Aalborg Universitet

An active defrost scheme with a balanced energy consumption and food quality loss in
supermarket refrigeration systems

Cai, Junping; Stoustrup, Jakob; Rasmussen, Bjarne Dindler

Published in:
Elsevier IFAC Publications / IFAC Proceedings series

Publication date:
2008

Document Version
Early version, also known as pre-print

Link to publication from Aalborg University

Citation for published version (APA):

Cai, J., Stoustrup, J., & Rasmussen, B. D. (2008). An active defrost scheme with a balanced energy
consumption and food quality loss in supermarket refrigeration systems. Elsevier IFAC Publications / IFAC
Proceedings series, (1).

General rights
Copyright and moral rights for the publications made accessible in the public portal are retained by the authors and/or other copyright owners
and it is a condition of accessing publications that users recognise and abide by the legal requirements associated with these rights.

? Users may download and print one copy of any publication from the public portal for the purpose of private study or research.
? You may not further distribute the material or use it for any profit-making activity or commercial gain
? You may freely distribute the URL identifying the publication in the public portal ?
Take down policy
If you believe that this document breaches copyright please contact us at [email protected] providing details, and we will remove access to
the work immediately and investigate your claim.
Proceedings of the 17th World Congress
The International Federation of Automatic Control
Seoul, Korea, July 6-11, 2008

An Active Defrost Scheme with a Balanced Energy

Consumption and Food Quality Loss in
Supermarket Refrigeration Systems
J. Cai, J. Stoustrup ∗ B. D. Rasmussen ∗∗
∗ Automation and Control, Department of Electronic Systems, Aalborg
University, 9220 Aalborg, Denmark. (e-mail: [email protected],
[email protected])
∗∗ Business Unit Industry and Water Services, Grundfos Management A/S,
8850 Bjerringbro, Denmark. (e-mail: [email protected] )

Abstract: This paper introduces food quality as a new parameter, together with energy, to determine
an optimal cooling time between defrost cycles. A new defrost-on-demand scheme is proposed. It
uses a feedback loop consisting of on-line model updating and estimation as well as a model based
optimization. This scheme automatically adjusts the time interval between defrost cycles with varying
operating conditions, continuously seeking an optimal time interval, featuring either an energy optimal
time, or a trade-off between energy consumption and food quality loss. This adaptive approach is
compared with traditional defrost schemes, found to be able to reduce energy consumption significantly.

1. INTRODUCTION quality. Simulation results are presented in Section 5. Gains by

the new defrost-on-demand control scheme is demonstrated in
Supermarkets are one of the most energy-intensive types of Section 6. Finally some discussions and conclusions are given
commercial buildings. Refrigeration is the largest component in Section 7.
of their electric energy usage, accounting for half or more of
the store total. Among others, energy associated with defrosts 2. SIMPLE INTRODUCTION OF REFRIGERATION
and anti-sweat heaters in supermarkets may exceed 30% of SYSTEM AND DEFROST
the total energy requirement, see Howell et al. [1999]. Thus,
from an energy point of view, determining how often to defrost
2.1 Refrigeration of foodstuffs in a supermarket
is an important issue. From a food quality point of view,
it is also important. During defrosting, the air temperature
inside a display cabinet will normally increase, and so will the The display cabinet depicted in Fig. 1 consists of a food
food temperature. Depending on the defrost method, the food container and an air tunnel. The evaporator inside the air tunnel
temperature will stay out of the usually controlled range for a cools the passing air, which circulates around the food container
period of time, which is harmful to the food quality, see Cai and creates an air blanket on the top of foodstuffs.
et al. [2006]. To control properly the temperature inside a cabinet, one or
Defrost approaches can be classified in two major schemes: more thermal sensors are required in the system. The number
one is scheduled defrost, another is defrost-on-demand. Fahlen and function of those sensors differ from one application to
[1996] studied and compared these two schemes, and found another. For example in our system, there are 3 sensors: S3 , S4 ,
that both of them have some excellent features, but various S5 . S3 and S4 are used to measure the temperature of air inlet
limitations and drawbacks. Food quality has never been used Ta,i and outlet Ta,o respectively, S5 is a defrost stop sensor.
as an active decision factor in either of these schemes. To increase sales, any physical obstacles between the products
Biotechnology provides us the knowledge of food quality and customers should be avoided. As a consequence, for most
change during the refrigerated storage. Research on frost for- display cabinets, one or more forced air blankets (for horizontal
mation enables us to predict the system performance degrada- display) or air curtains (for vertical display) are the only barrier
tion under frosting conditions. between the refrigerated foodstuffs and warm ambient air. Due
to the disturbance of air flow, a more or less significant amount
This paper introduces food quality as a new parameter, together of warm ambient air is always entrained, which introduces the
with energy, to determine an optimal time between defrost frost formation, reduces the temperature control capabilities
cycles depending on ambient parameters. Based on this, a new and increases energy consumption.
defrost-on-demand scheme is proposed.
The paper starts with a simple introduction to supermarket 2.2 Requirements on food storage temperature
refrigeration systems and the most commonly used defrost
methods, defrost schemes, which is in Section 2. We propose a In supermarkets, there are general requirements regarding the
new defrost-on-demand control scheme in Section 3. In Section storage temperature for different foodstuffs in display cabinets.
4, we discuss the problem connected with traditional defrost For example in Denmark, according to Announcement [2004]
schemes by using three models, two for energy and one for food and DSK et al. [2004].

