.707 4.7 47 0.3081 6.8156 4τm 1.0 4.0 52 0.3770 9.1775
5τ m 0.707 5.9 50 0.2526 8.2176 5τ m 1.0 4.8 56 0.3164 11.166
6τ m 0.707 7.1 52 0.2140 9.6196 6τ m 1.0 5.6 59 0.2726 13.155
7τm 0.707 8.2 54 0.1856 11.022 7τm 1.0 6.3 61 0.2394 15.143
8τ m 0.707 9.2 55 0.1639 12.424 8τ m 1.0 7.2 62 0.2135 17.132
9τm 0.707 10.4 56 0.1467 13.826 9τm 1.0 8.0 64 0.1926 19.120
10τ m 0.707 11.5 57 0.1328 15.228 10τ m 1.0 8.7 65 0.1754 21.109
11τm 0.707 12.7 58 0.1213 16.630 11τm 1.0 9.6 66 0.1611 23.097
12τ m 0.707 13.8 59 0.1117 18.032 12τ m 1.0 10.4 67 0.1489 25.086
13τ m 0.707 14.9 59 0.1034 19.434 13τ m 1.0 11.2 67 0.1384 27.074
14τ m 0.707 16.0 60 0.0963 20.836 14τ m 1.0 12.0 68 0.1293 29.063
15τ m 0.707 17.0 60 0.0901 22.238 15τ m 1.0 12.7 68 0.1213 31.051
16τ m 0.707 18.2 60 0.0847 23.640 16τ m 1.0 13.6 69 0.1143 33.040
0.9588 Closed loop time constant =
Km τ m 30425
. τm τm
Cluett and Wang [44]
Km τ m 52586
. τm 2τ m
Model: Method 2
Km τ m 7.2291τ m 3τ m
Km τ m 91925
. τm 4τ m
Km τ m 111637
. τm 5τ m
Km τ m 131416
. τm 6τ m
Rotach [77] 0.75 Damping factor for
K mτ m . τm
241 oscillations to a disturbance
Model: Method 4 input = 0.75.
Poulin and Pomerleau [78] 2.13 Maximum sensitivity
Model: Method 2 034
. K u or 104
. Tu = 5 dB
Km Tu
Rule Kc Ti Comment
ωp 1 A m φ m + 0.5πA m (A m − 1)
ωp =
Gain and phase margin –
Kookos et al. [38]
A m Km (
ω p 05
. π − ωpτ m ) (A m
2
)
− 1 τm
Model: Method 2 0.942 . τm

4510 Gain Margin = 1.5
K mτ m Phase Margin = 22 .5 0
Representative results 4.098τ m Gain Margin = 2
0.698
0.491 6942
. τm Gain Margin = 3
0384
. 18.710τm Gain Margin = 4
Other methods
0.58 Maximum closed loop gain =
Penner [79] K mτ m 10τ m 1.26
Model: Method 2
0.8 Maximum closed loop gain =
K mτ m . τm
59 2.0
Srividya and Chidambaram 0.67075
[80] K mτ m 36547
. τm
Model: Method 5
K me− sτ m  
- controller G c ( s) = Kc  1 + 1  1
Table 9: PI tuning rules - IPD model G m ( s) = . 1 tuning rule
s  Ts
i  1 + Tf s
Rule Kc Ti Comment
Robust
. λ + 0.277
0463 τm τm
Tf = ,
Tan et al. [81] K mτ m 0.238λ + 0.123 5.750λ + 0.590
Model: Method 2 λ = 0.5
K me− sτ m Kc
Table 10: PI tuning rules - IPD model G m ( s) = - controller U( s) = Kc Y(s) − E ( s) . 1 tuning rule
s Ti s
Rule Kc Ti Comment
Direct synthesis
Chien et al. [74] 1.414TCL + τ m Underdamped system
Model: Method 1
(
K m TCL + 1.414TCL τ m + τ m
2 . TCL + τ m
1414 response - ξ = 0.707 .
τ m ≤ 0.2Tm
K me− sτ m
Table 11: PI tuning rules - IPD model G m ( s) = - Two degree of freedom controller:
s
 1 
U(s) = K c 1 + E (s ) − α Kc R( s) . 1 tuning rule
 Tis 
Rule Kc Ti Comment
Servo/regulator tuning Performance index minimisation
τm
Taguchi and Araki [61a] ≤ 1.0 .
0. 7662 Tm
K m τm 4.091 τm
Overshoot (servo step)
Model: ideal process
α = 0. 6810 ≤ settling time of tuning
rules of Chien et al. [10]
0.049 Ku 2. 826 Tu α = 0.506 , φ c = −1640
Minimum ITAE -
Pecharroman and Pagola
0.066 Ku 2. 402 Tu α = 0.512 , φ c = −1600
[134b] 0.099 Ku 1.962 Tu α = 0.522 , φ c = −1550
Km =1
0.129 Ku 1.716 Tu α = 0.532 , φ c = −1500
0.159 Ku 1.506 Tu α = 0.544 , φ c = −1450
Model: Method 6 0.189 Ku 1.392 Tu α = 0.555 , φ c = −1400
0.218 K u 1.279 Tu α = 0.564 , φ c = −1350
0.250 Ku 1.216 Tu α = 0.573 , φ c = −1300
0.286 Ku 1.127 Tu α = 0.578 , φ c = −1250
0.330 Ku 1.114 Tu α = 0.579 , φ c = −1200
0.351K u 1.093Tu α = 0.577 , φ c = −1180
Km e− sτ  
- controller G c ( s) = Kc  1 + 1  . 6 tuning rules
m
Table 12: PI tuning rules – FOLIPD model G m (s) =

s(1 + sTm )  i 
Ts
Rule Kc Ti Comment
Ultimate cycle
2
 
McMillan [58]     T  0.65 
1477
. Tm  1  . τ m 1 +  m  
332 Tuning rules developed
Model: Method not relevant K m τ 2    0.65    τ m   from Ku , Tu
m  Tm 
 1 +   
  τm  
Regulator tuning Minimum performance index
Minimum IAE – Shinskey
[59] – page 75. 0.556 3.7( τ m + Tm )
Model: Method not K m ( τ m + Tm )
specified
Minimum IAE – Shinskey
[59] – page 158 0.952 4(Tm + τ m )
Model: Open loop method Km (Tm + τ m )
not specified
Minimum ITAE – Poulin and b Tm
2
Pomerleau [82] – deduced K m ( τ m + Tm ) a τ + T 2 + 1

( m m) a( τ m + Tm )
from graph
Model: Method 2 τm Tm a b τm Tm a b τm Tm a b
0.2 5.0728 0.5231 1.0 4.7839 0.5249 1.8 4.6837 0.5256
Output step load 0.4 4.9688 0.5237 1.2 4.7565 0.5250 2.0 4.6669 0.5257
disturbance 0.6 4.8983 0.5241 1.4 4.7293 0.5252
0.8 4.8218 0.5245 1.6 4.7107 0.5254
(2 tuning rules)
0.2 3.9465 0.5320 1.0 4.0397 0.5311 1.8 4.0218 0.5313
0.4 3.9981 0.5315 1.2 4.0337 0.5312 2.0 4.0099 0.5314
Input step load disturbance
0.6 4.0397 0.5311 1.4 4.0278 0.5312
0.8 4.0397 0.5311 1.6 4.0278 0.5312
Direct synthesis
Poulin and Pomerleau [78] 2.13 Maximum sensitivity
034
. K u or 104
. Tu = 5 dB
Km Tu
Model: Method 2
Km e− sτ  
- controller G c ( s) = Kc  b + 1  . 1 tuning rule
m
Table 13: PI tuning rules – FOLIPD model G m (s) =

s(1 + sTm )  Ti s 
Rule Kc Ti Comment
Direct synthesis
. e− 0.23τ + 0.019τ b = 0.33e2.5τ −1.9 τ .
2 2
041
,
Km ( Tm + τ m ) . τ m e1.7 τ − 0.69τ
2
57 τ
Astrom and Hagglund [3] - . ≤ m ≤ 55
Ms = 1.4; 014 .
maximum sensitivity – pages τ = τ m ( τ m + Tm ) Tm
210-212
. e −1.1τ + 0.76 τ . e−1.9 τ + 1. 2τ .
2
b = 078
2
081
Km (Tm + τ m ) 3.4τ me 0.28τ − 0.0089 τ
2
τ
Model: Method 1. . ≤ m ≤ 55
Ms = 2.0; 014 .
Tm
Km e− sτ m
Table 14: PI tuning rules - FOLIPD model G m (s) = - Two degree of freedom controller:
s(1 + sTm )
 1 
U(s) = K c 1 + E (s ) − α Kc R( s) . 1 tuning rule.
 Tis 
Rule Kc Ti Comment
Servo/regulator tuning Minimum performance index




τm
1  ≤ 1.0 .
Taguchi and Araki [61a] 0 .1787 +
0 .2839  τ τ 
2
Tm
Km  τm  4. 296 + 3. 794 m + 0. 2591 m 
 + 0. 001723 Tm  Tm  Overshoot (servo step)
τ
α = 0. 6551+ 0. 01877 m
Tm
≤ settling time of tuning
rules of Chien et al. [10]
0.049 Ku 2. 826 Tu α = 0.506 , φ c = −1640
Minimum ITAE -
0.066 Ku 2. 402 Tu α = 0.512 , φ c = −1600
[134b] 0.099 Ku 1.962 Tu α = 0.522 , φ c = −1550
K m = 1 ; Tm = 1
0.129 Ku 1.716 Tu α = 0.532 , φ c = −1500
0.159 Ku 1.506 Tu α = 0.544 , φ c = −1450
0.218 K u 1.279 Tu α = 0.564 , φ c = −1350
0.250 Ku 1.216 Tu α = 0.573 , φ c = −1300
0.286 Ku 1.127 Tu α = 0.578 , φ c = −1250
0.330 Ku 1.114 Tu α = 0.579 , φ c = −1200
0.351K u 1.093Tu α = 0.577 , φ c = −1180
K m e− sτ m
Km e− sτ m
Table 15: PI tuning rules - SOSPD model or

Tm1 s2 + 2ξ m Tm1s + 1
2
(1 + Tm1s)(1 + Tm2 s)
 1
- controller G c ( s) = Kc  1 +  . 11 tuning rules.
 i 
Ts
Rule Kc Ti Comment
Robust
λ varies graphically with
Tm1 + Tm 2 + 0.5τ m Tm1 + Tm 2 + 05
. τm τm (Tm1 + Tm 2 ) -
K m τ m ( 2λ + 1) . ≤ τ m ( Tm1 + Tm 2 ) ≤ 10
01
Brambilla et al. [48]
τ m (Tm1 + Tm 2 ) λ τ m (Tm1 + Tm 2 ) λ τ m (Tm1 + Tm 2 ) λ
Model: Method 1 0.1 3.0 1.0 0.6 10.0 0.2
0.2 1.8 2.0 0.4
0.5 1.0 5.0 0.2
Direct synthesis
τm
( )
1
Gain and phase margin - Tan 2
et al. [39] – repeated pole β Ti ω φ 1 + βTm ω φ [

βω φ tan − 2 tan −1 βTm ω φ − βτ m ω φ − φ m ] β = 0.8,
Tm
< 0.33 ;
ωφ < ω u
Model: Method 11
( )
2
A m 1 + βTi ω φ τm
β = 0.5, > 0.33
Tm
Minimum IAE - Shinskey 100Tm1   −
3 Tm1

τ m  0.5 + 35
. 1 − e ( m m2 ) 
 τ +T
[59] – page 158  −
T m1 
 
K m (τ m + Tm2 ) 50 + 551 − e τ m + T m2 
 
Model: Open loop method    
  
not specified
α
Minimum ISE – McAvoy Km βτ m
and Johnson [83] – deduced
ξm τ m Tm 1 α β ξm τ m Tm 1 α β ξm τ m Tm 1 α β
from graph
1 0.5 0.8 1.82 4 0.5 4.3 3.45 7 0.5 7.8 3.85
Model: Method 1 1 4.0 5.7 12.5 4 4.0 27.1 6.67 7 4.0 51.2 5.88
1 10.0 13.6 25.0
α
Minimum ITAE – Lopez et Km βTm
al. [84] – deduced from
ξm τ m Tm 1 α β ξm τ m Tm 1 α β ξm τ m Tm 1 α β
graph
0.5 0.1 3.0 2.86 1 0.1 7.0 2.00 4 0.1 40.0 0.83
Model: Method 7 0.5 1.0 0.2 0.83 1 1.0 0.95 2.22 4 1.0 6.0 3.33
0.5 10.0 0.3 4.0 1 10.0 0.35 5.00 4 4.0 0.75 10.0
0.77 Tm1 τm T
= 02
. , m2 = 01
. ( τ m + Tm 2 )
.
Minimum IAE – Shinskey K m τm 283 Tm1 Tm 1
[17] – page 48. Model:
070
. Tm1 2.65( τ m + Tm 2 ) τm T
method not specified. = 02
. , m2 = 0.2
Km τ m Tm1 Tm1
080
. Tm1 2.29( τ m + Tm 2 ) τm T
= 02
. , m2 = 05
.
Km τ m Tm1 Tm 1
080
. Tm1 . ( τ m + Tm2 )
167 τm T
= 02
. , m2 = 10
.
Km τ m Tm1 Tm 1
Rule Kc Ti Comment
Minimum IAE - Huang et al. Tm 2 τ
1 0< ≤ 1 ; 01
. ≤ m ≤1
[18] K c ( 31) Ti ( 31) Tm1 Tm1
Model: Method 1
1  
− 0.9077 −0.063
τ Tm2 τ mTm2  τm   τm 
1
Kc (31) = 6.4884 + 4.6198 m − 3491
. − 253143
. + 0.8196  − 52132
.   
Km  Tm 1 Tm1 Tm1
2
 Tm1   Tm1  

1  
0.5961 0.7204 1.0049 1.005
 τm   Tm2   Tm2   Tm 2  T
+ − 7.2712  − 180448
.   + 5.3263  + 139108
.   + 0.4937 m2 
Km   Tm1   Tm1   Tm1   Tm1  τm 

1  Tm 2  τ m 
0.8529
Tm 2  τ m 
0.5613
τm  Tm2 
0.557
τ m  Tm 2 
1.1818

+ 191783
.   + 12.2494   + 8.4355   − 17.6781   
Km  Tm1  Tm1  Tm1  Tm1  Tm1  Tm1  Tm1  Tm1  
1  τ m Tm 2 τ m Tm 2

+ − 0.7241eT m1
− 2.2525e Tm 1
+ 54959
. e Tm1 2 
Km  
 
 τm Tm2  τm 
2
τ m Tm2  Tm2 
2
 τm  
3
Ti ( 31)

= Tm1 0.0064 + 3.9574 + 4.4087 − 6.4789  − 128702
. − 15083
.   + 9.4348   
  Tm1   Tm1   Tm1  
2
Tm1 Tm1 Tm1
 T  τm 
2
τ  Tm2 
2
T 
2
τ 
4

+ Tm 1 17.0736 m2   + 15.9816 m   − 3909
.  m 2  − 10.7619 m  
 Tm1  m1 
T Tm1  m1 
T  m1 
T  Tm1  
 T τ 
3 2
 τ  T 
2
τ
3
 Tm2  T 
4

+ Tm 1 − 10.684 m2  m  − 22.3194 m   m 2  − 6.6602 m   + 6.8122 m 2  
 Tm1  Tm1   Tm1   Tm1  Tm1  Tm 1   Tm1  
  τm 
5
Tm2  τ m 
4 2
 Tm2   τ m 
3
 τ m   Tm 2  
2 3
+ Tm1  75146
.   + 28724
.   + 114666
.     + 111207
.     
  Tm1  Tm1  Tm1   Tm1   Tm 1   Tm1   Tm1  
  τm   Tm 2 
4
 Tm 2 
5
 τm 
6
Tm2  τm 
5
 Tm2  
6

+ Tm 1 − 12174
.    − 4.3675  − 2.2236  − 0112
.   + 10308
.   
  Tm1   Tm1   Tm1   Tm1  Tm1  Tm1   Tm1  
 4
 τ  T 
2 3
 τm   Tm2 
3 2
 τ m   Tm2 
4
τm  Tm 2 
5

+ Tm1 − 1. 9136 m   m2  − 34994
.     − 15777
.     + 11408
.   
  m1   m1 
T T  m1   m1 
T T  m1   m1 
T T Tm1  Tm1  
Rule Kc Ti Comment
Minimum IAE - Huang et al. 0.4 ≤ ξ m ≤ 1 ;
2
[18] K c ( 32) Ti ( 32) τm
0.05 ≤ ≤1
Tm1
Model: Method 1
1  τm τm  τm 
1.4439
 τm 
0.1456

2
Kc ( 32)
=  − 10.4183 − 209497
. − 55175
. ξ m − 265149
. ξm + 42.7745  + 105069
.   
Km  Tm1 Tm1  Tm1   Tm1  
1  
0.3157 −0.0541
τ  τ 
+ 15.4103 m  + 34.3236ξ m 3.7057 − 17.8860ξ m 4 .5359 − 54.0584 ξ m 1.9593 + 22.4263ξ m  m  
Km   Tm1   Tm1  

1  T 
4. 7426
τ  τ  τm
+ 2.7497ξ m  m  + 50.2197ξ m 1.8288  m  − 171968
. ξ m 2.7227 + 10293
. ξ m m1 
Km   Tm1   Tm1  Tm1 τm 
 
1  
τ τm
ξ
m
− 167667 e T + 14.5737eξ − 7.3025 e 

m
+ . m1 m Tm1
Km  

 τm  τm 
2
τm  τm  
3
Ti ( 32)

= Tm1 11447
. + 45128
. − 75.2486ξ m − 110807
.   − 12.282ξ m + 345.3228ξ m + 191.9539
2
 
 Tm1  Tm1  Tm1  Tm1  
 τ 
2
τ τ  
4
+ Tm1 359 .3345ξ m  m  − 158.7611 m ξ m 2 − 770.2897ξ m2 − 153633

.  m 
  Tm1  Tm1  Tm1  
 τ 
3
τ 
2
τm 
+ Tm1 − 412 .5409 ξ m  m  − 414 .7786ξ m 2  m  + 4850976
. ξ m 3 + 864.5195ξ m 4 
  Tm1   Tm1  Tm1 
 τ 
5
τ 
4
τ 
3
 τm 
2

+ Tm 1 55.4366 m  + 222 .2865 ξ m  m  + 275166
. ξ m 2  m  + 2052493
.   ξm 
3
  Tm1   Tm1   Tm1   Tm1  
  τm  4  τm 
6
 τm 
5


+ Tm 1 − 479 .5627  ξ m − 4731346
. ξ m − 6.547
5
 − 432822
. ξ m  + 99.8717ξ m 
6
  Tm1   Tm1   Tm1  
  τm 
4
 τm 
3
 τm 
2
τ 
+ Tm1 − 735666
.   m ξ 2
− 56.4418  m ξ 3
− 37 .497  ξ m + 160.7714 m ξ m 
4 5
  Tm1   Tm1   Tm1  Tm1 

Rule Kc Ti Comment
Servo tuning Minimum performance index
Minimum IAE - Huang et al. Tm 2 τ
3 0< ≤ 1 ; 01
. ≤ m ≤1
[18] K c ( 33) Ti ( 33) Tm1 Tm1
Model: Method 1
1  
− 1.0169
τ   τm 
3.5959
τ T τ T
3
Kc (33) = − 130454
. − 9.0916 m + 2.6647 m2 + 9.162 m m2 2 + 0.3053 m  + 11075
.   
Km  Tm1 Tm 1 T  Tm1   Tm1  
 m1
1  T 
3.6843 0.8476 2.6083 2.9049
τ   Tm2  T  T 
+ − 2.2927 m  − 310306
.   − 13.0155 m2  + 9.6899 m 2  − 0.6418 m 2 
Km   Tm1   Tm1   Tm1   Tm1  τm 
 
1  
−0.2016 1.3293 0.801
Tm2  τ m  Tm 2  τ m  τ m  Tm 2 
+ 189643
.   − 39.7340   + 28155
.   
Km  Tm1  Tm1  Tm1  Tm1  Tm 1  Tm1  

1  
3.956 τ T τ T
τ m  Tm2 
m m2 m m2
− 2.0067 eT 
2
+   + 4.8259e + 2.1137e + 84511
T m1 T
m1
. m1
Km  Tm1  Tm1  
 
2
τ m Tm2  Tm 2 
2
 τm  
3
= Tm1 0.9771− 0.2492 + 0.8753 + 3.4651  − 38516 + 7.5106   − 7 .4538  

( 33)
Ti .
  Tm1   Tm1   Tm1  
2
Tm1 Tm1 Tm1
 
 Tm2  τ m 
2
τ m  Tm 2 
2
 Tm 2 
3
 τm  
4
+ Tm1 116768
.   − 10.9909   − 161461
.   + 82567
.   
 Tm1  Tm1  Tm1  Tm1   Tm1   Tm1  

 Tm2  τ m 
3 2
 τ  T 
2
τ m  Tm 2 
3
 Tm 2  
4
+ Tm 1 − 181011
.   + 6.2208 m   m2  + 219893
.   + 158538
.   
 Tm1  Tm1   Tm1   Tm1  Tm 1  Tm1   Tm1  
 τ 
5
T τ 
4 2
T   τ 
3
 τ  T  
2 3
+ Tm1  − 4.7536 m  + 14.5405 m 2  m  − 2.2691 m2   m  − 8.387  m   m 2  

  Tm 1  Tm1  Tm1   Tm1   Tm1   Tm1   Tm1  

  τ m   Tm2 
4
 Tm 2 
5
 τm 
6
Tm2  τ m 
5
 Tm 2  
6

+ Tm 1 − 16.651   − 71990
.   + 11496
.   − 4.728   + 11395
.   
  Tm1   Tm 1   Tm1   Tm1  Tm 1  Tm1   Tm1  
 4
 τ  T 
2 3
 τ m   Tm 2 
3 2
 τ m   Tm2 
4
τ  Tm 2 
5

+ Tm 1 0.6385 m   m 2  + 10885
.     + 31615
.     + 4.5398 m   
  m1   m1 
T T  m1   m1 
T T  m1   m1 
T T Tm1  Tm1  
Rule Kc Ti Comment
Minimum IAE - Huang et al. 0.4 ≤ ξ m ≤ 1 ;
4
[18] K c ( 34 ) Ti ( 34) τm
0.05 ≤ ≤1
Tm1
Model: Method 1
1  τm τm  τm 
3
 τm  
2
4
Kc ( 34)
= − 10.95 − 18845
. − 3.4123ξ m + 4.5954ξ m − 17002
.   − 21324
. ξ m  
Km  Tm1 Tm1  Tm1   Tm 1  
1  
0.421 0.1984 1.8033
τ  τ  τ   τm 
+  − 14.4149ξ m 2  m  − 0.7683ξ m 3 + 7.5142 m  + 3.7291 m  + 53444
.   
Km   Tm1   Tm1   Tm1   Tm1  

1  
−0.6753 −0.1642
τ  τ 
+  − 0.0819 ξ m 19 .5419 − 3603
. ξ m 1.0749 + 71163
. ξ m 1.1006 + 3206
. ξm m  − 7.8480ξ m  m  
Km   Tm1   Tm1  

1  Tm1 
τ ξ m τm
τ 
m
+ 113222
. ξ m1.9948  m  + 2.4239e T m1
+ 34137
. eξ + 10251
m
. e Tm1
− 05593
. ξm 
Km   Tm1  τm 
 
 τm  τm 
2
τm  τm  
3
Ti ( 34)

= Tm1 2.4866 − 233234
. + 53662
. ξ m + 656053
.   + 29.0062ξ m − 24.1648ξ m − 83.6796
2
 
 Tm1  Tm1  Tm1  Tm1  
 τ 
2
τm τ  
4
+ Tm1  − 1359699
. ξ m  m  + 431477
. ξ m 2 + 519749
. ξ m 3 + 86.0228 m  
  Tm1  Tm 1  Tm1  

 τ 
3
τ 
2
τm 
+ Tm 1 704553
. ξ m  m  + 1534877
. ξ m 2  m  − 1250112
. ξ m 3 − 685893
. ξ m4
  Tm1   Tm1  Tm1 
 τ 
5
τ 
4
τ 
3
τ 
2

+ Tm 1 − 62.7517 m  + 27.6178 ξ m  m  − 152 .7422 ξ m 2  m  + 20.8705 m  ξ m 3 
  Tm1   Tm1   Tm1   Tm 1  
  τm  4  τm 
6
 τm 
5


+ Tm1 54.0012  ξ m + 58.7376ξ m + 131193
5
.   + 202645
. ξm   − 232064
. ξm 6 
  Tm1   Tm1   Tm1  
  τm 
4
 τm 
3
 τm 
2
τm 
+ Tm 1 − 616742
.   m ξ 2
+ 136.2439  m ξ 3
− 954092
.   ξ m + 204168
4
. ξm 5 
  Tm1   Tm1   Tm1  Tm1 
Rule Kc Ti Comment
Ultimate cycle
Decay ratio = 0.15 - ε < 2.4 ,
Regulator - nearly minimum Kc ( 7)
Ti ( 7) τm
0.2 ≤ ≤ 2.0 , 0.6 ≤ ξ m ≤ 4.2
IAE, ISE, ITAE – Hwang Tm1
[60] Decay ratio = 0.15 -
Kc( 8) Ti (8) τm
2.4 ≤ ε < 3 , 0.2 ≤ ≤ 2.0 ,
Model: Method 3 Tm1
0.6 ≤ ξ m ≤ 4.2
Decay ratio = 0.15 -
Kc ( 9 ) 5 Ti ( 9) τm
3 ≤ ε < 20 , 0.2 ≤ ≤ 2.0 ,
Tm1
0.6 ≤ ξ m ≤ 4.2
6Tm1 + 4ξ m Tm1 τm + K H Km τ m
2 2
9  τm 2 ξ m Tm1τ m 
2
5
ε= , KH =  − T
2
− ,
2 Tm1 τ mω H
2 m1
2τ m K m
2
 18 18 
1 + KH K m
ωH =
2T τ ξ K K τ 2
Tm12 + m1 m m + H m m
3 6
( 7)


= 1 −
[
0.674 1 − 0.447ω H τ m + 0.0607 (ω H τ m )
2
] K , Ti ( 7) =
K c (1 + K H K m )
( )
Kc
K H K m (1 + K H K m )
H
  0.0607ω H K m 1 + 1.05ω H τ m − 0.233ω H τ m
2 2
 
(8 )


= 1 −
[
0.778 1 − 0.467ω H τ m + 0.0609(ω H τ m )
2
] K , Ti (8 ) =
K c (1 + K H K m )
( )
Kc
K H K m (1 + K H K m )
H
  0.0309ω H K m 1 + 2.84ω H τ m − 0.532 ω H τ m
2 2
 
( 9)


= 1 −
. ( 0519
131 . )
ω τ
1 − 103
H m
[
. ε + 0514
. ε 2 
K H , Ti =
]
( 9) K c (1 + KH K m )
( ) ( )( )
Kc 
 KH Km 1 + KH K m  0.0603 1 + 0.929 ln[ ω H τ m ] 1 + 2.01 ε − 12
. ε2
Rule Kc Ti Comment
Kc ( 10 ) 6
Ti ( 10) Decay ratio = 0.15 - ε > 20 ,
τm
0.2 ≤ ≤ 2.0 , 0.6 ≤ ξ m ≤ 4.2
Regulator – nearly minimum Tm1
IAE, ISE, ITAE - Hwang [60] Decay ratio = 0.2 - ε < 2.4 ,
Kc ( 11)
Ti ( 11) τm
0.2 ≤ ≤ 2.0 , 0.6 ≤ ξ m ≤ 4.2
Model: Method 3 Tm1
Decay ratio = 0.2 -
Kc ( 12 )
Ti ( 12) τm
2.4 ≤ ε < 3 , 0.2 ≤ ≤ 2.0 ,
Tm1
0.6 ≤ ξ m ≤ 4.2
Decay ratio = 0.2 -
Kc(13 ) Ti (13) τm
3 ≤ ε < 20 , 0.2 ≤ ≤ 2.0 ,
Tm1
0.6 ≤ ξ m ≤ 4.2
Decay ratio = 0.2 - ε > 20 ,
Kc(14 ) Ti (14) τm
0.2 ≤ ≤ 2.0 , 0.6 ≤ ξ m ≤ 4.2
Tm1
Decay ratio = 0.25 - ε < 2.4 ,
Kc(15) Ti (15) τm
0.2 ≤ ≤ 2.0 , 0.6 ≤ ξ m ≤ 4.2
Tm1
Kc (16 )
Ti ( 16) τm
2.4 ≤ ε < 3 , 0.2 ≤ ≤ 2.0 ,
Tm1
0.6 ≤ ξ m ≤ 4.2
6 ( 10)


= 1 −
[
. 1 − 0.482ω H τ m + 0.068( ω H τ m )
114
2
] K , Ti (10) =
K c (1 + K H K m )
( )
Kc
K H K m (1 + K H K m )
H
  0.0694ω H K m − 1 + 2.1ω H τ m − 0.367ω H τm
2 2
 
( 11)


= 1 −
[
0.622 1 − 0.435ω Hτ m + 0.052(ω H τm )
2
] K , Ti (11) =
K c (1 + K H K m )
( )
Kc
K H K m (1 + K H K m )
H
  0.0697 ω H K m 1 + 0.752ω H τ m − 0.145ω H τ m
2 2
 
( 12 )


= 1 −
[
0.724 1 − 0.469ω H τ m + 0.0609 (ω H τ m )
2
] K , Ti (12 ) =
K c (1 + K H K m )
( )
Kc
K H K m (1 + K H K m )
H
  0.0405ω H K m 1 + 1.93ω H τ m − 0.363ω H τ m
2 2
 

Kc (13) =  1 −
. (0.506)
126
ω τ
1 − 107
H m
[
. ε + 0.616 ε 2  ]
K H , Ti (13) =
K c (1 + K H K m )
 K H Km 1 + KH K m ( )
 0.0661(1 + 0.824 ln[ω H τ m ])(1 + 171 . ε2 )
. ε − 117
( 14 )


= 1 −
109 [
. 1 − 0.497ω H τ m + 0.0724(ω Hτ m )
2
] K , Ti (14 ) =
K c (1 + K H K m )
( )
Kc
K H K m (1 + K H K m )
H
  0.054 ω H K m − 1 + 2.54ω H τ m − 0.457ω H τ m
2 2
 
( 15)


= 1 −
[
0.584 1 − 0.439ω H τ m + 0.0514 ( ω H τ m )
2
] K , Ti (15) =
K c (1 + K H K m )
( )
Kc
K H K m (1 + K H K m )
H
  0.0714ω H K m 1 + 0.685ω H τ m − 0.131ω H 2 τ m 2
 
( 16)


= 1 −
[
0.675 1 − 0.472ω H τ m + 0.061( ω H τ m )
2
] K , Ti (16) =
K c (1 + K H K m )
( )
Kc
K H K m (1 + K H K m )
H
  0.0484ω H K m 1 + 1.43ω H τ m − 0.273ω H τ m
2 2
 
Rule Kc Ti Comment
Regulator - nearly minimum Kc(17 ) 7 Ti (17) τm
3 ≤ ε < 20 , 0.2 ≤ ≤ 2.0 ,
IAE, ISE, ITAE - Hwang [60] Tm1
0.6 ≤ ξ m ≤ 4.2
Model: Method 3 Decay ratio = 0.25 - ε > 20 ,
Kc (18 ) Ti (18) τm
0.2 ≤ ≤ 2.0 , 0.6 ≤ ξ m ≤ 4.2
Tm1
Decay ratio = 0.1 - ε < 2.4 ,
Servo - nearly minimum IAE, τm
0.2 ≤ ≤ 2.0 , 0.6 ≤ ξ m ≤ 4.2
ISE, ITAE - Hwang [60] Kc ( 19 )
Ti ( 19) Tm1
2
τm τ 
ξ m ≤ 0.776 + 0.0568 + 018
.  m
Model: Method 3 Tm1  T m1 
Decay ratio = 0.1 - 2.4 ≤ ε < 3 ,

τm
0.2 ≤ ≤ 2.0 , 0.6 ≤ ξ m ≤ 4.2
( 20) ( 20) Tm1
Kc Ti
2
τm τ 
ξ m ≤ 0.776 + 0.0568 + 018
.  m
Tm1  T m1 
Decay ratio = 0.1 - 3 ≤ ε < 20 ,

τm
0.2 ≤ ≤ 2.0 , 0.6 ≤ ξ m ≤ 4.2
( 21) ( 21) Tm1
Kc Ti
2
τm τ 
ξ m ≤ 0.776 + 0.0568 + 018
.  m
Tm1  T m1 
Decay ratio = 0.1 - ε > 20 ,

τm
0.2 ≤ ≤ 2.0 , 0.6 ≤ ξ m ≤ 4.2
( 22 ) ( 22 ) Tm1
Kc Ti
2
τm τ 
ξ m ≤ 0.776 + 0.0568 + 018
.  m
Tm1  T m1 
 12. ( 0495
. ) [
. ε + 0.698 ε  ]
ω τ
1 − 11
H m 2
  K , T (17 ) = Kc (1 + KH K m )
= 1−
7 ( 17)
( )( )
Kc
 KH Km (1 + K H Km )  0.0702 1 + 0.734 ln[ ω H τ m ] 1 + 148
i
. ε − 11
. ε2
H
 
( 18)


= 1 −
[
1.03 1 − 0.51ω H τ m + 0.0759( ω H τ m )
2
] K , Ti (18) =
K c (1 + K H K m )
( )
Kc
K H K m (1 + K H K m )
H
  0.0386ω H K m − 1 + 3.26ω H τ m − 0.6ω H τm
2 2
 
( 19 )


= 1 −
[
0.822 1 − 0.549ω H τ m + 0.112(ω Hτ m )
2
] K , Ti (19 ) =
K c (1 + K H K m )
( )
Kc
K H K m (1 + K H K m )
H
  0.0142ω H K m 1 + 6.96ω H τ m − 177
. ωH τm
2 2
 
( 20)


= 1 −
[
0.786 1 − 0.441ω H τ m + 0.0569(ω H τ m )
2
] K , Ti ( 20) =
K c (1 + K H K m )
( )
Kc
K H K m (1 + K H K m )
H
  0.0172 ω H K m 1 + 4.62ω H τ m − 0.823ω H τ m
2 2
 
 128
Kc ( 21) = 1 −
. ( 0542
. )
ω τ
[
1 − 0.986 ε + 0558
H m
. ε 2  ] K c (1 + K H K m )
( ) KH , Ti = 0.0476 1 + 0.996 ln ω τ 1 + 213
( 21)
 K H K m 1 + K K
H m  ( [ H m ])( . ε − 113
. ε2 )
( 22)


= 1 −
114 [
. 1 − 0.466 ω H τ m + 0.0647 ( ω H τ m )
2
] K , Ti (22) =
K c (1 + K H K m )
( )
Kc
K H K m (1 + K H K m )
H
  0.0609ω H K m − 1 + 1.97ω H τ m − 0.323ω H 2 τ m2
 
Rule Kc Ti Comment
Decay ratio = 0.1 - ε < 2.4 ,
8 0.2 ≤ ≤ 2.0 , 0.6 ≤ ξ m ≤ 4.2
ISE, ITAE - Hwang [60] K c ( 23) Ti ( 23) Tm1
2
τm  τ 
ξ > 0889
. + 0.496 + 0.26 m 
Model: Method 3 Tm1  Tm1 
Decay ratio = 0.1 - 2.4 ≤ ε < 3 ,

τm
0.2 ≤ ≤ 2.0 , 0.6 ≤ ξ m ≤ 4.2
K c ( 24 ) Ti ( 24) Tm1
2
τm  τ 
ξ > 0889
. + 0.496 + 0.26 m 
Tm1  Tm1 
Decay ratio = 0.1 - 3 ≤ ε < 20 ,

τm
( 25) ( 25)
0.2 ≤ ≤ 2.0 , 0.6 ≤ ξ m ≤ 4.2
Kc Ti Tm1
2
τm  τ 
ξ > 0889
. + 0.496 + 0.26 m 
Tm1  Tm1 
Decay ratio = 0.1 - ε > 20 ,

τm
0.2 ≤ ≤ 2.0 , 0.6 ≤ ξ m ≤ 4.2
K c (26) Ti (26) Tm1
2
τm  τ 
ξ > 0889
. + 0.496 + 0.26 m 
Tm1  Tm1 
Decay ratio = 0.1 - ε < 2.4 ,

τm
0.2 ≤ ≤ 2.0 , 0.6 ≤ ξ m ≤ 4.2
K c ( 27 ) Ti ( 27) Tm1
2
τm τ 
ξ m > 0.776 + 0.0568 + 018
.  m 
Tm1  T m1 
2
τm τ 
ξ m ≤ 0.889 + 0.496 + 0.26 m 
Tm1  Tm1 
8 ( 23) 
= 1 −
[
 0.794 1 − 0541
. ω H τ m + 0.126( ω H τ m )
2
] K , Ti (23) =
K c (1 + K H K m )
( )
Kc
K H K m (1 + KH K m )
H
  0.0078ω H K m 1 + 8.38ω H τ m − 197
. ω H2 τ m 2
 
( 24)
= 1 −
[
 0.738 1 − 0.415ω τ + 0.0575 ω τ 2
 H m ( H m) ]  K , Ti (24) =
K c (1 + K H K m )
( )
Kc
KH K m (1 + KH K m )
H
  0.0124ω H K m 1 + 4.05ω H τ m − 0.63ω H 2 τ m 2
 
( 25)
 115
= 1 −
[
. (0.564) H m 1 − 0.959 ε + 0.773 ε 
ω τ 2
 K , T (25) =] K c (1 + K H K m )
( ( )( )
Kc

 K K
H m 1 + K K
H m )  H i
 0 .0355 1 + 0 .947 ln[ω H τ m ] 1 + 19
. ε − 107
. ε2
 107

Kc ( 26) =  1 −
[
. 1 − 0.466ω H τ m + 0.0667( ω H τ m )
2
] K , Ti (26) =
K c (1 + K H K m )


K H Km (1 + K H Km ) 

