BM2101

MULTICRITERIA DECISION MAKING

Goal Programming
A goal programming model is very similar to a linear programming model, with an objective function,
decision variables, and constraints.
EXAMPLE: Beaver Creek Pottery Company example. Recall that this model was originally formulated
in week 2 as follows:
𝑚𝑚𝑚𝑚𝑚𝑚𝑚𝑚𝑚𝑚𝑚𝑚𝑚𝑚𝑚𝑚 𝑍𝑍 = 𝑃𝑃40𝑥𝑥1 + 50𝑥𝑥2
𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠 𝑡𝑡𝑡𝑡
𝑥𝑥1 + 2𝑥𝑥2 ≤ 40 ℎ𝑟𝑟. 𝑜𝑜𝑜𝑜 𝑙𝑙𝑙𝑙𝑙𝑙𝑙𝑙𝑙𝑙
4𝑥𝑥1 + 3𝑥𝑥2 ≤ 120 𝑙𝑙𝑙𝑙. 𝑜𝑜𝑜𝑜 𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐
𝑥𝑥1, 𝑥𝑥2 ≥ 0
This is a standard linear programming model; as such, it has a single objective function for profit.
However, let us suppose that instead of having one objective, the pottery company has several
objectives, listed here in order of importance:
1. To avoid layoffs, the company does not want to use fewer than 40 hours of labor per day.
2. The company would like to achieve a satisfactory profit level of P1,600 per day.
3. Because the clay must be stored in a special place so that it does not dry out, the company
prefers not to keep more than 120 pounds on hand each day.
4. Because high overhead costs result when the plant is kept open past normal hours, the
company would like to minimize the amount of overtime.
Labor Goal

The first goal of the pottery company is to avoid underutilization of labor—that is, using fewer than 40
hours of labor each day. To represent the possibility of underutilizing labor, the linear programming
constraint for labor, 𝑥𝑥1 + 2𝑥𝑥2 ≤ 40 hours of labor, is reformulated as
𝑥𝑥1 + 2𝑥𝑥2 + 𝑑𝑑1− + 𝑑𝑑1+ = 40 ℎ𝑟𝑟
This reformulated equation is referred to as a goal constraint. The two new variables, 𝑑𝑑1− and 𝑑𝑑1+ ,
are called deviational variables. They represent the number of labor hours less than 40 (𝑑𝑑1− ) and
the number of labor hours exceeding 40 (𝑑𝑑1+ ,). More specifically, 𝑑𝑑1− represents labor underutilization,
and 𝑑𝑑1+ represents overtime. For example, if 𝑥𝑥1 = 5 bowls and 𝑥𝑥2 = 10 mugs, then a total of 25 hours
of labor have been expended. Substituting these values into our goal constraint gives
5 + 2(10) + 𝑑𝑑1− + 𝑑𝑑1+ = 40
25 + 𝑑𝑑1− + 𝑑𝑑1+ = 40
Because only 25 hours were used in production, labor was underutilized by 15 hours (40 - 25 = 15).
Thus, if we let 𝑑𝑑1− = 15 hours and 𝑑𝑑1+ = 0 (because there is obviously no overtime), we have
25 + 15 + 0 = 40
40 = 40

