End Term Ques Paper - Prof. Sawik - FINAL

You might also like

Download as docx, pdf, or txt
Download as docx, pdf, or txt
You are on page 1of 8

National Institute of Industrial Engineering

Vihar Lake, Mumbai

Batch: PGP 2020-22, PGP 2021-23, Fellow 1st Year (2021)

Subject: Global Online Certification Course on Supply Chain Operations and


Disruptions Management: A Way Forward

Prof. Prof. Tadeusz Sawik

Max. Marks: 40 Date: 14/11/2021

 All seven questions are compulsory.


 Keep Installed the AMPL Software for Running the Codes. Use “option
solver gurobi” for while solving. Give command “reset” before running a
fresh code. Ensure mod and dat files are loaded in the appropriate path.
 All answers are to be filled in the MS Quiz Link Floated
Question 01 (6 points)
Consider a scheduling flowshop with single stations and unlimited in-process buffers to minimize schedule
length. Given are following parameters:

The mathematical formulation is given as:

1
Complete the AMPL code for the above problem for the missing variables (1 x 4 = 4 points):

#SETS AND PARAMETERS#


param m:=3;
param n:=10;

set I:=1..?; #Question1(A)#


set K:=1..n;
param p{i in I,k in K};
param Q;
#END SETS AND PARAMETERS#
#VARIABLES#
var CMAX>=0;
var c{i in I, k in K} >= 0;

var y{k in K,l in K:?<l} binary; #Question1(B)#


#END VARIABLES#
#OBJECTIVE FUNCTION#
minimize SCHEDULE_LENGTH: CMAX;
#CONSTRAINTS#
subject to constr_1{k in K}: c[1,k] >= p[1,k];
subject to constr_2{i in I,k in K: i>1}: c[i,k] - c[i-1,k] >= p[i,k];

subject to constr_3{i in I,k in K,l in K:k<l}: c[i,k] + Q*y[k,l] >= c[?,l]+p[i,k]; #Question1(C)#
subject to constr_4{i in I,k in K,l in K:k<l}: c[i,l] + Q*(1 - y[k,l]) >= c[i,k]+p[i,l];

subject to constr_5{k in K}: c[m,k] <=?; #Question1(D)#


#END OF MODEL#

Question 1(E): Run the AMPL code for the above model for the dataset given below and find
the optimal solution (2 points):
param p:=
: 1 2 3 4 5 6 7 8 9 10:=
14215435218
22586747362
3 2 8 6 2 2 4 2 7 1 8;
param Q:=100;

2
Question. 02 (4 points)

Consider the problem for finding a subset of m outcomes with minimum total value, out of n, (n >
m), outcomes. Given are following parameters:

The mathematical formulation is given as:

Complete the AMPL code for the above problem for the missing variables (1 x 4 = 4 points):

#SETS AND PARAMETERS#


param n:=5; #number of all outcomes
param m:=3; #number of outcomes to be selected
set K:= 1..?; #set of all outcomes # #Question 2(A)#
param a{k in ?}:= if k=1 then 10 else if k=2 then 1 else if k=3 then 11 else if k=4 then 3 else 5 ; #Question
2(B)#
#example of outcome values
#END SETS AND PARAMETERS#
#VARIABLES#
var z{k in K} binary; #=1, if outcome a[k] is added to sum; z[k]=0, otherwise (outcome selection variable)
#END VARIABLES#
#OBJECTIVE FUNCTION#
minimize SUM_OF_OUTCOMES:
sum{k in K}?[k]*z[k]; #Question 2(C)#
#CONSTRAINTS#
subject to constr_1: sum{k in K}? [k]=m; #Question 2(D)#
#END OF MODEL#
3
Question 03 (10 points)

Consider a selection of risk-averse supply portfolio. Assume that the manufacturer does not pay for
undelivered orders from suppliers and needs to pay contractual penalty cost for unfulfilled customer
demand.
Given are following parameters:

The mathematical formulation is given as:

4
Complete the AMPL code for the above problem for the missing variables (1 x 8 = 8 points):

#SETS AND PARAMETERS#


param m:=3; #number of part suppliers
set I:=1..?; #set of suppliers# #Question3(A)#
param f{i in I}:=if i=1 then 500 else if i=2 then 1000 else 1500; #fixed ordering cost for supplier i
param o{i in I}:= if i=1 then 15 else if i=2 then 10 else 5; #unit price of of parts purchased from supplier i
param g:=30; #per unit loss of unfulfilled customer demand
param D:=12000; #total demand for parts
param c{i in I}:= ceil(D/(m/2)); #capacity of supplier i
set S:=1..2^m; #set of disruption scenarios
param dp{i in I,s in S}; #disruption pattern in scenario s (dp[i,s]=1 normal operation, dp[i,s]=0 disruption)
param p{i in I}:=if i=1 then 0.005 else if i=2 then 0.01 else 0.05;
#probability of local disruption for supplier i
param pg:=0.01;#global disruption probability
param P{s in S}:=if sum{i in I}dp[i,s]>0 then (1-pg)*(prod{i in I:dp[i,s]=1}(1-p[i]))*(prod{i in I:dp[i,s]=0}
(p[i]))
else pg+(1-?)*(prod{i in I}(p[i])); #Question3(B)#
#probability of disruption scenario s
param alpha:=0.95; #confidence level
#END SETS AND PARAMETERS#
#VARIABLES#
var u{i in I} binary;#=1, if parts are ordered from supplier i, otherwise u[i]=0
var v{i in I}>=0;#order allocation variable (the fraction of total demand for parts purchased from supplier i)
var Q{s in S}>=0;#tail cost per part: amount by which cost in scenario s exceeds VaR
var VaR; #Value-at-Risk of cost per part
#END VARIABLES#
#OBJECTIVE FUNCTION#
minimize ConditionalCost_at_Risk:

5
VaR+sum{s in S}P[s]*Q[?]/(1-?); #Question3(C) and Question 3(D)#
#CONSTRAINTS#
subject to constr_1: sum{i in I}v[?]=1; #Question3(E)#
subject to constr_2{i in I}: v[i]<=u[i];
subject to constr_3{i in I}: D*v[i]<=c[i]*u[i];
subject to constr_4{s in S}: Q[s]>=
sum{i in I}f[i]*u[i]/?
+sum{i in I}?[i]*v[i]
+sum{i in I:dp[i,s]=0}(g-o[i])*v[i]
-?; #Question3(F), Question 3(G), and Question 3(H)#
#END OF MODEL#
Question 3(I): Run the AMPL code for the above model for the dataset given below and find the
optimal solution [*Report your answer up to 2 decimal places only] (2 points):
param dp(tr):=
: 1 2 3 :=
1000
2001
3010
4011
5100
6101
7110
8 1 1 1;

6
Question. 04 (5 points)

Consider a problem of scheduling customer orders to maximize service level and balance
production. This is Model 7 discussed in the class.
Re-run the AMPL code with the following sensitivity analysis (one at a time) and mention your
solution for each change: [1 x 5 = 5 points]
*After each sub-question, restore back to the original code, before attempting the next sub-
question.
Question 4(A): Lambda = 0.5
Question 4(B): Gamma = 0.3
Question 4(C): Distribution related to due date of customer order changes between Uniform
(11,20)
Question 4(D): Distribution related to size of customer order changes between Uniform (5,10)
Question 4(E): Capacity of output buffer changes to 40000

Question. 05 (5 points)

Consider a problem of balancing and batch scheduling of SMT lines. This is Model 3 discussed in
the class.
Re-run the AMPL code with the following sensitivity analysis (one at a time) and mention your
solution for each change: [1 x 5 = 5 points]
*After each sub-question, restore back to the original code, before attempting the next sub-
question.
Question 5(A): Increasing the number of buffers for J[6] from 2 to 3
Question 5(B): Changing parameter r for stage 12 for all g board types to 40 (equal for all)
Question 5(C): Changing parameter r for stage 1 for all g board types to 15 (equal for all)
Question 5(D): Reducing placement time of both stages 5 and 9 for component type 3 from 10 to 5
[HINT: #PLACEMENT TIMES q(i,f) for component types param q:=)].
Question 5(E): Increasing placement time of both stages 5 and 9 for component type 10 from 14 to
20 [HINT: #PLACEMENT TIMES q(i,f) for component types param q:=)].

Question. 6 (5 points)

Consider a problem of optimization of cybersecurity investments in Supply Chains to Minimize


expected cost. This is Model 16 discussed in the class.
Re-run the AMPL code with the following sensitivity analysis (one at a time) and mention your
solution for each change: [1 x 5 = 5 points]

7
[*Report your answer up to actual value of first 2 decimal places only, that is no rounding up
related to third decimal place]
*After each sub-question, restore back to the original code, before attempting the next sub-
question.
Question 6(A): Changing strong implementation as strong implementation = light cost*20
Question 6(B): Changing medium implementation to same as light implementation, that is,
medium implementation = light cost*1
Question 6(C): Changing the probability of attack on supply chain nodes as 0.35 throughout
Question 6(D): Changing the loss from breach of node i for node 8 from 2,500 to 10,000
Question 6(E): Changing the intrinsic vulnerability of supply chain node, for node 1 from 0.6713
to 0.5

Question. 07 (5 points)

Consider a problem of selection of primary and recovery supply portfolios and production
scheduling under disruption risks. This is Model 13 discussed in the class.

Re-run the AMPL code with the following sensitivity analysis (one at a time) and mention your
solution for each change: [1 x 5 = 5 points]
[*Report your answer up to actual value of first 2 decimal places only, that is no rounding up
related to third decimal place]
*After each sub-question, restore back to the original code, before attempting the next sub-
question.
Question 7(A): Local non-disruption probability param p1 = 0.85 for all suppliers i
Question 7(B): param pr{r in R}:=if r=1 then 0.005, else 0.01
Question 7(C): param f:= #*1000, becomes 10 throughout (for all 4)
Question 7(D): param p becomes 10 throughout (for all 4)
Question 7(E): probability of disruption scenario s:
param P{s in S}:=if ts[s]=1 then 0.2*(prod{r in R}P1[r,s]) else if ts[s]=2 then 0.3*(prod{r in
R}P1[r,s])else 0.5*(prod{r in R}P1[r,s]);

You might also like