978-3-902661-00-5/08/$20.00 © 2008 IFAC 9374 10.3182/20080706-5-KR-1001.1934

17th IFAC World Congress (IFAC'08)
Seoul, Korea, July 6-11, 2008

Tstore , RH store stop temperature while with a maximum defrost time as a

security.
Infiltration • Defrost-on-demand: Initiating the defrost cycle only when
necessary, see Llewelyn [1984]. This approach normally
S3 uses one parameter to initiate and terminate the defrost
process, such parameters could be: air pressure difference
Ta ,i across the evaporator, fan power sensing etc.
Food storage
Features and shortcomings of the current defrost schemes are:
• Scheduled defrost is simple and uses a low cost controller,
S4 S5 so it is the most commonly used defrost scheme in today’s
. supermarkets, but the time schedule is normally deter-
Ta ,o Va mined based on experience and observation, most cases,
based on worst case conditions. It is configured during
the commission stage and can not automatically adapt to
the varying shop conditions under which the system is
working, so the time between two defrost cycles may be
Fig. 1. A simplified display cabinet in a supermarket either too long or too short.
• Existing defrost-on-demand schemes typically involve the
• Frozen food, the max. temperature is -18◦ C. installation of an additional sensor to detect frost build-
• Fresh fish and fish products, the max. temperature is +2◦ C. up, and use one parameter to initiate and terminate the
• Milk, the max. temperature is +5◦ C. defrost cycle. The threshold of this detected parameter is
Usually, the legally constrained temperature is the air tempera- determined mainly to ensure a safe operation, or maintain
ture. the performance degradation of the system within fixed
limits over the whole range of operating conditions. No
2.3 Frost formation and defrost methods energy optimality is guaranteed.
• None of the existing schemes have used food quality as a
Frosting is a well known and undesirable phenomenon on decision factor.
evaporator coils. It happens whenever the surface temperature
of the evaporator is below 0◦ C and humid air passes by. Frost 3. DEFROST CONTROLLER DESIGN
decreases the performance of the heat exchanger. In order to
maintain a satisfactory performance, defrosting needs to be To overcome the shortcomings of the current defrost schemes,
done regularly. and realize an objective with a balanced system energy con-
sumption and food quality loss, we propose a new defrost-on-
The most common defrost methods for medium and low tem- demand control scheme, see Fig. 2. It uses a feedback loop
perature applications are: consisting of an on-line model updating and estimation by
• Off-cycle defrost: This is the simplest defrost methods: an Extended Kalman Filter (EKF), as well as a model based
refrigeration is stopped, evaporator fans continue to move optimization (see below). zm is the measured output. It could
room air over the frosted coil surface, which warms and advantageously use some extra sensors, but this scheme would
melts the frost. also work just with the existing sensors in the system. For
• Hot/cool gas defrost: During a hot/cool gas defrost, the example, for the refrigeration system in Fig. 1, if S3 is the
normal supply of cold refrigerant is stopped. The former controlled temperature by the normal controller, S4 could be
involves the circulation the hot gas from the compressor a good candidate for on-line measurement. Here we assume
discharge manifold directly to the display cabinet, and that the store temperature Tstore and relative humidity RHstore
the latter utilizes cooler gas from the liquid receiver. The are measured, such as at every half or at every full hour, de-
cool or hot gas condenses in the evaporator, releasing heat pending on the stability of store indoor conditions. d f r,opt,ave
which melts the ice from the coil. is the average optimal frost thickness for defrosting generated
• Electric defrost: This approach uses the electric heater by the optimization, but defrost will only be initiated when
embedded on the fin surface to supply heat, warm and melt the estimated frost thickness d f r from the EKF is equal to or
the frost. larger than d f r,opt,ave . This initiating signal is sent to the normal
defrost controller, defrosting starts, and it is terminated in a
Normally, for medium temperature applications, the cheapest normal way. After the defrost is complete, a reset signal is sent
and least energy consuming means of defrost is off-cycle. to the EKF, and the process for the next defrost cycle starts
For low temperature applications, electric defrost is the most again.
commonly used.
3.1 Model based optimization
2.4 Defrost schemes
The optimization objective is described as the following:
In a simple taxonomy, there are the following two classes of
defrost schemes. min E(t) + k · Q f ood,loss (t) (1)
topt

