Application of Goal Programming in Project

Selection Decision
A Case Study from Indian Coal Mining Industry

Submitted by:
Deepika Bharadwaj

Divya Suryavanshi

Kartik Soni

Kriti Keshav
INTRODUCTION
A wide variety of problems in engineering, industry, and many other fields, involves the
simultaneous optimization of several objectives. In many cases, the objectives are defined in
incomparable units, and they present some degree of conflict among them (i.e., one objective
cannot be improved without deterioration of at least another objective). These problems are
called Multi-objective Optimization Problems (MOPs).

Analysis of decision making situations brings out the fact that the project selection decision
very often involves different influence groups, hence different objectives/goals. Although
various techniques have been reported in the literatures which are capable of tackling a multi-
objective decision environment, the goal programming technique has been identified as the
most appropriate for this research study.

Page | 2
LITERATURE REVIEW
Goal Programming Problem:

Goal programming Approach establishes a scientific numeric goal for each of the objective
and then attempts to achieve each goal sequentially up to a satisfactory level rather than an
optimal level. It is an extension of linear programming in which targets are specified for a set
of constraints.
THE GENERALISED GOAL PROGRAMMING MODEL
The goal programming technique, originally developed by Charnes and Cooper which can be
expressed mathematically as:

Minimize : Z=∑ di+ + di-

Subject to linear constraints
Goal Constraints: ∑aijxj-di+ + di-=bi , for i = 1,2,3,....,m
System Constraints: ∑aijxj ≤ bi , for i = m+1,....,m+p
∑aijxj = bi, for i = m+1,....,m+p
∑aijxj ≥ bi, for i = m+1,....,m+p
+ -
With di ,di ,xj ≥ 0,for i = 1,2,....m; for j = 1,2,.....,n

Where, there are m goals, p system constraints and n decision variables

Z = objective function = Summation of all deviations
aij = the coefficient associated with variable j in the ith goal xj = the jth decision variable
bi = the associated right hand side value i
di- = negative deviational variable from the ith goal (underachievement)
di+ = positive deviational variable from the ith goal (overachievement).
Both overachievement and underachievement of a goal cannot occur simultaneously.
Hence, either one or both of these variable must have a zero value
that is,
di+× di- = 0
. Both variables apply for the non-negativity requirement as to all other linear programming
variables
that is,
di+, di- ≥ 0

In goal programming there are two basic models: the pre–emptive (lexicographic) model and
the non  -pre–emptive (Archimedean)  model.
In the pre–emptive model, goals are ordered according to priorities. The goals at a certain
priority level are considered to be infinitely more important than the goals at the next level.
With the Archimedean model, weights or penalties for not achieving targets must be specified
and one attempts to minimize the weighted sum of goal under–achievement.

Page | 3
In pre-emptive goal programming, the objectives can be divided into different priority
classes. Here, it is assumed that no two goals have equal priority. The goals are given ordinal
ranking and are called pre-emptive priority factors. These priority factors have the
relationship P1 >>> P2.....>>>Pi >>> Pi+1>>>....>>>Pm where the P1 goal is so much more
important than the P2 goal and P2 goal will never be attempted until the P1 goal is achieved
to the greatest extent possible. The priority relationship implies that multiplication by n,
however large it may be, cannot make the lower-level goal as the higher goal (that is, Pi >
Pi+1).
The model can be stated as:
Minimize: Z=∑ Pi(di+ + di-)
Subject to linear constraints
Goal Constraints: ∑aijxj-di+ + di-=bi , for i = 1,2,3,....,m
System Constraints: ∑aijxj ≤ bi , for i = m+1,....,m+p
∑aijxj = bi, for i = m+1,....,m+p
∑aijxj ≥ bi, for i = m+1,....,m+p
+ -
With di ,di ,xj ≥ 0,for i = 1,2,....m; for j = 1,2,.....,n
Where, there are m goals, p system constraints, k priority levels and n decision variables Pi =
the pre-emptive priority factors of the ith goal.

