Goal Programming

Applications for Services

References:
‘OperationsResearch’ byWinston
‘OperationsResearchModelsAndMethods’byPaulA.JensenAndJonathanF.Bard
 Introduction
• Many service operations management decisions involve trying to make the most
effective use of an organization's resources.

• Resources typically include

• labor,
• money,
• storage space/capacity, or
• materials.
• These resources may be used to produce services such as
• schedules for shipping and production,
• advertising policies,
• investment decisions, or
• hospital meal plans.

• Linear programming (LP) and Goal programming (GP) are widely used
mathematical techniques designed to help operations managers in planning and
decision making relative to the tradeoffs necessary to allocate resources.
Overview of Linear Programming
All LP problems have four properties in common.

1. All problems seek to maximize or minimize some quantity.

2. The presence of restrictions, or constraints, limits the

degree to which we can pursue our objective.

3. There must be alternative courses of action to choose from.

4. The objective and constraints in linear programming

problems must be expressed in terms of linear equations or
inequalities.
Applications of Linear Programming
• Ingredient Blending Applications: The diet problem
• Shift Scheduling Problem
• Workforce Planning Problem
• Transportation Problem
• Employee Scheduling Applications: An Assignment
Problem
• Labor Planning
• Marketing Applications: Media Selection
•…
Goal Programming

• Linear programming models allow for only one objective, and all
constraints must be met absolutely.

• In real life, virtually all problems have several objectives in different

units of measure. This makes construction of a consolidated single
objective difficult.

• For example, an emergency ambulance system could have the

following objectives:
1. Maintain an average response time of approximately 4 minutes.
2. Ensure that 90 percent of all calls receive aid in less than 10 minutes.
3. Try not to exceed a budget of $100,000 per year.
4. Allocate calls to ambulance crews in an equitable manner.
Goal Programming
• is a variation of LP that permits multiple and conflicting goals with
different dimensions.
• Multiple goals are rank-ordered and treated as preemptive priorities.

• The solution approach:

• higher-ranked goals are not sacrificed to achieve lower-ranked goals.
• is equivalent to solving a series of nested LP problems in which higher-
ranked goals become constraints on lower-ranked goals.

• While LP optimizes a single objective, goal programming

minimizes deviations from goals.

• This solution approach is known as satisfying, because not all

goals necessarily will be met. Instead, the goals will be achieved
as closely as possible.
Goal Programming
Goal Types
Three possible types of goals:
1. Lower One-Sided Goal: For the 𝑘𝑡ℎ objective function, a lower limit, 𝐿𝑘,
that the decision maker does not want to fall below. It is desired to
achieve a value of "at least" 𝐿𝑘 for the objective. Falling under this value
is permissible if it leads to the best compromise.

2. Upper One-Sided Goal: For the 𝑘𝑡ℎ objective function, an upper limit, 𝑈𝑘,
that the decision maker does not want to exceed. It is desired to achieve
a value of "at most" 𝑈𝑘 for the objective. Exceeding this value is
permissible if this leads to the best compromise.
Goal Programming
Goal Types
3. Two-Sided Goal: Sets a specific target value, 𝐺𝑘, for the 𝑘𝑡ℎ objective so
the value of 𝑓𝑘(𝒙) should be "equal to" some 𝐺𝑘. The goal might be
written as an equality constraint:

The solution process will allow for deviations from this goal in either
direction.
Goal Programming
Constraints
• stated as equalities
• contain both plus (+) and minus (-) deviational variables in
addition to decision variables

𝑦𝑘+ : positive deviation or the amount

by which the 𝑘𝑡ℎ goal is exceeded

𝑦𝑘− : negative deviation or the amount

by which the 𝑘𝑡ℎ goal is underachieved
Goal Programming
Objective function
• contain only deviational variables (i.e., plus or minus
deviations from goals) (𝑦𝑘+, 𝑦𝑘−)

• deviational weights can be used to distinguish relative

importance within a priority level
𝑝+: the penalty applied to the positive component 𝑦+.
𝑘 𝑘
𝑝 : the penalty applied to the negative component 𝑦−.
−
𝑘 𝑘

• Objective is to minimize the sum of the deviations at each

priority level
• The penalties measure the relative importance of the goals
Goal Programming
Objective function
• The three kinds of goals are associated with the following penalty
assignments:
1. lower one-sided goal: 𝑝− +
𝑘 > 0, 𝑝𝑘 = 0
▪ underachievement is penalized.
▪ we set 𝑝𝑘− > 0 because we want to penalize the underachievement of the goal 𝐿 𝑘.
▪ we set 𝑝𝑘+ = 0 because we don’t want to penalize its achievement.

2. upper one-sided goal: 𝑝− = 0, 𝑝+ > 0

𝑘 𝑘
▪ exceeding the value is penalized.
▪ we set 𝑝𝑘+ > 0 because we want to penalize the exceeding the goal 𝑈 𝑘.
▪ we set 𝑝𝑘− = 0 because we don’t want to penalize its achievement.

