2

CIVIL ENGINEERING DEPARTMENT

COLLEGE OF ENGINEERING, ARCHITECTURE, TECHNOLOGY, AND SCIENCES
JOHN PAUL COLLEGE

CASE STUDY 2
A BEGINNER’S GUIDE TO AHP

PREPARED BY:
MABUNGA, TRYK CIV C.
BSCE – 2B

SUBMITTED TO:
ENGR. MARIANNE S. MENDEJA, RCE
INSTRUCTOR

March 16, 2024

 1
Published in:
Inter’l Journal of Industrial Engineering: Applications and Practice, Vol. 2, No. 1, pp. 35-44,

USING THE ANALYTIC HIERARCHY PROCESS FOR DECISION

MAKING IN ENGINEERING APPLICATIONS: SOME CHALLENGES
Evangelos Triantaphyllou
Department of Industrial and Manufacturing Systems
Engineering Louisiana State University
3128 CEBA Building
Baton Rouge, LA 70803-6409, U.S.A.

Stuart H. Mann
School of Hotel, Restaurant and Recreation
Management The Pennsylvania State
University
201E Mateer Building
University Park, PA 16802,
U.S.A.

In many industrial engineering applications the final decision is based on the evaluation of a number of
alternatives in terms of a number of criteria. This problem may become a very difficult one when the criteria
are expressed in different units or the pertinent data are difficult to be quantified. The Analytic Hierarchy
Process (AHP) is an effective approach in dealing with this kind of decision problems. This paper examines
some of the practical and computational issues involved when the AHP method is used in engineering
applications.

Significance: In many engineering applications the final decision depends on the evaluation of a set of
alternatives in terms of a number of decision criteria. This may be a difficult task and the
Analytic Hierarchy Process seems to provide an effective way for properly quantifying the
pertinent data. However, there are many critical issues that a decision maker needs to be aware
of.

Key words: Multi-Criteria Decision-Making, Analytic Hierarchy Process, Pairwise Comparisons.

(Received August 23, 1994; Accepted in revised form 14

January 1995)

1. INTRODUCTION

The Analytic Hierarchy Process (AHP) is a multi-criteria decision-making approach and was introduced by Saaty
(1977 and 1994). The AHP has attracted the interest of many researchers mainly due to the nice mathematical
properties of the method and the fact that the required input data are rather easy to obtain. The AHP is a
decision support tool which can be used to solve complex decision problems. It uses a multi-level hierarchical
structure of objectives, criteria, subcriteria, and alternatives. The pertinent data are derived by using a set of
pairwise comparisons. These comparisons are used to obtain the weights of importance of the decision criteria,
 4

and the relative performance measures of the alternatives in terms of each individual decision criterion. If the
comparisons are not perfectly consistent, then it provides a mechanism for improving consistency.
Some of the industrial engineering applications of the AHP include its use in integrated manufacturing
(Putrus, 1990), in the evaluation of technology investment decisions (Boucher and McStravic, 1991), in flexible
manufacturing systems (Wabalickis, 1988), layout design (Cambron and Evans, 1991), and also in other
engineering problems (Wang and Raz, 1991).
As an illustrative application consider the case in which one wishes to upgrade the computer system of a
computer integrated manufacturing (CIM) facility. There is a number of different configurations available to
choose from. The different systems are the alternatives. A decision should also consider issues such as: cost,
performance characteristics (i.e.,)
CPU speed, memory capacity, RAM, etc.), availability of software, maintenance, expendability, etc. These
may be some of the decision criteria for this problem. In the above problem we are interested in determining the
best alternative (i. e., computer system). In some other situations, however, one may be interested in
determining the relative importance of all the alternatives under consideration. For instance, if one is interested
in funding a set of competing projects (which now are the alternatives), then the relative importance of these
projects is required (so the budget can be distributed proportionally to their relative importance).
Multi-criteria decision-making (MCDM) plays a critical role in many real life problems. It is not an
exaggeration to argue that almost any local or federal government, industry, or business activity involves, in one
way or the other, the evaluation of a set of alternatives in terms of a set of decision criteria. Very often these
criteria are conflicting with each other. Even more often the pertinent data are very expensive to collect.

