You are on page 1of 14

Introduction

Academic papersFAn
Facility location decision involves organiza-
An AHP decision model tions seeking to locate, relocate or expand
for facility location their operations. The facility location decision
process encompasses the identification, analy-
selection sis, evaluation and selection among alterna-
tives. Plants, warehouses, retail outlets, termi-
nals, and storage yards are typical facilities to
Jiaqin Yang and
be located. Site selection starts normally with
Huei Lee the recognition of a need for additional capac-
ity. A decision is then made to start the search
for the “best” location.
Facility location decision has drawn
increased attention from both academic and
business communities in the past two
decades. Many large corporations even do
The authors
sufficient site selections to warrant a full-time
Jiaqin Yang is Associate Professor in the Department of executive for these functions. Facility location
Management, University of North Dakota, Grand Forks, problems have attracted researchers with
North Dakota, USA. diverse backgrounds such as economics,
Huei Lee is Associate Professor in the Department of industrial engineering, and geography (Ghosh
Management and Marketing, Lamar University, Beaumont, and Harche, 1993).
Texas, USA. It has been well recognized that facility
location selection has important strategic
Abstract implications for the operations to be located,
Presents an AHP (analytical hierarchy process) decision because a location decision normally will
model for facility location selection from the view of involve long-term commitment of resources
organizations which contemplate locations of a new and be irreversible in nature. Specifically, the
facility or a relocation of existing facilities. The AHP model location choice for a manufacturing facility
provides a framework to assist managers in analysing may have a significant impact on the firm’s
various location factors, evaluating location site alterna- strategic competitive position in terms of
tives, and making final location selections. The primary operating cost, delivery speed performance,
principle of the AHP model is to match decision-makers’ and firm’s flexibility to compete in the
preferences with location site characteristics. The model marketplace. For example, selecting a pro-
requires that a number of potential sites have been duction facility location that will allow the
proposed. Alternatives are then evaluated and compared company to achieve proximity to suppliers has
under both quantitative and qualitative factors to allow become a critical strategic advantage in the
managers to incorporate managerial experience and recent marketplace, since proximity to suppli-
judgement in the solution process. Uses an example ers is important to JIT (Just-In-Time) pro-
problem to illustrate the solution process. Addresses duction systems and flexible distribution
managerial implications for future research. systems for reduced inventories and improved
delivery performance. Therefore, the final
selection of a facility location must contribute
to the success of corporate strategic plans for
financing, marketing, human resource, and
production objectives (Mount, 1990).
The increased attention to business logis-
tics has also contributed to the interest in
location decision (Ballou and Masters, 1993).
Location decision, inventory management,
transportation system design, and customer
service level requirement are four major
components of logistic planning. These com-
Facilities
Volume 15 · Number 9/10 · September/October 1997 · pp. 241–254 ponents should be planned collectively due to
© MCB University Press · ISSN 0263-2772 their close interrelations. Location decision
241
An AHP decision model for facility location selection Facilities
Jiaqin Yang and Huei Lee Volume 15 · Number 9/10 · September/October 1997 · 241–254

research is also promoted by various public relocation of an existing facility. Both qualita-
and private interest groups which have a high tive and quantitative location factors are
stake behind location selection decisions considered in the model. The model is illus-
(Galbraith, DeNoble and Estavillo, 1990). In trated through an example problem.
the private sector, for instance, location deci- This paper consists of five sections. The
sions concern many stakeholders including next section presents findings from selected
venture capitalists, financial investors, indus- literature on facility location. There then
trial site developers, as well as hotel and con- follows a description of the AHP model devel-
vention centre managers. In contrast, the oped in this paper, and an illustration of the
primary public interest about location deci- solution procedure through an example prob-
sions is from local governments or their eco- lem. Finally, managerial implications are
nomic development agencies, who view discussed.
attracting new business as a key to the eco-
nomic wellbeing of their region and commu-
nities. Literature review
Facility location selection is a typical multi- A selected literature is reviewed in this sec-
criteria decision-making (MDM) problem in tion. First, the recent trends in facility loca-
which managerial preference among perfor- tion selection are discussed. Key location
mance criteria plays a key role in the final factors that have been identified in recent
decision. To assess the decision-maker’s literature are then addressed. Finally, an
preference explicitly with a preference model, overview of reported facility location models
many efforts have been made to develop the and solution procedures are summarized.
theory and methodology for preference
assessment. In the current literature, the most Recent trends in facility location
preferred approaches are multi-attribute selection
utility theory (MAUT) and analytic hierarchy
There are two major trends in recent facility
process (AHP) (Falkner and Benhajla, 1990;
location selection. First, there has been an
Saaty, 1990).
increased interest in locating a new facility
The AHP approach was developed in early
abroad to gain potential competitive edge in a
1970s in response to military contingency
more global marketplace. Second, small to
planning, scarce resources allocation, and the
medium-sized communities, traditionally in a
need for political participation in disarma-
“disadvantageous” position, have been
ment agreements (Saaty, 1980). All these
becoming more attractive to many businesses
problems rely heavily on measurement and
as new facility locations, as the new trans-
tradeoff of intangibles in a multicriteria
portation and communication technologies
process. The AHP is a structured method to
are diminishing the traditional distance barri-
elicit preference opinion from decision mak-
er in location selection consideration.
ers. Its methodological procedure can easily
be incorporated into multiple objective pro-
gramming formulations with interactive
solution process. The AHP approach involves ‘…The consideration of locating
decomposing a complex and unstructured production facilities abroad is
problem into a set of components organized primarily motivated by the fact that in
in a multilevel hierarchic form (Saaty, 1982). the last two decades, traditional trade
A salient feature of the AHP is to quantify barriers have been tumbling down…’
decision makers’ subjective judgements by
assigning corresponding numerical values
based on the relative importance of factors The consideration of locating production
under consideration. A conclusion can be facilities abroad is primarily motivated by the
reached by synthesizing the judgements to fact that in the last two decades, traditional
determine the overall priorities of variables trade barriers have been tumbling down with
(Saaty, 1994a). the creation of multilateral international trade
An AHP location decision model is devel- blocks involving several countries, such as:
oped in this research. The location decision is NAFTA (North American Free Trade Agree-
addressed from the perspective of the organi- ment), EC (European Community), and PR
zations seeking a site for a new facility, or a (Pacific Rim). The suggested benefits from
242
An AHP decision model for facility location selection Facilities
Jiaqin Yang and Huei Lee Volume 15 · Number 9/10 · September/October 1997 · 241–254

