European Journal of Operational Research 115 (1999) 5976

Theory and Methodology

A modeling approach to logistics in concurrent engineering

Shad Dowlatshahi
*

Department of Information and Decision Sciences, College of Business Administration, University of Texas at El Paso, El Paso, TX 79968-0544, USA Received 21 July 1997; accepted 30 October 1997

Abstract This paper advocates logistics involvement in the early phases of product design and development in a concurrent engineering environment. A concurrent engineering environment and the benets of such involvement are explained in detail. The paper focuses on facilitating an interface and a collaboration between the designer and the logistician. A conceptual interface for design for logistics is presented. Four areas of interface are considered: (a) logistics engineering, (b) manufacturing logistics, (c) design for packaging, and (d) design for transportability. A set of detailed design factors pertaining to each area are presented. A modeling approach, Bond Energy Algorithm, is used to accomplish the design for logistics concerns developed throughout the paper. An example is provided to test and validate the algorithm. The results are analyzed and appropriate perspectives for managerial implication of the methodology are provided. Finally, some conclusions and assessments are presented. 1999 Elsevier Science B.V. All rights reserved. Keywords: Design for logistics; Concurrent engineering; Transportability; Manufacturing product design; Bond energy algorithm

1. Introduction Product design in a concurrent engineering environment focuses on an interdisciplinary approach that utilizes methods, procedures, and rules to plan, analyze, select, and optimize the design of products. In the early stages of the design process concurrent engineering considers and includes various product design attributes such as aesthetics, durability, ergonomics, interchangeability,

Tel.: 915 747 7759; fax: 915 747 5126; e-mail: sdowlats@utep.edu.

logistics, maintainability, marketability, manufacturability, procurability, reliability, remanufacturability, safety, schedulability, serviceability, simplicity, testability, and transportability. The greatest impact and benets of concurrent engineering are realized at the design stage of product development. The design decisions made in the early phases of product design and development will have signicant impact upon future manufacturing and logistical activities. The following examples illustrate the relationship. A study at Rolls Royce revealed that design determined 80% of the nal production cost of 2000 components (Corbett, 1986).

0377-2217/99/$ see front matter 1999 Elsevier Science B.V. All rights reserved. PII: S 0 3 7 7 - 2 2 1 7 ( 9 8 ) 0 0 1 8 4 - 2

60

S. Dowlatshahi / European Journal of Operational Research 115 (1999) 5976

According to General Motors executives, 70% of the cost of manufacturing truck transmissions is determined in the design stage (Whitney, 1988). Ford Motor Company has estimated that among the four manufacturing elements of design, material, labor, and overhead, 70% of all production savings stem from improvements in design (Cohodas, 1988). A study revealed that product design is responsible for only 5% of a product's cost; it can, however, determine 75% or more of all manufacturing costs and 80% of a product's quality performance (Huthwaite, 1988). Yet another study shows that 70% of the life cycle cost of a product is determined at the design stage. The life cycle cost here refers to cost of materials, manufacture, use, repair, and disposal of a product (Nevins and Whitney, 1989). Fig. 1 depicts the system approach to the design of a product. This gure provides a conceptual framework where interactions among various functional areas in a concurrent engineering environment can be explored and analyzed. The

functional areas presented in Fig. 1 encompass all relevant areas pertaining to product life cycle from product inception to disposal. This includes business, production, and support requirements. Each of these requirements consists of several related functional areas as depicted in the gure. Fig. 1 also signies that product design is at the core of concurrent engineering activities and it provides the most signicant benets when considered in conjunction with any of a combination of several functional areas. The paper focuses on the linkage between product design and logistics as highlighted in Fig. 1. 1.1. Objectives and organization of the paper The focus of this paper remains on the integration of logistics concerns in the early stages of product design and development. The objectives of this paper are threefold. To demonstrate the importance of early logistics involvement in the product design process. To present a conceptual as well as an analytical basis for integrating logistics concerns,

Fig. 1. Product design in a concurrent engineering environment.

S. Dowlatshahi / European Journal of Operational Research 115 (1999) 5976

61

constraints, and contributions in the design process. The conceptual and analytical bases are not an end in themselves, but are used to facilitate the design for the logistics process. To provide a framework where managerial implications of design for logistics can be explored. Furthermore, the eciency and eectiveness of the methodology, results, and their managerial implications are analyzed. After a presentation of system approach to concurrent engineering and the literature review, a conceptual framework for design for logistics is presented in Section 2. This section consists of a systems approach to design for logistics. It also addresses the components of the design for logistics and their associated modules and design factors. A design for logistics model is developed in Section 3. The details of the algorithm are outlined as well. This section will also present an example to validate and test the approach and methodology oered in this paper. Section 4 provides the advantages and implications as well the conclusion and assessment.