H
(
0.0328ω H K m − 1 + 2.21ω H τ m − 0338
. ω H2 τ m 2 )
( 27)
= 1 −
[
 0.789 1 − 0.527ω τ + 0.11 ω τ 2
 H m ( H m) ]  K , Ti (27) =
K c (1 + K H K m )
( )
Kc
K H K m (1 + K H K m )
H
  0.009ω H K m 1 + 9.7ω H τ m − 2.4ω H 2 τ m 2
 
Rule Kc Ti Comment
Decay ratio = 0.1 - 2.4 ≤ ε < 3 ,
9 0.2 ≤ ≤ 2.0 , 0.6 ≤ ξ m ≤ 4.2
ISE, ITAE - Hwang [60] K c (28) Ti (28) Tm1
2
τm τ 
ξ m > 0.776 + 0.0568 + 018
.  m 
Model: Method 3 Tm1  T m1 
2
τm τ 
ξ m ≤ 0.889 + 0.496 + 0.26 m 
Tm1  Tm1 
Decay ratio = 0.1 - 3 ≤ ε < 20 ,

τm
0.2 ≤ ≤ 2.0 , 0.6 ≤ ξ m ≤ 4.2
K c ( 29 ) Ti ( 29) Tm1
2
τm τ 
ξ m > 0.776 + 0.0568 + 018
.  m 
Tm1  T m1 
2
τm τ 
ξ m ≤ 0.889 + 0.496 + 0.26 m 
Tm1  Tm1 
Decay ratio = 0.1 - ε > 20 ,

τm
0.2 ≤ ≤ 2.0 , 0.6 ≤ ξ m ≤ 4.2
K c ( 30) Ti (30) Tm1
2
τm τ 
ξ m > 0.776 + 0.0568 + 018
.  m 
Tm1  T m1 
2
τm τ 
ξ m ≤ 0.889 + 0.496 + 0.26 m 
Tm1  Tm1 
Regulator - minimum IAE -

Shinskey [17] – page 121. 048
. Ku 083
. Tu τm T
Model: method not = 02
. , m2 = 0.2
Tm1 Tm1
specified
9 ( 28) 
= 1 −
[
 0.76 1 − 0426
. ω H τ m + 0.0551( ω H τ m )
2
] K , Ti (28) =
K c (1 + K H K m )
( )
Kc
K H Km ( 1 + K H Km )
H
  0.0153ω H K m 1 + 4.37ω H τ m − 0.743ω H 2 τ m 2
 
( 29) 
= 1−
[
. (0.55) H m 1 − 0.978 ε + 0.659 ε 
 122 ω τ 2
]
 K , T (29) = K c (1 + K H K m )
( ( )( )
Kc

 K K
H m 1 + K K
H m )  H i
 0.0421 1 + 0.969 ln[ω H τ m ] 1 + 2.02 ε − 111
. ε2
( 30)
 111

= 1 −
[
. 1 − 0.467ω H τ m + 0.0657( ω H τ m )
2
] K , Ti (30) =
K c (1 + K H K m )
( )
Kc
KH K m ( 1 + K H Km )
H
  0.0477ω H K m − 1 + 2.07ω H τ m − 0.333ω H2 τ m 2
 
K m e− sτ m
Table 16: PI tuning rules - SOSPD model -

Tm1 s2 + 2ξ m Tm1s + 1
2
 
- Two degree of freedom controller: U(s) = K c 1 + 1 E (s ) − α Kc R( s) . 3 tuning rules.
 Tis 
Rule Kc Ti Comment




τm
1 
≤ 1.0 ; ξ m = 1
Taguchi and Araki [61a] 0.5613  Ti
( 30a ) 10
Tm
0. 3717+
K m  τm 
 + 0 .0003414 Overshoot (servo step)
2 3
τm τ  τ 
α = 0. 6438− 0. 5056 + 0. 3087 m  − 0. 1201 m 

Tm  Tm   Tm  rules of Chien et al. [10]




τm
≤ 1.0 ; ξ m = 0.5
1  0.05627  Ti
( 30b )
Tm
 0. 1000+ 
Km τm
 [ + 0.06041] 
2

 T 
m
2 3 4
τ τ  τ  τ 
α = 0. 6178− 0. 4439 m − 7. 575 m  + 9. 317 m  − 3. 182 m
 T


Tm  Tm   Tm   m  rules of Chien et al. [10]
Minimum ITAE - α = 0. 4002 ,
Pecharroman and Pagola 0.1713 Ku 1.0059 Tu φ c = − 139.65 0
[134a]
K m = 1 ; Tm = 1 ; ξ m = 1
Model: Method 15
0.147 K u 1.150 Tu α = 0. 411 , φ c = −1460
Minimum ITAE -
0.170 K u 1.013Tu α = 0. 401 , φ c = −1400
[134b] 0.195 K u 0.880 Tu α = 0.386 , φ c = −1330
K m = 1 ; Tm = 1 ; ξ m = 1
0.210 K u 0.720 Tu α = 0.342 , φ c = −1250
0.234 K u 0.672 Tu α = 0.345 , φ c = −1150
Model: Method 15 0.249 K u 0.610 Tu α = 0.323 , φ c = −1050
0.262 K u 0.568 Tu α = 0.308 , φ c = −940
0.274 K u 0.545 Tu α = 0. 291 , φ c = −840
0.280 K u 0.512 Tu α = 0. 281 , φ c = −730
0.291K u 0.503 Tu α = 0.270 , φ c = −630
0.297 K u 0.483 Tu α = 0.260 , φ c = −520
0.303 K u 0.462 Tu α = 0.246 , φ c = −410
0.307 K u 0.431Tu α = 0.229 , φ c = −300
 τm  τm 
2
 τ m  
3
10
Ti
( 30a ) 
= Tm 2.069 − 0.3692 + 1.081  − 0.5524  
 

Tm  Tm   Tm  
 τ τ 
2
τ 
3
τ  
4
= Tm  4. 340 − 16. 39 m + 30. 04 m  − 25 .85 m  + 8.567  m  

( 30b )
Ti
 Tm  Tm   Tm   Tm  

Rule Kc Ti Comment
Minimum ITAE - 0.317 K u 0.386 Tu α = 0. 171 , φ c = −190
[134b] - continued
0.324 K u 0.302 Tu α = 0.004 , φ c = −100
0.320 K u 0.223 Tu α = −0.204 , φ c = − 60
K m e −sτ m
Table 17: PI tuning rules - SOSIPD model (repeated pole) -

s (1 + Tm1s) 2
 
- Two degree of freedom controller: U(s) = K c 1 + 1 E (s ) − α Kc R( s) . 1 tuning rule.
 Tis 
Rule Kc Ti Comment




τm
≤ 1.0 ;
Taguchi and Araki [61a] 1 
+
0 .3840  τ Tm
K m 
0. 07368
τm  8.549 + 4.029 m
 + 0.7640 Tm Overshoot (servo step)
τm ≤ settling time of tuning
α = 0. 6691 + 0.006606
Tm rules of Chien et al. [10]
0.049 Ku 2. 826 Tu α = 0.506 , φ c = −1640
Minimum ITAE -
0.066 Ku 2. 402 Tu α = 0.512 , φ c = −1600
[134b] 0.099 Ku 1.962 Tu α = 0.522 , φ c = −1550
K m = 1 ; Tm = 1
0.129 Ku 1.716 Tu α = 0.532 , φ c = −1500
0.159 Ku 1.506 Tu α = 0.544 , φ c = −1450
0.218 K u 1.279 Tu α = 0.564 , φ c = −1350
0.250 Ku 1.216 Tu α = 0.573 , φ c = −1300
0.286 Ku 1.127 Tu α = 0.578 , φ c = −1250
0.330 Ku 1.114 Tu α = 0.579 , φ c = −1200
0.351K u 1.093Tu α = 0.577 , φ c = −1180
Km e − sτ  
- controller G c ( s) = Kc  1 + 1  . 1
m
Table 19: PI tuning rules - TOLPD model

(1 + sTm1 )(1 + sTm2 )(1 + sTm 3 )  i 
Ts
tuning rule
Rule Kc Ti Comment
0.7  Tm1 
0.333
τm
  > 0.04 ; Tm1 ≥ Tm 2 ≥ Tm3
Hougen [85] . τm Tm1( Tm 2 + Tm3 )
0.08
Km  τm  15 Tm1
Model: Method 1
τm
≤ 0.04 ; Tm1 ≥ Tm 2 ≥ Tm3
1   Tm 1 
Tm1( Tm 2 + Tm3 )
0. 333 
T + Tm2 + Tm 3
. τm
0.08
0.7   + 0.8 m1  15 Tm1
2K m   τ m  ( Tm 1Tm 2 Tm 3 ) 
0. 333
 
K m e −sτ m
Table 20: PI tuning rules - TOLPD model (repeated pole) -

(1 + Tm1s ) 3
 
- Two degree of freedom controller: U(s) = K c 1 + 1 E (s ) − α Kc R( s) . 1 tuning rule.
 Tis 
Rule Kc Ti Comment




τm
1  0.7399  ( 30 c ) 1 ≤ 1.0 ;
Taguchi and Araki [61a] 0. 2713+ Ti Tm
K m  τm 
 + 0.5009 Overshoot (servo step)
2
τ τ  ≤ settling time of tuning
α = 0. 4908 − 0. 2648 m + 0. 05159  m  rules of Chien et al. [10]
Tm  Tm 
 τ τ  
2
= Tm  2.759 − 0. 003899 m + 0.1354  m  

1 ( 30c )
Ti
 

Tm  Tm  
K me − sτ  
- controller G c ( s) = Kc  1 + 1  . 6 tuning rules
m
Table 21: PI tuning rules - unstable FOLPD model

1 − sTm  i 
Ts
Rule Kc Ti Comment
Direct synthesis
1 τm
De Paor and O’Malley [86] ( 35) gain margin = 2; <1
Kc  1 − Tm τm  Tm
Tm   tan( 0.5φm )
Model: Method 1  Tm τ m  φ m = tan − 1
1 − Tm τ m
Tm τ m
(
− T m τ m 1 − Tm τ m )
Venkatashankar and Kc( 36) 25( Tm − τ m ) τm

< 0.67
Chidambaram [87] Tm
Model: Method 1
Chidambaram [88] 1  T  τm
Model: Method 1  1 + 0.26 m  25Tm − 27τ m < 06
.
Km  τm  Tm
. A m (A m − 1)
A m φ m + 157
ωp = ,
Ho and Xu [90] ω p Tm 1 (A m
2
)
−1 τm
A m Km 1
1.57ω p − ω p τ m −
2
τm
Model: Method 1 Tm < 0.62
Tm
Robust
 λ   λ 
Rotstein and Lewin [89] Tm λ + 2 λ + 2 λ obtained graphically –
 Tm   Tm 
sample values below
λ Km
2
Km uncertainty = 50% τm Tm = 0.2 λ = [0.6Tm ,1.9Tm ]

Model: Method 1 τm Tm = 0.2 λ = [ 0.5Tm , 45
. Tm ]
Km uncertainty = 30% τm Tm = 0.4 λ = [15
. Tm ,4.5Tm ]
τm Tm = 0.6 λ = [ 39
. Tm ,41
. Tm ]
Ultimate cycle
Maximum closed loop log
Luyben [91] 031
. Ku 2.2Tu modulus = 2 dB; closed
Model: Method not relevant loop time constant =
. τm
316
2
Tm  1 − Tm τ m 
= cos (1 − Tm τ m )Tm τ m + (1 − Tm τ m )Tm τ m 
2 ( 35)
Kc sin
K m  Tm τ m 
β 2 Tm2 (Tm − τ m )
2
1+
 2   (Tm − τ m ) 2 τ 2m : τm
Kc ( 36)
=
1  0.98 1 + 0.04Tm   25  β( T − τ ) β = 1373
. , < 025
.
   τm 
( Tm − τ m )
2 m m
625
Km   1+β2 ( Tm − τ m ) 2 Tm
τm
2
τm
β = 0.953, 0.25 ≤ < 0.67
Tm
K m e− s τ  
- controller G c ( s) = Kc  1 + 1  . 3
m
Table 22: PI tuning rules - unstable SOSPD model

(1 − sTm1 )(1 + sTm 2 )  i 
Ts
tuning rules.
Rule Kc Ti Comment
Ultimate cycle
McMillan [58] Kc( 37) 3 Ti (37 ) Tuning rules developed
Model: Method not relevant from Ku , Tu
Minimum ITAE – Poulin and aTm2 2 4Tm1 ( τ m + Tm2 )
bTm1 1 +
aTm1 − 4( τ m + Tm2 )
Pomerleau [82] – deduced 4( τm + Tm 2 )
2
from graph
(
K m aTm1 − 4[ τ m + Tm2 ] )
Model: Method 1 τm Tm a b τm Tm a b τm Tm a b
0.05 0.9479 2.3546 0.25 1.4905 2.5992 0.45 2.0658 2.9004
Output step load
0.10 1.0799 2.4111 0.30 1.6163 2.6612 0.50 2.2080 2.9826
disturbance
0.15 1.2013 2.4646 0.35 1.7650 2.7368
0.20 1.3485 2.5318 0.40 1.9139 2.8161
(considered as 2 tuning
rules)
0.05 1.1075 2.4230 0.25 1.5698 2.6381 0.45 2.1022 2.9210
0.10 1.2013 2.4646 0.30 1.6943 2.7007 0.50 2.2379 3.0003
Input step load disturbance 0.15 1.3132 2.5154 0.35 1.8161 2.7637
0.20 1.4384 2.5742 0.40 1.9658 2.8445
2
 
 
    (T + T )T T  
0.65
1477
. Tm1Tm 2 1 
3
Kc ( 37) =   , T ( 37)
= 332
. τ 1 + 
m1 m2 m1 m 2
 
  (Tm1 − Tm 2 )(Tm1 − τ m )τ m  
i m
K m τ m2   ( Tm1 + Tm 2 )Tm1Tm 2  
0.65

1+   
  (Tm1 − Tm2 )(Tm1 − τ m )τ m  
Kc ( 38) =
Ti A max
2 2 6
(
Tm1 Tm2 ω max + Tm1 + Tm 2 ω max + ω max
2 2
) 4 2
1 + Ti ω max
2 2
Km
 1
Table 23: PI tuning rules – delay model e − sτ m - controller G c ( s) = Kc  1 +  . 2 tuning rules
 i 
Ts
Rule Kc Ti Comment
Direct synthesis
Hansen [91a]
Model: Method not 0.2 0.3τm
specified
Shinskey [57] – minimum
IAE – regulator - page 67. 0.4 0.5τ m
Model: method not Km
specified
K m e− sτ
m
 1 
Table 25: PID tuning rules - FOLPD model - ideal controller G c ( s) = Kc  1 + + Td s . 56.
1 + sTm  Ti s 
Rule Kc Ti Td Comment
Process reaction
Ziegler and Nichols 12
. Tm 2Tm
[ , ] . τm
[8] K m τ m K m τm 2τ m 05 Quarter decay ratio
Model: Method 1
Astrom and 0.94Tm
Hagglund [3] – page K mτ m 2τ m . τm
05 Ultimate cycle Ziegler-
139 Nichols equivalent
Model: Method 6
Parr [64] – page 194 125
. Tm
Model: Method 1 K mτ m 2.5τ m 0. 4τ m
0% overshoot;
Chien et al. [10] – 095
. Tm 2.38τ m 0.42τ m τ
. < m <1
011
regulator Km τ m Tm
20% overshoot;
Model: Method 1
12
. Tm 2τ m 0.42τ m τ
. < m <1
011
K mτ m Tm
0% overshoot;
Chien et al. [10] - 0.6Tm Tm 0.5τ m τ
. < m <1
011
servo K mτ m Tm
20% overshoot;
Model: Method 1
095
. Tm 1.36Tm 0.47τ m τ
. < m <1
011
Km τ m Tm
Three constraints Quarter decay ratio;
method - Murrill [13]- 0.950 0.738 0.950 minimum integral error
1370
.  Tm  Tm  τm  τ 
page 356     0.365Tm  m  (servo mode);
Km  τm  1351
.  Tm   Tm  Kc K m Td
Model: Method 3 = 05
. ;
Tm
τm
. ≤
01 ≤1
Tm
   
2
 2.5 τ m + 0.46 τ m   0.37 τ m

Cohen and Coon [11] 1  Tm 
 135
. .  T  Tm
+ 025  Tm  
 τ Quarter decay ratio
Km  τm  m τm  1 + 0.2 m
Model: Method 1 1 + 0.61 Tm
 Tm 
 
Astrom and T90% = 90% closed

Hagglund [93]- 3 T90% . τm
05 loop step response
pages 120-126 Km time. Leeds and
Northrup Electromax V.
Model: Method 19
Sain and Ozgen [94] 1  T  0.8647Tm + 0.226τ m 0.0565Tm
 0. 6939 m + 01814
.  Tm T
Model: Method 5 K m  τm  + 0.8647 0.8647 m + 0.226
τm τm
Minimum IAE –  Tm 
0.921
Tm  τm 
0.749
τ 
1.137
τm
Murrill [13] – pages
1435
.
    0.482 Tm  m  . <
01 ≤1
Km  τ m   Tm  Tm
358-363 0878
.  Tm 
Model: Method 3
Modified minimum 0 .921 1.137
3  Tm  τ 
0.749
Tm  τm 
IAE - Cheng and     0.482 Tm  m 
Hung [95] K m  τm  0878
.  Tm   Tm 
Model: Method 7
Minimum ISE -  Tm 
0.945
Tm  τm 
0.771
τ 
1. 006
τm
1495
.
    0.56Tm  m  . <
01 ≤1
358-363 1101
.  Tm 
Model: Method 3
Minimum ISE -  Tm 
0.970
Tm  τm 
0.753
τ 
0.948
τm
Zhuang and
1473
.
    0.55Tm  m  . ≤
01 ≤ 10
.
Atherton [20] 1115
.  Tm 
 Tm 
0.735
Tm  τ m 
0.641
τ 
0.851
τm
Model: Method 6
1524
.
    0.552Tm  m  . ≤
11 ≤ 2.0
Km  τm   Tm  Tm
1130
.  Tm 
Minimum ITAE -  Tm 
0 .947
Tm  τ m 
0.738
τ 
0.995
τm
1357
.
    0.381Tm  m  . <
01 ≤1
358-363 0842
.  Tm 
Model: Method 3
Minimum ISTSE -  Tm 
0.970
Tm  τm 
0.725
τ 
0.939
τm
Zhuang and
1468
.
    0.443Tm  m  . ≤
01 ≤ 10
.
Atherton [20] 0942
.  Tm 
 Tm 
0.730
Tm  τm 
0.598
τ 
0 .847
τm
Model: Method 6
1515
.
    0.444 Tm  m  . ≤
11 ≤ 2.0
0957
.  Tm 
Minimum ISTES -  Tm 
0.960
Tm  τ m 
0.746
τ 
0.933
τm
Zhuang and
1531
.
    0.413Tm  m  . ≤
01 ≤ 10
.
Atherton [20] 0971
.  Tm 
 Tm 
0.705
Tm  τm 
0.597
τ 
0.850
τm
Model: Method 6
1592
.
    0.414 Tm  m  . ≤
11 ≤ 2.0
0957
.  Tm 
Minimum error - step
load change - Gerry 03
. τm
[96] 0.5τm Tm >5
Km Tm
Model: Method 6
Minimum IAE -  Tm 
0.869
Tm  τm 
0.914
τm
1086
.
  0.348 Tm   . <
01 ≤1
Rovira et al. [21] τ Tm
Model: Method 3 Km  τm  0.740 − 0.13 m  Tm 
Tm
Minimum IAE -  0.6032 
 (Tm + 05
. τm)
0.5Tm τ m τm
 0.7645 +
 τ m Tm  Tm + 0.5τ m 0.05 < <6
Wang et al. [97] Tm + 05. τm Tm
K m ( Tm + τ m )
Model: Method 6
Minimum ISE - Wang  0.7524 
 (Tm + 0 .5τ m )
τm
 0.9155 +
Tm + 0.5τ m 0.05 < <6
et al. [97]  τ m Tm  0.5Tm τ m Tm
K m (Tm + τ m )
Tm + 05. τm
Model: Method 6
 Tm 
0.897
Tm τ 
0.888
τm
1048
.
  0.489 Tm  m  . ≤
01 ≤ 10
.
Minimum ISE - τ Tm
Zhuang and Km  τ m  1195
. − 0.368 m  Tm 
Tm
Atherton [20]
 Tm 
0.567
Tm τ 
0.708
τm
1154
.
  0.490 Tm  m  . ≤
11 ≤ 2.0
τ Tm
Model: Method 6 K m  τm  1.047 − 0.220 m  Tm 
Tm
0.85
Tm τ 
0.929
τm
0965
.
  0.308 Tm  m  . ≤
01 ≤1
Rovira et al. [21] τ Tm
Tm
Modified minimum 0.855 0.929 Damping factor of
.  Tm 
12 Tm τ 
ITAE - Cheng and   τ 0.308 Tm  m  closed loop system =
Hung [95] Km  τm  0.796 − 0.147 m  Tm  0.707.
Tm
Model: Method 7
Minimum ITAE –  05307
. 
 (Tm + 0.5τ m )
τm
 0.7303 +
Tm + 0.5τ m 0.05 < <6
Wang et al. [97]  τ m Tm  0.5Tm τ m Tm
K m (Tm + τm )
Tm + 05. τm
Model: Method 6
 Tm 
0 .897
Tm τ 
0.906
τm
1042
.
  0.385Tm  m  . ≤
01 ≤ 10
.
Minimum ISTSE – τ Tm
Zhuang and Km  τm  0.987 − 0.238 m  Tm 
Tm
Atherton [20]
 Tm 
0.579
Tm τ 
0 .839
τm
1142
.
  0.384 Tm  m  . ≤
11 ≤ 2.0
τm Tm
Model: Method 6 Km  τ m  0.919 − 0.172  Tm 
Tm
 Tm 
0.904
Tm τ 
0.892
τm
0968
.
  0.316 Tm  m  . ≤
01 ≤ 10
.
Minimum ISTES – τ Tm
Tm
Atherton [20]
 Tm 
0 .583
Tm τ 
0.832
τm
1061
.
  0.315Tm  m  . ≤
11 ≤ 2.0
τ Tm
Model: Method 6 Km  τ m  0.892 − 0.165 m  Tm 
Tm
Ultimate cycle
Regulator – minimum τm
IAE – Pessen [63] 0.7K u 0.4Tu 0149
. Tu . ≤
01 ≤ 10
.
Tm
Model: Method 6
Servo – minimum
ISTSE – Zhuang and 0.509Ku 0.051( 3. 302 K m Ku + 1) Tu 0125
. Tu τm
Atherton [20] . ≤
01 ≤ 2.0
Tm
Model: Method not
relevant
Servo – minimum
ISTSE – Pi-Mira et 0.604 K u 0.04 (4. 972 K m K u + 1)T 0.130 Tu
al. [97a]
Model: Method 27
Regulator - minimum
ISTSE - Zhuang and 4.434 K m K u − 0.966 . K m K u − 0.612
1751 0144
. Tu τm
Ku Tu . ≤
01 ≤ 2.0
Atherton [20] . K m K u + 1734
512 . . K m K u + 1388
3776 . Tm
Model: Method not
relevant
1
Model: Method 8
Regulator - nearly Kc( 38) Ti (38 ) 0471
. Ku Decay ratio = 0.15 -
minimum IAE, ISE, K mω u τm
0.1 ≤ ≤ 2.0 ; ε < 2.4
ITAE - Hwang [60] Tm
Model: Method 8 ( 39) ( 39 ) τm
Kc Ti 0471
. Ku
0.1 ≤ ≤ 2.0 ;
K mω u Tm
2.4 ≤ ε < 3
Kc( 40) Ti (40 ) 0471
. Ku τm
0.1 ≤ ≤ 2.0 ;
K mω u Tm
3 ≤ ε < 20
Kc( 41) Ti (41) 0471
. Ku τm
0.1 ≤ ≤ 2.0 ; ε ≥ 20
K mω u Tm
Decay ratio = 0.2 -

Kc ( 42)
Ti ( 42 ) 0471
. Ku τm
0.1 ≤ ≤ 2.0 ; ε < 2.4
K mω u Tm
Decay ratio = 0.2 -

Kc( 43) Ti (43) 0471
. Ku τm
0.1 ≤ ≤ 2.0 ;
K mω u Tm
2.4 ≤ ε < 3
2
 2 10τ m3 16 Tm τ m2  1775 
1
KH =
9  τm − τm Tm + 1884
. Km Kc Ku τ m
+
τ m4 49 τm 2Tm 2 7Tm τm 3 0.471Ku K c
+ + −  + −
. Kc 2 Ku 2 τ m2 
2  
2 K mτ m 18 18 9ωu 324 324 162 ωu  81 81  81ω u 2
 
1+ KH Km 2Tm ω u + K H K m τ m ω u − 1884
. Ku Kc
ωH = ,ε=
τ m Tm K H K m τ m 0.942 K c K u τ m
2
. K c K uω H τ m
0471
+ −
3 6 3ω u
( 38)


= KH −
[
0.674 1 − 0.447ω H τ m + 0.0607ω H τ m
2 2
] , T ( 38)
=
Kc
( 38)
(1 + KH K m )
( )
Kc
 K m (1 + K H K m )  i
ω H Km 0.0607 1 + 105
. ω Hτ m − 0233
. ωH 2τ m2
 
( 39)

=  KH −
[
0.778 1 − 0.467ω H τ m + 0.0609ω H τ m
2 2
] , T ( 39)
=
Kc
( 39)
(1 + K H Km )
( )
Kc
 K m (1 + KH K m )  i
ω H K m 0.0309 1 + 2.84ω H τ m − 0532
. ω H2 τ m 2
 
2

Kc ( 40) =  K H −
. ( 0519
131
ω τ
[
. ) H m 1 − 103
. ε + 0.514 ε 2  ]
, Ti (40) =
K c (1 + K H K m )
( 40)

 K m (1 + K H Km ) 
 ω H K m 00603
. (
1 + 0.929 ln[ ω Hτ m ] 1 + 2.01 ε − 12
. ε2 )( )

Kc ( 41) =  KH −
114 [
. 1 − 0.482ω H τ m + 0.068ω H 2 τ m 2 ] , T ( 41)
=
Kc
( 41)
(1 + KH K m )

K m (1 + K H Km ) 

i
ω H K m 0.0694 − 1 + 21 (
. ω H τ m − 0.367ω H 2 τ m2 )

Kc ( 42) =  K H −
0622
. [
1 − 0.435ω H τ m + 0.052ω H τ m 2 2
]  , T ( 42)
=
Kc
( 42)
(1 + KH K m )

K m (1 + K H K m ) 

i
ω H K m 00697
. 1 + 0752 (
. ω H τ m − 0.145ω H 2 τ m 2 )

Kc ( 43) =  KH −
[
0.724 1 − 0469
. ω H τ m + 00609
. ωH 2 τm2 ] , T ( 43)
=
Kc
( 43)
(1 + K H K m )

K m (1 + K H K m ) 

i
ω H K m 0.0405 1 + 193 (
. ω H τ m − 0.363ω H 2 τ m 2 )
3
Decay ratio = 0.2 -
Regulator – nearly Kc ( 44)
Ti ( 44 ) 0471
. Ku τm
0.1 ≤ ≤ 2.0 ;
minimum IAE, ISE, K mω u Tm
ITAE - Hwang [60] 3 ≤ ε < 20
(continued) Decay ratio = 0.2 -
0471
. Ku
Kc( 45) Ti (45) K mω u τm
Model: Method 8 0.1 ≤ ≤ 2.0 ; ε ≥ 20
Tm
0471
Kc ( 46)
Ti ( 46 )
K mω u τm
0.1 ≤ ≤ 2.0 ; ε < 2.4
Tm
Kc ( 47)
Ti ( 47 ) 0471
. Ku τm
0.1 ≤ ≤ 2.0 ;
K mω u Tm
2.4 ≤ ε < 3
Kc( 48) Ti (48 ) 0471
. Ku τm
0.1 ≤ ≤ 2.0 ;
K mω u Tm
3 ≤ ε < 20
0471
Kc ( 49)
Ti ( 49 )
K mω u τm
0.1 ≤ ≤ 2.0 ; ε ≥ 20
Tm
Servo - nearly min. 0471
IAE, ISE, ITAE - Kc ( 5 0)
Ti ( 50)
K mω u τm
. ≤
01 ≤ 2.0 ; ε < 2.4
Hwang [60] Tm
Model: Method 8
3

Kc ( 44) =  K H −
126
ω τ
[
. ( 0.506) H m 1 − 107
. ε + 0616
. ε 2  ]
, Ti (44) =
K c (1 + K H K m )
( 44)

 K m (1 + K H K m ) 
 ω H K m 00661
. (
1 + 0.824 ln[ ω H τ m ] 1 + 171
. ε − 117
. ε2 )( )
( 45)


=  KH −
109 [
. 1 − 0.497ω H τ m + 0.0724ω H2 τ m 2 ]  , T ( 45)
=
K c (45) (1 + K H K m )
( )
Kc
K m (1 + K H Km )  i
ω H K m 0.054 − 1 + 2.54ω H τ m − 0.457 ω H τ m
2 2
 

Kc ( 46) =  K H −
[
0.584 1 − 0439
. ω H τ m + 00514
. ωH 2 τm2 ] , T ( 46)
=
Kc
( 46)
(1 + K H K m )

K m (1 + K H Km ) 

i
(
ω H K m 0.0714 1 + 0.685ω H τ m − 0131
. ωH 2 τm2 )

Kc ( 47) =  K H −
0675
. [
1 − 0.472ω H τ m + 0061
. ω H 2 τm2 ]  , T ( 47)
=
Kc
( 47)
(1 + K H K m )

K m (1 + KH K m ) 

i
ω H K m 00484
. (
1 + 143
. ω H τ m − 0.273ω H 2 τ m 2 )

Kc ( 48) =  K H −
12
ω τ
[
. ( 0.495) H m 1 − 11
. ε + 0.698 ε 2 ], Ti (48) =
K c (1 + K H K m )
( 48)

 K m (1 + K H K m ) 
 (
ω H K m 0.0702 1 + 0.734 ln[ ω H τ m ] 1 + 148
. ε − 11
. ε2 )( )
( 49)


= KH −
103 [
. 1 − 051
. ω H τ m + 0.0759ω H 2 τ m 2 ]  , T ( 49)
=
Kc
( 49)
(1 + K H K m )
( )
Kc
K m (1 + KH K m )  i
ω H K m 0.0386 −1 + 3.26ω H τ m − 0.6ω H 2 τ m2
 

Kc (50) =  KH −
[
0.822 1 − 0.549ω H τ m + 0112
. ω H2 τ m 2 ]  , T ( 50)
=
Kc
( 50)
(1 + K H K m )

K m (1 + K H K m ) 

i
(
ω H K m 0.0142 1 + 6.96ω H τ m − 177
. ω H 2 τm2 )
4
Decay ratio = 0.1 -
Servo - nearly min. Kc(51) Ti (51) 0471
. Ku τm
0.1 ≤ ≤ 2.0 ;
IAE, ISE, ITAE - K mω u Tm
Hwang [60] 2.4 ≤ ε < 3
(continued) Decay ratio = 0.1 -
Kc(5 2) Ti (52) 0471
. Ku τm
Model: Method 8 . ≤
01 ≤ 2.0 ;
K mω u Tm
3 ≤ ε < 20
0471
Kc(53) Ti (53) K mω u τm
. ≤
01 ≤ 2.0 ; ε ≥ 20
Tm
Servo – nearly    
2 Kc    
2
c + c τ m + c 9  τ m   K u
τm
c1 + c2 τ m + c3  τ m   K u   τ m    7 8 Tm
2 . ≤
01 ≤ 2.0 -
minimum IAE, ITAE  τm  Tm   ω u Kc
Tm  Tm   K uω u c 4 + c 5 + c6      Tm
– Hwang and Fang    Tm  Tm  
  c = 0.350, c = − 00344 . decay ratio = 0.03
[61] c 1 = 0.537, c 2 = −0.0165 7 8
c 4 = 0.0503 , c 5 = 0.163 c 9 = 0.00644
Model: Method 9 c 3 = 0.00173
c 6 = −0.0389
Regulator – nearly  τm  τm  
2 Kc    
2
c + c τ m + c  τ m   K u
τm
minimum IAE, ITAE c 1 + c 2 T + c 3  T   K K uω u c 4 + c 5 τ m + c 6  τ m    7 8 Tm 9  T m   ω u Kc
2 . ≤
01 ≤ 2.0 -
  Tm
– Hwang and Fang  
m m  Tm  Tm  
 
c = 0802
. , c = − 0154
. c 7 = 0.421, c 8 = 0.00915 decay ratio = 0.12
[61] 1 2 c 4 = 0.190 , c 5 = 0.0532
c 3 = 0.0460 c 9 = −000152
.
Model: Method 9 c 6 = −0.00509
Simultaneous  τ τ  
2 Kc    
2
c 1 + c 2 m + c 3  m   K u c 7 + c 8 τ m + c 9  τ m   K u
   2
Servo/regulator -  Tm  Tm   τ  τ  
K u ω u c 4 + c 5 m + c 6  m   
Tm  Tm   ω u K c τm
nearly minimum IAE,
 
 Tm  Tm  
 . ≤
01 ≤ 2.0
c 1 = 0.713, c 2 = −0176 . c 7 = 0.371, c 8 = −00274 . Tm
ITAE - Hwang and c 4 = 0.149, c 5 = 0.0556
c 3 = 0.0513 c 9 = 0.00557
Fang [61] c 6 = −0.00566
Model: Method 9
McMillan [58]  
 
1.415 Tm     T 
0. 65
   T 
0 .65
 Tuning rules
1    
τ m 1 +   0 .25 τ m 1 +  
m m
 0 .65   
Model: Method 1 or K m τm   Tm    + τ m   + τ m 
developed from Ku , Tu
   
Tm
Tm
Method 18 1+   
  Tm + τ m  
Ku Tu Specify gain margin

Astrom and Am arbitrary 4π 2 Ti Am
Hagglund [98]
4 Specify phase margin
tan φ m + + tan 2 φ m
Ku cosφ m αTd α φ m ; α = 4 (Astrom et
Model: Method not Tu
4π al. [30])
relevant
4

Kc (51) =  K H −
0786
. [
1 − 0441
. ω H τ m + 00569
. ω H 2 τm2 ]  , T ( 51)
=
Kc
( 51)
(1 + K H K m )

Km (1 + KH K m ) 

i
(
ω H K m 0.0172 1 + 4.62ω H τ m − 0.823ω H2 τ m 2 )

Kc (52 ) =  K H −
. ( 0.542)
128
ωH τm
[
1 − 0.986 ε + 0.558 ε 2 
, Ti (52) =
] K c (1 + K H K m )
( 52)

 K m (1 + K H K m ) 
 (
ω H K m 0.0476 1 + 0.996 ln[ ω H τ m ] 1 + 2.131 ε − 113
. ε2 )( )

Kc (53) =  KH −
114 [
. 1 − 0.466ω H τ m + 0.0647ω H 2 τ m 2 ]  , T ( 53)
=
Kc
( 53)
(1 + K H K m )

K m (1 + KH K m ) 

i
(
ω H K m 0.0609 − 1 + 197
. ω H τ m − 0.323ω H 2 τ m2 )
T  4
η + η + 
2
Li et al. [99] Kc(5 4) 5 εTd
4π  η 
Model: Method not φm Am η φm Am η φm Am η φm Am η
relevant 150
2 2.8 30 0
1.67 3.8 45 0
1.25 4.6 60 0
1.11 5.4
20 0 1.67 3.2 35 0 1.43 4.0 50 0 1.25 4.9 65 0 1.11 5.5
25 0 1.67 3.5 40 0 1.43 4.2 55 0 1.25 5.2
4h
πAA m 03183
. T 00796
. T simplified algorithm
Ku 4 Am = 2 , φ m = 450 ;
cosφ m tan φ m + + tan 2 φ m
Tan et al. [39] Am αTd α α chosen arbitrarily
Tu
4π
Model: Method 10 Kφ (
rK φ ω u − ω φ
2 2
) 1 Arbitrary A m , φ m at
Am ω φ Ti
2
ωφ ;
ω uωφ K u − r 2 Kφ
2 2 2
r = 01
. + 09
. (K u K φ )
τm
Friman and Waller < 0.25 , Am = 2 ,
0.25 05774
. 01443
. Tm
[41] G p ( jω u ) ωu ωu φ m = 60 0
Model: Method not τm
0.25 ≤ ≤ 2.0
relevant 04830
. 3.7321 0.9330 Tm
(
G p jω 150 0 ) ω150 0 ω 150 0
Am = 2 , φ m = 450
5
Kc =( 54 ) 4h cos φm
 tan φ − b A 2 − b2 

, η= 
m  ( )
with ±h = amplitude of relay, ±b =
 2
πA m  A − b +

2 2πb
T

(Ti − Td )

1 + b A2 − b 2 
  ( )
deadband of relay, A, T = limit cycle amplitude and period, respectively.
Direct synthesis
Pole is real and has
Regulator - Gorecki ( 55) Ti ( 55)
Td ( 5 5) maximum attainable
Kc
et al. [28] τ
multiplicity; m < 2
Tm
Model: Method 6
Low frequency part of
( 56) 6 Ti (56) Td (5 6) magnitude Bode
(2 tuning rules) Kc
diagram is flat.
Regulator - minimum
IAE - Smith and Tm τm
0.5τ m . ≤
01 ≤ 15
.
Corripio [25] – page K mτ m Tm Tm
343-346
Model: Method 6
Servo – minimum 5Tm
IAE – Smith and 6Km τ m Tm 0.5τ m τm
Corripio [25] – page . ≤
01 ≤ 15
.
Tm
343-346
Model: Method 6
Servo – 5% Tm
overshoot – Smith 2K m τ m Tm 0.5τ m
and Corripio [25] –
page 343-346
Model: Method 6
 2
2
2  τ  τ
( 55) 2 Tm  τm  τm   τm   τm   3+  m  − 3− m
 2Tm 
= 6+ 3+  −9−   −
2 Tm
 e
6
Kc
Km τ m  2Tm  2Tm   2Tm   2Tm  
 
