A multiple-objective decision model for the evaluation of advanced manufacturing system technologies

Journal of Manufacturing Systems Volume I l/No. 3

A Multiple-Objective Decision Model for the Evaluation of Advanced Manufacturing System Technologies Johann G. Demmel, Hughes Aircraft Company, Tucson, AZ Ronald G. Askin, University of Arizona, Tucson, AZ

Abstract

turing (CAM), computer-aided design (CAD), computer-integrated manufacturing (CIM), robotics, and flexible manufacturing systems (FMSs)

Ordinary financial measures oversimplify the evaluation of advanced manufacturing system technologies (AMSTS). In this paper, a multiple-objective decision model is developed that avoids the shortcomings of traditional evaluation methods. The model is comprised of three objectives--pecuniary, strategic, and tactical. The pecuniary objective is based upon traditional discounted cash flow (DCF) techniques, with the results normalized to a [-1, + 1] (worst-best) scale. The strategic and tactical objectives are based upon the concept of qualitative flows, and a qualitative discounting method is employed to discount the qualitative costs and benefits to a present value. The three objectives are traded off using the composite programming technique, resulting in a rank ordering of the alternatives under consideration. The three objectives of the model are broken down into attributes that define the objective. These attributes are mapped into the organization of a manufacturing environment. It is shown that the model covers the entire manufacturing organization in accounting for the costs and benefits of the proposed AMST alternatives. In addition to providing a ranking among alternative AMST projects, the influence of the three objectives on the final score can be analyzed using a mixture experiment. The mixture experiment provides insight into the effect of varying the importance of each objective and its effect on the final rankings. This provides the analyst a method to determine which attributes and objectives are critical for the AMST alternative being investigated.

were unknown then. Today, these are common technologies in the manufacturing environment and make up what is called advanced manufacturing system technologies (AMSTs). Investments in AMSTs are difficult to justify using ordinary financial measures. The common discounted cash flow (DCF) measures (net present value, return on investment, and internal rate of return) oversimplify the investment decision. Their inability to account for intangible benefits such as greater flexibility, shorter lead time, and increased knowledge in the use of new technologies makes them inept for most strategic decision making processes. The methods also assume a static environment for the do-nothing alternative. Thus, the justification process requires enhancement to include these intangible factors. 1 Changes in the cost structure of manufacturing and competitive strategies point to the need for improved economic evaluation of AMSTs and cost management systems) These changes include the following conditions: traditional financial techniques are inappropriate for technology planning, direct labor is no longer the driver for production control, cost patterns have changed, and cost accounting systems no longer reflect the manufacturing process. The DCF approach is criticized in Reference 3 because companies tend to set arbitrarily high hurdle rates for evaluating investments into new projects. These rates cause companies to invest in incremental rather than revolutionary projects. As depicted in Table 1, investment trends have a direct impact on productivity. Therefore, in order for the US to maintain its competitiveness, it must find a method to assist in improved evaluation of

Keywords: CIM, Manufacturing System Justification, Multiple Objective, Strategic Planning

Introduction Manufacturing technology has progressed considerably in the past 30 years. Technologies such as numerical control (NC), computer-aided manufac-

179

Journal of Manufacturing Systems Volume l l/No. 3

capital projects. Various work is currently being undertaken in the area of economic justification and planning with relation to AMSTs. Some methods deal specifically with the justification of AMSTs, while others address planning with respect to certain financial measures. A weighting scheme method to perform a net present worth analysis to evaluate CIM opportunities is described in Reference 5. Included are nonmonetary attributes; however, they are not integrated with the monetary measures. In Reference 6, an expert system for use in the justification of FMSs is presented. This system provides a consultant service to evaluate advanced manufacturing technology, where the results are a categorization of alternatives rather than a rank ordering. Multicriteria Q-Analysis (MCQA) is employed in Reference 7 to rank eleven possible automation projects for selection in a printed circuit card manufacturing area. Analysis of flexible versus nonflexible automation is performed using two-stage convex quadratic program, mixed integer programming, and control theory. 7-11 Simulation is used in References 12-14 to analyze the AMST evaluation problem. Another popular evaluation methodology for AMSTs is the analytic hierarchy process (AHP). A methodology is presented in Reference 15 to compare and evaluate manufacturing systems based on a system value model utilizing the AHP. In Reference 16, the AHP is used to analyze tangible and intangible benefits that result from the implementation of an FMS. AHP is used in Reference 17 to arrive at a nonfinancial evaluation method for CIM in terms of risk, benefits, and time.

In this paper, a methodology for the evaluation of AMSTs is proposed. This methodology quantifies relevant factors using cash flows and subjective estimates of system effects. The generic methodology may be tailored to fit the uniqueness of any manufacturing related company. The model captures the individuality of the company by depending upon the company's strategic plan, organization, and operations.

Model Formulation The proposed model takes a fundamental view of the AMST evaluation problem as being composed of three objectives--strategic, tactical, and pecuniary. The pecuniary objective contains quantitative factors and is represented by a monetary index. The monetary index considers the cash flows created by an alternative, and uses the net present value to arrive at a single value, which describes the pecuniary objective. The strategic and tactical objectives represent qualitative factors, each represented by the attributes that describe the respective objectives. These intangible costs and benefits will be analyzed using the concept of qualitative flows, and each objective is represented by a nonmonetary index array. The following sections provide detailed definitions of these objectives. These definitions are provided in a hierarchical manner to provide a clear understanding of what each objective encompasses. The hierarchy is set up in three levels with the objective at the top level, attributes at the next level, and a description of the attributes (elements) at the lowest level.

Table 1 Capital Investment and Related Productivity Increase of Western Countries, 1974-19824

JAPAN FRANCE WEST GERMANY UNITED KINGDOM UNITED STATES

CAPITAL INVESTMENT (as % of output)

PRODUCTIVITY INCREASE

17.1% 13.6% 11.2% 13.0% 11.1%

79% 43% 29% 16% 18%

Strategic Objective The strategic objective relates to the planning performed by a company before any action takes place. The attributes that define this objective are internal relations, market position, mission, organization, public relations, and technology. The strategic objective is a qualitative measure. The attributes of this objective are formulated from the ideas presented in References 15, 18,and 19. Internal relations are the factors that revolve around the people who make up the company--the employees. This attribute is described by issues

180

Journal of Manufacturing Systems Volume l I/No. 3

such as employee morale, employee development, quality of work life, absenteeism, and use of employee skills. The second strategic attribute is market position. Market position refers to how the company views itself and the products it manufactures with respect to the rest of the world. Market position is described by market share, demand, competition assessment, market penetration, vulnerability, and survivability. These issues focus on how the company stands in its market, where it wants to go, and where it can go. Mission is the company's road map for its future. It is represented by the five-year plan used by most companies, and possibly even a longer range plan (10 + years). The issues under the mission are how well the company is performing compared to the current plan, evolution of the company, economic feasibility of the goals, and a focused factory. Organization defines the structure of the company. This attribute describes how well the company controls its actions, how responsive it is to changes, and the general management of information. Issues such as duplication of functions and barriers between departments are included in this objective. Public relations is an attribute that transmits the operations of the company to the outside world. This attribute is described by the image of the company, company prestige, and service. The final attribute of the strategic objective is technology. This attribute deals with the position the company is in relative to AMSTs, and where it would like to be. Issues such as scientific information yield, technological position, and availability describe technology.

Flexibility deals with how well the company reacts to changes. This attribute is defined by issues such as versatility, response to change, batch size, lead-times, and throughput. AMSTs are characterized by their flexibility. Flexibility can be thought of in aggregate terms such as low, medium, or high. Low flexibility is defined as having a series of stand-alone machining cells that are not interconnected. Each cell can only handle a specific part family. Medium flexibility carries the interconnection a step further and connects the machining cells with conveyors or tow lines. Rudimentary control logic is employed, which optimizes part routings through the cells. Some cells take on the capability to handle more than a single part family, yet some operator setup may be required. High flexibility uses interconnected, general purpose cells that can be controlled in real time. The integration attribute concerns interaction among the operations and departments within a company. Issues such as communication, duplication, synergism, data requirements, and data usefulness describe integration. Material deals with the control and movement of product through the manufacturing environment. Issues such as control, scheduling, planning, expediting, and handling encompass the material attribute. The personnel attribute deals with the people aspect required to carry out the operations of the tactical objective. Skill requirements and training of the people, along with the type (direct/indirect) of people required for the task are delineated. Human factor issues such as workplace design and employee safety are included. The final attribute of the tactical objective is producibility. This attribute deals with how the product is made; the issues involved include consistency, compatibility, feasibility, reliability, capacity, and external interface.