2. STRUCTURE OF THE DECISION PROBLEM UNDER CONSIDERATION

The structure of the typical decision problem considered in this paper consists of a number, say M, of alternatives
and a number, say N, of decision criteria. Each alternative can be evaluated in terms of the decision criteria and
the relative importance (or weight) of each criterion can be estimated as well. Let a ij (i= 1,2,3,...,M, and N=
1,2,3,...,N) denote the performance value of the i-th alternative (i.e., Ai) in terms of the j-th criterion (i. e.,
Cj). Also denote as Wj the weight of the criterion Cj. Then, the core of the typical MCDM problem can be
represented by the following decision matrix:

Criterion
C1 C2 C3 ... CN
Alt. W1 W2 W3 ... WN

A1 a11 a12 a13 ... a1N

A2 a21 a22 a23 ... a2N
A3 a31 a32 a33 ... a3N
. . . . . .
. . . . . .
. . . . . .
AM aM1 aM2 aM3 ... aMN

Given the above decision matrix, the decision problem considered in this study is how to determine which is
the best alternative. A slightly different problem is to determine the relative significance of the M alternatives
when they are examined in terms of the N decision criteria combined.
In a simple MCDM situation, all the criteria are expressed in terms of the same unit (e. g., dollars). However,
in many real life MCDM problems different criteria may be expressed in different dimensions. Examples of such
dimensions include dollar figures, weight, time, political impact, environmental impact, etc. It is this issue of
multiple dimensions which makes the typical MCDM problem to be a complex one and the AHP, or its variants,
may offer a great assistance in solving this type of problems.

3. THE ANALYTIC HIERARCHY PROCESS

The AHP and its use of pairwise comparisons has inspired the creation of many other decision-making methods.
Besides its wide acceptance, it also created some considerable criticism; both for theoretical and for practical
reasons. Since the early days it became apparent that there are some problems with the way pairwise
comparisons are used and the way the AHP evaluates alternatives. First, Belton and Gear (1983) observed that
the AHP may reverse the ranking of the alternatives when an alternative identical to one of the already existing
alternatives is introduced. In order to overcome this deficiency, Belton and Gear proposed that each column of
the AHP decision matrix to be divided by the maximum entry of that column. Thus, they introduced a variant of
the original AHP, called the revised-AHP. Later, Saaty (1994) accepted the previous variant of the AHP and
now it is called the Ideal Mode AHP. Besides the revised-AHP, other authors also introduced other variants of
the original AHP. However, the AHP (in the original or in the ideal mode) is the most widely accepted method
and is considered by many as the most reliable MCDM method.
The fact that rank reversal also occurs in the AHP when near copies are considered, has also been studied by
Dyer and Wendell (1985). Saaty (1983a and 1987) provided some axioms and guidelines on how close a near
copy can be to an original alternative without causing a rank reversal. He suggested that the decision maker has
to eliminate alternatives from consideration that score within 10 percent of another alternative. This
recommendation was later sharply criticized by Dyer
 6

(1990). The first step in the AHP is the estimation of the pertinent data. That is, the estimation of the aij and Wj values of
the decision matrix. This is described in the next sub-section.

Table 1: Scale of Relative Importances (according to Saaty (1980))

Intensity of
Importance Definition Explanation
1 Equal importance Two activities contribute
equally to the objective
3 Weak importance of one Experience and judgment
over another slightly favor one
activity over another
5 Essential or strong Experience and judgment
importance strongly favor one
activity over another
7 Demonstrated An activity is strongly
importance favored and its dominance
demonstrated in practice
9 Absolute importance The evidence favoring one
activity over another is
of the highest possible
order of affirmation
2,4,6,8 Intermediate values When compromise is needed
between the two
adjacent judgments
Reciprocals of If activity i has one
above nonzero of the above nonzero
numbers assigned to it
when compared with
activity j, then j has
the reciprocal value
when compared with i.

