Biegler/Grossmann/Westerberg:

Systematic Methods of Chemical Process Design

Design and Scheduling of
Batch Plants
Cheng-Liang Chen
PSE
LABORATORY
Department of Chemical Engineering
National TAIWAN University
Chen CL 1
Single Product Batch Plants
EX: A Processing Recipe
1. Mix raw materials A and B. Heat to 80
o
C and react during 4
hours to form product C.
2. Mix with solvent D for 1 hour at ambient conditions.
3. Centrifuge to separate solid product C for 2 hour.
4. Dry in a tray for 1 hour at 60
o
C.
Chen CL 2
Single Product Batch Plants
Gantt Charts (with/without Transfer Times)
Chen CL 3
Single Product Batch Plants
Cycle Times
CT =

j=1

j
for non-overlapping operation
max
j=1,M

j
for overlapping operation
Chen CL 4
Multiple Product Batch Plants
Flowshop/Jobshop Plants
Flowshop (Multi-product) Plants:
all products require all stages following the same sequence of
operations
Jobshop (Multi-purpose) Plants:
not all products require all stages and/or follow the same sequence
Chen CL 5
Single/Mixed-Product Campaigns
Chen CL 6
Eect of Cleanup Time On Cycle Time
Chen CL 7
Transfer Policies
Zero-wait (ZW) transfer
No-intermediate storage (NIS)
Unlimited storage (UIS)
CT
UIS
= max
j=1,M

i=1
n
i

ij

Chen CL 8
Cycle Times for Various Transfer Policies
ZW
NIS
UIS
Chen CL 9
Parallel Units and Intermediate Storage
We assume zero-wait transfer and that the size of the
batch in each stage is the same (1000 kg).
Chen CL 10
Parallel Units and Intermediate Storage
CT = max
j=1,M

j
NP
j
If a large number of batches are to be produced, then to
produce the same amount we can reduce the batch size
to 500 kg since the cycle time has been halved.
Chen CL 11
Parallel Units and Intermediate Storage
The other alternative is to introduce storage between
stages. This has the eect of decoubling the two stages
si that each stage can operate with dierent cycle times
and batch sizes. Stage 1 has a cycle time of 12 hours and
handles batches of 1000 kg; Stage 2 has a cycle time of
3 hours and handles batches of 250 kg. The intermediate
storage must hold up to three batches (750 kg).
Chen CL 12
Sizing of Vessels in Batch Plants
An Example
Problem:
Design a two-stage plant to produce 500, 000 b/yr of product C
(operating 6, 000 hours per year) with the following recipe
Mix 1 b A, 1 b B, and react for 4 hours to form C. The yield is
40% in weight and the density of the mixture is
m
= 60b/ft
3
.
Add 1b solvent and separate by centrifuge during 1 hour to
recover 95% of product C. Density of the mixture is
m
=
65b/ft
3
.
Chen CL 13
Sizing of Vessels in Batch Plants
An Example
Size Factor S
j
for stage j:
S
j
= volume vessel j required to produce 1b of nal product
S
1
=
1 ft
3
60 b mix