locating the operations abroad include: low technologies on facility location consideration
cost land, skilled but relative cheaper labours, have made the facility location selection
closeness to raw materials, early access to become more strategically important and the
potential markets, and favourable trade regu- location decision process become more com-
lations by host nations as localized industries. plex.
Major difficulties from locating a facility
abroad have also been addressed, such as: Location factors
additional costs due to cultural traditions, The suitability of a specific site for proposed
language barriers, business climate differ- facility operations depends largely on what
ences, political pressure, as well as difficulty in location factors are selected and evaluated, as
dealing at a distance with administrative and well as their potential impact on corporate
legal procedures. Locating a facility in a for- objectives and operations. There are a large
eign country is obviously a strategic decision number of location factors that have an influ-
for a firm in balancing possible risks and ence on location decisions. Unimportant
potential benefits and opportunities (Gal- factors, including those which are not sensi-
braith et al., 1990). tive to location site and those that the differ-
ence in degree of factor achievement is
insignificant, will be first identified and elimi-
‘…New technologies including improved nated from the consideration. The location
highway transportation system and factors that have been widely used in industri-
advanced computer and al location research generally can be grouped
into the following categories: Market, Trans-
telecommunication techniques have
portation, Labour, Site considerations, Raw
changed facility location
materials and services, Utilities, Governmen-
consideration…’
tal regulations, and Community environment,
as shown in Table I (Levine, 1991).
Location factors have been also addressed
New technologies including improved high-
based on the uniqueness of industry type,
way transportation system and advanced
facility type, and product life-cycle stage
computer and telecommunication techniques
(HBS, 1989). For example, labour intensive
have changed facility location consideration.
industries will emphasize more on labour-
Improved highway system reduces the trans-
related factors such as labour availability and
portation costs for shipping goods between
rate, while high-tech industries will look at
traditional remote locations and major indus- factors that attract highly skilled labours such
trial distribution centres. New telecommuni- as quality of life concerns. Management also
cation technologies expand business bound- will prioritize factors differently for locating a
aries by increasing the distance over which new manufacturing facility from locating a
labour, product, capital, and everything else distribution centre. In terms of the product
can be moved around. These new technolo- life-cycle stage, during the new product devel-
gies have redefined the notion of time and opment stage, facilities will often be located in
distance in business and thus have changed major R&D centres where facility costs are
the character, scale, and location of business usually high but be justified with the prompt
operations. As a result, the business world is new product introduction. In comparison,
experiencing the “end of geography” (ADR, after reaching the growth stage, facilities are
1993) implying that many businesses are no preferred to locate close to major markets to
longer tied to their original locations, once minimize transportation costs and delivery
considered too remote for potential develop- time.
ment. With the new technologies, for exam- The relative importance of location factors
ple, distance is no longer a major concern for is also changing as the decision process stages
delays in customer service, information shar- proceed (Haigh, 1990). In the early stage of
ing, and financial transactions. So firms can identifying preferred geographical areas, only
open new branches far from their base to take a few priority items are considered to identify
advantage of proximity to markets and distri- those regions that satisfy most important
bution channels. criteria, such as the availability of labour or
In summary, the increased attractiveness to proximity to market. When reaching the stage
locate abroad and the impact of new of selecting specific locations, site-specific
243
An AHP decision model for facility location selection Facilities
Jiaqin Yang and Huei Lee Volume 15 · Number 9/10 · September/October 1997 · 241–254

Table I Major facility location factors Table II A general ranking of location factors

1 Access to markets/distribution centres A Pivotal


Cost of serving markets Worker productivity
Trends in sales by areas Receptivity to business and industry
Ability to penetrate local market by plant presence Market access
Skills/technical/professional workers
2 Access to supplies/resources
Transportation access
Transportation costs
Trends in supplier by area B Vital
Living amenities
3 Community/government access Market growth potential
Ambience/cost of living Preference of company executive
Co-operation with established local industry Industrial building available
Community pride Water supply
Housing/churches Unskilled/semi-skilled workers
Schools and colleges
C Important
4 Competitive considerations Proximity to services
Location of competitors Energy supplies
Likely reaction to the new site Attitude towards business and industry taxes
5 Environmental factors Energy costs
Community attitude Raw materials/supplies accessibility
Waste water facilities
State/local governmental regulations
D Secondary
6 Labour
Cost of property and construction
Prevailing wage rates
Personal income tax structure
Extent and militancy of unions in the area
Attitudes on environmental control
Productivity
Financial health of region
Availability Financial incentives
Skill levels available Proximity to other company facilities
7 Taxes and financing
State income tax/local property and income taxes
Unemployment and compensation premiums factors that impact on the ease of doing busi-
Tax incentive concessions ness in a particular area, such as labour atti-
Industrial pollution control revenue bonds tude, business climate, and quality of life. Such
factors cannot readily be expressed in numeri-
8 Transportation cal values and evaluated by quantitative mod-
Trucking service els. Location selection problems become more
Rail service complex when qualitative factors are consid-
Air freight service ered, because subjective judgements must then
9 Utilities services be adopted.
Quality and price of water and sewerage One major concern in facility location
Availability and price of electric and natural gas selection is to allow adequate time to investi-
Quality of police, fire, medical services gate a variety of potential sites. The collection
of information on potential sites and location-
al factors can be a huge burden on a compa-
factors such as land costs and access to roads ny’s resources. Research is thus critical to
may dominate. In the final stage of evaluating arrive at a sound choice in location decision.
a few selected communities, some qualitative Companies should be aware of various infor-
factors may tip the scales in favour of one mation sources for different location factors.
community over another. A general ranking of
location factors is presented in Table II. Location solution process and decision
Location factors, based on the measurabili- models
ty, can also be addressed from quantitative and Facility location decision involves the deter-
qualitative categories. The quantitative type mination of a geographic site to locate a firm’s
can be measured in numerical values, such as operations. Firms typically approach the
the cost of lands and tax incentives. The quali- location decision as a two-stage process
tative type incorporates non-quantifiable (SME, 1993). The first stage establishes the
244
An AHP decision model for facility location selection Facilities
Jiaqin Yang and Huei Lee Volume 15 · Number 9/10 · September/October 1997 · 241–254

