April 1999
| This issue of OR/MS Today gives ample evidence that OR is alive and well throughout the world. But is it the same OR, or are there differences in different parts of the world? If we traveled and observed OR implementations, would we say, "This is OR, but not as we know it!"? Within the last year, these global differences were debated at the EURO Conference in Brussels and at the CLAIO Congress in Buenos Aires. I have brought together three separate contributions to the debate in the brief articles that follow. Andres Weintraub, president of IFORS, professor in the Department of Industrial Engineering at the University of Chile, and a member of the of 1998 Edelman Prize-winning team, was on the panel in Brussels. He gives some examples from his experience in Chile of the similarities and differences in implementing OR. Jonathan Rosenhead, professor of OR at the London School of Economics, responded to the discussion in Brussels with a thought-provoking piece that turns the question around — Is OR local? And finally, I have included a short extract from Jaime Barcelo's contribution to the panel in Buenos Aires in which he introduces the concept of "culture broth." Barcelo is a professor in Barcelona, and was the organizer of the EURO/INFORMS conference there in 1997. — Graham K. Rand |
Is OR Global?
By Andres Weintraub
From a basic methodological viewpoint the obvious conclusion is that OR is universal. The requirements, characteristics of basic research and algorithms used to solve problems are applicable everywhere.
Differences appear when real problems are tackled, and an OR/MS approach and its different tools are used. I will concentrate on OR/MS applications. What is common, what is different?
First, some of the problems faced in most developing countries are different. Typical differences exist in education, health and urban transportation. In these cases, the problems themselves are different. Second, there are important differences, in many fields, in the conditions under which OR/MS is applied. For example:
- Political and entrepreneurial instability. To some degree, varying by countries and along time, this instability plays a role, and often implies a decreased interest in long-term, more sophisticated solution approaches, such as those provided by OR.
- Weaknesses in general management. At times the basic organizational order and control needed before the use of quantitative tools requires stronger development. Often the firm's problems are more serious than the 5 percent to 15 percent improvement that OR can typically provide.
- Lack of qualified personnel in enterprises. Usually enterprises lack OR-educated people who can adapt and implement OR tools. It has been our experience that academics usually need not only to introduce OR concepts and techniques, but also to handle the implementation.
- Insufficient information or data needed to feed the models properly.
- Weaknesses in the use of hard frontier technology, such as data communication, that limit the use of OR.
- In some industries, the level of competition is weak, with little incentive to become highly efficient.
Enterprises that face a high level of competition tend to be motivated to use OR and information technologies, creating a culture of high-level management and use of technology. These firms typically compete in global markets. In Chile, the mining and forestry industries are good examples. On the other hand, firms that face a low level of competition are less likely to use OR. Some manufacturing sectors, the building industry, and public services such as health and education are typical examples in Chile. I want to illustrate these points with specific examples we have encountered.
Forest transportation system
Around 1989 I was a member of a team that developed a system to schedule daily transportation for trucks in the forest industry. Transportation accounts for about 40 percent of operational costs. Forest firms export most of their production, and thus compete in tough, global markets.
The basic problem: how to best transport logs of different lengths and diameters from origins in the forest to destinations such as sawmills, pulp plant and port. The objective is to move the timber to satisfy daily demands at the different destinations at minimum cost. It is also important to coordinate well with downstream operations, such as the unloading on a conveyor belt.
Daily decisions involve how to satisfy demand for each product from the different origins, the fleet of trucks and loaders needed and, fundamentally, to create a schedule for each truck and loader. The traditional program was carried out manually, with severe limitations — long queues, often unsatisfied demands, high transportation costs.
The new system was based on an organizational change, a central transportation office to schedule and control all trips, and a simulation model with heuristic rules to do the daily scheduling. The system was extremely successful; it was used by most Chilean forest firms and reduced costs between 15 percent to 35 percent (see reference 1). The tools used in the system are of global nature, and are also in use in Brazil and South Africa, where they enabled a forest firm, Mondi, to win the South Africa Logistics Award of 1996.
The natural next step in the development of the system is to schedule trucks online, using global positioning systems (GPS) to locate trucks at each moment, and possibly satellite communications to carry data and instructions between trucks and the central transportation office. At this stage, however, the weak technological development in Chile hampers further progress.