(1+1) b mix
(20.40.95) b prod
= 0.0438
ft
3
b prod
S
2
=
1 ft
3
65 b mix

(1+1+1) b mix
(20.40.95) b prod
= 0.0604
ft
3
b prod
Chen CL 14
Sizing of Vessels in Batch Plants
An Example
Cycle Time, # of Batches, Batch Size, Vessel Volumes
CT = max{4, 1} = 4 hours
# Batches =
6000 hours
4 hrs/batch
= 1500 batches
Batch Size =
500,000 b
1500 batches
= 333 b/batch
Vessel V
1
= S
1
ft
3
b prod
B
b
batch
= 0.0438
ft
3
b prod
333
b
batch
= 14.6 ft
3
Vessel V
2
= S
2
ft
3
b prod
B
b
batch
= 0.0604
ft
3
b prod
333
b
batch
= 20.1 ft
3
Total Vol. = V
1
+V
2
= 14.6 + 20.1 = 34.7 ft
3
Chen CL 15
Sizing of Vessels in Batch Plants
An Example
Bottleneck: stage 1 two units operating in parallel
CT = max{4/2, 1} = 2 hours
# Batches =
6000 hours
2 hrs/batch
= 3000 batches
Batch Size =
500,000 b
3000 batches
= 166 b/batch
Vessel V
1
= S
1
ft
3
b prod
B
b
batch
= 0.0438
ft
3
b prod
166
b
batch
= 7.3 ft
3
Vessel V
2
= S
2
ft
3
b prod
B
b
batch
= 0.0604
ft
3
b prod
166
b
batch
= 10. ft
3
Total Vol. = 2V
1
+V
2
= 2 7.3 + 10 = 24.6 ft
3
Chen CL 16
Sizing of Vessels in Two-Product Batch Plants
Problem:
Design a two-stage plant to produce 500, 000 b/yr of product A
and 300, 000 b/yr of product B (operating 6, 000 hours per year).
Processing Time (hr.) Size Factor (ft
3
/b prod)
Stage 1 Stage 2 Stage 1 Stage 2
A 8 3 0.08 0.05
B 6 3 0.09 0.04
Cleanup Time = 4 hours A to B, B to A
Length of production cycle = 1000 hours (repeat 6 times)
1000 4 4 = 992 hr.s for production operation (eective time)
Chen CL 17
Sizing of Vessels in Two-Product Batch Plants
Single product campaign; Production of A and B in each campaign:
for A:
500,000 b
6
= 83, 333 b
for B:
300,000 b
6
= 50, 000 b
Batch size B
i
of product i; # of batches, total production time t
i
(equating the batch sizes and constraining production times to 992 hours)
B
i
=
production i
# batches i
=
production i
t
i
/CT
i
83,333
t
A
/8
=
50,000
t
B
/6
t
A
+t
B
= 992
t
A
= 684, t
B
= 308 hours
Chen CL 18
Sizing of Vessels in Two-Product Batch Plants
Vessel volumes (select largest volume in each stage)
V
ij
= S
ij
B
i
V
j
= max
i=1,N
{V
ij
}
Volumes V
ij
= S
ij
B
i
Stage 1 Stage 2
A 77.9 48.7
B 87.7 39.0
V
j
87.7 48.7
Chen CL 19
Sizing of Vessels in Two-Product Batch Plants
Inventory: A trade-o production cycle inventories
Demand Rate: d
A
=
83,333 b
1000 hr
= 83.3 b/hr
d
B
=
50,000 b
1000 hr
= 50.0 b/hr
Produc Rate: p
A
=
83,333 b
684 hr
= 121.8 b/hr
p
B
=
50,000 b
308 hr
= 162.3 b/hr
Inv Prole: p
A
d
A
= 121.8 83.3 = 38.5 b/hr (accum rate, 0 684 hrs)
d
A
= 83.3 b/hr (depletion rate, 684 1000 hrs)
d
B
= 50.0 b/hr (depletion rate, 0 688 hrs)
p
B
d
B
= 162.3 50 = 112.3 b/hr (accum rate, 688 996 hrs)
Max Inventory I
Amax
= 38.5 (b/hr) 684 (hr) = 26334 (b)
I
Bmax
= 112.3 (b/hr) 308 (hr) = 34598 (b)
Ave Inventory I
A
=
1000(26334)
2(1000)
= 13167 (b)
I
B
=
1000(34600)
2(1000)
= 17300 (b)
Invent Cost: C
inv
= (13167 + 17300) (b) $1.25/ (b-yr) = $38048/yr
Chen CL 20
Sizing of Vessels in Two-Product Batch Plants
Chen CL 21
Synthesis of Flowshop Plants
Three Major Decision Levels
Structural Level
Assignments of tasks to equipment
Merge adjacent tasks whose sum of processing times does not exceed cycle
time (construction materials must meet all required functions)
Chen CL 22
Synthesis of Flowshop Plants
Three Major Decision Levels
Chen CL 23
Synthesis of Flowshop Plants
Three Major Decision Levels
Structural Level (cont.)
Number of parallel units or intermediate storage
Use parallel units whenever there is a stage with a much larger processing
times (a smaller # of bigger equip. a larger # of smaller pieces)
Chen CL 24
Synthesis of Flowshop Plants
Three Major Decision Levels
Sizing Level
Equipment sizing
To select the same batch size for all products
Scheduling Level
Nature of production campaigns, transfer policies
single products if cleanup or transition times are large;
ZW, NIS, or UIS ?
Length of production cycles
(requires detailed evaluation and optimization)
Sequencing of products
Chen CL 25
Multiproduct Batch Plant Design
Problem
Given a multiproduct batch plant that consists of three processing
stages: mixing, reaction, and centrifuge separation. Two products,
A and B, are to be manufactured in such a plant using production
campaigns of single products. The data for processing times, size
factors for the units, demands, and cost data are given in the next
page. Assuming continuous sizes, and that the plant is operated
with single-product campaigns, determine the sizes of the units
required at each processing stage, as well as the number of units
that ought to be operating in parallel to minimize the investment
cost. Furthermore, resolve the problem for the following cases:
1. The demands of A and B are increased by 20%.
2. For the above demands the time available for production is increased from 6000
hours to 7500 hours.
Chen CL 26
Multiproduct Batch Plant Design
Data
Demand A = 200, 000 kg
Demand B = 150, 000 kg
Horizon time = 6, 000 hr
Processing Time (hrs) Size Factors (L/kg)
Mixer Reactor Centrifuge Mixer Reactor Centrifuge
A 8 20 4 A 2 3 4
B 10 12 3 B 4 6 3
Mixer cost = 250V
0.6
Reactor cost = 500V
0.6
Centrifuge cost = 340V
0.6
Min size = 250 L
Max size = 2500 L
Chen CL 27
Multiproduct Batch Plant Design
Notation
N = number of products to be produced
M = number of stages in the batch plant
B
i
= size of product i at the end of the M stages
S
ij
= size factor of product i in stage j (L/kg)
t
ij
= processing time of product i in stage j (hr)
H = horizon time (hr)
Q
i
= demand of product i (kg)