If the decision maker is more interested in direct comparisons of the objectives then weighted
goal programming should be used. The weighting of deviational variables at the same
priority level shows the relative importance of each deviation. Charnes and Cooper (1977)
stated the weighted goal programming model as:
Minimize: Z=∑ (Wi+ di+ + Wi- di-), for i = 1,2,...,m
Subject to linear constraints
Goal Constraints: ∑aijxj-di+ + di-=bi , for i = 1,2,3,....,m
System Constraints: ∑aijxj ≤ bi , for i = m+1,....,m+p
∑aijxj = bi, for i = m+1,....,m+p
∑aijxj ≥ bi, for i = m+1,....,m+p
+ -
With di ,di ,xj ≥ 0,for i = 1,2,....m; for j = 1,2,.....,n
Where, Wi+ and Wi- are non-negative constants representing the relative weight to be assigned
to the respective positive and negative deviation variables. The relative weights may be any
real number, where the greater the weight the greater the assigned importance to minimize
the respective deviation variable to which the relative weight is attached. This model is a non-
preemptive model that seeks to minimize the total weighted deviation from all goals stated in
the model.

Page | 4
ABOUT THE COMPANY
Project selection is a crucial decision in capital management, involves various issues, and it is
an important consideration in the mining industries. So far in the Indian mining industry, little
attempt has been made for development of any rational framework for project selection
decisions. In this research report, it is intended to formulate a project selection problem of a
coalmining company, Indian Mines Limited (IML).

IML is the largest coal producing company in India, producing different grades of coal.
Although the nationalisation of IML took place phase-wise, its final formulation as a public
sector corporate body came up in the early 70’s, and included 425 coal mines. Today this
company contributes more than 90% of national coal production. Being a public sector
industry and coal playing a vital role in the Indian economy, the need for effective and
efficient management of IML is beyond doubt. However, a critical survey on its past
performance does not exhibit a bright picture in terms of its financial efficiency or capital
utilization. Further, IML has accepted a challenging task of production of 375 million tonnes
of coal per annum by the turn of this century (present annual production being close to
200million tonnes). In this pursuit, it is planning to invest around two thousand crores of
rupees on capital project.

Page | 5
PROBLEM IDENTIFICATION order to identify the appropriate goals for
According to operating managers, in order to identify the appropriate goals for the project
selection problem of IML some of the economic criteria are considered during feasibility
analysis of project proposals. Some factors that could be considered (maximisation of
customer satisfaction, maximisation of safety, minimisation of environmental hazards and
many more) were ignored due to its qualitative nature, involving intangible factors which
could have created difficulties in data collection or data validation.

The following five goals have been identified as appropriate for their inclusion in the
proposed goal programming model:

Goal 1: Capital Investment goal

Goal 2: Cost of production goal

Goal 3: Profit goal

Goal 4: Manpower goal

Goal 5: Demand goal

Analysis of the past data indicates that demand is random and it is included in the proposed
model as a stochastic constraint. The top management of IML wants to keep production at a
level where at least 80% of demand should be satisfied.

Page | 6
 MODEL FORMULATION

The proposed model depicts a capital investment decision situation involving selection
among a set of alternative mine projects at the level of a mine area. A mine area involves
management of operations of a group of mines. There are two types of mine projects -
reconstruction projects and new mines. The reconstruction mine project represents an
investment strategy for enhancing production of the existing mine system design.

The feasibility report and economic data of each mine project are available and the
management is to rationally distribute the total capital among any combination of these two
types of mine projects. The proposed goal programming model aims at selecting the most
suitable subset of mine projects among a set of feasible project proposals as an accepted
investment plan with simultaneous attainment of all goals as best as possible.

