Computers ind. Engng Vol.15, Nos 1-4, pp.85-90, 1988 Printed in Great Britain. All rights reserve 0360-8352/88 $3.00+0.00 Copyright c 1988 Pergamon Press plc

Industrial engineers face problems requiring complex and time-consuming analyses every day. The task of collecting and organizing data, and calculating relationships becomes a Significant portion of each day's workload. The mathematical models employed by industrial engineers require sophisticated knowledge and analysis techniques.

The literature presents many different models for analyzing operations problems. But the calculating power required when implementing any of these models in a real working environment exceeds manual capabilities. Fortunately, the availability of computers has provided industrial engineers the opportunity to model some very complex situations, and compute fairly sophisticated solutions.

In the last several years, computer technology has developed the power to provide such computational power in a local, desktop environment. Unfortunately, software for such purposes was lacking. Now, however, several general purpose analyses packages, and other, more specific modeling software packages have been offered to help industrial engineers in their analyses.

The decisions being made now are based on analyses made by engineers using desktop computers in a local environment. They are using recently developed software which provides the sophisticated models used in such analyses. This approach reflects the decision support systems process now being implemented in many situations.

My purpose in this paper is to describe the Decision Support System (DSS) approach to problem solving in an industrial engineering environment. The recent development of micro-computers and related software to aid the engineering problem solver will also be presented. The industrial engineer is ideally suited to implementing a DSS using the micro-computer hardware and modeling software now available. Finally, I will review a number of general and specific DSS software packages which can assist the industrial engineer when undertaking the complex analyses necessary to make good decisions.

DECISION SUPPORT SYSTEMS (DSS):

The early rational approach to management advocated by such writers as Weber, Fayol, and Gulick proposed a universal set of functions and principles which would lead to organizational success. The Hawthorne studies of Mayo, Rothlisberger, and Dixon were in reaction to the mechanistic environment inherent in such approaches to management. The industrial engineer, however, works in a management environment that developed from the Frederick W. Taylor school of scientific management. All of these approaches to the management topic embrace the importance of decision-making.

Recently, the concept of Decision Support Systems has been advanced to integrate the several foci of the various approaches to the study of management. DSS has been described in the literature by several authors. For the purpose of this paper, I will define DSS as a management environment which systematically integrates information, analytical models, computational hardware and software, and human judgment in support of the decision process.

The decision process involves six steps (Harrison). First is the problem definition and setting of objectives. The second step is concerned with searching for alternative solutions for the problem defined in step one. Step three evaluates each of the alternatives generated in the previous step. It is important to separate these two steps. Choice of the most appropriate alternative is step four. Next comes implementation of the chosen alternative. Finally, evaluation and follow-up of the chosen strategy leads back to the first step, definition of the problem, in a circular fashion.

There are three criteria for good decisions. Reiss suggests two of them, efficiency and effectiveness. Efficiency is a concept that suggests the ratio of output to input can be used to measure decision success. Effectiveness suggests that goal achievement is an important criterion of good decisions. Finally, Vroom and Yetton add to these two criteria that of timeliness. In addition to the cost-benefit ratio and the degree of objective attainment, (also characterized as the quality and acceptance of the chosen alternative by Maier) the amount of time available to consider alternatives becomes an issue. The choice step may be imposed on the decision process earlier than optimally when action on an issue must be taken.

The final foundation I will present for this review describes the components of a Decision Support System. My definition suggests the four components of information, analytical models, computer hardware and software, and human judgment. Information can be described as the systematic accumulation and organization of data which addresses relevant decision issues. Often large amounts of information are required, and must be manipulated in efficient and effective ways to assist the engineer in an analysis. Sources of information in the electronic age include organizationally specific and locally generated data derived when processing routine accounting and manufacturing records. Also, information resources available through electronic media often can benefit an analysis.

Analytical models are defined as simplified representations of reality which can be manipulated by the analyst to observe outcomes without risking the real system. Industrial engineers usually use symbolic, quantitative models which are either prescriptive or descriptive, and stochastic or deterministic. Though simplified, these models still require massive computational power when applied to problems of interest.

