INTRODUCTION
A revolution has occurred over the last forty years in computer applications for business. Emphasis has grown from simply processing data to much more sophisticated uses of computer data and models for managerial decision making. In particular, over the last fifteen years or so, managers/decision-makers, and system specialists have begun to understand that standard reports and large amounts of data cannot always provide the specific information needed. Since the late 1970‘s, a great deal of effort has been devoted to systems generically called decision support systems (DSS), to provide the manager/decision maker with specific data about problems faced. These systems often take the form of models which may be useful in evaluating possible alternative solutions. In this article we discuss the concept and evolution of decision support systems as they have been developed for business applications. In particular, we are interested in how to assist the decision maker in project management. In addition, we provide some examples of decision support systems used in broad business applications.
WHAT ARE DECISION SUPPORT SYSTEMS?
There is a major difference between decision support systems and the more traditionally-understood “management information systems” (MIS). The MIS usually provides fairly large amounts of processed information to managers in the form of pre-defined reports and query outputs. The decision support system allows the manager to have access to data and information of specific interest from numerous sources, both internal and external to the organization. The manager can then evaluate that data using computerized models. These models, an integral part of most DSS, frequently allow the decision maker the option of assessing the outcomes of numerous alternatives. This requires little effort and allows decisions based not only on your own experience, but also on the simulated evaluation of the model-produced alternative outcome. DSS models cannot be considered“decision makers” themselves, but only support tools for you, the ultimate decision maker.
The model is the center of any DSS. The user is on one side of the model and the needed data is on the other side. When the user wants something, the model is accessed and the user explains what is wanted. The model then goes about arranging the data in the desired format expressed by the user. Models will never perfectly reflect the real world but will give microcosms of real world situations. The user will then build and combine from various model attempts to get the desired result. Models are thus an attempt to gain a better understanding of the real world or actual system.
“Decision Support Systems”
Rudy Boznak, United Research
Today's executives face an unprecedented rate of technological innovation and accelerated product obsolescence, which open and close “windows of opportunity” faster than ever before.
But what has not changed is the executive's role. Executives must continue to formulate and operationalize effective strategies to meet the challenges of a changing business environment. “While both efforts are critical to corporate vitality and survival, operationalizing strategic initiatives is by far the most difficult task for most executives,” states Rudy Boznak, Senior Management Consultant for United Research, a New Jersey- and London-based management consulting firm.
“One of the ways we help executives accelerate strategic change is by creating a decision support model called SIMPLEX (Strategic Initiative Management and Planning for Executives),” explains Boznak. “Using this methodology and Viewpoint—project management software, executives can ‘model’ if, how, and when their organization can accomplish multiple strategic initiatives.”
Boznak believes that its been only in the last one to two years that project management packages have had true “multiproject” capabilities and graphics to enable executives to visualize their resources, The firm's methodology combines the Optimizing and Prescriptive models used in DSS to help executives align strategic needs and organizational capacities. As are-suit, executives can better predict the likelihood of implementing their strategies, and of achieving competitive advantage.
Decision support systems and models come in varying degrees of completeness and complexity. Perhaps the simplest DSS, and the one with which most managers are probably familiar is the computer augmented spreadsheet. Examples would be Lotus 1-2-3®, SuperCalc®, etc . . . . Spreadsheets allow the user to develop their own models for particular decision situations and use various scenario data inputs to evaluate outcomes on a “what-if” basis. Spreadsheets have proven to be very popular and very effective. This is particularly true in financial and accounting environments. Recent years have seen a widening of their use in all functions of businesses, including manufacturing and project management.
More sophisticated decision support. systems and models can integrate data from several sources and use specialized situations. These are used to predict outcomes and in many cases to recommend specific action to you the decision maker. As computer development becomes more and more sophisticated and as artificial intelligence and expert systems begin to be integrated into decision support systems, the manager continues to have greater opportunities. The future will see DSS used in searching for solutions to less-and-less structured decisions.
