A recent article in The Economist magazine talked about the increasing use of computer models and simulations to make political forecasts. [“Game theory in practice,” 3 September 2011]. The article begins:

“Bruce Bueno de Mesquita, an academic at New York University, has made some impressively accurate political forecasts. In May 2010 he predicted that Egypt’s president, Hosni Mubarak, would fall from power within a year. Nine months later Mr Mubarak fled Cairo amid massive street protests. In February 2008 Mr Bueno de Mesquita predicted that Pakistan’s president, Pervez Musharraf, would leave office by the end of summer. He was gone before September. Five years before the death of Iran’s Ayatollah Khomeini in 1989, Mr Bueno de Mesquita correctly named his successor, and, since then, has made hundreds of prescient forecasts as a consultant both to foreign governments and to America’s State Department, Pentagon and intelligence agencies. What is the secret of his success? ‘I don’t have insights—the game does,’ he says.”

Political upheaval is just one of the concerns facing supply chain professionals. Just look at how the civil war in Libya disrupted oil supplies coming from that country. While I’m a fan of computer models and simulations, the danger is that those who use them will accept the outcomes as facts rather than possibilities. No one, not even the smartest computers, can predict the future with extreme accuracy. The reason, of course, is that the future is not a straight line extrapolation from the past. There are breaks and discontinuities that simply cannot be predicted based on past experience. The actions of a single individual can change the course of history. Bueno de Mesquita’s computer model “uses a branch of mathematics called game theory, which is often used by economists, to work out how events will unfold as people and organisations act in what they perceive to be their best interests.” Political scientists and military analysts have also studied game theory for decades. The article explains that “numerical values are placed on the goals, motivations and influence of ‘players’—negotiators, business leaders, political parties and organisations of all stripes, and, in some cases, their officials and supporters. The computer model then considers the options open to the various players, determines their likely course of action, evaluates their ability to influence others and hence predicts the course of events.”

Professor Bueno de Mesquita, like many other of today’s academics, has managed to turn his academic expertise into a business. His consulting firm is called Mesquita & Roundell. The article states that it “is just one of several consulting outfits that run such computer simulations for law firms, companies and governments. Most decision-making advice is political, in the broadest sense of the word—how best to outfox a trial prosecutor, sway a jury, win support from shareholders or woo alienated voters by shuffling a political coalition and making legislative concessions.” Simulations and models, however, are only as good as the data that feeds them. Obtaining that information is normally the biggest challenge faced by those who run such simulations and models. The article notes that “gathering and handling the necessary data can require expensive expertise or training. Decide, the Dutch consultancy, usually charges €20,000-70,000 ($28,000-100,000) to solve a problem using its software, called DCSim, because it must first conduct lengthy interviews with experts.” While the cost may deter some companies from using models and simulations, Stephen Black, a modeller with PA Consulting, told The Economist that the “use of modelling makes business clients more inclined to adopt longer-term strategies.” When it comes to supply chain risk management, long-term strategies are a good thing.

Supply chain risk management expert Daniel Stengl notes that Tuncer I. Ören, a researcher in the Computer Science Department at the University of Ottawa, and Bernard P. Zeigler, a researcher in the Department of Applied Mathematics at The Weizmann Institute of Science in Israel, wrote “one of the first scientific papers on simulation … in the late 70s.” [“Advanced Concepts in Simulation,” Supply Chain Risk Management, 6 April 2011]. That paper was entitled Concepts for advanced simulation methodologies. Stengl writes that Ören and Zeigler aggregated “some fundamental knowledge about simulation” then suggested a “conceptual model for simulation.” Stengl goes on to write about their suggestions “from the perspective of a supply chain.”

“Simulation programs can be described completely using the following terms:

• “Model Structure, describes the static and dynamic aspects of the model, like components of the model, variables and rules for interaction (supply chain context: echelons and elements of the echelon)
• “Model Outputs, contain all variables and functions that can be observed during and after execution (e.g., lead times, profit of the chain)
• “Input Scheduling, when is the model fed with what inputs (e.g., how is demand generated)
• Initialization of Simulation, what are the starting parameters of the model (e.g., are queues already loaded?)
• “Termination of Simulation, what are the conditions for model termination, may be after specific time or specific model states
• “Data Collection, how and when is the data during simulation collected
• Simulator, the simulator carries out the model’s instructions to generate the new model states (nowadays the Simulator is usually a software like Arena, Anylogic or similar)

“All those elements can be found within modern simulation tools, since they do not only include the Simulator component, but also aids to model and control the supply chain.”

Stengl makes it clear that simulation models aren’t simply a matter of plugging data into an easy formula and waiting for an answer to be spit out. He notes that simulation models “can be … characterized by the ‘language’ they are using. … Is the model using a continuous or discrete time scale[? Is it] … block- or expression-oriented[?] … What is the world view expressed in the model by differential equations, events or process interactions[?]” He continues:

“After modeling a supply chain you can do experiments using the model, but of course also in reality (e.g., increasing buffers or changing inventory policies). This would allow the experimenter to validate and compare the models output with the real changes. Of course in a supply chain context experiments in reality are confined to a very small range of parameter variation without compromising the ongoing operations. … The experimental frame is defined as: a limited set of circumstances under which the system is to be observed or subject to experimentation, so the settings which are used for above mentioned components of a simulation.”