site requirements and the relative importance decision problems. In particular, the current
of these requirements. In the second stage, location decision model literature lacks a
these criteria are applied to sites under con- framework that can present and organize all
sideration and unqualified locations are elimi- related location factors (both quantitative and
nated until the most suitable one is found. qualitative) into a solution structure and
Feasibility study defines objectives and selects analyse these location factors with specific site
a solution methodology. Macro-economic requirements. That is the primary motivation
analysis consists of a preliminary screening to of this research.
identify geographical areas that warrant fur-
ther investigation. In evaluation stage, specific An AHP location decision model
sites within a selected geographical area are
examined to narrow choices. Site selection is The analytic hierarchy process (AHP) has
the final step, involving a detailed analysis of been proposed in recent literature as an
the selected sites based on a thorough investi- emerging solution approach to large, dynam-
gation of all related location factors. ic, and complex real world multi-criteria
decision-making problems. The AHP has
been used in a wide variety of complex deci-
‘…Various location decision models have sion-making problems, such as the strategic
been developed to conduct planning of organizational resources (Saaty,
comparative evaluation of potential 1990), the evaluation of strategic alternatives
location sites through examining (Tavana and Banerjee, 1995), or the justifica-
tion of new manufacturing technology
related location factors and site
(Albayrakoglu, 1996). An earlier survey
requirements…’
provided over 200 known applications of the
AHP (Zahedi, 1986). Up-to-date successful
applications of the AHP have been reported in
Various location decision models have been
marketing, finance, education, public policy,
developed to conduct comparative evaluation
economics, medicine, and sports (Saaty,
of potential location sites through examining
1990, 1994a, 1994b). The AHP has been
related location factors and site requirements.
applied in a variety of formats such as: the
In addition to traditional mathematical
design tool for large-scale systems or
models (e.g. mixed integer programming and
composite ratio scales (Weiss and Rao, 1987),
decision analysis), new solution methodolo-
the instrument for pairwise comparison in the
gies have been applied in recent facility loca- application of artificial neural networks (Sun,
tion research, such as simulation models, Stam, and Steuer, 1996), or the primary
expert systems, and neural network tech- structure of decision-support systems (Tavana
niques (Benjamin, Chi, Gaber and Riordan, and Banerjee, 1995). As a convenient
1995; Ghosh and Harche, 1993; Kathawala methodology, the AHP approach has been
and Gholamnezhad, 1987). A detailed review used to determine goal priorities and objective
on recent solution procedures for facility function weights in a goal LP formulation
location problem has been summarized in (Gass, 1986), to examine the weighting vector
Houshyar and White (1997). within the reference framework and search
There are also limitations in the current reference direction in a visual interactive
location decision model literature. First, system (Korhonen, 1987), and to identify
application-oriented solution procedures that objective coefficient and parameter values in
are practical to handle large real-world prob- multiple-objective LP problems (Korhonen
lems are rare due to the complexity involved. and Wallenius, 1990). In addition, the soft-
Second, qualitative location factors are basi- ware package that implements the original
cally not incorporated in the majority of AHP (Saaty, 1980, 1988), Expert Choice, has
reported location decision models. But in been promoted in recent publications (Expert
many cases, qualitative factors are the primary Choice, 1992).
concerns in the location selection. Finally, Selecting a facility location, in most real
most reported location decision models are world applications, is a complex process
deterministic in nature and thus lack the involving multiple facilities (to be located),
dynamic solution capability necessary to deal multiple sites (to be considered), multiple
with rapid changes in today’s location criteria (to be evaluated), and multiple stages
245
An AHP decision model for facility location selection Facilities
Jiaqin Yang and Huei Lee Volume 15 · Number 9/10 · September/October 1997 · 241–254