STRUCTURED VERSUS UNSTRUCTURED DECISIONS
Most early computer applications for “decision support” could probably be categorized as structured in nature. The models dealt with highly quantifiable decisions such as determining inventory levels, economic order quantity, etc. Most current development is being done toward the less structured end of the decision continuum. Decision making in the future may have some quantifiable components but, in general, will be much more subjective in nature. In fact, some of the most valuable decision support systems developed in the past few years have been used for such purposes as determining plant location, making judgments concerning extension of credit, capacity planning, and allocation (or reallocating) organizational resources. Most of these models were used to improve productivity and profitability.
One major difficulty in the evolution of DSS has been that most organizations need personalized, non-standard, systems to support the types of decisions they make. This, of course, means every decision support system must be unique. This causes the cost of developing DSS to increase dramatically. Furthermore, different decision makers in an organization have their own decision-making “style.” For an organization's DSS to be effective, it must be able to respond to each decision maker in a manner which is most effective, else it will not be utilized to its potential.
Decision support systems, while originating with models for operational management decisions developed by management scientists, have now evolved to include the full spectrum, including decisions at both the tactical and strategic levels in the organization. They have, therefore, become useful not only for helping managers determine operations requirements but also for controlling manpower, material, and machinery resources for all types of projects.
The following are examples of models used in DSS with a brief description of what each does.
Descriptive models. Concentrates on identifying the problem. An example would be the use of queuing theory to determine where problem lines may occur in supermarket check-out lines.
Perscriptive models. Directs activities. An example would be to tell you how many extra checkers are needed during peak periods to solve the supermarket line problem.
Satisficing models. Allows you to achieve a satisfactory solution within given constraints. An example would be in scheduling production runs when operating under constraints such as limits on inventory, manpower, etc. The solution is thus not the best but the one that at least satisfies requirements under the constraints in which you must operate.
Optimizing models. Finds the best solution when given a set of performance criteria. An example would be in designing production runs when optimum efficiency is desired.
A Decision Support Tool for Successful Project Implementation
Kathy D. Phipps of EDS/Simulation Services
The concept of pro-active instead of re-active management is sometimes difficult to sell, especially to managers of production systems. The prevalent attitude is: “I‘m too busy fighting fires to have time to p1an.” But recently, simulation helped to change this attitude by selling the concept of planning to the managers of an automotive paint shop and by performing as a tool for communications and team building.
The traditional method of designing a new system is to have the process engineers design the manufacturing process, give contracts out to bidders, build the manufacturing line and then turn the line over to plant personnel to run. At that point, the engineers often exit the picture and let the plant personnel learn how to operate the system. This method usually leads to re-active management.
However, the simulation process alters the standard practice, builds teams, and results in a better system. The simulation model requires data from all aspects of the shop, i.e. layout, scheduling, process, routings, and control logic. Consequently, input is necessary from many sources. In the paint shop project, this was accomplished through open forum meetings with personnel of various responsibilities. The focus of these meetings was operational philosophy. Differences of opinion that came to light were resolved. These meetings helped to develop a team concept and a greater understanding of the entire system.
The model was constructed in an animated simulation language. The original question asked of the model was: “Will the system produce the required throughput?” As the model development progressed and the question was answered, the management team discovered new uses for the model. One of the more obvious uses was as a communication tool. The model was used to instruct the line supervisors on plans for the new shop, giving them an overall view of the system.
The simulation language allowed execution to be stopped and parameters changed at any time. This capability allowed the managers and other users to set up specific scenarios and examine the behavior of the system. This proved to be a valuable tool for crisis planning and training of personnel. One experiment examined the effect of a major fault in the paint system (five hours of 100% rejects) and allowed the management to plan for such an occurrence.
At the beginning of the project, most team personnel had no experience with simulation or planning. At the end, all were convinced that to do any system modifications without modeling them first was not a good idea. They also concluded that crisis planning was performed much easier on a machine than on the system.
Predictive models. Uses past data to predict future events. An example would be the use of life and mortality data to predict the life span of now living people.
A few of the above models along with tools for implementation are discussed in more detail below.
