Академический Документы
Профессиональный Документы
Культура Документы
PROCESS DESIGN
SIMULATIONS
by
ANOUSHTAKIN ffARMAN
Bachelor of Science in
Chemical Engineering
Oklahoma State University
Stillwater, Oklahoma
May, 1985
Thesis Approved:
1263914
ii
ABSTRACT
iii
ACKNOWLEDGEMENTS
lV
TABLE OF CONTENTS
Chapter Page
I . INTRODUCTION. . . • • . • . . . . . . . . . . . • . . • . . . • . . . . . . . • . . . 1
Direct Substitution..................... 5
Wegstein's Method....................... 6
Dominant Eigenvalue Method.............. 8
General Dominant Eigenvalue Method ...... 11
Newton and. Quasi-Newton's Methods ..•.... 14
II I. DISCUSSION AND RESULTS.......................... 17
APPEND I XES . . . . . . . . . . . . . • . . • . • . . • . . . . . . . . . • . . • • . • • . . . . . . 52
v
LIST OF TABLES
Table Page
I. Results of the GDEM Algorithm in Model 2 ••••••• 26
II. Results of the GDEM Algorithm in Model 3 ••••••• 28
I I I. Results of the GDEM Algorithm in Model 4 ••••••• 30
IV. Results of the GDE~ Algorithm in Model 5 ••••••• 33
v. Results of the GDEM Algorithm in Model 6 ••••••• 35
VI. Results of the GDEM Algorithm in Model 7 ••••••• 37
VII. Results of the GDEM Algorithm in Model 8 (A) •••• 40
VI I I. Feed Streams to the Example Models ••••••••••••. 65
Vl
LIST OF FIGURES
Figure Page
1. Sequential Modular Architecture................... 2
2. Graphical Illustration of the Wegstein's Method... 7
3. The Pipe Network •••.•.•..••...•••••.•••..•.•.••..• 18
4. Comparison of the Three Methods
in the. Pipe Network............................... 21
5. A Typical Chemical Process Model ..••••.••..•••..•. 22
6. Comparison of the Three Methods in Model 2 •..•..•. 23
7. A Heat Exchange Dominated System ..•••...••••.•.... 27
8. A Heat and Mass Dominated System ..•....•.•.•.•..•. 29
9. A Stream Divider Dominated System •......•....•..... 32
10. A Flash Dominated System.......................... 34
11. System Containing a Distillation Column ••.......•. 36
12. System With Four Flash Drums in Series ....•.....•. 38
13. Oscillatory Behavior of Flash Drums in Series •.... 41
14. Distillation Column Replacing Flashes in Series ... 42
15. Stability of Stream# 4 in Model 8 (B) ....•..•.... 43
16. History of the Sum of the Mass of Stream #3 ....... 54
17. Oscillatory History of the Sum of the Mass
0 f s t ream # 4 ... 0 • • • • • • • • • • • • • •••• ., • • • • e • • • • • • • • • • • 55
vii
22. Orbach and Crowe's DEM Applied at Different
Damping Factors................................... 62
23. Flowchart for the GDEM Algorithm ..••.••.•..••..... 64
Vlll
LIST OF SYMBOLS
A - Linearized approximation to F
E - Accuracy desired
F(X) - Feed stream
FR - Calculated feed of a broken recycle stream
H - The negative inverse of the Jacobian matrix
I - The identity matrix
PR - Calculated product of a broken recycle stream
T - The transpose matrix
W - Weighting matrix
X - Feed stream to a process unit
e - An arbitrary point near n
n - Number of iterations performed
o - The initial guess
s - The absolute solution
·1'-
v - Number of coefficients to estimate ,f"- j
- Damping factor in DEM
- Eigenvalue of A
- Forward difference operator
,,
_,P-j - jth eigencoefficient of A
- Estimated value
- Temperature, F
- Pressure, psia
- Stream number
ix
CHAPTER I
INTRODUCTION
1
2
IPl l F2
F3 • Rl
[
1' Pl J,F2 _
Fl Xl AI1---X2--*~ B I
~-I-X3-~>___c--'J P2 >
FRl
(calculated)
FRl
(estlmated) TPl t F2 t
FF Xl ~ A ~---1-X2=---'~>I B I~--=X3,.----:,.t)~ c I PZ )
until the feed and the product agree within a set tolerance
(Figures l(B and C)).