3.1. The Use of Pairwise Comparisons

One of the most crucial steps in many decision-making methods is the accurate estimation of the pertinent data. This is
a problem not bound in the AHP method only, but it is crucial in many other methods which need to elicit qualitative
information from the decision-maker. Very often qualitative data cannot be known in terms of absolute values. For
instance, "what is the worth of a specific computer software in terms of a user adaptivity criterion?" Although information
about questions like the previous one are vital in making the correct decision, it is very difficult, if not impossible, to
quantify them correctly. Therefore, many decision-making methods attempt to determine the relative importance, or
weight, of the alternatives in terms of each criterion involved in a given decision-making problem.
An approach based on pairwise comparisons which was proposed by Saaty (1980) has long attracted the interest of many
researchers. Pairwise comparisons are used to determine the relative importance of each alternative in terms of each
criterion. In this approach the decision-maker has to express his opinion about the value of one single pairwise comparison
at a time. Usually, the decision-maker has to choose his answer among 10-17 discrete choices. Each choice is a linguistic
phrase. Some examples of such linguistic phrases are: "A is more important than B", or "A is of the same importance as
B", or "A is a little more important than B", and so on (see also table 1).
The main problem with the pairwise comparisons is how to quantify the linguistic choices selected by the decision
maker during their evaluation. All the methods which use the pairwise comparisons approach eventually express the
qualitative
answers of a decision maker into some numbers which, most of the time, are ratios of integers. A case in which pairwise
comparisons are expressed as differences (instead of ratios) was used to define similarity relations and is described by
Triantaphyllou (1993). The following paragraphs examine the issue of quantifying pairwise comparisons. Since pairwise
comparisons are the keystone of these decision-making processes, correctly quantifying them is the most crucial step in
multi-criteria decision-making methods which use qualitative data.
Pairwise comparisons are quantified by using a scale. Such a scale is an one-to-one mapping between the set of discrete
linguistic choices available to the decision maker and a discrete set of numbers which represent the importance, or weight,
of the previous linguistic choices. The scale proposed by Saaty is depicted in table 1. Other scales have also been proposed
by others. An evaluation of 78 different scales appears in Triantaphyllou et al. (1994). All the alternative scales depart from
some psychological theories and develop the numbers to be used based on these psychological theories.
In 1846 Weber stated his law regarding a stimulus of measurable magnitude. According to his law a change in sensation
is noticed if the stimulus is increased by a constant percentage of the stimulus itself (Saaty, 1980). That is, people are
unable to make choices from an infinite set. For example, people cannot distinguish between two very close values of
importance, say 3.00 and 3.02. Psychological experiments have also shown that individuals cannot simultaneously compare
more than seven objects (plus or minus two) (Miller, 1956). This is the main reasoning used by Saaty to establish 9 as the
upper limit of his scale, 1 as the lower limit and a unit difference between successive scale values.
The values of the pairwise comparisons in the AHP are determined according to the scale introduced by Saaty (1980).
According to this scale, the available values for the pairwise comparisons are members of the set: {9, 8, 7, 6, 5, 4, 3, 2,
1, 1/2, 1/3, 1/4, 1/5, 1/6, 1/7, 1/8, 1/9} (see also table 1).
As an illustrative example consider the following situation. Suppose that in the previous example of selecting the best
computer system, there are three alternative configurations, say A, B, and C. Also, suppose that one of the decision criteria
is hardware expandability (i.e., the flexibility of attaching to the system other related peripheral devices, such as printers,
new memory, etc.). Suppose that system A is much better than system B, and system C is the least desired one as far as
the hardware expandability criterion is concerned. Suppose that following is the judgement matrix when the three
alternative configurations are examined in terms of this criterion.

C1: Hardware A B C
Expandability

A 1 6 8
B 1/ 1 4
6
C 1/ 1/4 1
8

For instance, when system A is compared to system B then the decision-maker has determined that system A is between
to be classified as "essentially more important" and "demonstrated more important" than system B (see also table 1). Thus,
the corresponding comparison assumes the value of 6. A similar interpretation is true for the rest of the entries.
The next step is to extract the relative importances implied by the previous comparisons. That is, how important are the
three alternatives when they are considered in terms of the hardware expandability criterion? Saaty asserts that to answer
this question one has to estimate the right principal eigenvector of the previous matrix. Given a judgment matrix with
pairwise comparisons, the corresponding maximum left eigenvector is approximated by using the geometric mean of each
row. That is, the elements in each row are multiplied with each other and then the n-th root is taken (where n is the number
of elements in the row). Next the numbers are normalized by dividing them with their sum. Hence, for the previous matrix
the corresponding priority vector is: (0.754, 0.181, 0.065).
An evaluation of the eigenvalue approach can be found in (Triantaphyllou and Mann, 1990). An alternative approach
for evaluating the relative priorities from a judgment matrix is based on a least squares formulation and is described in
(Triantaphyllou et al., 1990a and 1990b). One of the most practical issues in the AHP methodology is that it allows for
slightly non-consistent pairwise comparisons. If all the comparisons are perfectly consistent, then the following relation
should always be true for any combination of comparisons taken from the judgment matrix: aij = aik akj.
However, perfect consistency rarely occurs in practice. In the AHP the pairwise comparisons in a judgment matrix are
considered to be adequately consistent if the corresponding consistency ratio (CR) is less than 10% (Saaty, 1980). The CR
coefficient is calculated as follows. First the consistency index (CI) needs to be estimated. This is done by adding the
columns in the judgment matrix and multiply the resulting vector by the vector of priorities (i. e., the approximated
eigenvector) obtained earlier. This yields an approximation of the maximum eigenvalue, denoted by 8max. Then, the CI
value is calculated by using the formula: CI = (8max - n)/(n - 1). Next the consistency ratio CR is obtained by dividing the
 8

CI value by the Random Consistency index (RCI) as given in table 2.

When these approximations are applied to the previous judgment matrix it can be verified that the following are derived:
8max = 3. 136, CI = 0. 068, and CR = 0. 117. If the CR value is greater than 0. 10, then it is a good idea to study the
problem further and re-evaluate the pairwise comparisons (this was not done in the numerical example in this paper).