DSS AND PROJECT MANAGEMENT
Many of the mid-century-developed models such as PERT and CPM have been integrated into decision support systems as a means of unifying a computerized base of historical data. Use is also made of operating and condition data and forecast data. These models use this data to make planning and managing projects easier and more readily responsive to changing organizational and environmental requirements. In fact, with the advent of the microcomputer, project management software, which may be considered a specialized application of decision support systems, has become available for even relatively small organizations at a low cost. These software packages make use of many of the traditional models to help managers develop schedules for resources and activities, monitor levels of materials and supplies, track schedules, and detect potential schedule failures. This is done so early action may be taken to reallocate resources, often with suggestions from the project management software. These are general examples of support a DSS can give to the project manager.
TOOLS AND TECHNIQUES FOR PROJECT MANAGEMENT
A number of tools which were developed to handle fairly structured problems have been incorporated into the decision support systems of many organizations and are particularly useful in project management. Among them are inventory control, network analysis models (for communications systems), decision trees, purchasing management and manufacturing control models.
Most of these tools are useful in managing projects in all functional types of models described above. These tools assist in engineering, marketing, manufacturing, and systems development. As technology has improved over the past decade, almost all of the models and tools have become available, not only in mainframe environments, but also in microcomputer environments. The desk top PC is the area in which most development is currently being made. A few examples of current and potential applications will highlight the importance of DSS in project management.
Project schedule uncertainty.
While PERT was developed to assess uncertainty and thus the probability of an event occurring on or before a specific date, its logical basis was flawed. The uncertainty was assumed to be dependent only upon the most critical path leading to that event. Recent additions to project management scheduling software have included packages such as Welcom Software's OPERA®—which uses a Monte Carlo simulation approach to assessing the same probability but based cm the uncertainty of all paths leading to the event of concern. It can be expected that other elements of project risk will be addressed in future software developments.
Decision trees. In dealing with decision making under conditions of risk, DSS which include the capability of evaluating decision trees allow the evaluator to assess multiple-outcome sequential decisions. Use of sensitivity (what-if) analysis allows for evaluating conditions not only as probabilities for potential outcomes but also as new alternatives. This is done as new states of nature enter and current ones leave the model. In using decision trees, the optimal decision has traditionally been the alternative having the greatest expected monetary value. With more sophisticated models incorporated into some DSS, noneconomic quantifiable values can be used. It should be noted that expected value theory is based on the assumption of repeated occurrences of the same decisions. The “one time nature” of project management and related decisions also requires analysis of worst case scenarios, a inherent capability of decision trees.
Purchasing management models.
One of the tasks in the environment of project management is to schedule not only personnel but also the purchase and delivery of materials. In those projects in which large quantities of numerous items are required, a DSS which allows the purchaser to evaluate delivery schedules, competing vendors’ prices, warranties, quality, etc., is virtually indispensable. With service (penalty) costs as well as both ordering and holding costs to be considered, unstructured decisions (in addition to the usual structured ones) can be supported with a DSS that allows us to perform sensitivity analysis on intangible components. Models have been developed to allow the project manager to evaluate the risk of ordering low-cost but long-lead-time components for projects not yet approved or for which the contract has not been awarded, These types of decision support systems can support two scenarios. One is those cases in which a project is -already committed and the goal is to minimize expected cost, A second is those cases in which the decision maker is bidding for the project and wishes to evaluate associated costs before proceeding with the bid.
Manufacturing control models.
Decision support systems have been developed which can integrate realtime decision making and control with concurrent simulation capability in manufacturing environments. The goal is to derive the maximum utility from the flow of raw materials. By viewing the long-term manufacturing process as a project and incorporating merchandising of output as an integral part, managers can determine the effects of the DSS logic as well as explore alternative capability, this type of model (as well as others discussed) would lack the characteristics needed to make it a decision support system, It would instead, be another management information system, albeit in real time.
In each of the above cases, one of the major characteristics which distinguishes the decision support system from simply the structured model is its capacity to allow the decision-maker to evaluate alternatives using processes of simulation. This is accomplished with data from several sources and not just from the organization's internal data base.
DSS AND ALTERNATIVE DECISION CRITERIA.
