Volume Compression Ratio Determination using Newton Raphson

Problem Statement

Determine the volume compression ratio (r 1) if the efficiency of the engine is 0.651 given the ratio of
specific heats of air (k) is 1.4, volume expansion ratio (r 2) is 11.82 and ratio of pressure at states 3 and 2
(p3/p2) is 1.3657. The efficiency of the engine is expressed as,

1 p 3 r1 k

η=1−
( )( ( ) )
r 1k−1 p 2 r2
−1

p3 p r1
p2 ( ){ }
−1+k 3
p2 r2
−1

Use any root-finding method except bracketing method to solve the problem. Tolerance is set at
0.00001 and provide valid and logical initial guesses to solve the problem.

Results and Discussion

Given the values of some parameters, these were substituted in the efficiency equation to be
manipulated giving a revised equation in terms of the volume compression ratio.
1.4
1 r1

0.651=1−
( r 11.4−1 )( 1.3657(11.82 ) −1)
r1
1.3657−1+1.4 ( 1.3657 ) { 11.82
−1 }
1

1−0.651=
( )
r1 0.4 ( 0.0430220244 r 11.4−1 )
0.3657+0.1617580372r 1−1.91198

0.349 ( 0.1617580372 r 1−1.54628 ) =0.0430220244 r 1−r 1−0.4

0=0.01343153058 r 1+ r 1−0.4−0.53965172

f (r 1)=0.01343153058 r 1+ r 1−0.4−0.53965172
Getting the derivative of the derived equation gives,

f ' ( r 1 ) =0.01343153058−0.4 r 1−1.4

Using the equation for Newton-Raphson,

f (x n)
x n+1=x n −
f ' (x n)
 0.01343153058r 1 +r 1−0.4−0.53965172
r 1 , n+1=r 1 , n−
0.01343153058−0.4 r 1−1.4

Looking at the plotted graph for the derived equation shows two points that gives y=0. Therefore, two
initial guesses that could determine the roots were used, r1=7 and r1=15.

Figure 1. Graph of the derived equation

Source: http://www.meta-calculator.com/?panel-101-equations&data-bounds-xMin=-15&data-bounds-
xMax=15&data-bounds-yMin=-1398.0111697040409&data-bounds-yMax=1417.98883029596&data-equations-0=
%22y%3D0.01343153058*x%2B(x%5E-0.4)-0.539617%22&data-rand=undefined&data-hideGrid=false

Using VBA, Table 1.1 and Table 1.2 show the iterations done to arrive at the final answer:

Table 1.1. Iterations using initial guess of , r1=7

Iterations r1,n f(r1,n) F’(r1) r1,n+1 f(r1,n+1) Relative

Error
1 7.00000 0.01353 -0.01281 8.05619 0.00261 0.131102824
2 8.05619 0.00261 -0.00812 8.37806 0.00019 0.038418402
3 8.37806 0.00019 -0.00697 8.40504 0.00000 0.003209321
4 8.40504 0.00000 -0.00688 8.40522 0.00000 2.13982E-05
5 8.40522 0.00000 -0.00688 8.40522 0.00000 9.46501E-10

Table 1.2. Iterations using initial guess of , r1=15

Iterations r1,n f(r1,n) F’(r1) r1,n+1 f(r1,n+1) Relative

Error
 1 15.00000 0.00032 0.00440 14.92622 0.00000 0.004943204
2 14.92622 0.00000 0.00434 14.92569 0.00000 3.55239E-05
3 14.92569 0.00000 0.00434 14.92569 0.00000 1.84927E-09

There are two possible answers however to obtain a constant volume decrease of pressure from state 5
to state 1 the volume compression ratio needs to be greater than the volume expansion ratio.
Therefore, the answer chosen is 14.92568642.

Problems Encountered

The simplification of the equation is one the most difficult in the problem since one wrong calculation
would lead to a wrong solution. It is also required to plot the equation to determine how many roots to
solve and to determine the logical initial guesses to use to minimize the iterations involved. The
graphing of the equation was done using a graphical calculator from the internet. The link can be seen
under the graph and in the reference section of problem 1.