1.2. System approach to concurrent engineering There are two issues at the core of successful implementation of concurrent engineering: 1. All activities related to the development of a product should be focused in the early stages of product design (conceptual design) so that the greatest benets of such an integration are achieved. The information requirements and exchanges at the conceptual design are not well-dened and usually fuzzy. This poses a challenge for implementing concurrent engineering. 2. The impact and constraints associated with various functional requirements should be communicated to the designer on a timely, accurate, and relevant basis. Fig. 2 represents an integrated logistics system as it relates to product design. An eective design for logistics cuts across a number of functional areas as illustrated in Fig. 2. These activities converge to product design as the embodiment of all future activities. As the design for logistics aects

Fig. 2. Integrated logistics system for product design.

62

S. Dowlatshahi / European Journal of Operational Research 115 (1999) 5976

other functional areas, other areas in turn aect logistics considerations. This process is inherently a dynamic one requiring negotiation and trade-o among the functional areas in a concurrent engineering environment. Dierent functional areas of Figs. 1 and 2 and their interrelationships can be explored and culminated into several other research papers. The focus of this paper, however, is on exploring and analyzing the link between logistics and product design. 1.3. Review of literature A review of literature reveals that little or no work is being done on the interface of product design and logistics. The early and crucial interface and collaboration between product design and logistics has been largely disregarded. Some related literature for the design for logistics topics are presented in Table 1. The review of literature indicates that no direct, signicant work has been done in bridging the gap between product design and logistics. It is important to note that most papers focused on the qualitative and non-numerical evaluation of logistics. There was only one paper that focused on the quantitative and analytical evaluation of logistics manufacturing within a global manufacturing environment. This paper, however, did not address the crucial logistics factors in the early stages of product design. Additionally, it only considered the role of logistics manufacturing and not other factors such as logistics engineering, packaging, and transportability issues. This paper combines conceptual and quantitative approaches to design for logistics by rst developing a conceptual framework and then by using a modeling approach to create self-contained clusters that are easily manageable and can be implemented simultaneously. 2. A conceptual framework to design for logistics A system approach to design for logistics essentially includes a designer's functional requirements as well as a logistician's requirements of

availability, supportability, cost, quality, volume changes, timely delivery, order frequencies, and the like. The design for logistics in Fig. 3 is decomposed into four subsystems. Fig. 3 suggests that the design for logistics has four essential subsystems whose presence and interactions largely determines the content of design for logistics. These subsystems encompass design, manufacturing, marketing, logistics, distribution, packaging and other similar aspects of the overall product design. The scope of this paper does not include design for environment and recycling issues. In other words, these four subsystems are the embodiment of the functional areas presented in Fig. 2. To achieve exibility, design economics, and overall design optimization, the design for logistics presented in Fig. 3 needs to be decomposed further into manageable and homogeneous units. For simplicity purposes, these manageable and homogeneous units are called modules. Therefore, each subsystem of Fig. 3 is, in turn, divided into modules. Each subsystem can be composed of several modules. Modules are the building blocks of design for logistics. Each module is further decomposed into design factors. Each module, then, has several respective design factors that must be considered in the design of that particular module. Fig. 4 shows the entire decomposition process of design for logistics and, as such, provides a conceptual framework for the inclusion of logistics concerns, constraints, and contributions in the design process. Each module addresses a specic, manageable and homogenous aspect of design for logistics. The logic is that modules are easier to develop, manage, and implement than a monolithic system with all the complexity it oers. The modules presented in Fig. 4 are intended to be generic, yet comprehensive, for the design for logistics. The composition, nature, and the number of modules may be modied (new ones may be added or the existing ones be deleted) based on the unique requirements and the complexity of each design for logistics environment. One may arguably question the necessity of having all the modules presented in Fig. 4 or judiciously propose adding new ones. These decisions are largely dependent upon the discretion of the designer and

S. Dowlatshahi / European Journal of Operational Research 115 (1999) 5976 Table 1 Review of literature Reference Foo et al. (1990) Focus/theme Product design from a materials logistics standpoint. The ideal product has: the minimum number of parts, standard or ``preferred'' parts, a modular and reusable physical architecture, a limited set of end item congurations, and a modular BOM structure. The eect of design decisions on logistics operations of corrosion inhibition. The design decisions spanned from conception to decommissioning of the product. The role of material selection in manufacturing processes, quality control, product packaging, product failure and product safety was discussed. Informational issues related to an integrated logistics system. Interoperability and collaboration with regard to material accessibility. Surface Mount Technology (SMT), in order to be successful, requires the implementation of both concurrent engineering and design for manufacturability. The basis for design for logistics (DFL) is the rm's ability to give a customer the required products when wanted. It is imperative that the manufacturing processes change (as needed) to suit an environment in which the manufacturing changes can be most eective. US manufacturers must do more than improve their internal eciency and eectiveness. They must also develop an understanding of how to better structure productive resources. A discussion of the ndings of a path analytic investigation of the interrelationships between a rm's perception of economic globalization, a rm's emphasis of logistics and manufacturing considerations in the design and management of global manufacturing networks, and a rm's competitive/nancial performance. Industrial engineers are the ideal professionals to design and integrate logistics, transportation and distribution systems. The relationships between logistics technology adoption and organizational design practices are examined and the logistics technologies of several organizations are discussed. For factories with design capabilities, a plan to reduce cost with design for manufacturability and documentation is proposed. A comprehensive hierarchical structure of design for logistics is presented. An algorithm is used to model logistics problems in concurrent engineering. Methodology Conceptual

