Applications of Operations Research

Presented By:Chhatrapal Surve (160)

S.Y.BBAINDEX

SR DESCRIPTION PAGE NO.

1

NO.1. Introduction to Operations Research (OR) 3

2. Applications of management science 4

3. Airlines and Aviation Industry 5

4. Two cases of General Motors 6-8

5. Case study of Ford Motors 9-14

6. Ibm’s OR software 15

Introduction to Operations Research (OR)

Definition of operations research:

The analysis of problems in business and industry involving the construction of models and the application of linear programming, critical path analysis, and other quantitative techniques .

 The terms management science and decision science are sometimes used as more modern-sounding synonyms.

Operational Research (OR) is the use of advanced analytical techniques to improve decision making. It is sometimes known as Operations Research, Management Science or Industrial Engineering.

2

Employing techniques from other mathematical sciences, such as mathematical modeling, statistical analysis, and mathematical optimization, operations research arrives at optimal or near-optimal solutions to complex decision-making problems.Originating in military efforts before World War II, its techniques have grown to concern problems in a variety of industries.

The major subdisciplines in modern operational research, as identified by the journal Operations Research are:

1. Computing and information technologies2. Environment, energy, and natural resources3. Financial engineering4. Manufacturing, service sciences, and supply chain management5. Marketing Engineering6. Policy modeling and public sector work7. Revenue management8. Simulation9. Stochastic models10. Transportation.11. Aviation

Applications of management science

Applications of management science are abundant in industry as airlines, manufacturing companies, service organizations, military branches, and in government. The range of problems and issues to which management science has contributed insights and solutions is vast. It includes:

1. Scheduling airlines, including both planes and crew,2. Deciding the appropriate place to site new facilities such as a warehouse, factory

or fire station,3. Identifying possible future development paths for parts of the telecommunications

industry,4. Establishing the information needs and appropriate systems to supply them within

the health service, and5. Identifying and understanding the strategies adopted by companies for their

information systems

3

Some OR methods and techniques

1. Computer simulation : Allowing you to try out approaches and test ideas for improvement.

2. Optimisation : narrowing your choices to the very best when there are so many feasible options that comparing them one by one is difficult.

3. Probability and statistics: helping you measure risk, mine data to find valuable connections and insights in business analytics, test conclusions, and make reliable forecasts.

Airlines and Aviation Industry

We're all aware of the huge growth of low cost airlines in recent .

You may wonder how the low cost carriers manage to offer such surprisingly low fares. One thing, of course, is that by keeping their overheads down and cutting out frills like 'free' meals, they are able to run much leaner operations than the longer established airlines. But even so, the low cost carriers still can't afford to sell every seat at the headline prices.

There's another challenge that confronts every airline. Seats on any flight are perishable - once the plane has taken off, there is no possibility of selling any empty seats (the same applies to theatre tickets, hotel rooms, package holidays and seats on trains, for example). This being so, it pays an airline to fill a seat, even at a very low fare, rather than have it take off empty. But, as we've said, it simply isn't viable to sell every seat at a low price, so a mechanism has to be found for selling seats at different prices.

Fortunately for the airlines, there are different markets - some passengers want the lowest price and to get it are prepared to book well in advance, travel off peak and commit to using a specific flight; others want the flexibility to book at the last minute, and perhaps

4

to switch flights, and are prepared to pay more - often as much as ten times more - for the convenience. So the skill is to offer a range of fares - very low fares to catch the headlines, pretty low fares for those on a tight budget, through to much higher fares for those prepared to pay for flexibility and convenience.

The real skill comes in working out how many tickets to sell at each fare. Ideally, on any flight, the airline would first like to see how many people are willing to pay the highest price, sell as many tickets as possible to them, then sell as many as possible at the next highest price, and so on, filling up any remaining seats at the cheapest price. Unfortunately they can't do it that way because, as we have seen, the kind of people who are willing to pay the higher fares often want to book at the last minute. So the airline has to do it the other way round, selling the cheapest tickets first and holding back some places at the higher prices. The difficulty with that is knowing how many seats to hold back, given that the number of bookings for any flight varies from day to day.