The availability of computers and software have supplied the required computational power. Originally this power was centralized in the hands of systems analysts and programmers. The availability of the desktop environment has made this power accessible to users more directly. Suitable software, analyzing the problems of interest to the industrial engineering environment has made this power usable by the analyst directly. No longer must an industrial engineer await the support of a technician to program a remote and sophisticated computer to solve a relatively simple problem requiring large amounts of computations. The immediate availability of a desktop computer and software designed to analyze the particular problem at hand affords the engineer an opportunity to implement decisions in a more timely fashion, and to evaluate more alternatives before making a final choice. This capability allows for the input of the human judgment so necessary in DSS. A timely manipulation of data and models analyzing a system provides the engineer an opportunity to get a feel for the way reality might act when manipulated in various ways. The judgment of the analyst and others responsible for implementing the chosen alternative can be integrated with the results of the analysis for the most informed decision. The components of DSS are integrated in this way to make efficient, effective, and timely decisions.

INDUSTRIAL ENGINEERING ENVIRONMENT:

The environment in which industrial engineers work involves designing systems integrating people and machines for the purpose of providing goods and/ or services. A major effort on the part of most IE's is the productivity improvement effort necessary. This can often be accomplished by designing for system optimization. Focusing on such criteria as productivity improvement and system optimization, IE's may benefit by adapting the DSS approach to their systems design efforts.

Both the hardware and the software available to the IE have evolved from rather centralized, and unknown resources, to the current state. IE's may have on their own desks both the computer and the problem specific software to implement a relevant DSS to employ when designing the human/ machine systems for which they are responsible.

The original development of hardware led to physically large, well secured computing machines isolated from the users. Access to these resources was restricted to the systems analysts and computer programmers. IE's found it necessary to enlist the aid of these people in developing their analyses. Such approaches were time consuming and sometimes inaccurate.

The original software available for analyses required a unique computer program for each analysis. The code written for these models was often difficult to verify and became costly to generate and modify. Subsequent development of the higher level programming languages such as FORTRAN, PASCAL, and BASIC assisted analysts. But the generation of code programming computers to analyze the desired models was an inefficient way to operate.

Both of the above environments required highly sophisticated technical support. In the late 1970's we saw the introduction of the micro-computer technology. This technology soon resulted in the hands-on availability of a computer for nearly anyone who had a need for computational power. Likewise, software was developed which greatly enhanced the ability of the individual user to develop and analyze custom designed models. The early spreadsheet packages like Visi-calc and Multiplan were key software packages in this stage of evolution. These were augmented by file management and data based management packages which helped IE's collect and organize data they would otherwise not have access to.

There has been a recent development of model-specific software packages. This has been coupled with the recent development of desktop computers. The result provides IE analysts with the technology necessary to implement a DSS when addressing the systems design for which they are responsible. Specific problems which have been modeled include resource allocation and transportation problems using linear programming models. Also included are demand forecasting, using both time and linear regression models. Waiting line and queuing theory models are also included, using models directly, or instead, applying monte-carlo simulation. The design of operating characteristics curves and X-bar and R charts is also available in some of these software packages.

There are several modeling packages available for the IE analyst. The general purpose analytical software is also helpful to an analyst. Many reviews of spreadsheet, data base management, and graphics software packages are available elsewhere. The purpose of the remaining section of this paper is to present a review of the features available in a number of packages which are specific to the analyst interested in modeling the real system.

SOFTWARE PACKAGE REVIEW:

There are two types of software packages available for analyzing operations based models. The first is developed as templates for the general analysis software packages. This usually is in the form of templates for LOTUS 123 which are designed to input data in specific ways, and then use the capabilities of the general program to compute the model results. Examples of this approach include 123 Forecasting templates for LOTUS 123, a package by Nathan and Cicilioni, published by West Publishing Company, and the Moore and Hammer package available from the Institute of Industrial Engineers.

The other type of package usually consists of a group of programs, often sharing a common data input format, which calls up a particular users model of interest for analysis. The packages in this group include STORM, the IIE packages, a package by Erickson and Hall (2nd Ed) published by Addison-Wesley, one by Lee and Shim (MicroManager) published by W. C. Brown, another by Dennis and Dennis (MicroModels for DSS) published by West, and finally, one by Chang and Sullivan (Quantitative Systems for Business) published by Prentice-Hall.