63

Starsmore (1990)

Conceptual

Colling (1991)

Conceptual

Kyng (1991)

Conceptual

Implementing Concurrent Engineering for SMT (1992)

Conceptual

Mather (1992)

Conceptual

Fawcett and Closs (1993)

Conceptual and quantitative (path analysis)

Petersen (1993)

Conceptual

Germain et al. (1994)

Conceptual and empirical (36 logistics technologies) Conceptual

Cherf (1995)

``A Modeling Approach to Logistics in Concurrent Engineering'' (1997)

Conceptual and quantitative (Bond Energy algorithm)

based on the managerial implications of each manufacturing organization. Notwithstanding this statement, Fig. 4 may be used as a suitable conceptual framework for any design for logistics of

mechanical products. Additionally, design factors are the driving force behind the modules. Design factors are the smallest functional requirements in the overall design of logistics. The design factors in

64

S. Dowlatshahi / European Journal of Operational Research 115 (1999) 5976

Fig. 3. Components of design for logistics.

Fig. 4. Hierarchical decomposition of design for logistics.

S. Dowlatshahi / European Journal of Operational Research 115 (1999) 5976

65

a logistics design are synonymous to ``parts'' in product design. Each subsystem of the design for logistics is subsequently explained and design factors associated with each module are delineated. 2.1. Logistics engineering Logistics engineering is a eld of logistics that deals with the supportability of product and systems throughout their life cycle. Logistics engineering is concerned with the design process in that it establishes requirements to which the ultimate design conguration must comply. Logistics must be intuitively an integral part of the design process along with performance, size and weight, reliability, safety, manufacturability, cost, and the like. Logistics engineering also takes into consideration environmental impact, energy conservation, and solid waste transportation and disposal. Design and logistics engineers should collaborate by having sound and complete design specications, periodic visits, eective and continuing communication and dialogue, and development of eective planning documents. This area addresses system, product design, and development. Logistics engineering criteria support the life-cycle aspects of product design (design for product supportability). The focus remains on inuencing and incorporating changes in the product design to meet product supportability requirements before the design is nalized. The design factors associated with the logistics engineering subsystem are presented in Table 2. The list of design factors presented in Table 2 (and subsequent Tables 35) are not intended to be exhaustive. These generic lists merely present the design factors that may be appropriate to use in the design of these subsystems. Additional design factors may be added or deleted from Tables 25 depending on the unique requirements and complexity of each logistics systems design. Notwithstanding this statement, Tables 25 still present the most comprehensive list of design factors developed for the modules specied in Fig. 4. Also, noteworthy is that there are no priorities in the ordering of these design factors, nor

is any design factor more important than any other. Only the designer's experience and judgment, in terms of a factor's relevance or importance, determines the inclusion and/or exclusion of certain design factors in Tables 25.

2.2. Manufacturing logistics The characteristics of the manufacturing processes and activities are a major determinant of the logistics activities and logistics system design. Manufacturing processes and activities often create a number of constraints and opportunities for a logistics system. Manufacturability, as a main feature of product design, is a signicant contributor to the design for logistics. The design factors associated with the modules of the manufacturing engineering subsystem are presented in Table 3.

2.3. Design for packaging This area addresses the issues related to packaging requirements in the product design process. Packaging is an important feature of a product as it creates or enhances the product's image. Product packaging is an important marketing tool and has a signicant impact upon overall product cost, its ease of use, and its perception to customers. Packaging also protects the product from breakage, spillage, etc. Logistics requirements for packaging must be incorporated at the design stage with those of marketing and manufacturing requirements. In the past, designers have attempted to incorporate only marketing and manufacturing requirements in the design instead of including the logistics requirements as well. This has resulted in an ineciency in the overall product and system performance, as well as higher operational and usage costs. Poor packaging results in lower sales, damaged contents, customer dissatisfaction, and higher cost of material handling, warehousing and transportation. Packaging absorbs approximately 12% of the logistics dollar (Ballou, 1987). The design factors associated with