Routing of emergency vehicles
Santiago's Electric Utility has a fleet of vehicles to handle emergencies, from high-priority fallen cables to individual home breakdowns. At present, failures are communicated by phone to the central dispatch office, which manually schedules vehicles to cover failures and communicates by radio to the drivers. We developed a computerized scheduling system, based on heuristic approaches for scheduling and an exponential smoothing approach, to try to predict future failures. Tests carried out for two weeks showed that the computerized system could improve response time by 16 percent on normal days and by 45 percent on rainy days, when more failures occur (see reference 2). The use of models to schedule vehicles has led to major savings worldwide. In Chile, implementation of these systems has often been hampered by a reduced need of a high level of service, and by administrative weaknesses in control and dispatching. Further, as in the electric utility case, a real-time dispatching system requires good links to a city GIS system and data instructions communications system, which are not easily achieved.
Urban transportation
This is an area where OR models have been amply used to determine traffic equilibrium and passenger behavior, and thus evaluate the consequences of different proposals, from changing the direction of traffic in a street to adding a subway line. The models include public transportation as a user alternative. In developing countries such as Chile, buses constitute a much larger proportion of trips than in developed countries. This leads to different models to determine traffic and passenger equilibrium.
In Chile, the Ministry of Transportation defines bus routes and fares. Within these rules, buses are free to serve any route they choose. Because the average bus owner possesses less than two buses, the system could be approximated by considering each bus driver making his own decisions. So, to determine the equilibrium in these cases, we also needed to consider buses as additional players, leading to a different formulation (see reference 3).
Forest management
During the last nine years a set of systems has been developed to support decisions in the Chilean forest industry, including the transportation system described above. Other systems deal with:
- strategic decisions, to determine long-range sustainable timber production consistent with existing or planned industrial installations;
- tactical decisions, dealing with the sequencing of yearly and seasonal harvesting; and
- operational decisions dealing with short-term harvesting decisions to satisfy precise orders and to define the location of harvesting machinery (see reference 4).
- tactical decisions, dealing with the sequencing of yearly and seasonal harvesting; and
The systems are based on well-known OR techniques such as LP, column generation, MIP and heuristics. Friendly interfaces were developed, including GIS-based data and graphic interactive interfaces.
The implementations were similar to what would be carried out in any country. These systems won the 1998 Edelman Prize competition. This is one more indication that OR work in the developing countries can compete on the same level as OR being carried out in the developed world. These models are currently being applied to evaluate the implementation of environmental protection policies that have been in place in developed countries for many years. So, while operational management is very similar for firms in developed and developing countries, the approach to environmental measures marks a difference in the problems to be handled. In developing countries the implementation of environmental protection measures is being implemented now and is likely to be different.
Finally, while there are in some instances important differences in implementing OR in developing countries — both in the type of problems handled and the possibilities of using OR — the use of OR is perfectly possible and at a very high-quality level. Moreover, there are enormous challenges and potential for the use of OR that have not been exploited in these countries.
References
- Weintraub A., Epstein R., Morales R., Serón J. and Traverso P., 1996, "A Truck Scheduling System Improves Efficiency in the Forest Industries" Interfaces, Vol. 26, No. 4.
- Weintraub, A, Aboud, J, Fernandez C and Laporte G., "An emergency vehicle dispatching system for an electric utility in Chile," to appear in The Journal of the Operational Research Society.
- Fernandez E., Marcotte P., Mondschein S., Vera J and Weintraub A.,1993, "Solution Approaches to the Bus Operator Problem". Transportation Research B. Vol. 27B, No.1.
- Weintraub, A., Epstein R., Morales, R., Seron J., 1998, "Use of OR Systems in the Chilean Forest Industries", 1998, to appear in Interfaces.
Is OR Local?
By Jonathan Rosenhead
Since IFORS is dedicated to the global unity of OR, it seems churlish to assert in these pages that OR practice is in certain essential characteristics different in different societies. However, better an honest clarity about the heterogeneity of our practice, than a pretence of homogeneity.