2 2
 τm   τm  τ  τ 
6 +  3+  −9− m − m 
 2Tm   2Tm  2Tm  2Tm 
= τm
( 55)
Ti ,
 τm  τm  
2
 τm 
2
τm
2
 τm   τ m 
3
21 + 3 +   3+   − 36 − 4.5 − 6  − 
 Tm  2Tm    2Tm  Tm  2Tm   2Tm 

2
 τ 
3+  m  − 1
 2Tm 
= 0.5τ m
( 55)
Td
2 2
 τm   τm  τ  τ 
6 +  3+  −9− m − m 
 2Tm   2Tm  2Tm  2Tm 
2 3 4
Tm T  T  T 
7 + 42 + 135 m  + 240  m  + 180 m 
1 1 τm τ
 m τ
 m  τm 
Kc ( 56) = , Ti (56 ) = τ m
τ m  Tm    Tm  
2
Km
2 (56)  + 1 − 2  Tm Tm
152 + 1 1 + 3 + 6  
Ti  τ m   τm   τm  τm  

2 3 4
Tm T  T  T 
1+ 7 + 27 m  + 60 m  + 60 m 
τm  τm   τm   τm 
= τm
( 56 )
Td 2 3 4
Tm T  T  T 
7 + 42 + 135 m  + 240  m  + 180 m 
τm  τm   τm   τm 
Suyama [100] 1  Tm  2.236Tm τ m
 0.7236 + 0.2236
Model: Method 6 Km  τm  Tm + 0.309 τ m 7.236Tm + 2.236τ m
2 2 2
τ  τ   τ 
τm τ  τm τ
0.5 m  Tm α + m − 0.5α m 
Juang and Wang α+ + 05
.  m α+ + 05
.  m  Tm   Tm Tm  Closed loop time
Tm  Tm  Tm  Tm 
τ m    τ m   constant = αTm ,
2
[101] Tm  τm
 α +  α + + 0.5  
2  τ   Tm    Tm  Tm  
 τ  α + m  0<α <1
Model: Method 6 Km  α + m   Tm 
 Tm 
0.019952 τ m + 0. 20042Tm 0.099508τ m + 0.99956Tm −0.0069905τ m + 0.029480Tm Closed loop time
τm τm
Km τm 0.99747 τ m − 8 .7425.10 − 5 Tm 0.029773τ m + 0.29907 Tm
Cluett and Wang constant = 4τm
[44] 0.055548 τ m + 0. 33639Tm 0 .16440τ m + 0.99558 Tm −0.016651τ m + 0.093641Tm Closed loop time
τm τm
Km τm 0.98607τ m − 15032
. .10 − 4 T m 0.093905τm + 0.56867 Tm
constant = 2τm
Model: Method 6
0.092654 τ m + 0.43620Tm 0.20926 τ m + 0 .98518Tm −0.024442τ m + 0.17669Tm Closed loop time
τm τm
0.96515τ m + 4255010 . − 3 Tm 0.17150τm + 080740
Km τm . Tm
. τm
constant = 133
012786
. τ m + 0.51235Tm 0.24145τ m + 0 .96751Tm −0.030407 τ m + 0 .27480Tm Closed loop time
τm τm
. − 2Tm
0.93566 τ m + 2 .298810 0.25285 τ m + 1.0132Tm
K mτ m constant = τ m
016051
. τ m + 057109
. Tm 0 .26502τ m + 0 .94291Tm −0.035204τ m + 0.38823Tm Closed loop time
τm τm
0.89868τ m + 6.935510. − 2 Tm 0.33303 τm + 11849
K mτ m . Tm
constant = 08. τm
019067
. τ m + 0.61593Tm 0.28242τ m + 0.91231T m −0.039589τ m + 0.51941Tm Closed loop time
τm τm
0.85491τ m + 0.15937 Tm 0.40950τ m + 13228
K mτ m . Tm
. τm
constant = 067
Gain and phase mK u cos(φ m ), tan(φ m ) +
4
+ tan2 (φ m ) tan(φm ) +
4
+ tan2 ( φm )
Gain margin = 2, phase
margin - Zhuang and −0 .347 K m K u α α margin = 60 0
m = 0 .614(1 − 0.233 e ) α ,
2ω u 2ω u
Atherton [20] τ
Model: Method not
φ m = 338
0
(
. 1 − 0.97e
− 0. 45K m Ku
) α = 0.413( 3.302K mK u + 1) 01. ≤ m ≤ 2.0
Tm
relevant
− 1.002
τ  V = fractional
Abbas [45] 0177
. + 0.348 m  Tm + 05
. τm Tm τ m overshoot
 Tm 
2Tm + τ m 0 ≤ V ≤ 0.2
K m ( 0531
. − 0.359V 0.713 )
Model: Method 6 τ
. ≤ m ≤ 5.0
01
Tm
Camacho et al.[102] 1 Tm + τ m 4Tm τ m Tm τ m
Model: Method 6 Km Tm τm Tm + τ m Tm + τ m
Am ∈[ 2 ,5] ,
( 57)
[ ]
− 0.9471 1.0264
Servo – minimum ISE 18578
. φm
0.0821
 τm  Ti 7
0.4899Tm φ m 0.1457  τ m 
- Ho et al. [103]     φ m ∈ 30 0 ,60 0 ,
K A 0.9087 m m  Tm  Am
0.0845
 Tm 
τm
01 ≤ 10
. .
Tm
621189
. 403182
. τ m 76.2833τ m
Given A m , ISE is minimised when φ m = 29.7985 + + −
Am Tm A m Tm
(Ho et al [104])
Am ∈[ 2 ,5] ,
− 0.908
[ ]
0.3678 1.0317
Regulator - minimum 10722
. φ m −0.116  τ m  12497
. Tm φ m 1.0082  τ m  0.4763Tm φ m −0.328  τ m 

ISE – Ho et al. [103] Km A m 0.8432  Tm       φ m ∈ 30 0 ,60 0 ,
Am
0.2099
 Tm  Am
0.0961
 Tm 
τm
01 ≤ 10
.
Tm
Given A m , ISE is minimised when φ m = 46.5489A m0.2035 ( τ m Tm )
0.3693
(Ho et al [104])
7
Ti
( 57 )
=
[
0.0211Tm 1 + 0.3289 A m + 6.4572φ m + 251914
. ( τ m Tm ) ]
1 + 0.0625A m − 0.8079φ m + 0.347( τm Tm )
Morilla et al. [104a] Kc
( 57a ) 8
τm α Ti
( 57 a ) α = 0. 1; δ 0 = [ 0. 2, 0. 5]
Model: Method 24 1 + 1 − 4α
Robust
Robust - Brambilla et τm
Tm + 0.5τ m Tm τ m . ≤
01 ≤ 10 and no
al. [48] 1  Tm + 05
. τm  Tm
  2Tm + τ m
Km  λτ m  model uncertainty -
Model: Method 6
λ ≈ 0.35
Rivera et al. [49] 1  Tm + 05
. τm  Tm τ m λ > 01
. Tm ,
  Tm + 0.5τ m 2Tm + τ m
Km  λ + 05
. τm  λ ≥ 0.8τ m .
Model: Method 6
5Tm τm
≤ 0.5τ m < 033
.
Fruehauf et al. [52] 9τ m Km 5τ m Tm
Tm τm
Model: Method 1
Tm ≤ 0.5τ m ≥ 0.33
2τ m Km Tm
Ti τm
2
τm
2
 τ  τm
Tm + 1 − m  λ=
Lee et al. [55] K m ( λ + τm ) 2( λ + τ m ) 2( λ + τ m )  3Ti  3
τm ατ
3 2
Model: Method 6 Two degrees of
Tm + α − − m freedom controller;
Ti Tmα − 6 2
α = T −
λ + ατ m − 05
2
. τm
2
2λ + τ m − α m
K m ( 2λ + τm − α ) -

2
λ  −T ;
τ m
2λ + τ m − α Ti Tm 1 −  e m
 Tm 
λ2 + ατ m − 05
. τm
2
Desired response =
2λ + τ m − α
e− τ m s
1 + λs
Ti
( 57 a )
ωn0 − δ 0 Tm + δ 0 Tm + Tm τ m − Ti
2 2
( ( 57a )
)− αT ( 57a ) 2
)] , ω =
i
=
(
8 ( 57a )
( )− αT
Kc n0
( 57 a ) 2
K m [ 2δ 0 + ω n 0 τ m − Ti
( 57 a )
Tm τ m − Ti
( 57 a )
i
1 b
with δ 0 = , = desired closed loop response decay ratio
2 a
 2π 
1 +  
 log e [b a ] 
K m e− sτ m
Table 26: PID tuning rules - FOLPD model - ideal controller with first order filter
1 + sTm
 1  1
G c ( s) = Kc  1 + + Td s . 3 tuning rules.
 Ti s  Tf s + 1
Robust
1  Tm + 05
. τm  λτ m
  TF = ,
** Morari and
K m  λ + τm 
Tm + 0.5τ m Tm τ m 2( Tm + τ m )
Zafiriou [105] 2Tm + τ m λ > 0.25τ m ,
λ > 0.2Tm .
Horn et al. [106] λτ m
Tf = ;
2Tm + τ m Tm + 0.5τ m τm Tm 2( λ + τ m )
Model: Method 6 2 (λ + τ m ) K m τ m + Tm λ > τ m , λ < Tm .
τm
Tf =
H∞ optimal – Tan et 0.265λ + 0.307  Tm  Tm + 0.5τ m τ m Tm 5.314 λ + 0.951
 + 05
. 
al. [81] Km  τm  τ m + 2Tm λ = 2 - ‘fast’
response
Model: Method 6 λ = 1 - ‘robust’ tuning
λ = 1.5 - recommended
K m e− sτ
m
Table 27: PID tuning rules - FOLPD model - ideal controller with second order filter
1 + sTm
 1  1 + b1s
G c ( s) = Kc  1 + + Td s . 1 tuning rule.
  1 + a 1s + a 2s
2
Ti s
Robust
2Tm + τ m Filter
,
2( 2λ + τm − b1) Km Tm + 05
. τm τ m Tm 1 + b1 s
Horn et al. [106]
2λτ m + 2 λ + b1 τ m
2
1+ s + a 2s 2
[
λ τ m + 2 Tm τ m ( τ m − λ )
2
] τ m + 2Tm 2( 2λ + τ m − b 1 )
Model: Method 6 b1 =
Tm (τ m + 2 λ) a2 =
λ 2τ m
; λ > τm ,
2( 2λ + τ m − b1 )
+
[
2 λ (2T m − λ) ] λ < Tm .
( τ m + 2λ)
K m e− sτ m
Table 28: PID tuning rules - FOLPD model - ideal controller with set-point weighting
1 + sTm
 1 
G c ( s) = Kc  b + + Td s . 1 tuning rule.
 Ti s 
−8.4 τ + 7.3τ 2 − 2.5τ − 1.4 τ 2
. τ me 0.89 τm e −0.37 τ − 4.1τ or b = 0.40e 0.18 τ + 2 .8 τ ;
2 2
38
. e Tm 52 or
,
Astrom and
K mτ m M s = 1.4 ;
0.46Tme 2.8τ − 2.1τ 0.077 Tm e5.0τ − 4 .8 τ
2 2
Hagglund [3] –
pages 208-210 τ = τ m ( τ m + Tm ) . ≤
014
τm
≤ 55
.
Tm
Model: Method 3 b = 0.22e0.65τ + 0.051τ ;
2
8.4e− 9.6 τ + 9.8τ Tm . τ m e− 1.5τ − 0.93τ or 0.86τ m e− 1.9 τ − 0.44 τ or

2
Ms = 2.0 ;
2 2
32
Km τ m 0.28Tm e3.8τ − 1.6 τ
2
0.076 Tm e3.4 τ − 1.1τ
2
τm
. ≤
014 ≤ 55
.
Tm
K m e− sτ
m
Table 29: PID tuning rules - FOLPD model - ideal controller with first order filter and set-point
1 + sTm
 1  1  1 + 0.4Trs 
weighting U(s) = K c 1 + + Td s  R (s ) − Y( s)  . 1 tuning rule.
 Tis  Tf s + 1  1 + sTr 
Direct synthesis
Normey-Rico et al.
[106a] 0. 375(τ m + 2Tm ) Tm + 0. 5τ m Tm τ m Tf = 0.13τ m
Model: Method not K m τm 2Tm + τ m Tr = 0.5τ m
specified
 
K m e− sτ m
 1   1 + Td s 
Table 30: PID tuning rules - FOLPD model - classical controller G c ( s) = Kc  1 +   . 20.
1 + sTm  Ti s   1 + Td s 
 
 N 
Process reaction
Hang et al. [36] – 083
. Tm Foxboro EXACT
page 76 Km τ m . τm
15 0.25τ m controller ‘pretune’;
Model: Method 1 N=10
Witt and Waggoner Equivalent to Ziegler
[107] 0.6Tm Tm τm τm and Nichols [8];
N = [10,20]
[ , ]
Model: Method 1 K mτ m K mτ m
Witt and Waggoner Equivalent to Cohen
[107] Kc ( 5 8) 1
Ti ( 58)
Td ( 5 8) and Coon [11];
Model: Method 2 N = [10,20]
Tm
St. Clair [15] – page Km τ m 5τ m 0.5τ m ‘aggressive’ tuning;
21
Model: Method 1 05
. Tm
K mτ m 5τ m 0.5τ m ‘conservative’ tuning;
Shinskey [15a] 0. 889Tm τm
1.75τ m 0.70 τm = 0.167
Model: Method 1 K m τm Tm
Minimum IAE - Kaya  Tm 
0.76167
Tm  τm 
1.05221
τ 
0.89819
τm
and Scheib [108]
098089
.
    0.59974 Tm  m  0< ≤ 1 ; N=10
.  Tm 
Minimum IAE – Witt
and Waggoner [107] τm
Kc(5 9) 2 Ti (59) Td (59 ) 0.1 < < 0.258 ; N = [10,20]
Model: Method 1 Tm
2
T T τ τ 
1.350 m + 0.25 ± m 0.7425 + 0.0150 m + 0.0625 m 
τm τm Tm  Tm 
=
1 ( 58 )
Kc
2 Km
Tm
Ti (58 ) =
2
Tm τ τ 
1350
. + 0.25 m 0.7425 + 0.0150 m + 0.0625 m 
τm Tm  Tm 
Tm
Td (58 ) =
2
Tm τ τ 
1350
. + 025
. ± 07425
. + 0.0150 m + 0.0625 m 
τm Tm  Tm 
−0.921  − 1.886 
0718
.  τm  1 ± 1 − 1693 τm  ,
=   .  
2 ( 59 )
Kc
Km  Tm    Tm  
 
1.137 1.137
τ  τ 
0.964 Tm  m  0.964 Tm  m 
 Tm   Tm 
= =
( 59 ) ( 59 )
Ti , Td
1.886 1.886
τ  τ 
1 − 1 m 1693
.  m 1 + 1 ± 1.693 m 
 Tm   Tm 
Minimum ISE - Kaya  Tm 
0.89711
Tm  τ m 
0.9548
τ 
0.87798
τm
and Scheib [108]
111907
.
    0.54766 Tm  m  0< ≤ 1 ; N=10
.  Tm 
1.06401
Tm  τ m 
0.70949
τ 
1.03826
τm
Kaya and Scheib
077902
.
    0.57137 Tm  m  0< ≤ 1 ; N=10
K m  τm   Tm  Tm
[108] 114311
.  Tm 
Model: Method 3
Minimum ITAE -
τm
Witt and Waggoner Kc( 60) Ti( 60) 3 Td ( 60) 0 .1 < < 0.379 ; N = [10,20]
Tm
[107]
Model: Method 1
Minimum IAE - Kaya .  Tm 
1.04432
Tm  τm 
1.08433
τm
065
  0< ≤ 1 ; N=10
and Scheib [108] τ 0.50814 Tm  Tm
Model: Method 3 Km  τm  0.9895 + 0.09539 m  Tm 
Tm
Minimum IAE - Witt
τm
and Waggoner [107] Kc( 61) Ti (61) Td ( 61) 0 .1 ≤ ≤ 1 ; N = [10,20]
Tm
Model: Method 1
1.03092
Tm τ 
0.86411
τm
071959
.
  0.54568 Tm  m  0< ≤ 1 ; N=10
and Scheib [108] τ Tm
Model: Method 3 Km  τ m  112666
. − 0.18145 m  Tm 
Tm
0.80368
Tm τ 
1.0081
τm
112762
.
  0.42844Tm  m  0< ≤ 1 ; N=10
Kaya and Scheib τ Tm
[108] Km  τ m  0.99783 + 0.02860 m  Tm 
Tm
Model: Method 3
0.995
τ 
0.947  −1.733  0.762Tm  m 
0.679  τ m  1 ± 1 − 1.283 τ m   T ( 60) =  Tm 
=
( 60)
 
3
Kc ,
K m  Tm    Tm   i 1.733
   τ 
1 − 1 m 1283
.  m
 Tm 
0.995
τ 
0.762Tm  m 
 Tm 
=
( 60 )
Td .
1.733
τ 
1 + 1 ± 1.283 m 
 Tm 
− 0.869  0.914 
1086
.  τm  1 ± 1 − 1392  τm   τm 
=   .   . − 013
( 61)
Kc
 074 .  ,
Km  Tm   Tm   Tm  
 
0.914 0.914
τ  τ 
0.696 Tm  m  0.696Tm  m 
 Tm   Tm 
= =
( 61) ( 61)
Ti , Td
0.869 0.869
τ   τm  τ   τm 
1 m 1 − 1.392 m   0.74 − 0.13  1 ± 1 − 1.392  m  0.74 − 0.13 
 Tm   Tm   Tm   Tm 
Minimum ITAE -
τm
Witt and Waggoner Kc( 62) 4 Ti (62 ) Td ( 62) 0 .1 ≤ ≤ 1 ; N = [10,20]
Tm
[107]
Model: Method 1
Direct synthesis
Tsang and Rad [109] 0809
. Tm Tm . τm
05 Overshoot = 16%; N=5
Model: Method 13 Km τ m
aTm
Tsang et al. [111] Km τ m Tm . τm
025 N = 2.5
a ξ a ξ a ξ a ξ a ξ a ξ
Model: Method 6
1.681 0.0 1.161 0.2 0.859 0.4 0.669 0.6 0.542 0.8 0.456 1.0
8 0 4 3 9 9
1.382 0.1 0.991 0.3 0.754 0.5 0.600 0.7 0.495 0.9
9 6 2 0 7
Robust
1  Tm 
Chien [50]   Tm . τm
05 λ = [ τ m , Tm ] ; N=10
K m  λ + 05
. τm 
Model: Method 6 1  05 . τm 
  . τm
05 Tm λ = [ τ m , Tm ] ; N=10
K m  λ + 05
. τm 
Ultimate cycle
Minimum IAE
regulator – Shinskey K mτ m  Tu 
Tu  015
. − 0.05
[59] – page 167. 3τ m − 0.32 Tu  τm 
014
. Tu
Model: Method not
specified
. Tm Km τ m
095 1.43τ m . τm
052 τ m Tm = 0.2
Minimum IAE . Tm Km τ m
095 . τm
117 . τm
048 τ m Tm = 0.5
regulator - Shinskey
[16] – page 143. . Tm Km τ m
114 1.03τ m 0.40τ m τ m Tm = 1
139 . τm
077 0.35τ m τ m Tm = 2
− 0.85  0.929 
0.965  τ m  1 ± 1 − 1232  τm   τm 
=   .   0.796 − 01465
( 62)
 ,
4
Kc .
K m  Tm    Tm   Tm  
 
0.929 0.929
τ  τ 
0.616 Tm  m  0.616Tm  m 
 Tm   Tm 
= =
( 62) ( 62)
Ti , Td
0.85 0.85
τ   τm  τ   τm 
1 m 1 − 1.232 m   0.796 − 0.1465  1 ± 1 − 1.232 m  0.796 − 0.1465 
 Tm   Tm   Tm   Tm 
K m e− sτ
m
Table 31: PID tuning rules - FOLPD model - non-interacting controller

1 + sTm
 
  
1  Td s  . 2 tuning rules.
U (s) = K c 1 +  E( s) − Y (s )
 Tis  Td s 
 1+ 
 N 
Minimum IAE – τm
Huang et al. [18] . ≤
01 ≤ 1 ; N =10
Kc( 63) 5 Ti (63) Td ( 63) Tm1
Model: Method 6
Minimum IAE - τm
Huang et al. [18] . ≤
01 ≤ 1 ; N =10
Kc( 64) 6 Ti (64 ) Td ( 64) Tm
Model: Method 6
1  
−0.8058 0.6642 2.1482 τ
τ  τm   τm   τm  m
5
Kc ( 63) = 01098
. − 8.6290 m + 11863
.   + 231098
.   + 20.3519   − 191463
. e T m

Km  Tm  Tm   Tm   Tm  

 τ τ 
2
τ 
3
τ 
4
τ  
5
= Tm − 0.0145 + 2 .0555 m − 4.4805 m  + 7 .7916 m  − 7 .0263 m  + 2.4311 m  

( 63)
Ti
 Tm  Tm   Tm   Tm   Tm  
Tm   τm  
2 3 4 5 6
τm  τm   τm   τm   τm 
Td
( 63)
=  −0.0206 + 0.9385 − 2 .3820  + 7.2774   − 111018
.   + 8.0849   − 2.274  
 Tm   Tm   Tm   Tm   Tm  
( 63)
K c  Tm
1  
0.0865 −0.4062 2 .6405 τ
τ  τm  τ   τm  m
6
Kc ( 64) = 7.0636 + 66.6512 m + 261928
.   + 7.3453 m  + 336578
.   − 57.937e T m

Km  Tm  Tm   Tm   Tm  

 τ  τm 
2
τ 
3
 τm 
4
τ  
5
= Tm  0.9923 + 0.2819 m − 14510  + 2.504 m  − 18759  + 0.5862 m  

( 64 )
Ti . .
 Tm  Tm   Tm   Tm   Tm  
Tm   τm  
2 3 4 5 6
τm  τm   τm   τm   τm 
Td
( 64 )

= ( 64) 0.0075 + 0.3449 − 0.0793  + 0.8089   − 1.0884  + 0.352  + 0.0471  
K c  Tm  Tm   Tm   Tm   Tm   Tm  
K m e− sτ
m

1 + sTm
 1  Tds
U (s) = K c 1 + E (s ) −
 Y (s) . 5 tuning rules.
 Tis  sT
1+ d
N
 Tm 
0.887
Tm  τm 
0.886
τm
1260
.
    . ≤
01 ≤ 10
. ; N=10
Minimum ISE - τ 0.375T Tm
Km  τm   Tm 
m
Zhuang and 0.701 − 0.147 m
Tm
Atherton [20]
 Tm 
0.619
Tm τ 
0.756
τm
1295
.
  0.378 Tm  m  . ≤
11 ≤ 2.0 ; N=10
τ Tm
Tm
 Tm 
0.930
Tm τ 
0 .907
τm
1053
.
  0.349 Tm  m  . ≤
01 ≤ 10
. ; N=10
Minimum ISTSE - τm Tm
Zhuang and Km  τ m  0.736 − 0.126  Tm 
Tm
Atherton [20]
 Tm 
0.625
Tm τ 
0.811
τm
1120
.
  0.350 Tm  m  . ≤
11 ≤ 2.0 ; N=10
τ Tm
Tm
 Tm 
0.933
Tm τ 
0.897
τm
0942
.
  0.308 Tm  m  . ≤
01 ≤ 10
. ; N=10
Minimum ISTES - τ Tm
Tm
Atherton [20]
 Tm 
0 .624
Tm τ 
0.813
τm
1001
.
  0.308 Tm  m  . ≤
11 ≤ 2.0 ; N=10
τ Tm
Tm
Ultimate cycle
Servo – minimum
ISTSE – Zhuang and 4.437K m K u − 1587
. 0.037(589
. K m K u + 1)Tu 0112
. Tu τm
Ku . ≤
01 ≤ 2.0 ; N=10
Atherton [20] 8.024K m Ku − 1435
. Tm
Model: Method not
relevant
Regulator - minimum 0.5556Ku 039
. Tu 014
. Tu τ m Tm = 0.2
IAE - Shinskey [16] – 0.4926Ku 0.34Tu 014
. Tu τ m Tm = 0.5
page 148.
05051
. Ku 0.33Tu 013
. Tu τ m Tm = 1
Model: Method 6
0.4608Ku 0.28Tu 013
. Tu τ m Tm = 2
K m e− sτ m

1 + sTm
 1  Td s
U (s) =  K c + E (s) −
 Y ( s) . 6 tuning rules.
 Ti s  sT
1+ d
N
0.8826
Tm  τm 
1.3756
τ 
0.4576
τm
and Scheib [108]
131509
.
    0.5655Tm  m  0< ≤ 1 ; N=10
Km  τm  Tm
.  Tm   Tm 
0.9308
Tm  τ m 
1.25738
τ 
0.41941
τm
and Scheib [108]
13466
.
    0.79715Tm  m  0< ≤ 1 ; N=10
Km  τm  Tm
.  Tm   Tm 
0.7937
Tm  τ m 
1.42603
τ 
0.41932
τm
Kaya and Scheib
13176
.
    0.49547Tm  m  0< ≤ 1 ; N=10
Km  τm  Tm
[108] 112499
.  Tm   Tm 
Model: Method 3
0.81314
Tm  τm 
0.17707
τm
113031
.
  0.32175Tm   0< ≤ 1 ; N=10
Tm
0.8388
Tm τ 
0.24572
τm
126239
.
  0.47617Tm  m  0< ≤ 1 ; N=10
Tm
0.49851 Tm τ 
0.16768
τm
Kaya and Scheib
098384
.
  τ 0.21443Tm  m  0< ≤ 1 ; N=10
Km  τ m  2.71348 − 2.29778 m  Tm 
Tm
[108] Tm
Model: Method 3
K m e− sτ
m
Table 32: PID tuning rules - FOLPD model - non-interacting controller with set-point weighting
1 + sTm
 1 
 E( s) − c d Y( s) + Kc ( b − 1)Y(s) . 3 tuning rules.
K Ts
U( s) = Kc  b +
 Ti s  sT
1+ d
N
Ultimate cycle
τm
Hang and Astrom < 0.3 .
Tm
[111] 0.6Ku 0.5Tu 0125
. Tu
. τm
166
b = 2( x − 01
. )+ ,
Tm
N=10
x = overshoot.
Model: Method 1 τ
0.3 ≤ m < 0.6 .
Tm
0.6Ku 0.5Tu 0125
. Tu
τm
b = 2x +
Tm
τm
0.6 ≤ < 0.8 .
 τm  Tm
0.6Ku . −.083
0.5 15 .  Tu 0125
. Tu
 Tm  τm
b = 16
. −
Tm
 τm  τm
0.6Ku . −.083
0.5 15 .  Tu 0125
. Tu . ≤
08 < 10
. ; b=0.8
 Tm  Tm
τm
0.6Ku 0.335Tu 0125
. Tu . <
10 ; b=0.8
Tm
0.6Ku 0.5Tu 0125
. Tu τm
Hang et al. [65] . ≤
016 < 0.57 ;
15 − K '
36 K ='
Tm
b= , b= ,
15 + K 27 + 5K '  11[τ m Tm ] + 13
'
N=10
Model: Method 1
2 
 37[ τ m Tm ] − 4 
10% overshoot - servo 20% overshoot - servo
0.6Ku 0.222 K ' Tu τm

0125
. Tu 0.57 ≤ < 0.96 ;
8 4 '  K = ' Tm
b=  K + 1 ,
17  9   11[τ m Tm ] + 13 N=10
2 
 37[ τ m Tm ] − 4 
20% overshoot,
10% undershoot;
servo
Hang and Cao [112]  τ   τ T τm
Model: Method 11 0.6Ku . − 0.22 m  Tu
 053 . − 0.22 m  u
 053 . ≤
01 < 0.5 ; N=10
 Tm   Tm  4 Tm
K m e− sτ
m
Table 33: PID tuning rules - FOLPD model - industrial controller

1 + sTm
 
 1  1 + Td s 
U( s) = Kc 1 +   R( s) − Y(s)  . 6 tuning rules.
 Ti s   1+ d
Ts 

 N 
Minimum IAE - Kaya .  Tm 
0 .7938
Tm  τ m 
1.00403
τ 
0.7848
τm
and Scheib [108]
091
    0.5414Tm  m  0< ≤ 1 ; N=10
Km  τm  Tm
.  Tm   Tm 
0.8992
Tm  τ m 
0.8753
τ 
0.91107
τm
and Scheib [108]
11147
.
    0.56508 Tm  m  0< ≤ 1 ; N=10
Km  τm  Tm
.  Tm   Tm 
0.8872
Tm  τ m 
0.99138
τ 
0.971
τm
Kaya and Scheib
07058
.
    0.60006Tm  m  0< ≤ 1 ; N=10
Km  τm  Tm
[108] 103326
.  Tm   Tm 
Model: Method 3
1.004
Tm  τm 
0.97186
τm
081699
.
  0< ≤ 1 ; N=10
and Scheib [108] τ 0.44278Tm   Tm
Tm
0.9365
Tm τ 
0.78088
τm
11427
.
  0.35308Tm  m  0< ≤ 1 ; N=10
Tm
0.7607
Tm τ 
1.11499
τm
08326
.
  0.44243Tm  m  0< ≤ 1 ; N=10
Kaya and Scheib τ Tm
[108] Km  τm  1.00268 + 0.00854 m  Tm 
Tm
Model: Method 3
K m e− sτ  1 
 (1 + sTd ) . 3 tuning
m
Table 34: PID tuning rules - FOLPD model - series controller Gc ( s) = Kc  1 +

1 + sTm  Ti s 
rules.
Autotuning **
Astrom and
Hagglund [3] – page 5Tm . τm
15 0.25τ m Foxboro EXACT
246 6Km τ m controller
Model: Not specified
Ultimate cycle
Pessen [63] 035
. Ku 0.25 0.25 τm
01 ≤1
Tu Tu Tm
Direct synthesis
aTm
Tsang et al. [110] Km τ m Tm . τm
025
Model: Method 13
1.819 0.0 1.269 0.2 0.949 0.4 0.748 0.6 0.617 0.8 0.541 1.0
4 0 2 2 0 3
1.503 0.1 1.089 0.3 0.837 0.5 0.675 0.7 0.570 0.9
9 4 8 6 9
K m e− sτm
Table 35: PID tuning rules – FOLPD model - series controller with filtered derivative
1 + sTm
 
 1  sTd 
Gc ( s) = Kc  1 +   1 +  . 1 tuning rule.
 Ti s  sT
1+ d 

 N 
Robust
Tm
Chien [50] K m (λ + 0.5τm ) Tm . τm
05 λ = [ τ m , Tm ] ; N=10
Model: Method 6 0.5τ m

K m ( λ + 0.5τm ) . τm
05 Tm λ = [ τ m , Tm ] ; N=10
K m e− sτ m
Table 36: PID tuning rules – FOLPD model - controller with filtered derivative
1 + sTm
 
 1 Td s 
G c ( s) = Kc  1 + +  . 3 tuning rules.
 Ti s 1 + s Td 
 N 
Robust
Chien [50] Tm + 0.5τ m Tm τ m
K m ( λ + 0.5τm ) Tm + 05
. τm 2Tm + τ m λ = [ τ m , Tm ] ; N=10
Model: Method 6
Tm + 0.3866τ m 03866
. Tm τm
Gong et al. [113] K m (λ + 1.0009τm ) Tm + 0.3866τ m Tm + 0.3866τ m N = [3,10]
Model: Method 6
λ=
( 0.1388 + 0.1247N )Tm + 0.0482 Nτ m τ
0.3866 ( N − 1)Tm + 0.1495 Nτ m
m
Direct synthesis
1  τm   τm  Closed loop response
 0186
. + 0532
.  Tm 0.25 0186
. + 0532
.  Tm
Davydov et al. [31]  τm   T   T  damping factor =
Km  1552
. + 0078
.  m m
 Tm  . ≤ τ m Tm ≤ 1 ;
0.9; 02
Model: Method 12 N = Km
1  τ   τ  Closed loop response
 0.382 m + 0.338 Tm 0.4 0.382 m + 0.338 Tm
 τm   Tm   Tm  damping factor =
K m  1209
. + 0103
. 
 Tm  . ≤ τ m Tm ≤ 1 ;
0.9; 02
N = Km
K m e− sτ m
Table 37: PID tuning rules – FOLPD model - alternative non-interacting controller 1 -
1 + sTm
 1 
U( s) = Kc 1 +  E( s) − Kc Td sY ( s) . 6 tuning rules.
 Ti s 
Ultimate cycle
Ku 2 Tu
Tu < 2 .7
Regulator - minimum Tu 0125
. 012
. Tu τm
3.73 − 0.69 τm
IAE - Shinskey [59] – τm
page 167.
Ku 2 Tu
Model: method not Tu ≥ 2.7
T 0125
. 012
. Tu τm
specified 2.62 − 0.35 u τm
τm
Minimum IAE -
Shinskey [17] – page . Tm K m τ m
132 . τm
180 0.44τm τ m Tm = 01
.
117. Model: Method
6
. Tm K m τ m
132 1.77τm 0.41τ m τ m Tm = 0.2
Minimum IAE - . Tm Km τ m
135 1.43τ m 0.41τ m τ m Tm = 0.5
Shinskey [16] – page
143. . Tm K m τ m
149 . τm
117 0.37τm τ m Tm = 1
182 0.92τ m 0.32τ m τ m Tm = 2
Regulator - minimum 0.7692 Ku 0.48Tu 011

. Tu τ m Tm = 0.2
IAE - Shinskey [16] – 0.6993Ku 0.42Tu 012
. Tu τ m Tm = 0.5
page 148.
0.6623Ku 0.38Tu 012T
. u τ m Tm = 1
Model: Method 6
0.6024K u 0.34Tu 012
. Tu τ m Tm = 2
Regulator – minimum
IAE - Shinskey [17] –
page 121. Model: 0.7576 Ku 0.48Tu 011
. Tu τ m Tm = 0.2
method not specified
Process reaction – 15
. Tm
VanDoren [114] Km τ m . τm
25 . τm
04
Model: Method 1
K m e− sτ m
Table 38: PID tuning rules – FOLPD model - Alternative filtered derivative controller -
1 + sTm
 1   1 + 05
. τm s + 0.0833τ m s 2 
2
G c ( s) = Kc  1 +  . 1 tuning rule.
 Ti s   [1 + 01. τ ms]
2 

Direct synthesis
aTm
Tsang et al. [110] Km τ m Tm 025
. Tm
Model: Method 13
1.851 0.0 1.329 0.2 1.028 0.4 0.841 0.6 0.695 0.8 0.552 1.0
2 3 0 1 3 7
1.552 0.1 1.159 0.3 0.924 0.5 0.768 0.7 0.621 0.9
0 5 6 0 9
K m e− sτ m
- I-PD controller U(s) = c E (s) − K c (1 + Td s)Y(s) . 2 tuning

K
Table 39: PID tuning rules – FOLPD model
1 + sTm Ti s
rules.
Direct synthesis
Chien et al. [74] Underdamped system
K c ( 64a )
1
Ti ( 64a ) Td ( 64 a ) response - ξ = 0.707 .
Model: Method 6 τ m ≤ 0.2Tm
Minimum ISE – Tm τ m First order Pade
Argelaguet et al. 2Tm + τ m Tm + 0. 5τ m 2Tm + τ m approximation for τm
[114a]. Model: 2K m τ m
Method not defined
1.414TCLTm + τ m Tm + 0.25τ m 2 − TCL2 . TCL Tm + τ mTm + 0.25τ m2 − TCL2

1414
1
Kc (64a ) = , Ti (64 a ) =
(
Km TCL2 + 0.707TCL τ m + 0.25τ m 2
) Tm + 05
. τm
. TmTC Lτ m + 0.25Tm τ m2 − 05
0707 . τ m TCL2
Td (64 a ) =
Tm τ m + 025
. τ m + 1414
2
. TCLTm − TC L2
K me − sτ m
Table 40: PID tuning rules - FOLPD model - G m ( s) = – Two degree of freedom controller:
1 + sTm
   
   
1 Td s β T s
U (s) = K c 1 + +  E( s) − K c  α + d  R (s) . 1 tuning rule.
 Tis T   Td 
 1+ d s  1+ s
 N   N 
Servo/regulator Minimum performance index
tuning
τm
Taguchi and Araki ≤ 1.0 ;
Tm
[61a] ( 64b ) 2 ( 64 b) ( 64b )
Kc Ti Td Overshoot (servo step)
≤ settling time of
tuning rules of Chien
et al. [10]
 
 
 2 3

2 ( 64 b)
=
1 
0.1415 +
1. 224  ( 64 b)
=  0.01353 + 2.200 τm − 1. 452  τm  + 0. 4824  τm  
Kc , T T     
Kc  τm  i m
Tm  Tm   Tm  
 − 0.001582  
 Tm 
 τm  τm  
2
τm τm 
2
Td
( 64 b) 
= Tm 0.0002783 + 0. 4119 − 0. 04943   , α = 0. 6656 − 0. 2786 + 0. 03966   ,
   Tm 

Tm  Tm   Tm
2
τm τ 
β = 0. 6816 − 0.2054 + 0.03936  m 
Tm Tm 
 1 
Table 41: PID tuning rules - non-model specific – ideal controller G c ( s) = Kc  1 + + Td s . 25 tuning rules.
 Ti s 
Ultimate cycle
Ziegler and Nichols [ 0.6Ku , Ku ] 0.5Tu 0125
[8]
Blickley [115] 0.5K u Tu [ 0125 . Tu ]
. Tu , 0167 Quarter decay ratio
0.5K u Tu 0.2Tu Overshoot to servo
Parr [64] – pages response ≈ 20%
190-191 05
. Ku 034
. Tu 008
De Paor [116] 0.866Ku 0.5Tu 0125

. Tu phase margin = 30 0
Corripio [117] – page 0. 75Ku 0. 63Tu 01
27
Mantz andTacconi 0.906K u 0.5Tu 0125
[118]
0.4698 K u 0.4546Tu 01136
. Tu Gain margin = 2, phase
margin = 20 0
Astrom and Gain margin = 2.44,
Hagglund [3] – page 01988
. Ku 12308
. Tu 0.3077Tu phase margin = 61 0
142
Gain margin = 3.45,
0.2015K u 0.7878Tu 01970
Astrom and Gain margin ≥ 2 , phase
Hagglund [93] 0.35Ku 0.77Tu 0.19Tu margin ≥ 450
Atkinson and Davey 0. 25Ku 0.75Tu 0. 25Tu 20% overshoot - servo
[119] response
** Perry and Chilton 033
. Ku 05
. Tu 033
. Tu ‘Some’ overshoot
[120] 02
. Ku 05
. Tu 033
. Tu ‘No’ overshoot
Yu [122] – page 11 033
. Ku 05
. Tu 0125
. Tu ‘Some’ overshoot
02
. Ku 05
. Tu 0125
. Tu ‘No’ overshoot
Luo et al. [121] 0.48Ku 0.5Tu 0125
. Tu
McMillan [14] – 0.5K u 0.5Tu 0125
. Tu
page 90
McAvoy and 054
. Ku Tu 0.2Tu
Johnson [83]
Karaboga and Kalinli  04267
. K u 15 . Ku 
[123] [ 032
. Ku ,0.6Ku ] 
 Tu
,
Tu 
 [ 008 . Ku Tu ]
. Ku Tu ,015
Hang and Astrom Ku sinφ m Tu (1 − cos φ m ) Tu (1 − cos φ m ) φ m = phase margin
[124] π sin φm 4π sin φ m
Astrom et al. [30] Ku cos φ m Tu  
 tan φ m + 1 + tan φ m 
2 Tu  
 tan φ m + 1 + tan φ m 
2 ‘small’ time delay
4π   16π  
St. Clair [15] - 05

. Ku 12
. Tu 0125
. Tu ‘aggressive’ tuning
page 17 0.25Ku 12
. Tu 0125
. Tu ‘conservative’ tuning
Pole placement - Shin Kc ( 65) 1
Ti ( 65)
αTi ( 65) Typical α : 0.1
et al. [125] Typical ξ : [0.3,0.7]
Other rules
Harriott [126] – K25% 0167
. T25% 0.667T25% Quarter decay ratio
pages 179-180
K25% 0.67T25% 017
. T25% Quarter decay ratio
Parr [64] – pages 05.
191, 193 Tu 0.25Tu
G p ( jω u )
McMillan [14] - 083
. K25% 0.5T25% 01
. T25% ‘Fast’ tuning
Page 43 0.67K25% 0.5T25% 01
. T25% ‘Slow’ tuning
1 1 Ti φ m = 450 ,‘small’ τ m
Calcev and Gorez 2 2 G p ( jω u ) ωu 4 φ m = 15 0 , ‘large’ τ m
[69]
Zhang et al. [127] Kc( 66) 2 Ti (66 ) Td ( 66)
ρ
1
K c ( 65) = ,
 1   1 
a 1  αω uTi −  − a 2  αω 1Ti −  − b2 − ρ a 2
( 65) ( 65)
 ω u Ti   ω 1Ti 
( 65) ( 65)
 
( a2 − a1 ) + (a 1 − a 2 ) 2 + 4(αρa1ω u + αb2 ω 1 ) ρωa1 + ωb 2 
 1
Ti ( 65) =
u
2(αρa1ω u + αb2 ω 1 )
( ω − ω1 ) 1 − ξ2 , b = 1 sin ∠G ( jω ) , a = − 1
a2 =
1
K1
(
cos ∠G p ( jω 1 ) , ρ = u
ωu
) ξ
2
K1
p 1 1
Ku
( )
K1 , ω1 = modified ultimate gain and corresponding angular frequency
1
2
Kc ( 66) = , d, ε = relay amp. and deadband, A = limit cycle amp.
(
π 2 A2 − ε 2 ) + sin
2
(
1 + tan φm − φ p
16d 2
) φm 2
Crossing point of the Nyquist curve and relay with hysteresis is outside the unit circle:
2
 ωu 
  −1
 ωc  ωu ω
Ti ( 66) =
 ω 
. +
, 10
ωc
( )
tan φ m − φ p ≤ β ≤ 1.2 + u tan φ m − φ p
ωc
( )
ω u  β − u tan φ m − φ p 
 ωc 
( )
Td ( ) =
66
βω u − ω c tan φm − φ p ( )
, ω c = frequency when the open loop gain equals unity.
ωu − ωc
2 2
Crossing point of the Nyquist curve and relay with hysteresis is within the unit circle:
2
 ωu 
  −1
 ωc  ωu − ωr ω − ωr
Ti ( 66) = , − 0.4 ≤ β ≤ u + 0.4
 ω  ωr ωr
ω u  β + u tan φ p − φ m 
 ωc 
( )
Td ( ) =
66
βω u − ω c tan φm − φ p ( )
, ω r = oscillation frequency when a pure relay is switched into closed loop.
ωu − ωc
2 2
Garcia and Castelo ( 66a ) 3 ( 66a ) ( 66a )
Kc Ti T d
[127a]
( 66a )
=
(
cos 180 0 + φm − ∠G p ( jω1 ) ), T ( 66a )
=
( )
2[sin 180 0 + φ m − ∠G p ( jω1 ) + 1]
( )
3
Kc ,
G p ( jω1 ) ω1 cos 180 + φ m − ∠G p ( jω1 )
i 0
=
(
sin 180 0 + φm − ∠G p ( jω1 ) + 1 )
) , ω = oscillation frequency when a sine function is placed in series
( 66a )
Td
(
2ω1 cos 180 0 + φm − ∠G p ( jω1 )
1
with the process in closed loop; ω1 < ωu .

