Multiple Criteria Decision Analysis for Industrial Engineering

Methodology and Applications

Gerald William Evans

• Publisher: CRC Press
• Year: 2016
• ISBN:
  9781498739856
• Edition: 1
• Topic: Technology & Engineering
• Subtopic: Engineering General
• Index: Technology & Engineering

This textbook presents methodologies and applications associated with multiple criteria decision analysis (MCDA), especially for those students with an interest in industrial engineering. With respect to methodology, the book covers (1) problem structuring methods; (2) methods for ranking multi-dimensional deterministic outcomes including multiattribute value theory, the analytic hierarchy process, the Technique for Order Preference by Similarity to Ideal Solution (TOPSIS), and outranking techniques; (3) goal programming,; (4) methods for describing preference structures over single and multi-dimensional probabilistic outcomes (e.g., utility functions); (5) decision trees and influence diagrams; (6) methods for determining input probability distributions for decision trees, influence diagrams, and general simulation models; and (7) the use of simulation modeling for decision analysis.

This textbook also offers:

· Easy to follow descriptions of how to apply a wide variety of MCDA techniques

· Specific examples involving multiple objectives and/or uncertainty/risk of interest to industrial engineers

· A section on outranking techniques; this group of techniques, which is popular in Europe, is very rarely mentioned as a methodology for MCDA in the United States

· A chapter on simulation as a useful tool for MCDA, including ranking & selection procedures. Such material is rarely covered in courses in decision analysis

· Both material review questions and problems at the end of each chapter. Solutions to the exercises are found in the Solutions Manual which will be provided along with PowerPoint slides for each chapter.

The methodologies are demonstrated through the use of applications of interest to industrial engineers, including those involving product mix optimization, supplier selection, distribution center location and transportation planning, resource allocation and scheduling of a medical clinic, staffing of a call center, quality control, project management, production and inventory control, and so on. Specifically, industrial engineering problems are structured as classical problems in multiple criteria decision analysis, and the relevant methodologies are demonstrated.

1

Introduction

1.1 Decision Making Is Important

Decision making is an important aspect of any human endeavor, whether that endeavor is accomplished by an individual or as part of an organization. Analyzing a decision situation and then making a “good decision” is a prelude to accomplishing important endeavors. As such, both individuals and organizations make decisions in order to solve problems and to take advantage of opportunities. Some examples of personal decisions made by individuals include the following:

1. A high school student decides where to attend college and what to study.

2. A consumer decides which type of automobile to purchase.

3. A newly married couple decides whether to honeymoon in Florida or Hawaii.

4. A patient, in conjunction with his or her doctor, decides whether to use chemotherapy or radiation to treat his or her cancer.

In fact, personal decisions, especially those having to do with our health and safety, can even result in death (Keeney, 2008).

Here are some examples of decision problems faced by private industry:

1. General Electric decides whether or not to develop a new MRI machine with a larger opening for the patient (Welch, 2005, pp. 74, 75).

2. An electric utility cooperative decides whether or not to add an additional transmission line to link with an electric utility (Borison, 1995).

3. A corporation decides where to locate a new production facility.

4. An automobile manufacturer decides whether or not to produce a new line of cars.

Examples of decision problems faced by individuals (within professional settings) include the following:

1. A machine operator decides which part to process next.

2. An industrial engineer designs a layout for a new production facility.

3. A plant manager selects a new quality engineer from among several applicants.

4. An information systems manager selects an accounting system for his or her organization.

Finally, some examples of decisions made by public institutions such as local or federal governments include the following:

1. A state government (in conjunction with the local metropolitan government and the federal government) chooses a location for a new bridge in the metropolitan area.

2. The federal government decides whether or not commercial airplanes should have special protection against surface-to-air missile attacks by terrorists (von Winterfeldt and O’Sullivan, 2006).

3. The Department of Energy decides how to allocate a limited budget among various R&D projects (Parnell, 2001).

In its most elemental form, a decision problem involves a situation in which one alternative must be selected from among several feasible alternatives. In fact, the word “decide” is derived from the Latin decidere, which means “to cut off.” When we decide on something, we “cut off” the other alternatives from further consideration by selecting one alternative. For example, the corporation in the aforementioned example might be considering three alternatives for the location of a new production facility: Evansville, Indiana; Lexington, Kentucky; and Nashville, Tennessee. When they choose one of the locations (i.e., they make a decision), they cut off the other two from further consideration.

This book is concerned with problems and situations involving decision making and optimization. Of particular interest are situations in which multiple objectives and uncertainty (and resulting risk) must be considered. When the number of alternative decisions in a particular situation is so large that evaluation of all alternatives is not possible, then some type of optimization technique is desirable as part of the solution process. Examples of situations involving such a large number of alternatives would be one involving continuous “decision variables,” such as the amount of product produced by a company in the next month, or a combination of discrete/integer decision variables, such as the number of machines of various types to include in a production facility.

1.2 Characteristics of Decision Situations

1.2.1 Decisions Are Made to Solve Problems or to Take Advantage of Opportunities

As mentioned in the previous section, a decision is made to solve a problem or, alternatively, to take advantage of an opportunity. A problem can be defined as a gap between a current state of affairs and some desired state of affairs. For example, in the situation described earlier involving the protection of commercial airplanes, the “problem” might be defined as “there is a gap between the perceived current level of safety and the desired level of safety because of a possible terrorist attack on a commercial airplane.” As another example, a personal problem might be defined as “there is a gap between my current salary and the salary I would like to receive.”