This is where the airline needs O.R. inside. By observing the day to day variations in the number of high priced tickets sold, the number of seats that need to be reserved to give high fare passengers the best chance of being able to get on the flight, even if they book at the last minute, can be estimated. But if all of those seats were always sold until the last minute, then more often than not some of them would be left unsold. So the profile of bookings - how bookings come in over time - is monitored on a continuous basis, compared with the typical profile for the flight, and the number of seats held back is adjusted according to whether bookings are heavier or lighter than the typical profile. To do all this accurately and in such a way as to produce the best achievable results is difficult, and calls for some sophisticated analysis, which the O.R. inside provides.

Two cases of General Motors

5

The First Problem  

OnStar is GM's two-way vehicle communication system that provides a variety of services enhancing safety, security, entertainment, and productivity. In 1997, GM faced fundamental strategic decisions with respect to OnStar and the market for telematics (the provision of communications services to cars).

GM had to decide whether to view OnStar as a car feature or as a service, and choose between an evolutionary and a revolutionary strategy. Complicating the decision-making process was the almost complete uncertainty GM faced regarding technological approaches, major competitors, and what competitive and complementary technologies would emerge. The challenge required operations research expertise in developing

6

decision support systems, creating mathematical models of uncertainty, and strategic planning.

The O.R. Solution The GM/OnStar O.R. team developed a multi-method modeling approach to evaluate strategic alternatives. They used dynamic modeling to address some of the critical decisions GM faced in 1997, such as the company's choice between incremental and aggressive marketing strategies for OnStar. They also used an integrated simulation model for analyzing the new telematics industry, consisting of six sectors: customer acquisition, customer choice, alliances, customer service, financial dynamics, and dealer behavior.

The Value 

In 2001, OnStar had amassed 2 million subscribers – an 80% market share of the emerging telematics market – and had been valued at between $4 and $10 billion. The OnStar project set the stage for a broader GM initiative in service businesses that ultimately could yield billions of dollars in incremental earnings. Even more important than its financial benefits for GM, OnStar has saved lives that otherwise would have been lost in vehicle accidents.

The Second Problem  

At a time when General Motors was facing increasing competitive pressures, higher quality demands and a sluggish economy, the giant automaker was unable to effectively analyze the productivity and throughput of its manufacturing operations. Many of its factories during the late 1980s and early 1990s were missing production goals, working unscheduled overtime and experiencing high scrap costs. Although GM overall had excess production capacity at the time, production bottlenecks and other problems were causing the company to lose money even on high-demand products. In addition to creating problems for existing products, the lack of an effective throughput analysis capability was impeding the launch of new products, resulting in lost sales. In 1991, GM reported a $4.5 billion loss – a business record at that time.

The O.R. Solution 

7

Recognizing the broad scope of the problem, GM responded with a three-pronged O.R.-based effort focusing on production line data collection, throughput modeling and algorithms, and throughput improvement processes. In tackling the production data collection component, GM recognized the importance of limiting data inputs from the massive universe of production data available, to avoid bogging down the analytical processes. The goal was to develop models and algorithms with modest data requirements that still produce meaningful results. Repeated trial and validation efforts established the appropriate data inputs in the areas of workstation speed changes, scrap counts and classification of workstation stoppages (such as equipment failure and safety stops).

In the interest of making analysis be fast, accurate and easy to use, GM built analytic simulation models for simpler situations, as well as detailed discrete-event simulation models for more complex situations. They were built into an analysis tool that revealed, for each production line evaluated, the hidden bottlenecks that were impeding throughput.