Tactical Objective The tactical objective includes those qualitative issues that arise due to the actions or operations of the company. They are the actions taken that serve the purpose laid out by the strategic objective. The attributes that describe this objective are design, flexibility, integration, material, personnel, and producibility. These attributes are derived from References 3, 15, 18, 19, and 20. The design attribute identifies those aspects of the company that are involved in the design of a product. Design is defined by issues such as efficiency, features, lead times, and standardization.

Pecuniary Objective The third objective addresses the financial attributes (costs) involved in the AMST decision, and will be referred to as the pecuniary objective. The attributes that make up the pecuniary objective are operation and maintenance, plant and equipment, and product. This breakdown follows the broad categories presented in Reference 21.

181

Journal of Manufacturing Systems Volume l I/No. 3

Table 2 Objectives of the AMST Evaluation Problem

Operation and maintenance are the costs related to the operations of the company. Included in this attribute are operating labor, maintenance labor, overtime, labor turnover, absenteeism, training, direct and indirect costs (including health care and savings incentive benefits), supervision, setup, maintenance tools and supplies, production rates, routings, shop floor control, expediting, interrupted production, insurance, and documentation. The plant and equipment attribute includes the costs incurred due to the physical resources of a company. The attribute is made up of the following costs: equipment, startup, installation, tooling, hardware development (including modification of existing equipment for compatibility with AMST equipment), software development, spare parts, space, safety equipment, energy, depreciation, and taxes. The final attribute of the pecuniary objective is product. These are the costs directly related to the product(s) the company manufactures. Examples of costs included under the product attribute are design changes, inventory, engineering, quality, and sales.

OBJECTIVE/ Attribute STRATEGIC Internal Relations

Market Position

Mission Organization Public Relations Technology

TACTICAL Design Flexibility Integration

Material

Personnel Produeeability

ELEMENTS

employee morale, employee development, use of skills, quMity of work life, compensation, absenteeism. market share, competition assessment, market penetration, survivability, vulnerability, demand. match to plan, evolution, economic feasibility, focused factory. control, responsiveness, information management. company image, company prestige, service. scientific information yield, technological position, and availability.

et~ciency, features, lead-times, standardization. versatility, response to change, batch size~ lead-times, throughput. communications, reduced duplication, synergism, data requirements, data usefulness. process control, scheduling, part tracking, shop floor control, cost tracking, expediting, handling, planning. skill requirements, training, safety, direct/indirect, human factors. consistency, compatibility, feasibility, reliability, capacity, external interface.

PECUNIARY Operation & Maintenance operating labor, maintenance labor, direct/ indirect costs, absenteeism, training, supervision, insurance, overtime, labor turnover, setup, maintenance tools/supplies, production rates, documentation, routings, shop floor control. Plant & Equipment equipment, startup, installation, tooling, spare parts, energy, space, safety equipment, hardware development, software development, depreciation, taxes. design changes, inventory, quality, Product engineering, sales.

Model Completeness Table 2 lists the three objectives used in this model, along with the attributes of each objective and the elements that describe each of the attributes. These objectives represent a high level breakdown of the company. There are monetary (pecuniary) and nonmonetary considerations broken down by planning (strategic) and implementation (tactical). Before presenting a proposition that provides the basis for using the objectives listed in Table 2, the following definition is required: • Definition: A typical manufacturing enterprise is represented by the functions and subfunctions depicted in Figure 1. The functions are based upon organizational hierarchies presented in References 22 and 23, and the authors' professional experience. Additional insight was provided by the idealized organizational model of Reference 24. Figure 1 is a representation of the functions found in a typical manufacturing enterprise. The structure of the functions may be allocated differently for various organizations; however, these functions will be present. A brief description of each of the functions is provided in the following paragraphs.

The executive function represents the top level in a company. It is within this function that plans for the future of the company are made, along with keeping the current operations within previously defined plans. Human resources deals primarily with the recruitment and training of people required to operate the company and support the manufacturing operations. Determining future demands and customer relations are part of the marketing function. The financial function is best described in Reference 23 as a scorekeeper of the company, "to see how well the firm and its component departments are scoring in the business competition game." Procurement deals with the acquisition of material from outside sources. Maintaining the existing equipment and plant site falls under the

182

Journal of Manufacturing Systems Volume I I/No. 3

Rannlng t Strataglc Policy Dec~aions

I Personnai Development Recruitment Training Health, Safety Wage Administration Labor Relations

fi

FBudg,"ng

~-Sales ~-Future Demand ~-Advertising L CustomerRelations Customer Service

I

l fpRODU,SSURA.OSai,

AO, Ur,ES ,

, : Equipment Installation Maintenance& Repair

~

Energy Building/Site

F~,~ui~ Re~

~-Acquire Equipment L Vendor Certification

~-Investment Review L Cost Data

: . ~Reliability

I

sYSi

[

I

I- Support Engineering ~- Production Control ~- Matedai Handling ~- Production I- Inventory L Tooling

~-Data Processing ~-Data Management L Communications

I

1

~'New 'Pr'oduct . . . . . L New Technology

t Product Design Product Enhancement Bill of Material

Figure 1 Organizational Structure of a Manufacturing Enterprise

facilities function. This function also has responsibility for installing newly acquired equipment. The product assurance function monitors manufacturing operations to ensure that the products produced meet the required specifications. Management information systems deals with the handling and reporting of information required by the company. The heart of the manufacturing enterprise is manufacturing operations. It is within this function that the actual product is built. The subfunctions for manufacturing operations include production control, material handling, inventory, tooling, production, and support engineering. The support engineering subfunction includes industrial and manufacturing engineering. The research and development function is where new technologies are investigated. Finally, design engineering deals with actually specifying the components required to build a product and how the components fit together.

Proposition: The pecuniary, strategic, and tactical objectives allow all of the costs and benefits associated with AMSTs to be evaluated. TEST: To confirm that the objectives allow all of the costs and benefits to be evaluated, the objectives will be related to the organization of a typical manufacturing enterprise. We define a cover of the enterprise to be a set of measurable objective elements such that each subfunction is associated with at least one element of the set. If the objectives cover all of the functions in the organization, then we have a feasible scheme of accounting for all of the costs and benefits. The traditional monetary costs are captured in the pecuniary objective. The intangible issues are accounted for in the strategic and tactical objectives, which address the planning and implementation aspects of the company as a whole. The coverage test is performed by identifying an association of the elements in Table 2 to the

183

Journal of Manufacturing Systems Volume l l/No. 3

subfunctions of the manufacturing enterprise that encompass the element. This covering is shown in Figures 2-4. The minimal aspect of the cover is shown by the fact that all objective elements have an associated subfunction. Note also that each element of Table 2 is associated to only one subfunction, representing the primary subfunction to which the element contributes costs and benefits. This facilitates data collection and reduces ambiguity in element definition. Some subfunctions are associated with more than one element. Nonetheless, these elements describe complementary aspects of the subfunction and do not duplicate costs and benefits. All of the elements of our three objectives are needed to describe the entire firm and all subfunctions of the firm are considered in our multi-objective model. For further details regarding the test, see Reference 25. In addition to providing an integrated quantitative and qualitative analysis, we believe this demonstration of coverage makes the following methodology an improvement over most existing methodologies.