i
,
j
= cost coecient and cost exponent for unit j
V
L
j
, V
U
j
= lower and upper bounds of volumes
T
Li
= cycle time for product i
N
j
= number of parallel units in stage j

i
= production time
n
i
= number of batches
Chen CL 28
Multiproduct Batch Plant Design
Formulation
Volume V
j
can process all products
V
j
S
ij
B
i
i = 1, , N; j = 1, , M
Cycle time
T
Li
= max
j
t
ij
N
j
i = 1, , N; j = 1, , M
T
Li

t
ij
N
j
Chen CL 29
Multiproduct Batch Plant Design
Formulation
Example: N
1
= N
2
= N
3
= 1
T
LA
= max{8, 20, 4} = 20 hr
T
LB
= max{10, 12, 3} = 12 hr
Chen CL 30
Multiproduct Batch Plant Design
Formulation
Example: N
1
= N
3
= 1, N
2
= 2
T
LA
= max{8, 20/2, 4} = 10 hr
T
LB
= max{10, 12/2, 3} = 10 hr
Chen CL 31
Multiproduct Batch Plant Design
Formulation
Production time

i
= n
i
T
Li
=
Q
i
B
i
T
Li
i = 1, , N

i=1

i
=
N

i=1
Q
i
B
i
T
Li
H
Investment cost
C =
M

j=1
N
j

j
V

j
j
Chen CL 32
Multiproduct Batch Plant Design
Summary
min
x
C =
M

j=1
N
j

j
V

j
j
X = {B
i
, V
j
, N
j
, T
Li
; i = 1, , N; j = 1, , M}
=

V
j
S
ij
B
i
T
Li

t
ij
N
j
N

i=1
Q
i
B
i
T
Li
H
V
L
j
V
j
V
U
j
B
i
0, T
Li
0
i = 1, , N; j = 1, , M

Chen CL 33
Multiproduct Batch Plant Design
Given Maximum K Parallel Units
T
L
Li
T
Li
T
U
Li
B
L
i
B
i
B
U
i
T
L
Li
= max

t
ij
K

j = 1, , M
T
U
Li
= max {t
ij
} j = 1, , M
B
L
i
=
Q
i
T
L
Li
H
B
U
i
= min
j

V
U
j
S
ij

Chen CL 34
Multiproduct Batch Plant Design
Variable Transformation Convex Constraints
let v
j
ln V
j
n
j
ln N
j
b
i
ln B
i
t
Li
ln T
Li
C =
M