06 Handout 1 *Property of STI

 [email protected] Page 1 of 9
 BM2101

Now consider the case where 𝑥𝑥1 = 10 bowls and 𝑥𝑥2 = 20 mugs. This means that a total of 50 hours
has been used for production, or 10 hours above the goal level of 40 hours. This extra 10 hours is
overtime. Thus, 𝑑𝑑1− = 0 (because there is no underutilization) and 𝑑𝑑1+ = 10 hours.
In each of these two (2) brief examples, at least one (1) of the deviational variables equaled zero. In
the first example, 𝑑𝑑1+ = 0, and in the second example, 𝑑𝑑1− = 0. This is because it is impossible to use
fewer than 40 hours of labor and more than 40 hours of labor at the same time. Of course, both
deviational variables, 𝑑𝑑1− and 𝑑𝑑1+ , could have equaled zero if exactly 40 hours were used in production.
These examples illustrate one of the fundamental characteristics of goal programming: At least one
(1) or both of the deviational variables in a goal constraint must equal zero.
The next step in formulating our goal programming model is to represent the goal of not using fewer
than 40 hours of labor. We do this by creating a new form of objective function:
𝑚𝑚𝑚𝑚𝑚𝑚𝑚𝑚𝑚𝑚𝑚𝑚𝑚𝑚𝑚𝑚 𝑃𝑃1 𝑑𝑑1−
The objective function in all goal programming models is to minimize deviation from the goal constraint
levels. In this objective function, the goal is to minimize 𝑑𝑑1− , the underutilization of labor. If 𝑑𝑑1− equaled
zero, then we would not be using fewer than 40 hours of labor. Thus, it is our objective to make 𝑑𝑑1−
equal zero or the minimum amount possible. The symbol 𝑃𝑃1 in the objective function designates the
minimization of 𝑑𝑑1− as the first-priority goal. This means that when this model is solved, the first step
will be to minimize the value of 𝑑𝑑1− before any other goal is addressed.
The fourth-priority goal in this problem is also associated with the labor constraint. The fourth goal,
𝑃𝑃4 , reflects a desire to minimize overtime. Recall that hours of overtime are represented by 𝑑𝑑1+ ; the
objective function, therefore, becomes
𝑚𝑚𝑚𝑚𝑚𝑚𝑚𝑚𝑚𝑚𝑚𝑚𝑚𝑚𝑚𝑚 𝑃𝑃1 𝑑𝑑1− + 𝑃𝑃4 𝑑𝑑1+
As before, the objective is to minimize the deviational variable 𝑑𝑑1+ . In other words, if 𝑑𝑑1+ equaled zero,
there would be no overtime at all. In solving this model, the achievement of this fourth-ranked goal
will not be attempted until goals one, two, and three have been considered.
Profit Goal

The second goal in our goal programming model is to achieve a daily profit of P1,600. Recall that the
original linear programming objective function was
𝑍𝑍 = 40𝑥𝑥1 + 50𝑥𝑥2
Reformulate this objective function as a goal constraint with the following goal level:
𝑍𝑍 = 40𝑥𝑥1 + 50𝑥𝑥2 + 𝑑𝑑2− + 𝑑𝑑2+ = 𝑃𝑃1,600
The deviational variables 𝑑𝑑2− and 𝑑𝑑2+ represent the amount of profit less than P1,600 (𝑑𝑑2− ) and the
amount of profit exceeding P1,600 (𝑑𝑑2+ ). The pottery company’s goal of achieving P1,600 in profit is
represented in the objective function as
𝑚𝑚𝑚𝑚𝑚𝑚𝑚𝑚𝑚𝑚𝑚𝑚𝑚𝑚𝑚𝑚 𝑃𝑃1 𝑑𝑑1− , 𝑃𝑃2 𝑑𝑑2− , 𝑃𝑃4 𝑑𝑑1+
Notice that only 𝑑𝑑2− , is being minimized, not 𝑑𝑑2+ , because it is logical to assume that the pottery
company would be willing to accept all profits in excess of P1,600 (i.e., it does not desire to minimize
𝑑𝑑2+ , excess profit). By minimizing 𝑑𝑑2− , at the second-priority level, the pottery company hopes that
𝑑𝑑2− , will equal zero, which will result in at least P1,600 in profit.
Material Goal