• Scheduled defrost: Initiating the defrost cycle by a timer, Where E(t) is the system energy consumption, which includes
normally with a fixed number of defrost cycles per day. two parts, one is the energy used direct for defrosting, another
Defrost is terminated either based on a fixed time or on a is the extra energy used for compensating the degraded system

9375
17th IFAC World Congress (IFAC'08)
Seoul, Korea, July 6-11, 2008

• It increases the thermal resistance between the fin and

airflow, and decreases the cooling capacity of heat ex-
RH store , Tstore changers.
Reset signal • It substantially reduces the airflow through heat exchang-
ers, and increases the air pressure drop through heat ex-
Defrost Refrigeration zm EKF changers. Depending on the characteristic of the fan, sev-
Controller System eral hours later, the airflow path may be nearly or com-
Yes
d fr pletely blocked.
Update state:
d fr >= d fr ,opt ,ave
frost thickness Frost build up is a complex process even on a flat plate. It
Optimization Update is affected by many factors, such as air flow velocity, air and
d fr ,opt ,ave Q/E Model parameters
plate temperature, air humility ratio etc. Frost growth on a
real evaporator becomes even more complex, limited modeling
exists, see Yang et al. [2006].
Fig. 2. On line new defrost-on-demand control scheme
This paper is not aiming at developing a detailed model to
efficiency due to frost build-up. Q f ood,loss (t) is the quality loss predict the frost formation under varying conditions. Instead
of foodstuffs, k is a weighing factor based on costs or shop it uses energy correlations to calculate how much water is
owners’ priorities, topt is the optimal cooling time between condensed on the surface of evaporator as frost. The purpose of
defrosting. modeling here is for controlling, so the model itself is extremely
To determine an optimal frost thickness threshold under dy- simple but still captures the main dynamical features seen from
namic situations is not easy; a simple method is proposed, an input / output point of view.
d f r,opt,ave is approximated by the average value of optimal Frost growth rate estimation: The nominal cooling demand
frost thickness under the whole working range. Simulation re- Qnom for a display cabinet can be calculated as follows (without
sults under the following conditions: [20◦ C, 25◦ C] [50%, 60%] special notification, all the units are standard SI units):
showed that by using the average to initiate defrosting, the
4
maximal energy loss compared with using the true optimal Qnom = Q0 ∑ (Gi · Xi ·Yi ) (2)
value is less than 4%. The alternative using a detailed model for i=1
determining the value given by Eq. (1) has from a control point Where Q0 is the standard cooling demand, can be calculated
of view a serious lack of robustness, and realistic modeling according to cabinet category and dimension. Gi is the correc-
errors could easily cause larger deviations than 4%. Moreover tion factor for the difference between testing conditions and
this 4% in worst-case is insignificant relative to the potential measured actual operating conditions, Xi is the load distribution
savings demonstrated below. factor, Yi is the load reduction factor related to the covering of
the display case, if no physical covering Yi = 1. i indicate the
3.2 Features and advantages of the new controller
load type, 1 for the load from heat conduction, 2 for infiltration,
• Adaptivity: The approach suggested uses real time dis- 3 for radiation and 4 for the load from electric equipments, such
turbance measurements (store temperature and humidity) as light, fan, anti-sweat etc.
for on-line model updating, estimation and optimization,
The correction factor for the infiltration and the infiltration load
continually seeking an optimal time for defrosting under
can be calculated as follows:
dynamic conditions.
Istore,m − Icab,m
• Optimality: The approach suggested is based on an op- G2 = (3)
timality condition which is a weighted function between Istore,test − Icab,test
system energy consumption and food quality loss, so Qin f = Q0 · (G2 · X2 ·Y2 ) (4)
the resulting closed-loop system will always operate on Where I are specific enthalpy, which can be calculated based on
the Pareto-optimal trade-off curve between system energy temperature and RH. Subscript store, cab represent store and
consumption and food quality loss. cabinet respectively, test and m refer to test and measurement.
• Feasibility: The proposed method is a model based control Parameters and details see Holm et al. [1996].
method, and introduction of an EKF can avoid some spe-
cial sensors. Those sensors are normally expensive, and Infiltration is caused by an amount of hot humid air from the
often some reliability and feasibility problems are asso- store entrained in the display cabinet. The load of infiltration
ciated with the complex and unreliable sensing methods. can also be calculated in another way:
The EKF can infer those values of interest from some Qin f = ṁa,ent (Istore,m − Icab,m ) (5)
available measurements. Thus the new controller can be From the above correlation, we can calculate the mass flow rate
implemented directly on the top of existing systems, no of air entrained ṁa,ent .
physical rearrangements or extra components installations
are required. Infiltration is the only source of water condensates on the
surface of evaporator and eventually becomes frost. As time
4. ENERGY AND QUALITY MODELING goes, it will increase both the thickness and density of frosts.
ṁ f r = ṁa,ent (xstore − xcab ) (6)
4.1 Extra energy to compensate for the reduced performance
under frosting ṁ f r = ρ f r · δ̇ f r · A f r (7)
Where x is the specific humidity or humidity ratio of air, based
Frosting of the heat exchanger surface affects its thermal per- on temperature and RH, ṁ f r is the frost mass growth rate. ρ f r
formance in the following ways, see Chen et al. [2003]: and A f r are the frost density, frosting area respectively. δ̇ f r is