66

S. Dowlatshahi / European Journal of Operational Research 115 (1999) 5976

Table 2 Hierarchical structure of logistics engineering Module Design for supportability (a1 ) Design factors                                        Future redesigns Product variation Logistics performance data Product performance Product weight Design specications Operating height Market characteristics Storage and movement systems Equipment type Carrier type Product test and evaluation Self maintenance Standardized parts Part counts Interchangeable parts Product physical dimensions and bulk Product value Kinematics (type, direction of motion, velocity, acceleration, etc.) Forces (direction, magnitude, and frequency of force) Ease of assembly Type of manufacturing processes Tolerances Surface nish Number of changes in product lines Number of product lines Conguration management Demand variability Design leadtimes Demand forecast error Product seasonality Product prot margin Duration of downtime Measures of availability Mean time to rst failure Mean idle time Failure rate per unit time Service time Time to respond to service Notation (a1X1 ) (a1X2 ) (a1X3 ) (a1X4 ) (a1X5 ) (a1X6 ) (a1X7 ) (a1X8 ) (a1X9 ) (a1X10 ) (a1X11 ) (a1X12 ) (a1X13 ) (a2X1 ) (a2X2 ) (a2X3 ) (a2X4 ) (a2X5 ) (a2X6 ) (a2X7 ) (a2X8 ) (a2X9 ) (a2X10 ) (a2X11 ) (a3X1 ) (a3X2 ) (a3X3 ) (a3X4 ) (a3X5 ) (a3X6 ) (a3X7 ) (a3X8 ) (a4X1 ) (a4X2 ) (a4X3 ) (a4X4 ) (a4X5 ) (a4X6 ) (a4X7 )

Design for manufacturability (a2 )

Product lines (a3 )

Design attributes (a4 )

the modules of the design for packaging subsystem are presented in Table 4. 2.4. Design for transportability Transportation costs represent the most single important element in logistics costs for most rms. An eective design for transportability stimulates direct competition among rms at dierent loca-

tions; creates greater economies of scale; and reduces the price of goods and services. The design factors associated with the modules of the design for transportability subsystem are presented in Table 5. 3. A modeling approach to design for logistics This paper utilizes Bond Energy Algorithm (BEA) a clustering approach to the design of

S. Dowlatshahi / European Journal of Operational Research 115 (1999) 5976 Table 3 Hierarchical structure of manufacturing logistics Module Manufacturing processes (a5 ) Design factors                                                   Robustness of product design Manufacturing tooling Critical process Size and dimensions of WIP Utilization rate Degree of automation Volume exibility Product exibility Stressstrain analysis Shape, size, and thickness of the parts Tolerances and surface nish requirements Production volume Product expected service life Type of manufacturing process used Assembly operations Length of production run Size of production run Leadtimes Production schedules Set-up times Throughput time Inventory turnover rate Seasonal inventory Idle times (machine and labor) Due dates (earliest, latest) Slack times Inventory level Loading Average ow time Average number of jobs in the system Changeover cost Inventory levels Work in process Material delivery time Critical and proprietary materials Unit load size Part type (unit, bulk, liquid, gas) Piece rate (quantity) Part orientation (design) Logistics of the materials move (internal, external) Material properties Material availability Material cost Plant capacity Demand schedules Number of plants Location of plants Distribution of costs Multiple warehouses Transportation costs Notation (a5X1 ) (a5X2 ) (a5X3 ) (a5X4 ) (a5X5 ) (a5X6 ) (a5X7 ) (a5X8 ) (a5X9 ) (a5X10 ) (a5X11 ) (a5X12 ) (a5X13 ) (a5X14 ) (a5X15 ) (a6X1 ) (a6X2 ) (a6X3 ) (a6X4 ) (a6X5 ) (a6X6 ) (a6X7 ) (a6X8 ) (a6X9 ) (a6X10 ) (a6X11 ) (a6X12 ) (a6X13 ) (a6X14 ) (a6X15 ) (a6X16 ) (a6X17 ) (a6X18 ) (a7X1 ) (a7X2 ) (a7X3 ) (a7X4 ) (a7X5 ) (a7X6 ) (a7X7 ) (a7X8 ) (a7X9 ) (a7X10 ) (a8X1 ) (a8X2 ) (a8X3 ) (a8X4 ) (a8X5 ) (a8X6 ) (a8X7 )

67

Production planning and control (a6 )

Materials (a7 )

Plant location (a8 )