The review of these packages will include a listing of the models available in each, their limits and capacities, the user-friendly features provided, price considerations, and a commentary about the ease of use and helpfulness one experiences when employing these models.

While a number of these packages are intended for academic use in the classroom, their modest prices make them very attractive to the practitioner also. Many are menu-driven, allowing even the naive computer user to maneuver through the programs and derive benefits from their use. The preliminary figure provided below suggests examples of the way in which I will summarize the results of my review.

SOFTWARE REVIEW:

The software packages reviewed for this presentation are designed for use as teaching software in the classroom. They are often employed as supplements to regular texts in standard Production/ Operations Management or Operations Management courses. The publisher's packages described below are relatively inexpensive, and may be used in many operations environments. These will be compared to the Microsoftware packages offered by the Institute of Industrial Engineers.

Certain characteristics of all of these packages make them suitable for use in Decision Support Systems. They all contain several analytical models, can be used on computers with local data, and require intelligent interpretation by the analyst in their application. These basic packages include several time-series and correlational forecasting approaches, linear programming and some of its specialized applications, deterministic and stochastic inventory analyses, project management and network flow models, and many other less obvious modeling topics.

In addition, each package provides documentation, data entry and saving/ retrieval, printing of results, and some even provide a degree of graphic presentation. Most of the packages are provided as executable programs organized by a main menu initial program. One is a set of LOTUS 1-2-3- templates with auto-executing macro's and custom menus (Nathan and Cicilioni). The IIE set are provided as many independent stand-alone programs in interpreted basic. Below is provided a summary of each package.

Quantitative Systems for Business (Chang and Sullivan); This package contains several pattern based forecasting models, a simple linear regression correlational model, a full complement of linear programming applications (normal, integer, transportation and assignment problem models), project management and more general flow analysis models, and a limited number of inventory models. In addition, queuing models and waiting line simulation models are provided, as are decision theory analyses.

The programs in this package are provided on two copy keyed 5 1/4" diskettes making it difficult to run them on PC's equipped with hard drives. The original master diskette must be kept in drive A: to begin the program. Programs and subroutines are accessed through menus. Color is supported. Printed output is accomplished through a printscreen routine either invoked directly from a menu or through the F8 key at any time. Extended ASCII screen characters cause some printers to print non-standard output. The forecasting routines include moving averages, exponential smoothing, and adaptive EXSM. Nothing is provided to include seasonal variation in the time-series models. Data entry is interactive and intuitively organized for most models. The small paperback book provided with this package is adequate to describe how the user is to operate each model, but is not sufficient to learn the analysis techniques.

Microcomputer Model for Management Decision-Making (Dennis and Dennis); The second edition of Dennis and Dennis contains a wide variety of pattern based forecasting and simple linear regression models, an adequate complement of linear programming and its transportation and assignment applications, a complete set of probabilistic and deterministic inventory models, and PERT project and network flow analyses. Also available are decision theory, markov, and queuing analyses.

This package is provided on two copy keyed 5 1/4" diskettes making it necessary to run the original in drive A:. Menus are used to move from program to program, and to choose specific models. Color is supported. Printing of output is available after analysis is called from a menu. only subproblems and their data can be saved to disks. Data alone is not savable, making it necessary to reenter data when comparing different forecasting models. The seasonal variation is not analyzed by these models directly. The full sized paperback book provided with these models adequately describes how to use them, but requires additional instruction to understand data entry and interpret results.

Storm (Emmons, et al.); The Personal Version 1.1 of-this package provides exponential smoothing time series forecasting models which incorporate average, trend, and seasonal components of variation, and an extensive implementation of multiple linear regression for correlational forecasting analyses. Also available are linear programming, transportation, and assignment models. A limited number of inventory and PERT/ CPM project management models are provided. Material Requirements Planning, queuing theory, investment analyses, and facilities location models are also available.

This package is also provided on two copy keyed 5 1/4" diskettes making it necessary to run this package with the original disk in drive A:. It is also menu driven. Results may be printed by choosing that function using function keys. Time series forecasting includes seasonal analysis. Data entry is accomplished easily, and the data may be stored and retrieved by menu commands. The user also has the opportunity to view the directory of files to recall data file names. The 300+ page paperback documentation is an excellent and detailed source of information on these models, and teaches their use on the computer well.