Table 2: RCI values for different values of n.

n 1 2 3 4 5 6 7 8 9
RCI 0 0 0.58 0.90 1.12 1.24 1.32 1.41 1.45

After the alternatives are compared with each other in terms of each one of the decision criteria and the individual
priority vectors are derived, the synthesis step is taken. The priority vectors become the columns of the decision matrix (not
to be confused with the judgment matrices with the pairwise comparisons). The weights of importance of the criteria are
also determined by using pairwise comparisons. Therefore, if a problem has M alternatives and N criteria, then the decision
maker is required to construct N judgment matrices (one for each criterion) of order MxM and one judgment matrix of
order
NxN (for the N criteria). Finally, given a decision matrix the final priorities, denoted by Ai , of the alternatives in terms
AHP
of all the criteria combined are determined according to the following formula (1).
N
i
A
AHP ' jai wj , for i ' 1, 2, ,3, ,... ... (1)
j ' j
1 M.
Recall that in the case of the ideal mode AHP the columns of the decision matrix are normalized by dividing by the largest
entry in each column. All the above are further illustrated in terms of a numerical example. The numerical data for this
example were adapted from an example given in (Saaty, 1983b).

3.2. A Numerical Example

Suppose that the three alternative computer systems described earlier need to be evaluated in terms of the four decision
criteria: hardware expandability, hardware maintainability, financing available, and user friendly characteristics of the
operating system and related available software. If more criteria are required to be considered, then this example can be
expanded accordingly. Suppose that the following matrices represent the corresponding judgment matrices with the
pairwise comparisons. Note that the corresponding priority vectors (for the individual criteria) and the consistency
coefficients are given as well. The first such matrix is the same as the one analyzed in the previous sub-section and
therefore it is omitted in this section (recall that the priority vectors is: (0. 754 0. 181 0. 065) and 8max = 3.136, CI =
0.068, and CR = 0.117).

C2: Hardware A B C Priority

Maintainability Vector

A 1 7 1/ 0.233
5
B 1/ 1 1/ 0.055
7 8
C 5 8 1 0.713

8max = 3.247, CI = 0.124, and CR = 0.213.

C3: Financing A B C Priority

Available Vector

A 1 8 6 0.745
B 1/ 1 1/ 0.065
 8 4
C 1/ 4 1 0.181
6
 10

8max = 3.130, CI = 0.068, and CR = 0.117.

C4: User A B C Priority

Friendly Vector

A 1 5 4 0.674
B 1/ 1 1/ 0.101
5 3
C 1/ 3 1 0.226
4
8max = 3.086, CI = 0.043, and CR = 0.074.

Finally, the following is the judgment matrix for the case of comparing the importances of the four decision criteria.

The four Criteria C1 C2 C3 C4 Priority

Vector

C1 1 5 3 7 0.553
C2 1/ 1 1/ 5 0.131
5 3
C3 1/ 3 1 6 0.271
3
C4 1/ 1/ 1/ 1 0.045
7 5 6
8max = 4.252, CI = 0.084, and CR = 0.093.

As it was mentioned earlier, the previous priority vectors are used to form the entries of the decision matrix for this
problem. The decision matrix and the resulted final priorities (which are calculated according to formula (1)) are as follows:

Decision Matrix and Solution when the Original AHP is used: Criterion

C1 C2 C3 C4 Final Priority
Alt. (0.553 0.131 0.271 0.045)
A 0.754 0.233 0.745 0.674 0.680
B 0.181 0.055 0.065 0.101 0.130
C 0.065 0.713 0.181 0.226 0.190

Decision Matrix and Solution when the Ideal Mode AHP is used:
Criterion
C1 C2 C3 C4
Alt. (0.553 0.131 0.271 0.045) Final Priority After Normalization

A 1.000 0.327 1.000 1.000 0.912 0.678

B 0.240 0.077 0.087 0.150 0.173 0.129
C 0.086 1.000 0.243 0.335 0.260 0.193

Therefore, the best system is A followed by system C which is followed by system B. Observe that although both the
original AHP and the ideal mode AHP yielded the same ranking for the three alternatives, they assigned different final
priorities for these alternatives. The next section describes some cases in which the original AHP (and in a similar manner
the ideal mode AHP) can exhibit ranking abnormalities for some decision making problems.