Project Invest Produc Produc Increas Cost/ye Cost/ye Profit/y Profit/y Manpo Manpow
ment tion tion ed ar(Rs. ar(Rs. ear(Rs. ear(Rs. wer er
(Rs. (MT) (MT) Produc Crore) Crore) Crore) Crore)
Crore) tion(M
T)
Present Future Present Future Present Future Present Future
1 70.87 0 0.6 0.6 0 26.55 0 -4.55 0 2196
2 10.62 0.225 0.315 0.09 11.54 12.72 -4 -1.89 1792 1615
3 9.47 0 0.195 0.195 0 5.01 0 -8.38 0 958
4 19.62 0 0.293 0.293 0 8.64 0 -0.87 0 728
5 11.8 0.27 0.45 0.18 16.06 22.04 -6.62 -4.03 3042 3042
6 69.48 0 0.563 0.563 0 24.17 0 2.27 0 1643
7 13.06 0.18 0.45 0.27 7.26 10.6 -3.03 4.88 293 417
8 12.31 0 0.24 0.24 0 7.55 0 -1.04 0 237

We are further provided with the following information:

 A case problem is developed for selection among eight mine projects, out of which
three are reconstruction projects and the remaining five are new mines.
 A total of 110 crores of rupees is available for capital investment and target average
cost of production is expected to be 350 per ton of coal production.
 The financial performance of coal company under consideration is quite poor and it is
funded by the government to cover up the losses. However, the target loss for the
project selection case study is supposed be 6 crores of rupees per annum.
 The lower limit and upper limit goals of manpower are expected to be 8,000 and
9,000 respectively.
 Analysis of past demand data of coal shows expected annual demand to be 2 million
ton and standard deviation to be 50,000 ton only.

Page | 7
The model can be formulated as follows

Goal 1: Capital investment goal

70.87x1 + 10.62x2 + 9.47x3 + 19.62x4 + 11.80x5 + 69.48x6 + 13.06x7 + 12.31x8 ≤ 110

70.87x1 + 10.62x2 + 9.47x3 + 19.62x4 + 11.80x5 + 69.48x6 + 13.06x7 + 12.31x8 - d1+ + d1- =
110

Goal 2: Production cost goal

26.55x1 + 12.72x2 + 11.54(1 - x2) + 5.01x3 + 8.64x4 + 22.04x5 + 16.06(1 - x5) + 24.17x6 +
10.60x7 + 7.26(1 - x7) + 7.55x8≤ 350(0.6x1 + 0.09x2 + 0.195x3 + 0.293x4 + 0.18x5 + 0.563x6 +
0.27x7 + 0.24x8 + 0.225 + 0.27 + 0.18)

-183.45x1 - 30.32x2 - 63.24x3 - 93.91x4 - 57.02x5 - 172.88x6 - 91.16x7 - 76.45x8 ≤ 201.39

-183.45x1 - 30.32x2 - 63.24x3 - 93.91x4 - 57.02x5 - 172.88x6 - 91.16x7 - 76.45x8 – d2+ + d2- =
201.39

Goal 3: Profit goal

-4.55x1 - 1.89x2 - 4(1 - x2) - 8.38x3 - 0.87x4 - 4.03x5 - 6.62(1 - x5) + 2.27x6 + 4.88x7 - 3.03(1 -
x7) - 1.04x8≥ - 6

-4.55x1 + 2.11x2 - 8.38x3 - 0.87x4 + 2.59x5 + 2.27x6 + 7.91x7 - 1.04x8 - d3+ + d3- = 7.65

Goal 4: Manpower goal

Page | 8
2196x1 + 1615x2 + 1792(1 - x2) + 958x3 + 728x4 + 3042x5 + 3042(1 - x5) + 1643x6 + 417x7 +
293(1 - x7) + 237x8≥ 8000

2196x1 + 1615x2 + 1792(1 - x2) + 958x3 + 728x4 + 3042x5 + 3042(1 - x5) + 1643x6 + 417x7 +
293(1 - x7) + 237x8≤ 9000