(to be conducted). In addition, location reached by overall composite weights. A


decisions are conducted in a dynamic fashion. detailed description of the AHP method-
That is, key factors may change over time due ology can be found in Saaty (1994a).
to unexpected events (e.g. an emerging energy
Some concerns have been raised regarding the
crisis may cause fuel and power costs to
AHP for the arbitrary ranking occurred when
assume greater importance over other fac-
two or more alternatives have similar or quasi-
tors). Furthermore, under multiple criteria
similar characteristics, or the rank reversal
comparison, more often no single location site
caused by the addition or deletion of alterna-
could dominate all other alternatives under
tives (Dyer, 1990; Perez, 1995; Tversky and
consideration in a clear-cut fashion. Instead,
Simonson, 1993). These undesirable effects,
each site candidate may have an appealing
however, do not invalidate the AHP method
advantage in its favour; as such the final selec-
(Harker and Vargas, 1987; Saaty and Vargas,
tion will be the result of a compromise, other
1993). In fact, almost all ordinal aggregation
than an “optimal” decision. In this regard,
methods exhibit rank reversal (Perez, 1995).
facility location decision problem is a particu-
It has been shown that the rank reversal will
larly suitable application of the AHP solution
not be a troublesome issue in real world appli-
approach. Carlsson and Walden (1995) report
cations because it is very rare to encounter
an example of applying the AHP model
two alternatives with very similar characteris-
(Saaty, 1980) in the location selection process
tics, or special precautions (e.g. grouping
for a grand new ice-hockey arena in a river
similar alternatives) can easily be taken to
valley of Finland. Owing to the conflicting
avoid any rank reversal (Saaty, 1994b). Saaty
political pressure from the parties involved in
(1994a) further argued that the AHP avoids
the decision, the AHP model was used in a
the rank reversal by dealing directly with
group decision-making environment and the
paired comparisons of the priority of impor-
third ranked site (not the “optimal” one) was
tance, preference, or likelihood of pairs of
actually chosen under final political compro-
elements in the decision hierarchy, that is just
mise.
the natural method that people follow in
making actual decisions. Specifically, three
steps have been identified in terms of ranking
‘…Pairwise comparison is a key step in
preservation:
the AHP location model to determine
(1) allow rank to reverse by using the distrib-
priority weights of location factors and
utive model of the relative measurement
provide a rating for site candidates
approach;
based on qualitative factors…’
(2) preserve rank by using the ideal mode (in
case of irrelevant alternatives); and
(3) preserve rank absolutely by using the
Three major steps in the AHP solution
absolute measurement mode (Saaty,
approach are (Saaty, 1994b):
1994a).
(1) Problem decomposition – the problem is
decomposed into elements (which are Pairwise comparison is a key step in the AHP
grouped on different levels to form a location model to determine priority weights
chain of hierarchy) and each element is of location factors and provide a rating for site
further decomposed into sub-elements candidates based on qualitative factors. The
until the lowest level of the hierarchy. procedure focuses on two factors at a time
(2) Comparative analysis – the relative impor- and their relation to each other, so decision
tance of each element at a particular level makers will be more comfortable to offer
will be measured by a procedure of pair- relative than absolute preference information.
wise comparison. The decision makers The relative importance of each factor is rated
provide numerical values for the priority by a measurement scale to provide numerical
of each element using a rating scale. judgements corresponding to verbal judge-
(3) Synthesis of priorities – the priority weights ments. The instrument used in this research is
of elements at each level will be computed a discrete scale, from 1 to 10 with 1 represent-
using eigenvector or least square analysis. ing the equal importance of two factors and
The process is repeated for each level of 10 being the highest possible importance of
the hierarchy until a decision is finally one factor over another (Saaty, 1994a). The
246
An AHP decision model for facility location selection Facilities
Jiaqin Yang and Huei Lee Volume 15 · Number 9/10 · September/October 1997 · 241–254

priority weights are then presented in a com- (4) clarifying the necessary participants
parison matrix. associated with the problem (Saaty,
To determine priority weights is an impor- 1990).
tant step in pairwise comparison. Three
Organizing objectives and attributes in a
methods have been suggested in the literature
hierarchy serves two purposes:
to calculate such priority weights, including
(1) provides an overall view of the complex
normalized eigenvalues (EM), logarithmic relationship inherent in the situation; and
least squares, and least squares methods. (2) helps decision makers assess whether the
While it has proved that the suggested three issues in each level are of the same order
methods will generate identical solutions in of magnitude, so homogeneity in compar-
terms of result consistency, the EM is recom- isons is preserved.
mended when the data are not entirely consis-
tent (Saaty and Vargas, 1984). Approximation The priority weights of structured location
of the eigenvector can also be used, such as factors are then determined through pairwise
using the geometric mean (Saaty, 1988). comparison to reflect the judgements and
The solution process of the AHP location relative preferences of different stakeholders.
decision model is depicted in Figure 1. With a Not surprisingly, the priority weights may
vary from stakeholder to stakeholder. When
number of potential sites proposed for a
there are several levels of factors and subfac-
facility location consideration, the AHP
tors, the weight vectors of higher level factors
model starts with identifying pertinent loca-
are first computed. The weight of the corre-
tion factors. These factors are then structured
sponding higher level element is then used to
into a hierarchy descending from an overall
weight the factors at the lower level in the
objective to various criteria and subcriteria in
hierarchy (composite weight). The procedure
successive levels.
is repeated by moving downward along the
Important guidelines for selecting perti-
hierarchy, computing the weight of each
nent factors and constructing the hierarchy
element at a particular level and using these to
structure have been suggested:
determine composite weights for succeeding
(1) representing the problem as thoroughly as
levels. When multiple decision makers are
possible, but not so thoroughly as to lose involved in developing the priority weights,
sensitivity to change in the elements; achieving consensus may be difficult. Weight
(2) considering the environment surrounding analysis can be used to assess the extent of
the problem; differences and the potential impact on final
(3) identifying the issues or attributes that decision. For instance, the means and vari-
contribute to the solution; and ances can be calculated and the significance of
the differences among sets of weights can be
Figure 1 Solution process of the AHP location model statistically tested.
To conduct a comparative analysis, two
Step 0: Facility location project data sets:
(facility need identified and justified) (1) the ranks of each location factor by the
decision makers; and
Step 1: Identify pertinent facility
location factors (2) the scores of each site for each location
factor;
Step 2: Develop priority weights will be collected for all candidate sites. Each
candidate site will be rated according the
Step 3: Collect data and rank each structured location factors (quantitative and
potential location qualitative). Quantitative factors are mea-
sured by their corresponding values (i.e.
Step 4: Analyse comparative results dollars) while the qualitative factors will be
measured by the (1 to 10) rating scale instru-
Step 5: Identify preferred site(s) ment. The sites are first compared and priori-
tized based on each factor of the lowest level
in the hierarchy. Qualitative analysis is
Step 6: Final recommendations conducted based on the pairwise comparison
of all the sites relative to each factor and
247
An AHP decision model for facility location selection Facilities
Jiaqin Yang and Huei Lee Volume 15 · Number 9/10 · September/October 1997 · 241–254