First, I take it as agreed that there are some defining characteristics shared by all activities that can reasonably be labeled operational research. Unfortunately it is by no means easy to be precise about these characteristics — witness the many attempted definitions of the subject, none of which has achieved wide currency. However let us assume that somewhere between "model-based decision aid" and "quantitative common sense" there is an uncontested minimum requirement for membership of the set.
In what senses, then, can OR be said to be non-universal in method? Let me dispose of two statements that I am not making. I am not asserting that equations which have one solution in country A will come out differently in country B, or that a feasible solution of a mathematical formulation will prove infeasible if repeated on another continent. This is patently untrue; quite evidently, correct mathematics or well-posed computations are indeed the same the whole world over.
The second statement that I am not making is that the particular systems which operational researchers may study in different countries can lead to distinctively different models. The reason why I am not saying this is because it patently is true, and so not worth saying. Iceland, say, has a unique mix of energy sources, so that models supporting energy planning there are likely to be "one-offs." Even where the natural resources and basic technology are common to many countries (for example, in forestry), there can be sharp disparities in the way the processes are carried out or interact. For example, locally imposed environmental or legal constraints, or different influences in the labor market, can result in the systems being studied, and hence any models developed to support decision-making, having specific national flavors. Ça va sans dire.
Having I hope eliminated two non-arguments, I will move to an argument worth having. This concerns neither the invariance of arithmetic nor the diversity of models. It concerns OR practice — the way in which the approach, tools and methods of OR are deployed with the well-founded intention of assisting people with the burden of decisions. My proposition is that the way operational researchers deploy their skills, techniques and methods is different in different societies, or needs to be to secure success. Within the confines of this note, I can support this point only in very general terms.
We may define any instance of OR practice as an analytic intervention in an ongoing system (composed of human and non-human elements) with a view to influencing the way in which decisions influencing the future of that system are taken. There are thus normally at least three categories of human actors associated with OR practice: 1. the operational researchers; 2. those who between them determine the commitment of resources (the "decision-makers"); and 3. the other human elements in or impacted on by the system (the workers, beneficiaries, victims). All of these have abilities, aspirations, attitudes (and a number of other attributes starting with other letters of the alphabet) which are the product of their socialization within a particular society, and are constrained by the social structures of that society.
To put it more simply (and leaving a lot out), OR workers in country A encounter managers who have expectations about the decision-making process and the possible roles of analysis within it which will differ (sharply or subtly) from those encountered in country B. And these managers need to take into account their beneficiaries/victims in ways which have country A-specific elements. And those OR workers we started with are themselves a variable ingredient in the cauldron of practice — through their career expectations, attitudes of deference to power, concerns for equity, etc. The brew is, of course, far headier than this. OR workers are themselves embedded in an OR hierarchy; managers are parts of systems of managers; and the dynamic within both these structures affects the form of information/proposal/advice/recommendation that it is thought appropriate to give and to receive. In all these ways and more, the specific culture of a society affects what will be done in OR practice.
Let me give one small example of how these influences can play out, and then a big one. The small one is the differential diffusion of problem structuring methods (PSMs). These methods, predominately developed in Britain (why?), have obtained a wide usage in their country of origin, and a certain currency in some of the more northerly states of Europe. Penetration in Mediterranean countries or the United States is negligible.
Simple explanations ("not invented here") do not fit the observed facts. It is at least plausible that it is the social requirements of these methods that have affected take-up. PSMs require that a group of managers interact, facilitated by one or more operational researchers. The willingness of managers, or operational researchers, to expose themselves in this way is doubtless affected by quite a range of unarticulated but powerful assumptions about roles and identities, as well as by internal political considerations.
Let us move to a larger arena and larger cultural differences. The developed and less developed worlds diverge in many aspects of economic and social structures. (Less developed countries are by no means identical with each other either, but I will just have to lay this complication on the table and pass on.) Rather than elaborate on these general differences, I will focus on some contrasts in the make-up of the OR community within a (representative) developed and less developed country. For definiteness, let us assume that the representative developed country's OR community has lively academics participating actively in international fora (conference, journals); in-house or consultancy-based practitioners serving industry and commerce; and similar arrangements for government and other public sector agencies.
