Globalization, economic volatility, digital commerce, and constantly changing product mix and customer preferences have made the traditional supply chain—with its fixed facilities, links, and rules, designed around stable markets and consistent relationships—obsolete. In today’s market, variability is the norm.
The modern supply chain must be more sophisticated and more flexible. It needs to “bend” into different shapes, and transform itself into new configurations, based on market need. This transformation must take place quickly and reliably, to meet cost and service level demands both at the operational and strategic levels.
Transportation problems, for example, are among the most complex operational issues that supply chain planners and executers face. The combinatorial nature of these problems results in numerous decision choices, each with its own cost and service level outcome. Any shipment generation process requires solving a variant of the vehicle routing problem (VRP). Copious numbers of variables—costs, resource availability, common carrier rates, time windows, service-level requirements—make trial and error impractical. There really isn’t an “average” day in transportation, and the math behind the supply chain must adapt accordingly.
In today’s complex environment, tactical and strategic decisions must also be supported by proven optimization algorithms; companies cannot afford to make strategic, high-value decisions with simplistic analysis and rules. Mathematical models and optimization algorithms provide these answers, with sophistication that increases with the intricacy of the supply chain configuration to support a holistic, end-to-end approach.
Executives and managers, of course, need to have confidence when making mission-important corporate decisions. However, they often treat decision support systems as wizards, or oracles. They tend to forget that mathematical models and optimization algorithms are based on inputs (in other words, “garbage in – garbage out.”).
Quality and comprehensiveness of the input impacts the quality and reliability of the output recommendation. But you must look beyond simply examining the quality of the input, particularly for strategic business problems. Here again, variability is a big factor. What happened in the past may not – and will not – happen in the future. Using averages, means or medians will not be enough to capture the variability. Tools that utilize algorithms that work in stochastic environments—settings characterized by probability, and random variables— and have scenario management and simulation capabilities, are in high demand, but low supply.
The modern supply chain is defined less by assets than by the relationships between these assets. Setting up these relationships involves many decisions, mostly to be made in a hierarchy. The number of distribution centers (DC) to operate, which vendor should supply which DC, which stores to service from which DC, pickup and delivery frequencies and schedules, fleet territory decisions, and common carrier contracts are some of the critical links that define the overall supply chain. With tools powered by math, managers can set up hypothetical supply chains, generate different scenarios with the help of configurable optimization, and evaluate them under possible probabilistic or futuristic conditions.
In summary, today’s complex supply chains have to be powered by math that factors in variability to make fast and correct decisions. Without it, supply chain managers are bound to remain stuck with tedious data massaging, long manual analyses, and risky decisions.
Aykagan Ak has a PhD in Industrial Engineering from Georgia Institute of Technology. He is a senior manager in the Science Team of Manhattan Associates (www.manh.com), responsible for the design of optimization algorithms.