Human

Resources

i

Design

Rexibility

Marketing

Financial -- Integration Procucement

Facilities

Product Assurance

Manz=gement Information

-- Pe/sonnel

Systems

Manufacturing Operations

--

Producibility

Research & Development

Design Engineering

Figure 3 Covering of Tactical Objective

I

L__

J

Ptanning Policy Decisions

Strategic

Executive

StrategicRanning Policy Decisions

Per

Pe~x~nnelDevelopment

Human

Resources

Human

Resources

i

Marketing

--

f

Operations

Marketing

:°Lo,!Facilities

~

Manufacturing

Gperations --

Engineering

I

M

FaciiRies

Da~ F Data Ma ~Communications .~ Sup~eering

Intarmatio~ Systems

Design

Procurement

Reliability

Management

Research & Development

~nancial

~uaHty

ProduCt A.ssurance

Production Control Material Handling Produc~on

,Responsbveness ~ -information m Manage _ ent

Organization

~Company Pfestlge ~$ervlce ,~cientilic Information ITechnological Position

,Availabi~i~

Plant and Equ,pmen~

Product t /~surance Management

=CompanyImage

Information Systems

Public Relations Technology

Manufacturing Operations

-__

an~/ Maintenance

Research & Development Product Design Product Enhancement Bill ct Material

Design

Engineering

Figure 4 Covering of Pecuniary Objective

Figure 2 Covering of Strategic Objective

184

Journal of Manufacturing Systems Volume I l/No. 3

An additional advantage is that it is structured along the organizational functions of a typical manufacturing enterprise. This allows the method to be tailored to fit a specific organization's structure, including those attributes and elements that are important to the organization and eliminating those attributes and elements that are not applicable. The proposed structure resembles that presented in Reference 15. Our tactical objective encompasses their capability and performance attributes. These are factors that are small scale actions serving a larger purpose. The strategic objective parallels the suitability attribute. The productivity attribute is similar to the pecuniary objective. The structures differ in that we clearly divide monetary and nonmonetary issues. Also, our model more closely matches organizational structure and its completeness has been documented. Finally, we will present an evaluation and sensitivity analysis procedure based on composite programming.

(best) scale. This is required for the preference ordering proposed later. A range is determined based upon the maximum and minimum NPV of all alternatives, and each NPV will be scaled by this range. The monetary index for a particular alternative j is calculated by: mlj =

RMt = (max(O,maxNPVj)-min(O,minNPVj) ) j~J j~,l • T h e o r e m : The m o n e t a r y i n d e x M I j NPVj --yields a result in the range [-1, + 1].

=

RMt

The proof of this result is provided in Appendix A. The greater the monetary index for a particular alternative, the better that alternative meets the decision maker's value. Alternatively, the decision maker can define target minimum and maximum NPVs. In this situation, the target values must be at, or exceed, minNPVi and maxNPVj. j~J jd

The monetary index of this model will use traditional net present value (NPV) methods to set up a cash flow analysis for each alternative under consideration, including the do-nothing alternative. The interest rate used will be the true cost of capital to the firm. In general, the interest rate used for NPV has been cause for much discussion. Companies tend to set an arbitrarily high interest rate. 3 Interest rates for several corporations have been determined in Reference 26, and of the 14 manufacturing firms identified in that paper, the interest rate averaged between 5% and 8%. Let Yo be the after tax cash flow of alternative j(j = 1 . . . . . J) in period t(t = 1 . . . . . T) and it be the interest rate in period t. The NPV is determined by:

Nonmonetary Evaluation (Strategic and Tactical Objectives) Where the pecuniary objective resulted in a single attribute of the monetary index to describe the objective, the nonmonetary evaluation will arrive at indices for each of the attributes of the strategic and tactical objectives. For each alternative, we begin by specifying a qualitative flow rating, by time period, for each attribute. These ratings will then be combined into a net present qualitative flow (NPQF) value for further analysis. • Definition-Qualitative Flow: The expected performance of alternative j f o r attribute k(k = 1, .... K) in period t is Qtkj, with -100 < -- Qtkj <+ 100. -100 is the worst rating, 0 is a neutral rating, and + 100 is the best rating. For each alternative j, the decision maker rates the expected qualitative flow Qa,j. The aforementioned definition extends the methodology of Reference 5 to permit dynamic effects and negative evaluations. The decision maker may prefer to

T

Y,j(l+i,)-'

(2)

where:

Monetary Index (Pecuniary Objective)

N P Vj = Z

NPV)

(1)

Monetary Index Formulation

An index for the monetary analysis will be required to scale the results into a dimensionless value that can be compared on a -1 (worst) to + 1

185

Journal of Manufacturing Systems Volume I l/No. 3

Composite Programming Solution

specify a unique Qtks for each element, because he/she may have differing views for each of the elements that describe an attribute. This provides the decision maker with some flexibility in the level of detail used in the model.

The structure of the proposed AMST evaluation problem lends itself to a solution using composite programming, z8 Composite programming (a form of compromise programming29) is a two-level tradeoff analysis. The first level is a trade off between the attributes and elements of an objective, and the second level is a trade off among the objectives. Composite programming lends itself well to this situation, and to our knowledge it has not been previously applied to the AMST evaluation problem. The formulation for evaluation of AMST using composite programming2s is:

• Definition: The NPQF of attribute k for alternative j is: T

N P Q Fkj = Z

Qtkj (1 + ht) -t

(3)

t=l

where ht qualitative discount rate indicating a decision makers' impatience for benefits, z7 Generally, k = 1 . . . . . 6 denotes attributes from the strategic objective and k = 7 . . . . . 12 denotes attributes from the tactical objective. =

min pj = J

N o n m o n e t a r y Index F o r m u l a t i o n A nonmonetary index will be required to scale the results into a dimensionless value that can be compared on a -1 (worst) to + 1 (best) scale. For attribute k of alternative j, this index will be determined by:

+

[3n n=2

keS ,,

where Composite evaluation of alternative j. = Weighting factor for the importance of an objective, n = 2: strategic, and n = 3: tactical. ant, = Weighting factor for the importance of the attributes and elements of an objective. S, = Set of attributes and elements for objective n. p = Parameter reflecting the distance measure for importance of the attributes and elements from the ideal (L-norm). q = Parameter reflecting the distance measure for importance of the objectives from the ideal (L-norm). =

13n

(4)

where T

RN1 = 1 0 0 ~

oql

(5)

oj

NPQF&i Nlkj = RM

[31

( l + h t ) -t

t=l

• Theorem: The nonmonetary i n d e x N l k j = NPQFkj yields a result in the range [-1, + 1]. RNI

The proof of this result is similar to that for the monetary index. To summarize the steps to obtain the NPQF, each appropriate element suggested in Table 2 under the strategic and tactical objectives will first be rated on performance (Qtkj) by the decision maker for each period t. Next, the NPQF is determined per Equation 3. This is performed separately for each attribute, or for each element. The decision maker then determines the Nlkj. The result is a vector of Nlkj for each alternative j being evaluated.

The resulting composite evaluation is a score for each alternative that describes the distance that the alternative is away from an ideal solution. The ranking of the composite evaluation scores is based upon the min PS being the best alternative from the s set of alternatives being evaluated. The weights O~nkindicate the importance of the attributes and elements for an objective as compared to the other attributes and elements for the same

186

Journal of Manufacturing Systems Volume I l/No. 3

objective. The weights [3. indicate the importance of the objective n to the overall result. These weights are determined by the decision maker on a scale of 0 to 100, and normalized to 100. This is done within each objective for a,t,, and across the objectives for 13,. The weight allows the decision maker to define the importance of the attributes and elements to obtain an accurate measure of their effects upon the specific objective, and the importance of the objectives to the overall composite evaluation. Similar weighting techniques are employed by References 5, 18, and 30. In composite programming, p (and also q) has special meaning. A p value of 1 signifies group utility in the decision-making process, or simple majority rule. The group in this case is the set of attributes. Setting p -- 1 emphasizes the sum of individual utilities, instead of any single regret. A p value of 2 signifies a move toward individual utility, or more emphasis on the individual regret of choice. When p equals ~, the interpretation is complete individual utility, or most emphasis on the largest individual regret of choice.29

Table 3 Current Operation Cash Flows Labor Mainten~ce

Training Documentation

Space

Cash Inflow

Maintenance

Training Documentation Space Rework Transportation

Cash Inflow

1 ($109,200) ($2,000) ($1,000) ($5,000)

2 ($119,600) ($2,000) ($1,150) ($5,000)

3 ($130,000) ($2,200) ($1,323) ($5,000)

4 ($140,400) ($2,200) ($1,521) ($5,000)

$0 $0 $0

(175,600) ($10,000) 1816,000

($84,000) ($10,000) $912,000

($92,400) ($10,000) $998,400

($100,800) ($10,000) $1,094,400

$0

Rework Transportation

Labor

O $0 $0 $0 $0

5 ($150,800) ($2,420) ($1,749) ($5,000) ($10,000) ($113,400) ($15,000) $1,228,800

($s,ooo)