68

S. Dowlatshahi / European Journal of Operational Research 115 (1999) 5976

Table 4 Hierarchical structure of design for packaging Module Packaging materials (a9 ) Design factors                                          Strength of material Weight per cubic measure Packaging density Transportation rate Cost of materials Cost of processing the package's materials Type of materials used Amount of materials reduced Material disposability and reusability Shock levels Moisture Heat resistance Vibrations Corrosiveness Pressure Tension Compression strength Impact Allowable level of damage/protection Material fragility Packaging shape, size, and modules Packaging overall cost Loss and damage historical data Packaging specications Packaging interior and exterior factors Packaging ease of opening, closing, and reusing Packaging ease of handling Package identication Amount of dead space in stacking Weight-to-protection shipping ratio Temper-resistant packaging Packaging physical dimensions Element environmental factors (temperature, humidity, etc.) Cubic utilization rate Packaging shape and structure Inventory requirements Transportation requirements Warehousing requirements Shipping and handling requirements Order picking requirement Linear footage of shelf space Notation (a9X1 ) (a9X2 ) (a9X3 ) (a9X4 ) (a9X5 ) (a9X6 ) (a9X7 ) (a9X8 ) (a9X9 ) (a10X1 ) (a10X2 ) (a10X3 ) (a10X4 ) (a10X5 ) (a10X6 ) (a10X7 ) (a10X8 ) (a10X9 ) (a10X10 ) (a10X11 ) (a11X1 ) (a11X2 ) (a11X3 ) (a11X4 ) (a11X5 ) (a11X6 ) (a11X7 ) (a11X8 ) (a11X9 ) (a11X10 ) (a11X11 ) (a12X1 ) (a12X2 ) (a12X3 ) (a12X4 ) (a12X5 ) (a12X6 ) (a12X7 ) (a12X8 ) (a12X9 ) (a12X10 )

Packaging testing (a10 )

Packaging design features (a11 )

Functional packaging requirements (a12 )

logistics systems. The clustering algorithms have been widely used in the decomposition of complex design problems. The purpose of clustering algorithms is to decompose a large system into subsystems and then cluster them into manageable modules and activities. BEA is concerned with the grouping of objects into homogeneous clusters

(groups) based on common features or activities. Clustering Analysis Methods have been widely used in various disciplines such as biology, data recognition, medicine, pattern recognition, production ow analysis, task selection, control engineering, automated systems, market segmentation, and expert systems.

S. Dowlatshahi / European Journal of Operational Research 115 (1999) 5976 Table 5 Hierarchical structure of design for transportability Module Transportation mode (a13 ) Design factors                                             Transit time (speed) Timely deliveries (dependability) Number of carriers The length of transportation contracts Transportation mode of operational availability Types of goods (sensitive, fragile, perishable, gas, liquid, etc.) Size of shipment Length of shipment Degree of vibration and acceleration Sectionalization and disassembly capacity Frequency of use Transportation (freight) rate for weight/mode/class Wage rate Exchange rate Trac rate Product density Transportation economy Transportation versatility Capacity Average transit times Delivery time variability Product time to market Product value to weight ratios Transit-time variability Average delivery times Degree of protection provided Low storage cost High movement cost Low-value weight ratio Total sales Transport method Economic storing, handling, and shipping Safe storing, handling, and shipping Work-in-process handling Physical properties (width, height, length, center of gravity, etc.) Dynamic limitations (acceleration, vibration, deection, leaking etc.) Life cycle cost Maximum weights Environmental limitations (temperature, pressure, humidity etc.) Hazardous eects (radiation, explosives, electrostatic, personal safety, etc.) Product value Type of packaging Product weight Shipping and handling characteristics (weight, bulk, compactness, etc) Notation (a13X1 ) (a13X2 ) (a13X3 ) (a13X4 ) (a13X5 ) (a13X6 ) (a13X7 ) (a13X8 ) (a13X9 ) (a13X10 ) (a13X11 ) (a13X12 ) (a13X13 ) (a13X14 ) (a13X15 ) (a13X16 ) (a13X17 ) (a13X18 ) (a13X19 ) (a14X1 ) (a14X2 ) (a14X3 ) (a14X4 ) (a14X5 ) (a14X6 ) (a14X7 ) (a14X8 ) (a14X9 ) (a14X10 ) (a14X11 ) (a15X1 ) (a15X2 ) (a15X3 ) (a15X4 ) (a15X5 ) (a15X6 ) (a15X7 ) (a15X8 ) (a15X9 ) (a15X10 ) (a15X11 ) (a15X12 ) (a15X13 ) (a15X14 )

69

Design criteria (a14 )

Transportability issues (a15 )

3.1. Solution procedure for Bond Energy Algorithm BEA is an interchange clustering algorithm that seeks to create a block diagonal form by maximizing some measure of eectiveness. The purpose

of BEA, in general, is to identify and display natural variable clusters that occur in complex data arrays. In particular, the objective of BEA in a logistics design is to group design factors into Design Factor Families (DFF) and modules into

