Muckstadt Retirement Ceremonies Include Ph.D. Symposium
Festivities during Jack Muckstadt’s fall retirement weekend included two scholarly components - a Ph.D. Reunion Symposium with talks by some of Muckstadt’s many successful Ph.D. students and colleagues, and a poster session at which Ph.D. students presented their current research. Symposium speakers included former Muckstadt students Howard Singer, Ph.D. ’79; Andrew Loerch, Ph.D. ’90; Ganesh Janakiraman, Ph.D. ’03; and Retsef Levi, Ph.D. ‘05, as well as his colleagues Robin Roundy, Nathaniel Hupert and Peter Jackson. Presentations spanned a broad array of applications of operations research. Several built on Muckstadt’s supply chain management ideas.
Singer is Chief Strategic Technologist at Warner Music Group. Loerch is Associate Professor in the Department of Systems Engineering and Operations Research at George Mason University. Janakiraman is Associate Professor of Operations Management at the University of Texas at Dallas. Levi is J. Spencer Standish (1945) Professor of Management at the Sloan School of Management at MIT.
Roundy, for many years an ORIE colleague of Muckstadt, is now Professor of Mathematics at Brigham Young University. Hupert is Associate Professor of Public Health and Medicine at Weill Cornell Medical College (WCMC) and Muckstadt’s collaborator in building models for public health response logistics. Jackson, Professor in ORIE, has coauthored many papers with Muckstadt and is co-developer of their popular Design of Manufacturing Systems course.
Rock music and OR
OR talks rarely start by quoting Hunter S. Thompson, the 1970’s counter-culture Gonzo journalist. But then, it is unusual for an OR Ph.D. to become a major music company executive, which Howie Singer is. He described his career path from that of a typical OR analyst at Bell Laboratories, with work reading like a list of ORIE course titles, to a division of AT&T that was a pioneer in using compression and encryption technology to distribute music over the Internet. That brought him into an industry Thompson calls “a cruel and shallow money trench…where …good men die like dogs. There’s also a negative side.”
For Singer, the music business has shown its positive side, as he used his OR and business background to understand and influence technology trends in e-commerce, digital rights management, music consumption habits, social media, peer-to-peer technologies, and copyright law. Singer concluded by illustrating that the stunning evolution of music media — from LP records and tape cassettes to compact disks and digital downloads — has been accompanied by a sharp rise and fall in music revenues despite a dramatic rise in the amount of music that is distributed.
Past, present and future of military OR
Both Muckstadt and his student Loerch are retired colonels, Muckstadt from the Air Force and Loerch from the Army. In his talk, Loerch traced OR to its beginnings as a “scientific method of providing executive departments with a quantitative basis for decisions regarding the operations under their control,” in the words of a seminal 1950 OR text by Morse and Kimball that began as a classified military report.
Loerch pointed out that the early military use of OR had no established methodology or standard models, but “sought to understand underlying phenomenology through observation and data collection.” He described military OR contributions ranging across determining the optimal size for military ocean convoys in World War II; racial integration of the Army in the Korean War; control of disease in the Vietnam War (where analysts supported commanders in the field); the use of simulation analysis in the Cold War; and real-time OR support in the Persian Gulf. Loerch said insurgency and “the forgotten lessons of Vietnam” are among the current areas where OR is being applied to military problems, noting that the military now is facing both recurring analytical problems and “new and different problems that don’t fit a standard paradigm,” so that it is once again necessary to “collect data and seek understanding before anything else.”
Low cost regular and high-cost emergency suppliers
Former Muckstadt student Janakiraman discussed an inventory problem, but not before thanking Muckstadt for his kindness, understanding, openness, dedication to teaching, encouragement, and foresight. He confessed to nearly abandoning his Ph.D. when the math became overwhelming, but “warm mentoring” by Muckstadt and Emeritus Professor Uma Prabhu got him to stay.
Janakiraman discussed the inventory problem of a firm that has two possible sources for a particular good: one a regular supplier that offers a lower price but requiring a longer lead time and the other a “surge” supplier that is able to deliver more quickly but at a higher price. The challenge is to balance purchase cost against lead times in the face of the firm experiencing possible surges is demand.
In practice, balance is often attempted by procuring a fixed quantity from the regular supplier and “tailoring” the amount ordered from the surge supplier to best meet surges in demand. Janakiraman and colleagues investigated how effective this “Tailored Base-Surge (TBS)” policy is, when compared with the best that can be theoretically achieved. They discovered that when surges in demand are rare but large, relative to variability in the base demand, the best TBS policy is nearly optimal. They also showed that the performance of TBS policies increases as the regular supplier’s lead time increases.
Achieving savings in the distribution of print magazines and newspapers
Levi described a collaboration among faculty and students at Technion and MIT, and analysts at Yedioth Group, the largest media company in Israel.