subfactor. The numerical rating values of (2) Decision makers involved in the solution
each factor are normalized considering all process should have strong insights or
other ratings of the factors at the same level of preferences about one or more geographi-
the hierarchy. The ratings of qualitative fac- cal regions prior to the selection process,
tors are the eigenvalues of the pairwise com- in order to select a reasonable number of
parison matrix. The results of both quantita- site candidates.
tive and qualitative analysis will be combined (3) The decision makers have a good under-
for each factor at the lowest possible level in standing of the operations and are knowl-
the hierarchy. The priority weights of each edgeable about the location variables so
factor are the eigenvalues in the correspond- that their evaluation will reflect the prefer-
ing eigenvector of each matrix. This eigenvec- ence of the company as a whole.
tor is weighted with the weight of the higher (4) The decision makers will provide their
level element which was used as the criterion managerial judgements and specialized
in making the pairwise comparison. The knowledge as the input to be integrated
procedure is repeated by moving downwards into the solution process.
along the hierarchy, computing the weights of
each element at every level and using these to An example problem
determine composite weights for succeeding
levels. If the factors at a particular level do not An example problem is used to illustrate the
have any subfactors, their priorities remain AHP location decision model. Assuming three
unchanged in the next level of the hierarchy. potential sites (Sites 1, 2, and 3) have been
The overall priority scores for each location proposed for location consideration, four key
candidate are the sum of individual products location factors that are most pertinent to
facility location selection problems (Levine,
of rating scores by the corresponding priority
1991) are selected in the example problem.
weight for each location factor from the lowest
Each key factor is then disaggregated into
level in the hierarchy.
three major subfactors, which are further
described in more detailed characteristics that
apply to each factor. The four key factors and
‘…The composite indices of the previous
12 subfactors (two levels) are then structured
matrix represent the consistency of the
into a decision hierarchy, as shown in Figure 2
data in the matrix with regard to the
and Table III. Table III also indicates if the
eigenvalues…’
corresponding factor is quantifiable or not.
For example, Factor A1 is the Market growth
potential and can be measured by the expected
The consistency of the data may also be inves- percentage of market growth over the next two
tigated during the analysis. The composite or three year period, considered as a quantita-
indices of the previous matrix represent the tive measure. Among 12 subfactors, five are
consistency of the data in the matrix with quantifiable but another seven are not. (Note:
regard to the eigenvalues. In the final step of the degree of detail and the number of levels
the proposed AHP location decision model, can certainly vary with the complexity of the
the site that has the highest overall priority decision, the amount of available resources,
score is identified as the preferred site. The and the desire of decision makers.)
location decision process ends when the Two data sets: (1) location factor data of
decision makers provide their final recom- each site (both quantitative and qualitative);
mendations for the most suitable location(s) and (2) decision makers’ ratings on each site;
with the analysis results. are needed in the AHP location model. For
Four key assumptions underlying the demonstration purposes, simulated data are
application of the proposed AHP location developed from several actual location deci-
decision model are: sion problems (Rose and Yang, 1996). The 12
(1) The need to search for a facility location location factors described in Table III are
has been justified. The selection task is used as decision criteria. The priority weights
complex enough to require consideration of each location factor are computed by using
of the model, and that sufficient resources pairwise comparison between each factor at a
are available to conduct a thorough analy- particular level (for the detailed results, see
sis. Appendix). Data analysis and matrix
248
An AHP decision model for facility location selection Facilities
Jiaqin Yang and Huei Lee Volume 15 · Number 9/10 · September/October 1997 · 241–254

Figure 2 Hierarchical representation of location factors composite weights computation. The overall
priority score for each location candidate is
Market growth potential computed by multiplying the rating score of
Market Proximity to market
the site concerning the given factor by the
priority weight of corresponding factor and
Proximity to raw materials summarizing these products. As shown in
Table V, Site 2 is the preferred location
Land transportation because it has the highest score (0.7559)
among the three candidate sites. Site 1 is the
Transportation Water transportation
next best location with the next highest score
Air transportation (0.4834).
Location
site
Cost of labour
Table IV Composite priorities of factors and subfactors
Labour Availability of skilled workers
A B C D Composite
Availability of semi-skilled workers Factor 0.0639 0.3817 0.1427 0.4115 priority
A 0.0639 0.0639
Housing A1 0.1842 0.0117
A2 0.4826 0.0308
Community Education
A3 0.3331 0.0213
Business climate B 0.3817 0.3817
B1 0.3748 0.1430
B2 0.4493 0.1715
Table III Location factors and subfactors B3 0.1758 0.0671
Factor Quantifiable C 0.1427 0.1427
A Market C1 0.1566 0.0223
A1 – Market growth potential Y C2 0.4783 0.0682
A2 – Proximity to market N C3 0.3650 0.0521
A3 – Proximity to raw materials N D 0.4115 0.4115
B Transportation D1 0.9502 0.9029
B1 – Land transportation N D2 0.0175 0.0166
B2 – Water transportation N D3 0.0322 0.0305
B3 – Air transportation N
C Labour Table V Overall rating of three sites
C1 – Cost of labour Y Factor Priority
C2 – Availability of skilled labour Y (j ) (wj ) Site 1 Site 2 Site 3
C3 – Availability of semi-skilled labour Y
A
D Community
A1 0.0117 0.123 0.614 0.263
D1 – Housing Y
A2 0.0308 0.4783 0.3650 0.1566
D2 – Education N
A3 0.0213 0.6707 0.2855 0.0437
D3 – Business climate N
B
B1 0.1430 0.1210 0.8681 0.0108
calculation are conducted with a mathemati- B2 0.1715 0.8147 0.0175 0.1676
cal software specialized in matrix computa- B3 0.0671 0.0171 0.0628 0.9199
tion, called GAUSSIAN. It provides the C
eigenvalues of a matrix. (Note: The software C1 0.0223 0.423 0.332 0.246
package of the AHP – Expert Choice (Expert C2 0.0682 0.606 0.151 0.242
Choice, 1992), is certainly recommended for C3 0.0521 0.315 0.209 0.475
D
real world large-sized problems). Composite
D1 0.9029 0.230 0.615 0.154
weights of location factors are computed by
D2 0.0166 0.1563 0.6338 0.2098
multiplying the weight of the subfactor by the
D3 0.0305 0.5691 0.1864 0.2443
weight of the factor directly above in the
Score: = wj.rij 0.4834 0.7559 0.2975
hierarchy. Table IV presents the results of
249
An AHP decision model for facility location selection Facilities
Jiaqin Yang and Huei Lee Volume 15 · Number 9/10 · September/October 1997 · 241–254