4. SOME CASES OF NUMERICAL INSTABILITY WITH THE AHP

In Triantaphyllou and Mann (1989) the four multi-criteria decision-making methods: the AHP, revised AHP (i.e., the ideal
mode AHP), the Weighted Sum Model (WSM) (Fishburn, 1967), and the Weighted Product Model (WPM) (Miller and
Starr, 1969) were examined in terms of two evaluative criteria. That study focused on the last step of any MCDM method
which involves the processing of the final decision matrix. That is, given the weights of relative performance of the
decision criteria, and the performance of the alternatives in terms of each one of the decision criteria, then determine what
is the ranking (or relative priorities) of the alternatives.
As it was shown in Triantaphyllou and Mann (1989), however, these methods can give different answers to the same
problem. Since the truly best alternative is the same regardless of the method chosen, an estimation of the accuracy of
each method is highly desirable. The most difficult problem that arises here is how one can evaluate a multi-dimensional
decision-making method when the true best alternative is not known. Two evaluative criteria were introduced for the above
purpose.
The first evaluative criterion has to do with the premise that a method which is accurate in multi-dimensional
problems should also be accurate in single-dimensional problems. There is no reason for an accurate multi-
dimensional method to fail in giving accurate results in single-dimensional problems, since single-dimensional
problems are special cases of multi-dimensional ones. Because the first method, the WSM, gives the most acceptable
results for the majority of single-dimensional problems, the result of the WSM was used as the standard for evaluating the
other three methods in this context.
The second evaluative criterion considers the premise that a desirable method should not change the indication of the
best alternative when an alternative (not the best) is replaced by another worse alternative (given that the importance of
each criterion remains unchanged).
In the following paragraphs we briefly present some numerical cases in which the original AHP fails when it is examined
in terms of the previous two evaluative criteria. The case of the ideal mode AHP is similar (although different numerical
examples are needed to expose cases of numerical instability), and thus it is omitted for simplicity.

4.1 Testing the original AHP Using the First Evaluative Criterion

Example 1: Suppose that the matrix below depicts the actual values, measured in the same units (for instance, in US
dollars) of three alternatives A1, A2, and A3, in terms of three criteria with the following weights of importance: w1 = 8/13,
w2 = 2/13, and w3 = 3/13. Suppose that these criteria are benefit criteria. That is, the higher the value the better it is.
Problems like this one are very common when one wishes, for instance, to perform an engineering economy study.

Criteria
C1 C2 C3

Alter. ( 8/13 2/13 3/13)

A1 1 9 9
A2 5 2 2
A3 1 5 9

Given the above data it is easy to see that the final scores of the alternatives in terms of the three criteria are 53/13,
50/13, and 45/13, respectively. For instance, the score for the first alternative is: 1(8/13) + 9(2/13) + 9(3/13) = 53/13.
Therefore, alternative A1 is the best one (since it corresponds to the score 53/13). The decision method used above is the
weighted sum model (WSM) which is the most commonly used approach when all the units are the same (as it was
assumed in this example).
When the AHP approach is used, the previous data have to be normalized. If the original AHP is to be used, then the
above data are normalized by dividing each column by the sum of the elements in that column. If the data cannot be
obtained directly, then the method which is based on the pairwise comparisons needs to be employed. Since there are three
criteria, the decision maker needs to construct three matrices with pairwise comparisons of size 3x3 each. The three 3x3
matrices with the pairwise comparisons that correspond to this problem are as follows (perfect consistency in the pairwise
comparisons is assumed in order to block out any effects due to inconsistent comparisons):

Criterion C1 Criterion C2 Criterion C3

 12

1 1/5 1/1 1 9/2 9/5 1 9/2 9/9

5/1 1 5/1 2/9 1 2/5 2/9 1 2/9
1/1 1/5 1 5/9 5/2 1 9/9 9/2 1
Therefore, the MxN (i.e., 3x3) matrix with the relative (i.e., normalized) importances of the alternatives in terms of each
criterion that is used in the final step of the AHP is:

Criteria
C1 C2 C3

Alter. ( 8/13 2/13 3/13)

A1 1/7 9/16 9/20
A2 5/7 2/16 2/20
A3 1/7 5/16 9/20

Applying the last step of the AHP it turns out that the alternative A 2 is the best one (A 2AHP = A*AHP = 0.48). Obviously,
this is in contradiction with the conclusion derived earlier by using the WSM.

4.2 Testing the original AHP Using the Second Evaluative Criterion

Example 2: Suppose that the following is a matrix of the eigenvectors produced by the AHP process. That is, the matrix
contains relative values for the importance of the alternatives instead of the actual values. Assume that the criteria have
weights w1 = 2/7, w2 = 2/7, and w3 = 3/7.