With print magazines and newspapers under threat around the world, distribution in Israel is still significant. But it suffers from the classic OR newsvendor problem – how many copies of an issue to stock when running out of copies means losing sales but unsold copies are perishable, hence worthless? Yedioth distributes periodicals through a network of retailers via sales agents. They send out copies at the beginning of each periodical’s distribution cycle (e.g., once a week) and accept returns of unsold copies at the end of the cycle for a full refund, a costly process especially when retailers order more than enough copies.
Levi and his collaborators found value in providing an additional delivery during the week from pooled stock, with the size of the initial delivery based on a two-stage stochastic linear programming model and the size of the additional delivery based on information collected by the sales agent, who uses a decision support system that incorporates the model. When a pilot study showed a 1% increase in sales, 42% fewer stock outs and 38% fewer returns despite a reduction of 9% in the number of copies distributed, the approach was implemented by Yedioth on a large scale.
A colleague returns
Roundy, who worked closely with Muckstadt during the 24 years when they were colleagues at Cornell, discussed a supply chain approach to which he, Muckstadt and ORIE Professor Emeritus William Maxwell have contributed. The problem, known as EOQ, was formulated in the early days of OR, when analysts focused on determining how much of a product a retailer should order, and at what time intervals. Although EOQ is easy to solve for simple situations, it is notoriously difficult for multilevel (“multiechelon”) and multistage supply chain networks.
Roundy’s early work on this problem, which won the prestigious Lanchester Prize, showed that if reorders are permitted only at time intervals that are power-of-two multiples of some fixed base planning period (such as a week) it becomes easier to compute solutions while sacrificing only about 2% of optimality on average. Roundy described how Muckstadt and Maxwell used two powerful intuitive notions to compute solutions to the resulting model: “divide and conquer” and an analogy employing weighted balloons floating in an atmosphere whose density diminishes steeply with altitude. Although Roundy’s “magical powers of 2” approach was developed a quarter century ago, he described important multilevel and multistage supply chain questions about it that are yet to be resolved.
Sometimes the egg comes before the chicken
Jackson discussed one of Muckstadt’s long time inventory policy principles: “don’t duplicate safety stocks at every level in a distribution system.” Drawing an analogy with a model (the chicken) and inventory operating policy (the egg), he noted that policy is often derived from modeling results, but in their research Muckstadt’s policy principle preceded the development of the model that supports it.
They looked at the relationship between inventory investment and the rate at which products are sold from that inventory. Because of “the curse of variety,” assets can be deployed disproportionately in products that contribute relatively little to sales. One reason is that applying fill rate formulas too simplistically results in holding safety stock in a way that violates Muckstadt’s principle. Jackson presented a new way of articulating use of a multilevel inventory system to avoid that.
Jackson and Muckstadt have proposed changing the way the inventory is modeled by introducing a criterion for the fulfillment of demand (the fill rate) for each item at the regional distribution center rather than incorporating such a requirement at the central distribution center. Consequently the mathematical model selects the distribution system level that is appropriate for each part number and concentrates (pools) the safety stock there. Through an example Jackson demonstrated that this approach achieves a better alignment between inventory investment and product demand.
Preparing for disaster
Muckstadt’s repeated assertions in recent years that he was about to retire has not prevented him from embarking on investigations of a new research area. Over the past decade he has developed a research partnership devoted to health care logistics with Hupert, who is a Senior Medical Advisor on preparedness modeling to the Centers for Disease Control (CDC).
Hupert’s talk, “A (Probabilistic) Walk through the Sciences of Disaster Preparedness, or Lessons from Jack,” was laced with Muckstadt aphorisms (see below) and showed the wide range of systems and processes involved in their work together – from natural/biological phenomena to transport resources to information and communication systems and business processes. He noted that it can take a while for Muckstadt’s ideas to become accepted by the establishment. For example in 2009 Muckstadt urged the CDC to use the established wholesale network to distribute influenza prophylaxis. Only now has the CDC asked Muckstadt to model this approach for them.
Hupert described the work that he and Muckstadt have been doing with Master of Engineering students on the logistics and scheduling of urology operating rooms, with results that are winning wide acceptance at WCMC. The students showed that multiple process changes, most focused on information sharing, could save 60 minutes per operating room per day, worth millions of dollars.
The World According to Jack
- “You don’t know much until you know something!”
- “Things fail when you ignore the details.”
- “The way you are going to design your system is a reflection of how much you think you are going to know about the system.”
- “Keep your assets in the most flexible form for as long as is economically and operationally possible.”
- “Always ensure that the unit of measurement you are using is the same as that used by the person you are trying to convince."
- “One thing that is certain is unpredictability.”
- “If you have no information, you need more inventory.”
- “If your face is in the fridge and your fanny is in the fire, on average you are comfortable.”
- "You never learn anything when your mouth is open.”
- “You have to know what people do to teach about it.”