70

S. Dowlatshahi / European Journal of Operational Research 115 (1999) 5976

Module Families (MF). The forming of these families allows a designer to simultaneously consider and design the design factors common to a set of modules. This modularized approach increases the eciency of the logistics design. The interactions between modules and design factors can be represented in a binary module-design factor incidence matrix. In the formulation of the incidence matrix [aij ], the entries of 0 and 1 are used. Entry 1 (0) indicates that a particular design factor does (does not) belong to a module. The binary assignment is accomplished by ``the designer''. The designer may be interpreted as one expert or a team of experts. No attempt is made to dene the composition or the working dynamics of the teams. This topic is beyond the scope of this paper. The topic of team decision-making and group negotiations has received considerable attention in current literature. The much talked about rating and assignment techniques (such as the Delphi method) are applicable to this paper as well. The assignment of binary values is inherently a technical task and requires the possession of knowledge and skills as they pertain to a particular logistics design. The methodology presented here allows the generation of multiple incidence matrices. Each of these matrices can be explored and solved, and the results can be evaluated for their eectiveness. On the other hand, discussions can be held by the design team members and changes can be made to only one incidence matrix until a consensus is achieved. The opportunities for developing sound incidence matrix(ces) abound. The Bond Energy Algorithm shown in Fig. 5 is an enhancement of the work by McCormick et al. (1972). Fig. 5 outlines the basic BEA algorithm by presenting two basic procedures for the columns and rows. The eciency of this algorithm is enhanced since the procedure used for the columns and rows is virtually similar. The measure of eectiveness for BEA is calculated as follows: ME
m n 1 ij iYj1 iYj1 i1Yj i1Yj X 2 i1 j1

Fig. 5. Flow chart of BE algorithm.

The ME measures the density of the bond between any two elements of the matrix. The greater

the numerical value of ME, the greater the bond between the elements. The purpose of this algorithm is to permutate the rows and columns and place elements adjacent to one another which have the largest ME values. By using the BEA, an unstructured incidence matrix can be transformed into a new structured

S. Dowlatshahi / European Journal of Operational Research 115 (1999) 5976

71

matrix with a clear grouping of activities. The clusters (mutually exclusive submatrices) obtained are considered the building blocks of the logistics system design. For procedural simplicity only two design factors per module were selected for this paper. The incidence matrix (1)(Fig. 6) represents the assignment of design factors to modules. In the incidence matrix (1), the aij entries correspond to design factors presented in Tables 25. Entry 1 signies that the inclusion of a particular design factor in a module is a necessary and essential requirement of forming that module. Each module inherently consists of a set of cohesive and bounded design factors whose interactions determine the overall design and eectiveness of the module. The selection and assignment of design factors to the appropriate modules pose a challenge as well as an opportunity for the designer. These decisions are virtually interrelated and must be made with caution. In assigning the binary values of 0 and 1, two issues must be considered. First, not all binary decisions require a similar level of diculty, skills, and judgment. Some binary assignments are fairly obvious and easy to make. Secondly, the specic requirements of each logistics design, the complexity of the design process, and nally the number of design factors and modules involved largely determine the nature of the binary decisions. Justications can be made for the binary assignments in the incidence matrix (1). A few examples illustrate this situation. The design factor a13X1 (transit time) is assigned to a module 13 (transportation mode) and module 15 (transportability requirements). These assignments are fairly obvious and easy to make. By the same logic, the design factor a1X5 (product weight) is assigned to module 1 (design for supportability), module 12 (functional packaging requirements), and module 13 (transportation mode) because it is a necessary and essential design factor for forming these modules. On the other hand, for example, assignment of the design factor a9X7 (human power: time, skill) to module 2 (design for manufacturability) requires justication and assumption, as it is not easily apparent as to why a9X7 is a necessary and essential requirement of forming module 2. The

designer may conclude that the type of materials used is directly related to design for manufacturability concerns. The materials selected in the design stage need to be fabricated or assembled as a part of the overall manufacturing processes. Certain materials (e.g. thin, fragile, and brittle components and materials) may require special equipment, tools, and processes. Although these considerations are important in many operations, they may not be necessary and essential for assignment of a design factor to a module. Designers ought to make a decision on these cases by using their skills and judgment. By so doing, the crucial decisions are deliberately made at the design stage and not by default at the implementation stage. Furthermore, the design factors outlined in Tables 25 may vary from one design to another depending upon the unique requirements of each logistics system design. Care, however, must be taken that only the most appropriate and relevant design factors are employed in any particular design. The incidence matrix (1) is an unstructured matrix that will be transformed into a new structured matrix by the use of the BEA. First, column 1 of incidence matrix (1) is selected arbitrarily and the measure of eectiveness for each column is calculated and presented in Table 6. From Table 6, columns 21 and 26 with the ME value of 5 (the highest ME value) are placed next to column 1. This procedure is repeated until all columns are accounted for and placed with one another. The results are represented in incidence matrix (2) Fig. 7. By the use of the BE algorithm and selecting row 1 arbitrarily, the measure of eectiveness for all other rows are calculated and presented in Table 7. Rows with the largest ME values are placed together. From Table 7, rows 13 and 15 with the ME value of 8 and 6, respectively are placed next to row 1. This procedure is repeated until all rows are accounted for and placed with one another. The detailed solution for columns and rows are available from the author. The combined results of both columns and rows for the BEA are presented in incidence matrix (3) is shown in Fig. 8. The clusters (mutually exclusive submatrices) in

72

S. Dowlatshahi / European Journal of Operational Research 115 (1999) 5976

Fig. 6. Incidence matrix (1).