Summary and managerial implications For example, with their first input (not
updated) about market, labour, and business
Facility location selection problem can be
climate, Site 1 is the “preferred” location
directed either towards an organization in
with a highest overall score of 0.785, while
search of a site to locate/relocate its facility
Site 2 only scored 0.624. After updated
or a community trying to attract new busi-
information about new market growth esti-
nesses to its area. This paper presents an
mates and the availability of skilled labour
AHP location decision model for the organi-
and changes in the business climate prefer-
zations seeking a site for a new facility, or a
ences among the three sites, however, Site 1
relocation of an existing facility. The solution
is replaced by Site 2 as the most preferred
procedure is illustrated through a simplified
“location” with the new highest score of
example problem. The result of the example
0.7559 (as shown in Table V) compared to
problem demonstrates that the proposed
Site 1 score of 0.4834. To test the model, the
AHP model can provide a framework to
firm’s management has attempted to change
assist decision makers in analysing various
the input, both quantitative values (i.e.
location factors, evaluating location site
labour cost) and subjective qualitative rat-
alternatives, and making final location selec-
ings (e.g. business climate), based on the
tions. Through organizing all location fac-
sensitivity analysis of market growth poten-
tors into a chain of hierarchy from an overall tial and managerial priority shifting among
objective to various criteria and subcriteria factors under consideration. Each time, the
in successive levels, the AHP location deci- AHP model is able to react accordingly in
sion model offers the flexibility to match the either changing its final site recommendation
preferences of the majority of stakeholders (or not change). Based on the feedback from
involved in the decision with specific site this firm, the managerial implications from
characteristics, as well as to incorporate the proposed AHP location decision model
decision makers’ managerial experience and are summarized below. (Note: The detailed
judgement. results from this firm’s application will be
presented in another report later.)
First, the AHP location decision model
‘…The primary motivation is to develop will assist managers to identify all informa-
a decision model that will stratify tion sources for required input data. Under
normally complex and numerous an organized hierarchical structure, the AHP
locational factors into a hierarchy and model will prepare the managers well in
help decision makers focus on key advance for their information need, collec-
areas of usually large-scaled location tion timing, potential cost, and specific
selection problems…’ sources. The manager of the consulting firm
reported several facility location selection
cases where, due to the lack of data prepara-
This research is initiated by a request from a tion, when the managers asked for some
local business consulting firm for advice on extra information in their location considera-
facility location projects. The primary moti- tion, they were told that the requested infor-
vation is to develop a decision model that mation was not available for various reasons;
will stratify normally complex and numerous such as: the time needed to collect the infor-
locational factors into a hierarchy and help mation was longer than expected; or the
decision makers focus on key areas of usually information currently available was out of
large-scaled location selection problems. date (e.g. three years old) and the updated
The data sets used in the example problem information would not be available until (six
are derived with simplification from the data months later); or the information asked was
sets provided by this local consulting firm either confidential or unreasonably expen-
from their location selection projects. For sive to acquire. Consequently, several site
that specific project, the firm’s management selections had to be made based on subjec-
was very satisfied with the proposed AHP tive estimates, and a few later became regret-
model, for its ability to organize the great table and costly mistakes.
number of location factors and especially its Another managerial implication is in rela-
flexibility to incorporate their managerial tion to a more relevant and systematic evalua-
inputs into the final site selection process. tion about site candidate characteristics and
250
An AHP decision model for facility location selection Facilities
Jiaqin Yang and Huei Lee Volume 15 · Number 9/10 · September/October 1997 · 241–254

location factor requirements. In practical have a stake at the location decisions, such as
location selection problems, a difficult task local governments, economic development
for managers is to evaluate and compare agencies and industrial site developers. If the
consistently those qualitative location factors. attraction of new companies is a goal,
A common complaint (or explanation) from members of these groups should understand
practitioners is that it is unfair or impossible the location decision process, as well as the
to ask an individual to offer a consistent differentiating factors used by an industrial
judgement or comparison on those qualita- organization contemplating location/reloca-
tive location factors (e.g. community atti- tion of its operations. This understanding
tude vs the quality of life) because an will enable a better allocation of efforts and
“absolute” subjective consistency is beyond resources to encourage (discourage) differ-
the nature of human beings. In a practical ent types of industry to locate (from locat-
sense, the proposed AHP location model is ing) in a specific area. That is, the AHP
very attractive to managers because its pair- approach can be adapted to help specific
wise comparison procedure will allow man- parties identifying what their area has to
agers only to offer relative other than offer and the types of industries it can best
absolute preference assessment, one at a support. The business consulting firm
time, on those qualitative factors. The proce- involved in this research reports two cases, in
dure will then scale these relative compar- both of which a small community was able to
isons uniquely to ensure the consistency of win the site selection for a larger manufac-
these values. In fact, a built-in inconsistency turing facility (involving 500+ job opportu-
checking mechanism has been established nities and millions of tax revenue for local
within the AHP to identify all inconsistencies government) from among a large number of
competing communities. The two successes
at very early stages of the solution process
have been mainly attributed to the fact that
(Saaty, 1994a). It is believed that this special
the community had successfully identified
capacity makes the proposed AHP location
the major locational advantage it possessed,
model extremely appealing and practical to
for that specific manufacturing facility,
industrial managers.
which the competition did not, and prepared
their package accordingly.
The methodological approach of organiz-
‘…An excellent application of such a
ing all locational factors into a chain of hier-
relative and systematic evaluation is
archy in the proposed AHP location model
the assessment of environmental
has also received positive confirmation from
factors in location selection
the practice. One advantage of such a hierar-
consideration…’ chical structure is to provide a framework in
which it is critical to seek input about the
factors and subfactors within the hierarchy
An excellent application of such a relative from different levels of managers in the
and systematic evaluation is the assessment organization. Another frequently mentioned
of environmental factors in location selection advantage is that such a hierarchical struc-
consideration. As environmental factors have ture can easily be incorporated into an inter-
become more important in firms’ facility active solution procedure which allows more
location search and most of those environ- active participation of the managers involved
mental factors can only be measured qualita- in the solution process. It has been reported
tively, it has been highly promoted that the that using the AHP procedure to estimate
pairwise comparison procedure in the AHP the initial “relative” weights or preferences
location model will be able to capture the about related locational factors requires very
relative importance for every pair of environ- little training for managers, and the use of
mental factors in a trustworthy manner and software “Expert Choice” can even allow
prioritize those environmental factors from a managers to make pairwise comparisons
more reasonable and consistent basis. (and “what-if ” analysis) on the screen (while
More importantly, perhaps, the applica- complex mathematical manipulations are
tion of the proposed AHP location decision performed by computers) (Saaty, 1994b).
model will also provide significant manageri- The flexibility to meet industrial unique-
al insights to the various interest groups who ness and the high acceptability to managers
251
An AHP decision model for facility location selection Facilities
Jiaqin Yang and Huei Lee Volume 15 · Number 9/10 · September/October 1997 · 241–254