6 ($161,2(X)) ($2,420) ($2,011) ($5,000) ($10,000) ($121,800) ($15,000) $I,315,200

($5,000)

7 ($171,600) ($2,662) ($2,313) ($5,000) ($15,000) ($134,400) ($15,000) $1,459,200

($5,000) ($10,000)

8 ($182,000) ($2,662) ($2,660) ($5,000) ($15,000) ($147,000) ($15,000) $1,593,600

9 10 ($192,400) ($202,800) ($2,928) ($2,929) ($3,059) ($3,519) ($5,000) ($5,000) ($15,000) ($15,000) ($163,800) ($180,600) ($20,{X}0) ($20,000) $ 1,776,000 $1,958,400

Table 4 Semi-Automatic Alternative Cash Flows

Depr Investment Nondepr Investment SMv&gevaiue Labor Maintenance Training Documentation

O ($120,000) ($30,000)

1

2

($270,400) ($228,800) ($6,660) ($6,660) ($20,000) ($20,000) ($2,500) ($2,500) ($6,250) ($6,250) ($75,600) ($75,600) ($50,000) ($50,000) $806,400 $921,600

Sp~e Rework Prograxrmting

C~h Inflow

3

($291,200) ($6,660) ($15,CO0) ($2,500) ($6,250) ($73,920) ($30,000) $1,927,200

4

($291,200) ($6,660) ($15,000) ($2,500) ($6,250) ($70,560) ($20,000) $1,132,800

Depr Investment Nondepr Investment

Salvage value Labor Maintenance

Training Documentation Space Rework

Numerical Example

Pr oKr~rcm~ng

Cash Inflow

To demonstrate this technique, a numerical example is provided. The intent of the example is to show how the qualitative issues can impact the evaluation. This example deals with a hypothetical electronics manufacturing company that is evaluating two alternatives over its current operations of component insertion. The current operations are entirely manual, and are performed offshore due to attractive labor rates. Two alternatives deal with bringing the operations back to the US, the other with leaving the operations offshore and simply adding people. Alternative one employs semiautomatic component insertion machines, the second uses a totally automatic component insertion machine. The cash flows for each alternative are shown in Tables 3-5. A project window of ten years is used, with an interest rate of 8% and an inflation rate of 5%, yielding a net interest rate of 13.4%. Currently, the offshore operations utilize 8 people to perform component insertion. An additional operator and engineer will be needed each year to keep up with the expected demand. Due to the

($291,200) (1353,6CO) ($353,600) ($6,660) ($6,660) ($6,660) ($10,000) ($15,000) ($10,000) ($2,500) ($2,500) ($2,500) ($6,250) ($6,250) ($6,250) ($68,040) ($60,900) ($60,480) ($20,000) ($20,000) ($20,000) $1,276,800 $1,372,800 $1,526,400

(1353,600) ($6,660) ($10,000) ($2,500) ($6,250) ($58,800) ($20,000) $1,670,400

($353,600) ($6,660) ($10,000) {$2,500) ($6,250) ($65,520) ($20,000) $1,862,400

$5,500 ($353,600) ($6,660) ($10,000) ($2,500) ($6,250) ($72,240) ($20,000) $2,054,400

Table 5 Automatic Alternative Cash Flows

Depr Investment Nondepr Investment SM~ge vMue Labor

Mainten~ce

Traini~ Documentation Space Rework

Pro~ammnig C~h Inflow Depr Investment Nondepr Investment Salvage value Labor Maint en~nce

Training

Documentation Space Rework

Prograxamnig

Cnsh Inflow

0 ($600,000) ($50,000)

1

($364,000) ($16,000) ($40,000) ($5,000) ($6,250) ($75,600) ($50,000) $777,600

2

($353,600) ($16,C00) ($40,000) ($3,000) ($6,250) ($75,6CO) ($50,000) $931,200

($301,600) ($301,eO0) ($301,600) ($16,000) ($16,000) ($16,000) ($20,000) ($20,0~) [$20,600) ($3,000) ($3,000) ($3,000) ($6,250) ($6,250) ($6,250) ($56,700) ($30,450) ($33,600) ($90,~300) ($20,000) ($20,G00) $1,286,400 $1,382~400 $1,526,400

3

($301,600) ($16,000) ($30,000) ($3,000) ($6,250) ($64,680) ($30,G00) $1,036,800

4

($301,600) ($16,000) (120,000) ($3,000) ($6,250) ($50,400) ($20,000) $1,132,800

($301,600) ($301,600) ($16,C00) ($16,000) ($20,000) ($20,000) ($3,000) ($3,000) ($6,250) ($6,250) ($29,400) ($32,760) ($20,000) ($20,000) $1,670,400 $1,862,400

$30,600 ($301,600) ($16,000) ($20,000) ($3,000) ($6,250) ($27,090) ($20,000) $2,054,400

increase of employees, space requirements increase over time. There is also a cost incurred in transporting the circuit boards back to the US. The remaining costs presented in Table 3 are self explanatory.

187

Journal of Manufacturing Systems Volume I l / N o . 3

The alternative of semi-automatic insertion equipment will require purchasing four semiautomatic insertion machines to replace offshore operations. These machines will require site preparation and miscellaneous peripherals. The equipment is depreciated according to the accelerated cost recovery system rate for seven year equipment. In the first year, the offshore production will be maintained in full while the semi-automatic machinery is installed and placed into the manufacturing environment. In the second year, the offshore operation will begin to be phased out, and by the third year, the offshore operations will be eliminated. Additionally, operators for two of the machines will be required. A third operator will be added in the third year, and a fourth operator in the seventh year. The idle machines will be used for programming and training. A full time support engineer will also be required. The rework rate of boards from this operation are anticipated to decline over the life of the project. Other costs presented in Table 4 are self explanatory. The fully automatic component insertion option will require one machine with appropriate site preparation and peripherals. This equipment is also depreciated over seven years. In the first year, offshore production will be maintained in full while the automatic equipment is installed and placed into the manufacturing environment. One operator and engineer will be required starting in the first year. In the second year, the offshore operation will be eliminated and two people will be required to operate the automated machinery with an additional support engineer. In the third year, only one and a half support engineers will be required, and beginning in the fifth year only one support engineer will be needed. The rework rate of boards from this operation is anticipated to improve over the semiautomatic alternative. Other costs are shown in Table 5. Next, the nonmonetary attributes of the situation are examined. Performance of each of the alternatives was performed on the attributes of the strategic and tactical objectives. The qualitative flows are shown in Tables 6-8. These tables show in general that current operations provide for a declining rating of the nonmonetary attributes. For example, internal relations deteriorate as pressure is applied from within the

Table 6

Current Operation Qualitative Flows Strategic Internal Relations Market Position Mission Organization Public Relations Technology