GM’s “Throughput Improvement Process” (TIP) is a procedure whereby plant-floor personnel and management can use the O.R.-based analyses to study and remove bottlenecks. TIP steps include identifying the problem, analyzing it, generating action plans, implementing the solution, and subsequent evaluation.

The Value 

For GM, the quantitative benefits of the Throughput Improvement Process and its underlying O.R.-based analytics include a 26% increase in manufacturing productivity between 1997 and 2004 and ongoing annual combined savings and incremental revenue improvements of over $2 billion. Qualitative benefits include creating standard data definitions and manufacturing performance measures, facilitating communication and transferring lessons learned throughout GM’s global operations.

Case study of Ford Motors

8

The prototype vehicles that Ford Motor Company uses to verify new designs are a major annual investment. A team of engineering managers studying for master’s degrees in a Wayne State University program taught at Ford adapted a classroom set-covering example to begin development of the prototype optimization model (POM). Ford uses the POM and its related expert systems to budget, plan, and manage prototype test fleets and to maintain testing integrity, reducing annual prototype costs by more than $250 million. POM’s first use on the European Transit vehicle reduced costs by an estimated $12 million. The model dramatically shortened the planning process, established global procedures, and created a common structure for dialogue between budgeting and engineering.

In 1903, Henry Ford founded the Ford Motor Company with an initial investment of $100,000. In 1910, Ford opened a large factory in Highland Park, Michigan, to meet the growing demand for the Model T. Here, Henry Ford combined precision manufacturing, standardized and interchangeable parts, division of labor, and, in 1913, a continuous moving assembly line. The introduction of the moving assembly line revolutionized automobile production by drastically cutting assembly time per vehicle and thus lowering costs. Ford’s production of Model Ts made his company the largest automobile manufacturer in the world at that time. Today, Ford Motor Company is the world’s second largest manufacturer of cars andtrucks with 111 manufacturing plants in 38 countries and more than 350,000 employees worldwide.

9

From its inception, Ford Motor Company has carefully designed its vehicles to meet and, in many cases, exceed customer expectations. These designs are time consuming and extremely expensive to produce,partly because each design must be painstakingly verified. For some verification tests, Ford must construct prototype vehicles so that it can examine the interactions of systems in their operating environment. The cost of building a prototype routinely exceeds $250,000. Complex vehicle programs commonly require over 100 full-vehicle prototypes and sometimes require over 200 in the course of product development.

Product Development at Ford Motor CompanyDeveloping a new vehicle or even a moderately modified vehicle requires capital, time, and resources. Product development (Table 1) begins at the macro level with definitions of vehicle descriptions, target markets, and costs. At this level, planners identify target markets and vehicle use—such as family sedan, sport utility, pickup truck, or commercial van. The overall plan includes vehicles with no modification, new vehicles, and all levels of change in between. Planners develop this cycle plan in an iterative process, determining the capital and resources required to deliver all the strategicwants, prioritizing the resources needed, and setting final targets and budgets. Once complete, the cycle plan is not static. Planners can modify it to meet changing customer demands or market needs or to balance one program’s expansion by reducing another. In the second stage of the productdevelopment cycle, teams assigned to the vehicle use the initial product intent and marketing strategy to define in detail customerwants, vehicle configurations, and strategies to ensure customer satisfaction. They refine the product, developing its general direction, sense, and feeling. Designers start work on computer and physicalmodels and establish vehicle dimensions, including length, width, height, and wheelbase. Designing a line of vehicles is complicated by the need to meet customers’ demands for a wide variety of options. The number of combinations of features in a complex vehicle can number in the billions. To reduce the number of combinations, designers define the buildable combinations, that is, what combinations offeatures will be available to consumers.

Once management approves the overall plan, the vehicle team must deliver the product. The product-development cycle moves into the design, build, and test stages (stages 3 and 4 in Table 1). Designingthe vehicle requires testing each individual component, the systems containing these components, and the vehicle as a whole. Design engineers determine how to test components and systems to prove outtheir design parameters, and they prepare the design-verification plan (DVP). After this, the team decides what specific configurations— of the billions of potential combinations— to build as prototypes to test and evaluate.