Criteria
C1 C2 C3

Alter. ( 2/7 2/7 3/7 )

A1 9/19 2/12 2/7
A2 5/19 1/12 4/7 = Matrix M1
A3 5/19 9/12 1/7

It can be shown (by multiplying the matrix with the relative importances by the vector with the weights of the 3 criteria
followed by normalization) that the priority vector of the alternatives (according to the original AHP) is (0. 305, 0. 344,
0.351).
Apparently, the best alternative is A3. If in the above problem the alternative A 1 (which is not the best one and was
defined by the relative values (9/19 2/12 2/7), is replaced by A /1 which is worse than the original alternative A 1, then, the
above matrix is modified as follows:

Criteria
C1 C2 C3

Alter. ( 2/7 2/7 3/7 )

A/1 8/18 1/11 1/6
A2 5/18 1/11 4/6 = Matrix M2
A3 5/18 9/11 1/6

Matrix M1 can be considered as the matrix with the relative values obtained from the following three 3x3 matrices with
pairwise comparisons (perfect consistency in the pairwise comparisons is again assumed as before):

Criterion C1 Criterion C2 Criterion C3

 1 9/5 9/5 1 2/1 2/9 1 2/4 2/1
5/9 1 5/5 1/2 1 1/9 4/2 1 4/1
5/9 5/5 1 9/2 9/1 1 1/2 1/4 1

Matrix M2 has been derived from matrix M1 by substituting the alternative A 1 with the lesser A/1 = (8/18 1/11 1/6) <
(9/19 2/12 2/7). That is, instead of 9 it is now 8, instead of 2 it is now 1, and instead of 19 it is now 18.
Similarly, the priority vector for matrix M2 is (0.224, 0.391, 0.385). It is clear that now the best alternative is A 2. The
last statement is, obviously, in contradiction with the original result namely, that the best alternative is A 3. Thus, by
introducing a new worse alternative (different from the best one) it is possible to have a change in the indication of the best
alternative.
In (Triantaphyllou and Mann, 1989) the previous two evaluative criteria were applied on random test problems with the
numbers of decision criteria and alternatives taking the values 3, 5, 7, ..., 21. In those experiments it was found that all
the previous four MCDM methods (i. e., the AHP; both original and in ideal mode, the WSM and the WPM) were
inaccurate. Furthermore, these results were used to form a decision problem in which the four methods themselves were
the alternatives. The decision criteria were derived by considering the previous two evaluative criteria.
Since the best decision making method was not known, and each alternative was a decision making method, we solved
the previous problem by using each one of the rival methods. To our greatest surprise, one method would recommend
another, rival method, as being the best method! This was reported in (Triantaphyllou and Mann, 1989) as a decision
making paradox. However, the final results seemed to suggest that the revised AHP (i. e., the ideal mode AHP) was the
most efficient MCDM method of the ones examined. Finally, a different approach of evaluating the performance of the
original AHP and the ideal mode AHP, by using a continuity assumption, is described by Triantaphyllou and Mann
(1994a). In that treatment it was found that these two methods may yield dramatically inaccurate results (more than 80% of
the time).

5. CONCLUSIONS AND DISCUSSION

The AHP provides a convenient approach for solving complex MCDM problems in engineering. It should be noted that
there is a software package, called Expert Choice (1990), which has significantly contributed to the wide acceptance of the
AHP methodology. However, as this paper demonstrated with some illustrative examples, its use to engineering problems
should be a cautious one. There is sufficient evidence to suggest that the recommendations made the AHP should not be
taken literally. In matter of fact, the closer the final priority values are with each other, the more careful the user should
be.
This is true with any MCDM method. The numerical examples in this paper, along with the extensive research of the
authors in this area (please also see the reference list for more details), strongly suggest that when some alternatives appear
to be very close with each other, then the decision-maker needs to be very cautious. An apparent remedy is to try to
consider additional decision criteria which, hopefully, can assist in drastically discriminating among the alternatives. A
summary of the results of a number of studies on the AHP and pairwise comparisons by the authors can be found in
(Triantaphyllou and Mann, 1994b).
The above observations suggest that MCDM methods should be used as decision support tools and not as the means for
deriving the final answer. To find the truly best solution to a MCDM problem may never be humanly possible. The
conclusions of the solution should be taken lightly and used only as indications to what may be the best answer. Although
the search for finding the best MCDM method may never end, research in this area of decision-making is still critical and
very valuable in many scientific and engineering applications.

6. REFERENCES

1. Belton, V. and Gear, T. (1983). On a Short-coming of Saaty' s Method of Analytic Hierarchies. Omega, 228-230.

2. Boucher, T.O. and McStravic, E.L. (1991). Multi-attribute Evaluation Within a Present Value Framework and
its Relation to the Analytic Hierarchy Process. The Engineering Economist, 37: 55-71.
 14

3. Cambron, K.E. and Evans, G.W., (1991). Layout Design Using the Analytic Hierarchy Process. Computers &
IE, 20: 221-229.