Fig. 7. Incidence matrix (2).

S. Dowlatshahi / European Journal of Operational Research 115 (1999) 5976 Table 6 ME values for column 1 Position Column number ME j1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 j+1 2 3 4 5 6 7 8 9 10 11 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 Value 4 0 0 0 0 3 2 0 0 0 0 0 3 0 0 0 4 0 0 0 5 0 2 2 2 5 0 0 3 0

73

Table 7 ME values for row 1 Position Row number ME j1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 j+1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 Value 0 0 0 0 0 0 0 0 0 0 4 8 0 6

incidence matrix (3) present a structured matrix with a clear grouping of activities. The incidence matrix (3) indicates that the overall design of logistics can be accomplished in four self-contained modules. Module Family 1 addresses transportation issues. It consists of design for supportability, functional packaging requirements, the transportation mode, and transportability issues. Module Family 2 addresses manufacturing issues. It consists of design for manufacturability, materials, and manufacturing processes. Furthermore, Module Family 3 and Module Family 4 address production planning and control and design characteristics, respectively. For example, Module Family 2 focuses on the issues and problems related to manufacturability, materials, and the actual manufacturing processes. In designing this module family, the designer considers only the necessary and relevant design factors. By so doing, the scope and degree of complexity of the design become more manageable. The logistics designer simply focuses on such design factors as: standardized parts, type of manufacturing processes, production volume, material properties, type of materials used, and physical properties (width, height, length, center of gravity, etc.). As is evident, this module family not only considers the commonly used manufacturing design factors, but also cuts across other modules (manufacturing processes a5 , packaging materials a9 , transportability issues a15 ) and includes other relevant and necessary design factors. This simply signies that the manufacturability module is not an isolated activity and may not be designed in a vacuum-like environment. On the contrary, this modular approach enhances the eectiveness of the manufacturability module by reaching and incorporating other relevant and essential design factors which may have otherwise been disregarded. The design eectiveness is further enhanced by the manageability and ease of implementation of such a modular design. Similarly, one may consider Module Family 3 which combines ve separate modules from three dierent subsystems of the logistics design. This module family proposes that logistics engineering, manufacturing logistics, and design for packaging are interrelated and must be considered

74

S. Dowlatshahi / European Journal of Operational Research 115 (1999) 5976

simultaneously. By the same logic, Module Families 1 and 4 may be analyzed and explained. 4. Advantages and implications The following represents a summary of advantages and practical managerial implications of the methodology of design for logistics. 1. The methodology presented here allows the designer to be an active participant in the design of logistics systems. The initial incidence matrix is developed by the designer. The nal solution to the problem is dependent upon the initial matrix. The designer is an inseparable part of the identication and development of subsystems, modules, and design factors and, for that matter, the entire logistics decision-making process. The designer has the exibility to select and assign design factors to modules based upon a designer's judgment, skills, and the unique requirements of each logistics design. In this methodology, the designer controls the parameters and the direction of the problem at hand and not a predetermined and rigid model. 2. Various solutions can be developed based on diering initial solutions. This capability enhances the eectiveness, as well as the eciency of the solution procedure, and provides a great deal of exibility for the designer. 3. This methodology is applicable to matrices of any size or shape. The only requirement is that the elements of a matrix be non-negative. The BEA solutions are nite. This makes the BEA algorithm applicable to new designs as well as currently existing designs. The nal solution obtained by using this algorithm is independent of the order in which the rows and columns are presented. 4. This methodology generates modules that are cohesive, bounded, or contain a self-contained group of activities. For eective implementation of integrated logistics design, each module solves one clearly dened segment of the total system. Modularity, as a basic rule of good design, is more easily changed, expanded, or contracted than monolithic system designs. Most managers nd modular system designs easier to understand and apply. These module families can be designed and

Fig. 8. Incidence matrix (3).

S. Dowlatshahi / European Journal of Operational Research 115 (1999) 5976

75

implemented simultaneously. This can potentially reduce the total design cycle time and bring about the advantages of concurrent engineering to the design of integrated logistics. 5. The analytical approach presented in this paper is capable of managing and organizing a large number of design factors involved in the design of integrated logistics with considerable ease and less computer CPU time. The computation time, which depends solely on the size of the matrix, is usually minimal and insignicant. The simple program written to run this example is available upon request from the author. The BEA is fairly easy to apply and understand by logisticians and higher level management. 6. The design of large-scale manufacturing systems such an logistics design usually requires numerous components that must be expressed by specications. There may be a variety of specications necessary to provide the guidance and controls associated with the development of the system and its components (Blanchard, 1991). The development of these specications and criteria can be eectively accomplished by modular design. The undertaking of a very large-scale system as a whole, with all its complex components, may be inecient or, at the very least, cumbersome. 7. The BEA allows the inclusion of values other than 0 and 1 in the incidence matrix (1). The design factors in this paper are assumed to be reduced to their lowest format allowing for no further decomposition. It is, however, possible to consider higher importance to certain combinations of module-design factors by assigning a higher value than 1 to these combinations. The BEA is equipped to handle these situations well since the placement of the adjacent columns and rows are based on the ME values, and since the ME values use aij in their calculations. The preceding points provide some advantages and managerial implications for the use of the methodology of design for logistics. The selection of any design tool, however, is largely dependent upon the objectives and complexities of the logistics design system. This selection is also aected by the level of expertise and willingness of the designer to actively participate in the process.