3. two-sided goal: 𝑝− > 0, 𝑝+ > 0

𝑘 𝑘
▪ both underachievement and exceeding are penalized.
▪ we set 𝑝𝑘− > 0, 𝑝+
𝑘 > 0 because we want to penalize any underachievement of or
exceeding the goal 𝐺𝑘.
Goal Programming
Case Study 1: DeWright Company
• Company is considering three new products.
• Their goals are
(1)achieving a long-run profit (net present value) of at least $125 million from
these products,
(2) maintaining the current employment level of 4,000 employees,
(3) holding the capital investment to less than $55 million.
• Each new product’s contribution to profit, employment level, and capital
investment level is proportional to the rate of production. These
contributions, the goals and penalty weights are given below:
Goal Programming
Case Study 1: DeWright Company
• Decision Variables:
𝑥1: production amount of product 1
𝑥2: production amount of product 2
𝑥3: production amount of product 3

• Goals:

• Then, any feasible solution to the following LP meets the goals of DeWright:
Maximize (or Minimize) 0𝑥1 + 0𝑥2 + 0𝑥3 (or any other objective function)
𝑠𝑢𝑏𝑗𝑒𝑐𝑡 𝑡𝑜 12𝑥1 + 9𝑥2 + 15𝑥3 ≥ 125
5𝑥1 + 3𝑥2 + 4𝑥3 = 40
5𝑥1 + 7𝑥2 + 8𝑥3 ≤ 55
𝑥1 , 𝑥2 , 𝑥3 ≥ 0
Goal Programming
Case Study 1: DeWright Company
• Objective Function of Goal Programming Model:
 Goal Programming
Case Study 1: DeWright Company
• Rewrite the goal constraints as equalities using deviation
variables:
1) achieving a long-run profit (net present value) of at least $125
million from these products
12𝑥1 + 9𝑥2 + 15𝑥3 ≥ 125

• 𝑦− : profit underachievement amount (𝑝−=5)

1 1
• 𝑦+ : profit overachievement +
amount (𝑝 =0) → Not required in the objective function!
1 1
 Goal Programming
Case Study 1: DeWright Company
• Rewrite the goal constraints as equalities using deviation
variables:
2) maintaining the current employment level of 4,000 employees
5𝑥1 + 3𝑥2 + 4𝑥3 = 40

• 𝑦− : amount under employment goal (𝑝−=4)

2 2
• 𝑦+ : amount over +
employment goal (𝑝 =2)
2 2
Goal Programming
Case Study 1: DeWright Company
• Rewrite the goal constraints as equalities using deviation
variables:
3) holding the capital investment to less than $55 million.
5𝑥1 + 7𝑥2 + 8𝑥3 ≤ 55

• 𝑦− : amount under capital investment goal (𝑝−=0)

3 3
• 𝑦+ : amount over capital investment +
goal (𝑝 =3)
3 3
Goal Programming
Case Study 1: DeWright Company
• Goal Programming Model:
Goal Programming
Case Study 2: TV Advertising Schedule Problem
Goal Programming
Case Study 2: TV Advertising Schedule Problem
• Decision Variables:

• Goals:
Goal Programming
Case Study 2: TV Advertising Schedule Problem
• Then, any feasible solution to the following LP meets Priceler’s goals:

• It is found that no point that satisfies the budget constraint meets all three
goals.
→ No feasible solution!
→ So we need to determine a cost for each goal that is incurred for failing to
meet the goal.
Goal Programming
Case Study 2: TV Advertising Schedule Problem
• Information on penalty coefficients of the three goals:

• We can now formulate a goal programming model that

minimizes the cost incurred in deviating from three goals:
Goal Programming
Case Study 2: TV Advertising Schedule Problem
• Information on penalty coefficients of the three goals:

• We can now formulate a goal programming model that

minimizes the cost incurred in deviating from three goals:

(!) Only 𝑠 −′ 𝑠 in the objective funtion (all goals are ≥)

Goal Programming
Case Study 2: TV Advertising Schedule Problem
• Goal Programming Model:
Goal Programming
Case Study 2: TV Advertising Schedule Problem
with a Budget Goal
• A new goal: Budget restriction of $600,000
• $1 penalty is assessed for each thousand dollar by which this goal is unmet,
→ penalty coefficient of 4th goal: $1

• 𝑠4+ should be in the objective funtion (the budget goal is ≤)

• The updated Goal Programming Model:

Goal Programming
Case Study 2: TV Advertising Schedule Problem
with a Budget Goal
Preemptive Goal Programming
• If the decision maker is not able to determine precisely the relative
importance of the goals, preemptive goal programming tool can
be used.

• The decision maker must rank his goals from the most important
(goal 1) to least important (goal 𝑛).