References

Graph of the equation: http://www.meta-calculator.com/?panel-101-equations&data-bounds-xMin=-

15&data-bounds-xMax=15&data-bounds-yMin=-1398.0111697040409&data-bounds-
yMax=1417.98883029596&data-equations-0=%22y%3D0.01343153058*x%2B(x%5E-0.4)-
0.539617%22&data-rand=undefined&data-hideGrid=false

Appendix
Public Function XFunction(ByVal x As Double) As Double
XFunction = 0.01343153058 * x + x ^ (-0.4) - 0.53965172
End Function
Public Function XPrime(ByVal x As Double) As Double
XPrime = 0.01343153058 - 0.4 * x ^ (-1.4)
End Function
Sub Newton()
Dim x1, x2 As Double
Dim tolerance, error As Double
Dim i As Integer
tolerance = Sheet2.Cells(10, 3).Value
x1 = Sheet2.Cells(9, 3).Value
If Sheet1.Cells(10, 3).Value = "" Or Sheet1.Cells(9, 3).Value = "" Then
MsgBox ("Give values for absolute error and initial guess.")
Else
i=0
Do
x2 = x1 - (XFunction(x1) / XPrime(x1))
i=i+1
Sheet2.Cells(7 + i - 1, 9).Value = i
Sheet2.Cells(7 + i - 1, 10).Value = x1
Sheet2.Cells(7 + i - 1, 11).Value = XFunction(x1)
Sheet2.Cells(7 + i - 1, 12).Value = XPrime(x1)
Sheet2.Cells(7 + i - 1, 13).Value = x2
 Sheet2.Cells(7 + i - 1, 14).Value = XFunction(x2)
error = Math.Abs((x2 - x1) / x2)
If error > tolerance Then x1 = x2
Sheet2.Cells(7 + i - 1, 15).Value = error
Loop Until error < tolerance
Sheet2.Cells(16, 2).Value = x2
Sheet2.Cells(17, 2).Value = error
End If
End Sub

Determination of the Displacement Components of Trusses

using LU Decomposition and SOR Method

Problem Statement

Determine the displacement components (x j) given the equilibrium equations shown below:

[ A ] ⃗x =⃗b
Where

0 1 2 3 4 5 6 7 8 9
0 8 0 -4 0 0 0 -2 3.464 0 0
1 0 12 0 0 0 0 3.464 -6 0 0
2 -4 0 12 0 -4 0 -2 -3.464 -2 3.464
3 0 0 0 12 0 0 -3.464 -6 3.464 -6
A  4 0 0 -4 0 8 0 0 0 -2 -3.464
5 0 0 0 0 0 12 0 0 -3.464 -6
6 -2 3.464 -2 -3.464 0 0 12 0 -4 0
7 3.464 -6 -3.464 -6 0 0 0 12 0 0
8 0 0 -2 3.464 -2 -3.464 -4 0 12 0
9 0 0 3.464 -3.464 -3.464 -6 0 0 0 12

 0 
 
 1000
 0 
 1000
 
4 l  0 
b 
E  1000
 0 
 
 1000
 0 
 
 1000
*Note: The following values are already substituted to the matrix and b for simplification

(a) Use LU decomposition to find the displacement components (x j). Provide the element matrices
for L and U upon decomposition.
(b) Use SOR method and find the optimum value of ω . Set the TOL=0.00001 and use relative
approximation and L-infinity norm as stopping conditions.

Results and Discussion

The given matrix, A, was decomposed into two matrices, L and U using Doolittle’s Method. Calculations
and programming were all done using MATHCAD with Trahan (2006) as reference to get the two
matrices. The resulting L and U matrices are shown below.

0 1 2 3 4 5 6 7 8 9
0 1 0 0 0 0 0 0 0 0 0
1 0 1 0 0 0 0 0 0 0 0
2 -0.5 0 1 0 0 0 0 0 0 0
3 0 0 0 1 0 0 0 0 0 0
lower  4 0 0 -0.4 0 1 0 0 0 0 0
5 0 0 0 0 0 1 0 0 0 0
6 -0.25 0.289 -0.3 -0.289 -0.187 0 1 0 0 0
7 0.433 -0.5 -0.173 -0.5 -0.108 0 0.026 1 0 0
8 0 0 -0.2 0.289 -0.437 -0.289 -0.493 0.289 1 0
9 0 0 0.346 -0.289 -0.325 -0.5 -0.042 -0.327 -0.122 1