Table 42: PID tuning rules - non-model specific – controller with filtered derivative
 
 1 Td s 
Gc ( s) = Kc 1 + +  . 8 tuning rules.
 Ti s 1 + Td s 

 N 
Direct synthesis
Ti A3
2( A 1 − Ti ) Td
( 67 ) 2
Td ( 67) 4 N < 10
A 2 − Td A1 −
( 67)
N
Vrancic [72]
A3 A3 A3A 4 − A2 A 5 N ≥ 10
(
2 A 1A 2 − A 3K m − Td A 1
2
) A 2 − Td A1 A 3 − A1A 5
2
05
. Ti A2 − A 2 2 − 4χ A1A 3 N = 10
A1 − Km Ti 2χ A1
χTi χ = [ 02
. , 0.25]
A3
8 ≤ N ≤ 20
2
Vrancic [73] Kc ( 67a ) Td (67 a) Td (67 a)
A 2 − Td (67 a) A 1 − Km
N
(A ) 4
( A 3A 2 − A5 )( A5 A2 − A4 A 3 )
2
− A3 − A 5A1 + − A 5A1 −
2 2
3
4
Td ( 67) = N
2
N
( A 3A2 − A5 )
t t t t
 y( τ ) 
y1( t) =  Km −
 ∫
0
∆u 
 dτ , y2 ( t) = ∫ (A 1 − y1 (τ))dτ ,
0
y3 ( t) = ∫ ( A 2 − y2 (τ))dτ ,
0
y 4 (t ) = ∫ [A
0
3 ]
− y 3 ( τ ) dτ ,
y5 ( t) = ∫ [A
0
4 ]
− y 4 ( τ) dτ , A1 = y1( ∞) , A 2 = y 2 ( ∞) , A 3 = y3 ( ∞) , A4 = y 4 ( ∞) , A 5 = y5 ( ∞)
1 + b1s + b2 s 2 + b3 s3 + b4 s 4 + b5 s5 − sτ
Alternatively, if the process model is G m ( s) = K m e , then m
1+ a 1s + a 2 s2 + a 3s 3 + a 4 s4 + a 5s 5
A1 = Km ( a1 − b1 + τ m ) , A2 = K m b 2 − a2 + A1a1 − b1 τm + 05
. τ m2 , ( )
(
A3 = Km a 3 − b3 + A 2 a1 − A1a 2 + b2 τ m − 05
. b1τ m 2 + 0167
. τ m3 , )
A4 = Km (b − a
4 4 + A 3a 1 − A 2a 2 + A 1a 3 − b3 τ m + 05
. b2 τ m 2 + 0167
. b1τ m 3 + 0.042 τ m 4 , )
A5 = Km (a − b
5 5 + A 4 a1 − A 3a 2 + A2 a 3 − A1a 4 + b 4 τm − 05
. b3 τ m2 + 0167
. b2 τ m 3 − 0.042b1τ m 4 + 0.008τm 5 )
A3
K c ( 67a ) =
 [ T (67 a ) ]2 
2 A 1A 2 − A 0A 3 − Td A1 − d A 0A 1 
( 67 a ) 2
 N 
Td (67a ) is obtained from a solution of the following equation:
A0 A3
[ ] A1A 3
[ ] A 0A 5 − A 2A 3
[ ] + (A )[ ] + (A A − A 3A4 ) = 0
4 3 2
+ − − A 1A5 Td
( 67a ) ( 67 a ) ( 67 a ) 2 ( 67a )
Td Td Td 3 2 5
N3 N2 N
Direct synthesis
Lennartson and Kc(111) 5 Ti (111) 0.4Ti (111) Ku Km ≤ 0.6
Kristiansson [157]
Model: Method 1
Kc(111a ) Ti (111a) 0.4Ti (111a) Ku Km > 10
Kristiansson and ω u K u Km ≤ 0.4
Lennartson [157]
Kc(111b ) Ti (111b ) 0.4Ti (111b) Ku K m > 10
Model: Method 1 ω u Ku K m ≥ 0.4
Kc(111c) 6 Ti (111c) 0.4Ti (111c) Ku Km > 10
ω u K u Km ≤ 0.4
Kc(111d ) Ti (111d) 0.4Ti (111d ) Ku Km > 10
ω u Ku K m ≥ 0.4
Ku K m > 167
.
ω u Ku K m > 0.45
( 111e) ( 111e) ( 111e)
Kc Ti 0.4Ti
N = 2.5
12Ku K m − 35K u Km + 30
2 2
K c (111)
Kc (111) = Ku K m =
5 ( 111)
[ ]
, Ti ,
−0.053ω u + 0.47ω u − 014
. ω u + 0.11
3 2
Ku K m + 25
. 12K u Km − 35Ku K m + 30
2 2
2.5  35 30 
N= 12 − + 
Km K u  K m K u K m 2 Ku 2 
12 Ku Km − 35Ku Km + 30
2 2
K c (111a )
Kc (111a ) = Ku Km =
(111a )
[ ]
, T ,
−0.525ω u3 + 0.473ω u2 − 0.143ω u + 0.113
i
Ku Km + 3 12Ku 2K m2 − 35KuK m + 30
3  35 30 
N= 12 − + 2 2
Km Ku  K mK u Km Ku 
12Ku Km − 35Ku Km + 30
2 2 ( 111b)
Kc
Kc (111b ) = Ku Km =
( 111b)
[ ]
, Ti ,
K uKm + 3 12Ku2 Km 2 − 35Ku Km + 30 −0185
. ω u + 1052
. ω u − 0854
. ω u + 0.309
3 2
3  35 30 
N= 12 − + 2 2
Km Ku  K mK u Km Ku 
( 111c )
12Ku K m − 35K u K m + 30
2 2
Kc
Kc (111c ) = Ku K m =
6 ( 111c )
[ ]
, Ti ,
−0525
. ωu + 0.473ω u − 0143
. ω u + 0113
3 2
Ku K m + 25
. 12K u Km − 35Ku Km + 30
2 2
.
2.5  35 30 
N= 12 − + 
( 111d)
12Ku K m − 35K u Km + 30
2 2
Kc
Kc (111d ) = Ku K m =
( 111d )
[ ]
, Ti ,
−0185
. ωu + 1052
. ω u − 0.854ω u + 0309
3 2
Ku K m + 25
. 12K u Km − 35Ku K m + 30
2 2
.
2.5  35 30 
N= 12 − + 
7.71 914
. K c (111e )
= 2 − + 314 =
( 111e ) (111e )
Kc . , Ti
2
Km K u K m Ku −0.63ω u + 0.39ω u 2 + 0.15ω u + 0.0082
3
Kristiansson and
Lennartson [158] Kc(111f) 7 Ti (111f ) 0.4Ti (111f )
Model: Method 1
(111g)
Tf given below;
Kristiansson and (111g ) (111g ) (111g )
Kc Ti Td 0.1 ≤ K u K m ≤ 0.5
Lennartson [158a]
(111h )
Tf given below;
(111h ) 8 (111h ) (111h )
Model: Method not Kc Ti Td K u K m > 0.5
specified
12 Ku Km − 35Ku Km + 30Ku
3 2 2 ( 111f ) 3 2
Kc Km Ku
7
Kc (111f ) = , Ti (111f ) =
[ ] [ ]
,
Km Ku + 25
3 3
. 12K u Km − 35Ku Km + 30
2 2
ω u 095
. Km Ku − 2Km Ku + 14
2 2
.
2.5  35 30 
N= 12 − + 
(1. 1K u K m − 2. 3K u K m + 1.6)
2 2  1. 6( −20 + 13 K m K u )(1+ 0.37 K m K u ) 
2
=  − 1
(111g )
Kc
K m K u ( −20 + 13K m K u )(1 + 0.37 K m K u ) 2  1.1K m K u − 2.3K m K u + 1. 6
2 2 2

K K (1. 1K u K m − 2. 3K u K m + 1.6) 1.6( −20 + 13K m K u )(1 + 0. 37K m K u ) 
2 2
= m u  − 1
(111g )
Ti
ω u ( −20 + 13K m K u )(1 + 0. 37K m K u ) 2  1. 1K m 2 K u 2 − 2. 3K m K u + 1.6 
 ( −20 + 13K m K u ) 2 (1 + 0.37 K m K u ) 2 
 
K K (1. 1K u K m − 2.3K u K m + 1. 6)  (1.1K m K u − 2.3K m K u + 1.6)
2 2 2 2 2

= m u −
(111g )
Td 1
ω u ( −20 + 13K m K u )(1 + 0.37 K m K u ) 2  1.6( −20 + 13K m K u )(1 + 0. 37K m K u ) 
 − 1 
 1.1K m K u − 2. 3K m K u + 1.6 
2 2
K m K u (1.1K u K m − 2.3K u K m + 1.6)

2 2
=
(111g)
Tf
ω u ( − 20 + 13 K m K u )(1 + 0.37 K m K u ) 2
(1.1K u K m − 2.3K u K m + 1. 6)  4. 8K m K u (1 + 0.37 K m K u ) 
2 2 2
=  − 1
8 (111h )
Kc
3ω u K m K u (1+ 0.37 K m K u )  1. 1K m K u − 2. 3K m K u + 1.6 
2 2 2 2 2
(1.1K u K m − 2.3K u K m + 1.6)  4.8K m K u (1+ 0.37 K m K u ) 

2 2
=  − 1
(111h )
Ti
3ω u (1 + 0.37 K m K u ) 2
1.1K m K u − 2.3K m K u + 1.6 
2 2
 3K m 2 K u 2 (1+ 0.37 K m K u ) 2 
 
(1.1K u K m − 2.3K u K m + 1.6)  (1.1K m K u − 2.3K m K u + 1. 6)
2 2 2 2

2
= − 1
(111h )
Td 
3ω u (1 + 0. 37K m K u ) 2
4 .8 K K ( 1 + 0. 37 K K )
 m u m u
−1 
 1. 1K m 2 K u 2 − 2.3K m K u + 1.6 
 
1.1K u K m − 2. 3K u K m + 1.6
2 2
=
(111h )
Tf
3ω u (1 + 0.37 K m K u ) 2
Kristiansson and (111i )
Tf given below;
Lennartson [158a] (111i) 9 (111i ) (111i )
Kc Ti Td K u K m < 0.1
(continued)
Model: Method not
specified
Other
ωTi (
tan φm − φ ω − 0.5π ) 2π β = 10
G p ( jω ) (1 − ω 2
Ti Td ) 2
+ ω Ti 2
2
ω 1+
2π
(
tan φ m − φ ω − 0.5π ) βω φω > φm − π
β
Leva [67]
ωTi
αTd (68) 10 Td ( 68) 6 ≤ α ≤ 10
G p ( jω ) (1 − ω 2
Ti Td ) 2
+ ω Ti
2 2
φω < φm − π
Astrom [68] Ku cosφ m
[
2 tan φ m + 1 + tan 2 φ m ] Ti
4
Parameters determined
at φ m = 300 ,450 ,600
ωu
( −6 + 3.7K 1350 K m ) 2  20 .8(1.8 + 0.3K m K135 0 ) 

=  − 1
9 (111i )
Kc
13 K m (1.8 + 0.3K m K 1350 )  ( −6 + 3. 7K m K 1350 )
2

( −6 + 3.7K 1350 K m ) K 1350 K m  20. 8(1.8 + 0. 3K m K 1350 ) 
=  − 1
(111i )
Ti
13ω1350 (1. 8 + 0.3K m K135 0 ) 2  (− 6 + 3.7K m K 1350 ) 
 169 (1. 8 + 0.3K m K135 ) 2
 0
 
( −6 + 3.7K m K 135 ) K m K 135  ( −6 + 3. 7K m K 135 ) 2  ( −6 + 3. 7K m K135 )K m K135 0 0
= − 1 , Tf =
(111i ) 0 0 ( 111i) 0
2  20 .8(1. 8 + 0 .3K K
Td
13ω135 (1. 8 + 0.3K m K135 )  m 135 ) 13ω 135 (1.8 + 0.3K m K 135 ) 2
−1 
0 0 0 0 0
 ( −6 + 3.7K m K135 ) 
 
0
− αω + α 2ω 2 + 4αω 2 tan 2 ( φ m − φ ω − 05
. π)
10
Td ( 68) =
2αω 2 tan(φ m − φω − 0.5π )
Table 43: PID tuning rules - non-model specific – ideal controller with set-point weighting
( )
U( s) = Kc Fp R(s) − Y(s) +
1
Ts
( Fi R(s) − Y(s)) + Td s( Fd R(s) − Y(s)) . 1 tuning rule.
i
Ultimate cycle
Mantz and Tacconi 0.6K u 0.5Tu 0125
[118] Fp = 017
. Fi = 1 Fd = 0.654
Table 44: PID tuning rules - non-model specific – ideal controller with proportional weighting
 1 
 Ti s 
Direct synthesis
(0.33e )K . e −1.3κ + 3.5κ
b = 058
2
− 0.31κ − κ
2
,
(0.76e )T (017 )T
u
−1.6κ − 0.36κ −0.46κ − 2.1κ
0 < Km Ku < ∞
2 2
. e
Astrom and κ = 1 Km Ku u u
maximum Ms = 1.4
Hagglund [3] – page
. e 0.56 κ − 0.12 κ
b = 025
2
217
(0.72e − 1.6κ + 1.2 κ
2
)K u (0.59e −1.3κ + 0.38 κ

2
)T
u (015
. e −1.4κ + 0.56κ
2
)T
u
0 < Km Ku < ∞
maximum Ms = 2.0
Table 45: PID tuning rules - non-model specific – non-interacting controller
 1  K Ts
U( s) = Kc 1 +  E( s) − c d Y( s) . 1 tuning rule.
 Ti s  sT
1+ d
N
Ultimate cycle
Fu et al. [128] 0.5K u 0.34Tu 0.08Tu
 1
Table 46: PID tuning rules - non-model specific – series controller U( s) = Kc  1 +  (1 + sTd ) . 3 tuning rules.
 Ti s 
Ultimate cycle
05
. ‘optimum’ servo
Pessen [131] 025
. Ku 033
. Tu Tu response
05
. ‘optimum’ regulator
0.2K u 0. 25Tu Tu response - step
changes
0.2K u 0.5Tu 0. 33Tu No overshoot; close to
Pessen [129] optimum regulator
0. 33Ku 0. 33Tu 0.5Tu ‘Some’ overshoot
Grabbe et al. [130] 025
. Ku 033
. Tu 0.5Tu
Table 47: PID tuning rules - non-model specific – series controller with filtered derivative
 
 1  sTd 
U( s) = Kc 1 + 1 +  . 1 tuning rule.
 Ti s   1 + d 
sT

 N 
Ultimate cycle
Hang et al. [36] 0.35Ku 113
. Tu 0. 20Tu
- page 58
 1  1 + sTd
Table 48: PID tuning rules - non-model specific – classical controller U( s) = Kc  1 +  . 1 tuning rule.
 Ti s  T
1+ s d
N
Ultimate cycle
Corripio [117] – page 0.6K u 05
. Tu 0125
. Tu 10 ≤ N ≤ 20
27
Table 49: PID tuning rules - non-model specific – non-interacting controller
 1
U( s) = Kc  1 +  E ( s) − Kc Td sY (s) . 1 tuning rule.
 Ti s
Ultimate cycle
VanDoren [114] 0.75K u 0.625Tu 01

. Tu
K m e− sτ m  
Table 50: PID tuning rules - IPD model - ideal controller Gc ( s) = K c 1 + 1 + Td s . 5 tuning rules.
s  Ti s 
Direct synthesis
α
Wang and Cluett [76] K mτ m βτ m χτ m
– deduced from
Closed loop Damping Gain margin Phase margin
graph χ
time Factor, ξ Am φ m [degrees] α β
constant
Model: Method 2
τm 0.707 2.0 32 0.9056 2.6096 0.3209
2τm 0.707 3.1 40 0.5501 4.0116 0.2205
3τ m 0.707 4.4 46 0.3950 5.4136 0.1681
4τm 0.707 5.5 49 0.3081 6.8156 0.1357
5τ m 0.707 6.7 52 0.2526 8.2176 0.1139
6τ m 0.707 7.8 54 0.2140 9.6196 0.0980
7τm 0.707 8.9 55 0.1856 11.0216 0.0861
8τ m 0.707 10.0 56 0.1639 12.4236 0.0767
9τm 0.707 11.2 57 0.1467 13.8256 0.0692
10τ m 0.707 12.2 58 0.1328 15.2276 0.0630
11τm 0.707 13.4 59 0.1213 16.6296 0.0579
12τ m 0.707 14.5 59 0.1117 18.0316 0.0535
13τ m 0.707 15.6 59 0.1034 19.4336 0.0497
14τ m 0.707 16.7 60 0.0963 20.8356 0.0464
15τ m 0.707 17.8 60 0.0901 22.2376 0.0436
16τ m 0.707 19.0 60 0.0847 23.6396 0.0410
τm 1.0 2.0 37 0.8859 3.2120 0.3541

2τm 1.0 2.9 46 0.6109 5.2005 0.2612
3τ m 1.0 3.8 52 0.4662 7.1890 0.2069
4τm 1.0 4.6 56 0.3770 9.1775 0.1713
5τ m 1.0 5.5 58 0.3164 11.1660 0.1462
6τ m 1.0 6.4 61 0.2726 13.1545 0.1275
7τm 1.0 7.1 62 0.2394 15.1430 0.1311
8τ m 1.0 8.0 64 0.2135 17.1315 0.1015
9τm 1.0 8.7 65 0.1926 19.1200 0.0921
10τ m 1.0 9.5 66 0.1754 21.1085 0.0843
11τm 1.0 10.4 67 0.1611 23.0970 0.0777
12τ m 1.0 11.1 67 0.1489 25.0855 0.0721
13τ m 1.0 12.0 68 0.1384 27.0740 0.0672
14τ m 1.0 12.8 68 0.1293 29.0625 0.0630
15τ m 1.0 13.4 69 0.1213 31.0510 0.0592
16τ m 1.0 14.4 69 0.1143 33.0395 0.0559
0.9588 Closed loop time
Cluett and Wang [44] Km τ m 30425
. τm 0.3912 τ m constant = τ m

Model: Method 2
Km τ m 52586
. τm 0.2632 τ m constant = 2τm

Km τ m 7.2291τ m 0.2058τ m constant = 3τ m

Km τ m 91925
. τm 01702
. τm constant = 4τ m

Km τ m 111637
. τm 01453

Km τ m 131416
. τm 01269
Damping factor for
Rotach [77] 121
. . τm
160 0.48τ m oscillations to a
K mτ m disturbance input =
Model: Method 4 0.75.
Process reaction
Ford [132] 1.48
Model: Method 2 K mτm 2τ m 0.37τ m Decay ratio: 2.7:1
Astrom and 094
.
Hagglund [3] – page Km τ m 2τ m 0.5τ m Ultimate cycle Ziegler-
139 Nichols equivalent
Model: Method not
relevant
K m e− sτ m
Table 51: PID tuning rules - IPD model - Ideal controller with first order filter, set-point weighting and
s
 1  1  1 + 0.4Trs 
output feedback U(s ) = K c 1 + + Tds   R( s) − Y(s )  − K 0 Y(s) . 1 tuning rule.
 Ti s  Tf s + 1  1 + sTr 
Direct synthesis
Normey-Rico et al. Tf = 0. 13τm
[106a] 0.563 1
K m τm 1.5τm 0.667 τm K0 =
2Km τ m
Model: Method not
specified Tr = 0.75τ m
K m e− sτ m
Table 52: PID tuning rules - IPD model - ideal controller with filtered derivative
s
 
 1 Td s 
G c ( s) = Kc  1 + +  . 1 tuning rule.
 Ts
i
sT
1+ d 
 N 
Robust
Chien [50] 2 τ m ( λ + 025
. τm ) 1
λ= ; N=10
K m ( λ + 0.5τm ) 2λ + τm
2 λ + τm Km
Model: Method 2
K m e− sτ m
Table 53: PID tuning rules - IPD model - series controller with filtered derivative
s
 
 1  Td s 
G c ( s) = Kc  1 +  1 +  . 1 tuning rule.
 i 
 1 + d 
Ts Ts
 N 
Robust
1  2λ + 0.5τ m   1
λ=

, τ m  ; N=10
Chien [50] 2λ + 05
. τm . τm
05
K m  [ λ + 05
. τ m ] 
2
K m 
Model: Method 2
1  . τm
05 

 1
λ=

, τ m  ; N=10
05τ m 2λ + 05
. τm
K m  [ λ + 05
. τ m ] 
2
K m 
 
K e− sτ m  1   1 + Td s 
Table 54: PID tuning rules - IPD model m - classical controller G c ( s) = Kc  1 +   .
s  Ti s   1 + Td s 
 
 N 
5 tuning rules.
Ultimate cycle
Luyben [133] maximum closed loop
Model: Method 2 0.46Ku 2.2Tu 016
. Tu log modulus of +2dB ;
N=10
Belanger and
Luyben [134] 311
. Ku 2.2Tu 364T
. u N=10
Model: Method 2
Robust
 05 . τ m   1 
Chien [50]
1  0.5τm 2λ + 05
. τm λ= , τ m  ; N=10
 λ + 0.5τ 2 
[ m] 
 m 
Km K
Model: Method 2
1  2λ + 0.5τ m   1
λ=

, τ m  ; N=10
2λ + 0.5τ m 0.5τm
K m  [ λ + 05
. τ m ] 
2
K m 

Minimum IAE -
121. Model: Method 056
. Ku 0.39Tu 015
. Tu
not specified
Minimum IAE -
Shinskey [17] – page 0.93 . τm
157 . τm
056
117. Model: Method Km τ m
1
Minimum IAE – 09259
.
Shinskey [59] – page Km τ m . τm
160 0.58τ m N=10
74
Model: Method not 09259
.
Km τ m 1.48τ m 0.63τm N=20
specified
K m e− sτ m
Table 55: PID tuning rules - IPD model - Alternative non-interacting controller 1 -
s
 1 
U( s) = Kc 1 +  E( s) − Kc Td sY ( s) . 2 tuning rules.
 Ti s 
Minimum IAE -
. . τm
19 . τm
046
74. K m τm
Model: Method not
specified
Minimum IAE –
121. Model: Method 0. 77K u 0. 48Tu . τm
015
not specified
Minimum IAE –
. 1.90τ m 0.48τ m
117. Model: Method K mτ m
1
K m e− sτ m
Table 56: PID tuning rules - IPD model - I-PD controller U(s) = Kc E (s) − K c(1 + Td s)Y(s) .
s Ti s
1 tuning rule.
Direct synthesis
Chien et al. [74] 0.25τ m2 + 0.707TC Lτ m Underdamped system
Kc ( 68a ) 1 . TCL + τ m
1414 1.414TCL + τ m response - ξ = 0.707 .
Model: Method 1 τ m ≤ 0.2Tm
1 ( 68a ) 1.414 TCL + τ m

( )
Kc
+ 0.707TCL τ m + 0.25τ m
2 2
Km T CL
K me− sτ m
Table 57: PID tuning rules - IPD model G m ( s) = - controller
s
1
U(s) = Kc (1 + ) E( s) + Kc ( b − 1) R (s) − Kc TdsY( s) . 1 tuning rule.
Tis
Direct synthesis
Hansen [91a]
Model: Method not 0.938 K m τm 2. 7τm 0.313 τm b = 0.167
specified
K m e− sτ m
Table 58: PID tuning rules - IPD model - Two degree of freedom controller:
s
   
   
1 Td s β T s
U (s) = K c 1 + +  E( s) − K c  α + d  R (s) . 2 tuning rules.
 Tis T   Td 
 1+ d s  1+ s
 N   N 
tuning
τm
≤ 1.0 ;
Taguchi and Araki 1  1.253  Tm
  2.388 τ m 0.4137 τm
K m  τ m 
[61a]
α = 0. 6642 ≤ settling time of
β = 0.6797 tuning rules of Chien
et al. [10]
Minimum ITAE - α = 0. 601 , β = 1 ,
Pecharroman and 1.672 K u 0.366 Tu 0.136 Tu
N = 10, φ c = −1640
Pagola [134a]
α = 0.607 , β = 1 ,
φ c = phase 1.236 K u 0.427 Tu 0.149 Tu
N = 10, φ c = −1600
corresponding to the α = 0.610 , β = 1 ,
crossover frequency; 0.994 Ku 0.486 Tu 0.155 Tu
Km =1 N = 10, φ c = −1550
α = 0.616 , β = 1 ,
0.842 Ku 0.538 Tu 0.154 Tu
N = 10, φ c = −1500
Model: Method 6
α = 0.605 , β = 1 ,
0.752 Ku 0.567 Tu 0.157 Tu
N = 10, φ c = −1450
α = 0.610 , β = 1 ,
0.679 Ku 0.610 Tu 0.149 Tu
N = 10, φ c = −1400
α = 0.612 , β = 1 ,
0.635 K u 0.637 Tu 0.142 Tu
N = 10, φ c = −1350
α = 0.610 , β = 1 ,
0.590 Ku 0.669 Tu 0.133 Tu
N = 10, φ c = −1300
α = 0.616 , β = 1 ,
0.551K u 0.690 Tu 0.114 Tu
N = 10, φ c = −1250
α = 0.609 , β = 1 ,
0.520 K u 0.776 Tu 0.087 Tu
N = 10, φ c = −1200
α = 0. 611 , β = 1 ,
0.509 Ku 0.810 Tu 0.068 Tu
N = 10, φ c = −1180
Km e − sτ  
- ideal controller Gc ( s) = K c 1 + 1 + Td s .
m
Table 59: PID tuning rules - FOLIPD model

s(1 + sTm )  Ti s 
1 tuning rule.
Ultimate cycle
McMillan [58] Kc( 69) 1 Ti ( 69 ) Td ( 69) Tuning rules
Model: Method not developed from Ku , Tu
relevant
2
 
 
1111
. Tm  1    T  0.65    T  0.65 
( 69 ) ( 69 )
1
Kc ( 69) =   , Ti = 2τ m 1 +    , Td = 05
m
. τ m 1 +  m  
K m τ m 2   T  0.65    τm     τ m  
 
1 +  τ  
m
 m 
Km e − sτ
m
Table 60: PID tuning rules - FOLIPD model - ideal controller with filter
s(1 + sTm )
 1  1
G c ( s) = Kc  1 + + Td s . 3 tuning rules.
 Ti s  1 + Tf s
Robust
τm
Tm τ m
Tf =
0.463λ + 0277. τm 5.750λ + 0.590
Tm +
Tan et al. [81] [
Km τ m
2 ]
0.238λ + 0123
. ( 0238
. λ + . ) Tm + τ m
0123 λ = 0.5 - performance
. λ + 0123
([ 0238 . ]Tm + τ m ) λ = 0.1 - robustness
Model: Method 2
λ = 0.25 - acceptable
λ3
Tf =
Zhang et al. [135] 3λ + Tm + τ m 3λ + Tm + τm ( 3λ + τ m ) Tm 3λ 2 + 3λτ m + τ m 2
Model: Method 2
(
K m 3λ 2
+ 3λτ m + τ m2 ) 3λ + τ m + Tm . τ m ≤ λ ≤ 4.5τ m
15
λ = 15. τ m ….Overshoot = 58%, Settling time = 6τ m
λ = 25. τ m ….Overshoot = 35%, Settling time = 11τ m Obtained from graph
λ = 35
. τ m ….Overshoot = 26%, Settling time = 16τm
λ = 45
. τm ….Overshoot = 22%, Settling time = 20τm
0. 0337Tm  τm  Tmτ m
1 + 
Tan et al. [136] Km τ m 
2
01225
. Tm 
Tm + 81633
. τm 01225
. Tm + τ m Tf = 05549
. τm
0. 0754Tm  τm  Tm τ m
Model: Method 2 2 
1+ 
Km τ m  0.1863Tm 
Tm + 53677
. τm 01863
. Tm + τ m Tf = 0. 4482τ m
01344
. Tm  τm  Tm τ m
 1+  Tm + 3.9635τ m 0.2523Tm + τ m Tf = 0. 2863τm
K mτ m2  0.2523Tm 
Km e − sτ m
Table 61: PID tuning rules - FOLIPD model - ideal controller with set-point weighting
s(1 + sTm )
 1 
 Ti s 
Direct synthesis
. e −8 .8 τ + 6.8τ . τm e 6.7 τ − 4.4 τ . τm e −6.4 τ + 2.0 τ
2
56 11
2
17
2
Maximum Ms = 1.4
,
K m ( Tm + τ m )
Astrom and
. e 6.9 τ − 6.6 τ
b = 012
2
Hagglund [3] - pages

212-213 τ = τ m ( τ m + Tm )
. e −7.1τ + 5.4 τ . τ m e3.3τ − 2 .33τ 0.38τ me 0.056τ − 0.60 τ
2
86 10
2 2
Maximum Ms = 2.0
Model: Method 1 or
K m ( Tm + τ m ) . e −2 .2τ + 1.2 τ
b = 056
2
Method 2.
 
