Operations Research and Management Science at Shell Oil Company in the 1950s-1980’s
Shell Oil Company was established as the largely independent American subsidiary of the Royal Dutch Shell Company in 1912 (30% of Shell Oil was owned by third parties and publicly traded on the New York Stock Exchange). Among the benefits of this pseudo-independence was the ability Shell Oil to direct significant R&D efforts. Pre-OR/MS research efforts at Shell primarily centered on novel petrochemical studies. Shell started its first research laboratory in 1914 in Amsterdam, Netherlands. The Emeryville Research Center, in California, of the Shell Development Company opened in 1928 and was recognized by the American Chemical Society as one of the preeminent corporate research groups in the Western United States. With the advent of operations research following World War II, big oil, like most industries, saw an opportunity for competitive advantage.
There were and continue to be numerous operations research and management science problems pertinent to the oil and gas industry. R. C. McCurdy, Shell Oil’s President in the 1960s, wrote an article ( McCurdy 1968) on OR applications for the petrochemical space. At the mid-century point, one such application was the optimization of scheduling oil tankers. When Michael Rothkopf first joined Shell’s Applied Math Department as a fresh MIT PhD in 1964, he was given the freedom to look for operations research problems that would help Shell’s bottom line. Given that his dissertation covered the scheduling of multiple processes, he felt this would be an ideal starting point. Unfortunately, Rothkopf found that most scheduling problems in the industry hardly required significant analysis. He eventually identified an important and at the time, difficult problem: how to coordinate large tankers carrying oil products from Gulf coast refineries in Texas and Louisiana to East coast markets. While his priorities were directed elsewhere, this work was picked up years later as the ability to solve large optimization problems improved (Rothkopf 2001). Naturally, his solutions were applicable beyond the energy space as international maritime trade skyrocketed with the arrival of container shipping, namely between the United States, Europe, and Japan.
OR work at Shell predates the research during McCurdy’s tenure. In the late 1950s, Shell researchers J.C. Ornea and Paul Tillson noticed the growth of operations research use cases in their industry and recognized a “lack of consistency” between petroleum and chemical OR groups. They wrote an article in Operations Research defining operational, technological, and investment objectives for OR problems within the petroleum industry and determining which popular methods would best fit those parameters (Ornea & Stillson 1960). One such use case involved fuel-efficient air cargo transportation, which was featured within a 1957 edition of Shell Aviation News, a publication of Shell Oil (Brewer 1957).
Management science techniques were also developed at Shell Oil. In the journal Management Science, Rothkopf published “A Model of Rational Competitive Bidding”, presenting a mathematical model of the bidding process, explicitly dealing with the uncertainty of bidders and the uncertainty of auctioned items’ values. This article challenged the existing game theory and decision theory bidding models of the era, taking into consideration the optimality of all bidders’ decisions (Rothkopf 1969). Prior to this, Rothkopf authored three Shell reports on the subject between 1965 and 1969 and seven additional reports and articles in the decade following, both at Shell and his next professional home, the Palo Alto Research Center (PARC) of the Xerox Corporation. This work attracted the attention of senior chemical managers and was applied in government procurement bids. In 1979, Rothkopf co-authored a bibliography of over 500 publications on the subject, including those published by Shell, with the University of Delaware’s Robert Stark in Operations Research (Stark & Rothkopf 1979). Bidding plays a major role in the oil and gas industry as companies bid for resources, drilling rights, and chemical construction. The models in Rothkopf’s articles, however, are not limited to petrochemical businesses and offer application across multiple fields. This displays the commitment of Shell’s Emeryville facility toward expanding the discipline, not solely the profits of the company.
At the Houston Refinery’s Technological Department and the Mathematics and Computing Group of the Houston Research Laboratory, one problem of interest was developing a method for solving nonlinear programming problems, using the then available IBM 650 and IBM 740 systems. Nonlinear models were encountered in such situations as the blending of gasoline, where the octane rating may be a nonlinear function of the decision variables. The resulting Method of Approximate Programming (MAP) utilized linear programming algorithms and was first applied in 1959, before being improved upon and implemented on newer IBM hardware in the early 1960s. A team of fourteen individuals across both organizations were attributed as having made significant contributions to the effort (Griffith & Stewart 1961). Linear programming work at Shell continued into the 1970s as it was applied to company-wide distribution problems (Zierer et al. 1976).