06 Handout 1 *Property of STI

 [email protected] Page 2 of 9
 BM2101

The third goal of the company is to avoid keeping more than 120 pounds of clay on hand each day.
The goal constraint is
4𝑥𝑥1 + 3𝑥𝑥2 + 𝑑𝑑3− + 𝑑𝑑3+ = 120 𝑙𝑙𝑙𝑙
Because the deviational variable 𝑑𝑑3− represents the amount of clay less than 120 pounds, and 𝑑𝑑3+
represents the amount in excess of 120 pounds, this goal can be reflected in the objective function
as
𝑚𝑚𝑚𝑚𝑚𝑚𝑚𝑚𝑚𝑚𝑚𝑚𝑚𝑚𝑚𝑚 𝑃𝑃1 𝑑𝑑1− , 𝑃𝑃2 𝑑𝑑2− , 𝑃𝑃3 𝑑𝑑3+ , 𝑃𝑃4 𝑑𝑑1+
The term 𝑃𝑃3 𝑑𝑑3+ represents the company’s desire to minimize 𝑑𝑑3+ , the amount of clay in excess of 120
pounds. The 𝑃𝑃3 designation indicates that it is the pottery company’s third most important goal.
The complete goal programming model can now be summarized as follows:
𝑚𝑚𝑚𝑚𝑚𝑚𝑚𝑚𝑚𝑚𝑚𝑚𝑚𝑚𝑚𝑚 𝑃𝑃1 𝑑𝑑1− , 𝑃𝑃2 𝑑𝑑2− , 𝑃𝑃3 𝑑𝑑3+ , 𝑃𝑃4 𝑑𝑑1+
Subject to
𝑥𝑥1 + 2𝑥𝑥2 + 𝑑𝑑1− + 𝑑𝑑1+ = 40 ℎ𝑟𝑟
40𝑥𝑥1 + 50𝑥𝑥2 + 𝑑𝑑2− + 𝑑𝑑2+ = 𝑃𝑃1,600
4𝑥𝑥1 + 3𝑥𝑥2 + 𝑑𝑑3− + 𝑑𝑑3+ = 120 𝑙𝑙𝑙𝑙
𝑥𝑥1 , 𝑥𝑥2 , 𝑑𝑑1− , 𝑑𝑑2− , 𝑑𝑑2+ , 𝑑𝑑3− , 𝑑𝑑3+ ≥ 0
The one (1) basic difference between this model and the standard linear programming model is that
the objective function terms are not summed to equal a total value, 𝑍𝑍. This is because the deviational
variables in the objective function represent different units of measure. For example, 𝑑𝑑1− , and 𝑑𝑑1+
represent hours of labor, 𝑑𝑑2− represents dollars, and 𝑑𝑑3+ represents pounds of clay. It would be illogical
to sum hours, dollars, and pounds. The objective function in a goal programming model specifies only
that the deviations from the goals represented in the objective function be minimized individually, in
order of their priority.
NONLINEAR PROGRAMMING
Nonlinear Profit Analysis
When problems fit the general linear programming format but include nonlinear functions, they are
referred to as nonlinear programming problems.
Nonlinear programming problems are given a separate name because they are solved in a different
manner than are linear programming problems. In fact, their solution is considerably more complex
than that of linear programming problems, and it is often difficult, if not impossible, to determine an
optimal solution, even for a relatively small problem.
In linear programming problems, solutions are found at the intersections of lines or planes, and though
there may be a very large number of possible solution points, the number is finite, and a solution can
eventually be found.
However, in nonlinear programming there may be no intersection or corner points; instead, the
solution space can be an undulating line or surface, which includes virtually an infinite number of
points. For a realistic problem, the solution space may be like a mountain range, with many peaks
and valleys, and the maximum or minimum solution point could be at the top of any peak or at the
bottom of any valley. What is difficult in nonlinear programming is determining if the point at the top

06 Handout 1 *Property of STI

 [email protected] Page 3 of 9
 BM2101

of a peak is just the highest point in the immediate area (called a local optimal, in calculus terms) or
the highest point of all (called the global optimal).
Nonlinear programming starts by determining the optimal value for a single linear function.
Recall that in break-even analysis the profit function, 𝑍𝑍, is formulated as
𝑍𝑍 = 𝑣𝑣𝑣𝑣 − 𝑐𝑐𝑓𝑓 − 𝑣𝑣𝑐𝑐𝑣𝑣

Where
𝑣𝑣 is the sales volume
𝑝𝑝 is the price
𝑐𝑐𝑓𝑓 is the fixed cost
𝑐𝑐𝑣𝑣 is the variable cost
One important but somewhat unrealistic assumption of this break-even model is that volume, or
demand, is independent of price (i.e., volume remains constant, regardless of the price of the product).
It would be more realistic for the demand to vary as price increased or decreased.
EXAMPLE: For our Western Clothing Company example on Page 5 of 01 Handout 1, let us suppose
that the dependency of demand on price is defined by the following linear function:
𝑣𝑣 = 1,500 − 24.6𝑝𝑝
The linear relationship is shown in Figure 1. The figure illustrates the fact that as price increases,
demand decreases, up to a particular price level (P60.98) that will result in no sales volume.