Km e − sτ m
 1   1 + Td s 
Table 62: PID tuning rules - FOLIPD model - classical controller G c ( s) = Kc  1 +   .
s(1 + sTm )  Ti s   1 + Td s 
 
 N 
5 tuning rules.
Robust
1  T 
  2λ + τm λ = [ τ m , Tm ] ; N=10
m
Chien [50] Tm
 λ+τ 2
Km [ ]
m 
Model: Method 2
1  2λ + τ m 
2λ + τm Tm λ = [ τ m , Tm ] ; N=10
K m  [ λ + τ m ] 2 
Minimum IAE –
Shinskey [59] – page 0. 78 1.38( τ m + Tm ) 0.66(τ m + Tm ) τ m = Tm ; N=10
75. K m ( τ m + Tm )
Model: Method not
specified
100   T
− m  Tm
  > 05
Minimum IAE -
Shinskey [59] –
108K m τ m 122 (
. − 0.03 τ
m
Tm
) 157


. 1 − e τ m
. τ m  1 + 12


 

0.56τ m + 0.75Tm τm
.
pages 158-159 100   T

 Tm
− m
  ≤ 05
Model: Open loop
108K m τ m 1 + 0. 4 τ m
T
m
( ) 
. 1 − e τ m
. τ m 1 + 12
157




0.56τ m + 075
. Tm τm
.
Minimum ITAE - τm
0≤ ≤2;
Poulin and b Tm 2
+1
a(τ m + Tm ) Tm ( Td N)
Pomerleau [82], [92] – K m (τ m + Tm ) a (τ m + T m )
2
Td
deduced from graph . Tm ≤
01 ≤ 033
. Tm
N
Model: Method 2
τm ( Td N) a b τm ( Td N) a b
0.2 5.0728 0.5231 1.2 4.7565 0.5250
Output step load 0.4 4.9688 0.5237 1.4 4.7293 0.5252
disturbance 0.6 4.8983 0.5241 1.6 4.7107 0.5254
0.8 4.8218 0.5245 1.8 4.6837 0.5256
1.0 4.7839 0.5249 2.0 4.6669 0.5257
0.2 3.9465 0.5320 1.2 4.0397 0.5312

Input step load 0.4 3.9981 0.5315 1.4 4.0278 0.5312
disturbance 0.6 4.0397 0.5311 1.6 4.0278 0.5312
0.8 4.0397 0.5311 1.8 4.0218 0.5313
1.0 4.0397 0.5311 2.0 4.0099 0.5314
Km e − sτ
m
Table 63: PID tuning rules - FOLIPD model - series controller with derivative filtering
s(1 + sTm )
 
 1  Td s 
G c ( s) = Kc  1 +  1 +  . 1 tuning rule.
 i 
 1 + d 
Ts Ts
 N 
Robust
1  2λ + τ  λ = [ τ m , Tm ] ; N=10
Chien [50]  m 
2λ + τm Tm
 λ + τ 2
[ m] 
Km (Chien and Fruehauf
[137])
Model: Method 2
1  Tm 
 Tm 2λ + τm λ = [ τ m , Tm ] ; N=10
K m  [ λ + τ m ] 
2
Km e − sτ m
Table 64: PID tuning rules - FOLIPD model - alternative non-interacting controller 1
s(1 + sTm )
 1 
U(s) = Kc 1 +  E (s) − K c Td sY ( s) . 2 tuning rules.
 Ti s 
Minimum IAE -
. 1.38( τ m + Tm ) 0.55( τ m + Tm )
75. K m (τ m + Tm )
Model: Method not
specified
Minimum IAE -
Shinskey [59] – page 1.28  −
Tm 
. τ m 1 + 0.75[1 − e τ m ]
2
T 
159. K m τ m (1 + 0.24
Tm
− 014
.  m )
19
  0.48τ m + 0.7Tm
τm  τ m   
Model: Open loop
method not defined
Km e − sτ
m
Table 65: PID tuning rules - FOLIPD model - ideal controller with filtered derivative
s(1 + sTm )
 
 1 Td s 
 Ti s 1 + s Td 

 N
Robust
Chien [50] 2λ + τ m + Tm Tm ( 2λ + τ m )
K m( λ + τ m)
2 2λ + Tm + τ m λ = Tm ; N = 10
2λ + Tm + τ m
Model: Method 2
Km e − sτ m
Table 66: PID tuning rules - FOLIPD model - ideal controller with set-point weighting
s(1 + sTm )
( )
U(s) = Kc Fp R ( s) − Y( s) +
Kc
Ti s
[ Fi R(s) − Y(s)] + Kc Td s[ Fd R (s) − Y(s)] . 1 tuning rule.
Ultimate cycle
Oubrahim and 06
. Ku 05. Tu 0125
. Tu τm
0.05 < < 0.8 ;
Leonard [138] Fp = 01
. Fi = 1 Fd = 0.01 Tm
Model: Method not
20% overshoot
relevant
Km e − sτm
Table 67: PID tuning rules - FOLIPD model - Alternative classical controller
s(1 + sTm )
  
 1+ T s   1 + T s 
G c ( s) = Kc  i  d 
. 1 tuning rule.
 1 + Td s   1 + Td s 
 N  N 
Direct synthesis
Tsang and Rad [109] 0.809 Maximum overshoot =
K mτ m Tm 0.5τ m 16%; N = 8.33
Model: Method 3
Km e − sτ
m
Table 68: PID tuning rules – FOLIPD model - Two degree of freedom controller:
s(1 + sTm )
   
   
1 Td s β T s
 Tis T   Td 
 1+ d s  1+ s
 N   N 
tuning
τm
Tm
[61a] ( 69a ) 2 ( 69a ) ( 69a )
et al. [10]
Minimum ITAE - α = 0. 601 , β = 1 ,
N = 10, φ c = −1640
Pagola [134a]
α = 0.607 , β = 1 ,
φ c = phase 1.236 K u 0.427 Tu 0.149 Tu
N = 10, φ c = −1600
Km = 1 ; Tm = 1 ; N = 10, φ c = −1550
0.05 < τm < 0. 8 . α = 0.616 , β = 1 ,
0.842 Ku 0.538 Tu 0.154 Tu
N = 10, φ c = −1500
Model: Method 4 α = 0.605 , β = 1 ,
0.752 Ku 0.567 Tu 0.157 Tu
N = 10, φ c = −1450
α = 0.610 , β = 1 ,
0.679 Ku 0.610 Tu 0.149 Tu
N = 10, φ c = −1400
α = 0.612 , β = 1 ,
0.635 K u 0.637 Tu 0.142 Tu
N = 10, φ c = −1350
α = 0.610 , β = 1 ,
0.590 Ku 0.669 Tu 0.133 Tu
N = 10, φ c = −1300
α = 0.616 , β = 1 ,
0.551K u 0.690 Tu 0.114 Tu
N = 10, φ c = −1250
 
 
 2

2 ( 69a )
=
1 
0.7608 +
0.5184  ( 69 a )
=  0.03330 + 3.997 τ m − 0.5517  τm  
Kc , T T   
Tm  Tm  
 [ − 0.01308 ] 2  
 Tm 
 τm  τm 
2
 τm  
3
Td
( 69 a ) 
= Tm 0.03432 + 2. 058 − 1. 774   + 0.6878   ,
 

Tm  Tm   Tm  
2
τm τ 
α = 0. 6647 , β = 0. 8653 − 0. 1277 + 0.03330  m 
Tm  Tm 
Minimum ITAE - α = 0.609 , β = 1 ,
N = 10, φ c = −1200
Pagola [134a] -
continued α = 0. 611 , β = 1 ,
0.509 Ku 0.810 Tu 0.068 Tu
N = 10, φ c = −1180
Km e− sτ m
K m e− sτ m
Table 69: PID tuning rules - SOSPD model or - ideal controller

(1 + sTm1 )(1 + sTm 2 ) Tm12 s2 + 2ξ m Tm1s + 1
 1 
Gc ( s) = K c  1 + + Td s . 27 tuning rules.
 Ti s 
Minimum ITAE – τm
Sung et al. [139] ( 70) 0.05 < ≤2
Kc( 70) Ti (70 ) Td Tm1
Model: Method 2
Minimum ITAE – τm
Sung et al. [139] ( 71) 0.05 < ≤2
Kc( 71) Ti (71) Td Tm1
Model: Method 2
1
1    
− 0.983
τ 
1
Kc
( 70)
= −0.04 + 0333
. + 0.949  m  ξ m  , ξ m ≤ 0.9 or
Km    Tm1   
 
1  
−0.832
τm  τm 
Kc ( 70)
= −0.544 + 0308
. + 1408
.   ξ m  , ξ m > 0.9 .
Km  Tm1  Tm1  

 τ  τ  τ  τ
Ti ( 70) = Tm1 2.055 + 0.072 m  ξ m , m ≤ 1 or Ti ( 70) = Tm1 1768
. + 0.329 m ξ m , m > 1
 Tm1  T m1  Tm1  T m1
Tm1
Td ( 70) =
 
1.060
τ 
 m
 ξ
 −
T 
m1
m
 T  
1.090
1 − e 0.870  0.55 + 1.683 m1  

    τm  
  
 
1   τ
−2.001 − 0.766
 τm   τm 
Kc ( 71)
=  −067
. + 0297
.   + 2189
.   ξ m  , m < 0.9 or
Km   Tm1   Tm1   Tm1

1   τ
2 −0.766
τ  τ 
Kc ( 71) = − 0365
. + 0.260  m − 14
.  + 2189
.  m ξ m  , m ≥ 0.9
Km   Tm1   Tm1   Tm1

 τ 
0.520
 τ
Ti ( 71) = Tm1  2.212 m  − 03
.  , m < 0.4 or
  Tm1   Tm1
 −
ξm

  
2 τ 2
 τm   0.15+0.33 m  τm τ
Ti
( 71)
= Tm1{−0.975 + 0.910 − 1.845 + 1 − e Tm1
 5.25 − 0.88 − 2.8  } , m ≥ 0.4
 Tm1     Tm1   Tm1

 
Tm1
Td ( 71) =
 −
ξm
−1.121

  τ 
− 0.15+ 0.939 m   T  
1.171
τ 
−0.530
1 − e  Tm 1 
145
. + 0.969 m1   − 19 . + 1576
.  m
   τm    Tm1 
  
 
ξm τm Tm1 Kc Ti Td
25
Minimum ITAE – 0.5 0.1 05
. Tm1 0.25Tm1
Km
Lopez et al. [84] Representative results
- taken from plots 07
.
0.5 1.0 Km 1.3Tm1 12
. Tm1
Model: Method 12
0.35
0.5 10.0 Km 5Tm1 1.0Tm1
25
1.0 0.1 Km 0.5Tm1 0.2Tm1
18
.
1.0 1.0 Km 1.7Tm1 0.7Tm1
9.0
4.0 1.0 Km 2Tm1 0.45Tm1
Ultimate cycle
Regulator – nearly Kc ( 72) 2
Ti ( 72 ) 1.45(1 + Ku K m )  . 
116 τ
minimum IAE, ISE, 2
Km ω u
1 −


ε0 
0.2 ≤ m ≤ 2.0 and
Tm1
ITAE – Hwang [60] ( 1 − 0.612ω u τ m + 0103
. ω u2τ m2 )
0.6 ≤ ξ m ≤ 4.2 ; ε < 2.4
Model: Method 3 Decay ratio = 0.15 -
Kc ( 73)
Ti (73) 1.45(1 + Ku K m )  . 
116 τ
2
Km ω u
1 −


ε0 
0.2 ≤ m ≤ 2.0 and
Tm1
( 1 − 0.612ω u τ m + 0103
. ω u2τ m2 )
0.6 ≤ ξ m ≤ 4.2 ;
2.4 ≤ ε < 3
2 9  τm 2 ξ τ T 4 K mTd Kc τ m 
KH = − Tm1 − m m m +
2
2  +
2 K m τm  18 9 9ω u 
 
 T 4 + τm + 49 τm ξ m Tm − τm Tm1 + 7Tmξ m τ m + 10 τm1 Tm1 ξ m − K T K  10 τm + 4 Tm1 τm (ξ m τ m + Tm1 )  − 8Kc Km Td τ m
4 2 2 2 2 2 3 3 3 2 2 2 2
9 
2  m1 c d m

2 Km τm 324 81 9 81 9  81 9  81ω u 2
 
1 + K HK m 6T 2 + 4Tm1ξ mτ m + K uK mτ m 2
ωH = , ε 0 = m1
2ξ τ T K K τ 2
2 Kc KmTd τ m 2τ m Tm12ω u
Tm12 + m m m+ H m m −
3 6 3ω u
6Tm12 + 4 Tm1ξm τ m + KH Km τ m 2 − 4 τ mK mK cTd
ε=
( Km Kc Td τ m 2 + 2 τ mTm12 )ω H
( 72)


= KH −
0674
. [
1 − 0.447ω H τ m + 0.0607ω H2 τ m 2 ]  , T ( 72)
=
Kc
( 72)
(1 + K H K m )
( )
Kc
K m (1 + KH K m )  i
ω H K m 00607
. 1 + 105
. ω H τ m − 0.233ω H 2 τ m 2
 

Kc ( 73) =  KH −
[
0.778 1 − 0.467ω H τ m + 0.0609ω H 2 τ m 2 ]  , T ( 73)
=
Kc
( 73)
(1 + K H K m )

K m (1 + K H K m ) 

i
(
ω H K m 0.0309 1 + 2.84ω H τ m − 0.532ω H 2 τ m 2 )
Regulator – nearly Kc( 74) 3 Ti (74 ) 1.45(1 + Ku K m )  . 
116 τ
minimum IAE, ISE, 2
Km ω u
1 −


ε0 
0.2 ≤ m ≤ 2.0 and
Tm1
ITAE – Hwang [60] – ( 1 − 0.612ω u τ m + 0103
. ω u2τ m2 )
0.6 ≤ ξ m ≤ 4.2 ;
continued
3 ≤ ε < 20
Model: Method 3 Decay ratio = 0.15 -
Kc( 75) Ti (75) 1.45(1 + Ku K m )  . 
116 τ
2
Km ω u
1 −


ε0 
0.2 ≤ m ≤ 2.0 and
Tm1
( 1 − 0.612ω u τ m + 0103
. ω u2τ m2 )
0.6 ≤ ξ m ≤ 4.2 ; ε ≥ 20
Decay ratio = 0.2 -
Kc( 76) Ti (76 ) 1.45(1 + Ku K m )  . 
116 τ
2
Km ω u
1 −


ε0 
0.2 ≤ m ≤ 2.0 and
Tm1
( 1 − 0.612ω u τ m + 0103
. ω u2τ m2 )
0.6 ≤ ξ m ≤ 4.2 ; ε < 2.4
Decay ratio = 0.2 -
Kc ( 77)
Ti ( 77 ) 1.45(1 + Ku K m )  . 
116 τ
2
Km ω u
1 −


ε0 
0.2 ≤ m ≤ 2.0 and
Tm1
( 1 − 0.612ω u τ m + 0103
. ω u2τ m2 )
0.6 ≤ ξ m ≤ 4.2 ;
2.4 ≤ ε < 3
Decay ratio = 0.2 -
Kc( 78) Ti (78) 1.45(1 + Ku K m )  . 
116 τ
2
Km ω u
1 −


ε0 
0.2 ≤ m ≤ 2.0 and
Tm1
( 1 − 0.612ω u τ m + 0103
. ω u2τ m2 )
0.6 ≤ ξ m ≤ 4.2 ;
3 ≤ ε < 20
Decay ratio = 0.2 -
Kc( 79) Ti (79 ) 1.45(1 + Ku K m )  . 
116 τ
2
Km ω u
1 −


ε0 
0.2 ≤ m ≤ 2.0 and
Tm1
( 1 − 0.612ω u τ m + 0103
. ω u2τ m2 )
0.6 ≤ ξ m ≤ 4.2 ; ε ≥ 20
3

Kc ( 74) =  K H −
. ( 0519
131 . [
) ω H τm 1 − 103
. ε + 0514
. ε 2  ]
, Ti (74) =
K c (1 + K H K m )
( 74)

 K m (1 + KH K m ) 
 (
ω H K m 0.0603 1 + 0.929 ln[ ω H τ m ] 1 + 2.01 ε − 12
. ε2 )( )

Kc ( 75) =  KH −
114 [
. 1 − 0.482ω H τ m + 0.068ω H 2 τ m 2 ] , T ( 75)
=
Kc
( 75)
(1 + K H K m )

K m (1 + K H K m ) 

i
(
ω H K m 0.0694 − 1 + 21
. ω H τ m − 0.367ω H2 τ m 2 )
( 76)


= KH −
[
0.622 1 − 0.435ω H τ m + 0.052ω H τ m
2 2
] , T ( 76)
=
K c (76) (1 + K H Km )
( )
Kc
 K m (1 + K H K m )  i
  ω H Km 0.0697 1 + 0.752ω H τ m − 0145
. ω H 2 τm2
( 77)


= KH −
[
0.724 1 − 0.469ω H τ m + 0.0609ω H 2 τ m 2 ]  , T ( 77)
=
Kc
( 77 )
(1 + KH K m )
( )
Kc
 K m (1 + K H K m )  i
ω H K m 00405
. 1 + 193
. ω H τ m − 0363
. ωH 2 τm2
 
( 78)


= KH −
126
ω τ
[
. ( 0.506) H m 1 − 1.07 ε + 0.616 ε 2  ]
, Ti (78) =
K c (1 + K H K m )
( 78)
)( )
Kc

 K m (1 + KH K m ) 
 (
ω H K m 0.0661 1 + 0.824 ln[ ω H τ m ] 1 + 171
. ε − 117
. ε2

Kc ( 79) =  K H −
109 [
. 1 − 0497
. ω H τ m + 00724
. ω H 2 τm2 ]  , T ( 79)
=
Kc
( 79 )
(1 + KH K m )

Km (1 + KH K m ) 

i
(
ω H K m 0.054 − 1 + 2.54ω H τ m − 0457
. ω H 2 τm2 )
1.45(1 + Ku K m )  . 
116 Decay ratio = 0.25 -
1 − 
Regulator – nearly Kc(8 0) 4 Ti (80)
2
Km ω u  ε0  τ
0.2 ≤ m ≤ 2.0 and
minimum IAE, ISE, ( 1 − 0.612ω u τ m + 0103
. ω u2τ m2 ) Tm1
ITAE – Hwang [60] –
0.6 ≤ ξ m ≤ 4.2 ; ε < 2.4
continued
1.45(1 + Ku K m )  . 
116 Decay ratio = 0.25 -
1 − 
Model: Method 3 Kc ( 81)
Ti ( 81) 2
Km ω u  ε0  τ
0.2 ≤ m ≤ 2.0 and
( 1 − 0.612ω u τ m + 0103
. ω u2τ m2 ) Tm1
0.6 ≤ ξ m ≤ 4.2 ;
2.4 ≤ ε < 3
Kc(8 2) Ti (82) 1.45(1 + Ku K m )  . 
116 τ
2
Km ω u
1 −


ε0 
0.2 ≤ m ≤ 2.0 and
Tm1
( 1 − 0.612ω u τ m + 0103
. ω u2τ m2 )
0.6 ≤ ξ m ≤ 4.2 ;
3 ≤ ε < 20
Kc(83) Ti (83) 1.45(1 + Ku K m )  . 
116 τ
2
Km ω u
1 −


ε0 
0.2 ≤ m ≤ 2.0 and
Tm1
( 1 − 0.612ω u τ m + 0103
. ω u2τ m2 )
0.6 ≤ ξ m ≤ 4.2 ; ε ≥ 20
τ 
2
Servo– nearly Kc(8 4) Ti (84 ) 0471
. Ku ξ ≤ 0613
. + 0613
.
τm
+ 0117
.  m 
Tm1  Tm1 
minimum IAE, ISE, K mω u
ITAE – Hwang [60].
In all, decay ratio = τ 
2
τm
0.1 with Kc(85) Ti (85) 0471
. Ku ξ ≤ 0613
. + 0613
. + 0117
.  m 
Tm1  Tm1 
τ K mω u
0.2 ≤ m ≤ 2.0 and
Tm1
0.6 ≤ ξ m ≤ 4.2
Model: Method 3
4 ( 80)


=  KH −
[
0.584 1 − 0.439ω H τ m + 0.0514ω H 2 τ m 2 ]  , T ( 80)
=
Kc
( 80)
(1 + KH K m )
( )
Kc
K m (1 + K H K m )  i
ω H K m 0.0714 1 + 0.685ω H τ m − 0.131ω H 2 τ m 2
 

Kc (81) =  K H −
0675
. [
1 − 0.472ω H τ m + 0061
. ω H 2 τm2 ]  , T ( 81)
=
Kc
( 81)
(1 + K H K m )

K m (1 + KH K m ) 

i
ω H K m 00484
. 1 + 143 (
. ω H τ m − 0.273ω H 2 τ m 2 )

Kc (82 ) =  K H −
12
ω τ
[
. ( 0.495) H m 1 − 11
. ε + 0.698 ε 2  ]
, Ti (82) =
K c (1 + K H K m )
( 82)

 K m (1 + K H K m ) 
 ω H K m 0.0702 1 + 0.734 ln[ ω H τ m ] 1 + 148 (
. ε − 11
. ε2 )( )

Kc (83) =  KH −
103 [
. 1 − 0.51ω H τ m + 0.0759ω H 2 τ m 2 ] , T ( 83)
=
Kc
( 83)
(1 + KH K m )

K m (1 + K H K m ) 

i
(
ω H K m 0.0386 −1 + 3.26ω H τ m − 0.6ω H 2 τ m 2 )

Kc (84 ) =  K H −
[
0.822 1 − 0.549ω H τ m + 0112
. ωH τm 2 2
]  , T ( 84)
=
Kc
( 84 )
(1 + K H K m )

K m (1 + K H K m ) 

i
(
ω H K m 0.0142 1 + 6.96ω H τ m − 177
. ω H 2 τm2 )

Kc (85) =  KH −
[
0.786 1 − 0.441ω H τ m + 0.0569ω H 2 τ m 2 ]  , T ( 85)
=
Kc
( 85)
(1 + K H Km )

K m (1 + K H K m ) 

i
(
ω H K m 0.0172 1 + 4.62ω H τ m − 0.823ω H 2 τ m 2 )
τ 
2
Servo– nearly 0471
. Ku ξ ≤ 0613
. + 0613
.
τm
+ 0117
.  m 
Tm1  Tm1 
( 8 6) 5 ( 86)
Kc Ti
ITAE – Hwang [60]
(continued) 0471
. Ku
τ 
2
Kc(8 7) Ti (87 ) K mω u ξ ≤ 0613

. + 0613
.
τm
Tm1
+ 0117
.  m 
 Tm1 
2
0471
. Ku τm  τm 
ξ > 0649
. + 058
. . 
− 0005 
Tm1  Tm1 
Servo – nearly K mω u
( 8 8) ( 88)
Kc Ti
minimum IAE, ISE,
ITAE – Hwang [60] 0471
. Ku
2
Kc(8 9) Ti (89) K mω u ξ > 0649
. + 058
.
τm
Tm1
. 
− 0005
 τm 

 Tm1 
6
2
0471
ξ > 0649
. + 058
. . 
− 0005 
Tm1  Tm1 
Model: Method 3
Kc( 90) Ti (90) K mω u
2
0471
ξ > 0649
. + 058
. . 
− 0005 
Tm1  Tm1 
K mω u
( 91) ( 91)
Kc Ti
0471
. Ku τm  τ 2
ξ ≤ 0649
. + 058
.
Tm1
.  m 
− 0005
 Tm1 
,
K mω u
( 92) ( 92 )
Kc Ti
τm  τ 2
ξ > 0613
. + 0613
. .  m 
+ 0117
Tm1  Tm1 
0471
. Ku τm  τ 2
ξ ≤ 0649
. + 058
.
Tm1
.  m 
− 0005
 Tm1 
,
Kc( 93) Ti (93) K mω u  τ 2
τm
ξ > 0613
. + 0613
. .  m 
+ 0117
Tm1  Tm1 
5

Kc (86) =  KH −
128
ω τ
. [
. ( 0.542) H m 1 − 0986 ε + 0558
. ε 2 
, Ti (86) =
] K c (1 + K H K m )
( 86)

 K m (1 + K H Km ) 
 ω H K m 0.0476 1 + 0.996 ln[ ω H τ m ] 1 + 2.13 ε − 113
. ε2 ( )( )

Kc (87 ) =  K H −
114 [
. 1 − 0466
. ω H τ m + 0.0647ω H 2 τ m 2 ]  , T ( 87)
=
Kc
( 87)
(1 + K H K m )

K m (1 + K H K m ) 

i
(
ω H K m 0.0609 − 1 + 197
. ω H τ m − 0.323ω H 2 τ m 2 )
6 ( 88)


= KH −
[
0.794 1 − 0.541ω H τ m + 0126
. ω H 2 τ m2 ]  , T ( 88)
=
Kc
( 88)
(1 + KH K m )
( )
Kc
 K m (1 + K H K m )  i
ω H K m 0.0078 1 + 8.38ω H τ m − 197
. ω H 2 τm2
 

Kc (89 ) =  K H −
[
0.738 1 − 0415
. ω H τ m + 0.0575ω H 2 τ m 2 ]  , T ( 89)
=
Kc
( 89 )
(1 + K H K m )

K m (1 + K H K m ) 

i
(
ω H K m 0.0124 1 + 4.05ω H τ m − 0.63ω H 2 τ m 2 )

Kc ( 90) =  K H −
115 [
. ( 0.564) H m 1 − 0.959 ε + 0.773 ε 2 
ω τ
]
, Ti (90) =
K c (1 + K H K m )
( 90)

 K m (1 + K H K m ) 
 ω H K m 0.0335 1 + 0947
. ln[ ω H τ m ] 1 + 19 (
. ε − 1.07 ε 2 )( )

Kc ( 91) =  KH −
107 [
. 1 − 0.466ω H τ m + 0.0667ω H 2 τ m 2 ] , T ( 91)
=
Kc
( 91)
(1 + KH K m )

K m (1 + K H K m ) 

i
(
ω H K m 0.0328 −1 + 2.21ω H τ m − 0.338ω H 2 τ m 2 )

Kc ( 92) =  K H −
[
0.789 1 − 0.527ω H τ m + 011
. ω H τm 2 2
]  , T ( 92)
=
Kc
( 92)
(1 + K H K m )

K m (1 + KH K m ) 

i
ω H K m 0009
. (
1 + 9.7ω H τ m − 2.4ω H2 τ m 2 )

Kc ( 93) =  KH −
[
0.76 1 − 0.426ω H τ m + 0.0551ω H 2 τ m 2 ] , T ( 93)
=
Kc
( 93)
(1 + KH K m )

K m (1 + K H K m ) 

i
(
ω H K m 0.0153 1 + 4.37ω H τ m − 0.743ω H 2 τ m 2 )
Servo – nearly 0471
. Ku ξ ≤ 0649
. + 058
.
τm  τ 2
.  m 
− 0005 ,
Tm1  Tm1 
( 94) 7 ( 94 )
Kc Ti
τm τ  2
ITAE – Hwang [60] ξ > 0613
. + 0613
.
Tm1
.  m 
+ 0117
 Tm1 
(continued)
0471
. Ku τm  τ 2
ξ ≤ 0649
. + 058
.
Tm1
.  m 
− 0005
 Tm1 
,
K mω u
( 95) (95)
Kc Ti
τm  τ 2
ξ > 0613
. + 0613
. .  m 
+ 0117
Tm1  Tm1 
Minimum IAE 0.6173Ku 0.38Tu 015

. Tu Tm 2 Tm2 + τ m = 0.25
regulator – ultimate 0.6766Ku 0.33Tu 019T
. u Tm 2 Tm2 + τ m = 0.5
cycle - Shinskey [16]
– page 151. 0.7874K u 0. 26Tu 0.21Tu Tm 2 Tm2 + τ m = 0.75
Model: Method 10.
Direct synthesis
πTm1 2Tm1 , 0.5Tm1
,
Gain and phase Am Km τ m Sample A m ,φ m
margin – Hang et al. provided. Model has a
[35] A m = 1.5, φm = 30 o A m = 3.0 , φm = 60 o A m = 5.0, φ m = 72 o
repeated pole ( Tm1 )
Model: Method 4 A m = 2 .0, φm = 45o A m = 4.0 , φ m = 67 .5o
ω p Tm1 1
Gain and phase A m Km 4ω p τ m
2
1 Tm1 > Tm 2 . Sample
margin – Ho et al. 2ω p − + Tm 2 A m ,φ m provided
[140] π Tm1
. πA m (A m − 1)
A m φ m + 05
A m = 2 .0, φm = 45 o
A m = 3.0 , φm = 60 o A m = 4.0, φ m = 70 o ωp =
Model: Method 1 (A m
2
)
−1 τm
Gain and phase Kc( 96) Ti (96 ) Td ( 96) τm τm

margin – Ho et al. 2ξ m ≥ , ≤1
Tm Tm
[141]
A m = 2 .0, φm = 45o A m = 3.0 , φm = 60 o A m = 4.0, φ m = 70 o Sample A m ,φ m
Model: Method 1
provided *
Gain and phase Kc (97) Ti (97 ) Td ( 97) ω nτ m < 2ξ m ; Sample
margin – Ho et al. A m ,φ m provided
[142]
A m = 2 .0, φm = 45o A m = 3.0 , φm = 60 o A m = 4.0, φ m = 60 o
Model: Method 1
7

Kc ( 94) =  K H −
122 [
. ( 0.55) H m 1 − 0.978 ε + 0.659 ε 2 
ω τ
]
, Ti (94) =
K c (1 + K H K m )
( 94)

 K m (1 + K H K m ) 
 ω H K m 0.0421 1 + 0969
. (
ln[ ω H τ m ] 1 + 2.02 ε − 111
. ε2 )( )

Kc ( 95) =  K H −
111 [
. 1 − 0467
. ω H τ m + 0.0657ω H 2 τ m 2 ] , T ( 95)
=
Kc
( 95)
(1 + KH K m )

K m (1 + KH K m ) 

i
(
ω H K m 0.0477 − 1 + 2.07ω H τ m − 0.333ω H 2 τ m 2 )
 τ  π Tm1
2
 πξ m + π − 2 m  , Ti = Tm1( πξ m + π − 2τ m ) , Td =
2 2
Kc ( 96) =
( 96) ( 96)
πA m K m  Tm1  π 2( πξ mTm1 + π Tm1 − 2τ m )
 πξ m
2ω pTm1
2
 ( 97) 2 2  πξ m 
Kc
( 97)
=  + π − 2ω p τm  , Ti = Tm1  + π − 2ω p τ m  ,
π Am  ω p Tm1  π  Tm1ω p 
πTm1
Td ( 97) =
 πξ 
2 m + π Tm1 − 2Tm1ω p τm 
 ωp 
 8πτm ξ m 
 π + π2 +
 Tm1 
( )
 A m 2 − 1 − 2πA m ( A m − 1)
* φm <
4A m
ξ m > 0.7071 or
07071
. τm
0.05 < < 015
.
ξ m Tm1 05
. Tm1 Tm1 2ξ m 2 − 1
Wang et al. [143] Km τ m 2ξ m Tm1 ξm 0.7071τm
> 1, ξ m ≥ 1
Model: Method 8 Tm1 2 ξ m 2 − 1
ξ mτ m
or 0.05 < < 015
. ,
Tm1
ξmτ m
> 1, ξ m < 1
Tm1
Minimum of 05
. Tm1 ξ m ≤ 0.7071 and
2 Tm1
2 −
Tm τ m
ξm 07358
. ξ m Tm1 2ξ m Tm1 ξm ξmτ m
e , 0.15 ≤ ≤1
Km Km τ m Tm1
ξ m ≤ 1 or
Gain and phase ξ m > 1 with
( 98) 8 ( 98 ) ( 98)
Kc Ti 0.25Ti
margin – Leva et al.
. π − φm >
05
[34]
3τ m 
ξ m + ξ m − 1
2
Model: Method 5 Tm1  

ξ m > 1 with
3τ m 
ξ m − ξ m −1 
2
Tm1  
Kc( 99) Ti (99 ) Td ( 99)
< 0.5π − φm ≤
3τ m 
ξ m + ξ m − 1 ,
2
Tm1  
φ m = 70 0 at least
( 1 + ω T ) + 4ξ ω
2
 2ξ τ T ( 05
1  m m m1 . π − φm ) 
ω cn Ti cn
2
m1
2
m
2
cn
2
Tm1
2

, ω cn = 4.07 − φ m + tan −1 
8
=
( 98)
Kc 
(1 − T T ω ) + T ω τm 
 ( 0.5π − φ m ) Tm1 − τ m 
2 
2 2 2
Km 2 2 2
i d cn i cn 
1 
  ω cn 2 +  2ξ m ξ m 2 − 1 − 1
 2ξ mω cnTm1   Tm12  
  , Kc ( 99) = ω cnTm(99
2
2
Ti( 98) = tan 05 . π − tan− 1
. ω cn τ m + φ m − 05 1 ,
ω cn    ω cn Tm1 − 1 
2 2
K mTd ) ω cn + z
2 2
Tm1z +  ξ m + ξ m 2 − 1
ω cn   Tm1
z= , Ti (99 ) = , Td (99) =
      
 z ξ m + ξ m 2 − 1 Tm1z +  ξ m + ξ m2 − 1 
−1  ω cn Tm1     
tan φ m − 05
. π + ω cnτ m + tan  
   ξ m + ξm 2 − 1  
    
Gain and phase ξ m Tm1 Tm1
1047
. 2ξ m Tm1 05
.
margin – Wang and Km τ m ξm Am = 3 , φ m = 60 0
Shao [144]
ξ mTm1 Tm1
2.094 2ξ m Tm1 05
.
Model: Method 8 K mτ m ξm A m = 15
. , φ m = 30 0
ξ mTm1 Tm1
1571
. 2ξ m Tm1 05
.
Authors quote K mτ m ξm A m = 2 , φ m = 450
tuning rule for
Am = 3 , φ m = 60 0 ; ξ m Tm1 Tm1
0.785 2ξ m Tm1 05
.
K mτ m ξm A m = 4 , φ m = 67.50
other tuning rules
obtained using ξ mTm1 Tm1
0.628 2ξ m Tm1 05
.
authors method K mτ m ξm A m = 5 , φ m = 72 0
Pemberton [145] 2( Tm1 + Tm2 ) Tm1 + Tm2 Tm1Tm2
Model: Method 1 3Km τ m Tm1 + Tm 2
τm
Tm1 + Tm2 . ≤
01 ≤ 10
. ;
Pemberton [24] (Tm1 + Tm2 ) Tm 1Tm2 Tm1
K mτ m Tm1 + Tm 2 τm
Model: Method 1 0.2 ≤ ≤ 10
.
Tm 2
τm
Tm1 + Tm2 . ≤
01 ≤ 10
.
Pemberton [145] 2( Tm1 + Tm 2 ) Tm1 + Tm 2 Tm1
3Km τ m 4 τm
Model: Method 1 0.2 ≤ ≤ 10
.
Tm 2
Suyama [100] Tm1 + Tm 2 Tm 1Tm2
Model: Method 1 2K m τ m Tm1 + Tm2 Tm1 + Tm 2
Smith et al. [146] Tm1 + Tm 2 Tm 1Tm2
Model: Method not K m ( λ + τm ) Tm1 + Tm2 Tm1 + Tm 2
stated
Chiu et al. [29] λTm1 + Tm2 Tm 1Tm2 λ variable; suggested
K m (1 + λτm ) Tm1 + Tm2 Tm1 + Tm 2 values are 0.2, 0.4, 0.6
Model: Method 6 and 1.0.
λTm 1 Tm1 > Tm2 . λ = pole of
Wang and Clemens K m (1 + λτ m ) Tm1 Tm2 specified closed loop
[147] overdamped response.
2λξ m Tm1 Tm1 Underdamped
K m (1 + λτ m )
Model: Method 9 2ξ m Tm1 response
2ξ m
Gorez and Klan 2ξ m Tm1 Tm1 Non-dominant time
[147a] K m (2ξ m Tm1 + τ m ) 2ξ m Tm1 2ξ m delay
Model: Ideal
Miluse et al. [27a] 0. 368 (Tm1 + Tm 2 ) Tm1 + Tm2 Tm 1Tm2 Closed loop overshoot
K mτ m Tm1 + Tm 2 = 0%
Model: Method not
0. 514 (Tm1 + Tm 2 ) Tm1 + Tm2 Tm 1Tm2 Closed loop overshoot
specified
K mτ m Tm1 + Tm 2 = 5%
Overdamped 0. 581( Tm1 + Tm 2 ) Tm1 + Tm2 Tm 1Tm2 Closed loop overshoot
process; Km τ m Tm1 + Tm 2 = 10%
Tm1 > Tm 2
0. 641( Tm1 + Tm 2 ) Tm1 + Tm2 Tm 1Tm2 Closed loop overshoot
Km τ m Tm1 + Tm 2 = 15%
Miluse et al. [27a] 0. 696 (Tm1 + Tm 2 ) Tm1 + Tm2 Tm 1Tm2 Closed loop overshoot
(continued) K mτ m Tm1 + Tm 2 = 20%

K m τm Tm1 + Tm 2 = 25%
0. 801( Tm1 + Tm 2 ) Tm1 + Tm2 Tm 1Tm2 Closed loop overshoot

Km τ m Tm1 + Tm 2 = 30%
0. 853( Tm1 + Tm2 ) Tm1 + Tm2 Tm 1Tm2 Closed loop overshoot

K m τm Tm1 + Tm 2 = 35%

K mτ m Tm1 + Tm 2 = 40%

K mτ m Tm1 + Tm 2 = 45%
1.008 ( Tm1 + Tm 2 ) Tm1 + Tm2 Tm 1Tm2 Closed loop overshoot

K m τm Tm1 + Tm 2 = 50%
Miluse et al. [27a] 0. 736ξ mTm1 2ξ m Tm1 Tm1 Closed loop overshoot
Km τ m 2ξ m = 0%
Model: Method not
specified 1.028 ξ m Tm1 2ξ m Tm1 Tm1 Closed loop overshoot
K m τm 2ξ m = 5%
Underdamped 1.162 ξ m Tm1 2ξ m Tm1 Tm1 Closed loop overshoot
process; K m τm 2ξ m = 10%
0.5 < ξ m ≤ 1
1.282 ξ m Tm1 2ξ m Tm1 Tm1 Closed loop overshoot
K m τm 2ξ m = 15%

K m τm 2ξ m = 20%

K m τm 2ξ m = 25%

K m τm 2ξ m = 30%

K m τm 2ξ m = 35%

K m τm 2ξ m = 40%

K m τm 2ξ m = 45%

K mτ m 2ξ m = 50%
Miluse et al. [27b] 0. 736 Tm1 2Tm1 0.5Tm1 Tm1 = Tm 2
Model: Method 14 K m τm
‘Re-tuning’ rule. ω 0 =
Seki et al. [147b] ωc
2
tan θ c + 2 + tan 2 θ c freq. when controlled
Model: Method not G c ( jω 0 ) cos θ c 0.5Ti system goes unstable
ω0
2
specified ωc
(crossover freq.);
ω c = new crossover
freq. ; θ c = phase lag
4+β β 4+β 1 4 1 1≤ β ≤ 2 ;
Autotuning – 4 2 2 G p ( jω ) 0
β ω135 0 4 + β ω135 0 τm < 0.25Tm1
135
Landau and Voda
[148]
3 3.2 08
.
Model: Method not G p ( jω u ) ωu ωu ω uτ m ≤ 016
.
relevant
19
. 4 1
G p ( jω u ) ωu ωu . < ω u τ m ≤ 0.2
016
Robust
λ varies graphically
Brambilla et al. [48] – Tm1 + Tm 2 + 0.5τ m Tm 1 + Tm2 + 05
. τm Tm1 Tm2 + 0.5( Tm1 + Tm2 ) τ m with τm (Tm1 + Tm 2 )
values deduced from K m τ m ( 2λ + 1) Tm1 + Tm 2 + 05
. τm 0.1 ≤ τ m ( Tm1 + Tm 2 ) ≤ 10
graph τ m (Tm1 + Tm 2 ) τ m (Tm1 + Tm 2 ) τ m (Tm1 + Tm 2 )
λ λ λ
Model: Method 1
0.1 0.50 1.0 0.29 10.0 0.14
0.2 0.47 2.0 0.22
0.5 0.39 5.0 0.16
1.00 ξ m Tm τm A m = 3.14 ,
Chen et al. [53] τ mK m 2ξ m τm 2ξ m φ m = 61. 40 ,
Model: Method 1 Ms = 1. 00
1.22 ξ m Tm τm A m = 2.58 ,
τ mK m 2ξ m τm 2ξ m φ m = 55. 00 ,
Ms = 1. 10
1.34 ξ m Tm τm A m = 2. 34 ,
τ mK m 2ξ m τm 2ξ m φ m = 51.6 0 ,
Ms = 1. 20
1.40 ξ m Tm τm A m = 2. 24 ,
τ mK m 2ξ m τm 2ξ m φ m = 50.0 0 ,
M s = 1. 26
1.44 ξ m Tm τm A m = 2.18 ,
τ mK m 2ξ m τm 2ξ m φ m = 48. 70 ,
Ms = 1. 30
1.52 ξ m Tm τm A m = 2. 07 ,
τ mK m 2ξ m τm 2ξ m φ m = 46. 50 ,
Ms = 1. 40
1.60 ξ m Tm τm A m = 1.96 ,
τ mK m 2ξ m τm 2ξ m φ m = 44.10 ,
Ms = 1. 50
2
Tm1
Ti − 2ξ mTm1 + −
Lee et al. [55] 2 λ − τm Desired closed loop
2
Ti 2
Ti
2ξ mTm1 −
K m ( 2λ + τ m ) 2( 2λ + τ m ) e− τ m s
τm response =
3
Model: Method 1
( λs + 1) 2
6Ti ( 2λ + τm )
Tm1Tm2
Ti − Tm1 − Tm2 +
Lee et al. [55] Ti 2 λ2 − τ m 2 Ti Desired closed loop
Tm1 + Tm2 −
(continued) K m ( 2λ + τ m ) 2( 2λ + τ m ) τm 3 e− τ m s
− response =
6Ti ( 2λ + τ m ) ( λs + 1) 2
 τ m3  Desired closed loop
 Tm1 2 − 
6(λ + τ m ) 
Ti τm
2 
Ti − 2ξ m Tm1   e− τ m s
2ξ mTm1 +  
K m( λ + τ m )
response =
2( λ + τ m ) ( λ s + 1)
Ti
 