1 15 20 -30 -50 20 -90

2 0 15 -30 -50 0 -90

3 -30 10 -35 -50 -10 -90

4 -30 0 -45 -50 -20 -90

5 -30 -25 -50 -50 -30 -90

6 -30 -45 -55 -50 -40 -90

7 -50 -65 -60 -50 -50 -90

8 -50 -75 -65 -50 -60 -90

9 -80 -80 -75 -50 -70 -90

10 -80 -85 -85 -50 -80 -90

Tactical Design Flexibility Integration Material Personnel Produceability

1 0 -40 -15 25 35 -20

2 0 -40 -20 15 30 -25

3 0 -40 -25 0 30 -30

4 0 -40 -30 -20 25 -35

5 0 -80 -35 -20 15 -40

6 0 -80 -40 -30 5 -45

7 0 -80 -50 -30 -10 -50

8 0 -80 -60 -40 -15 -55

9 0 -80 -70 -40 -25 -60

10 0 -80 -80 -50 -35 -65

Table 7

Semi-Automatic Alternative Qualitative Flows Strategic

10

Technology

1 -35 10 30 -35 25 10

2 -25 -20 30 -10 25 10

3 5 0 30 10 35 25

4 10 5 35 15 35 30

5 20 15 35 15 45 40

6 30 25 35 20 45 40

7 40 35 40 25 55 40

8 40 35 40 40 55 40

9 60 55 40 40 55 40

60 55 50 40 55 40

Tactical Design Flexibility Integration Material Personnel Produceability

1 0 -20 -15 -10 35 -25

2 0 -5 0 10 35 -15

3 0 0 5 15 40 5

4 0 5 10 20 40 15

5 0 15 15 25 45 25

6 0 20 20 30 45 35

7 0 30 25 40 55 45

g 0 30 30 50 55 45

9 0 40 35 50 55 55

10 0 40 40 50 55 55

Internal Relations

Market Position Mission Organization Public Relations

Table 8

Automatic Alternative Qualitative Flows Strategic

Technology

1 -35 10 40 -35 45 35

2 -25 -20 40 -10 45 45

3 5 10 60 10 55 45

4 10 15 60 15 55 60

5 20 30 80 25 65 60

6 30 45 80 45 65 80

7 40 80 90 65 75 80

8 40 80 90 85 85 80

9 60 80 90 95 85 80

I0 60 80 90 95 85 80

Tactical Design Flexibility Integration Material Personnel Produceability

1 20 -20 -15 -10 -20 -35

2 20 0 0 10 0 -25

3 20 10 5 15 15 0

4 20 20 20 25 30 40

5 20 30 30 35 45 80

6 20 50 45 45 55 80

7 20 80 55 60 70 90

8 20 90 80 75 90 90

9 20 90 80 75 90 90

I0 20 90 80 75 90 90

Internal Relations

Market Position Mission Organization Public Relations

company to bring work back to the US to improve internal control. The company has a plan for modernization and automation, of which the offshore operations do not match. Organization is affected by poor information management between the offshore operations and the company in the US. There is

188

Journal of Manufacturing Systems Volume l l/No. 3

some impact on design, as the density of components on the board has to allow for manual operations, Flexibility is poor and integration is almost impossible. It is anticipated that material will become a problem unless some updated controls and communications are established between the US and the offshore operations. Producibility is bad because of high rework and consistency problems reported by the offshore operations. The semi-automatic alternative's nonmonetary performance shows that internal relations are anticipated to be poor in the first year due to the new machinery and changes in operation. The market position should improve, as the company will be poised to better compete. This alternative is a better match to the mission of modernization. Bringing back offshore operations will look good from a public relations view, and the technology is better than the current operations. Flexibility should improve. Similarly, integration, material, and producibility will have expected difficulties during the start up, and then improve with time. Personnel will be average during the start up, and is also expected to improve. Internal relations for the automated alternative are anticipated to be poor in the first years again due to the new machinery and the changes in operation. Due to the complexity of the machinery, this will last a year longer than in the semi-automatic alternative. The market position should improve. Nonetheless, as in the semi-automated case, the transition will cause some problems. This alternative is the best match to the mission of modernization. Bringing back offshore operations will look good from a public relations view, and the use of automatic machinery should improve this over the semiautomatic equipment due to its higher technology. Flexibility should improve with time. Integration, material, and producibility will also improve with time. Personnel will be poor during the start up due to the complexity and frustrations associated with a new learning process, and then is expected to improve. Having gathered the aforementioned information, the decision maker (company management) is requested to assist in determining the maximum and minimum NPVs to be used in the analysis. NPVMAx is $5,000,000 and NPVMIN is $0. Using the cash flows presented in Tables 3-5, the NPV of each

alternative is determined. The monetary and nonmonetary indices are then used to create the attribute versus alternative matrix shown in Table 9. A value of 2 is used for p and q. All of the calculations in this example were performed using a software p r o g r a m d e v e l o p e d for this effort entitled "AMSTEPD--Advanced Manufacturing Systems Technology Evaluation Program, Deterministic." Program details are available from the first author. Next, the decision maker is asked to assign weights to each of the attributes within an objective. Since the pecuniary objective has only one attribute, the monetary index, its weight is 100 by default. The weights assigned to the attributes of the strategic and tactical objectives are shown in Table 10, along with the normalized weights. Additionally, the decision maker is requested to assign weights to the three objectives by importance. These weights and their normalized result are shown in Table 11. Results of the composite programming technique are summarized in Table 12. The automatic insertion machine is preferred to the semi-automatic and current operations but the automatic and semiautomatic alternatives are relatively close. Based entirely upon the monetary evaluation, the decision would have been the current operations. Yet, when the nonmonetary attributes of the alternatives were included in the evaluation, the current operations were no longer a contending alternative and the semi-automatic and automatic insertion machinery were in a more favorable light. Table 9 Attribute versus Alternative Matrix

ATTRIBUTE Monetary Internal Relations Market Position Mission Organization Public Relations Technology Design Flexibility Integration Material Personnel Produeeability

189

MANUAL

SEMI-AUTO

AUTOMATIC

0.6384 -0.3049 -0.2442 -0.4919 -0.5000 -0.2733 -0.9000 0.0000 -0.6026 -0.3810 -0.1393 0.1058 -0.3936

0.5394 0.1382 0.1700 0.3533 0.1126 0.4052 0.2916 0.0000 0.1132 0.1298 0.2390 0.4431 0.1812

0.4429 0.1382 0.3380 0.6792 0.2962 0.6276 0.6096 0.2000 0.3548 0.3069 0.3421 0.3829 0.3998

Journal of Manufacturing Systems

Volume I I/No. 3

Table 10 Attribute Weights

ATTRIBUTE

WEIGHT

NORMALIZED

Monetary

100

100

Internal Relations Market Position Mission Organization Public Relations Technology

80 95 90 70 60 85

17 20 19 15 13 18

Design Flexibility Integration Material Personnel Produceability

75 95 90 75 80 95

15 19 18 15 16 19

sis. An immediate question that arises is the effect of the weights e~ and [3 on the final result. To analyze the effect of changing the weights, a mixture experimental was performed on the eLs and 13s independently. The object of this experiment is to measure the effects of changing the proportions of the components (or and [3) on the response (final score). In accordance with mixture experiment terminology, e~ and [3 were normalized to sum to unity. Additionally, for the mixture experiment, the value of q in the composite programming equation was changed from 2 to 1 to create a linear equation in the objectives. Initially, the analysis was performed on the objective weights (13n, n = 1,2,3). The [3ns were varied between 0 and 1, while the e~nk values were used as shown in Table 10. A simplex-centroid design for three components was used to analyze the effects of varying the proportion of [31, 132, and [33 used in the model. This was done for the three alternatives previously described. Using the observations, the fitted third-degree model for each alternative is:

Table 11 Objective Weights

OBJECTIVE

WEIGHT

Y manual NORMALIZED

=

1.8131 + 7.4[32 + 6-4[33 + 8.8131132 "~ 1.6131133 + 0132133 + 1.2[31132[33

:9 s e m i - a u t o . = 2"3131 + 3"8[32 + 4"1133 + Pecuniary Strategic Tactical

90 80 80

36 32 32

1.8131132 + 2.0131133 + 0.2[32[33 + 4.2131132133

33 a u t o m a a c

Table 12 Final Results and Rankings

ALTERNATIVE

EVALUATION

RANK

Manual Semi-Automatic Automatic

56 35 30

3 2 1

----- 2"8131 + 2"9132 + 3"3[33 + 2.2131132 + 2.2131133 + 1.2132133 + 2.1131132[33

where 33 is the estimate of the composite programming score p. Consider the large coefficients of 132 and [33 for the manual alternative. The inclusion of the strategic and tactical objectives in the model worsens the score and makes the manual alternative less attractive. The coefficients of the semi-automatic and automatic alternatives are more balanced, indicating a greater balance between the objectives. The estimated shape of the surface for 131, [32, and [33 is approximated pictorially in Figures 5-7 by a contour representation 32 of the surface. The contour plots are based upon the fitted third-degree model for each alternative. Predicted scores of equal

Effects of the Weights One advantage of the proposed use of composite programming on the three-objective AMST evaluation problem is the availability of sensitivity analy-

190

Journal of Manufacturing Systems Volume l I/No.

magnitude are joined by the same contour line. Note that X1, X2, and X3 on the plots correspond to 131, 132, and 13a, respectively. The contour plot of the manual alternative (Figure 5) shows that the best score occurs when (131, 132, 133) = (1, 0, 0). As [32 and 133 are increased in value, the score moves further from the ideal of zero reaching a maximum at around (131,132, 133) --- (0.2, 0.8, 0). The slope of the contour is large compared to the other alternatives, moving from a score of 1.8 at (1, 0, 0) to the maximum of 7.7 at (0.2, 0.8, 0). This shows the effect the strategic and tactical objectives (132 and 133) have on the evaluation. The contour plot of the semi-automatic alternative (Figure 6) is more robust as the score varies from 2.4 at (1, 0, 0) to a maximum of 4.13 at (0.12, 0.22, 0.66). Again, the lowest score is achieved when only the monetary objective is used; however, the slope of the contour is less than the slope exhibited for the manual alternative. Whereas the peak of the manual alternative was near the point where the weight of the strategic objective dominates, the semi-automatic plot achieves its maximum closer to where the weight of the tactical objective dominates. Finally, the automatic alternative contour plot (Figure 7) varies by only 0.72, indicating that this alternative performs well in all three objectives. In addition to the contour plots, a comparison of the scores for the three alternatives is presented in Figure 8 for various combinations of 13,. The combinations of [3,, used in the plot are shown in Table 13 and represent the following variations of 13,:

i

SYMBOL SCORE - ..... +...... A 2.00

.X. . A. .