9376
17th IFAC World Congress (IFAC'08)
Seoul, Korea, July 6-11, 2008

the frost thickness growth rate. Here we assume the density of fresh fish products as an example, according to the requirements
frost is a constant. A review on frost properties and modeling from food authorities, the maximum storage temperature is
was given by Iragorry et al. [2004]. +2◦ C. It uses electric defrost. Details see Cai [2007].
Frosted fin efficiency: Frosted fin efficiency η f according to Power and extra energy consumption: In this system, we focus
Barrow [1985] can be calculated as follows. Eq. (8) and (9) on two power consuming components: compressor and evap-
applies to both dry and frosted conditions. orator fan (the power consumption of the condenser fan has
η f = tanh mL/(mL) (8) no direct relation with frosting). Extra energy means that if we
s do not defrost, the efficiency of the system will degrade with
ha frosting, in order to meet the same cooling demand, more power
m= (9)
k f r δ f r + k f t f /2 is needed compared with frost free conditions.
Where L is the effective length of fin, m is a fin parameter, k is Wtot (t) = Wcomp (t) +W f an (t) (15)
used for conductivity, h for heat transfer coefficient, δ and t for Wextra (t) = Wtot (t) −Wtot (0) (16)
thickness, subscript a, f r and f refer to air, frost and fin. Where 0 is the frost free time, t is the time for frost growth.
The overall heat transfer coefficient U based on the total air side 4.2 Direct energy use for defrosting
area is given by:
1 Aa
µ µ ¶
Af
µ
ha k f r
¶µ
Af
¶¶−1 In order to maintain a satisfactory performance of heat exchang-
= + ha ηf + 1− ers, a periodic defrosting is required to remove frost. During
U hr Ar Aa k f r + ha δ f r Aa
a defrost cycle, the cooling system is shut down, and heat is
(10)
supplied to the heat exchanger to raise its temperature well
Where A is used for area, subscript r refer to refrigerant. above freezing.
Evaporator fan: In refrigeration systems, axial fans or centrifu- Energy distribution in a defrost cycle is:
gal fans are commonly used. The operating point of the fan
installed in a system is established at the intersection of the fan • Energy used to warm and melt frost Ed f , f r .
and device curve. Fig. 3 shows a system and fan interaction. • Energy used to heat the coil of heat exchanger Ed f ,coil .
• Energy used to heat the refrigerant Ed f ,r .
shut off pressure
• Energy wasted (the defrosting efficiency).
normal operating range
Static Pressure P [Pa]

4.3 Food quality loss under defrosting

Food quality decay is determined by its composition factors and

fan operating point
cu
rve many environmental factors, such as temperature, relative hu-
midity, light etc. Of all the environmental factors, temperature
is the most important.
system resistant curve max. free delivery
Food temperature T f ood is determined by the cabinet air temper-
ature Tair . For simple calculations, we can lump the food into
3
Volume flow rate V [ m / s] one thermal mass.
dT f ood
(mCp) f ood = UA(T f ood − Tcab ) (17)
dt
Fig. 3. Fan and system interaction
Food quality loss Q f ood,loss can be calculated as follows:
Z tf
COP (Coefficient Of Performance): System COP and compres- T f ood − Tre f
Q f ood,loss = 100 · DT,re f exp( )dt (18)
sor power WC can be calculated as follow: t0 z
ηc,nom = Z(C + D · V̇r ) (11) Where DT,re f , Tre f , Z are quality parameters, see Cai et al.
Te + 273.15 [2006]. UA is the heat transfer coefficient and area from air
εc = (12) to products, mCp f ood is the thermal mass and properties of
Tc − Te foodstuffs. Here we assume Tcab is the same as Ta,i , during
COP = ηc,nom εc (13) normal operation, T f ood and Tcab are equal, UA/(mCp) f ood =
WC = Qnom /COP (14) 3.97 · 10−4 .
where Z,C, D are constants for compressors. V̇r is the volume
flow rate of refrigerant, Te is the evaporating temperature. 5. SIMULATION RESULT
Tc is the condensing temperature, assumed to be a constant.
Parameters and details see Holm et al. [1996]. We are aiming at finding an optimal time interval between
defrost cycles to meet our optimization objective, which is a
Overall calculation procedure: Here we use the heat ex- weighted function between system energy consumption and
changer from Blundell [1977] as an e example. It has wavy food quality loss. Regarding the energy consumption, we need
continuous fins on circular tubes, with a staggered array. Slight to consider two aspects: extra energy and energy for defrosting.
modifications have been made on the dimension to meet our
cooling demand. The overall calculation procedure includes 2 5.1 Extra energy
loops: one is the time integration loop for frost growth, another
is an internal iterative loop for finding the operating point of fan. Simulation is carried out for 11 hours under frost formation,
Here we use an island site medium temperature display case for with a store temperature of 25◦ C and RH of 55%. Fig. 4