F(X) = X (1-1)
F(X) = Xs (1-2)
4
Xn ~ Xs (1-3)
and
LITERATURE SURVEY
NUMERICAL METHODS
5
6
Weastein's Method
where
g = s I (s-1) (2-4)
and
s = (F(Xn) - FX(n-1)) I (Xn- X(n-1)) (2-5)
g = 0 successive substitution
g < 0 can speed convergence but
also introduces instability
q > 0 slow, stable convergence.
7
F(X)
Xn-1 Xn Xs +
X
X(n+1) = A Xn + b (2-6)
where
A = ( 6 F I ~ X) @X = Xe (2-7)
and
b = F(Xe) - A Xe (2-8)
(2-9)
n
Xn - Xs = A (Xo - Xs) (2-10)
10
and 1n particular
m n
Xn - Xs = L Cj Zj Aj (2-11)
j=l
Xs = (I - A) - 1 b (2-12)
and
T T
Cj = Wj (Xo - Xs) I (Wj z j) (2-13)
n
Xn - Xs = C1 Zl A1 (2-14)
I A1 I = II A Xn II I II A X(n-:) II (2-16)
X( n + 1 ) = X( n -1 ) + 0{ ( Xn - X( n -1 ) ) I ( 1 - A1 ) ( 2 -1 7 )
11
Xn = A X(n-1) (2-19)
12
~0 =1
From the Cayley-Hamilton theorem (10), A satisfies equation
2-20 so that
m m-j
2 ~ j A A X(n-m) = 0 (2-22)
j=O
Repeated use of equations 2-19 and 2-22 gives
m
~ /j Ax(n-j) = 0 (2-23)
j=O
If the eigenvalues are labeled in ascending order of magni-
tude and if we assume that only v of them were large enough
to dominate the iteration, it then follows that
m
L/j AX(n-j) = 0 (2-24)
j=v+l
13
where
i = n , (n+l) , ...
....
and~ J is an approximation to the real value of ~ j.
Also~ j is estimated by taking the derivative of the square
norm with respect to~ k and setting them equal to zero.
Thus
v
} ;. . bj k = 0 (2-26)
foo ./ J
where
k = 1 ,2 , ... ,v
and
bij = <A X(n-j) , A X(n-k) > (2-27)
Xs - X(n+l) = "f. A
i=n+l
Xi LA
i=n+l
X( i-j) .(2-29)
14
Xs = Xn + A Xn I ( 1 + /" 1 ) (2-31)
where
bOl I bll (2-32)
A
F(X) = 0 (2-35)
is
X(n+l) = Xn- Jn-l F(Xn) (2-36)
15
where
Hn =- Jn- 1 (2-38)
Pn = Hn Qn (2-39)
Qn = X(n+l) - Xn (2-40)
and
Yn = F(X(n+1)) - F(Xn) (2-41)
Pi - Pj = Fm ~ Urn 2 L 12 D (3-1)
Q = ( 1T D2 I 4 ) Urn (3-2)
Pi - Pj = 8 Fm .f Q2 L I TT 2 D5 (3-3)
=c L Q2 I Ds (3-4)
17
18
MODEL 1
~--------------~~2 _________ 3
4 5
Let
then
I Pi - Pj I= Cij Qij 2 (3-6)
where Qij is the flow rate between the nodes i and J. Equa-
tion 3-6 can be rearranged and since the sum of the flow
rates is zero at any node,
where
Aij = (Cij I Pi - Pj )1 / 2 (3-9)
Fi = Pj - A l..J •o·l I
1 L Ai j (3-10)
E = F2 2 + F4 2 + F5 2 (3-11)
20
DIRECT SUBSTITUTION +
BOUNDED WEGSTEIN -5 < 0 < 0 J]
OEM . DAMP I 0(. I m • 9 A ).. a 5 :110 A
GDEM . 4/-'-• 5 :110 *
5
-5
L
0
G
-to
E
-15
-20 T --.----~r--.--,--y--r-.~
0 tO 20 30 40 50
ITERATIONS
,_.