10

Prototypes can be very costly to produce because most of the parts and the manufacturing and assembly processes are unique. The tools that would be used during mass production do not yet exist, and each vehicle is specially built to prove out designs, processes, and vehicle performance. Since the costs for prototype vehicles depend solely on the number built, the goal is to build as few vehicles as possible to perform the testing and evaluations required.

In the fourth stage of the product development cycle, the vehicle team does testing. The results may lead to design modifications. The team may make improvements and changes to provide desired features, to meet market demands, and to satisfy test parameters. This stage can include several design iterations, retesting, and rebuilding vehicles.

The fifth and last stage of the product development cycle is the final manufacturing prove out and full vehicle launch and introduction. During this stage, Ford increases vehicle production to full volumes and then maintains the production system at a steady-state rate. Success in this stage depends on flawless execution in the earlier stages of the product-development cycle.

The Engineering-Management Masters Program The engineering-management masters program (EMMP) at WSU is a three-year master’s program Wayne State University developed specifically for Ford Motor Company as a technical alternative to an MBA degree [Chelst, Falkenburg, and Nagle 1998]. Engineering and business faculty designed the program to teach students to integrate engineering and business issues in their decision making and management. A committee of Ford executives, WSU faculty, and class representatives oversees the program. Students maintain their regular jobs, taking classes that start in the late afternoon two days a week for 10 months a year. They come from diverse parts of the organization and have, on average, 10 years of experience in automotive engineering. Graduating students have a network of contacts throughout the company that further enhance their ability to get things done. At the end of the program, students form teams to work on large-scale projects suggested by executives and students. The oversight committee receives project proposals, and students must find

11

sponsors to fund their projects at $50,000 to $100,000, which insures the sponsor’s interest and the projects’ value to Ford. Between 1994 and 2000, 300 students participated in over 50 leadership projects, gaining academic knowledge, practical experience, and valuable contacts throughout the company. Four of us were directly involved with the EMMP. Kenneth Chelst is the operations research instructor in the program, while Jeffrey Lockledge teaches information systems.

John Sidelko is a member of the class of 1996 and was directly responsible for POM’s transfer from his team’s project to a working system within Ford. Dimitrios Mihailidis was a PhD candidate working with the student team to develop and implement their model in GAMS.

The Prototype Optimization Model (POM) to reduce the number of prototype vehicles Ford needed to verify the designs of its vehicles and to perform all the necessary tests. Historically, prototypes sat idle much of the time waiting for various tests, so increasing their usage would have a clear benefit. The barrier to sharing these idle prototypes among design groups and thereby increasing their usage lay in determining an optimal set of vehicles that could be shared and used to satisfy all of the testing needs. Ford’s planning with respect to prototypes can be broken into two phases: (1) strategicpredic tions of the fleet size during cycle-plan development, and (2) tactical planning of specific prototype requirements for vehicle design and testing.

These two phases are tied together by the assumptions concerning the vehicle line under development and the budget established for prototypes during strategic planning. With POM and the related expert-system modules, planners and designer base all their decisions on similar assumptions and have a mechanism for tracing the testing requirements from the engineers’ level through strategic planning. In discussing the role of POM, we start with Phase 2 since we built the model to help the firm to deal with the tactical planning during vehicle design and testing. The model guides the planner in developing a fleet of prototypes and assigning tests to specific vehicles. In Phase 1, the preliminary strategic-planning phase, the model is an integral part of an expert system called POM-Predictor. Its primary purpose is to estimate the size of the prototype fleet and the costs for different scenarios of the strategic product cycle. In this, POM provided the conceptual structure for developing the expert system that forecasts the budget.

