INTRODUCTION
Warehousing has been a part of civilization for thousands of years. Warehousing is the function of storing goods between the time they are produced and the time they are needed [Ref. 1]. In practice, goods are sent to storage points close to the market and are issued to consumers from these points easily and in small amounts when needed. Although warehousing was initially a means of storing foodstuffs, today it is a broad and complex issue [Ref. 3, 4]. For example, there are more than 300,000 large warehouses and 2.5 million employees in the United States alone. The cost of American warehousing is more than 5 percent of the gross national product [Ref. 2].
A warehouse is a distribution factory. The warehousing functions far exceed the mere provision of a building to protect the stored goods from the elements. Furthermore, any warehouse is a complex, constantly evolving center, which must be able to cope with a myriad of expansions and expectations and must do so cost effectively. Adequate space, customer service, favorable traffic connections with suppliers and key markets, easy freeway access, proximity to trains and airports and a qualified work force–these are only some of the factors that a warehousing study must evaluate [Ref. 5].
In order to succeed in certain demand areas, organizations must have a warehouse presence [Ref. 11]. Naturally, capital investment, operating expenses, and customer service are all affected by decisions regarding site and structure [Ref. 5]. As a result, storage should be considered as a resource. Investments in storage facilities should be identified through an initial study and must be followed by a feasibility analysis. The location of warehousing must be studied carefully prior to undertaking the other complex issues inherent in a storage study [Ref. 5].
Before a site is selected, all management levels and business entities must participate in the analysis. Unfortunately, warehouse location projects frequently are understaffed, under-funded, and fail to consider fully the entire distribution network’s current capabilities and future requirements. The process of selecting a site requires a clear understanding of the underlying strategy to be developed and must communicate this research to all the stakeholders involved. Obtaining buy-in from all levels and departments of the organization to ensure a successful analysis and decision is necessary [Ref. 11]
.
Warehouse design begins with determining the best warehouse location. The design process also includes the layout, storage methods, equipment and automated systems, source and nature of the supplies, zones, and order receiving methods [Ref. 4, 6].
Clearly, owing to its complexity, site selection is one of the most challenging and important responsibilities of logistics managers. The task of site selection literally involves art as well as science. Site selection has a major impact on logistics costs and operational efficiency. A warehouse poorly located can deal a costly and even a mortal blow to the life of an organization heavily involved in physical distribution [Ref. 2, 3, 9]. The site selection process begins at the highest strategic (macro) level and descends until a specific real estate parcel is chosen [Ref. 9]. Then the site selection process usually involves weighing priorities, determining the critical features, and eliminating inadequate sites. Since every location has advantages and disadvantages, the final selection of a site is likely to involve some compromises [Ref. 2, 3].
The first step to consider is asking why one is seeking a new warehouse site. The following are four common reasons [Ref. 2, 3]:
• Relocating to an existing warehouse operation is necessary.
• Inventory must be moved to a new location due to expanding responsibilities.
• Additional warehouse space is needed to accommodate a growing inventory.
• Contingency planning requires decentralizing existing warehousing.
Depending on which of the above reasons is the primary motive for seeking a new warehouse, the site search can assume many different forms.
RELOCATING A WAREHOUSE OPERATION
Ackerman [Ref. 3] states that the user should develop a functional outline for the supply center, which includes a review of the existing customer service needs and how best to achieve this service level. Simulating the costs of operating the center may be a beneficial tool to employ. Simulations of any kind should be employed to the highest extent possible. Finally, the user should develop a detailed plan for opening the facility to allow proper lead times and to minimize confusion during the startup phase [Ref. 3].
Although the actual relocation of a warehouse operation is beyond the scope of this research, the basics of the subject are presented in the following paragraphs.
1. To Move or To Stay This question may arise when a more desirable building or location becomes available. Some other changes in transportation or other customer service considerations may also be the reason [Ref. 3]. In the case study of this research, selecting an alternative location and evaluating potential changes in the current location are the main reasons for this question. Chapter IV furnishes necessary data about the case.
2. Initial Planning
Pre-planning is essential when building a warehouse from the ground up [Ref. 10]. Early in the planning stages for the supply center, communications, packaging, transportation, security and perishability of the stored products must be considered [Ref. 3].
3. Good Timing
A target date must be set before the actual relocation. This date may be changed, but for planning purposes a target date is essential. Short-term weather forecasts, seasonal and climate factors and seasonal inventory level variations should be considered in time to set the date. The target date can also be essential because estimating the costs of the relocation may vary depending on the proposed relocation period [Ref. 3].
4. Movement Cost Estimation
After a target date is set, an on-hand inventory assessment of that date is necessary. This estimate can simply be calculated as a percentage of the existing inventory level. An important point is to plan the moving date to coincide with a lower inventory level period or to a period when shipping activity is minimal [Ref. 3].
Additionally, how to load trucks, the cost of transferring the load, and the total time needed to move are other issues that require prior considerations. Second in importance to the question of whether or not to relocate is deciding whether to continue services or whether to suspend operations during the move. Customer service considerations and communications are priorities in a warehouse move [Ref. 3]. The final phase of relocating a warehouse always ends with opening the new warehouse and the supply center according to careful planning.
SITE SELECTION CONSIDERATIONS 1. Project Team and Site Selection Strategy
Once it is determined that a warehouse is needed, the next step is to find the right location. Proper selection of the warehouse location is a highly important and complex task [Ref. 3]. Before beginning the project selection, the organizations must assemble a dedicated facility planning team. This team should be formed of qualified personnel according to the project’s specific requirements. First of all, this team should define the factors that affect the selection of the site, including service requirements, transportation and inventory costs. The team’s site selection strategy usually consists of three levels: macro–analysis, micro–analysis and the specific site selection [Ref. 9].
The macro–analysis determines in what parts of the country the warehouses should be located and defines the significant trade-offs and constraints. The organization can benefit by using various models, including spreadsheet cost calculators, network simulators, and mathematical optimization models [Ref. 9].