9377
17th IFAC World Congress (IFAC'08)
Seoul, Korea, July 6-11, 2008

shows the operating points of the fan as a function of time, frequent defrost. More precisely, it is the specific enthalpy and
due to frosts build up. When the evaporator is clean, the humidity ratio that determine the frost formation rate.
fan provides a reasonable high air flow. As pressure drop
increases, the air flow is dramatically decreased. After 11 hours, From Fig. 11, focusing on the energy aspect, we can see that
the fan is already working out of its normal operating range. if we configure the defrosting of the system at an optimal
When the air flow rate decreases, the overall heat transfer time interval of 9 hours, according to one initial condition of
coefficient between the air and evaporator will decrease. In 20◦ C, 50% RH, when the store temperature rises up to 25◦ C,
order to meet the same cooling demand, the temperature drop same RH, this 9 hours scheme will lead to a daily energy
of the air across the coil must increase. This, in turn requires consumption of 229.8 kJ. Compared with its actual energy
a lower evaporating temperature, see Fig. 5. The drop in the optimal point of 101.5 kJ at 6 hours, an extra 126.4% of energy
evaporating temperature will cause a lower COP and increased is wasted.
power consumption. Fig. 6 shows the compressor and fan power
consumption for the same cooling demand as a function of time 7. DISCUSSION AND CONCLUSION
under frosting.
This paper discussed the problems related to the traditional
defrost schemes. Through the analysis on both system energy
5.2 Defrost energy
and food quality, we propose a new way of determining the opti-
mal time between defrost cycles, and a new defrost-on-demand
Fig. 7 shows the energy used to warm the coil and to melt frost control scheme. It on-line adjusts the cooling time between
as a function of time between defrosting. From the figure we defrost cycles, according to the varying operation condition,
can see that the longer time we wait for initiating the defrost, the continuously seeking an optimal time interval, featuring either
more energy is needed both for melting the frost and warming an energy optimal point, or a trade-off between system energy
the coil. This is because, on one hand, frost accumulates with consumption and food quality loss.
time. On the other hand, the coil will become colder when the
evaporating temperature goes down, it needs more energy to be
REFERENCES
warmed up.
Announcement. Announcement 1271 of 13/12/2004 con-
5.3 Food quality loss cerning hygiene of foodstfuffs by Danish authorities,
http://www.foedevarestyrelsen.dk. 2004.
Fig. 8 shows the daily food quality loss under different defrost H. Barrow. A note on frosting of heat pump evaporator surfaces.
frequencies. Heat Recovery Systems, 5(3):195– 201, 1985.
C. J. Blundell. Optimising heat exchangers for air-to-air space-
5.4 Energy vs. Quality heating heat pumps in the united kingdom. Energy Research,
1:69– 94, 1977.
J. Cai. Model based control of refrigeration systems. PhD