Figure 1. Linear relationship of volume to price

Source: Introduction to Management Science, 2016, p. 508.

Insert the new relationship of volume into the original profit equation:
𝑍𝑍 = 𝑣𝑣𝑣𝑣 − 𝑐𝑐𝑓𝑓 − 𝑣𝑣𝑐𝑐𝑣𝑣 = (1,500 − 24.6𝑝𝑝)𝑝𝑝 − 𝑐𝑐𝑓𝑓 − (1,500 − 24.6𝑝𝑝)𝑐𝑐𝑣𝑣
= 1,500𝑝𝑝 − 24.6𝑝𝑝2 − 𝑐𝑐𝑓𝑓 − 1,500𝑐𝑐𝑣𝑣 + 24.6𝑝𝑝𝑐𝑐𝑣𝑣

Substituting values for fixed cost (𝑐𝑐𝑓𝑓 = 𝑃𝑃10,000) and variable cost (𝑐𝑐𝑣𝑣 = 𝑃𝑃8) into this new profit
function results in the following equation:

06 Handout 1 *Property of STI

 [email protected] Page 4 of 9
 BM2101

𝑍𝑍 = 1,500𝑝𝑝 − 24.6𝑝𝑝2 − 10,000 − 1,500 (8) + 24.6𝑝𝑝 (8)

= 1,696.8𝑝𝑝 − 24.6𝑝𝑝2 − 22,000
Because of the squared term, this equation for profit is now a nonlinear, or quadratic, function that
relates profit to price, as shown in Figure 2.

Figure 2. The nonlinear profit functions

Source: Introduction to Management Science, 2016, p508

In calculus, the slope of a curve at any point is equal to the derivative of the mathematical function
that defines the curve. The derivative of our profit function is determined as follows:
𝑍𝑍 = 1,696.8𝑝𝑝 − 24.6𝑝𝑝2 − 22,000
𝜕𝜕𝜕𝜕
= 1,696.8 − 49.2𝑝𝑝
𝜕𝜕𝜕𝜕
Given this derivative, the slope of the profit curve at its highest point is defined by the following
relationship:
0 = 1,696.8 − 49.2𝑝𝑝

Now we can solve this relationship for the optimal price, 𝑝𝑝, which will maximize total profit:
0 = 1,696.8 − 49.2𝑝𝑝
49.2 𝑝𝑝 = 1,696.8
1,696.8
𝑝𝑝 =
49.2
= 𝑃𝑃34.49
The optimal volume of denim jeans to produce is computed by substituting this price into our
previously developed linear relationship for volume:

06 Handout 1 *Property of STI

 [email protected] Page 5 of 9
 BM2101

𝑣𝑣 = 1,500 − 24.6𝑝𝑝
= 1,500 − 24.6(34.49)
= 651.6 𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝 𝑜𝑜𝑜𝑜 𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑 𝑗𝑗𝑗𝑗𝑗𝑗𝑗𝑗𝑗𝑗
The maximum total profit is computed as follows:
𝑍𝑍 = 𝑃𝑃1,696.8𝑝𝑝 − 24.6𝑝𝑝2 − 22,000
= 𝑃𝑃1,696.8 (34.49) − 24.6 (34.49)2 − 22,000
= 𝑃𝑃7,259.45
The maximum profit, optimal price, and optimal volume are shown graphically in Figure 3.

Figure 3. Maximum profit, optimal price, and optimal volume

Source: Introduction to Management Science, 2016, p510

An important concept we have yet to mention is that by extending the break-even model this way, we
have converted it into an optimization model. In other words, we are now able to maximize an objective
function (profit) by determining the optimal value of a variable (price). This is exactly what we did in
linear programming when we determined the values of decision variables that optimized an objective
function. The use of calculus to find optimal values for variables is often referred to as classical
optimization.