have been addressed as other practical man- variations in input values, it may suggest
agerial implications from the proposed AHP further research to obtain more accurate
location decision model. Its high acceptabili- estimates. Sensitivity analysis can also be
ty can be attributed to the fact that the man- used in assessing the impact of alternative
agers are required to participate actively in scenarios. Another extension is to apply the
the solution process. The flexibility to indus- proposed AHP model within a particular
trial uniqueness comes from the fundamen- industry or within an industry group, as
tal nature of hierarchical structure. Different such; the industry-specific importance
industries (or industry groups) view facility weights could then be developed to reflect
location selection from different perspec- the primary location factors within a specific
tives. While service operations (e.g. a fast- industry.
food chain) will use its location as a competi-
tive weapon, manufacturing operations (e.g.
a food processing plant) will consider its References
location more for transportation cost reduc- ADR (Area Development Report) (1993), Locating in Rural
tion. Similarly, different industries (or differ- Areas: A Growing Option, April, pp. 22-3.
ent industry groups) have different location Albayrakoglu, M.M. (1996), “Justification of new manu-
selection strategies. For example, a commu- facturing technology: a strategic approach using the
nity that has a good work ethic reputation analytical hierarchy process", Production and
will be preferred for a labour-intensive oper- Inventory Management Journal, first quarter,
ations facility, but a location site that is near pp. 71-6.
to a research university or industrial research Ballou, R.H. and Masters, J.M. (1993), “Commercial
centres will certainly be more attractive for a software for locating warehouses and other facili-
ties”, Journal of Business Logistics, Vol. 14 No. 2,
high-tech industrial facility. In short, each
pp. 71-107.
industry (or industry group) may have a
Benjamin, O.C., Chi, S., Gaber, T. and Riordan, C.A. (1995),
unique priority consideration in the facility “Comparing BP and ART II neural network classifiers
location selection process. The application of for facility location”, Computers & Industrial
the AHP location decision model will allow Engineering, Vol. 28 No. 1, pp. 43-50.
industrial managers to structure their Carlsson, C. and Walden, P. (1995), “AHP in political group
“uniqueness” into the priority weights com- decisions: a study in the art of possibilities”, Inter-
putation to reflect the desired unique priori- faces, Vol. 25 No. 4, pp. 14-29.
ty consideration. Dyer, J.S. (1990), “Remarks on the analytic hierarchy
process”, Management Science, Vol. 36 No. 3,
pp. 249-58.
‘…Future research extension could also Expert Choice (1992), Expert Choice, Version 8.0., Expert
examine the effect of changing input Choice, Inc., Pittsburgh, PA.
parameters, such as measurement Falkner, C.H. and Benhajla, S. (1990), “Multi-attribute
decision models in the justification of CIM systems”,
values and importance weights of
The Engineering Economist, Vol. 35 No. 2,
location factors…’
pp. 91-114.
Galbraith, C., DeNoble, A.F. and Estavillo, P. (1990),
“Location criteria and perceptions of regional
Finally, this research project can be extended business climate: a study of Mexican and US small
in several directions. First, extra research electronics firms”, Journal of Business
effort should be conducted to investigate the Management, October, pp. 34-47.
impact of industry-unique location factors. Gass, S.I. (1986), “A process for determining priorities and
Future research extension could also exam- weights for large-scale linear goal programmes”,
ine the effect of changing input parameters, Journal of Operations Research Society, Vol. 37
No. 8, pp. 779-85.
such as measurement values and importance
weights of location factors. As any changes in Ghosh, A. and Harche, F. (1993), “Location-allocation
models in the private sector: progress, problems, and
decision makers’ preferences could affect the
prospects”, Location Science, Vol. 1 No. 1,
desirability of a specific location, another pp. 81-106.
research need is the sensitivity analysis on
Haigh, R. (1990), “Selecting a US plant location: the
the effect of changes in decision makers’ management decision process in foreign compa-
preferences. For example, if the final ranking nies”, Columbia Journal of World Business, Autumn,
of alternatives changes greatly with slight pp. 22-31.
252
An AHP decision model for facility location selection Facilities
Jiaqin Yang and Huei Lee Volume 15 · Number 9/10 · September/October 1997 · 241–254