+

B

B+

.

+B .1 O.-C-9O---O 0 , . C . B, • 0. * C. . . 0. +. C . E. E .0, C .Z O. - - - - E - - - B O - O ....... Co .

F

E F

•

0.

3.00

D

3 5 0

E

4.00

H .

.

B.

J -J--.

+

•

.

......

9,"

.+ + .

.

.

L

L .L.

. . . . .

÷

.

L-.

L+

.

.

+

.

.

.

.

.

.

K. . . . . . .

.

.

.

.

K. . . . . . .

.

i ~-

H. .

+

.

.

.

. I .....

.3

+

.

.

.

. . . . . . . .

.

J

*

J .

-

÷

K K

.

.

.

.

.

.

.

J

.

.

9

.

.......

.

.O

+

.

_/ T : ÷

.

.

I

+

. . . . . . . .

.

. . . . . . . . .

.

'.

.

J

.....

.

O

I I.

.

.

.

.8

.

*

"

-

. [ . . . . . . . .

J "

H.

I

. . . . . . .

.

H---.0

H

+ . . J

.7

....

.

I

J

.

.

H

........ I

J

+

~

"

. . . I

+

....

+

.B G.0

H

.

.

.

.J

K-.---1O-

K..

.

l

I .

+. +

K

.

G

.

+

+

R

. .

.

.

.

.

.

'.

. . . . . . . .

.+. .+. 20

K

. . . . . .

.

.

".

.G .

. . . . . . . .

+l .

,

+ J +

+K

.

.

..

.

+++

K

, .

.

.........

.

+

. . . . . . .

+.

+H

.---30----

5 FO

"6+a+."

J + + . .

+.

tt..

•

•

.

F

.

G. . . . . . .

H

+

+..

+

.

.........

•

0.-

.

.

. . . . . . .

K.

.

.

.

t-,

•

.

.

L L.

8." O.-

K.

,

J+. J

. . .

K.-

.

G - . . . . . .

H.H

.+.1

.

K.

. . . . . . .

,

.L

•

K.

.

. . . . . .

.t

f

5-----40--q-i---.-----H--~

'

.

,

O.-L

.

.

~

.4

EO F.

. 6

.1+.

.J

. K.

/.L .

+H

.

---50---H

+ 3.

E ..........

'.

. l-

J

.

fi .

.f'l

+

K

O.------K-

"

l

.......

,

[

F

.

+

7.00 1.50

.

.F . . . . . .

G

H+

JJ'.

+

K

B.

H

,

.

O.----

)~ •

I. I

.

B. 50

K

E F

.÷.

J

.3

F.

.G.

H

6.00

00 .

E

H-----60-G .......

I

• .J

O

.

+.

H -

5.50

1

.

E

.

.F

500

H

t

.

-E . . . . . .

G

.

.

I

.

O

.E

+F +

.

4 5 0

G

. .

+

.F G.. . 6

F

0.

----G---+---7O---H.

o.--I ..... •

O .

+E.

6

3.

.H •

4

.+. E

.

.F

2.50

C

1

.

3

.

d

.

J

. J

. h,

Figure 5 Contour Plot of I~. for the Manual Alternative

weight sets 3, 9, 12, and 16. These weight sets are where 131 is at its highest values (1.00, 0.67, 0.67, and 0.50, respectively). Next, a similar sensitivity analysis was performed on the attribute weights ((x,k, n = 2,3; k = 1 . . . . . 12). The ot,,t,s were varied between 0 and 1, while the 13n values were used as shown in Table 11. A simplex-centroid design for six components was

SYMBOLI BCOR6 ---;, ..... ~:;

•i + + +

m~ 1.

.

. .1

O.---BO----.o •

. C +

.

+ Z.

1. All the weight is in only one 13, (set # 1,2,3). 2. The weight is split evenly between only two 13,s (set # 4 , 7 , 1 0 ) . 3. The weight is split two-to-one between two 13nS (set #5,6,8,9,11,12). 4. The weight is split evenly between all three 13,s (set #13). 5. The weight is split two-to-one where two 13nS share the lesser weight (set #14,15,16). 6. The weight is split two-to-one where two 13,s share the greater weight (set # 17,18,19).

.......

•

.. F.

_. G.

G

.

.

G

.

• •

.

.H H.

•

.H

.

.

.

.

•

.

I

9."

From Figure 8 it can be seen that the manual alternative generally scores worse than the semiautomatic and automatic except in four areas:

........ .

*

+'.+

•

.-÷x~

•

.

.' .

.

"

.

.

.

.I.

0..

+

.I ........ I I . .

I .

.

.

.

. .

.

,I

.

. .

. .

. x

.

. ......... . . . .

. I

.

.

.

.

.

.

.

. . .

.

.

.

.

.

J

.

.---lO----.-J * , + .

. .

.

.

.

.

.

.

.

.

.

.

.

.

.

J

.

.a

. J

.

.

.

.

.

. ......... ÷

.

. .

.

.

.

. J

. .

.

J

.

.

. .

J

. ......... .

.

. .

.

+ ,

.

.

.

.O

J

. . .........

.

.

. J . . . . . . . . .

+

.

.

.

0 .

*

3

.

.

.

. . . . . . . . . .

.

. . ....... .

.

. +..

,

. .

.

.

.

l

.

. . . . . . . . .

.

. . .........

.

.

.JJ .

.

. .

.

.

.

.

. .

.

. . . . . . . .

J

I

,

. .

.

+

. . . . . .

+ . +. .

.

.

.

+

. .

.

,

.S HO +

÷

.*

.

+. J

. .

.

41

. H ........

.

.

.

+

. ......... . . . .

3937

4 GO

.

.

+.J

.

. .

.

.

. . . . .

+

Z0--J

.

35

J

. . .

. . . . . . . .

.+. ,+.

. . . . . . . . .

.

.

.

÷

.

.

.

÷

. .

.

.

.

+ .

. .

.

*

+

. . . . . . . . . .

.

. . ........

........ . . . . .

.

+++

*. •

.

.---30----.

. +

.

+

. .

÷ . + . * . +÷+ .

.

.

. . . . . . . .

.

I+ .

t

.

.

.1 .I

I

H+

3.3

O

.........

..

H--.---BO----. .. + , . + . ÷ .

. . . . . . . .

I,

.I

. . . . . . . . .

.

+

.

. . . .

1 .

.

.

3.1

HI

....

..++

.

. I I

.----1----.

.

8.

. .

•

........ .

•

0.-

.

.

. ..

O.-

+

.

.B .

+ H H,*

.

l.

.

. ..... ,

2.9

G

3

~ . . ÷ . . --G---ro ........

. G

27

F

. .

.

.......

.+ .

.

.---lO----.

r

.

.

5.

.O

Z.5

D E

6".

+

+

.

........

4F O. -

BO . . . . . . . . .

. .

F

•

"Z

. "~E

"E

.'

.

8O . . . . . . . . .

"

3. "E

•

.

'O++f

O.3

O,-

.

B C

. .

. .

.

.

.

.

.

+

. .

.9 JO

*

.

J

.

.

.

.

.

Figure 6 Contour Plot of I~. for the Semi-Automatic Alternative

191

.

+ .

x3

Journal of Manufacturing Systems Volume I l/No. 3

i ,.

Table 13

!-L"B%+--s-c-°![ ~ 3.0o

+

.

:

,~

C o m b i n a t i o n s of W e i g h t s for 6,,

3.10

~a0~2°

g

0.---90----

0

~

3.40

3.50

SET

~ :-c-----8.+o~ :.o. 2.

