That same year, Dantzig mathematically stated the linear-programming problem and developed the simplex method to solve such problems. Needless to say, 1947 was a landmark year for the Air Force, for Project SCOOP, for George Dantzig and for the future of operations research.

As part of the Air Force's 60th anniversary celebration, Dr. Jacqueline Henningsen, director, Air Force Studies & Analyses, Assessments and Lessons Learned, recently recognized 60 years of superior mathematical analysis by Air Force operations research personnel. The highlight of the event: the unveiling of a commemorative display honoring Dantzig — the "Father of Linear Programming" and a legendary figure in the O.R. community — for his many contributions to the nation while serving as chief mathematician for the Air Force's Project SCOOP.

Prior to SCOOP, typical problems of the day were solved manually and, therefore, last-minute excursions requiring extensive recalculation were impossible, so one of the task force's enduring accomplishments was close coordination and financial support in the development of the world's first programmable computers.

In her welcoming remarks, Henningsen noted that visionary Air Force scientists and O.R. pioneers like Dantzig, Saul Gass and Alex Orden, along with other Project SCOOP members, routinely pushed the technical boundaries of their day to expand Air Force combat capability and secure maximum utility from limited budgets. She expressed deep appreciation for Dantzig's World War II service with the Army Air Forces Combat Analysis Branch of Statistical Control and later work with the RAND Corporation, much of which directly benefited the Air Force. Henningsen also professed profound respect and admiration for the quality of current Air Force analytic community work which she characterized as "invaluable and indispensable" in supporting numerous senior leadership decisions.

Although Project SCOOP is no longer an active program (in name), the operations analysis foundations laid by its original participants continue to pay significant dividends today.

Along with Gass and Orden, the list of special guests at the recognition event, held in Rosslyn, Va., included Dantzig's niece, Sharon Dantzig.

Orden, professor emeritus in the Graduate School of Business at the University of Chicago, and Gass, professor emeritus in the Robert H. Smith School of Business at the University of Maryland, spoke at the event. Following are excerpts from their remarks:

Saul Gass: Sixty years ago "something" occurred at the Pentagon that greatly influenced all of our professional and economic lives. That "something" was the hiring of George Dantzig as a mathematician for the Air Force. Here is how it happened.

In June of 1941, George, who had not quite finished his Ph.D. in mathematics from Berkeley, came to Washington to work for a newly formed organization, the Army Air Force Statistical Control. Over the next five years, George became an expert in Air Force reporting procedures and Air Force programming and planning methods using desk calculators. His boss was Tex Thornton, who later founded Litton Industries, and one of his co-workers was Robert McNamara, who became Secretary of Defense and president of the World Bank. Certainly, a high-powered group!

After the war, in 1946, George returned to Berkeley and received his Ph.D. in mathematics and began looking for a job. He really wanted to start an academic career. Berkeley did offer him a position at what could be termed a very low salary ($1,400/year). His colleagues from the Air Force countered with a better paying position (GS-12 $5,900/year) and challenged George with the problem of mechanizing the Air Force programming and planning process, that is to more rapidly compute a time-staged force deployment, training and logistical supply program. At that time, mechanization meant desk calculators and IBM accounting equipment. Fortunately for us, George took the Pentagon position.

Shortly afterward, in 1947, the Department of Defense and the Department of the Air Force were formed, and the powers that be decided that the George's group should be a major activity. George became mathematical advisor to the Air Force comptroller. In October 1948, the group was named Project SCOOP (Scientific Computation of Optimum Programs) with George as chief mathematician.

By mid-1947, George had already constructed a new mathematical model, what we now call a linear-programming (LP) model. It was the first time that someone had combined the mathematical relationships that defined production and consumption relationships of an economic enterprise, such as the Air Force's, with the aim of determining how to run the enterprise in an optimal fashion, that is, imposing on the model a mathematically defined objective function. That was new and novel. The only problem was there was no known way to solve such a model. Hence, to do that, George invented the simplex method. ...

SCOOP had a major impact on the development and application of electronic computers and their use by the Air Force.