0 1 2 3 4 5 6 7 8 9
0 8 0 -4 0 0 0 -2 3.464 0 0
1 0 12 0 0 0 0 3.464 -6 0 0
2 0 0 10 0 -4 0 -3 -1.732 -2 3.464
3 0 0 0 12 0 0 -3.464 -6 3.464 -6
upper  4 0 0 0 0 6.4 0 -1.2 -0.693 -2.8 -2.078
5 0 0 0 0 0 12 0 0 -3.464 -6
6 0 0 0 0 0 0 8.375 0.217 -4.125 -1.083
7 0 0 0 0 0 0 0 4.119 1.189 -2.597
8 0 0 0 0 0 0 0 0 6 0
9 0 0 0 0 0 0 0 0 0 4.498
To get the values of xj, forward substitution will be done using [L][Z]=[b] given the equation below:

b1
Z1 =
lower 1,1
i−1
b i−∑ lower i , j Z j
j=0
Zi = , where j≠ i∧i=1. . n
lower i , i
0
0 0
1 -4·10 -3
2 0
3 -4·10 -3
Z 4 0
5 -4·10 -3
6 0
7 -8·10 -3
8 2.309·10 -3
9 -9.491·10 -3
This gave the values for Z:

Then to get X, the values of Z were used during the back-substitution process: [U][X]=[Z]. The following
equations were then used to get X. Calculations started at the nth equation since this is the only
equation that can be solved initially. The subsequent calculations uses the second equation below.

Zn
X n=
upper n ,n
1
Z i− ∑ upper i , j X j
j =i +1
Xi= , where j ≠i∧i=( n−1 ) .. 1
lower i ,i
This gives a set of x shown below:

 0.001401266679
 
 0.002026278384 Therefore, the values of displacement components using LU
 0.0001297334855 decomposition method are:
 0.003189946264
  x 1=0.001401266679∈¿ x 2=−0.002026278384∈¿
 0.0008823327366

X float 10 
 0.001277261757
 0.000004300276937
 x 3=−0.0001297334855∈¿ x 4 =−0.003189946264∈¿
 
 0.003383407336
 0.0003849001795 x 5=−0.0008823327366∈¿ x 6=−0.001277261757∈¿
 
 0.002110079069
x 7=−0.000004300276937∈¿ x 8=−0.003383407336∈¿
x 9=0.0003849001795∈¿ x 10=−0.002110079069∈¿

1
 
1
1
1
 
1
Xguess   
1
1
 
1
1
 
(b)Using the SOR method, initial guesses were given since the method is iterative. 1

The algorithm for SOR method is:

(a) Determine initial guesses.

(b) Get the Gauss-seidel X using the following equation. The superscript denotes that the X are
taken using the relaxation equation.
n
bi−∑ A i , j Xguess jR
j=1
Xgauss= where j≠ i∧i=1. . n
A i, i
(c) Get the relaxation, X using the following equation
Xrelax i=w Xgaussi + ( 1−w )∗Xguess i
(d) Get the relative approximation error for each X i
Xrelax i− Xguessi
error i=
Xrelax i
(e) Assign the value of Xrelax to Xguess and repeat steps (b) to (d) to get all the values of X.
(f) Get the L∞ which is equal to the maximum of error. This will be the criterion to determine if
iterations will be stopped. Compare L∞ if it is less than TOL. If not, repeat steps (b) to (e). If it is
below the tolerance, stop the iterations because the values of x have been determined.

This algorithm was done to different values of ω to determine the least number of iterations. For SOR,
the range of ω is between 1 and 2. Shown below is the narrowed down range to get the optimum ω
(range from 1.26 to 1.45).

ω No. of Iterations ω No. of iterations

1.26 32 1.36 26
1.27 31 1.37 29
1.28 29 1.38 28
1.29 30 1.39 27
 1.3 30 1.4 30
1.31 29 1.41 30
1.32 28 1.42 29
1.33 25 1.43 28
1.34 28 1.44 32
1.35 27 1.45 31

Based on the table above, ω of 1.33 will give the least amount of iterations which is 25; therefore, it is
considered the optimum value.

Shown below is the first three and last three iterations using ω=1.33.