Stengl reports that not everyone believes that the supply chain is a good candidate for simulation models. He writes:

“There are strong opponents of using simulation in the supply chain context (e.g., Sodhi and Tang, 2009), since at least when using optimization there will always be an optimality gap. Instead mathematical modeling should be used. But the huge amount of literature using different kinds of simulation … shows that supply chain management can benefit very much from simulation, since the method is probably more easily understood, than some mathematical models.”

In a subsequent post, Stengl looks at a paper that actually used a simulation model involving a supply chain. The paper was entitled Simulation performance in the optimisation of the supply chain and was written by Riccardo Manzini, Emilio Ferrari, Mauro Gamberi, and Alberto Regattieri, all of whom are researchers in the Department of Industrial and Mechanical Plants at the University of Bologna, along with Alessandro Persona, a researcher with the Department of Management and Engineering at the University of Padova. Stengl concludes, “The authors make a strong case for ‘their’ simulation approach: visual interactive simulation and claim that it is especially useful from a cost/benefit point of view.”

Rafael Diaz and Joshua G. Behr, research assistant professors at Old Dominion University’s Virginia Modeling, Analysis & Simulation Center, agree with Stengl that simulation models can be useful for supply chain professionals. They believe, “Business modeling and simulation allows supply management professionals to test and improve processes.” [“Improving With M&S,” Inside Supply Management, August 2011] They continue:

“While it’s difficult to predict a supply disruption or a spike in commodity prices, organizations should have a plan to mitigate the impact of such events on the organization’s strategic objectives. Through modeling and simulation (M&S), supply management professionals can model a disruption and simulate how this event is likely to ripple through the supply chain. The modeler can also test the behavior of the system’s response to the disruption under different ‘what if’ scenarios, such as the availability of a secondary supplier or alternate transportation routes.”

Readers of this blog know that I’m a fan of “what if” analysis (see my posts entitled Modeling “What If” Scenarios and Examining the “What Ifs” in Life). Diaz and Behr claim that “the fundamental motivation behind modern business M&S is rooted in the human drive to identify patterns and regularities in our environments with the intent of better controlling or directing the behavior of the system.” I would venture to say that another motivation is the desire to discover the irregularities that might not be obvious but could have devastating consequences. They continue:

“Business modeling, in its most basic form, is the creation and arrangement of representational elements that approximate reality. Simulation is the potential interaction, or playing out, of these elements over time. The dynamic behavior of a system as its elements interact over time is captured through measures of performance such as revenue or productivity. Once the behavior of a system is understood, supply managers can realign assets with the intent of finding the arrangement that will yield the most optimal measures of performance.”

I think Stengl would probably agree with their assessment. Diaz and Behr assert that “when modeling a system, the goal is not to mirror the real-world system in all its detail but to represent those aspects that are expected to meaningfully affect the performance variables of most interest, such as customer satisfaction or revenue. How does a modeler know what to incorporate in the model? The answer involves the notion of fidelity.” Before fidelity can enter the picture, a company really needs to make an assessment of their critical processes and assets (i.e., those processes and assets whose loss or disruption could cripple the company). Once they are identified, fidelity then comes into the picture. Diaz and Behr explain:

“Model fidelity describes the degree to which the model matches reality or behaves like the real-world system it is representing. As a general rule, a more complete and realistic representation (high-fidelity) is best for those aspects that matter to a system’s performance. For example, within many distribution centers, only 20 percent of the warehoused items are frequently pulled for delivery, but they account for nearly 80 percent of the revenue. When modeling, the high-fidelity focus would be on the demand behavior of the frequently pulled items, because revenue is a key measure of performance. Likewise, a relatively lower fidelity is acceptable in the model for those parts of the warehouse/distribution system that have limited impact on revenue. The model of the real-world system, when simulated, will generate trends in measures of performance. A valid model yields measures of performance similar to the known measures obtained from the real-world system. If the model and simulation are producing behavior similar to the known real world, then the model most likely is an appropriate, or valid, reflection of that real-world system.”

Once you have a model with excellent fidelity its real value (experimentation) comes into play. Diaz and Behr note that “experimenting with the real system could be costly, counterproductive and meet organizational resistance. Experimenting with a model, on the other hand, is both reasonable and economical.” They continue:

“A model offers the opportunity to experiment with different combinations of inventory review schedules while avoiding unseen or unintended managerial costs and consequences associated with real-world changes to the organization’s supply chain. The process of modeling and simulating a real-world system assists in identifying major system strengths and weaknesses. Strengths can be further leveraged while weaknesses can be managed through targeted interventions that have been designed, tested and analyzed prior to real-world adoption. This process of identifying strengths and weaknesses and virtually experimenting with changes to the system allows analysts and supply managers to see unintended second- or third-order consequences that might not have been readily apparent otherwise.”

In previous posts, I’ve discussed recommendations by supply chain analysts to streamline and simplify supply chains as well as make them more transparent. Good modeling and simulation can help in those endeavors. As Diaz and Behr put it: “Selecting and analyzing various supply management activities and processes can be crucial in improving efficiencies and performance. Understanding and testing how an organization’s resources interact at the different stages of the supply chain is critical for improving operations.” They conclude:

“M&S uses scientific methods as cornerstones to informed decision-making. Because of its outstanding capabilities in capturing and processing complexities, M&S is the preferred methodological framework for the analysis, design, modification and improvement of operations. Many successful companies today use M&S as a competitive weapon to improve their processes and activities.”

Not every company needs (or can afford to use) computer modeling and simulation. The more complex your supply chain, however, the more likely it is that you will find M&S a useful (and profitable) tool.