j=1
N
j

j
V

j
j
C =
M

j=1

j
exp (n
j
+
j
v
j
)
n
j
= ln N
j
N
j
{1, 2, , K}
n
j
=
K

k=1
(ln k)y
jk
K

k=1
y
jk
= 1 j = 1, M
V
j
S
ij
B
i
e
v
j
S
ij
e
b
i
v
j
ln S
ij
+b
i
i, j
So, t
Li
(ln t
ij
) n
j

i
Q
i
e
(t
Li
b
i
)
H
Chen CL 35
Multiproduct Batch Plant Design
MINLP with Convex Constraints
min
x
C =
M

j=1

j
exp (n
j
+
j
v
j
)
x = {b
i
, v
j
, n
j
, t
Li
, y
jk
; i = 1, N; j = 1, M}
=

v
j
ln S
ij
+b
i
, t
Li
ln t
ij
n
j

i
Q
i
exp (t
Li
b
i
) H
n
j
=
K

k=1
(ln k)y
jk
,
K

k=1
y
jk
= 1
ln V
L
j
v
j
ln V
U
j
ln B
L
i
b
i
ln B
U
i
ln T
L
Li
t
Li
ln T
U
Li
y
jk
{0, 1}, n
j
0
i = 1, , N; j = 1, , M; k = 1, , K