• Let the objective function coefficient for the variable representing

goal i be Pi.
Assume: 𝑃1 >>>>> 𝑃2 >>>>> 𝑃3
Goal Programming
Case Study 2: TV Advertising Schedule Problem
with Preemptive Goal Programming
• The relative importance of goals are not determined precisely but
the Order of importance is: Goal 1, Goal 2, Goal 3.
(P1>>>>>P2>>>>>P3)
Goal Programming
Case Study 2: TV Advertising Schedule Problem
with Preemptive Goal Programming
Goal Programming
Case Study 2: TV Advertising Schedule Problem
with Preemptive Goal Programming
Goal Programming
Case Study 2: TV Advertising Schedule Problem
with Preemptive Goal Programming
Goal Programming
Case Study 2: TV Advertising Schedule Problem
with Preemptive Goal Programming
Preemptive Goal Programming
Case Study 3: Stereo Warehouse
• The retail outlet of Stereo Warehouse, a discount audio, audio-components store,
is planning a special clearance sale.

• The showroom has 400 square feet of floor space available for displaying this
week’s specials: a model X receiver and series Y speakers.
A receiver A pair of speakers
Wholesale cost ($) 100 50
Display space requirement (square feet) 2 4
Price ($) 150 70

• The budget for stocking the stereo items is $8000.

• The sales potential for the receiver is considered to be no more than 60 units;
however, the budget-priced speakers appear to have an unlimited appeal.

• The store manager, desiring to maximize gross profit, must decide how many
receivers and speakers to stock.
Preemptive Goal Programming
Case Study 3: Stereo Warehouse
• Let
𝑥: number of receivers to stock,
𝑦: number of speakers to stock

• This problem can be formulated as an LP problem in the following manner:

Max. 50𝑥 + 20𝑦 → gross profit
s.t. 2𝑥 + 4𝑦 = 400 → floor space
100𝑥 + 50𝑦 ≤ 8000 → budget
𝑥 ≤ 60 → sales limit
𝑥, 𝑦 ≥ 0
Note: Profit = Price-Cost; Receiver profit=150-100=50; Speaker profit=70-50=20
Preemptive Goal Programming
Case Study 3: Stereo Warehouse
• Suppose you prioritize your goals as follows:
• First Priority Goal: Exceeding sales limit of 60 receivers should be avoided (i.e.

stocking more receivers than the sales potential is to be avoided)

• Second Priority Goal:

• Exceeding budget limit is to be avoided

• Profit should be maximized

• Meeting budget is twice as important as maximizing profit.

• Third Priority Goal: Deviation from from floor space usage

Preemptive Goal Programming
Case Study 3: Stereo Warehouse
• Decision variables of Goal Programming formulation:
Preemptive Goal Programming
Case Study 3: Stereo Warehouse
• First Priority Goal: Exceeding sales limit of 60 receivers is
to be avoided
• Hence, include 𝑑+
4 in the objective function.
• Do not include 𝑑−
4 as falling below 60 receivers is no problem.
• Penalty coefficient: 𝑃1

• Third Priority Goal: Floor space used for display

• we want to use approximately 400 square feet – both more and less space
usage is to be avoided.
• both deviational variables are minimized
• Hence, include 𝑑+ and 𝑑− in the objective function.
2 2
 Preemptive Goal Programming
Case Study 3: Stereo Warehouse
• Second Priority Goal:
• Profit objective is translated into a goal at the 2nd priority.
– Only 𝑑1− is included in the objective function to be minimized
– Do not include 𝑑1+ as exceeding the profit goal is no problem ☺
– minimizing 𝑑1− forces the profit to approach 99,999, a large number far in
excess of expected profit.

• Exceeding the budget is to be avoided at the 2nd priority,

– so only 𝑑+
3 is found in objective function
– Do not include 𝑑−3 as falling below budget is no problem.
– deviational weight of 2: meeting budget is twice as important as maximizing
profit.
Preemptive Goal Programming
Case Study 3: Stereo Warehouse
• Goal Programming formulation:
Preemptive Goal Programming
Case Study: Stereo Warehouse
• Solve the problem with Excel Solver
• Step 1: Prepare the model for the First Priority Goal
Preemptive Goal Programming
Case Study 3: Stereo Warehouse
• Step 1: Solve the model for the First Priority Goal
• Solve and find d4plus
Preemptive Goal Programming
Case Study 3: Stereo Warehouse
• Step 2: Solve for the Second Priority Goal:
• Profit goal and Avoid exceeding budget limit ➔ 𝑑− + 2𝑑+
1 3
• (!) Add the d4plus value you found as a constraint
Preemptive Goal Programming
Case Study 3: Stereo Warehouse
• Step 3: Solve for the Third Priority Goal:
• Avoid exceeding space limit ➔𝑑+ + 𝑑−
2 2
• (!) Add the d4plus and d1minus + 2 * d3plus values you found as
constraints

RESULT:
𝑥 = 60 𝑦 = 40
Stock 60 receivers and 40 speakers
By using 120 sq feet less space
(used a total of 400-120=280 sqfeet)
And making a total profit of
profit=99999-96199=3800