Iteration X1 X2 X3 X4 X5 X6 X7 X8 X9 X10 L∞
no.
1 0.092 -0.049 0.597 1 0.976 0.718 0.669 0.496 0.207 0.676 21.251
2 0.304 0.089 0.495 0.626 0.466 0.292 0.255 0.385 0.129 0.2 2.389
3 0.092 0.128 0.24 0.23 0.164 0.086 0.086 0.168 0.03 0.05 3.369
…
23 1.401e- -2.026e- -1.297e- -3.19e-3 -8.823e- -1.277e- 4.3e-6 -3.383e- 3.849e- -2.11e-3 2.039e-
3 3 4 4 3 3 4 4
24 1.401e- -2.026e- -1.297e- -3.19e-3 -8.823e- -1.277e- 4.3e-6 -3.383e- 3.849e- -2.11e-3 4.747e-
3 3 4 4 3 3 4 5
25 1.401e- -2.026e- -1.297e- -3.19e-3 -8.823e- -1.277e- 4.3e-6 -3.383e- 3.849e- -2.11e-3 5.438e-
3 3 4 4 3 3 4 6

Therefore, the values of x are the following:

 0.001401266052 Therefore, the values of displacement components using SOR

 
 0.00202627697
method are:
 0.0001297327411
 0.003189944136 x 1=0.001401266052∈¿ x 2=−0.00202627697∈¿
 
 0.0008823320118

Final_Ans float 10  x 3=−0.0001297327411∈¿ x 4 =−0.003189944136∈¿
 0.001277261263
 0.000004300375185

  x 5=−0.0008823320118 ∈¿ x 6=−0.001277261263∈¿
 0.003383405712
 0.0003849000492 x 7=−0.00000430037518∈¿ x 8=−0.003383405712∈¿
 
 0.002110078426
x 9=0.0003849000492∈¿ x 10=−0.002110078426∈¿

These match with the answers obtained from LU decomposition method and the Isolve function in
Mathcad.
Problems Encountered

The problems encountered for item number 2 is the code of SOR. For the LU decomposition, there is
already a built-in code for the decomposition and solution for linear equations which made it easier to
check if the program done is correct. Additionally, there is an LU decomposition code available in the
internet in Mathcad form. As for the SOR, the only reference that is available in terms of coding in
MATHCAD is Gauss-Seidel. Since it is somewhat similar to that of SOR, this became the basis of the SOR
code.

References

LU decomposition:
http://mathforcollege.com/nm/simulations/mcd/04sle/mcd_sle_sim_ludecomp.pdf
Gauss-Seidel
http://mathforcollege.com/nm/simulations/mcd/04sle/mcd_sle_sim_gaussseidel.pdf

Appendix
XansIter  row  0
for w  1.251.26 1.45
Linf  1
iter  0
 01 
 
 01 
 01 
 01 
 
01
Xguess   
 01 
 01 
 
1
1
 
1
while Linf  TOL
for i  0  n
sum  0
for j  0  n
sum  sum  A  Xguess if j  i
i j j
b  sum
i
XgaussN 
i A
i i
Xrelax  w XgaussN  ( 1  w)  Xguess
i i i
Xrelax  Xguess
i i
error 
i Xrelax
i
Xguess  Xrelax
i i
Linf  max( error )
Answer  iter
row 1
iter  iter  1
Answer w
row 0
row  row  1
Answer
w  1.33
Xnew1  Linf  1
iter  0
while Linf  TOL
iteration  iter
iter 0
for i  0  n
sum  0
for j  0  n
sum  sum  A  Xguess if j  i
i j j
b  sum
i
Final_Ans  LinfXgaussN
1 
i A
i i
iter  0
iteration  XgaussN
while Linf iter i 1
TOL i
 iter  ( 1  w)  Xguess
Xrelax  w XgaussN
iteration
i iter 0 i i
for i  0  n
iteration  Xrelax
iter i 11 i
sum  0
Xrelax  Xguess
i i
errorforj  0  n
i Xrelax
sum  sum  A i  Xguess if j  i
i j j
Xguess  Xrelax
i bi  sum
i
XgaussN
Linf  max( error 
i ) A
i i
iteration  Linf
iter 21
Xrelax  w XgaussN  ( 1  w)  Xguess
i i i
iter  iter  1
Xrelax  Xguess
iteration i i
error 
i Xrelax
i
Xguess  Xrelax
i i
Linf  max( error )
iteration  Linf
iter 21
iter  iter  1
Xrelax