The outcome of the macro–analysis should be a set of alternative scenarios, and identification of a region or regions, which will meet the site selection objectives [Ref. 9].
The micro–analysis defines a geographic area of the country to locate the new warehouses. The micro–analysis addresses the trade-offs involved in comparing the potential sites within a geographic region. The project team weighs such regional factors as zoning laws, government investment incentives, accessibility to highways, air and rail transportation, utility services, land values, and climate [Ref. 9].
Specific site analysis identifies the particular location where the facility will be. The selected site must meet the objectives. After selecting the final location, the project team must determine whether to erect a new structure or to adapt an existing one [Ref. 9]. .
2. Selection Consideration and Constraints
Once all the required data has been collected, the actual analysis can take place. The actual site selection task consists of three steps: Setting priorities, determining the
critical features, and selecting the best location among alternatives that may have both advantages and disadvantages. The final selection of a location will probably consist of some trade-offs [Ref. 1, 2].
The first step is identifying the reasons for a new warehouse location. Sometimes the reason is the need to relocate an existing warehouse because operational objectives change or because stock is increasing [Ref. 3]. The reason may even be security or transportation problems. Determining these reasons defines or narrows the search for a new warehouse location.
Selecting a site is usually difficult and finding an outside consultant may be necessary. Real estate offices, sales representatives, railroad companies, utility companies, government agencies, and also engineers are possible consultants. In addition, assigning a project manager to oversee the location selection is wise [Ref. 2, 3]. Stated succinctly, for a decision of this importance, the best method is often accepting the least risk.
Geographical factors can also substantially affect the utility of a warehouse site. Access conditions can be even more critical when a supply center uses different modes of transportation, such as rail, airways, highway, and water. While roadways offer possibilities for the most extensive geographical coverage, the ability to extend waterways and railways is usually constrained by geography. Climate is another important factor, especially for the energy costs. The possibility that climate will disrupt transportation is also a major consideration. If airfreight service is critical to the operation the service record of the nearest airport under consideration should be analyzed [Ref. 2].
Eventually, for a successful site search, carefully defining user requirements is vital. A requirement list is a very helpful guide for location decision. Selecting critical features is the next step in choosing a warehouse location [Ref.2, 3]. For example, if reaching customers rapidly is crucial, then the location should be very close to the customer center. On the other hand, if security is more important, choosing a location far from population centers will probably be a better decision. Finally, preparing a list of
alternative locations and the characteristics of these alternatives are other critical considerations.
Transportation highly influences the selection and success of warehouses and is an essential part of the warehousing concept: Goods must be brought to the warehouse and from the warehouse to the customer [Ref. 5]. All warehouses use trucks or railroads, airlines, and waterlines to perform their distribution duties [Ref. 3]. The shipping time and cost heavily depend on the warehouse location and its ability to use various means of transportation easily. Consequently, the specific delivery-time requirements of the organizations must be established before any warehouse locations are determined.
Naturally, connected to the issue of transportation is the distance of the warehouse from each customer center. Normally, the customers served by the warehouse are separated in a region randomly. Creating a model based on optimization methods may minimize the total distance from all customer centers to the warehouses. In addition, while considering the location of the warehouse, forecasting new potential customers should be included in the decision process.
The availability of a rich labor market for the alternative site has usually minor importance, if the proposed operation has a high degree of automation and relatively little touch-labor. On the other hand, labor can be very important in other warehousing operations [Ref. 3].
Taxes can be a critical competitive factor for civilian companies when the warehouse inventory has high value. Variations in taxes, particularly inventory taxes, for different sites, can make the site costly [Ref. 3]. An example of this variation may be seen in free-trade zones and in some metropolitan areas.
Safety and security reasons are also common features for choosing a location. The probability of natural disasters should be researched. Many universities, private companies or government agencies have site safety test services that may be used [Ref. 7].
Security, somewhat different from safety, usually addresses physical security, such as protection against theft. For military and government warehouses, security has more priority than for private warehouses. At times, selecting a site far from a city center may help reduce such problems. Since the value of goods stored in warehouses can be significant, all necessary precautions must be taken before determining the location.
In searching for a supply center warehouse site, finding one that fits a general construction plan is more effective than attempting to adapt construction to the site. For instance, warehouse buildings with odd-shaped walls designed to fit a railroad curve or some other site constraint are usually more costly to operate. Enlargement opportunity of the site is also a critical factor in the planning and selection processes [Ref. 2].
Public utilities around the warehouse location area are becoming very important. Utilities include electricity, water, phone, or sewerage services, etc. Utilities not only affect operating costs, but also influence the risk of spoilage.
Industrial and technological environments are other influential characteristics for warehouse locations. To have the flexibility of easily applying new technologies to the warehouses and of keeping the education level of personnel high, the location should be near industrial or technological centers.
Every site seeker should also carefully study community attitudes toward the new warehouse. This is more important for military facilities since there may be an opposition against military installations in the proposed area. In most communities today, a clean and quiet warehouse development is considered preferable to operations that may cause pollution, congestion or other conditions perceived as detrimental to a community’s quality of life. Nowadays, some communities are opposed to any new industrial development, even warehousing [Ref. 3]. This kind of opposition obviously should be considered when one selects potential sites.
Preparing a list of the various constraints is the next phase before starting to eliminate the alternatives. Economic factors are usually the most essential constraints. The most important economic factor to measure in site selection is the cost versus the
value of a new warehouse site [Ref. 2]. Since companies do not have infinite resources to invest in a warehouse, they try to satisfy their needs within their available resources.
With all requirements and critical features defined, the user should now move through the selection process [Ref. 3]. By combining all these considerations, constraints and the advice from consultants, inadequate alternatives can be eliminated. The important issue here is to crosscheck the final proposed location from multi-sources to be sure that it is the best-choice location [Ref. 3].