N
22
MODEL 2
DIRECT SUBSTITUTION +
BOUNDED WEGSTEIN -5 < 0 < 0 -It
OEM • DAMP CC( I • • 9 • tl A • 5 X 0
GDEM • Jj ~ • 5 X 6
2 "1
0.0
L
~
0 -2.5
G
-5.0
-7.5i 1 u a I o u a I u u 1 a u u a u u u a u 1 u 1 a a a a u ; ; 1 u
~
; ; u a u u s u I ; a u a ; o 0 u 1 I
0 5 10 15 20 25
ITERATIONS
N
w
24
TABLE I
RESULTS OF THE GDEM ALGORITHM IN MODEL 2
A.
2 5 13
2 1 11
3 5 12
3 1 11
4 5 13
6 5 21
7 5 14
10 5 13
10 1 11
Tolerance = lE-4
Direct Substitution = 21 Iterations
Method = GDEM
27
TABLE II
RESULTS OF THE GDEM ALGORITHM IN MODEL 3
STREAM ! ITERATIONS
2 5 11
8 5 11
Tolerance = 1E-5
Direct Substitution = 13 Iterations
Method = GDEM
29
MODEL 4
.TABLE III
RESULTS OF THE GDEM ALGORITHM IN MODEL 4
STREAM .! ITERATIONS
2 5 14
4 5 14
6 5 14
10 5 14
Tolerance = 1E-5
Direct Substitution = 14 Iterations
Method = GDEM
31
MODEL 5
TABLE IV
RESULTS OF THE GDEM ALGORITHM IN MODEL 5
STREAM ! ITERATIONS
3 5 10
3 1 10
6 5 12
6 1 12
Tolerance = lE-5
Direct Substitution = 90 Iterations
Method = GDEM
34
MODEL 6
TABLE V
RESULTS OF THE GDEM ALGORITHM IN MODEL 6
II>
STREAM ! A_e(!) ·ITERATIONS
2 5 18
3 5 21
8 1 14
8 5 13
8 10 11
Tolerance = 1E-5
Direct Substitution = 22 Iterations
Method = GDEM
36
MODEL 7
*APPENDIX F
37
TABLE VI
RESULTS OF THE GDEM ALGORITHM IN MODEL 7
STREAl~ ! ITERATIONS
10 15
10 15
Tolerance = lE-5
Direct Substitution = 15 Iterations
Method = GDEM
38
MODEL 8(A)
TABLE VII
RESULTS OF THE GDEM ALGORITHM IN MODEL 8(A)
STREAM! ITERATIONS
10 1 33
16 1 33
Tolerance = lE-4
Direct Substitution = 33 Iterations
Method = GDEM
J/ODBL B (A)
STREAM f -4
DIRECT SUBSTITUTION
2.5
0.0
-2.5
L
0
v\
G
• ~ool \
-7.5
-SO.Oi
I I I I I I
V
I • • I 1 • 1 1 • • • • 1 1 1 ; a o a 8 1 11 1 1 1 ; ; e ; a 1 ; a ; 1 ; ; e ; e 6 1 1 1 I ; 1 1 1 ; 1 1 1 1 I
0 5 so 15 20 25 30
ITERATIONS
.....
....
42
MODEL 8(B)
~--T"'------1*
5
*APPENDIX G
** -112.43 KBTU/Hr
JIODEL B (B)
STREAM f 4
DIRECT SUBSTITUTION
l \
0
G
E
-5
-101~~~O~OTO~i~O~O~O~iTO,j~O~i~OTO~O~rT,j~O~O~O~OTO~O~O~O~OTjTO~O~O~i~OTO~O~O~OrTjTO~O~O~OrTOTO~O~O~O~jTOTO~O~O~i~OTO~O~O~jrTOTO~O~O~OTOTOTO~O~jrrCTITS~i~OrTITiTI~S~jrr>TOTO~O~O~ITOTO~O~jrrOTITO~O~I~OTOTITO~j
0 1 2 3 ... 5 6 7 8 9 10
ITERATIONS
The best results were for GDEM ranging from zero improve-
ment for a very oscillatory system to over 90 % reduction in
the number of iterations in the case of an non-oscillatory
system. Model 2 which represented a typical chemical pro-
cess had a reduction of almost 50 % in the number of itera-
tions.