06 Handout 1 *Property of STI

 [email protected] Page 6 of 9
 BM2101

Constrained Optimization
In the preceding section, the profit analysis model was developed as an extension of the break- even
model. Recall that the total profit function was
𝑍𝑍 = 𝑣𝑣𝑣𝑣 − 𝑐𝑐𝑓𝑓 − 𝑣𝑣𝑐𝑐𝑣𝑣

Where
𝑣𝑣 is the sales volume
𝑝𝑝 is the price
𝑐𝑐𝑓𝑓 is the fixed cost
𝑐𝑐𝑣𝑣 is the variable cost
and the demand function (i.e., volume as a function of price) was
𝑣𝑣 = 1,500 − 24.6𝑝𝑝
By substituting this demand function into our total profit equation, we developed a nonlinear function:
𝑍𝑍 = 1,500𝑝𝑝 − 24.6𝑝𝑝2 − 𝑐𝑐𝑓𝑓 − 1,500𝑐𝑐𝑣𝑣 + 24.6𝑝𝑝𝑐𝑐𝑣𝑣

Then, by substituting values for 𝑐𝑐𝑓𝑓 (P10,000) and 𝑐𝑐𝑣𝑣 (P8) into this function, we obtained

𝑍𝑍 = 1,696.8𝑝𝑝 − 24.6𝑝𝑝2 − 22,000

We then differentiated this function, set it equal to zero, and solved for the value of 𝑝𝑝 (P34.49), which
corresponded to the maximum point on the profit curve (where the slope equaled zero).
This type of model is referred to as an unconstrained optimization model. It consists of a single
nonlinear objective function and no constraints. If we add one or more constraints to this model, it
becomes a constrained optimization model. A constrained optimization model is more commonly
referred to as a nonlinear programming model. The reason this type of model is designated as a
form of mathematical programming is because all types of mathematical programming models are
actually constrained optimization models. That is, they all have the general form of an objective
function that is subject to one or more constraints. Linear programming has an objective function and
constraints that happen to be linear. A nonlinear programming model has the same general form as
a linear programming model, except that the objective function and/or the constraint(s) are nonlinear.
Nonlinear programming differs from linear programming, however, in one other critical aspect: The
solution of nonlinear programming problems is much more complex. In linear programming a
particular procedure is guaranteed to lead to a solution if the problem has been correctly formulated,
whereas in nonlinear programming no guaranteed procedure exists. The reason for this complexity
can be illustrated by a graph of our profit analysis model. Figure 3 shows the nonlinear profit curve
for the example model.
Now we will transform this unconstrained optimization model into a nonlinear programming model by
adding the constraint,
𝑝𝑝 ≤ 𝑃𝑃20
In other words, because of market conditions, we are restricting the price to a maximum of P20. This
constraint results in a feasible solution space, as shown in Figure 4.

06 Handout 1 *Property of STI

 [email protected] Page 7 of 9
 BM2101

Figure 4. A constrained optimization model

Source: Introduction to Management Science, 2016, p511

As in a linear programming problem, the solution is on the boundary of the feasible solution space
formed by the constraint. Thus, in Figure 4, point A is the optimal solution. It corresponds to the
maximum value of the portion of the objective function that is still feasible.
However, the difficulty with nonlinear programming is that the solution is not always on the boundary
of the feasible solution space formed by the constraint. For example, consider the addition of the
following constraint to our original nonlinear objective function:
𝑝𝑝 ≤ 𝑃𝑃40
This constraint also creates a feasible solution space, as shown in Figure 5.

Figure 5. A constrained optimization model with a solution point not on the constraint boundary
Source: Introduction to Management Science, 2016, p.512

06 Handout 1 *Property of STI

 [email protected] Page 8 of 9
 BM2101

But notice in Figure 5 that the solution is no longer on the boundary of the feasible solution space, as
it would be in linear programming. Point C represents a greater profit than point B, and it is also in the
feasible solution space. This means that we cannot simply look at points on the solution space
boundary to find the solution; instead, we must also consider other points on the surface of the
objective function. This greatly complicates the process of finding a solution to a nonlinear
programming problem, and this difficulty is aggravated by an increased number of variables and
constraints and by nonlinear functions of a higher order. You can imagine the difficulties of finding a
solution if you contemplate a model in space that is made up of intersecting cones, ellipses, and
undulating surfaces, as well as planes, and a solution that is not even on the boundary of the solution
space.

References
Taylor III, B. W. (2016). Introduction to Management Science (12th ed.). New York: Pearson
Education Limited.

06 Handout 1 *Property of STI

 [email protected] Page 9 of 9