We use one day as an example, and assume we defrost the
thesis, Automation and Control, Department of Electronic
system 2, 3... up to 6 times, then the cooling time between two
Systems, Aalborg University, Denmark, 2007.
defrost cycles will be 12, 8... 4 hours (defrost time is ignored),
J. Cai, J. Risum, and C. Thybo. Quality model of foodstuffs in
this is also the time that we allow frost to grow and the system
a refrigerated display cabinet. Purdue, USA, 2006. 11th Int.
performance to degrade. We plot the daily energy consumption
Refrigeration and Air Condition Conference.
and daily food quality loss in Fig. 9. From the figure, we can
H. Chen, L. Thomas, and R. W. Besant. Fan supplied heat
conclude that from an energy point of view, we should select an
exchanger fin performance under frosting conditions. Int. J.
optimal cooling time of 5 hours. But from the food quality point
Refrigeration, 26(1):140– 149, 2003.
of view, we should defrost at a longer interval, such as 11 hours.
DSK, Danish Supermarket Group, and COOP. Hygiene and Self
This is a conflicting requirement to supermarket owners. It is up
Control, Regulation for Supermarkets. 2004.
to them to make the final decision, based on their preference on
P. Fahlen. Frosting and defrost of air coils - results from
quality, or cost, or a trade-off.
laboratory testing. 1996.
H. V. Holm, P. O. Danig, and B. D. Rasmussen. Energy saving
6. GAINS FROM NEW DEFROST-ON-DEMAND for remote refrigeration and frozen equipments in trading
CONTROL SCHEME and service sector, calculations of norm energy consumption,
report nr. 3 (danish). 1996.
The above simulation is based on one specific situation, where R. H. Howell, L. Rosario, D. Riiska, and M. Bondoc. Potential
the store has a constant temperature and relative humidity, saving on display case energy with reduced supermarket rel-
which is more or less true for a store with air conditioning ative humidity. Sydney, Australia, 1999. 20th Int. Congress
systems, while in some European countries, such as Denmark, of Refrigeration, IIR/IIF.
this is not the case. The indoor environment will normally vary J. Iragorry, Y. X. Tao, and S. Jia. Review article: A critical
with outdoor condition, staff and customers’ activities. The review of properties and models for frost formation analysis.
fixed optimal cooling time which is determined off-line and Int. J. Refrigeration, 10(4):393– 420, 2004.
configured at the commissioning phase, as conditions change, D. S. Llewelyn. A significant advance in defrost control. Int. J.
may not be the best choice any more. Refrigeration, 7(5):334– 335, 1984.
Fig. 10 shows the energy optimal cooling time under different D. K. Yang, K. S. Lee, and S. Song. Modeling for predicting
store conditions. Generally speaking, a high store temperature frosting behavior of a fin tube heat exchanger. Int. J. Heat
and RH gives more load to the system, a faster frost growth, and Mass Transfer, 49:1472– 1479, 2006.
and a quicker performance degradation, which requires more