"C++, ' .

2

.

.

.

.

.

fl,

f12

f13

0.00 0.00 1.00 0.00 0.00 0.00 0.50 0.33 0.67 0.50 0.33 0.67 0.33 0.25 0.25 0.50 0.40 0.40 0.20

0.00 1.00 0.00 0.50 0.30 0.60 0.00 0.00 0.00 0.50 0.60 0.30 0.30 0.20 0.50 0.20 0.40 0.20 0.40

1.00 0.00 0.00 0.50 0.67 0.33 0.50 0.67 0.33 0.00 0.00 0.00 0.33 0.50 0.25 0.25 0.20 0.40 0.40

MAN SEM

AUT

.

D.÷.

3

+

D

.

.

.3

oo ...... E- ~E-7~......... :o . .

E E.

.

E

OiE...... . [

7~.-~ ......

E

.

6.C O.E ...... E

E

E ---E-----.F . E

.

.O 8. O.

0

.

--C--

A + a

D C

.B

+G

.+ .

. >--

-F---

F

f

.

. .G

.

.

......

.

. .

. .

. .

.

.

,8

G ---G---

.

O

.G

.

G

F

G

. .

G .

G+

. .

+

.

....... .

.

F

. .

.

g

-F . . . .

F

0

+

F

. .

+E O. . . . . . . . . . . . . .

.G. G

.

.......

......

. .

.

.7 .0

.

*.

.

.

. . ........ .

+ .

.

. ,

. .

.

+ .

H -H---.

H .H

.

---

+ .

.

+ F

E

o C .....

.

+ . +. .. . . . . . .

.

6 +

6 D

.H

+

.

---10

. .E

0.

.C

+ A . B . C xz . . . . A... B ....

.

H

.+.

G.*

.

. .- .......

. . . . . . . . . -+ . .

20 . . . . .

G . . . . .

.F F

O.

. H H

.+. G

, G

E

+

H. .

+ H +

F

.S GO

+

. +H+

F

--E .....

G

.......

. . +H , .....

.

---3OH--+ . .

+ F.- .......

.

> . .O .

C

.

F F

.

G

+ .

+

G *

+

o .

.

+ . +. H 4 0 H. . . . . . . +H ÷

+ H +

.

G. . . . . .

F

.

+.H

.G

E. .E

---D-----

+ C

+ , B

.

+.

.

.G .......

F

.E .E +

D O

c

.B .B+

. ~).

.

c 9 O.B--

E

+ .

G.

.F

.E. .E. --,E .....

O --O--

c, C

F

7 o

.O

,F . + G . . G . . F-.-----G-. ....... . . G . +

G

.4

~o 7 ................

.

1 2 3 4 5 6 7 8 9" 10 11 12 13 14 15 16 17 18 19

. F

. + G 5 .G+ . . . . G, + . . . . G . --50 --- ....... G . + . .

G

F

"

,

F, r

E

.

F+

"

*F .+

F

F.

,

.

.

.E F 5.E F OE . . . . . . F-- ..... E.+ F . . E

÷

E

+

.

E+

.

E.

+

. . E E . . . . . . . . . . . . . . .

+ X3

Figure 7 C o n t o u r Plot of I~n for the A u t o m a t i c Alternative

used to analyze the effects of varying the proportion of (x,,t, used in the model. This was again performed for the three alternatives previously described. A sixth-degree fitted model is obtained for the strategic and tactical objectives of each alternative. These models contain 63 terms each, and again the coefficients can be analyzed to determine the impact of the attributes upon the final score. Unfortunately, the estimated shape of the surface for the o%t,s cannot be depicted pictorially. Instead, a comparison of the scores for the three alternatives is plotted in Figure 9 (strategic) and Figure 10 (tactical) for

8T I 7 -

!

MAN SEM AUT

....

?~

l

~

/

6r

'

,

'~,

I

d 2

-

±

Data Manual

Set

3.3 2.9 2.8 3.4 3.2 3.1 3.6 3.2 3.0 3.4 2.9 2.8 3.7 3.8 3.7 3.5 3.6 3.7 3.8

Manual. Semi-Automatic. Automatic.

1. All the weight is in only one o~,,k (set # 1-6). 2. The weight is split evenly between only two o~,,ks (set #7-21). 3. The weight is split evenly between only three ot,,ks (set #22-41). 4. The weight is split evenly between only four OLnl,S (set #42-56). 5. The weight is split evenly between only five %,ks (set #57-62). 6. The weight is split evenly between all six (XnkS (set #63).

[

'.

4.1 3.8 2.3 4.0 4.0 3.9 3.7 3.5 2.9 3.5 3.3 2.8 4.0 4.2 4.1 3.6 3.8 3.9 4.2

various combinations of (x,,k. The combinations of ant` used in the plots represent the following variations of OL,k:

l /

6.4 7.4 1.8 6.9 6.7 7.1 4.5 4.9 3.3 5.0 5.5 3.7 5.6 6.0 6.2 4.7 5.4 5.2 6.3

Number

Semi-Automatic

Automatic

Results tend to be robust to the attribute weights. However, outcomes do highlight the strengths and weaknesses of each attribute. For the strategic

Figure 8 S c o r e s for the A l t e r n a t i v e s for V a r i o u s V a l u e s of I~.

192

Journal of Manufacturing Systems Volume l l/No. 3

58 T 544 52 I

II '

46 44

S

42~

c o r e

40 .

385 36

32

*

"

3O , / '"

5 []

10

15

Manual

20

25

30

35

40

Data Set Number Semi-Automatic

45

50

55

60 o

65

70

75

50 49 48 47 46 45 44 43 42 41 40 39 38 37 36 35 34 33 32 31 3O 5

Automatic

10

Manual

15

20

25

30

35

40

Data Set Number Semi*Automatic

45

50 ,:

55

60

Automatic

Figure 9 Scores of Alternatives for a.k Values of the Strategic Objective

Figure 10 Scores of Alternatives for o~.k Values of the Tactical Objective

objective, the manual alternative score is consistently higher, and the semi-automatic and automatic alternatives exhibit similar trends with respect to the weight of ant ` . The tactical objective is more sensitive to the weights of ant ` . The lower scores of the manual alternative result from higher weights of the design, flexibility, and producibility attributes. Higher weights of the design, flexibility, and personnel attributes drive down the semi-automatic scores, and higher weights of flexibility, integration, and personnel account for the lower scores of the automatic alternative. These shifts indicate that the tactical objective should be given careful consideration when evaluating these AMSTs. For comparison, the proposed three-objective model and example data were also analyzed using goal programming 33 and MCQA 1I. 34 Results were similar except the sensitivity analysis shown in the contour plots are not as readily obtained. 2s

ysis not only provides a ranking of the alternatives, but also an understanding of the objectives that make up the AMST evaluation, An example was presented to illustrate the basic methodology. Using a mixture experiment, it is possible to graphically portray the contours of each alternative and to investigate the tradeoffs among the pecuniary, tactical, and strategic objectives.

Acknowledgement This article is based upon work supported in part by the National Science Foundation under grant DMC 85-44993. Appreciation is also given to Hughes Aircraft Company and its Advanced Education Program.

References l. C.N. Madu and N.C. Georgantzas, "Strategic Thrust of Manufacturing Automation Decisions: A Conceptual Framework," liE Transactions, Vol. 23, No. 2, June 1991, pp. 138-48. 2. J.R. Canada and W.G. Sullivan, Economic and Multiattribute Evaluation of Advanced Manufacturing Systems, Prentice-Hall, 1989, pp. 6-10. 3. R.S. Kaplan, "Must CIM be Justified by Faith Alone?," Harvard Business Review, Vol. 64, No. 2, May-April 1986, pp. 87-95. 4. US Department of Labor, Trends in Manufacturing: A Chart Book, Bureau of Labor Statistics, 1985, pp. 25-27. 5. J.R. Canada, "Non-Traditional Method for Evaluating CIM Opportunities Assigns Weights to Intangibles," Industrial Engineering, Vol. 18, No. 3, March 1986, pp. 66-71. 6. W.G. Sullivan and S.E. LeClair, "Justification of Flexible Manufacturing SystemsUsing Expert System Technology," AUTOFACT '85 Conference Proceedings, Dearborn, MI, November4-7, 1985, CASA/SME, pp. 7-1-7-13. 7. M.L. Wymore, and L.D. Duckstein, "Prioritization of Capital Investments for Factory Automation Using Multicriterion Q-

Conclusion The analysis of AMSTs should not rely on quantitative measures alone. This paper presented a methodology for incorporating qualitative and quantitative measures in the evaluation of AMSTs. The pecuniary, strategic, and tactical objectives of the model were presented, along with supporting documentation for the completeness of the model for a manufacturing organization. Measures of the performance of the objectives were normalized and evaluated using composite programming. This anal-

Analysis," Proceedings of the International Conference on Multiple Criteria Decision Making: Applications in Industry & Service, Bangkok, Thailand, December 1989, pp. 317-31.