Contrast the situation for our hypothetical less developed country. Its academic system is geared (in prestige and promotion) to the international research agenda that is dominated by the metropolitan countries. Ph.D.s from France, the United States, Britain, etc. count more than local ones. Library subscriptions to journals published in other third world countries are given lower priority. Scholarships for travel or research visits abroad follow the same pattern. In all these ways the local research agenda is established outside the country, and can be expected to have only the barest relevance to local needs and conditions.
Local firms, even of considerable size, tend to be dynastic in nature, with a charismatic founder plus descendants who believe their genes guarantee good decision-making. (Comparably sized and controlled firms predominated in developed countries not so long ago; the evidence is that such conditions do not favor openness to technically based advice.) Multinationals occupy a strong position in our Ledevia; their OR of any significance is done back home at headquarters.
This leaves the public sector, which may well support a small number of practitioners attempting to assist the development process. But in the absence of support from a lively OR culture, they will find their task a lonely, isolated one. The style of decision-making among the relatively small elite is likely to be interpersonal rather than formal, presenting additional obstacles to OR.
To these OR-specific conditions we must add the manifold and manifest ways in which social structures and relationships take on a different character under conditions of development as opposed to under-development. Can we doubt, then, that OR is different from physics? Physics operates in the laboratory under controlled conditions of temperature and pressure. For OR the equivalent of the laboratory is the relevant segment of society under study, plus the "studiers" and deciders. The conditions under which this takes place cannot be controlled.
Culture Broth
By Jaime Barcelo
A first approach to an attempt to explain OR global differences can be found by analyzing the relationship between technology and society. In his thought-provoking book, "The Information Age: Economy, Society and Culture," (Vol. I: The Rise of the Network Society, Blackwell Publishers, [1996], Manuel Castells argues that "while society does not determine technology, it can suffocate its development, above all through the state ... what we must keep in mind for understanding the relation that exists between technology and society is that the role of the state in either holding back, unleashing or directing technological innovation is a decisive factor in the overall process, as it expresses and organises the social forces that dominate in a given space and time ..."
This is a statement with which I agree. Consequently, if from this perspective we analyze the historical development of the working methodology and the set of derived techniques that today we call OR, as well as the crucial milestones that have marked it out, it is clear that the development of OR has always been linked to processes of change and technological innovation. Hence, we can conclude that the culture broth that makes innovation possible is largely the nutrient that also makes possible the development of OR as a methodology and set of associated techniques that have sense, or rather come to make full sense, in the service of decision-making processes that involve the development and use of technology.
|
IFORS' ORbit Offers panORama View of OR By Celso C. Ribeiro IFORS invites you to visit its web site (http://www.ifors.org) called ORbit. The site consists of two parts. One is dedicated to static information related to IFORS such as the list of national OR societies, regional groupings, IFORS statutes and by-laws, list of officers, IFORS conferences and publications, archives, etc. If you plan a conference or workshop and seek international exposure, you might be interested in visiting the site to obtain information on ways IFORS can help you to promote your activities, including publication of refereed papers in the International Transactions in Operations Research (ITOR). The second part of the site consists of IFORS' online magazine called panORama. The new project is designed to provide timely information on what is happening on the international OR scene and provide links to OR resources worldwide. The current list of directories includes: Call for Papers, Conferences, Jobs, News, Prizes and Awards, and Recreational OR. As expected, the "Conferences" and "Jobs" directories are already quite large. Other directories (Developing Countries, Education, Organizations, Software and Tutorials) are being constructed. The site was designed to complement the information and services provided by the web sites of the national OR societies (e.g., INFORMS Online and the U.K's Live OR), promote OR internationally, and facilitate communication and cooperation between national OR societies and related societies. We encourage you to visit the site and have a global glance at OR. Celso C. Ribeiro is the IFORS vice president for ALIO. |
OR/MS Today copyright © 1999 by the Institute for Operations Research and the Management Sciences. All rights reserved.
Lionheart Publishing, Inc.
506 Roswell Street, Suite 220, Marietta, GA 30060, USA
Phone: 770-431-0867 | Fax: 770-432-6969
E-mail: lpi@lionhrtpub.com
URL: http://www.lionhrtpub.com
Web Site © Copyright 1999 by Lionheart Publishing, Inc. All rights reserved.