Computer Models for Management Science (Erickson and Hall); A major shortcoming of CMMS is a lack of forecasting models. It does provide an extensive set of linear programming models and applications (including integer programming, transportation and assignment algorithms), basic deterministic and stochastic inventory models, and network models (PERT/CPM and flow analysis). Other models include decision theory and Markov analysis, and queuing models.

This package is provided on one 5 1/4' floppy diskette which is not copy protected in any way. This makes it easy to transfer to a hard drive and run from that drive. It is also menu driven, and printing of results may be accomplished choosing a proper menu selection. Data entry, and the saving and retrieval of data is accomplished through menus. This package is also available for Apple II computers. The paperback documentation guides the user through the use of the package well, but requires supplemental instruction on the models for data entry and results interpretation.

The above four packages all are similar. They use menus, they present a group of analytical models, they provide for both keyboard and disk input of data, and they provide for hard copy output of results. They are also limited in problem size. A finite number of data points can be analyzed, which may inhibit their generalization in actual operations environments.

Also, their documentation is primarily for student learners and not operating practitioners. On these two criticisms, Storm can be pointed to as an alternative which has the potential to perform better. There is a professional version available at a higher cost, which increases the limits on data, and the documentation is far superior. Other packages exist which present different forms of analysis than the menu driven Decision Support System modeling. First I will present a review of the package of LOTUS 1-2-3 templates which provides many models similar to those described above. Finally, I will discuss the IIE Microsoftware and compare it to the packages discussed above.

A Spreadsheet Approach to Production and Operations Management (Nathan and Cicilioni); This package is the only one reviewed here which consists of a set of LOTUS 1-2-3 templates In order to use it you must first load LOTUS using the templates in the data disk. It consists of an auto-load worksheet which contains macro menus to analyze various problems using spreadsheet models. This particular package includes a limited set of time series forecasting templates (moving average and EXSM), nothing in linear programming models, simple EOQ and ELS deterministic inventory models, and nothing for network flow or PERT/CPM analysis. It does include a work measurement summary, two facilities location models, two production scheduling algorithms, an interesting materials requirements planning template, two statistical quality control templates. Macro's are used to load and save templates and to print resulting worksheets. Because of the nature of LOTUS 1-2-3 spreadsheet models, most of the templates are less generalizable than the models presented above; however, the LOTUS graphics are exploited well by using macro's and the F10 key to display information pictorially.

This package is also provided on an unprotected 5 1/4" diskette, allowing easy use on systems equipped with hard drives. The extensive use of menu macro's provides a great deal of user convenience, including the saving and retrieving of data and printing results all done from menus and macro commands. The package assumes familiarity with LOTUS 1-2-3, and requires the user to own a suitable copy of the spreadsheet software in order to be able to employ these worksheets. The paperback documentation is adequate to run these models.

Institute of Industrial Engineering packages (Whitehouse; By far the most versatile set of programs reviewed for this presentation is the Microsoftware provided by IIE. These programs come in separate packages of four or so in each. If you include all of them, a large number of models are available, including several EXSM and Winter's time series forecasting models, multiple linear regression for correlational modeling, a limited set of linear programming models, another set of deterministic inventory control models, and network flow and PERT/CPM models. In addition, an excellent set of work measurement analyses (time study summaries, work sampling scheduling and summaries, etc.,), some facilities location analyses, and excellent statistical quality control models are available.

These models are provided in each package on 5 1/4' floppy diskettes, unprotected, in interpretive basic. Each program is self-standing, and no attempt has been made to integrate the packages with menus of any sort. I would have expected them to at least be in compiled form, and the extra effort necessary to choose a proper model from a menu doesn't seem to me to be hard to program. The documentation for these models is most impressive. Ring binders with clear presentations about the models being used and how to run the programs are provided with the software. Most of the programs do include menus to execute various functions, including data saving and retrieving, and printing. However, the high cost of these packages seems to prohibit the average user from acquiring all of the models available. At $140 for IIE members ($175 for others) for each set of four programs, the cost of duplicating the models available in some of the publisher's packages seems extravagant. However, for some of the analyses (multiple linear regression in particular) the IIE packages are clearly superior. And the variety of applications available from the IIE models is also greater.