Harker, P.T. and Vargas, L.G. (1987), “The theory of ratio SME (Society of Manufacturing Engineers) (1993), “Chap-
scale estimation: Saaty’s analytic hierarchy ter 18: Facility Planning”, Handbook of Industrial
process”, Management Science, Vol. 33, Engineering, pp. 18.1-18.8.
pp. 1383-403. Sun, M., Stam, A. and Steuer, R.E. (1996), “Solving multi-
HBS (Harvard Business School) (1989), Note on Facility ple objective programming problems using feed -
Location, Publishing Division, Harvard Business forward artificial neural networks: the interactive
School, Boston, MA. FFANN procedure”, Management Science, Vol. 42
No. 6, pp. 835-49.
Houshyar, A. and White, B. (1997), “Comparison of
solution procedures to the facility location prob- Tavana, M. and Banerjee, S. (1995), “Strategic assessment
lem”, Computers and Industrial Engineering, Vol. 32 model (SAM): a multiple criteria decision support
No. 1, pp. 77-87. system for evaluation of strategic alternatives”,
Decision Science, Vol. 26 No. 1, pp. 119-43.
Kathawala, Y. and Gholamnezhad, H. (1987), “New
Tversky, A. and Simonson, I. (1993), “Context-dependent
approach to facility location decisions”, Internation-
preferences”, Management Science, Vol. 39 No. 10,
al Journal of Systems Science, Vol. 18 No. 2, pp. 389-
pp. 1179-89.
402.
Weiss, E.N. and Rao, V.R. (1987), “AHP design issues for
Korhonen, P. (1987), “The specification of a reference
large-scale systems”, Decision Sciences, Vol. 18 No.
direction using the analytic hierarchy process”,
1, pp. 43-57.
Mathematical Modelling, Vol. 9 Nos. 3-5, pp. 361-8.
Zahedi, F. (1986), “The analytic hierarchy process – a
Korhonen, P. and Wallenius, J. (1990), “Using qualitative survey of the method and its applications”, Inter-
data in multiple objective linear programming”, faces, Vol. 16 No. 4, pp. 96-108.
European Journal of Operational Research, Vol. 48
No. 1, pp. 81-7.
Levine, T.M. (1991), “How site seekers rate communities”, Appendix: Priority weights calculation
Plant Location, pp. 6-10. of example problem
Mount, S.M. (1990), “Strategic facility planning as a
component of the business plan”, Industrial Devel- Table AI Priority weights of location factors
opment Section, Vol. 1, August, pp. 879-82.
a) Overall priority weights of key location factors
Perez, J. (1995), “Some comments on Saaty’s analytic
hierarchy process”, Management Science, Vol. 41 A B C D Priority weight
No. 6, pp. 1091-5.
A 1 2 1/3 1/8 0.0639
Rose, C. and Yang, J. (1996), “A decision model for facility B 0.5 1 1/2 5 0.3817
location”, Working Paper, Management Depart-
C 3 2 1 1/4 0.1427
ment, #POM9601, University of North Dakota.
D 8 1/5 4 1 0/4115
Saaty, T.L. (1980), The Analytic Hierarchy Process,
McGraw-Hill, New York, NY. b) Priority weights of factor A: market
Saaty, T.L. (1982), Decision Making for Leaders: The A1 A2 A3 Priority weight
Analytical Hierarchy Process for Decisions in a
Complex World, Wadsworth, Belmont, MA. A1 1 1/5 7 0.1842
Saaty, T.L. (1988), Multicriteria Decision Making: The
A2 5 1 8 0.4826
Analytical Hierarchy Process. RWS Publications, A3 1/7 1/8 1 0.3331
Pittsburgh, PA. c) Priority weights of factor B: transportation
Saaty, T.L. (1990), “How to make a decision: the analytic
B1 B2 B3 Priority weight
decision process”, European Journal of Operational
Research, Vol. 48, pp. 9-26. B1 1 4 1/3 0.3748
Saaty, T.L. (1994a), “How to make a decision: the analytic B2 1/4 1 7 0.4493
hierarchy process”, Interfaces, Vol. 24 No. 6, B3 3 1/7 1 0.17581
pp. 19-43.
d) Priority weights of factor C: labour
Saaty, T.L. (1994b), “Highlights and critical points in the
theory and application of the analytic hierarchy C1 C2 C3 Priority weight
process”, European Journal of Operational C1 1 1/8 3 0.1566
Research, Vol. 74, pp. 426-47. C2 8 1 1/4 0.4783
Saaty, T.L. and Vargas, L.G. (1984), “Comparison of C3 1/3 4 1 0.3650
eigenvalue, logarithmic least squares, and least
squares methods in estimating ratios", Mathemati- e) Priority weights of factor D: transportation
cal Modelling, Vol. 5, pp. 309-24. D1 D2 D3 Priority weight
Saaty, T.L. and Vargas, L.G. (1993), “Experiments on rank D1 1 5 8 0.9502
preservation and reversal in relative measurement”,
D2 1/5 1 1/2 0.0175
Mathematical and Computer Modelling, Vol. 17
Nos. 4/5, pp. 13-18. D3 1/8 2 1 0.0322

253
An AHP decision model for facility location selection Facilities
Jiaqin Yang and Huei Lee Volume 15 · Number 9/10 · September/October 1997 · 241–254

Table AII Ratings on quantitative factors Table AIII Ratings on qualitative factors

a) A1: market growth potential Site 1 Site 2 Site 3 Ratings


Ti Ratings A2
Factor A1 (100/A1) (Ti / Σ Ti ) Site 1 1 1/4 5 0.4783
Site 1 15 6.66 0.123 Site 2 4 1 6 0.3650
Site 2 3 33.33 0.614 Site 3 1/5 1/6 1 0.1566
Site 3 7 14.28 0.263
A3
b) C1: cost of labour
Site 1 1 3 2 0.6707
Ti Ratings Site 2 1/3 1 5 0.2855
Factor C1 (100/C1) (Ti / Σ Ti) Site 3 1/2 1/5 1 0/0437
Site 1 5.5 18.18 0.423 B1
Site 2 7 14.28 0.332
Site 1 1 1/4 8 0.1210
Site 3 9.5 10.52 0.246
Site 2 4 1 1/3 0.8681
c) C1: availability of skilled labour
Site 3 1/8 3 1 0.0108
Ti Ratings
B2
Factor C2 (100/C2) (Ti / Σ Ti )
Site 1 1 3 5 0.8147
Site 1 200 0.50 0.606
Site 2 1/3 1 1/7 0.0175
Site 2 800 0.125 0.151
Site 3 500 0.20 0.242 Site 3 1/5 7 1 0.1676

d) C3: availability of semi-skilled labour B3


Site 1 1 1/2 1/7 0.0171
Ti Ratings
Site 2 2 1 1/4 0.0628
Factor C3 (100/C3) (Ti / Σ Ti )
Site 3 7 4 1 0.9199
Site 1 1,200 0.083 0.315
Site 2 1,800 0.055 0.209 D2
Site 3 800 0.125 0.475 Site 1 1 1/7 3 0.1563
e) D1: housing Site 2 7 1 1/2 0.6338
Site 3 1/3 2 1 0.2098
Ti Ratings
Factor D1 (100/D1) (Ti / Σ Ti ) D3
Site 1 1 8 1/3 0.5691
Site 1 8 12.50 0.230
Site 2 3 33.33 0.615 Site 2 1/8 1 4 0.1864
Site 3 12 8.33 0.154 Site 3 3 1/5 1 0.2443

254

You might also like