9378
17th IFAC World Congress (IFAC'08)
Seoul, Korea, July 6-11, 2008

60
40
Fan curve

35 defrost 4 times
50
after 11hrs frosting

30
Defrost 2 times
40
Static Pressure [Pa]

Food Quality Loss [%]

25
after 7 hrs frosting
30
20 No defrost

20 15

clean coil
10
10
5

0
0 200 400 600 800 1000 1200 0
0 4 8 12 16 20 24
Volume flow rate [m3/h] Time [hr]

Fig. 4. Operating point of fan as a function of time under frost Fig. 8. Food daily quality loss under defrosting frequencies
build-up
1400 21.8
−2
Ta,o
Te 1200 21.6
−4

Daily energy consumption [kJ]

1000 21.4

Daily food quality loss [%]

−6
Temperature [C]

800 21.2

−8
600 21

−10
400 20.8

−12 200 20.6

0 20.4
−14 4 5 6 7 8 9 10 11
0 1 2 3 4 5 6 7 8 9 10 11
Cooling time between two defrost cycles [hr]
Time [hr]

Fig. 5. Evaporation and air outlet temperature as a function of Fig. 9. Daily Energy consumption and Food quality loss as a
time under frost build-up function of cooling time between defrosting

600
1400
Tstore=25 C, RHstore=55%
500 Tstore=25 C, RHstore=50%
1200
Daily energy consumption [KJ]

400
Power consumption [W]

1000

Wcomp
300
Wfan 800
Wtot
200 Wextra 600
Energy optimal point

100
400

0
200

−100
0 1 2 3 4 5 6 7 8 9 10 11 0
4 5 6 7 8 9 10 11
Time [hr]
Cooling time between two defrost cycles [hr]

Fig. 6. Power consumption for compressor, fan, total and extra Fig. 10. Energy optimal time under different store RH
as a function of time as frost build-up
600
160
Edf,fr Tstore=25 C, RHstore=50%

140 Edf,coil Tstore=20 C, RHstore=50%

500
Edf,tot
Daily energy consumption [KJ]

120
Energy consumption [KJ]

400
100

Energy optimal point

300
80

60 200
δJ
40
100
20

0
0 4 5 6 7 8 9 10 11
0 1 2 3 4 5 6 7 8 9 10 11
Cooling time between two defrost cycles [hr]
Time for frost growth [hr]

Fig. 7. Energy consumption for melting frost, warming coil and Fig. 11. Potential gains on energy by the new defrost-on-
the total as a function of time demand control scheme

9379