4. Expert Choice Software, (1990). Produced by Expert Choice, Inc., 4922 Ellsworth Avenue, Pittsburgh, PA.

5. Dyer, J.S., and Wendell, R.E. (1985). A Critique of the Analytic Hierarchy Process, Working Paper, 84/85-4-24,
Department of Management, The University of Texas at Austin, Austin, TX.

6. Dyer, J.S. (1990). Remarks on the Analytic Hierarchy Process, Management Science, 36: 249-258.

7. Fishburn, P.C. (1967). Additive Utilities with Incomplete Product Set: Applications to Priorities and Assignments.
American Society of Operations Research (ORSA), Baltimore, MD, U.S.A..

8. Miller, D.W., and Starr, M.K. (1969). Executive Decisions and Operations Research. Prentice-Hall, Inc.,
Englewood Cliffs, NJ, U.S.A.

9. Miller, G. A. (1956). The Magical Number Seven Plus or Minus Two: Some Limits on Our Capacity for
Processing Information. Psychological Review, 63: 81-97.

10. Putrus, P. (1990). Accounting for Intangibles in Integrated Manufacturing (nonfinancial justification based on the
Analytical Hierarchy Process). Information Strategy, 6: 25-30.

11. Saaty, T. L. (1977). A Scaling Method for Priorities in Hierarchical Structures, Journal of Mathematical
Psychology, 15: 57-68.

12. Saaty, T.L. (1980). The Analytic Hierarchy Process. McGraw-Hill International, New York, NY, U.S.A.

13. Saaty, T.L. (1983). Axiomatic Foundations of the Analytic Hierarchy Process, Management Science, 32: 841-855.

14. Saaty, T.L. (1990). An Exposition of the AHP in Reply to the Paper ' Remarks on the Analytic Hierarchy Process'
, Management Science, 36: 259-268.

15. Saaty, T. L. (1994). Fundamentals of Decision Making and Priority Theory with the AHP. RWS Publications,
Pittsburgh, PA, U.S.A.

16. Triantaphyllou, E., and Mann, S. H. (1989). An Examination of the Effectiveness of Four Multi-Dimensional
Decision-Making Methods: A Decision-Making Paradox. International Journal of Decision Support Systems, 5:
303-312.

17. Triantaphyllou, E., Pardalos, P. M., and Mann, S. H. (1990a). A Minimization Approach to Membership
Evaluation in Fuzzy Sets and Error Analysis, Journal of Optimization Theory and Applications, 66: 275-287.

18. Triantaphyllou, E., Pardalos, P.M., and Mann, S.H. (1990b). The Problem of Evaluating Membership Values
in Real World Situations, Operations Research and Artificial Intelligence: The Integration of Problem Solving
Strategies, (D.E. Brown and C.C. White III, Editors), Kluwer Academic Publishers, 197-214.

19. Triantaphyllou, E., and Mann, S. H. (1990). An Evaluation of the Eigenvalue Approach for Determining the
Membership values in Fuzzy Sets. Fuzzy Sets and Systems, 35: 295-301.

20. Triantaphyllou, E., (1993). A Quadratic Programming Approach in Estimating Similarity Relations. IEEE
Transactions on Fuzzy Systems, 1: 138-145.

21. Triantaphyllou, E., Lootsma, F.A., Pardalos, P.M., and Mann, S.H. (1994). On the Evaluation and Application
of Different Scales For Quantifying Pairwise Comparisons in Fuzzy Sets. Multi-Criteria Decision Analysis, 3: 1-
23.
22. Triantaphyllou, E., and Mann, S. H. (1994a). An Evaluation of the AHP and the Revised AHP When the
Eigenvalue Method is Used Under a Continuity Assumption. Computers and Industrial Engineering, 26:
609-618.

23. Triantaphyllou, E., and Mann, S. H. (1994b). Some Critical Issues in Making Decisions With Pairwise
Comparisons, Proceedings of the Third International Symposium on the AHP, George Washington
University, Washington, DC, 225-235.

24. Wabalickis, R.N. (1988). Justification of FMS With the Analytic Hierarchy Process. Journal of
Manufacturing Systems, 17: 175-182.