SUMMARY OF SITE SELECTION PROCESS
One of the toughest decisions for a logistics manager is selecting the best location to establish a new warehousing facility. No manager wants to be remembered for locating a new warehousing facility in a ludicrous area, such as the vicinity of a toxic waste dump [Ref. 9]. The first step in the site selection process is to determine the macro-location and reduce it to the micro-location. Doing this is important since it forces the analyst to pursue a specific process of elimination. First, one should determine the largest possible universe that could be considered in selecting the site. Then one must, through elimination, systematically narrow the field as specific alternatives are considered. In narrowing the location from macro to micro, the decision makers should always keep in mind the main reason for seeking a new warehouse site [Ref. 2].
Sources of information must be crosschecked with other sources so that analysts are sure about the final site. If one consultant states that a specific site is earthquake safe with a good history of durability during past earthquakes, at least one other opinion from a trustable, objective and independent consultant should also be obtained.
Other general issues to be considered are zoning, topography, existing buildings or other improvements on the selected site, landscaping, access to the site, storm and sanitary infrastructure, water, sprinklers and other fire protection systems, power, and fuel [Ref. 2]. If any of these issues are ignored, serious problems could develop in the future.
Contingency plans are always quite valuable. Aside from the preferred site, selecting an alternative site, which is almost as good and equally available, is wise. Then
in the event that the bargaining for the preferred site should fail, letting the seller discover that an alternative site exists can improve one’s leverage during negotiations [Ref. 2].
As stated previously, site selection for warehouses is one of the most important decisions the related managers ever make. Although most of the decisions are correctable, a poor choice for a warehouse location is a very costly decision to correct [Ref. 2, 3]. In conclusion, before choosing a warehouse location, priorities are set, critical features are determined and finally, alternatives are eliminated. In other words, warehouse location decisions must be taken very carefully.
Appendix A contains a Location Analysis Checklist, a useful aid in analyzing a location [Ref. 3].
WAREHOUSE SITE SELECTION MODELS
BACKGROUND
Depending on the number of alternative locations, finding the best location can take a tremendous amount of time. Processing huge amounts of data requires decision makers to employ some of the site selection models in the large-scale problems. There are a number of site selection models to assist in analyzing various site selection scenarios [Ref. 11].
Some of the site selection methods are listed below:
1. Mathematical Optimization Models
Linear/Integer Programming Models fall in this category. Each model consists of an objective function and the constraints. The variables must be defined to represent each decision. Models can be concisely expressed using algebraic expressions with subscripted variables. Usually a solver software program can obtain the optimum solution.
2. Software Programs for Decision Analysis
There are some specific software programs designed to assist in making decisions. Logical Decisions for Windows (LDW) is one of them. This program isexplained in detail later in this chapter.
3. Simulation Models
Simulation is the process of designing and creating a model of a real system for the purpose of conducting numerical experiments to obtain a better understanding of the behavior of that system for a given set of conditions. Nowadays, simulations are usually created on a computer with appropriate software [Ref. 13]. The simulation models are quite economical and help avoid poor investments, but they can be expensive to develop.
4. Location-Allocation Models
Location-allocation modeling was originally developed to solve site selection problems in the public sector for facilities such as schools, fire stations and hospitals. Location-allocation is a very flexible approach, but the decision is usually restricted by the time and the cost required for computation [Ref. 14].
5. Center of Gravity Method
This method is also called the Minisum method. Minimizing transportation costs is quite important in site selection projects for a new warehouse, a logistics center or a new military base. The Center of Gravity method focuses on the transportation costs. Obviously, many other factors must be considered during the process of selecting a site. The Center of Gravity method presents a basic solution to the site selection problems. However, the outcome of this method is still quite valid [Ref. 8].
6. Multi-Criteria Decision-Making Models (MCDM)
Frequently, decision makers face numerous elements on a project that superficially seem mutually exclusive, yet in reality, each component is an intrinsic element of the system as a whole. As a consequence, every component in a system must be evaluated. This task entails many disciplines. Therefore, making decisions about any subject requires an interdisciplinary approach [Ref. 8].
Essentially, decision-making means solving problems. In other words, the decision maker is forever at a dichotomy. Sometimes, simple scientific methods are enough to solve the myriad of dichotomies, at other times, approaching the event multi-dimensionally is necessary [Ref. 8].
A decision maker often uses more than one criteria or objective to evaluate the alternatives in a decision problem. Usually, these criteria conflict with one another. There are many types of multi-criteria problems; they are very common in everyday life. For example, Multi-Criteria Decision-Making (MCDM) methods can be easily applied in choosing government projects, choosing new products, selecting candidates for a professional position, preparing equipment plans, selecting sites for various types of facilities, etc. MCDM refers to making decisions in the presence of these multiple criteria problems. MCDM methods employ multi-dimensional and interdisciplinary approaches [Ref. 8, 12, 15].
MCDM methods are categorized in many different ways. A general list of the commonly used MCDM methods is presented below:
a. Electré Method
Electré was originally developed by B. Roy [Ref. 33, 36] to incorporate the imprecise and uncertain nature of decision-making. Ranking and selecting projects are difficult and complicated tasks because there is usually more than one dimension for measuring the impact of each project and more than one decision maker. The Electré method has several unique features to handle multi-dimensional problems, namely the concepts of outranking and indifference and preference thresholds [Ref. 16, 17].
b. Weighted Sum Method (WSM)
Hwang and Yoon [Ref. 37] state that this procedure chooses, as the best alternative, the option that obtains the best global performance, as computed using a weighted sum of the performances of the alternatives along each criterion.
c. Multi-Attribute Utility Theory (MAUT)
A technique based on the paradigm of a decision tree and risk analysis based on cardinal utility. MAUT incorporates multiple viewpoints [Ref. 17, 21].
d. Analytic Hierarchy Process (AHP)
T. Saaty [Ref. 20] has proposed AHP as a systematic method for comparing a list of objectives or alternatives. AHP is especially suitable for complex decisions that involve the comparison of decision elements, which are difficult to quantify. AHP assumes that, when faced with a complex decision, the natural human reaction is to cluster the decision elements according to their common characteristics [Ref. 19, 20].