4.1. Conclusion and assessment A concurrent engineering environment provides a suitable venue to consider logistics problems since logistics concerns, contributions, and constraints are best addressed in the early phases of the product design cycle. The advantages outlined for concurrent engineering earlier in this paper can be realized if logistics requirements, as a part of the overall product design, are considered. The conceptual framework provides for an effective tool for the logistician to include the necessary and relevant subsystems, modules, and design factors. This approach allows the designer to become a full participant in the logistics systems design. BEA provides an ecient clustering algorithm for solving logistics problems. This algorithm generates self-contained clusters that can be more easily understood, changed, and implemented. This feature allows for a logistics system to be implemented more eectively and in a shorter period of time since the independent clusters need not be implemented sequentially. Fig. 9 illustrates how the four module-families obtained in incidence matrix (3) can be overlapped to reduce the design cycle time. Note that the bars are not drawn to scale. This paper has specically explored a large number of areas where collaboration and interface of logistics and design activities can result in signicant achievements for a manufacturing

Fig. 9. The Gantt chart of overlapping module families.

76

S. Dowlatshahi / European Journal of Operational Research 115 (1999) 5976 Cherf, R.M., 1995. Reducing board costs. Circuits Assembly 6 (3), 82. Cohodas, M.J., 1988. Make the most of supplier know-how. Electronics Purchasing, July, pp. 3839. Colling, D.A., 1991. Materials and product safety. Professional Safety 34 (4), 125. Corbett, J., 1986. Design for economic manufacture. Annals of C.I.R.P. 35 (1), 93. Implementing Concurrent Engineering for SMT, 1992. Electronic Packaging and Production, August, p. 22. Fawcett, S.E., Closs, D.J., 1993. Coordinate global manufacturing: The logistics/manufacturing interaction, and rm performance. Journal of Business Logistics 14 (1). Foo, G., Clancy, J.P., Kinney, L.E., Lindemulder, C.R., 1990. Design for material logistics. AT&T Technical Journal 69 (3), 61. Germain, R., Droge, C., Daugherty, P.J., 1994. A cost and impact typology of logistics technology and the eect of their adoption on organizational practice. Journal of Business Logistics 15 (2), 227. Huthwaite, B., 1988. Designing in quality. Quality 27 (11), 34 35. Kyng, M., 1991. Designing for cooperation: Cooperating in design. Communications of the ACM 34 (12), 64. Mather, H., 1992. Design for logistics (DFL) The next challenge for designers. Production and Inventory Management Journal 33 (1), 710. McCormick, W.T., Jr., Schweitzer, P.J., White, T.W., 1972. Problem decomposition and data reorganization by a clustering technique. Operations Research 20 (5), 993 1009. Nevins, J.L., Whitney, D.E., 1989. Concurrent Design of Products and Processes. McGraw-Hill, New York, pp. 23. Petersen, H.C., 1993. Logistics, transportation and distribution: Engineering service and prots. Industrial Engineering 25 (12), 21. Starsmore, R.P., 1990. History of a wet gas transportation pipeline from design through decommissioning. Corrosion Prevention and Control 37 (5), 117. Whitney, D.E., 1988. Manufacture by design. Harvard Business Review, JulyAugust, pp. 8391.

enterprise. Although these areas of collaboration are equally applicable to a variety of manufacturing logistics, care must be taken that a specic, relevant, and custom made program resulting from the unique requirements and environment of each rm is selected. This process allows for a better focus on the ensuing issues that require more immediate attention. The most important and essential prerequisite for an interface between logistics and design, however, remains to be the elimination of the ``over-the-wall-design'' concept. All collaboration and interface must be done on a long-term basis unless circumstances dictate otherwise. Logisticians should be given an essential role as the key player in the design process. This process is certain to fail, if legitimate authority and power is not delegated to the logistics function. Top management should genuinely encourage logistics involvement. If the design process is viewed as an abstract and vacuum-like process that is performed sequentially rather than concurrently, no signicant achievements are expected to occur. The eective dialogue between logistics and design can only occur when the barriers and walls whether real or imaginary are removed. The support of top management and a positive institutional culture are essential to instill and foster such an environment. References
Ballou, R.H., 1987. Basic Business Logistics, 2nd edition. Prentice-Hall, Englewood Clis, NJ. Blanchard, B.S., 1991. System Engineering Management. Wiley, New York, p. 65.