 
 τ m3  Desired closed loop

 Tm1Tm 2 − 
 (6 λ + τm ) e− τ m s
τm
2
Ti Ti − ( Tm1 + Tm2 )  
Tm1 + Tm2 +  
K m( λ + τ m )
Ti response =
2 ( λ + τm ) 


 ( λ s + 1)
Km e− sτ m
K m e− sτ m
Table 70: PID tuning rules - SOSPD model or - filtered controller

(1 + sTm1 )(1 + sTm 2 ) Tm12 s2 + 2ξ m Tm1s + 1
 1  1
G c ( s) = Kc  1 + + Td s . 1 tuning rule.
 Ti s  Tf s + 1
Robust
Model has a repeated
Hang et al. [35] 2Tm1 + τ m Tm1 + 05
. τm Tm1τm pole (Tm1 ) .
2( λ + τ m ) K m 2Tm1 + τm λ > 025
. τm .
Model: Method 4
λτ m
Tf =
2( λ + τ m )
Km e− sτ m
K m e− sτ m
Table 71: PID tuning rules - SOSPD model or - filtered controller

(1 + sTm1 )(1 + sTm 2 ) Tm12 s2 + 2ξ m Tm1s + 1
 1  b s+1
G c ( s) = Kc  1 + + Td s 1 . 1 tuning rule.
 Ti s  a 1s + 1
Robust
Jahanmiri and Fallahi 2ξ m Tm1 b1 = 0.5τ m
[149] K m (τ m + λ) 2ξ m Tm1 Tm1 λτ m
a1 =
λ = 0.25τ m + 01. ξ m Tm1 2ξ m 2( τ m + λ )
Model: Method 6
Km e− sτ m
K m e− sτ m
Table 72: PID tuning rules - SOSPD model or - classical controller

(1 + sTm1 )(1 + sTm 2 ) Tm12 s2 + 2ξ m Tm1s + 1
 
 1   1 + Td s 
G c ( s) = Kc  1 +   . 7 tuning rules.
 Ti s   1 + Td s 
 
 N 
 −
T m1
  −
1.2 T m1

Minimum IAE - 1
Kc ( 100 ) 
τm 15
. −e 1.5τ m  
0.56τ m 1 − e τ m + Tm2
    ≤3
Shinskey [59] – page     τm
159
  −
T m2
 0.6Tm2
 1 + 0.9 1 − e τ m

Model: Open loop   

25
. Tm1 Tm2
τm + 0.2Tm2 τm + 0.2Tm2 >3
K mτ m τm
0800
. Tm1 . ( Tm2 + τ m )
15 . ( Tm2 + τ m )
060 Tm 2
= 025
.
** Minimum IAE - Km τ m Tm 2 + τm
Shinskey [59]
0770
. Tm1 . ( Tm2 + τ m )
12 0.70( Tm 2 + τ m ) Tm 2
= 0.5
K mτ m Tm2 + τ m
0833
. Tm1 . ( Tm2 + τ m )
075 . ( Tm2 + τ m )
060 Tm 2
= 0.75
Km τ m Tm2 + τ m
** Minimum IAE - 059
. Ku 0. 36Tu 0. 26Tu τm T
Shinskey [17] = 0.2 , m 2 = 0.2
Tm1 Tm1
085
. Tm1 τm T
. τm . τm = 0.2 , m 2 = 01
.
** Minimum IAE K mτ m 198 086 Tm1 Tm1
Shinskey [17]
087
. Tm1 Tm 2
2.30τ m . τm
165 = 0.2
Km τm Tm1
100
. Tm1 Tm 2
2.50τm . τm
200 = 0.5
K mτ m Tm1
125
. Tm1 2.75τm 2.75τ m Tm 2
= 10
.
K mτ m Tm1
100
1
Kc (100) =
  −
1.2 Tm1
 Km τ m  Tm2  Tm 2  
2
 48 + 57 1 − e   
τm
1 + 0.34 − 0.2
   Tm1  τm  τ m  

Minimum ISE –
McAvoy and α βτ m χτ m
Johnson [83] – Km
deduced from graph χ χ χ
ξm τm α β ξm τm α β ξm τm α β
Model: Method 1 Tm1 Tm1 Tm1
1 0.5 0.7 0.97 0.75 4 0.5 3.0 1.16 0.85 7 0.5 5.4 1.19 0.85
N = 20 1 4.0 7.6 3.33 2.03 4 4.0 22.7 1.89 1.28 7 4.0 40.0 1.64 1.14
1 10.0 34.3 5.00 2.7
1 0.5 0.9 1.10 0.64 4 0.5 3.2 1.33 0.78 7 0.5 5.9 1.39 1.04
N = 10 1 4.0 8.0 4.00 1.83 4 4.0 23.9 2.17 1.17 7 4.0 42.9 1.89 1.37
1 10.0 33.5 6.25 2.43
Direct synthesis
3
Astrom et al. [30] –  3τ m  Tm1 + Tm2 Tm 1Tm2 N = 8 - Honeywell
Method 13 K m 1 +  Tm1 + Tm 2 UDC6000 controller
 Tm1 + Tm 2 
αTm1
Smith et al. [26] – K mτ m Tm1 Tm2
deduced from graph ρ ρ ρ N = 10
ξm α ξm α ξm α
[ ρ = τ m ( Tm1 + Tm2 ) ] 6 ≥ 0.02 0.51 3 ≥ 004
. 0.50 2 ≥ 006
. 0.45
1.75 0.27 0.46 1.75 0.13 0.42 1.75 0.07 0.39
Model: Method not 1.75 0.07 0.36 1.5 0.33 0.44 1.5 0.17 0.38
stated 1.5 0.11 0.33 1.5 0.08 0.28 1.0 0.50 0.40
1.0 0.25 0.46 1.0 0.17 0.48 1.0 0.13 0.49
Km e− sτ m
K m e− sτ m
Table 73: PID tuning rules - SOSPD model or - alternative classical

(1 + sTm1 )(1 + sTm 2 ) Tm12 s2 + 2ξ m Tm1s + 1
 1   1 + NTd s 
controller G c ( s) = Kc  1 +   . 1 tuning rule.
 Ti s   1 + Td s 
******
0.7 0. 3
08
. Tm1 Tm2
Hougen [85] . Tm1 + Tm 2
05 τm Tm1 Tm 2 N=10
τm
3
0.8 0.2
Model: Method 1 084
. Tm1 Tm 2
0.53Tm1 + 1.3Tm2 0.08( τ m Tm1Tm2 )
0.28
N=30
τm
Km e− sτ m
K m e− sτ m
Table 74: PID tuning rules - SOSPD model or - alternative non-

(1 + sTm1 )(1 + sTm 2 ) Tm12 s2 + 2ξ m Tm1s + 1
 1
interacting controller 1: U( s) = K c  1 +  E ( s) − Kc Td sY ( s) . 3 tuning rules.
 Ti s
 − m1 
T
 −
1.2 Tm1

Minimum IAE - 1 ( 101) τ m  0.5 + 1. 4[ 1 − e 1.5 τ m ] 0.42τ m  1 − e τm  + Tm2
Kc    
    ≤3
Shinskey [59] – page τm
159   T
− m2  0.6Tm2
 1 + 0.48 1 − e τ m 
  
  
Model: open loop
method not specified 333
. Tm 1 τ m + 0.2Tm2 τ m + 0.2Tm2 Tm 2
>3
Km τm τm
118
. Tm1 τm T
2.20τ m 0.72τ m = 0.2 , m 2 = 01
.
Minimum IAE - Km τ m Tm1 Tm1
119. Model: method 125
. Tm1 Tm 2
not specified. = 0.2
K m τm 2.20τm . τm
110 Tm1
167
. Tm1 Tm 2
2.40τ m . τm = 0.5
K mτ m 165 Tm1
25
. Tm1 Tm 2
2.15τ m 2.15τ m = 10
.
K mτ m Tm1
Minimum IAE -
Shinskey [17] – page τm T
121. Model: method 085
. Ku 035
. Tu 017
. Tu = 0.2 , m 2 = 0.2
Tm1 Tm1
not specified.
100
1
Kc (101) =
  −
1.5Tm1
 Km τ m  Tm2  Tm 2  
2
 38 + 401 − e   
τm
1 + 0.34 − 0.2
   Tm1  τm  τ m  

Km e− sτ m
K m e− sτ m
Table 75: PID tuning rules - SOSPD model or - series controller

(1 + sTm1 )(1 + sTm 2 ) Tm12 s2 + 2ξ m Tm1s + 1
 1 
G c ( s) = K c 1 +  (1 + Td s) . 1 tuning rule.
 Ti s 
Tm1 > Tm2 . Maximum
Least mean square 2Tm1 Tm1 Tm2 sensitivity = 1.9, Gain
error - Haalman [23] 3τ m K m margin = 2.36, Phase
Model: Method 1 margin = 50 0
Km e− sτ m
K m e− sτ m
Table 76: PID tuning rules - SOSPD model or - non-interacting

(1 + sTm1 )(1 + sTm 2 ) Tm12 s2 + 2ξ m Tm1s + 1
 
  
1  Td s  . 2 tuning rules.
controller U (s) = K c 1 +  E( s) −
 Y (s )
 Ti 
s T s 
 1+ d

 N 
Servo – Min. IAE - 0<
Tm 2
≤1;
Huang et al. [18] Kc ( 102 ) 2
Ti ( 102)
Td ( 102 ) Tm1
Model: Method 1 τm
0.1 ≤ ≤ 1 ; N =10
Tm1
1  
0.0865
τ T τ T  τm 
2
Kc (102 ) = 7.0636 + 66.6512 m − 137 .8937 m2 − 122 .7832 m m2 2 + 261928
.   
Km  Tm1 Tm1 Tm1  Tm1  

1  
2.6405 1.0309 2.345 1.0570
 τm   Tm 2  T  T  T
+ 336578
.   + 30098
.   − 10.9347 m 2  + 141511
.  m2  + 29.4068 m2 
Km   Tm1   Tm1   Tm1   Tm1  τm 

1  
− 0.9450 −0.9282 0.8866
Tm2  τ m  Tm 2  τ m  τ m  Tm2 
+ 34.3156   − 701035
.   + 152.6392   
Km  Tm 1  Tm1  Tm1  Tm1  Tm1  Tm1  

1  τ T 
0 .8148 τ m T m2
τ m Tm 2
τ 
− 0.4062 
+ − 47.9791 m  m 2  − 57.9370e + 10.4002e T
T m1 m1
+ 6.7646e Tm1 2
+ 7.3453 m  
Km  Tm1  Tm1   Tm  
 
2
τ m Tm2  Tm2 
2
 τm  
3
= Tm1 0.9923 + 0.2819 − 0.2679 − 1.4510  − 0.6712 + 0.6424   + 2 .504   

( 102)
Ti
 Tm1   Tm1   Tm1  
2
 Tm1 Tm1 Tm1
 T τ 
2
τ T 
2
T 
3
 τm 
4
τ m  Tm2  
4
+ Tm1 2.5324 m 2  m  + 2.3641 m  m 2  + 2.0500  m 2  − 18759

.   + 08519
.   
 Tm1  Tm1  Tm1  Tm1   Tm1   Tm1  Tm1  Tm1  

 Tm2  τm 
3 2
 τ m   Tm2 
2
τ T 
3
 Tm 2  
4
+ Tm 1 − 13496
.   − 34972
.     − 2.4216 m  m 2  − 31142
.   
 Tm1  Tm1   Tm1   Tm1  Tm1  Tm1   Tm1  
  τm 
5
Tm2  τ m 
4 2
 Tm2   τ m 
3 2
 τm   Tm 2 
3
 Tm 2  
5
+ Tm 1 05862
.   + 0.0797   + 0.985    + 12892
.     + 12108
.   
  Tm1  Tm1  Tm1   Tm1   Tm1   Tm1   Tm1   Tm1  
 τm Tm 2  τm 
2
τm Tm 2  Tm2 
2
 τm  
3
Td
( 102)

= Tm1 0.0075 + 0.3449 + 0.3924 − 0.0793  + 2.7495 + 0.6485  + 0.8089   
 Tm1   Tm1   Tm1  
2
 Tm1 Tm1 Tm1
 T τ 
2
τ  Tm2 
2
 Tm2 
3
 τm 
4

+ Tm1 − 9.7483 m 2  m  + 3.4679 m   − 58194
.   − 10884
.   
 Tm1  Tm1  Tm1  Tm1   Tm1   Tm1  
 T τ 
3 2
 τm   Tm 2 
2
τ T 
3
 Tm 2  
4
+ Tm 1 12.0049 m 2  m  − 14056
.     − 3.7055 m  m 2  + 100045
.   
 Tm1  Tm1   Tm1   Tm1  Tm1  Tm1   Tm1  
  τm 
5
Tm 2  τ m 
4 2
 Tm2   τ m 
3
 τ m   Tm 2  
2 3

+ Tm 1 0.3520  − 6.3603   − 32980
.     + 7.0404    
  Tm1  Tm1  Tm1   Tm1   Tm1   Tm1   Tm1  
  τ m   Tm 2 
4
 Tm 2 
5
 τm 
6
Tm 2  τm 
5
 Tm 2 
6

+ Tm1 14294
.    − 69064
.   + 0.0471  + 11839
.   + 17087
.   
  m1   m1 
T T  m1 
T  m1 
T Tm1  m1 
T  Tm1  
Servo-Min. IAE - ( 103) 3 ( 103) ( 103) τm
Kc (N=10) Ti Td 0 .4 ≤ ξ m ≤ 1 ; 0 .05 ≤ ≤1
Huang et al. [18] Tm1
 4
 τ m   Tm 2 
2 3
 τ m   Tm2 
3 2
 τ m   Tm 2 
4
τ m  Tm 2  
5

+ Tm1 1. 7444    − 12817
.     − 21281
.     + 15121
.   
  Tm1   Tm1   Tm1   Tm1   Tm1   Tm1  Tm1  Tm1  

1  
−1.9009
τ τ  τ 
3
Kc (103) = −81727
. − 32.9042 m + 319179
. ξ m + 38.3405ξ m  m  + 0.2079 m  
Km  Tm1  Tm1   Tm1  

1  
0.1571 1.2234
τ   τm 
+ 29.3215 m  + 359456
.   − 214045
. ξ m0 .1311 + 51159
. ξ m1.9814 − 219381ξ m1.737 
Km   Tm1   Tm 1  

1  
−0.1303 1.2008
 τm   τm  τ
+ − 17.7448ξ m   + 268655
. ξ m  − 52.9156 m ξ m1.1207 
Km   Tm1   Tm1  Tm1 

1  τ
τ m τ mξm
ξ m Tm1 
+ − 22.4297 m ξ m 0.3626 − 33331
. e T m1
+ 85175
. e ξm
− 15312
. e Tm1
+ 08906
. 
Km  Tm1 τm 

 τ τ 
2
τ τ 
2

= Tm1 11731 + 6.3082 m − 0.6937ξ m + 8.5271 m  − 24 .7291ξ m m − 6.7123ξ m  m  + 7 .9559ξ m 
( 103) 2
Ti .
 Tm1  Tm1  Tm1  Tm1  
  τm 
3
 τm 
4
 τm 
3
 τ 
+ Tm1  − 32.3937   − 271372
.   + 166.9272ξ m   + 363954
. ξ m2  m  
  Tm1   Tm1   Tm1   Tm1  

 2

2  τm  τm

+ Tm1 − 94.8879ξ m   − 22.6065 ξ m − 16084
3
. ξ m 3 + 29.9159ξ m 4 
  Tm1  Tm1 
 τ 
5
τ 
4
τ 
3
 τm 
2

+ Tm 1 49.6314 m  − 84.3776ξ m  m  − 938912
. ξ m 2  m  + 1101706
.   ξm 
3
  Tm1   Tm1   Tm1   Tm1  
  τm  4 τ 
6
τ 
5

+ Tm 1 − 251896
.   ξ m − 19.7569 ξ m − 12.4348  m  − 117589
5
. ξ m  m  + 55268
. ξ m6
  Tm1   Tm1   Tm1  
  τm 
4
 τm 
3
 τm 
2
τm 
+ Tm 1 68.3097  ξ m − 17.8663
2
 ξ m − 225926
3
.   ξ m + 95061
4
. ξm 5 
  Tm1   Tm1   Tm1  Tm1 
 τm  τm 
2
τm  τm  
3
Td
( 104)

= Tm1 0.0904 + 0.8637 − 0.1301ξ m + 4 .9601  + 14 .3899ξ m + 0.7170ξ m − 12.5311
2
 
 Tm1  Tm1  Tm1  Tm1  
 τ 
2
τ  τ  
4
+ Tm1 − 42.5012ξ m  m  − 214907

. ξ m2  m  − 69555
. ξ m 3 − 12.3016  m  
  Tm1   Tm1   Tm1  
 τ 
3
τ 
2
τ 
+ Tm1 102.9447ξ m  m  + 7.5855ξ m 2  m  + 19.1257 m ξ m 3 + 17.0952 ξ m 4 
  Tm1   Tm1  Tm1 
  τm 
5
 τm 
4 3 2

2  τm   τm 

+ Tm1 108688
.   − 17.2130ξ m   − 1100342
. ξm   + 50.6455  ξm 
3
  Tm1   Tm1   Tm1   Tm1  
 τ  τ 
6
τ 
5

+ Tm 1 − 16.7073 m  ξ m 4 − 16.2013ξ m5 − 0. 0979 m  − 109260
. ξ m  m  + 54409
. ξ m6
  Tm1   Tm1   Tm1  
Regulator – Min. IAE 0<
Tm 2
≤1;
- Huang et al. [18] Kc(104 ) 4 Ti (104) Td (104 ) Tm1
Model: Method 1 τm
0.1 ≤ ≤ 1 ; N =10
Tm1
  τm 
4
 τm 
3
 τm 
2
τm 5 
+ Tm 1 29.4445  ξ m + 216061
2
.   ξ m − 241917
3
.   ξ m + 62798
4
. ξm 
  Tm1   Tm1   Tm1  Tm1 
1  
− 0.8058
τm Tm2 τ m Tm2  τm 
4
Kc (104) = 01098
. − 86290
. + 766760
. − 33397
. + 11863
.   
Km  Tm1 Tm1 Tm1 2
 Tm1  

1  
0.6642 2 .1482 0.8405 2.1123
 τm   τm  T   Tm2 
+ 231098
.   + 203519
.   − 52.0778 m 2  − 121033
.   
Km   Tm1   Tm1   Tm1   Tm1  

1  
0.5306 −1.0781 −0.4500
τ T  T τ  T τ 
+ 9.4709 m  m2  + 13.6581 m 2  m  − 19.4944 m2  m  
Km  Tm1  Tm1  Tm1  Tm1  Tm1  Tm1  

1  Tm 2 
1.1427 τ T τ m Tm 2
τ T 
m m2
+ − 28.2766 m  m2  − 191463
. e T + 8.8420e T
m1 m1
+ 7.4298e Tm1 2
− 114753
. 
Km  Tm1  Tm1  τm 
 
 τ T τ 
2
τ T T 
2
τ  
3
= Tm1 − 0.0145 + 2 .0555 m + 0.7435 m2 − 4.4805 m  + 1.2069 m m2 2 + 0.2584 m 2  + 7.7916  m  

( 104)
Ti
 Tm1 Tm1  Tm1  Tm1  Tm1   Tm1  
 Tm 2  τ m 
2
τ m  Tm2 
2
 Tm 2 
3
 τm 
4
Tm 2  τ m  
3

+ Tm1 − 6.0330   + 3.9585   − 30626
.   − 7.0263  + 7.0004   
 Tm1  Tm1  Tm1  Tm1   Tm1   Tm1  Tm1  Tm1  

 2
T   τ 
2
 τm   Tm 2 
3
 Tm 2 
4
 τm  
5
+ Tm1 − 2.7755 m 2   m  − 15769

.    + 31663
.   + 2 .4311  
  T m1   Tm1   Tm1   Tm1   Tm1   Tm1  
 T 
5
T τ 
4
T   τ 
2 3
 τ m   Tm2  τ
2 3
 Tm2 
5

+ Tm 1 − 0.9439 m 2  − 2.4506 m 2  m  − 0.2227 m2   m  + 19228
.     − 0.5494 m   
  Tm1  Tm1  Tm 1   Tm1   Tm1   Tm1   Tm1  Tm1  Tm1  
 τm Tm 2  τm 
2
τ mTm 2  Tm2 
2
 τm  
3
= Tm1 −0.0206 + 0.9385 + 0.7759 − 2.3820  + 2 .9230 − 3.2336  + 7.2774   

( 104)
Td
 Tm1   Tm1   Tm1  
2
 Tm1 Tm1 Tm1
 T  τm 
2
τ  Tm 2 
2
 Tm2 
3
 τm 
4

+ Tm 1 − 9.9017 m 2   + 2.7095 m   + 61539
.   − 111018
.   
 Tm1  m1 
T Tm1  m1 
T  m1 
T  Tm1  
 T τ 
3 2
 τ m   Tm2 
2
τ  Tm 2 
3
T 
4

+ Tm 1 10.6303 m2  m  + 57105
.     − 7.9490 m   − 6.6597 m 2  
 Tm1  Tm1   Tm1   Tm1  Tm1  Tm1   Tm1  
  τm 
5
T τ 
4 2
T   τ 
3
 τ m   Tm2  
2 3
+ Tm 1 80849
.   − 4.4897 m 2  m  − 7.6469 m2   m  + 21155
.     
  Tm1  Tm1  Tm1   Tm1   Tm1   Tm1   Tm 1  
  τ m   Tm 2 
4
 Tm 2 
5
 τm 
6
Tm 2  τ m 
5
 Tm 2  
6

+ Tm 1 50694
.    + 4.1225  − 2.274   + 0519
.   − 11295
.   
  Tm1   Tm1   Tm1   Tm1  Tm1  Tm1   Tm1  
 4
 τ  T 
2 3
 τ  T 
3 2
 τm   Tm 2 
4
τm  Tm2 
5

+ Tm 1 2.2875 m   m2  + 0.9524 m   m 2  − 16307
.     − 09321
.   
  m1   m1 
T T  m1   m1 
T T  m1   m1 
T T Tm1  Tm1  
Regulator - Min. IAE - ( 105) 5 ( 105) ( 105) τm
Kc Ti Td [N=10] 0 .4 ≤ ξ m ≤ 1 ; 0 .05 ≤ ≤1
Huang et al. [18] Tm1
Model: Method 1
1  
3 0.086
τm  τm   τm   τm 
5
Kc ( 105)
= −357307
. − 1419
. + 14023
. ξ m + 6.8618ξ m   − 0.9773  + 555898
. ξ m  
Km  Tm1  Tm1   Tm1   Tm  

1  
2 −1. 6624 −0.6951
τ  τm  τm  τ 
+ − 33093
. ξ m  m  + 538651
. ξ m 2 + 114911
. ξ m3 + 08778
.   − 29.8822 m  
Km   Tm 1  Tm1  Tm1   Tm1  

1  
− 0.4762 −2.1208
τ  τ 
+ 53535
.  m − 16.9807 ξ m1.1197 − 254293
. ξ m1.4622 − 01671
. ξ m 58981 + 0.0034ξ m  m  
Km   Tm1   Tm  

1  ξ m Tm1 
τ τ ξ
τm τ m 1.2103
m m m
+ − 250355
. ξm 3.0836
− 54.9617 ξm − 01398
. e T
− 82721
. ξ
e + 63542
. e T + 10479
.
m1 m
 m1
Km  Tm1 Tm1 τm 
 
 τm  τm 
2
τm  τm  
3
Ti
( 105)

= Tm1 0.2563 + 11.8737 − 1.6547ξ m − 16.1913  − 9 .7061ξ m + 3.5927 ξ m + 19.5201
2
 
 Tm1  Tm1  Tm1  Tm1  
 τ 
2
τ  τ  
4
+ Tm 1 − 14.5581ξ m  m  + 2.939ξ m 2  m  − 0.4592ξ m 3 − 34.6273 m  

  Tm1   Tm1   Tm1  
  τm 
3 2

2  τm  τ
+ Tm 1 50.5163ξ m   + 8.9259ξ m   + 8.6966 m ξ m − 6.9436ξ m 
3 4
  Tm1   Tm 1  Tm1 
  τm 
5
 τm 
4 3 2

2 τm   τm 

+ Tm 1 27.2386  − 20.0697ξ m   − 42.2833ξ m   + 85019
.   ξm 
3
  Tm1   Tm1   Tm1   Tm1  
 τ  τ 
6
τ 
5

+ Tm 1 − 12.2957 m  ξ m 4 + 8.0694 ξ m 5 − 7.7887 m  + 2.3012 ξ m  m  − 2.7691ξ m 6 
  Tm1   Tm1   Tm1  
  τm 
4
 τm 
3
 τm 
2
τ 
+ Tm 1 88984
.   ξ m + 102494
2
.   ξ m − 54906
3
.   ξ m + 4.6594 m ξ m 
4 5
  Tm1   Tm1   Tm1  Tm1 
 τm  τm 
2
τm  τm  
3
= Tm1 −0.021 + 3.3385 + 0.185ξ m − 0.5164  − 0.9643ξ m − 0.8815ξ m + 0.584   

( 105) 2
Td
 Tm1  Tm1  Tm1  Tm1  
 τ 
2
 τm  
4
2 τ 
+ Tm1  − 12513
. ξ m  m  + 13468 ξ m  m  + 2.3181ξ m − 52368   
3
. .
  Tm1   Tm1   Tm1  
 
 τ 
3 2

2 τ  τ
+ Tm1 153014 ξ m  m  + 119607 ξ m  m  − 2.0411 m ξ m − 31988ξm 
3 4
. . .
  Tm1   Tm1  Tm1 
 
  τm 
5
 τm 
4 3 2

2 τm   τm 

+ Tm1 34675
.   − 0.8219ξ m   − 15.0718ξm   − 18859
.   ξm 
3
  Tm1   Tm1   Tm1   Tm1  
 
 τ  4 τ 
6
τ 
5

+ Tm1  0.4841 m  ξ m + 2.2821ξ m − 0.9315 m  + 0529
. ξ m  m  − 06772 ξm 
5 6
.
  Tm1   Tm1   Tm1  
 
  τm 
4
 τm 
3
τ 
2
τ 
+ Tm1  −14212   ξ m + 71176   ξ m − 2.3636 m  ξ m + 0.5497 m ξ m 
2 3 4 5
. .
  Tm1   Tm1   Tm1  Tm1 
 
Km e− sτ
m
K m e− sτ m
Table 77: PID tuning rules - SOSPD model or - ideal controller with
(1 + sTm1 )(1 + sTm 2 ) Tm12 s2 + 2ξ m Tm1s + 1
( )
set-point weighting U(s) = Kc Fp R ( s) − Y( s) +
Kc
Ti s
[ Fi R(s) − Y(s)] + Kc Td s[ Fd R (s) − Y(s)] . 1 tuning rule.
Ultimate cycle
06
. Ku Repeated pole
Oubrahim and 16 − Km Ku 05. Tu 0125
. Tu τ
Fp = 13
. . < m < 3;
01
Leonard [138]
17 + Km Ku Fi = 1 Fd = Fp
2 Tm
10% overshoot
Model: Method not
06
. Ku Repeated pole
specified
38 05. Tu 0125
. Tu τ
Fp = . < m < 3;
01
29 + 35Km K u Fi = 1 Fd = Fp
2 Tm
20% overshoot
Km e− sτ m
K m e− sτ m
Table 78: PID tuning rules - SOSPD model or - non-interacting

(1 + sTm1 )(1 + sTm 2 ) Tm12 s2 + 2ξ m Tm1s + 1
 1 
controller U( s) = Kc  b + [ R (s) − Y(s) ] − ( c + Tds)Y( s) . 1 tuning rule.
 i 
Ts
Direct synthesis
b = 0.198
Hansen [150] Kc(105a) 6 . Ti ( 105a )
154 Td (105a) c = K c ( 105a ) − 1 K m ;
zero overshoot
Model: Method 1 b = 0.289
Kc(105a) . Ti (105a )
127 Td (105a) c = K c ( 105a ) − 1 K m ;
minimum IAE
b = 0.143
Kc(105a) . Ti (105a )
175 Td (105a) c = K c ( 105a ) − 1 K m ;
conservative tuning
[
2 Tm1Tm 2 + (Tm1 + Tm2 )τ m + 0.5τ m2 ] [ . ( Tm1 + Tm 2 )τ m 2 + 0167 ],
3
3 Tm1Tm2 τ m + 05 . τm3
6 (105a )
= , Ti (105a ) =
[T ]
Kc
[
9 Km Tm1Tm2 τ m + 0.5(Tm1 + Tm2 )τ m + 0167
. τm 2
]
3 2
m1Tm 2 + (Tm1 + Tm2 )τ m + 05
. τm 2
2[ T + (Tm1 + Tm 2 )τ m + 05 ]
2
m1Tm2 . τ m2 Tm1 + Tm2 − τ m
= −
( 105a )
[ ]
Td
3K m Tm1Tm 2 τ m + 0.5(Tm1 + Tm2 )τ m + 0167
. τm2 3 Km
Km e− sτ m
K m e− sτ m
Table 79: PID tuning rules - SOSPD model or - ideal controller with
(1 + sTm1 )(1 + sTm 2 ) Tm12 s2 + 2ξ m Tm1s + 1
 
 1 Td s 
filtered derivative G c ( s) = Kc  1 + +  . 2 tuning rules.
 Ts
i
sT
1+ d 
 N 
Direct synthesis
ω p Tm 1 1 Sample A m , φ m
Hang et al. [151] A m Km 4ω p 2 τ m 1 Tm2 provided. N = 20.
2ω p − +
π Tm1 Tm1 > Tm2
Model: Method 1
A m φ m + 0.5πA m (A m − 1)
A m = 2 .0 , φ m = 45o A m = 4.0 , φ m = 67.5 o A m = 5.0, φ m = 72 o ωp =
(A m
2
)
−1 τm
A m = 3.0 ,φ m = 60 o
Robust
Hang et al. [151] Tm1
K m( λ + τ m) Tm1 Tm2 N = 20. Tm1 > Tm2
Model: Method 1
K m e− sτ m
Table 80: PID tuning rules – SOSPD model - Two degree of freedom controller:
Tm1 s2 + 2ξ m Tm1s + 1
2
   
   
1 Td s β T s
 Tis T   Td 
 1+ d s  1+ s
 N   N 
tuning
τm
Taguchi and Araki ≤ 1.0 ; ξ m = 1. 0
Tm
[61a] (105 b) 7 (105 b ) (105b )
et al. [10]
τm
≤ 1.0 ; ξ m = 0.5
Tm
(105 c ) (105 c ) (105 c )
et al. [10]
Minimum ITAE - α = 0. 5840 , β = 1 ,
Pecharroman and 0.7236 K u 0.5247 Tu 0.1650 Tu N = 10,
Pagola [134a]
φ c = − 139.65 0
Model: Method 15 K m = 1 ; Tm = 1 ;
ξm =1
 
 
 2 3

7 (105 b)
=
1 
1.389 +
0. 6978  (105 b)
=  0.02453 + 4.104 τ m − 3. 434  τm  + 1.231 τ m  
Kc , T T    
Kc  τm  i m  Tm  Tm   Tm  

 [ − 0.02295 ] 2  
 Tm 
 τm  τm 
2
 τm 
3
 τ m  
4
Td
(105 b) 
= Tm 0.03459 + 1. 852 − 2.741  + 2.359   − 0. 7962   ,
 Tm  Tm   Tm   T m  

2 3 2
τm τ  τ  τ τ 
α = 0. 6726 − 0.1285 − 0.1371  m  + 0.07345  m  , β = 0. 8665 − 0.2679 m + 0.02724  m 
Tm  m
T  m
T T m  Tm 
 
 
 2 3

(105c )
=
1 
0.3363 +
0.5013  (105 c )
=  − 0.02337 + 4.858 τm − 5.522  τm  + 2.054  τ m  
Kc , T T     
Tm  Tm   Tm  
 [ − 0.01147 ] 2  
 Tm 
 τm  τm 
2
 τ m  
3
Td
(105 c ) 
= Tm 0.03392 + 2. 023 − 1.161  + 0.2826   ,
 Tm  Tm   Tm  

2 3 2
τm τ  τ  τ τ 
α = 0. 6678 − 0. 05413 − 0. 5680  m  + 0. 1699  m  , β = 0. 8646 − 0.1205 m − 0.1212  m 
Tm Tm   Tm  Tm  Tm 
Minimum ITAE - α = 0.585 , β = 1 ,
N = 10, φ c = −1460
Pagola [134b]
α = 0.584 , β = 1 ,
φ c = phase 0.727 K u 0.524 Tu 0.165 Tu
N = 10, φ c = −1400
crossover frequency; 0.672 Ku 0.532 Tu 0.161Tu
K m = 1 ; Tm = 1 ; N = 10, φ c = −1340
0.1 < τm < 10 α = 0.550 , β = 1 ,
0.669 Ku 0.486 Tu 0.170 Tu
N = 10, φ c = −1250
α = 0.543 , β = 1 ,
Model: Method 15 0.600 Ku 0.498 Tu 0.157 Tu
N = 10, φ c = −1150
α = 0.528 , β = 1 ,
0.578 K u 0.481Tu 0.154 Tu
N = 10, φ c = −1050
α = 0.504 , β = 1 ,
0.557 K u 0.467 Tu 0.149 Tu
N = 10, φ c = −930
α = 0.495 , β = 1 ,
0.544 Ku 0.466 Tu 0.141Tu
N = 10, φ c = −840
α = 0.484 , β = 1 ,
0.537 K u 0.444 Tu 0.144 Tu
N = 10, φ c = −730
α = 0.477 , β = 1 ,
0.527 K u 0.450 Tu 0.131Tu
N = 10, φ c = −630
α = 0.454 , β = 1 ,
0.521K u 0.440 Tu 0.129 Tu
N = 10, φ c = −520
α = 0.445 , β = 1 ,
0.515 K u 0.429 Tu 0.126 Tu
N = 10, φ c = −410
α = 0.433 , β = 1 ,
0.509 Ku 0.399 Tu 0.132 Tu
N = 10, φ c = −300
α = 0.385 , β = 1 ,
0.496 Ku 0.374 Tu 0.123 Tu
N = 10, φ c = −190
α = 0.286 , β = 1 ,
0.480 Ku 0.315 Tu 0.112 Tu
N = 10, φ c = −100
α = 0.158 , β = 1 ,
0.430 Ku 0.242 Tu 0.084 Tu
N = 10, φ c = −6 0
K m e− sτ m
Table 81: PID tuning rules - I 2PD model G m ( s) = - controller
s2
1
U(s) = Kc (1 + ) E( s) + Kc ( b − 1) R (s) − Kc TdsY( s) . 1 tuning rule.
Tis
Direct synthesis
Hansen [91a]
Model: Method 1 3.75K m τ m
2 5.4τ m 2. 5τ m b = 0.167
K m e −s τ m
Table 82: PID tuning rules – SOSIPD model (repeated pole) - Two degree of freedom controller:
s (1 + Tm1s) 2
   
   
1 Td s β T s
 Tis T   Td 
 1+ d s  1+ s
 N   N 
tuning
τm
Tm
[61a] (105 d) 8 (105 d ) (105d )
Model: Method 2
et al. [10]
Minimum ITAE - α = 0. 601 , β = 1 ,
N = 10, φ c = −1640
Pagola [134a]
α = 0.607 , β = 1 ,
φ c = phase 1.236 K u 0.427 Tu 0.149 Tu
N = 10, φ c = −1600
K m = 1 ; Tm = 1 ; N = 10, φ c = −1550
0.1 < τm < 10 α = 0.616 , β = 1 ,
0.842 Ku 0.538 Tu 0.154 Tu
N = 10, φ c = −1500
α = 0.605 , β = 1 ,
Model: Method 1 0.752 Ku 0.567 Tu 0.157 Tu
N = 10, φ c = −1450
α = 0.610 , β = 1 ,
0.679 Ku 0.610 Tu 0.149 Tu
N = 10, φ c = −1400
α = 0.612 , β = 1 ,
0.635 K u 0.637 Tu 0.142 Tu
N = 10, φ c = −1350
α = 0.610 , β = 1 ,
0.590 Ku 0.669 Tu 0.133 Tu
N = 10, φ c = −1300
α = 0.616 , β = 1 ,
0.551K u 0.690 Tu 0.114 Tu
N = 10, φ c = −1250
 
 
1  0. 5667 
= 0.1778 +
8 (105d )
Kc ,
Kc  τm
 + 0. 002325 
 Tm 
 τm  τm 
2
 τm 
3
 τm  
4
Ti
(105 d) 
= Tm 0.2011 + 11.16 − 14. 98  + 13. 70  − 4.835  
 

Tm  Tm   Tm   Tm  
2
 τ  τ τ 
= Tm  1. 262 + 0.3620 m  , α = 0. 6666 , β = 0. 8206 − 0.09750 m + 0.03845  m 
(105 d)
Td
 Tm  Tm  Tm 
Minimum ITAE - α = 0.609 , β = 1 ,
N = 10, φ c = −1200
Pagola [134a] –
continued α = 0. 611 , β = 1 ,
0.509 Ku 0.810 Tu 0.068 Tu
N = 10, φ c = −1180
K m (1 − sTm3 ) e− s τ m
Table 83: PID tuning rules - SOSPD model with a negative zero - controller with filtered
(1 + sTm1 )(1 + sTm2 )
 
 1 Td s 
derivative G c ( s) = Kc  1 + +  . 1 tuning rule.
 Ti s 1 + Td s 

 N 
Robust
Tm1 > Tm 2 > Tm3
Tm1 + Tm2 − Tm 3 Tm1 + Tm2 − Tm 3 Tm1 Tm2 − (Tm1 + T m2 − Tm3 )Tm 3
Chien [50] N=10; λ = [ Tm1 , τm ]
K m( λ + τ m )
Tm1 + T m2 − Tm3
Model: Method 1
2ξTm 2 − Tm3 2ξTm1 − Tm 3 Tm2 1 − (2 ξTm1 − Tm3 )Tm3 N=10; λ = [ Tm1 , τm ]
Km (λ + τ m ) 2ξ Tm1 − Tm3
K m (1 − sTm3 ) e− s τ m
Table 84: PID tuning rules - SOSPD model with a negative zero - classical controller
(1 + sTm1 )(1 + sTm2 )
 