e. Preference Ranking Organization Method (PROMETHEE)
PROMETHEE is based on the same principles as Electré and introduces functions to describe the decision maker’s preferences along each criterion. This procedure provides a partial order of the alternatives [Ref. 18].
f. Technique for Order Preference by Similarity to Ideal Solution (TOPSIS)
This procedure chooses, as the best alternative, the option that is closest to an ideal solution and furthest from the worst solution [Ref. 18, 37].
g. Evidential Reasoning (ER)
The Evidential Reasoning (ER) approach is not only a method that combines both qualitative and quantitative assessments, but also handles uncertain and imprecise information or data. The state of an attribute (a criterion) may be determined by factors (sub-criteria) at a lower level [Ref. 12].
h. Factor Rating Method
This method is mostly used to evaluate the factors, which cannot be evaluated numerically with the other factors. Applying this method includes determining the highest scores for each factor (criterion), determining the various levels of each factor, and determining the appropriate scores for these levels [Ref. 8].
i. Sorting and Cost Convenience Method
Although this method does not always conclude with satisfactory solutions, it is still one of the most widely used methods. Applying this method includes assigning weights and scores to each factor, then finding the importance of each alternative location by calculating their total scores [Ref. 8].
B. LOGICAL DECISIONS FOR WINDOWS (LDW)
Logical Decisions for Windows (LDW) is a sophisticated system for evaluating alternatives that differ on a number of evaluation variables or criteria. Alternatives can be anything needed to choose between–jobs, potential employees, factory locations, or even what wine to have for dinner. LDW works best for decisions where many concerns must be considered at once, and where professional and value judgments will play a crucial role [Ref. 32].
LDW uses the judgments and preferences of the decision makers and the stakeholders to rank the alternatives. LDW helps decision makers review their preferences concerning the measures by guiding them through a series of questions. On the basis of their answers, Logical Decisions constructs the formula that ranks the alternatives [Ref. 32].
LDW uses powerful methods from the field of Decision Analysis to help decision makers evaluate the alternatives. Decision analysis was developed in the 1960s and
In summary, the following steps describes the development of a LDW model for a decision problem [Ref. 32]:
•
Define a set of alternatives to be ranked
•
Define measures to describe the alternatives
•
Enter the level on each measure for each alternative
•
Review your preferences so the measure levels can be combined
•
Rank the alternatives and choose the best one.
C. CENTER OF GRAVITY (MINISUM) METHOD
As stated before, the Center of Gravity Method is based on minimizing the total transportation costs. Doing this, the Center of Gravity Method makes two assumptions about the transportation costs [Ref. 8]:
•
Transportation cost is the only factor when selecting a site,
•
Transportation cost changes proportionally to the transportation distance.
The following data must be known to compute the coordinates of the best location that minimizes the transportation costs [Ref. 8]:
•
The amount of cargo to be transported to all customer locations,
•
The geographic coordinates of all locations on a specific grid system,
•
The unit transportation costs.
The mathematical notations of the variables are as follows [Ref. 8, 34]:
P
i (Xi; Yi) : The coordinates of the customer location i.
T (X; Y) : The coordinates of the best location.
D (T-P
i) : The distance between location T and Pi, (km).
C
i : The transportation cost to carry one unit load to one unit distance between T and Pi, (Cost/kg*km).
Q
i : The amount of load to transport between T and Pi, (kg).
N : The number of customer locations.
TC : The total transportation cost .
The objective function is defined as below:
The objective of this method is finding the coordinates of the location T (X; Y) that minimize the total transportation costs. C
i, Qi and n are constant values. The only variable is the distance: d (T-Pi). The distance can be calculated in three different ways [Ref. 34,35]:
a. Rectilinear Distance
Used for ground transportation when a rectilinear street network or aisle network needs to be considered. The rectilinear distance between the best location and the customer locations is calculated as follows [Ref. 34]:
d (T-P
i) = |X – Xi| + |Y – Yi|. (2)
b. Linear (Euclidean) Distance
Used for air transportation since the distances are assumed to be linear for this type of transportation. The linear distance between the best location and the customer locations is calculated as follows [Ref. 34]:
c. Square of Linear Distance (Centroid Problem)
In linear distance problems the convex hull is a line segment. The cases where convex hull is not a line segment, the distances are obtained by squaring each Euclidean (linear) distance. Finding a location to minimize the transportation costs from the square of linear distances is called the centroid problem. The solution to the centroid problem is called the centroid. The centroid is the unique new facility location that
inimizes the transportation cost function. The contour sets of this function are quite simple; in fact, it is known that contour sets are disks, each with an optimum point as a center. Thus the contour sets are particularly simple to construct, and just as with the rectilinear distance problem, are useful in evaluating other possible locations for the new facility. The closer the new facility can be to the centroid location, the better the solution will be. The distances are obtained from the following formula for this method [Ref. 34]:
1. Calculation with the Rectilinear Distance [Ref. 8, 34]
W
i is the transportation cost of carrying the entire load of customer “i” (Pi) to one unit distance between T and Pi. In short, “Wi” will be called “the weighted load values” and represented with the following formula:
W
i = Ci * Qi. (5)
The formula for the total transportation cost is below:
TC = Min f (x) + Min f (y).
f (x) is the total cost of transportation in the x-direction and f (y) is the total cost of transportation in the y-direction. Thus the minimum total cost can be obtained by solving the two independent problems of minimizing the cost of transportation in the x-direction and minimizing the cost of transportation in the y-direction.
The X and Y values minimizing both functions are obtained from the following formulas:
The largest cumulative sum of
and
The smallest cumulative sum of
Wm
is the cumulative weighted load value (cumulative sum of Wi) that satisfies the inequalities of (6) and (7). When Wm is greater than the median (½ * WΣ=ni1i ), the coordinates of the best site are the X and Y median values, which belongs to the Wm value (mth X value=Xm and mth Y value=Ym). If Wm is equal to the median, then these coordinates are expressed as an interval: [Xm, Xm+1] or [Ym, Ym+1]. An example of the calculation is explained later in this chapter.