193

Journal of Manufacturing Systems Volume I l/No. 3

Quality, American Society for Quality Control, 1983, pp. 1-18. 32. L.B. Hare and P.L. Brown, "Plotting Response Surface Contours for Three-Component Mixtures," Journal of Quality Technology, Vol. 9, No. 4, October 1977, pp. 193-97. 33. S.M. Lee, Goal Programming Methods for Multiple Objective Integer Programs, Vol. 2, American Institute of Industrial Engineers, 1979, pp. 1-3. 34. H. Hiessl, L. Duckstein, and E.J. Plate, "Multiobjective QAnalysis with Concordance and Discordance Concepts," Applied Math and Computation, Vol. 17, 1985, Elsevier, New York, NY, pp. 107-22.

8. C.H. Fine and R.M. Freund, "Optimal Investment in ProductFlexible Manufacturing Capacity," Management Science, Vol. 36, No. 4, April 1990, pp. 449-66. 9. M.C. Burstein, "Finding the Economical Mix of Rigid and Flexible Automation for Manufacturing Systems," Proceedings of the

Second ORSA/TIMS Conference on FMS: Operations Research Models and Applications, Ann Arbor, MI, August 12-15, 1986, pp. 43-54. 10. C. Gaimon, "The Dynamical Optimal Acquisition of Automation," Annals of Operations Research, Vol. 3, No. 1-4, 1985, pp. 59-79. 11. C. Gaimon, "The Strategic Decision to Acquire Flexible Technology," Proceedings of the Second ORSA/TIMS Conference on FMS: Operations Research Models and Applications, Ann Arbor, MI, August 12-15, 1986, pp. 69-81. 12. D.J. Nobel, "Using Simulation as a Tool for Making Financial Decisions in an Uncertain Environment," Industrial Engineering, Vol. 20, No. 1, January 1988, pp. 44-48. 13. W.E. Newman Jr., "Model to Evaluate the Benefits of FMS Pallet Flexibility," Proceedings of the Second ORSA/TIMS Conference on FMS: Operations Research Models and Applications, Ann Arbor, MI, August 12-15, 1986, pp. 209-20. 14. S.U. Randhawa and T.M. West, "Uncertainty Modeling in CIM Investment Analysis," CIM Review, Vol. 6, No. 1, Fall, 1989, pp. 32-36. 15. J.W. Troxler and L.T. Blank, "Value Analysis for Manufacturing Technology Investments," 1987 HE Integrated Systems Conference Proceedings, Nashville, TN, November 5-7, 1987, Institute of Industrial Engineers, pp. 24-29. 16. R.N. Wabalickis and B.K. Ghosh, "Analytic Hierarchy Process for Justification of FMS," 1988 liE Integrated Systems Conference Proceedings, St. Louis, MO, October 30-November 2, 1988, Institute of Industrial Engineers, pp. 298-303. 17. R. Putrus, "Accounting for Intangibles in CIM Justification," CIM Review, Vol. 6, No. 2, Winter 1990, pp. 23-29. 18. J.L. Noble, "Strategic Benefits of CIM in Cost Justification," CIM Review, Vol. 6, No. 4, Summer 1990, pp. 66-70. 19. M.T. Works, "Cost Justification and New Technology Addressing Management's 'NO' To the Funding of CIM," A Program Guide for CIM Implementation, CASA/SME, 1985, pp. 58-66. 20. Computer Integrated Manufacturing--An IBM Perspective, IBM Corporation, 1987. 21. G.A. Fleischer, " A Generalized Methodology for Assessing the Economic Consequences for Acquiring Robots For Repetitive Operations," 1982 Annual Industrial Engineering Conference Proceedings, New Orleans, LA, May 1982, Institute of Industrial Engineers, pp. 130-39. 22. J.L. Gibson, J.M. Ivancevich, and J.H. Donnelly, "Organizations," 6th ed., BPI-Irwin, 1988. 23. J.L. Riggs, Production Systems: Planning, Analysis and Control, John Wiley & Sons, 1981, pp. 40-44. 24. S.E. Garrett, "Strategy First: A Case in FMS Justification,"

Appendix A When both NPVMAx and NPVMtN are positive, the RMz will simply be NPVMax. In this case,

MIj

-

NPVj NPVMax and all of the MIj will be positive

and less than or equal to one. When NPVMax is positive and NPVmzN is negative, RMz will be NPVMax + [NPVMml. In this situation, M/j =

NPVj NPV~tAX+ INPVMmI"

Since NPVMAx + INPVmml

> INPVjl, the MIj must be between -1 and + 1. Finally, if NPVMax and N P V M I N should both be negative, RMz will be INPVmzNI. This will yield Mlj NPVj -

-

INPVMINI

and M/j will range between -1 and 0.

Therefore, the MIj is either [-1, 0], [0, + 1], or (-1, + 1), and as such the MIj will be in the range [-1,

+1].

Authors' Biographies Johann G. Demmel is a project engineer at Hughes Aircraft Company in Tucson, AZ. Dr. Demmel received a BS in industrial engineering from Rochester Institute of Technology, Rochester, NY, an MS in industrial engineering and management from Oklahoma State University, Stillwater, OK, and a PhD in systems and industrial engineering from the University of Arizona, Tucson, AZ. He has worked in the aerospace industry for six years in various manufacturing and engineering positions. His interests include technology management and computer-integrated manufacturing systems. Dr. Demmel is a Howard Hughes Doctoral Fellow, Tau Beta Pi Fellow, liE Gilbreth Fellow, and a member of liE, Tau Beta Pi, Alpha Pi Mu, and Phi Kappa Pi. Ronald G. Askin is an Associate Professor of Systems and Industrial Engineering at the University of Arizona, Tucson, AZ. Dr. Askin received a BS in industrial engineering from Lehigh University, and an MS in operations research and PhD in industrial and systems engineering from Georgia Institute of Technology. He has published in various professional journals, predominantly in the areas of manufacturing system design and operation and engineering statistics. Dr. Askin has received several awards, including the liE Transactions Development and Applications award (coauthor), the ASEE/IIE Eugene L. Grunt award (coauthor) and a National Science Foundation Presidential Young Investigator award. Dr. Askin is an active member of CASA/SME, liE, ORSA, ASQC, POMS, and ASA.

Proceedings of the Second ORSA/TIMS Conference on FMS: Operations Research Models and Applications, Ann Arbor, MI, August 12-15, 1986, pp. 17-29. 25. J.G. Demmel, " A Multiple Objective Decision Model for the Evaluation of Advanced Manufacturing System Technologies," The University of Arizona, UMI Dissertation Information Service, Ann Arbor, MI, 1991. 26. R. Shinnar, O. Dressier, C.A. Feng, and A.I. Avidan, "Estimation of the Economic Rate of Return for Industrial Companies," Journal of Business, Vol. 62, No. 3, 1989, pp. 417-45. 27. P.C. Fishburn, "Utility Theory for Decision Making," John Wiley & Sons, Inc., 1970, Chapter 2. 28. A. Bardossy, I. Bogardi, and L. Duckstein, "Composite Programming as an Extension of Compromise Programming," Mathematics of Multi-Objective Optimization, Springer-Verlag, Udine, Italy, 1985, pp. 375-408. 29. M. Zeleny, " A Concept of Compromise Solutions and the Method of the Displaced Ideal," Computers and Operations Research, Vol. 1, 1974, Pergamon Press, Great Britain, pp. 479-96. 30. F. Choobineh, "Justification of Manufacturing Systems," Material Handling Research in the US, June 1990, Material Handling Institute, Hebron, KY, pp 221-228. 31. J.A. Cornell, How to Run Mixture Experiments for Product

194