25. Wang, L., and Raz, T. (1991). Analytic Hierarchy Process Based on Data Flow Problem. Computers &
IE, 20: 355-365.
 16

AHP
A strong framework for multi-criteria decision-making that offers both methodological rigor and
practical feasibility is provided by the Analytic Hierarchy Process (AHP). Its hierarchical form
simplifies complex problems into manageable parts and allows for systematic decision analysis.
Because the approach relies on pairwise comparisons, it may objectively and transparently
evaluate criteria importance and alternative performance on a quantitative basis.The
applicability of AHP in a variety of sectors, including integrated manufacturing, technology
investment, and layout design, is a noteworthy testament to its adaptability across domains.
This flexibility highlights its applicability in handling many decision-making scenarios that arise in
real-life situations.
Furthermore, when making difficult decisions involving the simultaneous consideration of
several factors, such as sponsoring a project or updating a computer system, AHP is quite
helpful. AHP promotes informed decision-making by allowing stakeholders to balance the
relative importance of criteria and alternatives, ensuring that decisions are in line with overall
objectives and limitations.Indeed, techniques like AHP are essential in today's environment
when decision-makers must weigh a growing number of complex and competing factors. For
companies trying to make good, defendable decisions in the face of ambiguity and conflicting
priorities, it is an invaluable tool because of its capacity to manage trade-offs and uncertainties
while taking stakeholder preferences into account. Therefore, the Analytic Hierarchy Process is
a fundamental component of efficient decision support systems and is expected to keep making
important advances in decision science and application.

A clear framework for assessing options across several dimensions is provided by the structure
described for common multi-criteria decision-making (MCDM) situations. A set of criteria is used
to evaluate each choice, and weighting is used to account for the relative importance of each
criterion in the decision-making process. The basis for evaluating and contrasting options is this
organized representation, which is represented by the decision matrix.The fundamental task of
the MCDM problem is to identify the optimal option or evaluate each alternative's relative
importance in light of all relevant criteria. In order to reflect the variety of factors present in real-
life decision scenarios, this calls for a sophisticated understanding of how each alternative
performs across several dimensions.One of the primary issues with MCDM is the wide range of
criteria, which might include things like time, money value, political repercussions,
environmental effect, and environmental impact. This variability makes decision-making more
challenging, necessitating the employment of a scientific approach that can account for various
criteria and increase the relevance of comparisons. Prominent among multi-criteria decision-
making (MCDM) techniques, the Analytic Hierarchy Process (AHP) is praised for its well-
organized methodology and simplicity of use. Nevertheless, throughout time, it has drawn both
praise and criticism for its reliance on pairwise comparisons.

Unquestionably, the Analytic Hierarchy Process (AHP) is a strong paradigm for multi-criteria
decision-making (MCDM) that combines analytical rigor with practical viability. Its hierarchical
form makes it easier to analyze and evaluate complex decision problems by dissecting them
into smaller, more 0manageable parts. Decision-makers can objectively and openly evaluate the
relative merits of criteria and the relative performance of alternatives on a quantitative scale by
using AHP's pairwise comparisons.The adaptability of AHP to many industries and decision-
making contexts is among its most alluring features. Its successful implementation in areas
including technology investment, integrated manufacturing, and layout design highlights its
versatility and efficacy across a range of contexts. AHP is a useful tool for handling a variety of
decision problems that arise in real-world situations because of its versatility. AHP comes be
very handy when making decisions that incorporate several aspects of differing importance.
AHP promotes informed decision-making that is in line with overarching goals and limits by
letting stakeholders assess the relative importance of criteria and options. AHP helps decision-
makers navigate complicated trade-offs and uncertainties, ensuring that choices are well-
informed and defendable, whether they are choosing between competing projects or the best
computer system upgrade.Decision-makers deal with a growing number of complicated and
competing priorities in the fast-paced commercial world of today. Organizations can make wise
decisions in the face of uncertainty and ambiguity by using techniques like Analytic Hierarchy
Process (AHP), which offer an organized way to negotiate these challenges. AHP forms the
basis of effective decision support systems by taking stakeholder preferences into account and
methodically assessing trade-offs.

Nevertheless, despite its advantages, AHP has its detractors, especially when it comes to its
dependence on pairwise comparisons. Opponents have expressed worry about possible
problems such the quantification of qualitative judgments and rank reversal. Variants such as
the Ideal Mode AHP and revised-AHP attempt to solve some of these issues, but further study
and improvement are required to improve the robustness and dependability of the
approach.Even with these difficulties, AHP is still a vital part of decision science and its
applications. Because of its flexibility and organized approach, it is a vital tool for decision-
makers who want to efficiently navigate complex choice landscapes. AHP will be essential in
facilitating strategic and well-informed decision-making processes as businesses continue to
struggle with ever-more complicated scenarios requiring decision-making.

Finally, by fusing methodological rigor with real-world applicability, the Analytic Hierarchy
Process provides a strong framework for multi-criteria decision-making. Its usefulness and
versatility make it a valuable tool in today's decision-making landscape, notwithstanding
criticism, especially with regard to pairwise comparisons. It is anticipated that AHP will continue
to significantly advance decision science and practice with continued investigation and
improvement.