The following steps are used to determine the best location:
•
Create a data table to find the abscissa (X) and another table to find ordinate (Y) of the best location,
•
Sort ascending the abscissa and ordinates of the demand centers,
•
Calculate the Wi (Ci*Qi) value of the customer locations,
•
Calculate the cumulative sum of Wi values and list in a column on the data table,
•
Calculate ΣWi (the total of all weighted load values),
•
Calculate the median (½ ΣWi) value,
•
Find the number that is equal to the (½ ΣWi) from the cumulative sum of Wi column. If the (½ ΣWi) value does not exist in the cumulative sum of Wi, then find the number that is closest in value to the (½ ΣWi) but greater than (½ ΣWi). The abscissa and ordinate values in these rows give the point T (X; Y), which minimizes the total transportation costs.
2. Calculation with the Square of Linear Distance [Ref. 8, 34]
W
i (Wi=Ci*Qi) represents the same weighted load values in this method.
The distance function between T and P
i locations is:
d (T-P
i) = (X–Xi)2 + (Y–Yi)2.
Thus, the total transportation cost function is represented as below:
TC =
Σ{W=ni1i* [(X–Xi)2 + (Y–Yi)2]}.
The T (X; Y) point, which minimizes this function, must satisfy the following conditions:
The coordinates of the best location are obtained by setting the above partial derivatives to zero (0). Consequently, the abscissa (X) and the ordinate (Y) of the best location are derived and represented as below:
D. ELECTRÉ METHOD
1. Background
Bernard Roy [Ref. 33, 36] conceived the Electré method in 1966 in response to the deficiencies of the existing decision-making methods. This method has evolved through a number of versions (I through IV); all are based on the same principles but are operationally somewhat different [Ref. 16].
The Electré method may not always be the best decision aid; however, it is a proven approach. It has several unique features not found in other methods, including the concepts of outranking, as well as indifference and preference thresholds. Additionally, the Electré method can especially be used to convert qualitative data to quantitative [Ref. 8, 16].
With its dynamic characteristics, this method may be applied successfully to many problems. However, to obtain reliable results from this method, a decision maker must define the problem broadly, identify the main constraints, and most importantly identify the primary objective. Next, a decision maker should establish an interdisciplinary committee to address the problem, and the members of this committee should be experts in various fields related to the problem. This committee must have experience and the ability to handle the problem with an interdisciplinary approach. They must be unbiased. Likewise, the decision maker must eschew personal bias that could deviate the interdisciplinary committee from the primary objective [Ref. 8, 33].
Using the Electré method, the interdisciplinary committee generally follows the guideline given below to select the best location [Ref. 8, 33]:
•
Identify the alternative locations,
•
Identify the important criteria of the problem,
•
Assess each alternative location according to each criterion
•
Employ the Electré method to solve the problem.
2. Explanation of the Electré Method on an Example [Ref. 8]
The Electré method will be applied to a site selection problem. Alternative locations (options) and criteria will be shown with symbols. The following steps are used to solve the problem.
a. Step1: Identifying the Options [Ref. 8, 33]
There will be five alternative locations in this example problem. These alternatives (options) are represented by the following symbols: A, B, C, D, and E. These locations may be proposed for a supermarket, a hospital, an appliance main depot, etc.
b. Step 2: Identifying the Criteria [Ref. 8, 33]
Some of the criteria defined by the interdisciplinary committee may include:
1. Transportation
2. A supply of temporary labor
3. The potential for maintaining a stable workforce
4. Enlargement opportunity
5. The populace’s response to the project
6. Integration availability to the existing facilities
7. Closeness to raw material or supply
8. Water supply
9. Suitability of transportation facilities and costs
10. The culture and advantages of a metropolitan area
11. Closeness to markets
12. Sewage and garbage service
13. Proximity of universities as a source of high-caliber work force
14. Advertising availability
15. Topography of the site
16. Energy and electricity availability
17. Employer-employee relations
18. Fuel and heating and cooling expenses.
The criterion list can be expanded according to the particular characteristics of the problem. However, the example problem in this chapter will consist of only five criteria. These five criteria are represented by a, b, c, d, and e.
c. Step 3: Weighing the Criteria [Ref. 8, 33]
Weighing the criteria is one of the most vital points of this method. The interdisciplinary committee will sort the criteria by their levels of importance and score each criterion by considering the primary objective.
d. Step 4: Determining Scales [Ref. 8, 33]
Instead of using numerical grades to evaluate the options according to the criteria, the options must be evaluated with the qualitative measures, such as: “Very good, good, not bad, bad, very bad.” Next, these qualitative results will be converted to numerical values according to the predetermined scales. The upper and lower limits of the scales will match the “very good” and “very bad,” and the intermediate values (good, not bad, bad) will be calculated with the interpolation. For example, a scale from 10 to 0 can be represented as below:
•
Very good : 10.0
•
Good : 7.5
•
Not bad : 5.0
•
Bad : 2.5
•
Very bad : 0.0
The committee must adhere to the following requirements about the scales:
•
There must be as many different scales as the number of different weights, which are determined during Step 3.
•
The scale range of the highest weight must be the largest, and the scale range for the lowest weight must be the narrowest. Accordingly, the scale range of lower weights must be a subset of the scale ranges of higher weights. For example, assume that very important criterion “a” has a weight of 4, less important criteria “b and d” have a weight of 2, and the least important criteria “c and e” have a weight of 1. In this example, the scale ranges are chosen as 0 to 10 for “a,” 2 to 8 for “b and d,” and 3 to 7 for “c and e” (Table 3.3). The ranges can be chosen differently as long as the range of less important criteria is a subset of the higher important criteria. For example, when an option is evaluated as Very Good for two different criteria that have different weights (importance level), this option will get a higher score from the more important criterion and a lower score from the less important criterion. The number of different scales with different ranges is equal to the number of different weights used in the problem.
e. Step 5: Evaluating Options Regarding Criteria [Ref. 8]
Each option is evaluated regarding all the criteria. In the example problem, the options “A, B, C, D, E” are evaluated regarding to criteria “a, b, c, d, e” (Table 3.4).
f. Step 6: Forming the Concordance and Discordance Matrices
The fundamental assumption of the Electré method is the existence of an outranking operation between two options. For example, one defines option A to outrank option B if the following two conditions are fulfilled [Ref. 22]:
1. A is better or at least as good as B with respect to a major subset of the criteria,
2. A is not much worse than B with respect to the remaining criteria.
The first condition is a concordance condition and the second condition is a discordance condition [Ref. 22].