 1   1 + Td s 
G c ( s) = Kc  1 +   . 1 tuning rule.
 Ti s   1 + Td s 
 
 N 
Robust
Tm 2 Tm2 Tm1 Tm1 > Tm 2
Chien [50] K m( λ + τ m ) N=10; λ = [ Tm1 , τm ]
Model: Method 1 Tm1 Tm1 Tm2 Tm1 > Tm 2

K m( λ + τ m ) N=10; λ = [ Tm1 , τm ]
K m (1 − sTm3 ) e− s τ m
Table 85: PID tuning rules - SOSPD model with a negative zero - series controller with
(1 + sTm1 )(1 + sTm2 )
 
 1  Td s 
filtered derivative G c ( s) = Kc  1 +  1 +  . 1 tuning rule.
 i  1 + d 
Ts Ts
 N 
Robust
Tm 2 Tm2 Tm1 − Tm3 Tm1 > Tm 2 > Tm3
Chien [50] K m( λ + τ m ) N=10; λ = [ Tm1 , τm ]
Model: Method 1 Tm1 Tm1 Tm 2 − Tm3 Tm1 > Tm 2 > Tm3

K m( λ + τ m ) N=10; λ = [ Tm1 , τm ]
K m (1 + sTm 3 ) e− sτ m
Table 86: PID tuning rules - SOSPD model with a positive zero - ideal controller
(1 + sTm1 )(1 + sTm 2 )
 1 
Gc ( s) = K c  1 + + Td s . 1 tuning rule.
 Ti s 
Minimum performance index
( p0 q1 − p1 ) p
q1 − 1 ,
p0 q 2 − p 2 p1
− . ≤
01
τm
≤1
ρ , ,
Minimum IAE, ISE
p0
2 p0 p 0 q1 − p1 p 0 Tm1
and ITAE – Wang et Tm 2
p 0 = Tm1 + Tm2 + Tm3 − τ m p2 = 05
. Tm1Tm 2 τ m p1 = Tm1Tm2 + . ≤
01 ≤1
al. [97]
q 1 = Tm1 + Tm 2 + 05
. τm 1 0.5τ m (Tm1 + Tm2 ) − 0.5Tm3 τ m Tm 1
q 2 = Tm1Tm2 Tm 3
01 ≤1
+0.5τ m ( Tm1 + Tm 2 ) Tm1
τm Tm 2 Tm 3 τ  τm Tm2 Tm3 
ρ = 35550
. − 3.6167 + 21781
. − 55203
. + m 14704
. − 04685
. + 14746
. 
Tm1 Tm1 Tm1 Tm1  Tm1 Tm1 Tm1 
Tm3  τm T T 
+
Tm 2  τm T T  + 14746
. − 0.3318 m 2 + 2.5356 m 3  , minimum IAE
− 0.4685 − 0.4918 m 2 − 0.3318 m 3 
Tm 1  Tm 1 Tm1 Tm 1  Tm1  Tm1 Tm1 Tm1 
τm Tm 2 Tm 3 τ m  τm Tm2 Tm3 
ρ = 39395
. − 3.2164 + 16185
. − 58240
. + 10933
. − 02383
. + 13508
. 
Tm1 Tm1 Tm1 Tm1  Tm1 Tm1 Tm1 
Tm3  τm T T 
+
Tm 2  τm T T  + 13508
. − 0.0564 m2 + 2.5648 m3  , minimum ISE
− 0.2383 − 0.6679 m 2 − 0.0564 m3 
Tm1  Tm1 Tm1 Tm1 
Tm 1  Tm1 Tm1 Tm1 
τm Tm2 Tm3 τ  τm T Tm3 

ρ = 32950
. − 34779
. + 25336
. − 55929
. + m 14407
. − 0.5712 m2 + 15340
. 
Tm1 Tm 1 Tm1 Tm 1  Tm1 Tm1 Tm1 
Tm3  τm Tm2 T 
+
Tm 2  τm T T  + 15340
. − 05790
. + 2.7129 m 3  , minimum ITAE
− 0.5712 − 0.3268 m2 − 0.5790 m 3 
Tm 1  Tm1 Tm1 Tm 1  Tm1  Tm1 Tm1 Tm1 
K m (1 + sTm 3 ) e− sτ m
Table 87: PID tuning rules - SOSPD model with a positive zero - ideal controller with filtered
(1 + sTm1 )(1 + sTm 2 )
 
 1 Td s 
derivative G c ( s) = Kc  1 + +  . 1 tuning rule.
 Ti s 1 + Td s 

 N 
Robust
Tm1 + Tm2 +
T m3 τ m Tm3 τ m Tm1 > Tm 2 > Tm3
Tm3 τ m λ + Tm 3 + τ m
N=10; λ = [ Tm1 , τm ]
λ + T m3 + τ m
Tm1 + Tm2 +
Chien [50] K m ( λ + Tm 3 + τ m ) λ + Tm 3 + τ m +
Tm1 Tm2
Tm3 τ m
Tm1 + Tm2 +
λ + Tm 3 + τ m
Model: Method 1
2ξTm1 +
Tm3 τ m
2 ξTm1 +
Tm3 τ m Tm3 τ m Tm1 > Tm 2 > Tm3
λ + Tm3 + τ m λ + Tm3 + τ m λ + Tm 3 + τ m
N=10; λ = [ Tm1 , τm ]
K m ( λ + Tm 3 + τ m ) Tm2 3
+
Tm 3 τ m
2ξ Tm1 +
λ + Tm3 + τ m
K m (1 + sTm 3 ) e− sτ m
Table 88: PID tuning rules - SOSPD model with a positive zero - classical controller
(1 + sTm1 )(1 + sTm 2 )
 
 1   1 + Td s 
 Ti s   1 + Td s 
 
 N 
Robust
Tm 2 Tm2 Tm1 Tm1 > Tm 2 > Tm3
Chien [50] K m ( λ + Tm3 + τm ) N=10; λ = [ Tm1 , τm ]
Model: Method 1 Tm1 Tm1 Tm2 Tm1 > Tm 2 > Tm3

K m ( λ + Tm3 + τm ) N=10; λ = [ Tm1 , τm ]
K m (1 + sTm 3 ) e− sτ m
Table 89: PID tuning rules - SOSPD model with a positive zero - series controller with
(1 + sTm1 )(1 + sTm 2 )
 
 1  Td s 
filtered derivative G c ( s) = Kc  1 +  1 +  . 1 tuning rule.
 i  1 + d 
Ts Ts
 N 
Robust
Tm 2 Tm 3τ m N=10; λ = [ T, τ m ] , T =
Tm1 +
Chien [50] K m ( λ + Tm3 + τm ) Tm2 λ + Tm 3 + τ m dominant time
constant
Model: Method 1
Tm1 Tm 3τ m N=10; λ = [ T, τ m ] , T =
Tm 2 +
K m ( λ + Tm3 + τm ) Tm1 λ + Tm3 + τ m dominant time
constant
Km e− sτm
Table 90: PID tuning rules - TOLPD model - ideal controller

(1 + sTm1 )(1 + sTm 2 )(1 + sTm 3 )
 1 
 Ti s 
Minimum performance index
Standard form
optimisation -  
 1 − 4 Tm 2 + Tm2  Tm1 4 6Tm2 τ m + τ m τm Tm1 > 10( Tm2 + τm )
 τ 
binomial - Polonyi  6 m Tm3  τ m
[153]
Model: Method 1
Standard form
optimisation – 
 1 − 21
.
Tm2 T  T
+ m2  m1 2.7 3.4Tm2 τ m + τ m τm Tm1 > 10( Tm2 + τm )
minimum ITAE -  3.4 τ Tm3  τ m
 m
Polonyi [153]
Model: Method 1
K m e− sτ m
Table 91: PID tuning rules - TOLPD model - G m (s ) = – Two degree of freedom controller:
(1 + sTm ) 3
   
   
1 Td s β T s
U (s) = K c 1 + +  E( s) − K c  α + d  R (s) . 1 tuning rule.
 Tis T   Td 
 1+ d s  1+ s
 N   N 
tuning
τm
Tm
[61a] (105 e ) 2 (105 e ) (105 e )
et al. [10]
 
 
1  1.275 
= 0.4020 +
2 (105e )
Kc ,
Kc  τm
 − 0.003273 
 Tm 
 τm  τm 
2
 τm 
3
 τm  
4
Ti
(105 e ) 
= Tm 0. 3572 + 7. 647 − 12. 86  + 11 .77   − 4. 146  
 

Tm  Tm   Tm   Tm  
 τm  τm  
2
τm  τm 
2
Td
(105 e ) 
= Tm 0.8335 + 0.2910 − 0. 04000   , α = 0. 6661 − 0.2509 + 0. 04773   ,
   Tm 

Tm  Tm   Tm
2
τm τ 
β = 0. 8131 − 0.2303 + 0.03621  m 
Tm Tm 
K m e− sτ m
 1 
Table 92: PID tuning rules - unstable FOLPD model - ideal controller Gc ( s) = K c  1 + + Td s .
1 − sTm  Ti s 
3 tuning rules.
Direct synthesis
Tm  
(1 − T τ )T
1
m τm + ,
K m  
cos m m
  1− T τ 
De Paor and Tm 

m m
(
 tan 0.75φ

) 1  Tm τ m
2 
 1
 τm
Tm τ m
<1
m
 
  Tm  1 − Tm τ m  Ti
O’Malley [86] Tm  1 − Tm τ m
Km  Tm τ m
sin (1 − T τ )T
m m m τm


Tm
  1 − Tm τ m
φ m = tan − 1
Tm τm
(
− Tm τm 1 − Tm τ m )
Model: Method 1
Chidambaram [88] 1  T   τ  τm
. + 0.3 m 
13 Tm  25 − 27 m  . τm
046 < 0.6
Model: Method 1
 τm 
 τ 
0.245
1165
.  Tm   0176 + 0.36 T τm
Tm  m  0176 + 0.36 m  Tm < 06
.
Valentine and    . .
Km  τ m  τm  τm 
2
 Tm  Tm
Chidambaram [154] - 0.179 − 0.324 + 0.161 
Tm  Tm 
dominant pole
placement 1165
.  Tm 
0.245
 τ   τ  τm
 0176 + 0.36 m  25Tm  0176 + 0.36 m  Tm 0.6 ≤ ≤ 08
.
  . .
Km  τ m   Tm   Tm  Tm
Model: Method 1
1165
.  Tm 
0.245
0.176Tm + 0.36τ m  τ  τm
 0176 + 0.36 m  Tm . <
08 ≤1
  τ .
Km  τ m  0.12 − 0.1 m  Tm  Tm
Tm
K m e− sτ m
Table 93: PID tuning rules - unstable FOLPD model - non-interacting controller
1 − sTm
 1 KTs
U( s) = Kc  1 +  E ( s) − c d Y(s) . 2 tuning rules.
 Ti s sT
1+ d
N
Huang and Lin [155] τm
 τm τ m 
−1 .0251
  τ 
6.6423
  7.6983τ m
 0.01 ≤ ≤ 0.8 ; N=10
- minimum IAE τm  T  − 0. 0312+ 1. 6333 τm + 00399
1  
−
Km
− 0 .433 + 0 .2056 + 0 .3135 
 Tm  − 00018
. + 08193
. + 7. 7853 m  . e T m  Tm

Tm  T m     m
 Tm  Tm    Tm 
Model: Method 2
Huang and Lin [155]  τ  τm
1 
−1.004
  τ m 
2. 9123 Tm  0.0011 + 0.4759 m  0.01 ≤ ≤ 0.8 ; N=10
- minimum IAE τm τ  τ
−
Km 
0.2675 + 01226
.
Tm
+ 0.8781 m  
 Tm  00005

. + 2.4631 m + 95795
.   

 Tm  Tm
 Tm  
 Tm  m
T 
Model: Method 2
K m e− sτ
m
Table 94: PID tuning rules - unstable FOLPD model - classical controller
1 − sTm
 
 1   1 + Td s 
 Ti s   1 + Td s 
 
 N 
0.9091Tm . τm
170 0.60τm τ m Tm = 01
.
Shinskey [16] - Km τ m
minimum IAE – page
Tm 1.90τ m 0.60τm τ m Tm = 0.2
381.
K mτ m
Method: Model 1 08929
. Tm 2.00τ m 0.80τ m τ m Tm = 0.5
Km τ m
08621
. Tm 2.25τm 0.90τm τ m Tm = 0.67
K m τm
08333
. Tm 2.40τ m . τm
100 τ m Tm = 0.8
K m τm
Km e − sτ m
Table 95: PID tuning rules - unstable SOSPD model G m ( s) = - ideal controller
(1 − sTm1 )(1 + sTm 2 )
 1 
Gc ( s) = K c  1 + + Td s . 2 tuning rules.
 Ti s 
Ultimate cycle
McMillan [58] Kc(106 ) 3 Ti (106) Td (106 ) Tuning rules
Model: Method not developed from Ku , Tu
relevant
Robust
 λ 
  λ   λ 
λ + 2 Tm 2
Rotstein and Lewin   Tm1 
[89] Tm1  λ  + 2 + Tm 2  λ + 2 + Tm 2
  Tm1    Tm1   λ 
λ + 2 + Tm 2 λ determined
Model: Method 1 λ 2
 Tm1  graphically – sample
Km uncertainty = 50% τm Tm = 02
. λ = [0.5Tm ,19
. Tm ] values provided
τm Tm = 0.4 λ = [1.3Tm ,19
. Tm ]
τm Tm = 02
. λ = [0.4Tm ,4. 3Tm ]
Km uncertainty = 30% τm Tm = 0.4 λ = [11
. Tm ,4.3Tm ]
τm Tm = 0.6 λ = [ 2.2Tm , 4.3Tm ]
2
 
 
1111
. Tm1Tm2  1 
=
3 ( 106)
Kc  0.65  ,
Km τ m 2   ( Tm1 + Tm 2 )Tm1Tm2  
1 +  T − T T − τ τ  
  ( m1 m2 )( m1 m) m 
 
  (T + T ) T T  
0.65
  (T + T )T T  
0.65
 
= 2τ m 1 +  = 05
. τ m 1 + 
( 106) m1 m2 m1 m2 ( 106) m1 m2 m1 m 2
Ti   , Td  
  ( Tm1 − Tm2 )( Tm1 − τ m ) τ m     (Tm1 − Tm 2 )( Tm1 − τ m ) τ m  
Km e − sτ m
Table 96: PID tuning rules - unstable SOSPD model G m ( s) = - classical controller
(1 − sTm1 )(1 + sTm 2 )
 
 1   1 + Tds 
G c ( s) = Kc  1 +   . 2 tuning rules.
 Ti s   1 + Td s 
 
 N 
aTm2 2 τm
0≤ ≤2;
4Tm1 ( τ m + Tm2 )
bTm1 1 +
Minimum ITAE - 4( τm + Tm 2 )
2
( d N)
T
aTm1 − 4( τ m + Tm2 )
Poulin and Tm2
Pomerleau [82], [92] – K m ( aTm1 − 4[ τ m + Tm2 ]) . Tm ≤
01
Td
≤ 033
. Tm
deduced from graph N
Model: Method 1
(τ m + Td N) Tm1 a b (τ m + Td N) Tm1 a b
0.05 0.9479 2.3546 0.30 1.6163 2.6612
Output step load 0.10 1.0799 2.4111 0.35 1.7650 2.7368
disturbance 0.15 1.2013 2.4646 0.40 1.9139 2.8161
0.20 1.3485 2.5318 0.45 2.0658 2.9004
0.25 1.4905 2.5992 0.50 2.2080 2.9826
0.05 1.1075 2.4230 0.30 1.6943 2.7007

Input step load 0.10 1.2013 2.4646 0.35 1.8161 2.7637
disturbance 0.15 1.3132 2.5154 0.40 1.9658 2.8445
0.20 1.4384 2.5742 0.45 2.1022 2.9210
0.25 1.5698 2.6381 0.50 2.2379 3.0003
Km e − sτ m
Table 97: PID tuning rules - unstable SOSPD model G m ( s) = - series controller
(1 − sTm1 )(1 + sTm 2 )
 1
Gc ( s) = Kc 1 +  (1 + Td s) . 1 tuning rule.
 Ti s
Direct synthesis
Ho and Xu [90] ω p Tm1 1 . A m (A m − 1)
A m φ m + 157
. ω p − ω p τm − ωp =
2
A m Km
157
Tm1 Tm2 (A m
2
)
− 1 τm
Model: Method 1
Km e − sτ
m
Table 98: PID tuning rules - unstable SOSPD model G m ( s) = - non-interacting controller
(1 − sTm1 )(1 + sTm 2 )
 1 KTs
U( s) = Kc  1 +  E ( s) − c d Y(s) . 2 tuning rules.
 Ti s sT
1+ d
N
Huang and Lin [155] Tm2 ≤ Tm1 ;
- minimum IAE Kc(107 ) 4 Ti (107) Td (107 ) τm
0.05 ≤ ≤ 0.4
Tm1
Model: Method 2
1  T τ 
−1.344 0.995
τm  τm  Tm2  Tm 2 
4
Kc ( 107 )
=− 10.741 − 13363
. + 0099
.   + 727.914 − 708.481  + 9.915 m 2 2m 
Km  Tm1  Tm1  Tm1  Tm1  Tm1 
 
1  
1.031 0.997 τ T T m 2 τm
T   Tm 2 
m m2
Tm 2
− 84.273 m1  − + − − 
2
 
T T T
90.959 9.034 e
m1
2 .386e m1
16.304 e m1
Km   τm   Tm1  τm 

 τm τ 
2.12
T T 
0.985
T τ 
Ti (107 ) = Tm1  −149 .685 − 141418
. − 88.717 m  − 17.29 m 2 + 20518
.  m2  − 12.82 m 2 2m 
 Tm1  Tm1  Tm1  Tm1  Tm1 

 τ 
0.286
 Tm 2 
1.988
T
τ m T m2 T m 2τ m

+ Tm1 3.611 m  + 0.000805 m2 + 141.702e T − 2.032e T + 10.006e T 
2
  m1 m1 m1
  Tm1   Tm1  τm 
 τm  τm 
2
Tm2  Tm2 
2
T τ 
Td ( 107 )
= Tm1  −0.4144 + 15805
. − 142.327   + 0.7287 + 01123
.   − 18.317 m 2 2m 
 Tm1  Tm1  Tm1  Tm1  Tm1 

 τ 
3
T 
3
 τ 2T  τ T 2  τ 
4

+ Tm1 48695
.  m  − 10542
.  m2  + 204.009 m 3m 2  + 47.26 m m32  + 396349
.  m 
  Tm1   Tm1   Tm1   Tm1   Tm1  
  τ 3T  τ T 3  τ 2T 2  T 
4
τ 
5

+ Tm1  − 138.038 m 4m 2  + 52155
.  m m42  − 646.848 m m4 2  + 19.302 m 2  − 4731.72 m  
  Tm1   Tm1   Tm1   Tm1   Tm1  
  τ 4T  τ T 4  τ 3T 2   τ 2T 3   Tm 2 
5

+ Tm1 425789
.  m 5m 2  − 289 .746 m m52  − 841807
.  m m5 2  + 1313.72 m m5 2  − 37688
.   
  Tm1   Tm1   Tm1   Tm1   Tm1  
  τm 
6
 τ 5T   τ T 5  τ 4T 2  
+ Tm1 6264 .79  − 161469
.  m 6m2  + 204.689 m m62  + 25706
.  m m6 2  
  Tm1   Tm1   Tm1   Tm1  
  τ 2T 4  τ T 
3
T  
6
+ Tm1  − 791857
.  m m6 2  + 648.217 m m2 2  − 5.71 m2  
  Tm1   Tm1   Tm1  
 
Huang and Lin [155] Tm 1 < Tm 2 ≤ 10Tm1 ;
- minimum IAE - Kc(108 ) 5 Ti (108) Td (108 ) τm
continued 0.05 ≤ ≤ 025
.
Tm1
Model: Method 2
1  T τ 
− 0.3055 0.5174
τm τ  T T 
5
Kc (108) = − −1302
. + 85914
. + 34.82 m  + 10.442 m 2 − 22.547  m2  − 14.698 m 2 2m 
Km  Tm1  Tm1  Tm1  Tm1  Tm2 
 
1  T 
1.0077
 Tm 2 
0.9879
 Tm 2 
τ m T m2
Tm2 τ m

− 52.408  m1    − 5147
.   + 53.378e − 0.000001e T
T m1 m1
+ 0.286e Tm1 2 
Km  τ
 m  m1 
T τ
 m 
 
 τm  Tm 2   τm 
2
 Tm2 
2
Tm2 τ m 
Ti ( 108)
= Tm1 72.806 − 268.746 − 4.9221  + 246819
.   + 0.6724  + 151351
. 
 Tm1  Tm1   Tm1   Tm1  Tm12 

 τ 
3
 Tm 2 
3
 τ 2T  τ T 2 τ  
4
+ Tm1 − 6914.46 m  − 00092

.   − 795465 .  m 
.  m 3m 2  − 14.27 m m32  + 558017
  Tm1   Tm1   Tm1   Tm1   Tm1  
  τ 3T   τ T 3  τ 2T 2   Tm2 
4

+ Tm 1 1417 .65 m 4m 2  + 0.4057 m m42  + 55536
.  m m4 2  − 0001119
.   
  Tm1   Tm1   Tm1   Tm1  
 τ m T T
m2 m2 τ m
τ 
19.056
 Tm2 
7.3464 
+ Tm 1 − 44.903e + 0.000034e − 15694 
2
T m1 T
. eTm1 m1
+ 678778 m   
  Tm1   Tm1  
 
 τ τ 
1.1798
T T 
0.1064
T τ 
Td (108) = Tm1 175515
. − 86.2 m + 348.727 m  − 0.008207 m 2 − 55619
.  m2  + 0.0418 m 2 2m 
 Tm1  Tm1  Tm1  Tm1  Tm1 

 τ 
0.0355
 Tm2 
0.0827
T 
τ m Tm 2 Tm 2 τ m

+ Tm 1 78959
.  m   + 0.005048 m 2  − 187 .01e T m1
+ 0000001
. e Tm1
− 00149
. e Tm1 2 
  Tm 1   Tm1   τm  
 
Huang and Lin [155] Tm2 ≤ Tm1 ;
– minimum IAE Kc ( 109 ) 6
Ti ( 109)
Td ( 109 )
τm
0.05 ≤ ≤ 0.4
Tm1
Model: Method 2
1  T τ 
−1.164 2.54
τm  τm  Tm 2  Tm2 
6
Kc ( 109 )
=− − 174167
. − 31364
. + 0.4642  − 103069
. − 83916
.   − 66.962 m2 2m 
Km  Tm1  Tm1  Tm1  Tm1  Tm1 
 
1  1.065
T  T 
1.014
Tm 2
τ m Tm 2 Tm 2 τ m

− 59.496  m1   m 2  − 7079
. + 23121
. e T m1
+ 126.924e Tm 1
+ 26.944e Tm 1 2 
Km   τm   Tm1  τm 
 
 τm  τm 
2
Tm 2  Tm2 
2
Tm2 τ m  τm  
3
Ti ( 109)
= Tm1 0.008 + 2.0718 + 6.431  + 0.4556 + 0.7503  + 2.4484 − 18686
.   
 Tm1  Tm1  Tm1  Tm1  Tm12  Tm1  

 T 
3
T τ 2 τ T 2 τ 
4
 T τ 3
+ Tm 1 − 2.9978  m 2  − 21135 .  m 2 4m  
.  m2 m3  + 12.822  m m32  + 39.001 m  + 22848
  Tm1   Tm1   Tm1   Tm1   Tm1  
  T 3τ   T 2τ 2  T 
4

+ Tm1 − 4.754 m 2 4 m  − 0.527  m2 4m  + 164
.  m2  
  Tm1   Tm1   Tm1  
 τm τ 
2
Tm 2  Tm2 
2
T τ τ  
3
Td (109) = Tm1  −0.0301 + 11766

. − 4.4623 m  + 05284
. − 14281
.   + 4.6 m 2 2m + 11176
.  m 
 Tm1  Tm1  Tm1  Tm1  Tm1  Tm1  

  Tm 2 
3
 τ m2 Tm 2   τ m Tm2 2   τm 
4
 Tm 2 τm 3  

+ Tm 1 10886
.   − 5.0229  − 05039
.   − 98564
.   − 7528
.  
  Tm1   Tm1   Tm1   Tm1   Tm1  
3 3 4
 τ T 3  τ 2T 2   Tm 2  
4
+ Tm1 − 2.3542  m m42  + 9.3804  m m4 2  − 01457

.   
  Tm1   Tm1   Tm1  
Huang and Lin [155] Tm 1 < Tm 2 ≤ 10Tm1 ;
- minimum IAE – Kc(110 ) 7 Ti (110) Td (110 ) τm
continued 0.05 ≤ ≤ 025
.
Tm1
Model: Method 2
1  T τ 
2.1984 0.791
τm  τm  Tm 2  Tm 2 
7
Kc ( 110)
=− 1750.08 + 1637.76 + 1533.91  − 7.917 + 6187
.   − 6.451 m2 2m 
Km  Tm1  Tm1  Tm1  Tm1  Tm1 
 
1  
3.2927 1.0757 τ T Tm 2 τ m
 Tm1   Tm 2 
m m2
Tm 2
− 0002452
.     + 13729
. − 1739.77e − 0.000296e T
T m1 m1
+ 2.311e Tm1 2 
Km   τm   Tm1  τm 
 
 τ τ 
2
Tm 2  Tm 2 
2
T τ τ  
3
Ti (110) = Tm1 51678

. − 57.043 m + 1337.29 m  + 01742
. − 01524
.   + 7.7266 m 2 2m − 6011.57 m  
 Tm1  Tm1  Tm1  Tm1  Tm1  Tm1  

  Tm2 
3
 Tm2 τ m 2   τ m Tm2 2   τm 
4
 Tm2 τ m 3  

+ Tm 1 0.0213  − 65.283  + 0.0645  + 913552
.   + 274 .851 
  Tm1   Tm1   Tm1   Tm1   Tm1  
3 3 4
  T 3τ   T 2τ 2   Tm 2 
4 τ m T m2

+ Tm 1 0.003926 m 2 4 m  − 2.0997 m 2 4m  − 0001077
.   − 49.007 e T
+ 0.000026
m1
e T m1

  m1 
T  m1 
T  m1 
T 
 T τm2 m

+ Tm1 0.2977e T m1

 
 τ τ 
2
T T 
2
T τ τ  
3
Td (110) = Tm1  −0.0605 + 4.6998 m − 29.478 m  + 0.0117 m 2 − 0.0129 m 2  + 0.6874 m2 2m + 140.135 m  

 Tm1  Tm1  Tm1  Tm1  Tm1  Tm1  

  Tm 2 
3
 τ m 2 Tm 2   τ mTm 2 2   τm 
4
 Tm 2 τm 3  

+ Tm 1 0.002455  − 14712
.   − 01289
.   − 238.864   + 06884
.  
  Tm1   Tm1   Tm1   Tm1   Tm1  
3 3 4
  τ T 3  τ m 2Tm 2 2   Tm 2  
4
+ Tm 1 0.007725 m m42  − 01222

.   − 0.000135  
  Tm1   Tm1   Tm1  
4
K m e − sτ
m
Table 99: PID tuning rules – general model with a repeated pole G m ( s) = - ideal controller
(1 + sTm ) n
 1 
 Ti s 
Direct synthesis
Skoczowski and
(1 + ω )
n− 1
2
ω g Tm Tm 2
n−1
g
Tarasiejski [156] ≤ Tm
Km 1 + ω g 2 Td 2 Tm , n ≥ 2
n+2
1
Model: Method 1
 2n + 4 τ m 4n + 2 
− Tm n 2 − 2n − 2 + + φm  ± b
 π Tm π 
ωg =
1
 2 4 n + 2 τ m 2n − 2 
2Tm n − 4n + 3 +
2
+ φm 
 π Tm π 
with
2
 2 2n + 4 τ m 4 n + 2 2   2  4 n + 2 τm 2 n − 2 
b = Tm n − 2n − 2 + + φ m  + 4( n + 2) 1 − φ m   n2 − 4n + 3 + + φm 
 π Tm π   π  π Tm π 
Table 100: PID tuning rules – general stable non-oscillating model with a time delay - ideal controller
 1 
 Ti s 
Direct synthesis
(110 a ) 2 (110 a ) (110a )
Kc Ti Td
Gorez and Klan
τm  τm 
(110 a ) (110 a )
Kc Ti
[147a] Ti
(110 a )
1 − 
 T 
Model: not specified Tar  ar 
(110 a ) (110 a ) (110 a )
Kc Ti 0.25Ti
2
τ  τ
1+ 1 + 2 m  −2 m

 Tar 
( 110a )
Ti Tar
= = Tar
2 (110 a ) (110 a )
Kc , Ti ,
+ τm
( 110a )
Ti 2
 2 
2τ m  
Td
(110 a )
=
(110 a )

(Tar
) τ
2Tar 

 Ti (110a ) + Tcr cr + Ti (110a ) − Taa − m 1 − K c (110a ) 1 +
Tar T
 
 3Ti (110 a )  
Ti  
Tar = average residence time of the process (which equals Tm + τ m for a FOLPD process, for example); T aa =
  
additional apparent time constant; Tcr = 1 − K c
(110 a )
1 + τ m  τ m
  2T (110a ) 
 i 
Table 101: PID tuning rules – fifth order model with delay
 1 
G m ( s) = – ideal controller Gc ( s) = K c  1 + + Td s .
(
1 + a 1s + a 2 s + a 3s + a 4 s + a 5s
2 3 4 5
)  Ti s 
1 tuning rule.
Direct synthesis
3
Magnitude optimum
- Vrancic et al. [159] Kc ( 112 )
Ti (112) Td (112 )
Model: Method 1
Kc ( 112 ) =
( )
a13 − a1 2b 1 + a1 b 2 − 2 a1a2 + a2 b 1 + a 3 − b 3 + τ m a1 2 − a1b1 − a 2 + b 2 + 0.5( a1 − b1 ) τ m 2 + 0167
. τm 3
[
2 Km − a1 b 1 + a1a2 + a1b1 − a3 − b1b 2 + b 3 + ( a1 − b 1 ) τ m + ( a1 − b 1 ) τ m2 + 0.333τ m 3 − ( a1 − b 1 + τ m ) Td
2 2 2 2
]
Ti (112 ) =
2
( 2
)
a1 − a1 b 1 + a 1b 2 − 2 a1a 2 + a2 b 1 + a 3 − b 3 + τ m a 1 − a1b 1 − a 2 + b 2 + 0.5( a1 − b1 ) τm + 0.167τ m
3 2 3
[a 1
2
− a 1b 1 − a 2 + b 2 + ( a1 − b 1 ) τ m + 0.5τ m − ( a1 − b1 + τ m ) Td
2
]
Td ( 112)
… see attached sheet with δ = 0.
Table 102: PID tuning rules – fifth order model with delay
G m ( s) = – controller with filtered derivative

(
1 + a1s + a2 s2 + a3s 3 + a 4 s4 + a 5s5 )
 
 1 Tds 
 Ts
i 1 + d 
Ts
 N 
Direct synthesis
Magnitude optimum
- Vrancic et al. [73] Kc(113) 4 Ti (113) Td (113) 8 ≤ N ≤ 20
Model: Method 1
Kc ( 113) =
( )
a13 − a12 b1 + a1b2 − 2a1a2 + a 2b 1 + a 3 − b 3 + τ m a12 − a1b 1 − a2 + b 2 + 0.5( a1 − b1 ) τ m2 + 0.167τ m3
 2 T2 
2 K m −a1 b 1 + a1a2 + a1b 1 − a3 − b1b 2 + b 3 + ( a1 − b1) τ m + ( a1 − b1 ) τ m + 0.333τ m − ( a1 − b1 + τ m ) Td − d ( a1 − b1 + τ m ) 
2 2 2 3 2
 N 
Ti( 113 ) =
( )
a13 − a12 b1 + a1b 2 − 2a1a2 + a2 b1 + a 3 − b 3 + τ m a12 − a1b 1 − a 2 + b 2 + 0.5( a1 − b1 ) τ m2 + 0.167τ m3
 2 Td 2 
a1 − a1b 1 − a 2 + b 2 + ( a1 − b 1) τ m + 0.5τ m − ( a1 − b1 + τ m ) Td −
2

 N
Td (113) = see attached sheet
4. Conclusions
The report has presented a comprehensive summary of the tuning rules for PI and PID controllers that
have been developed to compensate SISO processes with time delay. Further work will concentrate on evaluating
the applicability of these tuning rules to the compensation of processes with time delays, as the value of the time
delay varies compared with the other dynamic variables.
Table 103: Tuning rules published by year and medium

Year Journal articles Conference Books/ Ph.D. Total
papers/ thesis
correspondence
1942 1 1
1950 1 1
1951 1 1
1952 1 1
1953 3 3
1954 1 1
1961 1 1
1964 1 1
1965 1 1
1967 1 1 2
1968 1 1
1969 2 2
1972 2 2
1973 1 1 2
1975 2 2
1979 1 1
1980 2 2
1982 1 1
1984 2 1 3
1985 1 1 2
1986 1 1
1987 1 1
1988 3 4 2 9
1989 4 3 7
1990 4 1 5
1991 5 5
1992 1 2 3
1993 5 4 1 10
1994 5 2 2 9
1995 14 2 1 17
1996 8 6 2 16
1997 11 4 1 16
1998 6 7 13
1999 12 1 1 14
2000 1 16 17
1941-1950 2 0 0 2
1951-1960 6 0 0 6
1961-1970 5 0 3 8
1971-1980 7 0 2 9
1981-1990 15 7 7 29
1991-2000 68 44 8 120
TOTAL 103 51 20 174
List of journals in which tuning rules were published and number of tuning rules published
Advances in Modelling and Analysis C, ASME Press 1

AIChE Journal 4
Automatica 12
British Chemical Engineering 1
Chemical Engineering Communications 5
Chemical Engineering Progress 1
Chemical Engineering Science 1
Control 1
Control and Computers 1
Control Engineering 7
Control Engineering Practice 4
European Journal of Control 1
Hydrocarbon Processing 2
Hungarian Journal of Industrial Chemistry 1
IEE Proceedings, Part D 9
(including IEE Proceedings - Control Theory and Applications, Proceedings of the IEE, Part 2)
IEEE Control Systems Magazine 1
IEEE Transactions on Control Systems Technology 4
IEEE Transactions on Industrial Electronics and Control Instrumentation 1
IEEE Transactions on Industrial Electronics 1
IEEE Transactions on Industry Applications 1
Industrial and Engineering Chemistry Process Design and Development 2
Industrial Engineering Chemistry Research 12
International Journal of Adaptive Control and Signal Processing 1
International Journal of Control 3
International Journal of Electrical Engineering Education 1
International Journal of Systems Science 1
Instrumentation 1
Instrumentation Technology 2
Instruments and Control Systems 2
ISA Transactions 2
Journal of Chemical Engineering of Japan 2
Journal of the Chinese Institute of Chemical Engineers 3
Pulp and Paper Canada 1
Process Control and Quality 1
Thermal Engineering (Russia) 2
Transactions of the ASME 7
(including Transactions of the ASME. Journal of Dynamic Systems, Measurement and Control)
Transactions of the Institute of Chemical Engineers 1
Classification of journals in which tuning rules were published and number of tuning rules
published
Chemical Engineering Journals 32

(AIChE Journal, British Chemical Engineering, Chemical Engineering Communications, Chemical Engineering
Progress, Chemical Engineering Science, Transactions of the Institute of Chemical Engineers, Hungarian Journal
of Industrial Chemistry, Industrial and Engineering Chemistry Process Design and Development, Industrial
Engineering Chemistry Research, Journal of Chemical Engineering of Japan, Journal of the Chinese Institute of
Chemical Engineers)
Control Engineering Journals 53
(Automatica, Control, Control and Computers, Control Engineering, Control Engineering Practice, European
Journal of Control, IEE Proceedings, Part D, IEE Proceedings - Control Theory and Applications, Proceedings of
the IEE, Part 2, IEEE Control Systems Magazine, IEEE Transactions on Control Systems Technology,
International Journal of Adaptive Control and Signal Processing, International Journal of Control, International
Journal of Systems Science, Instrumentation, Instrumentation Technology, Instruments and Control Systems,
ISA Transactions, Process Control and Quality)
Mechanical Engineering Journals 8
(Advances in Modelling and Analysis C, ASME Press, Transactions of the ASME, Transactions of the ASME.
Journal of Dynamic Systems, Measurement and Control)
Electrical/Electronic Engineering Journals 4
(EE Transactions on Industrial Electronics and Control Instrumentation, IEEE Transactions on Industrial
Electronics, IEEE Transactions on Industry Applications, International Journal of Electrical Engineering
Education)
Trade Journals 5
(Hydrocarbon Processing, Pulp and Paper Canada, Thermal Engineering (Russia))
5. References
1. Koivo, H.N. and Tanttu, J.T., Tuning of PID Controllers: Survey of SISO and MIMO techniques.
Proceedings of the IFAC Intelligent Tuning and Adaptive Control Symposium, Singapore, 1991, 75-80.
2. Hwang, S.-H., Adaptive dominant pole design of PID controllers based on a single closed-loop test.
Chemical Engineering Communications, 1993, 124, 131-152.
3. Astrom, K.J. and Hagglund, T., PID Controllers: Theory, Design and Tuning. Instrument Society of America,
Research Triangle Park, North Carolina, 2nd Edition, 1995.
4. Bialkowski, W.L., Control of the pulp and paper making process. The Control Handbook. Editor: W.S.
Levine, CRC/IEEE Press, Boca Raton, Florida, 1996, 1219-1242.
5. Isermann, R., Digital Control Systems Volume 1. Fundamentals, Deterministic Control. 2nd Revised Edition,
Springer-Verlag, 1989.
6. Ender, D.B., Process control performance: not as good as you think. Control Engineering, 1993, September,
180-190.
7. Astrom, K.J. and Wittenmark, B., Computer controlled systems: theory and design. Prentice-Hall
International Inc., 1984.
8. Ziegler, J.G. and Nichols, N.B., Optimum settings for automatic controllers. Transactions of the ASME, 1942,
November, 759-768.
9. Hazebroek, P. and Van der Waerden, B.L., The optimum adjustment of regulators. Transactions of the
ASME, 1950, April, 317-322.
10. Chien, K.-L., Hrones, J.A. and Reswick, J.B., On the automatic control of generalised passive systems.
Transactions of the ASME, 1952, February, 175-185.
11. Cohen, G.H. and Coon, G.A., Theoritical considerations of retarded control. Transactions of the ASME, 1953,
May, 827-834.
12. Wolfe, W.A., Controller settings for optimum control. Transactions of the ASME, 1951, May, 413-418.
13. Murrill, P.W., Automatic control of processes. International Textbook Co., 1967.
14. McMillan, G.K., Tuning and control loop performance - a practitioner’s guide. Instrument Society of
America, Research Triangle Park, North Carolina, 3rd Edition, 1994.
15. St. Clair, D.W., Controller tuning and control loop performance, Straight Line Control Co., Inc., 2nd Edition,
1997.
15a. Shinskey, F.G. (2000). PID-deadtime control of distributed processes, Preprints of the Proceedings of PID
’00: IFAC Workshop on digital control (Past, present and future of PID control), Terrassa, Spain, pp. 14-
18.
16. Shinskey, F.G., Process Control Systems - Application, Design and Tuning. McGraw-Hill Inc., New York, 3rd
Edition, 1988.
17. Shinskey, F.G., Process Control Systems - Application, Design and Tuning. McGraw-Hill Inc., New York, 4th
Edition, 1996.
18. Huang, C.-T., Chou, C.-J. and Wang, J.-L., Tuning of PID controllers based on second-order model by
calculation. Journal of the Chinese Institute of Chemical Engineers, 1996, 27(2), 106-120.
19. Yu, S. W., Optimal PI tuning for load disturbances. Journal of the Chinese Institute of Chemical Engineers,
1988, 19(6), 349-357.
20. Zhuang, M. and Atherton, D.P., Automatic tuning of optimum PID controllers. IEE Proceedings, Part D, 1993,
140(3), 216-224.
21. Rovira, A.A., Murrill, P.W. and Smith, C.L., Tuning controllers for setpoint changes. Instruments and Control
Systems, 1969, 42, December, 67-69.
22. Khan, B.Z. and Lehman, B., Setpoint PI controllers for systems with large normalised dead time. IEEE
Transactions on Control Systems Technology, 1996, 4(4), 459-466.