12

The Math Model

The prototype planners must determine the specific characteristics of the fleet and schedule all tests to that fleet. In the language of mathematical programming, their objective is to minimize the number of vehicles built subject to the constraint that every test be completed on an appropriate vehicle by a specific deadline. The test planners’ primary input to the model is a description for each test of the—requisite vehicle characteristics,—the number of hours of testing, and—the deadline or due date for completion.A separate input is a list of every realistic vehicle configuration, called the buildable combinations matrix (BCM). This is necessary because some combinations of features are incompatible, for mechanical or esthetic reasons (for example, the largest engine may not be in the same vehicle with the smallest transmission). The BCM reflects reduced product complexity that is a byproduct of the early decision-making process in the cycle planning process. It describes the combinations of major components (drive trains, body styles, and so forth) that are intended to serve specific market segments.

For example, in the F-350 truck series (Figure 1), the regular cab is the only cab style available in many of the specialized configurations. This isn’t because a dump truck with a crew cab is impossible but because the market for one would be so small that it would never be profitable. Other combinations may not be available because the components would be mismatched.

For example, the Ranger pickup is offered with 2.5, 3.0, and 4.0 liter engines. The smallest engine is not powerful enough for four-wheel drive and is therefore available only in a two-wheel drive.

In POM, the BCM is represented as a spreadsheet with ones and zeroes indicating the available combinations (Table 2). Each row corresponds to a set of possible combinations with all compatible components marked. Thus, one row may indicate a large number of vehicles. Multiple rows may represent the same configuration. While this is redundant information, planners judged it acceptable, because the people who specify the configurations are used to thinking in overlapping product segments. Users found this

13

representation compact and much more natural than entering a row for every available combination, especially as potential combinations may number in the tens of thousands.Each vehicle-configuration type is defined as vector of features, with elements drawn from a restricted set of values for each feature. The vehicle features are typically grouped into a limited number of major categories.

For example, on the Transit vehicle program the categories were (1) vehicle variant (body style), (2) engine, (3) roof height, (4) transmission,(5) rear closure, (6) gross vehicle weight,(7) wheelbase, and (8) right or left hand drive.

The test requester specifies which combinations of features need testing and identifies which features have no bearing on the test. For example, in an altitudedrivability test (to determine whether a vehicle will perform acceptably at high altitude), the test requester will generally specify a vehicle with the highest gross vehicle weight (GVW), the smallest engine, and an automatictransmission to examine the worst-case vehicle configuration. For this test, the vehicle variant, roof height, rear closure, wheelbase, or which hand drive would be insignificant factors.

14

Ibm’s OR software

OR Software

IBM developed a system that optimizes inventory levels in its PC manufacturing supply chain, saving the company $750 million in 1998. IBM now sells the system as its Asset Management Tool. But implementing that product is no slam dunk, says Brenda Dietrich, a senior manager at the IBM Research Division's Optimization Center. "You need some expertise in OR. . . . I'd say someone with a master's degree in OR could use it very effectively." Supply-chain software from the major enterprise resource planning (ERP) vendors has some OR capabilities. "Some are integrated, so you don't even see it," says John Birge, president of the Institute for Operations Research and the Management Sciences in Linthicum, Md. "It might just tell you, 'Route your trucks this way,' and you have no idea it's optimizing in the background." But Birge, who is also dean of the McCormick School of Engineering and Applied Science at Northwestern University in Evanston, Ill., says standard ERP packages can make unrealistic assumptions, such as unlimited parts access. So users may want to use an OR software specialty company, he says. For example, Ilog Inc., Manugistics Group Inc. and Aspen Technology Inc. have supply-chain optimization packages, some tailored for specific industries. "They have some general-purpose tools that you have to be fairly expert in to use," Birge says. "But they . . . can develop more application-specific tools for you." "IT managers should know what OR can do for them," Birge says. "They should know what applications have been developed by the specialized software houses and then decide if they need those capabilities." If they do, he says, they should consider having an OR-trained consultant install them and train users.

15

16