Company support of industry-wide collaboration and publication is displayed in Shell researchers’ participation in the OR/MS community at large. This participation expanded beyond the Emeryville Group as Robert Barton and Herbert Kanner of Houston’s Technical Services Division and Shell Research, respectively, Frank Cutting of Shell New York, and David Probst in Caracas, Venezuela, were all listed as Operations Research Society of American (“ORSA”) Members in 1956 (ORSA 1956). Barton, who did early development work on Shell’s Burroughs/Datatron 205 computer, left Shell for Burroughs before eventually joining the University of Utah’s Electrical Engineering Faculty in 1965. He went on to become a charter recipient of the IEEE Computer Pioneer Award in 1981 for developing language directed architecture. In professional societies, Shell personnel particularly played an active role in the Northern California operations research community. Emeryville’s Paul Stillson chaired the 20th National Meeting of ORSA, jointly held with The Institute for Management Science (“TIMS”) in San Francisco (ORSA 1961). Five years prior, he had served as a co-chair of the Operations Research section in the Chemical and Petroleum Industries committee for the 10th National Meeting (ORSA 1957). In 1956, Stillson simultaneously served as the Secretary of the Southern California Chapter of TIMS and Secretary Treasurer of the Western Section of ORSA. This specific instance of cross pollination between the two organization was used by John Lathrop in an argument for merging ORSA and TIMS into a single organization (Lathrop 1956), four decades before the eventual merger that established INFORMS.
Involvement with non-Shell researchers and the wider OR public expanded beyond the United States. In tandem with the 2nd International Conference on Operations Research in 1962, Shell Oil Company hosted attendees at their refinery in Berre, France and at a nearby chateau. John Lathrop, the ORSA Representative to the International Federation of Operational Research Societies (IFORS), lauded the “opulent Conference banquet” for adding an opportunity for industry networking and a chance for delegates to “become acquainted with each other, as well as the magnificent cuisine of France (Lathrop 1961).”
In 1972, Shell Emeryville closed, and research efforts were relocated to the Houston Research Laboratory in Texas. On a grander scale, Royal Dutch Shell fully acquired the independent shares of its subsidiary in 1984. Operations research efforts continued in Houston and elsewhere, relatively unimpeded by these transitions. In the 1980s, for instance, Shell Oil was among a handful of organizations, including the US Department of Energy, the U.S-Saudi Joint Commission on Economic Cooperation, and Stanford University to conduct simulations on a novel computerized model on the noncooperative international energy market (Salant 1981). Work headed into the 21st century included novel computational models, while continuing a strong focus on optimized fuel storage and logistics planning.
Compiled by: Reed E Devany
Edited by: Linus Schrage
Links and References
Links and References
Brewer S.H. (1957) Air Cargo – Next Ten Years. Shell Aviation News 234, 6-9.
Griffith R. E. & R. A. Stewart (1961) A Nonlinear Programming Technique for the Optimization of Continuous Processing Systems. Management Science, 7(4): 379-392.
Priest, T. (1981) Shell to Houston. The Houston Review, 3(1): 10-11.
Lathrop, J. B. (1956) Letter to the Editor: A Proposal for Merging ORSA and TIMS. Operations Research, 5(1): 123-125.
Lathrop, J. B. (1961) Back Matter: Announcements – The Second International Conference on Operations Research. Operations Research, 9(1): 142 -144.
McCurdy, R. C. (1968) Application of Operations Research to Chemical Technology. Industrial and Engineering Chemistry, 40(2).
Ornea J. C. & P. Stillson (1960) The Optimum Solution in Operations Research. Operations Research, 8(5): 616-629.
ORSA (1956) Members of the Society. Operations Research, 4(6) : 11-51.
ORSA (1957) Tenth National Meeting of the Society. Operations Research, 5(1): 132-153.
ORSA (1961) Back Matter: Announcements. Operations Research, 9(4): 600-601.
Rothkopf, M. H. (1969) A Model of Rational Competitive Bidding. Management Science, 15(7) Theory Series: 362-373.
Rothkopf, M. H. (2001) Tales from a Nonstandard Career in Operations Research. INFOR: Information Systems and Operational Research, 39(4): 367-393.
Salant S. W. (1981) Imperfect Competition in the International Energy Market: A Computerized Nash-Cournot Model. Operations Research, 30(2): 252-280.
Stark R. M. & M. H. Rothkopf (1979) Competitive Bidding: A Comprehensive Bibliography. Operations Research, 27(2): 364-390.
Zierer, T.K., Mitchell, W. A., & T. R. White (1976) Practical Applications of Linear Programming to Shell’s Distribution Problems. Interfaces, 6(4): 13-26.