For the example problem, the concordance matrix value of “E outranks B” assumption is calculated as below [Ref. 8].
•
Compare column E values of Table 3.4 with column B values.
•
Find the criteria in which the score of option E is equal or greater than the score of option B [using Table 3.4, E is at least as good as B for criterion b (5>3.5), criterion c (7>3), and criterion e (6=6)].
•
Calculate the sum of the weights of these criteria and divide it by the sum of all the weights of criteria. The weights of the criteria are respectively 2, 1, and 1 for criteria b, c, and e. The sum of these weights is 4, and it is divided by the sum of
the weights of all the criteria (10=4+2+1+2+1). Therefore the concordance matrix value for this assumption is 0.4 (4/10).
•
Write this value in the intersection cell of Row B and Column E on the concordance matrix.
The complete concordance matrix is presented below:
In the example problem, the discordance matrix value of “C outranks A” assumption is calculated as below [Ref. 8].
•
Compare column C values of Table 3.4 with column A values.
•
Find the criteria that the score of option C are less than the score of option A [using Table 3.4, the score of option C is less than option A for the criterion a (2.5<5) and criterion c (5<7)].
•
Subtract the scores of option A from the scores of option C and determine the greatest deviation among these pairs of scores [(5-2.5=) 2.5 and (7-5=) 2; therefore the greatest deviation is 2.5 (2.5>2)].
•
Divide the greatest deviation by the largest scale range (10 – 0 = 10). This value is the discordance indicator for the C outranks A assumption and will be inserted in the intersection cell of row A and column C (0.25=2.5/10).
•
Note this value (0.25) in the intersection cell of Row B and Column E on the concordance matrix. The outcome is called the first discordance matrix (s=1) (“s” is the discordance parameter).
The completed first discordance matrix is presented below:
If one option is much worse than another option (high discordance), then the outranking assumption between these two options will automatically be penalized. However, this may not be a favorable situation because the discordance indicator can meet the second condition when the second discordance matrix (s=2) is formed. The second discordance matrix helps to control the results of the first discordance matrix. The following steps are used to calculate the second discordance matrix (s=2) [Ref. 8]:
•
Follow the first two steps in the concordance procedure above.
•
Subtract the scores of option A from the scores of option C and determine the second greatest deviation among these pairs of scores. If the previous deviation (greatest deviation = 2.5) is the only deviation between the grades of the option pair, then the discordance indicator will be zero (0). Similarly, if the previous deviation (2.5) is equal to the second greatest deviation between the grades of the option pair, this deviation will be used to calculate the discordance indicator again. For the example problem, (5-2.5=) 2.5 and (7-5=) 2; therefore the second greatest deviation is 2.
•
Divide this value by the largest scale range (10 – 0 = 10). This value will be presented in the intersection cell of row A and column C of second discordance matrix (s=2). For the example problem, this value is 0.2 (2/10).
The completed second discordance matrix (s=2) is presented below:
g. Step 7: Electing and Decision
Before determining the best option, the threshold and nucleus concepts must be explained. There are two kinds of thresholds: the preference threshold (p) and the indifference threshold (q). The decision maker specifies the indifference thresholds. The choice of appropriate thresholds is not easy, but realistically, non-zero values should be chosen for p and q. While the introduction of this threshold goes some way toward incorporating how a decision maker actually feels about realistic comparisons, a problem remains. There is a point at which the decision maker changes from indifference to strict preference. Conceptually, there is good reason to introduce a buffer zone between indifference and strict preference, an intermediary zone where the decision maker hesitates between preference and indifference. This zone of hesitation is referred to as “weak preference”, and is modeled by introducing a preference threshold, p. Thus, the Electré Method is proposed as a double threshold model [Ref. 16].
Using the thresholds, the following preference relation can be defined for A outranks B assumption: If the concordance indicator (value in the concordance matrix) of this option pair is greater than or equal to “p” and the discordance indicator is less than
or equal to “q,” then A is preferred to B. These two conditions are reviewed for all pairs and the results will be shown in the solution figure. In this figure, each option is represented with a node and each option pair is connected with an arrow that points from the outranking (preferred) option to the other one. Thus, there will be two types of nodes in the chart, one of which is the node with at least one arrow-entering and the other type is a no-arrow-entering node. No-arrow-entering nodes are called nucleus. If a node is not connected to any other nodes, this node can also be an element of the nucleus. However, this situation can be risky and misleading and should be resolved [Ref. 8].
The actual solution process begins with the first concordance and discordance matrices (s=1) and the beginning threshold values are chosen. In the example problem, the beginning thresholds are chosen as p=0.7 and q=0.45 (notation: 0.7/0.45/1). If the discordance indicator is equal to 0.45 or less in the first discordance matrix, and the concordance indicator in the corresponding cell of the concordance matrix is also equal to 0.7 or greater, then this cell will be marked with “*” in the solution figure. This comparison is repeated for the entire concordance and discordance matrices. The results of the example problem are shown in Figure 3.1. B and D options (alternatives) are the two nuclei of the first iteration because no arrow is entering these nodes. Since there must be only one nucleus node, the indifference threshold (q) is increased to 0.6 in this example (notation: 0.7/0.6/1). This situation is shown in the second part of Figure 3.1. After this iteration, the only nucleus will be the option D [Ref. 8].