SUMMARY AND CONCLUSION:

The Decision Support Environment requires several components. Among them are data organized into information, human judgment, analytical models, and computer hardware and software. The current state of the technology in desktop computing hardware is far advanced over what was available only five years ago. Just now the software for integrating these components is being brought to market. In this review I concentrated on several packages intended primarily for teaching the application of desktop computers to Decision Support models. I concluded with a description of the IIE microsoftware available.

For the purpose of teaching DSS, I would consider most of these packages adequate. The only choice being the specific model preferences of the teachers. For the purpose of analyzing real operations in a DSS environment, only some of them are appropriate. I would conclude the Storm package is an excellent DSS choice, and particularly for the larger problems which could be analyzed with the more extensive professional version. The IIE packages are more elaborate in their coverage, but lack the user friendliness that could easily be programmed into them. Their price also seems prohibitive.

An analyst seems to have available many adequate choices to consider in picking among various alternatives. The convenience of the commercial and educational packages suggests that increased emphasis in modeling operations environments may lead to more effective decisions. These packages can assist the analyst in being both efficient and effective in DSS environments.

BIBLIOGRAPHY

1. Chang, Yih-Long, and Robert S. Sullivan, Quantitative Systems for Business (Prentice-Hall, Englewood Cliffs, NJ, 1986, $32.00).

2. Dennis, Laurie B., and Terry L. Dennis, Microcomputer Models for Management Decision-Making , second-Edition, West Publishing Co., St. Paul, 1988, $36.00) .

3. Emmons, Hamilton, A., Dale Flowers, Kamlesh Mathur, and Chandrashekhar M. Khot, STORM: Quantitative Modeling for Decision Support, (Holden-Day, Inc., Oakland, CA, 1986, $29.95).

4. Erickson, Warren J., and Owen P. Hall, Jr., Computer Models for Management Science, Second Edition, (Addison-Wesley Publishing Co., Reading, MA, 1986, $39.65).

5. Nathan, Jay, and Ruth Yaron Cicilioni, A Spreadsheet Approach to Production and Operations Management, (West Publishing Co., St. Paul, 1987, $18.90).

6. Whitehouse, Gary E., Ed. Forecasting, (Institute of Industrial Engineers, Norcross, GA, 1984, $140 IIE members, $175 non-members).

7. Whitehouse, Gary E., Ed. Operations Research, (Institute of Industrial Engineers, Norcross, GA, 1984, $140 IIE members, $175 non-members).

8. Whitehouse, Gary E., Ed. Plant Layout, (Institute of Industrial Engineers, Norcross, GA, 1986, $140 IIE members, $175 non-members).

9. Whitehouse, Gary E., Ed. Production Control (Institute of Industrial Engineers, Norcross, GA, 1983, $140 IIE members, $175 non-members).

10. Whitehouse, Gary E., Ed. Project Management, (Institute of Industrial Engineers, Norcross, GA, 1987, $140 IIE members, $175 non-members).

11. Whitehouse, Gary E., Ed. Statistical Quality Control, (Institute of Industrial Engineers, Norcross, GA, (Institute of Industrial Engineers, Norcross, GA, 1987, $140 IIE members, $175 non-member

12. Whitehouse, Gary E., Ed. Work Measurement, (Institute of Industrial Engineers, Norcross, GA, IIE members, $140, $175 non-members).

Dr. Eliot S. Elfner, P.E., Associate Professor of Business Administration, St. Norbert College. Dr. Elfner serve as a Consultant/Evaluator and Commissioner-at-Large for the North Central Association Commission on Institutions of Higher Education, and is the owner of an organization which provides computer systems for small businesses. He is a senior member of IIE and has served as an officer and president of Chapter 112. He is also a member of the Academy of Management and the Decision Sciences Institute.

To contact the St. Norbert College Web Administrator, send e-mail to: webmaster@mail.snc.edu.
 

This page approved by St. Norbert College. URL of this page: http://www.snc.edu/socsci/chair/333/333_dss.htm  Last modified: January 15, 2002