Numerous criteria exist for decision making, the most common being optimizing and satisficing. Optimizing, the so-called rational economic model of decision making, can exist only if all possible outcomes are known and the values of the outcomes are certain. This is highly unlikely, Under conditions of uncertainty we can make decisions by optimizing the expected value of the outcomes rather than the outcomes themselves. DSS models can help the manager derive a state of near-optimization. Sophisticated decision support systems can provide modeling and simulation capabilities to evaluate a very large set of alternatives quickly and with little effort on your part. All that is required, is for you to develop the various scenarios. Using such a DSS, you can narrow the alternatives to a few considered to be most relevant. Notice the DSS is not limited to quantitative data but also considers various qualitative factors. These are used as inputs from you to narrow the choices.
This process is more closely akin to the so-called administrative model of decision making called satisficing. Using this technique the decision maker considers only that subset of all possible outcomes which has relatively complete knowledge and makes a decision from among. these alternatives. Again the DSS provides the essential support to allow you to select the best from among the competing alternatives.
EXPECTATIONS FOR THE FUTURE
The several examples of decision support systems for project management discussed above are only a few of the many applications which already exist, In the future we can expect three major areas of improvement of DSS for the project manager.
More variety. We will see a wider variety of models for each of the different tasks associated with project management. This will allow an integration of the numerous modules into single systems appropriate for the individual user of DSS. You will be allowed to customize a system for your own needs from among the modules. This will permit much-needed flexibility and personalization of the decision support environment for both companies and individual users.
User-friendliness. There will be more “user-friendliness” in decision support models. This is strongly needed so less-experienced users can derive greater benefit. This user-friendliness will be accompanied by increasingly sophisticated users who understand the exceptional power and value of such systems. Managers will use these new models to derive maximum benefit in both everyday and extraordinary situations.
Lower cost. Progressively lower costs for standardized decision support software will make the benefits of DSS available to more and smaller organizations. This will create a self-fulfilling prophesy of more highly skilled managers even for the smallest and least costly projects. Organizations which previously hadn't put much formal effort into project management will find the DSS software as necessary as the microcomputer is today.
Gary A. Williams, a native Texan, received a BBA in finance and a MBE from the University of North Texas and his DBA in business administration from Texas Tech University. He has taught at several universities, including East Texas State University -Texar-kana, where be is currently an Associate Professor of Management Science and Information Systems. His academic interests center around management information systems and quantitative methods. Research completed has been in a wide range of areas including communication, real estate, and microcomputer applications in business.
William L. Boyd received his BA in economics from the University of South Florida, MBA from Texas Christian University, and Ph.D. from Texas A & M. He is currently Professor of Finance/Accounting at Western Carolina University. His background includes teaching, consulting and private business. He has served as Dean of a School of Business and President of a University. His primary research areas have been in the area of budgeting and various areas of decision support systems applications.
Further Reading
Aptman, Leonard H. 1986. “Project Management: Successful Use of PC Software,” Management Solutions. Vol. 31, (December) 20.
Dos Santos, Brian L. and Martin L. Bariff. 1988. “A Study of User Interface Aids for Model-Oriented Decision Support Systems,” Management Science. Vol. 34, No. 4 (April) 461.
Economies, Spyrosj and Mark Colen. 1985. “Microcomputer-Based Decision Support Systems Aid Managers in Evaluating Alternatives,” Industrial Engineering. Vol. 17 (September) 44.
Fick, Goran Pagels. 1982. “Decision Support Systems - A Development Strategy for Production Planning,” (Amsterdam, Vol. 6 April) 201-204.
Khwaja, Rashid Z. 1984. “The Criteria Approach to Major Project Management,” CA Magazine. Vol. 117, No. 9. (September) 68.
Leigh, W.E. and Doherty, M.E. 1986. Decision Support and Expert Systems. Southwest Publishing Co. Cincinnati.
Ronen, Boaz, and Dan Trietsch. 1988. “A Decision Support System for Purchasing Management of Large Projects,” Operations Research. Vol. 36, No. 6, (Nov-Dec) 882.
April 1990