The second discordance matrix (s=2) is used to control this result. When the discordance parameter is increased from s=1 to s=2, then the values of the discordance indicators in the discordance matrix will decrease. Therefore, decreasing the value of the indifference threshold q is wise. The indifference threshold is chosen as q=0.15 and preference threshold is not changed (p=0.7) (notation: 0.7/0.15/2). If the discordance indicator is equal to 0.15 or less in the second discordance matrix and also the concordance indicator in the corresponding cell of the concordance matrix is equal to 0.7 or greater, then this cell will be marked with “*” in the third part of Figure 3.1. The B and D options (alternatives) are again the two nuclei of the third iteration. Since there must be only one nucleus node, the indifference threshold (q) is increased to 0.2 in this
example (notation: 0.7/0.2/1). This situation is shown in the last part of the Figure 3.1. After this iteration, the only nucleus will be the option D again. This result is consistent with the first finding. Therefore the final location should be option D [Ref. 8].
3. Conclusion
The Electré method is a dynamic method that is used to solve multi-criterion problems. Forming an interdisciplinary committee is highly recommended for Electré method applications. The committee and the decision maker can also employ consultants when necessary to evaluate the options of the project.
Owing to its dynamic perspective, the Electré method can prevent most of the drawbacks of the other decision-making processes.
Ugur Erdemir
Logistics and Acquisition Professional, Industrial Engineer
ugurerdemir1975@hotmail.com
See also:
Risk Allocation and Motivation Functions of Contracts
http://contractriskallocationandmotivation.blogspot.com/
Excess Inventory and Disposal
http://excessinventoryanddisposal.blogspot.com/
See also:
Risk Allocation and Motivation Functions of Contracts
http://contractriskallocationandmotivation.blogspot.com/
Excess Inventory and Disposal
http://excessinventoryanddisposal.blogspot.com/
APPENDIX A - SITE ANALYSIS CHECKLIST
Ackerman, Kenneth B. [Ref. 3] in his book
Practical Handbook of Warehousing presents the following Site Analysis Checklist as a useful aid to analyzing location decisions:
1. Current: Proposed: Master plan: Anticipated:
Storm Drainage
1. Is adequate electric power available to site? ___yes ___no
1. Creed, H. Jenkins, Modern Warehouse Management, McGraw Hill, New York, 1968.
General Information
1. Site location (city, county, state):
2. Legal description of the site:
3. Total acreage:
Approximate cost per acre:
Approximate dimensions of site:
4.Owner(s) of site (give names and addresses);
Zoning
2. Is proposed use allowed? ___yes ___no
Check which, if any, is required: ___rezoning ___variance ___special exception
Indicate approximate cost:
Indicate probability of success: ___excellent ___good ___fair ___poor
3. Applicable zoning regulations (attach copy):
Parking/loading regulations:
Open space requirements:
Office portion:
Maximum number of buildings allowed:
Warehouse/Distribution Center portion:
Percent of lot occupancy allowed:
Height restrictions: Noise limits: Odor limits:
Are neighboring uses compatible with proposed use? ___yes ___no
4. Can a clear title be secured? ___yes ___no
Describe easements, protective covenants, or mineral rights, if any:
Topography
1. Grade of slope: Lowest elevation: Highest evaluation:
2. Is site: ___level ___mostly level ___uneven ___steep
3. Drainage ___excellent ___good ___fair ___poor
Is degrading necessary? ___yes ___no
Cost of regarding:
5. Are there any ___streams ___brooks ___ditches ___lakes ___ponds ___on site ___bordering site ___adjacent to site?
Are there seasonal variations? ___yes ___no
5. What is the 100-year flood plan?
6. Is any part of site subject to flooding? ___yes ___no
7. What is the ground water table?
8. Describe surface soil:
9. Does site have any fill? ___yes ___no
10. Soil percolation rate ___excellent ___good ___fair ___poor
11. Load-bearing capacity of soil: ___PSF
12. How much of site is wooded: How much to be cleared:
Restrictions on tree removal: Cost of clearing site:
Existing Improvements
1. Describe existing improvements.
2. Indicate whether to be ___left as is ___remodeled ___renovated ___moved ___demolished
1. Describe the landscaping requirements for building parking lots, access road, loading zones, and buffer if necessary.
Access to Site
1. Describe existing highways and access roads, including distance to site (include height and weight limits of bridges and tunnels, if any).
2. Is site visible from highway? ___yes ___no
3. Describe access including distance from site to
• Interstate highways
• Major local roads
• Central business district
• Rail
• Water
• Airport.
Describe availability of public transport?
4. Will an access road be built? ___yes ___no
If yes, who will build it? Who will maintain it? Cost?
Indicate curb cuts, median cuts, traffic signals, and turn limitations:
5. Is rail extended to site? ___yes ___no If yes, name of railroad(s):
If no, how far? Cost of extension to site:
Who will maintain it? Is abandonment anticipated? ___yes ___no
1. Location and size of existing storm sewers:
2. Is connection to them possible? ___yes ___no Tap charges:
4. Where can roof drainage be discharged?
5. Describe anticipated or possible long-range plans for permanent disposal of storm waters, including projected cost to company.
Sanitary Sewage
1. Is public treatment available? ___yes ___no. If no, what are the alternatives?
2. Is there sanitary sewage to site? ___yes ___no. Location of sewer mains:
3. Cost of materials (from building to main)–include surface restoration if necessary:
4. Tap charges:
5. Special requirements (describe fully);
6. Describe possible or anticipated long-range plans for permanent disposal of sewage, including projected cost to company.