46
47
50
51
52
APPENDIX A
53
MODEL 1
STREAM II 3
DIRECT SUBSTITUTION
t .0
X
N
I
X
s
O.B
0.7 ~~,-~----~~~~-r-T~--~~~~~~r-r-T-T-~~~~~~~-r-T~~-,~~~~~r-T
0 5 tO 15 20
ITERATIONS
VI
+:--
JIODBL B
STREAM • 4
DIRECT SUeSTITUTION
1.01
1.00 ~
0.99
X
N
I
X 0.98
s
0.97
0.96~~~~~~---r~-?-.~-,~~~~r-----T-~~,-~~~-r-r-r-r-,~-,~~~~r-r-~T-T
0 5 10 15 20
ITERATIONS
lJl
lJl
APPENDIX B
Figures 18, 19, and 20 show that for model 2 the best
range for q is between -5 and 0 • They also show that
including the temperature as a variable in the matrix and
applying Wegstein every other iteration will help improve
the convergence of the method.
56
JIODEL 2
STREAM f 6
DIRECT SUBSTITUTION +
-5 <
Q < 0 ~
-5 < Q < 5 0
-10 < Q < 0 ~
5.0
2.5
0.0
L
0
G
-2.5
E
-5.0
0 2 3 4 5 6 7 B 9 10
ITERATIONS
V1
'-1
JIODEL 2
STREAM f 3
DIRECT SUBSTITUTION +
-5 < 0 < 0 (TEMPERATURE INCLUDED) *
-5 < 0 < 0 (TEMPERATURE NOT INCLUDEDI D
5.0
2.5
0.0
L
0
G
-2.5
E
-5.0
-7.5~~,-~~~~-r-r-r~-T-T~~~~~~r-r-r-~T-~,-~~~-r-r-r~~~~-,~~r-r-r-~~
0 5 10 15 20
ITERATIONS
Vl
w
JIODEL 2
STREAM I 6
DIRECT SUBSTITUTION ~
.·~
2.5
~
_o.o
L
0 -2.5
G
E -5.0
-7.5
-10.0
T
0 5 10 15 20
ITERATIONS
VI
'-0
APPENDIX C
60
JIODEL 2
STREAM I 3
DIRECT SUBSTITUTION ~
DAMP I tit I • . 7
OEM . b.A • 20 I A
OEM t). )I a 5 I ll
OEM, A).,- 2 I fro
3-
0
l
0 -t
G
-2
E
-3
-4
-5 ..,..
0 5 so 15 20
ITERATIONS
~
0\
.......
JIODBL 2
STREAM f 3
DIRECT SUBSTITUTION ~
OEM, ,A.). .. 5 .
DAMP 1~1 • .9 0
DAMP (Gr.) • .7 !\-
DAMP IG(. I • .5 .::\
0
L
0 -1
G
-2
E
-3
l
-l
-5
0 5 10 15 20
ITERATIONS
a•
N
APPENDIX D
63
64
Acceleration Flowchart
Iteration • 0
Yes
Warning /
I
Yes
No
xs • Xn +Axn I l •J-'2
.
No
TABLE VIII
FEED STREAMS TO THE EXAMPLE MODELS
NC8H18 = 1.4776
temperature = 100 (F)
pressure = 485 (psia)
65
APPENDIX F
!:TXT # :
COUNT ?~TES Fn~M BOTTOM UP
!IJUMBER OF PLATES IN COLUMN
NUMBER OF FEED PI..ATES ..."
~ruMBER OF PRODUCTS 2
NUMBER OF SIDE CCOLE.!;:S/HEATERS t)
cmm8JSER/DISTILLATE SPECiF:CATIONS-
D:ST!L!..ATE RATE0.64000 iD/Fl
REBOILER:BOTTOMS SPES!FICATIONS-
REBOILER DUTY 6. 00 f~:BTULB
66
APPENDIX G
2.74
r), 019
'\
Ill l ')24
I
i:\. (t::3
i}, !38
,147
(1
• ...
• r:
1 ·~
..
~~'='
':C8H18 1.4766 0.1)01)0 1.4766 2.68 -+.-+: '). :•49 1;, 1-~: -. -
.. ,.,
,.4.. 1).0466 1).04~4 1),001:
.~,,
67
VITA
~
ANOUSHTAKIN ARMAN
Candidate for the Degree of
Master of Science