In 1948, the Air Force gave the National Bureau of Standards $400,000 for the development of the SEAC (Standards Eastern Automatic Computer). The SEAC was the first fully operational stored program computer in the U.S; it had a 512-word internal memory and external magnetic tapes. Under the direction of Alex Orden, working with the NBS staff, the SEAC was the first electronic computer to solve linear programming problems, e.g., in 1951, using the first stored-program simplex-method, a 48 equation by 71 variable Air Force deployment model was solved in 18 hours. NBS mathematicians, in particular Alan Hoffman, also worked on SCOOP linear programming issues.For SCOOP, the Air Force installed the second UNIVAC I in the basement of the Pentagon in Room BD944. It was officially accepted by the Air Force on June 25, 1952, and retired in 1958. The UNIVAC had a 1,000-word high-speed mercury delay line memory, accessed at ten-thousandths of a second, and eight magnetic read/write tapes accessed at 1,000 words per second. It was used to solve many Air Force linear programming deployment and scheduling problems (and to play Christmas songs during the holidays). The UNIVAC LP code could solve problems with 250 equations and 500 variables. ...

The total influence of the Air Force's Project SCOOP is truly amazing. ... George Dantzig and Project SCOOP and the Air Force did not just contribute the linear programming model and the simplex method — what evolved from the Air Force's Project SCOOP changed the way business, industry, government and, especially, the military forces, plan, analyze and carry out their activities.

Alex Orden: The simplex method is a powerful tool. It not only solves very large LP problems, but is also the workhorse for much of integer programming (IP) and other forms of mathematical programming. It is undoubtedly used in at least half of all operations research implementations. It originated in Project SCOOP.

The simplex method is used to solve problems in planning, scheduling, routing, resource allocation and other areas in great detail. It is commonly used for solving problems that may be described as bottom-up, as opposed to top-down representations of situations. Philosophically, the opening up of a capability to solve bottom-up representations of situations expressed in mathematical form was in a way revolutionary (which, of course, could not have happened without that larger revolution — digital computers).

That is a bridge to another major aspect of Project SCOOP — mathematization of very large planning/decision making problems — directed in SCOOP at Air Force logistics. By analogy, I will liken the growth of use of LP (including IP) from the time of SCOOP to the present, to the growth of a tree — the roots, the trunk, the branches.

The primary aim of Project SCOOP was to facilitate and accelerate formation of integrated Air Force logistics planning — conversion of actual missions or preparatory consideration of alternatives, whether in war or in times of peace, into time staged requirements for personnel, training, acquisition and maintenance of equipment and supplies, movements of personnel and supplies, and more. The then emerging digital computer triggered a vision of what could be accomplished.

It was a conceptual breakthrough in SCOOP that integrated planning was treated by mathematical modeling. Its substantive and computational work on Air Force planning focused on a form of modeling called "triangular models," straightforward simple systems of linear equations where, after evaluating an initial variable, all others are evaluated sequentially using values previously determined for their predecessors. That computational simplicity made it possible to solve very large, chain-of-command models of Air Force plans that encompassed personnel and training requirements, equipment and supplies, as well as acquisition/storage/utilization/maintenance, etc; models on about 25,000 activities in 36 time stages, so about 900,000 variables.

In the invention and substantial progress on the simplex method, and in its strong thrust into large-scale, real-world modeling, SCOOP provided the roots for mathematical programming.

Beginning in about 1952, when the tree so to speak sprouted from its roots, the major oil companies became impressed by the prospect that LP would be highly valuable for production planning and scheduling, transportation management, gasoline blending, etc. The liquid flow aspects of oil refineries and pipelines made it clear that LP models would become pervasive. ... The oil industry was the trunk of the LP tree; others were its branches. ...

From the mid-1980s to the early 2000s, the role of the airline industry in the field was like that previously of the oil industry. Models for assignment of planes to flights, crew scheduling, yield management (timing of reduced fares) and others — and advances in integer algorithms and software — often paved the way for models in other industries. The trunk rose higher in the airline companies.

Recently, say since about 2000, due considerably to rapidly rising industry globalization, mathematical modeling has become prominent in supply chain planning. Spreading into diverse industries, the tree from high on its trunk has grown a canopy. Large manufacturing firms in the automobile industry, in computers and others can model systems that coordinate locations/schedules/inventories for supply and processing of raw materials, subassemblies, final assembly, transportation and distribution involving hundreds of locations and hundreds of thousands of items.

The 1950 vision in Project SCOOP of bringing the simplex method and its elaborations to bear on planning areas as large as that of the determinate triangular models for comprehensive planning of Air Force missions has been fulfilled.



Sources: Lt. Col. Dave McCormick (Phalanx), Saul Gass, Alex Orden

Table of Contents

OR/MS Today Home Page


OR/MS Today copyright © 2007 by the Institute for Operations Research and the Management Sciences. All rights reserved.


Lionheart Publishing, Inc.
506 Roswell Rd., 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 2007 by Lionheart Publishing, Inc. All rights reserved.