Water
1. Is there a water line to site? ___yes ___no
2. Location of main: Size of main:
3. Water pressure: Pressure variation:
4. Hardness of water:
5. Source of water supply: Is supply adequate? ___yes ___no
6. Capacity of water plant: Peak demand:
7. Who furnishes water meters? Is master meter required? ___yes ___no
Preferred location of meters: ___outside ___inside
8. Are fire hydrants metered? ___yes ___no
9. Attach copy of meter rates, including sample bill for anticipated demand if possible.
Capacity available at site:
2. Describe high voltage lines at site:
3. Type of service available:
4. Service is ___underground ___overhead
5. Reliability of system ___excellent ___good ___fair ___poor
6. Metering is ___indoor ___outdoor
7. Is sub-metering permitted? ___yes ___no
8. Indicate if reduced rates are available for
• Heat pumps: ___yes ___no
• Electric heating: ___yes ___no
9. Attach copy of rates, including sample bill for anticipated demand, if possible.
Fuel
1. Type of gas available:
2. Location of existing gas lines in relation to site:
3. Existence of a refinery:
Taxes
1. Date of most recent appraisal:
2. Real estate tax rate history:
3. History of tax assessments:
4. Proposed increases:
5. Are there any abatement programs in effect? ___yes ___no If yes, describe.
6. Is site in an Enterprise Zone? ___yes ___no
8. Indicate anticipated or possible major public improvements:
9. Services provided for taxes paid: Local: County: State:
10. What is state policy on inventory tax?
11. Indicate rates for:
• Personal income tax,
• Corporate income tax,
• Payroll tax,
• Unemployment compensation,
• Personal property tax,
• Sales and use tax,
• Franchise tax,
• Other taxes.
LIST OF REFERENCES
2. Tomkins, A. James and Smith, D. Jerry,
The Warehouse Management Handbook, McGraw Hill, New York, 1988, pp.1, 73-90.
3. Ackerman, Kenneth B.,
Practical Handbook of Warehousing, 3rd ed., Van Nostrand Reinhold, New York, 1990, pp. 3,91-133.
4. Martin, James F., "Improving the Layout of a Warehouse at the Coast Guard Aircraft Repair and Supply Center," NPS Master’s Thesis, Monterey, 1999, p. 5.
5. Dimitris, N. Chorafas,
Warehousing, The Macmillan Press Ltd, London, 1974.
6. Ashayeri, J. and Gelders, L. F., "Warehouse Design Optimization," European Journal of Operations Research, V. 21, 1985, pp. 285-294.
7. Interview with Tanyas, Mehmet, PhD., Istanbul Technical University (ITU), Instructor, July 2002.
8. Course Handbook, "Warehouse Site Selection Models," Istanbul Technical University (ITU), 2000.
9. Thomas, L. Freese,
Site Selection: How to Choose the Best Location, Freese & Associates, Chagrin Falls, OH, 1997.
10. Bolger, Dan,
Operations Management: Warehouse & Delivery Savings, Towse Publishing Company, P. E., 1997.
11. Gold, Steven and Marvick, Peat, "Beyond Bricks and Mortar: Warehouse Site Selection," April 1997 [www.glscs.com/archives/4.97.opinion.htm?adcode=30] (December 2002).
12. "Evidential Reasoning Approach," [http://info.sm.umist.ac.uk/wp/papers/wp2005.htm] (December 2002).
13. Kelton, W. David, Sadowski, P. Randall, Sadowski, A. Deborah,
Simulation with Arena, 2nd ed., McGraw Hill, Boston, 2002, p.7.
Location-Allocation Models for Retail Site Selection, CSCA Publications, January 1995.
15. Ragsdale, T. Cliff,
Spreadsheet Modeling and Decision Analysis, 3rd ed., South Western College Publishing, 2001, p. 764.
16. Buchanan, John and Sheppard, Phil, "Ranking Projects Using the Electré Method," Hamilton, New Zeland.
17. "Multi-Criterion Decision Analysis for Site Selection" (lecture notes), [http://ecolu-info.unige.ch /~haurie/mutate/Mutate-web-page/Lect_1_1_2.doc] (December 2002).
18. Irène Abi-Zeid, Micheline, Bélanger Adel, Guitouni, Jean-Marc Martel, and Khaled Jabeur, "A Multi-Criteria Method for Evaluating Courses of Action in Canadian Airspace Violation Situations," Québec, Canada, [http://www.dodccrp.org/Proceedings /DOCS/wcd00000/wcd00091.htm] (December 2002).
19. "Analytical Hierarchy Process," [www.aoe.vt.edu/~chall/courses/aoe4065/AHP.pdf] and Ernest H. Forman, "Decision by Objectives," [http://mdm.gwu.edu/Forman/DBO.pdf] (December 2002).
20. Saaty, T. L.,
The Analytical Hierarchy Process, University of Cambridge Department of Engineering, McGraw Hill, NY, 1980 [http://www-mmd.eng.cam.ac.uk/people/ahr/dstools/choosing/ahp.htm] (December 2002).
21. "Multi-Attribute Utility Theory, Columbia University, April 1990," [www.cpmc.columbia.edu/edu/G4050/week11.ppt] (December 2002).
22. "Multi-Criterion Decision Making Using Electré" (lecture notes), [http://ecolu-info.unige.ch / ~haurie / mutate / Mutate_final / Lectures / Lect_1_3_2 / Lect_1_3_2.htm] (December 2002).
33. Roy, Bernard,
The Outranking Approach and the Foundations of Electre Methods, in: C.A. Bana e Costa (ed.), Readings in Multiple Criteria Decision Aid, Springer, Berlin, 1990.
34. Francis, L. Richard; McGinnis, F. Leon, Jr; White, A. John,
Facility Layout and Location: An Analytical Approach, 2nd ed., Prentice Hall, Englewood Cliffs, New Jersey, 1992, pp. 189-209.
35. Chase, B. Richard and Aquilano J. Nicholas,
Production and Operations Management Manufacturing and Services, Irwin, Chicago, 1995, pp.380-382.
36. Roy, Bernard,
Classement et Choix en presence de Points de Vue Multiples.(La Methode Electre), Revue Française d'Informatique et de Recherche Operationnelle, Paris: Dunod , No. 8-VI, 1968.
37. Hwang CL, Yoon K.,
Multiple Attribute Decision Making: Methods and Applications, a State-of-the-Art Survey, New York: Springer, 1981.
38. R. L. Keeney and H. Raiffa,
Decisions With Multiple Objectives: Preferences and Value Tradeoffs, John Wiley, 1976.
39. www.europa-tech.com (December 2002)