Note that in many cases, the “state of affairs” is defined in a fairly nebulous fashion so that in order to proceed in the evaluation of various alternatives, one must clearly define the performance measures that determine the state of affairs. The concept of a “problem” will be discussed in much more detail in Chapter 2.

1.2.2 A Good Decision May Not Always Result in a Good Outcome (and a Bad Decision May Not Always Result in a Bad Outcome)

The timed sequence of processes/events that occur in a decision situation might be defined as follows:

1. Decision analysis

2. Selection of a decision to implement

3. Implementation of a decision

4. Occurrence of the results of the decision

The results of a decision are typically uncertain in nature, often depending on many uncertain parameters.

As a simple example, consider a friend of yours named Harry who is down to his last $2000. Suppose that Harry is considering two alternatives for the use of this money:

1. Invest the money in an educational course that will give him a 9 out of 10 chance of procuring a better job with an increase in annual salary of $10,000 over his current salary.

2. Purchase lottery tickets that will give him a probability of 1 chance in 10 million of winning $1 million.

Most people would consider the first alternative the best one. But if Harry does not procure a better job after the course (a 10% chance), then the result (or outcome) would be bad. One of the purposes of a decision analysis process would be to maximize the chances of getting a good outcome. Although it may not always turn out this way, over the long run the results of your decisions should be better outcomes through the use of a good decision analysis process.

1.2.3 Decisions Can Often Be Categorized as Being Strategic, Tactical, or Operational in Nature

Decisions must be made in a variety of situations. For example, one categorization of decision situations is strategic (e.g., decisions that influence outcomes more than a period of 1–10 years or longer), tactical (affecting outcomes more than a period of 1–12 months), and operational (affecting outcomes more than a period of one to a few days). Typically, decisions made at the strategic level place constraints on decisions made at the tactical and operational levels, and decisions made at the tactical level constrain decisions made at the operational level.

As an example, an organization’s strategic decision of where to locate a new manufacturing facility will probably affect tactical decisions regarding transportation policy. Hence, one wants to at least implicitly consider the tactical and operational decisions resulting from strategic decisions in a strategic decision analysis. By the same token, one wants to implicitly consider the relevant operational decisions in an analysis for a tactical decision.

1.3 Steps in the Process of Decision Making

The major steps in the process of decision making are as follows:

1. Generate alternative solutions (or just alternatives) for the problem.

2. Determine the performance measures for the problem situation.

3. Rank the alternatives in terms of the performance measures.

4. Implement the first-ranked alternative.

Of course, each of these major steps involves several related activities. For example, steps 1 and 2 involve the consideration of the values and objectives associated with the person/organization making the decision as well as the decision situation. In order to generate the values and objectives of an organization or a decision situation, one must consider the decision makers and stakeholders involved.

For example, the first thing that the couple who were about to marry had to do was to generate potential locations for their honeymoon. This could have been accomplished through the gathering of information (e.g., through the Internet, by talking with friends and travel agents, and by visiting the library) about different possibilities. The next step, as described earlier, involves the determination of performance measures. The performance measures in this case could have involved such measures as the projected cost of the trip and the amount of “fun” that the couple could expect to have. Note that the first performance measure is quantitative (or objective) in nature, while the second one is qualitative. In addition, there may be some uncertainty about the outcome associated with a performance measure for a particular alternative. For example, the amount of “fun” that the couple has on a trip to Florida may be dependent on the weather in Florida while they are there.

The third step in the process involves ranking the alternatives. This could, and probably would, involve making trade-offs between pairs of performance measures. For example, the honeymooning couple would probably have to trade off between “fun/enjoyment” and “cost.”

Finally, the decision must be implemented. In the case of the honeymooning couple, this would involve the purchase of tickets, making reservations, and so on.

Of course, most decisions are made without the explicit consideration of these formal steps. The contention associated with decision analysis is that better decisions will be made through the explicit use of the methodologies and processes such as those described in this book.

1.4 Elements in a Decision Analysis Process

The steps of the decision-making process described earlier involve several elements/agents. These elements include the following:

1. The people involved: decision maker(s), stakeholders, and analysts

2. Alternative solutions to the problem

3. Ways to measure the performance of an alternative: criteria, performance measures, and attributes

4. Constraints on the alternatives

5. A forecast associated with various states of nature

6. Models/techniques for the evaluation of alternatives

7. Models/techniques for the ranking of feasible alternatives and/or the selection of a best alternative

8. An optimization technique for use in situations where it is impossible to explicitly enumerate all of the...

Table of contents

• Cover
• Half Title
• Series Page
• Title Page
• Copyright Page
• Dedication
• Table of Contents
• List of Examples
• Preface
• Acknowledgments
• Author
• 1. Introduction
• 2. Problem Structuring
• 3. Making Decisions under Conditions of Certainty with a Small Number of Alternatives
• 4. Goal Programming and Other Methodologies for Multiple Objective Optimization
• 5. A Brief Review of Probability Theory
• 6. Modeling Preferences over Risky/Uncertain Outcomes
• 7. Modeling Methodologies for Generating Probabilistic Outcomes: Decision Trees and Influence Diagrams
• 8. Determining Probabilistic Inputs for Decision Models
• 9. Use of Simulation for Decision Models
• References
• Index