Chen CL 36
Multiproduct Batch Plant Design
GAMS Code
$TITLE BATCH PROCESSING PROBLEM
$OFFSYMXREF
$OFFSYMLIST
*
* MINLP FORMULATION OF BATCH PROCESSING PROBLEM
*
* Reference: Kocis, G.R. and I.E. Grossmann, "Global Optimization of
* Nonconvex MINLP Problems in Process Synthesis",
* I&EC Research, 1988, 27, 1407.
SETS I PRODUCTS / A , B /
J STAGES /MIXER, REACTOR, CENTRIFUGE/
K POTENTIAL NUMBER OF PARALLEL UNITS / 1 * 3 /
SCALAR H HORIZON TIME (available time hrs) / 6000. /
VLOW LOWER BOUND FOR SIZE OF BATCH UNIT /250./
VUPP UPPER BOUND FOR SIZE OF BATCH UNIT /2500./
PARAMETERS Q(I) DEMAND OF PRODUCT i / A=200000 , B=150000 /
ALPHA(J) COST COEFFICIENT FOR BATCH UNITS
/ MIXER= 250., REACTOR = 500., CENTRIFUGE = 340. /
BETA(J) COST EXPONENT FOR BATCH UNITS
/ MIXER = 0.6, REACTOR = 0.6, CENTRIFUGE = 0.6 /
Chen CL 37
* The parameters below are defined to obtain the original variables
ACTV(J) ACTUAL VOLUME (L)
ACTB(I) ACTUAL BATCH SIZES (kg)
ACTTL(I) ACTUAL CYCLE TIMES (hrs)
ACTN(J) ACTUAL NUMBER OF UNITS IN PARALLEL
* The coefficients below are defined to represent number of parallel units
COEFF(K) ;
COEFF(1) = LOG(1.) ;
COEFF(2) = LOG(2.) ;
COEFF(3) = LOG(3.) ;
TABLE S(I,J) SIZE FACTOR FOR PRODUCT i IN STAGE j (Kg per L)
MIXER REACTOR CENTRIFUGE
A 2 3 4
B 4 6 3
TABLE T(I,J) PROCESSING TIME OF PRODUCT I IN BATCH J (hrs)
MIXER REACTOR CENTRIFUGE
A 8 20 4
B 10 12 3
VARIABLES Y(K,J) BINARY VARIABLE DENOTING STAGE EXISTENCE
V(J) VOLUME OF STAGE J (L)
B(I) BATCH SIZE OF PRODUCT i (kg)
TL(I) CYCLE TIME OF PRODUCT i (hrs)
Chen CL 38
N(J) NUMBER OF UNIT IN PARALLEL STAGE j
COST TOTAL COST OF BATCH PROCESSING UNITS ($);
BINARY VARIABLES Y(K,J) ;
POSITIVE VARIABLES V(J) , B(I) , TL(I) , N(J) ;
EQUATIONS VOL(I,J) CALCULATE VOLUME OF STAGE j
CYCLE(I,J) CALCULATE CYCLE TIME OF PRODUCT i
TIME TIME CONSTRAINT
UNITS(J) CALCULATE NUMBER OF PROCESSING UNITS PER STAGE
LIM(J) LIMIT SELECTION TO ONE NUMBER
OBJ OBJECTIVE FUNCTION DEFINITION;
* CONVEXIFIED FORMULATION OF MODEL EQUATIONS
* Volume requirement in stage j
VOL(I,J).. V(J) =G= LOG(S(I,J)) +B(I) ;
* Cycle time for each product i
CYCLE(I,J).. N(J) + TL(I) =G= LOG(T(I,J)) ;
* Constraint for production time
TIME.. SUM(I , Q(I) * EXP(TL(I) - B(I))) =L= H ;
* Relating number of units to 0-1 variables
UNITS(J).. N(J) =E= SUM(K , COEFF(K) * Y(K,J)) ;
* Only one choice for parallel units is feasible
LIM(J).. SUM(K , Y(K,J)) =E= 1 ;
Chen CL 39
* Defining objective function (Last as per DICOPT++ requirement)
OBJ.. COST =E= SUM(J ,ALPHA(J)*(EXP(N(J) + BETA(J)*V(J))));
* Lower bound specified for DICOPT++
COST.LO = 0. ;
* BOUNDS SECTION
V.LO(J) = LOG(VLOW) ;
V.UP(J) = LOG(VUPP);
N.UP(J) = LOG(3.) ;
* Tight lower bounds are computed below for cycle times and batch sizes
PARAMETERS TLLOW(I) LOWER BOUND ON TL(I)
TLUPP(I) UPPER BOUND ON TL(I) ;
TLLOW(I) = SMAX(J, T(I,J) / EXP(N.UP(J))) ;
TLUPP(I) = SMAX(J, T(I,J) ) ;
TL.LO(I) = LOG(TLLOW(I)) ;
TL.UP(I) = LOG(TLUPP(I)) ;
PARAMETERS BLOW(I) LOWER BOUND ON B(I)
BUPP(I) UPPER BOUND ON B(I) ;
BLOW(I) = Q(I) * ( SMAX(J, T(I,J) / EXP(N.UP(J))))/ H;
BUPP(I) = MIN( Q(I) , SMIN(J , EXP(V.UP(J))/S(I,J))) ;
B.LO(I) = LOG(BLOW(I)) ;
B.UP(I) = LOG(BUPP(I)) ;
Chen CL 40
* INITIAL POINT
* Binary variables set for 3 units per stage
Y.L(K,J) = 0.;
Y.L(3,MIXER) = 1;
Y.L(3,REACTOR) = 1;
Y.L(3,CENTRIFUGE) = 1;
N.L(J) = SUM( K , COEFF(K)* Y.L(K,J) ) ;
* Batch sizes are set at the mid-point between bounds
B.L(I) = (B.UP(I) + B.LO(I))/2 ;
V.L(J) = SMAX(I , LOG(S(I,J)) + B.L(I)) ;
TL.L(I) = SMAX(J , LOG(T(I,J))-N.L(J)) ;
MODEL BATCH / ALL / ;
OPTIONS LIMCOL = 0 ;
OPTIONS ITERLIM = 10000 ;
OPTION OPTCR=0;
SOLVE BATCH USING MINLP MINIMIZING COST ;
* Converting variables into original form
ACTV(J) = EXP(V.L(J)) ;
ACTB(I) = EXP(B.L(I)) ;
ACTTL(I) = EXP(TL.L(I));
ACTN(J) = EXP(N.L(J)) ;
* Optimal Design
DISPLAY ACTV ;
DISPLAY ACTB ;
DISPLAY ACTTL ;
DISPLAY ACTN ;
Chen CL 41
Multiproduct Batch Plant Design
Optimal Design
optimal inve. cost = 167, 428
Product A: batch numb = 320
batch size = 625 kg cycle time = 10 hr
Product B: batch numb = 467
batch size = 321.5 kg cycle time = 6 hr
Chen CL 42
Thank You for Your Attention
Questions Are Welcome