23. Haalman, A., Adjusting controllers for a deadtime process. Control Engineering, 1965, July, 71-73.
23a. Chen, C.-L. and Yang, S.-F., PI tuning based on peak amplitude ratio, Preprints of the Proceedings of PID
’00: IFAC Workshop on digital control (Past, present and future of PID control), Terrassa, Spain, 2000, pp.
195-198.
24. Pemberton, T.J., PID: The logical control algorithm. Control Engineering, 1972, 19(5), 66-67.
25. Smith, C.A. and Corripio, A.B., Principles and practice of automatic process control. John Wiley and Sons,
New York, 2nd Edition, 1997.
26. Smith, C.L., Corripio, A.B. and Martin, J. (1975). Controller tuning from simple process models,
Instrumentation Technology, December, 39-44.
27. Hang, C.C., Tan, C.H. and Chan, W.P., A performance study of control systems with dead time. IEEE
Transactions on Industrial Electronics and Control Instrumentation, 1980, IECI-27(3), 234-241.
27a. Miluse, V., Vitecek, A. and Smutny, L. (2000a). Controller tuning for controlled plants with time delay,
Preprints of the Proceedings of PID ’00: IFAC Workshop on digital control (Past, present and future of
PID control), Terrassa, Spain, pp. 283-288.
27b. Miluse, V., Vitecek, A. and Smutny, L. (2000a). Controller tuning for controlled plants with time delay,
PID control), Terrassa, Spain, pp. 289-294.
28. Gorecki, H., Fuska, S., Grabowski, P. and Korytowski, A., Analysis and synthesis of time delay systems,
John Wiley and Sons, New York, 1989.
29. Chiu, K.C., Corripio, A.B. and Smith, C.L., Digital controller algorithms. Part III. Tuning PI and PID
controllers. Instruments and Control Systems, 1973, December, 41-43.
30. Astrom, K.J., Hagglund, T., Hang, C.C. and Ho., W.K., Automatic tuning and adaptation for PID controllers -
a survey. Control Engineering Practice, 1993, 1(4), 699-714.
31. Davydov, N.I., Idzon, O.M. and Simonova, O.V., Determining the parameters of PID-controller settings using
the transient response of the controlled plant. Thermal Engineering, 1995, 42(10), 801-807.
32. Schneider, D.M., Control of processes with time delay. IEEE Transactions on Industry Applications, 1988, 24
(2), 186-191.
33. McAnany, D.E., A pole placement technique for optimum PID control parameters. Proceedings of the ISA/93
Advances in Instrumentation and Control Conference, McCormick Place, Chicago, Illinois, USA, 1993, 48,
1775-1782.
34. Leva, A., Maffezzoni, C. and Scattolini, R., Self-tuning PI-PID regulators for stable systems with varying
delay. Automatica, 1994, 30(7), 1171-1183.
35. Hang, C.C., Ho, W.K. and Cao, L.S., A comparison of two design methods for PID controllers. Proceedings
of the ISA/93 Advances in Instrumentation and Control Conference, McCormick Place, Chicago, Illinois,
USA, 1993, 48, 959-967.
36. Hang, C.C., Lee, T.H. and Ho, W.K., Adaptive Control. Instrument Society of America, Reseaech Triangle
Park, North Carolina, 1993.
37. Ho, W.K., Hang, C.C. and Zhou, J.H., Tuning of PID controllers based on gain and phase margin
specifications. Automatica, 1995, 31(3), 497-502.
38. Kookos, I.K., Lygeros, A.I. and Arvanitis, K.G., On-line PI controller tuning for integrator/dead time
processes. European Journal of Control, 1999, 5, 19-31.
39. Tan, K.K., Lee, T.H. and Wang, Q.G., Enhanced automatic tuning procedure for process control of PI/PID
controllers. AIChE Journal, 1996, 42(9), 2555-2562.
40. Voda, A. and Landau, I.D., The autocalibration of PI controllers based on two frequency measurements.
International Journal of Adaptive Control and Signal Processing, 1995, 9, 395-421.
41. Friman, M. and Waller, K.V., A two channel relay for autotuning. Industrial Engineering Chemistry Research,
1997, 36(7), 2662-2671.
42. Smith, L., A modified Smith predictor for extruded diameter control. InstMC Mini Symposium - Algorithms
and Architectures for Industrial Controllers (in UKACC International Conference on Control ’98), Swansea,
Wales, Lecture 5, 1998.
43. Cox, C.S., Daniel, P.R. and Lowdon, A., Quicktune: a reliable automatic strategy for determining PI and PPI
controller parameters using a FOLPD model. Control Engineering Practice, 1997, 5(10), 1463-1472.
44. Cluett, W.R. and Wang, L., New tuning rules for PID control. Pulp and Paper Canada, 1997, 3(6), 52-55.
45. Abbas, A., A new set of controller tuning relations. ISA Transactions, 1997, 36(3), 183-187.
46. Bi, Q., Cai, W.-J., Lee, E.-L., Wang, Q.-G., Hang, C.-C. and Zhang, Y., Robust identification of first-order plus
dead-time model from step response, Control Engineering Practice, 1999, 7(1), 71-77.
47. Wang, Y.-G. and Shao, H.-H. (2000). Optimal tuning for PI controller, Automatica, 36, 147-152.
48. Brambilla, A., Chen, S. and Scali, C., Robust tuning of conventional controllers. Hydrocarbon Processing,
1990, November, 53-58.
49. Rivera, D.E., Morari, M. and Skogestad, S., Internal Model Control. 4. PID controller design. Industrial and
Engineering Chemistry Process Design and Development, 1986, 25(1), 252-265.
50. Chien, I.-L., IMC-PID controller design - an extension. Proceedings of the IFAC Adaptive Control of
Chemical Processes Conference, Copenhagen, Denmark, 1988, 147-152.
51. Thomasson, F.Y., Tuning guide for basic control loops. Proceedings of the 1997 process control, electrical
and information conference (TAAPI), 1997, 137-148.
52. Fruehauf, P.S., Chien, I.-L. and Lauritsen, M.D., Simplified IMC-PID tuning rules. Proceedings of the ISA/93
Advances in Instrumentation and Control Conference, McCormick Place, Chicago, Illinois, USA, 1993, 48,
1745-1766.
53. Chen, C.-L., Huang, H.-P. and Hsieh, C.-T. (1999). Tuning of PI/PID controllers based on specification of
maximum closed-loop amplitude ratio, Journal of Chemical Engineering of Japan, 32, 6, 783-788.
54. Ogawa, S., PI controller tuning for robust performance. Proceedings of the IEEE Conference on Control
Applications, 1995, 101-106.
55. Lee, Y., Park, S., Lee, M. and Brosilow, C., PID controller tuning for desired closed-loop responses for SI/SO
systems. AIChE Journal, 1998, 44(1), 106-115.
56. Isaksson, A.J. and Graebe, S.F., Analytical PID parameter expressions for higher order systems, Automatica,
1999, 35, 1121-1130.
57. Chun, D., Choi, J.Y. and Lee, J., Parallel compensation with a secondary measurement, Industrial Engineering
Chemistry Research, 1999, 38(4), 1575-1579.
58. McMillan, G.K., Control loop performance. Proceedings of the ISA/84 International Conference and
Exhibition. Advances in Instrumentation, Houston, Texas, USA, 1984, 39, 1, 589-603.
59. Shinskey, F.G., Feedback controllers for the process industries. McGraw-Hill Inc., New York, 1994.
60. Hwang, S.-H., Closed-loop automatic tuning of single-input-single-output systems. Industrial Engineering
Chemistry Research, 1995, 34(7), 2406-2417.
61. Hwang, S.-H. and Fang, S.-M., Closed-loop tuning method based on dominant pole placement. Chemical
Engineering Communications, 1995, 136, 45-66.

61a. Taguchi, H. and Araki, M., Two-degree-of-freedom PID controllers – their functions and optimal tuning,
PID control), Terrassa, Spain, 2000, pp. 95-100.
62. Hwang, S.-H. and Chang, H.-C., A theoretical examination of closed-loop properties and tuning methods of
single-loop PI controllers. Chemical Engineering Science, 1987, 42(10), 2395-2415.
63. Pessen, D.W., A new look at PID-controller tuning. Transactions of the ASME. Journal of Dynamic Systems,
Measurement and Control, 1994, 116, 553-557.
64. Parr, E.A., Industrial Control Handbook, Vol. 3. BSP Professional Books, 1989.
65. Hang, C.C., Astrom, K.J. and Ho, W.K., Refinements of the Ziegler-Nichols tuning formula. IEE Proceedings,
Part D, 1991, 138(2), 111-118.
66. Hagglund, T. and Astrom, K.J., Industrial adaptive controllers based on frequency response techniques.
Automatica, 1991, 27(4), 599-609.
67. Leva, A., PID autotuning algorithm based on relay feedback. IEE Proceedings, Part D, 1993, 140(5), 328-338.
68. Astrom, K.J., Ziegler-Nichols auto-tuners. Report TFRT-3167, Department of Automatic Control, Lund
Institute of Technology, Lund, Sweden, 1982.
69. Calcev, G. and Gorez, R., Iterative techniques for PID controller tuning. Proceedings of the 34th Conference
on Decision and Control, New Orleans, LA ., USA, 1995, 3209-3210.
70. Cox, C.S., Arden, W.J.B. and Doonan, A.F., CAD Software facilities tuning of traditional and predictive
control strategies. Proceedings of the ISA/94 International Conference, Exhibition and Training Program.
Advances in Instrumentation and Control, Anaheim, CA., U.S.A., 1994, 49, 2, 241-250.
71. Vrancic, D., Peng, Y., Strmcnik, S. and Hanus, R., A new tuning method for PI controllers based on a process
step response. Proceedings of the CESA ’96 IMACS Multiconference Symposium on Control, Optimisation
and Supervision, Lille, France, 1996, 2, 790-794.
72. Vrancic, D., Design of anti-windup and bumpless transfer protection. Part II: PID controller tuning by
multiple integration method. PhD thesis, University of Ljubljana, J. Stefan Institute, Ljubljana, Slovenia, 1996.
73. Vrancic, D., Peng, Y., Strmcnik, S. and Juricic, D., A multiple integration tuning method for filtered PID
controller. Proceedings of the IFAC 1999 14th World Congress, Beijing, China, 1999, Preprints, Paper 3b-02-3.
74. Chien, I.-L., Huang, H.-P. and Yang, J.-C., A simple multiloop tuning method for PID controllers with no
proportional kick, Industrial Engineering Chemistry Research, 1999, 38(4), 1456-1468.

75. Tyreus, B.D. and Luyben, W.L., Tuning PI controllers for integrator/dead time processes. Industrial
Engineering Chemistry Research, 1992, 31(11), 2625-2628.
76. Wang, L. and Cluett, W.R., Tuning PID controllers for integrating processes. IEE Proceedings - Control
Theory and Applications, 1997, 144(5), 385-392.
77. Rotach, V. Ya. (1995). Automatic tuning of PID-controllers – expert and formal methods, Thermal
Engineering, 42, 10, 794-800.
78. Poulin, E. and Pomerleau, A.. PI settings for integrating processes based on ultimate cycle information. IEEE
79. Penner, A., Tuning rules for a PI controller. Proceedings of the ISA/88 International Conference and
Exhibition, Houston, Texas, USA, 1988, 1037-1051.
80. Srividya, R. and Chidambaram, M., On-line controllers tuning for integrator plus delay systems. Process
Control and Quality, 1997, 9, 59-66.
81. Tan, W., Liu, K. and Tam, P.K.S., PID tuning based on loop-shaping H∞ control. IEE Proceedings - Control
82. Poulin, E. and Pomerleau. A., PID tuning for integrating and unstable processes. IEE Proceedings - Control
83. McAvoy, T.J. and Johnson, E.F. (1967). Quality of control problem for dead-time plants, Industrial and
Engineering Chemistry Process Design and Development, 6, 4, 440-446.
84. Lopez, A.M., Smith, C.L. and Murrill, P.W. (1969). An advanced tuning method, British Chemical
Engineering, 14, 11, 1553-1555.
85. Hougen, J.O., Measurement and Control Applications. Instrument Society of America, Research Triangle
Park, North Carolina, 1979.
86. De Paor, A.M. and O'Malley, M., Controllers of Ziegler-Nichols type for unstable processes with time delay.
International Journal of Control, 1989, 49(4), 1273-1284.
87. Venkatashankar, V. and Chidambaram, M., Design of P and PI controllers for unstable first order plus time
delay systems. International Journal of Control, 1994, 60(1), 137-144.
88. Chidambaram, M., Design of PI and PID controllers for an unstable first-order plus time delay system.
Hungarian Journal of Industrial Chemistry, 1995, 23, 123-127.

89. Rotstein, G.E. and Lewin, D.E., Simple PI and PID tuning for open-loop unstable systems. Industrial
Engineering Chemistry Research, 1991, 30(8), 1864-1869.
90. Ho, W.K. and Xu, W., PID tuning for unstable processes based on gain and phase-margin specifications.
IEE Proceedings - Control Theory and Applications, 1998, 145(5), 392-396.
91. Luyben, W.L., Tuning temperature controllers on openloop unstable reactors, Industrial Engineering
Chemistry Research, 1998, 37, 4322-4331.
91a. Hansen, P.D. (2000). Robust adaptive PID controller tuning for unmeasured load rejection, Preprints of the
Proceedings of PID ’00: IFAC Workshop on digital control (Past, present and future of PID control),
Terrassa, Spain, pp. 487-494.
92. Poulin, E. and Pomerleau. A., Unified PID design method based on a maximum peak resonance specification.
IEE Proceedings - Control Theory and Applications, 1997, 144(6), 566-574.
93. Astrom, K.J. and Hagglund, T., Automatic tuning of PID Controllers, Instrument Society of America,
Research Triangle Park, North Carolina, 1988.
94. Sain, S.G. and Ozgen, C., Identification and tuning of processes with large deadtime. Control and Computers,
1992, 20(3), 73-78.
95. Cheng, G.S. and Hung, J.C., A Least-Squares Based Self-Tuning of PID Controller. Proceedings of the IEEE
South East Conference, Raleigh, North Carolina, USA, 1985, 325-332.
96. Gerry, J.P., How to control processes with large dead times, http://www.expertune.com/artdt.html, 1998.
97. Wang, F.-S., Juang, W.-S. and Chan, C.-T., Optimal tuning of PID controllers for single and cascade control
loops. Chemical Engineering Communications, 1995, 132, 15-34.
97a. Pi-Mira, J., Mateo, E., Sarrate-Estruch, R. and Quevedo-Casin, J., LS-3000 digital PID controller, Preprints of
the Proceedings of PID ’00: IFAC Workshop on digital control (Past, present and future of PID control),
Terrassa, Spain, 2000, pp. 465-471.
98. Astrom, K.J. and Hagglund, T., Automatic tuning of simple regulators with specifications on phase and
amplitude margins. Automatica, 1984, 20(5), 645- 651.
99. Li, Z., Su., X. and Lin, P., A practical algorithm for PID auto-tuning. Advances in Modelling and Analysis C,
ASME Press, 1994, 40(2), 17-27.

100. Suyama, K., A simple design method for sampled-data PID control systems with adequate step responses.
Proceedings of the International Conference on Industrial Electronics, Control, Instrumentation and
Automation, 1992, 1117-1122.
101. Juang, W.-S. and Wang, F.-S., Design of PID controller by concept of Dahlin’s Law. Journal of the Chinese
Institute of Chemical Engineers, 1995, 26(2), 133-136.
102. Camacho, O.E., Smith, C. and Chacon, E., Toward an implementation of sliding mode control to chemical
processes, Proceedings of the IEEE International Symposium on Industrial Electronics, 1997, 3, 1101-1105.
103. Ho, W.K., Lim, K.W. and Xu, W., Optimal gain and phase margin tuning for PID controllers. Automatica,
1998, 34(8), 1009-1014.
104. Ho, W.K., Lim, K.W., Hang, C.C. and Ni, L.Y., Getting more phase margin and performance out of PID
controllers, Automatica, 1999, 35, 1579-1585.
104a. Morilla, F., Gonzalez, A. and Duro, N., Auto-tuning PID controllers in terms of relative damping, Preprints of
the Proceedings of PID ’00: IFAC Workshop on digital control (Past, present and future of PID control),
105. Morari, M. and Zafiriou, E., Robust process control. Prentice-Hall Inc., Englewood Cliffs, New Jersey, 1989.
106. Horn, I.G., Arulandu, J.R., Gombas, C.J., VanAntwerp, J.G. and Braatz, R.D., Improved filter design in internal
model control. Industrial Engineering Chemistry Research, 1996, 35(10), 3437-3441.
106a. Normey-Rico, J.E., Alcala, I., Gomez-Ortega, J. and Camacho, E.F. (2000). Robust PID tuning application to a
mobile robot path tracking problem, Preprints of the Proceedings of PID ’00: IFAC Workshop on digital
control (Past, present and future of PID control), Terrassa, Spain, pp. 648-653.
107. Witt, S.D. and Waggoner, R.C., Tuning parameters for non-PID three-mode controllers. Hydrocarbon
Processing, 1990, June, 74-78.
108. Kaya, A. and Scheib, T.J., Tuning of PID controls of different structures. Control Engineering, 1988, July, 62-
65.
109. Tsang, K.M. and Rad, A. B., A new approach to auto-tuning of PID controllers. International Journal of
Systems Science, 1995, 26(3), 639-658.
110. Tsang, K.M., Rad, A.B. and To, F.W., Online tuning of PID controllers using delayed state variable filters.
Proceedings of the IEEE Region 10 Conference on Computer, Communication, Control and Power
Engineering, 1993, 4, 415-419.

111. Hang, C.C. and Astrom, K.J., Refinements of the Ziegler-Nichols tuning formuale for PID auto-tuners.
Proceedings of the ISA/88 International Conference and Exhibition. Advances in Instrumentation, 1988, 43,
3, 1021-1030.
112. Hang, C.-C. and Cao, L., Improvement of transient response by means of variable set-point weighting. IEEE
Transactions on Industrial Electronics, 1996, 43(4), 477-484.
113. Gong, X., Gao, J. and Zhou, C., Extension of IMC tuning to improve controller performance. Proceedings of
the IEEE International Conference on Systems, Man and Cybernetics, 1996, 1770-1775.
114. VanDoren, V.J., Ziegler-Nichols methods facilitate loop tuning. Control Engineering, 1998, December.
114a. Argelaguet, R., Pons, M., Quevedo, J. and Aguilar, J., A new tuning of PID controllers based on LQR
optimization, Preprints of the Proceedings of PID ’00: IFAC Workshop on digital control (Past, present and
future of PID control), Terrassa, Spain, 2000, pp. 303-308.
115. Blickley, G.J., Modern control started with Ziegler-Nichols tuning. Control Engineering, 1990, 2 October, 11-
17.
116. De Paor, A.M., A fiftieth anniversary celebration of the Ziegler-Nichols PID controller. International Journal
of Electrical Engineering Education, 1993, 30, 303-316.
117. Corripio, A.B., Tuning of industrial control systems. Instrument Society of America, Research Triangle Park,
North Carolina, 1990.
118. Mantz, R.J. and Tacconi, E.J., Complementary rules to Ziegler and Nichols' rules for a regulating and tracking
controller. International Journal of Control, 1989, 49(5), 1465-1471.
119. Atkinson, P. and Davey, R.L., A theoretical approach to the tuning of pneumatic three-term controllers.
Control, 1968, March, 238-242.
120. Perry, R.H. and Chilton, C.H., Chemical engineers handbook, McGraw-Hill, New York, 5th edition, 1973.
121. Luo, K.-N., Kuo, C.-Y. and Sheu, L.-T. (1996). A novel method for fuzzy self-tuning PID controllers,
Proceedings of the Asian Fuzzy Systems Symposium, 194-199.
122. Yu, C.-C. (1999). Autotuning of PID controllers, Advances in Industrial Control Series, Springer-Verlag
London Ltd.
123. Karaboga, D. and Kalinli, A., Tuning PID controller parameters using Tabu search algorithm. Proceedings of
the IEEE International Conference on Systems, Man and Cybernetics, 1996, 134-136.
124. Hang, C.C. and Astrom, K.J., Practical aspects of PID auto-tuners based on relay feedback. Proceedings of
the IFAC Adaptive control of Chemical Processes Conference, Copenhagen, Denmark, 1988, 153-158.
125. Shin, C.-H., Yoon, M.-H. and Park, I.-S., Automatic tuning algorithm of the PID controller using two Nyquist
points identification. Proceedings of the Society of Instrument and Control Engineers annual conference,
Tokyo, Japan, 1997, 1225-1228.
126. Harriott, P., Process Control, McGraw-Hill, New York, 1964.
127. Zhang, G., Shao, C. and Chai, T., A new method for independently tuning PID parameters. Proceedings of
the 35th Conference on Decision and Control, Kobe, Japan, 1996, 2527-2532.
127a. Garcia, R.F. and Castelo, F.J.P., A complement to autotuning methods on PID controllers, Preprints of the
128. Fu, M., Olbrot, A.W. and Polis, M.P., Comments on 'Optimal gain for proportional-integral-derivative
feedback'. IEEE Control Systems Magazine, 1989, January, 100-101.
129. Pessen, D.W., How to tune in a three-mode controller. Instrumentation, 1954, 7(3), 29-32.
130. Grabbe, E.M., Ramo, S. and Woolridge, D.E. (Editors), Handbook of automation, computation and control.
Vol 3: Systems and Components, John Wiley and Sons, New York, 1961.
131. Pessen, D.W., Optimum three-mode controller settings for automatic start-up. Transactions of the ASME,
1953, July, 843-849.
132. Ford, R.L., The determination of the optimum process-controller settings and their confirmation by means of
an electronic simulator. Proceedings of the IEE, Part 2, 1953, 101(80), April, 141-155, 173-177.
133. Luyben, W.L., Tuning proportional-integral-derivative controllers for integrator/deadtime processes.
Industrial Engineering Chemistry Research, 1996, 35(10), 3480-3483.
134. Belanger, P.W. and Luyben, W.L., Design of low-frequency compensators for improvement of plantwide
regulatory performances. Industrial Engineering Chemistry Research, 1997, 36(12), 5339-5347.
134a. Pecharroman, R.R. and Pagola, F.L., Control design for PID controllers auto-tuning based on improved
identification, Preprints of the Proceedings of PID ’00: IFAC Workshop on digital control (Past, present
and future of PID control), Terrassa, Spain, 2000, pp. 89-94.
134b. Pecharroman, R.R., Private communication, 9 May 2000.

135. Zhang, W., Xu, X. and Sun, Y., Quantitative performance design for integrating processes with time delay,
Automatica, 1999, 35, pp. 719-723.
136. Tan, W., Liu, J. and Sun, W., PID tuning for integrating processes. Proceedings of the 1998 IEEE
International Conference on Control Applications, Trieste, Italy, 1998, 2, 873-876.
137. Chien, I.-L. and Fruehauf, P.S., Consider IMC tuning to improve controller performance. Chemical
Engineering Progress, 1990, October, 33-41.
138. Oubrahim, R. and Leonard, F., PID tuning by a composed structure. Proceedings of the UKACC International
Conference on Control ’98, Swansea, Wales, 1998, 2, 1333-1338.
139. Sung, S.W., O, J., Lee, I.-B., Lee, J. and Yi, S.-H., Automatic tuning of PID controller using second-order plus
time delay model. Journal of Chemical Engineering of Japan, 1996, 29(6), 991-999.
140. Ho, W.K., Hang, C.C. and Cao, L.S., Tuning of PID controllers based on gain and phase margin
specifications. Automatica, 1995, 31(3), 497-502.
141. Ho, W.K., Hang, C.C. and Zhou, J., Self-tuning PID control of a plant with under-damped response with
specifications on gain and phase margins. IEEE Transactions on Control Systems Technology, 1997, 5(4),
446-452.
142. Ho, W.K., Hang, C.C., Zhou, J.H. and Yip, C.K., Adaptive PID control of a process with underdamped
response. Proceedings of the Asian Control Conference, Tokyo, Japan, 1994, 335-338.
143. Wang, Q.-G., Lee, T.-H., Fung, H.-W., Bi, Q. and Zhang, Y., PID tuning for improved performance. IEEE
144. Wang, Y.-G. and Shao, H.-H. (1999). PID autotuner based on gain- and phase-margin specification,
Industrial Engineering Chemistry Research, 38, 3007-3012.
145. Pemberton, T.J., PID: The logical control algorithm II. Control Engineering, 1972, 19(7), 61-63.
146. Smith, C.L., Corripio, A.B. and Martin, J., Controller tuning from simple process models. Instrumentation
Technology, 1975, December, 39-44.
147. Wang, T.-S. and Clements, W.C., Adaptive multivariable PID control of a distillation column with unknown
and varying dead time. Chemical Engineering Communications, 1995, 132, 1-13.
147a. Gorez, R. and Klan, P., Nonmodel-based explicit design relations for PID controllers, Preprints of the

147b. Seki, H., Ogawa, M. and Ohshima, M. (2000). Retuning PID temperature controller for an unstable gas-
phase polyolefin reactor, Preprints of the Proceedings of PID ’00: IFAC Workshop on digital control
(Past, present and future of PID control), Terrassa, Spain, pp. 473-478.
148. Landau, I.D. and Voda, A., An analytical method for the auto-calibration of PID controllers. Proceedings of
the 31st Conference on Decision and Control, Tucson, Arizona, USA, 1992, 3237-3242.
149. Jahanmiri, A. and Fallahi, H.R., New methods for process identification and design of feedback controller.
Transactions of the Institute of Chemical Engineers, 1997, 75(A), July, 519-522.
150. Hansen, P.D., Controller structure and tuning for unmeasured load disturbance. Proceedings of the American
Control Conference, Philadelphia, Pennsylvania, 1998, 1, 131-136.
151. Hang, C.C., Ho, W.H. and Cao, L.S., A comparison of two design methods for PID controllers. ISA
Transactions, 1994, 33, 147-151.
152. Polonyi, M.J.G., PID controller tuning using standard form optimisation. Control Engineering, 1989, March,
102-106.
153. Valentine, C.C. and Chidambaram, M., PID control of unstable time delay systems. Chemical Engineering
Communications, 1997, 162, 63-74.
154. Huang, C.-T. and Lin, Y.-S., Tuning PID controller for open-loop unstable processes with time delay.
Chemical Engineering Communications, 1995, 133, 11-30.
155. Skoczowski, S. and Tarasiejski, L., Tuning of PID controllers based on gain and phase margin specifications
using Strejc’s process model with time delay. Proceedings of the Third International Symposium on
Methods and Models in Automation and Robotics (MMAR ’96), Miedzyzdroje, Poland, 1996, 765-770.
156. Lennartson, B. and Kristiansson, B., Pass band and high frequency robustness for PID control. Proceedings
of the 36th IEEE Conference on Decision and Control, San Diego, California, U.S.A., 1997, 2666-2671.
157. Kristiansson, B. and Lennartson, B., Robust design of PID controllers including auto-tuning rules.
Proceedings of the American Control Conference, 1998, 5, 3131-3132.
158. Kristiansson, B. and Lennartson, B., Optimal PID controllers for unstable and resonant plants. Proceedings
of the Conference on Decision and Control, 1998.
158a. Kristiansson, B. and Lennartson, B., Near optimal tuning rules for PI and PID controllers, Preprints of the

159. Vrancic, D., Peng, Y. and Strmcnik, S., A new PID controller tuning method based on multiple integrations.
Control Engineering Practice, 1999, 7(5), 623-633.
160. Hwang, S.-H. (1995). Closed-loop automatic tuning of single-input-single-output systems, Industrial
Engineering Chemistry Research, 34, 2406-2417.
161. Ferretti, G., Maffezzoni, C. and Scattolini, R. (1991). Recursive estimation of time delay in sampled systems,
Automatica, 27, 653-661.
162. Nishikawa, Y., Sannomiya, N., Ohta, T. and Tanaka, H. (1984). A method for auto-tuning of PID control
parameters, Automatica, 20, 321-332.
163. Lee, J. and Sung, S.W. (1993). Comparison of two identification methods for PID controller tuning, AIChE
Journal, 39, 695-697.
164. Deshpande, P.B. (1980). Process identification of open-loop unstable systems, AIChE Journal, 26, 305-308.
165. Pecharroman, R.R. and Pagola, F.L. (1999). Improved identification for PID controllers auto-tuning,
Proceedings of the 5 th European Control Conference (ECC ’99), Karlsruhe, Germany, Paper F453, BA-12.
Appendix 1: List of symbols used (more than once) in the paper.
a 1, a2 = PID filter parameters
A m = gain margin
b = setpoint weighting factor
b1 = PID filter parameter
c = derivative term weighting factor
E(s) = Desired variable, R(s), minus controlled variable, Y(s)
FOLPD model = First Order Lag Plus time Delay model
FOLIPD model = First Order Lag plus Integral Plus time Delay model
G c ( s) = PID controller transfer function
G p ( jω ) = process transfer function at frequency ω
G p ( jω ) = magnitude of G p ( jω ) , ∠Gp ( jω) = phase of G p ( jω )
IAE = integral of absolute error
IMC = internal model controller
IPD model = Integral Plus time Delay model
ISE = integral of squared error
ISTES = integral of squared time multiplied by error, all to be squared
ISTSE = integral of squared time multiplied by squared error
ITAE = integral of time multiplied by absolute error
ki = integral gain of the parallel PID controller
k p = proportional gain of the parallel PID controller
kd = derivative gain of the parallel PID controller
Kc = Proportional gain of the controller
Km = Gain of the process model
Ku = Ultimate gain
K25% = proportional gain required to achieve a quarter decay ratio
m = multiplication parameter in the lead controller

Ms = closed loop sensitivity
N = determination of the amount of filtering on the derivative term
PP = perturbance peak = peak of system output when unit step disturbance is added
R(s) = Desired variable
RT= recovery time = time for perturbed system output (when a unit step disturbance is added) to come to its final
value
SOSPD model = Second Order System Plus time Delay model
Td = Derivative time of the controller
TCL = desired closed loop system time constant
Tf = Time constant of the filter in series with the PID controller
Ti = Integral time of the controller
Tm = Time constant of the FOLPD process model
Tm1, Tm2 ,Tm3 = Time constants of the higher order process models
Tu = Ultimate time constant
T25% = period of the quarter decay ratio waveform
TOLPD model = Third Order Lag Plus time Delay model
TS = settling time
U(s) = manipulated variable
Y(s) = controlled variable
λ = Parameter that determines robustness of compensated system.
ξ m = damping factor of an underdamped process model, ξ = damping factor of the compensated system
κ = 1 K m Ku
φ = phase lag, φ m = phase margin, φ ω = phase lag at an angular frequency of ω
φ c = phase corresponding to the crossover frequency
τ m = time delay of the process model, τ = τ m ( τ m + Tm )
ω = angular frequency, ω u = ultimate frequency, ω φ = angular frequency at a phase lag of φ

PI and PID Controller Tuning Rules For Time Delay Processes

Uploaded by

PI and PID Controller Tuning Rules For Time Delay Processes

Uploaded by

PI and PID controller tuning rules for time delay processes: a summary

Technical Report AOD-00-01, Edition 1

Keywords: PI, PID, tuning rules, time delay.

2. PID controller structures

E. Second order system plus time delay (SOSPD) model ( G m (s) = , )

Method 1: Km , Tm , τ m assumed known.

H. Third order system plus time delay (TOLPD) model ( G m (s) = ).

Method 1: Km , Tm1 , Tm2 , Tm3 , τ m known.

I. Unstable first order lag plus time delay model ( G m (s) = )

J. Unstable second order system plus time delay model ( G m (s) = )

K m (1 + b1s + b2 s2 + b3s3 + b4 s4 + b ss5 ) e− s τ m

M. Fifth order system plus delay model ( G m ( s) = )

N. General model with a repeated pole ( G m ( s) = )

O. General stable non-oscillating model with a time delay

4. Organisation of the report

Tables 1-3: PI tuning rules – FOLPD model - G m ( s) =

[3] - page 205-208.

Tables 4-7: PI tuning rules - non-model specific

Tables 12-14: PI tuning rules – FOLIPD model G m (s) =

Tables 15-16: PI tuning rules – SOSPD model or

Table 17: PI tuning rules – SOSIPD model (repeated pole) G m (s) =

Table 20: PI tuning rules - unstable FOLPD model

Table 21: PI tuning rules - unstable SOSPD model

Table 22: PI tuning rules – delay model e −s τ m

Tables 23-40: PID tuning rules - FOLPD model

Tables 41-48: PID tuning rules - non-model specific

Table 43: Ideal controller with set-point weighting:

defined by Mantz and Tacconi [118].

Tables 58-67: PID tuning rules - FOLIPD model

Table 65: Ideal controller with set-point weighting:

Table 67: Two degree of freedom controller:

Tables 68-79: PID tuning rules - SOSPD model or

Table 89-90: PID tuning rules - TOLPD model

Tables 91-93: PID tuning rules - unstable FOLPD model

Tables 94-97: PID tuning rules - unstable SOSPD model G m ( s) =

** some more information needed.

Process Minimise Direct Ultimate Robust

Process Minimise Direct Ultimate Robust

Process Minimise Direct Ultimate Robust

Table D: PI controller structure and tuning rules – a summary

Total 124 42 18 17 49 30 280

Minimum ITAE – Yu [19] − 0. 057

Model: Method 6  0.808 0.511 0.255  T Kc Km τm

0. 581Tm Closed loop response

0. 696Tm Closed loop response

0. 801Tm Closed loop response

0. 853Tm Closed loop response

0. 906Tm Closed loop response

0. 957 Tm Closed loop response

1.008 Tm Closed loop response

Tm Closed loop response

Tm  0. 3852 0.723 τ  KcKm τm

0.5  Tm 3 + Tm2 τ m + 0.5Tm τ m 2 + 0.167τ m 3   Tm3 + Tm 2τ m + 0.5Tmτ m 2 + 0.167τ m 3 

Model: Method 6 2Tm + τ m

Shinskey [59] – page 167. Tu  0.87 − 0.855 u + 0.172  u  

Model: Method 8 ( 1 − µ1 ) K u , Kc µ 2 Ku ω u , Decay ratio = 0.1, r = 0.75 -

Servo – minimum ISTSE – τm

 0.46 − 0.0835 τ m + 0.0305 τ m   K

dominant pole design – . ≤ m ≤ 55

. e− 4.1τ + 5.7 τ Tm 8.9τ m e −6.6 τ + 3.0 τ or b = 0.44e0.78 τ − 0.45τ ;

Table 3: PI tuning rules - FOLPD model - G m ( s) = – Two degree of freedom controller:

= Tm  0.06216 + 3.171 m − 3.058  m  + 1. 205  m  

0.5 4 Alfa-Laval Automation

0.25 1.6 Alfa-Laval Automation

0.4830 3.7321 Gain margin = 2,

1.18 K u K m − 1. 72 1.18K u K m − 1.72 0.1 ≤ Ku Km ≤ 0.5

1.9κ − 1.3κ 2 − 4.4κ + 2. 7κ 2

λ Overshoot TS PP RT Zhang et al. [135]

Model: Method 2 0.942 . τm

Table 12: PI tuning rules – FOLIPD model G m (s) =

Pomerleau [82] – deduced K m ( τ m + Tm ) a τ + T 2 + 1

Table 13: PI tuning rules – FOLIPD model G m (s) =

Table 15: PI tuning rules - SOSPD model or

et al. [39] – repeated pole β Ti ω φ 1 + βTm ω φ [

− 167667 e T + 14.5737eξ − 7.3025 e 

+ Tm1 359 .3345ξ m  m  − 158.7611 m ξ m 2 − 770.2897ξ m2 − 153633