2196x1 - 177x2 + 958x3 + 728x4 + 1643x6 + 124x7 + 238x8 - d4+ + d4- = 2873

2196x1 - 177x2 + 958x3 + 728x4 + 1643x6 + 124x7 + 238x8 - d5+ + d5- = 3873

System constraint:

0.6x1 + 0.09x2 + 0.225 + 0.195x3 + 0.293x4 + 0.18x5 + 0.27 + 0.563x6 + 0.27x7 + 0.18 +
0.24x8≥ 2.042

0.6x1 + 0.09x2 + 0.195x3 + 0.293x4 + 0.18x5 + 0.563x6 + 0.27x7 + 0.24x8≥ 1.367

xi = decision variable that ith project is selected, i = 1, 2, … , 8

dj+ = positive deviation variable for jth goal, j = 1, 2, … , 5, dj+≥ 0

dj- = negative deviation variable for jth goal, j = 1, 2, … , 5, dj-≥ 0

Objective function:

Minimize (P1d1+ + P2d2+ +P3d3- + P4(d4- + d5+) )

Hence, the goal programming problem will become:

Minimize (P1d1+ + P2d2+ +P3d3- + P4 (d4- + d5+) )

Page | 9
Subject to

70.87x1 + 10.62x2 + 9.47x3 + 19.62x4 + 11.80x5 + 69.48x6 + 13.06x7 + 12.31x8 - d1+ + d1- =
110

-183.45x1 - 30.32x2 - 63.24x3 - 93.91x4 - 57.02x5 - 172.88x6 - 91.26x7 - 75.36x8 -d2+ + d2- =
201.39

-4.55x1 + 2.11x2 - 8.38x3 - 0.87x4 + 2.59x5 + 2.27x6+ 7.91x6 - 1.04x6 -d3+ + d3- = 7.65

2196x1 - 177x2 + 958x3 + 728x4 + 1643x6 + 124x7 + 238x8 - d4+ + d4- = 2873

2196x1 - 177x2 + 958x3 + 728x4 + 1643x6 + 124x7 + 238x8 - d5+ + d5- = 3873

0.6x1 + 0.09x2 + 0.195x3 + 0.293x4 + 0.18x5 + 0.563x6 + 0.27x7 + 0.24x8≥ 1.367

xi, dj+, dj-≥ 0

i = 1, 2, … , 8, j = 1, 2, … ,5.

Page | 10
THE INPUT DATA

Footnote
In the above input data, Let x9 = d1+ , x10 = d1- , x11 = d2+ , x12 = d2-, x13 = d3+ , x14 =
d3- , x15 = d4+ and x16 = d4- , x17 = d5+ , x18 = d5-. The Tora software was applied on
given Table to obtain the table below.

Page | 11
Interpretation Of The Solution

The application of the simplex method (Big M – Method) by Tora package, gives the optimum
solution as follows:

x1 = 0

x2 = 0

x3 = 0.65

x4 = 1

x5 = 1

x6 = 0.83

x7 = 1

x8 = 0.15

Page | 12
 d1+ d1- d2+ d2- d3+ d3- d4+ d4- d5+ d5-
0 0 0 639.30 0 1.77 0 0 0 1000

DISCUSSION OF FINDINGS

Each of the inequalities in the goal formulation represents a goal that the institution aspires to
satisfy. However, compromised solutions among the conflicting goals were found, and the
approach to finding a compromise solution is to convert each inequality into flexible goal in
which the corresponding constraints may be violated.

The simplex method (Big–M Method) was used to solve the weighted goal programming
model formulated, and the optimal solution obtained by TORA package is Z = 0.44, x3 =
0.65, x4=1, x5=1, x6=0.83, x7=1, x8=0.15, d2- = 639.30, d3- = 1.77, d5- = 1000, all the
remaining variables are zero.

CONCLUSION

RECOMMENDATIONS

Page | 13