Malcolm McLean, a trucker from North Carolina, was tired of waiting. He had a load full of cotton bales, and while idling away hours at a shipyard watching stevedores load other cargo onto ships he dreamed up containers that transformed global supply chains. Loading and unloading trucks was inefficient, so his frustration led to the commercialization of containers that could be moved on and off trucks easily. Containerization eventually reduced shipping and loading costs by at least 75%. The trade with Asia we take for granted today was only possible by mitigating a significant supply chain trade-off – reducing costs without appreciable impacts to quality and service. Supply chain optimization has also improved in significant ways that can address these trade-offs better than before.
Trade-offs are inherent in supply chains, so companies that address them better gain significant competitive advantage. Operational innovations like the invention of containers led to the huge growth in global value chains, and today 95% of manufactured goods move on ships. Technical innovations can also impact supply chains, which is why many turn to supply chain planning solutions, which are designed to be trade-off machines. Work planners previously did manually can be calculated by software. But advances in this area have two limitations – the software itself and our ever-changing world.
The myth of the “perfect plan”
The promise of supply chain planning software has led leaders to chase the mirage of the perfect plan. Analytical techniques like linear programming can create the mathematically “optimal” plan, but these methods must be implemented well to avoid creating other challenges. A plan is a point in time, so any solution must be paired with agility to adjust when the inevitable changes occur. Optimization has often been applied too narrowly, because as AI and supply chain expert Professor Nada Sanders articulates, analytical techniques applied to just one node only create highly-efficient silos. Optimizing one link doesn’t optimize the entire supply chain – you must consider the impact of any decision on the entire network.
Poor implementation of optimization has also created biased perceptions of mathematical optimization algorithms as slow or rigid. Optimization is a very powerful approach to addressing supply chain trade-offs with complex interdependencies, but improper model formulation (how the math is set up to address the problem) can add hours to solve time. And if the tools don’t allow flexibility for the planner using them, the resulting rigidity can limit their utility and planners resort to manual estimates. The irony is that software designed to address trade-offs falls prey to a trade-off itself – the belief that a planner must choose between a perfect (but rigid) plan that requires a long wait or a suboptimal but flexible plan they can generate quickly.
Software written to solve yesterday’s problems also doesn’t account for the manifold ways the global value chains shipping containers helped enable have also increased in complexity. AND the pandemic unleashed a combinatorial series of impacts that are still working their way through the chains. AND with growing concern about climate change, trade-off considerations must account for cost, quality, service, and now sustainability. The problem space has gotten harder, not easier.
Supply chain optimization for today’s realities
Thankfully, technical innovations have emerged that can help narrow the problem space. Algebraic gymnastics cleverly cut down the compute time needed by optimization solvers, which run on modern machines capable of much faster compute times. Better problem formulation targets the math more precisely, speeding up run time. And fusing mathematical techniques together allows each approach to bring its best to problems. Fast heuristics can narrow the problem space with a near-optimal but feasible solution that is also adaptable in real-time. We can think of optimization to parameterize the heuristics, or in lay terms, optimization guides the forest and heuristics guides the trees. And then we can layer in AI and machine learning to speed up the quality and speed of the solve by intelligently selecting the best algorithmic approaches. Optimization is then positioned to run faster in a narrowed decision space but with flexible objectives and granularity across any horizon.
Fusing analytical approaches improves the math behind optimization, but to avoid highly-efficient silos it should be paired with concurrent planning. Common approaches to optimization involve using a spreadsheet solver or third-party optimization tool and then passing it to the planning solution, a workflow that increases the risks of errors in the handoffs. When the master data and results are all in the planner’s regular workflow, not only are these risks mitigated but visualization of the results is in context of the problems they are trying to solve, which greatly increases understanding and likelihood of adoption by making what might otherwise be viewed as “black box” output more explainable.
Supply chain complexity has many dimensions. One dimension is managing trade-offs with multiple interdependencies, a situation well-suited to optimization. In this use case, optimization helps a planner identify the best plan based on what is known. But planning for what might happen in the future, especially when there are the inevitable disruptions, is another dimension of complexity for which creating scenarios can be critical. Scenarios allow planners to ask all kinds of what-if questions to help determine the best course of action under a variety of future outcomes.
Why supply chain optimization matters
If you’re a business leader, why do you care about better math? Imagine you are a high-tech manufacturer with more inventory in stock than your forecasted production volume, which puts you in good company, since US manufacturers’ inventories continue to rise. Not only are these pre-purchased components and assemblies gobbling up cash, they raise the risk of obsolescence, which these days represents not only wasted resources but also a potential hit to sustainability metrics. You have a limited procurement budget but want to maximize revenue with a supply-driven production plan that minimizes waste. What additional components would allow you achieve your business objectives? Fusing fast heuristics, optimization, and machine learning can answer that question, and scenarios can help identify the budget worth allocating. One company using this approach was able to invest only $3,000 to purchase additional inventory but generate an incremental revenue gain of $110,000. Now that’s an ROI!
Other problems that benefit from this fusion of fast heuristics, optimization and machine learning include the common blend challenge prevalent in life sciences – how do you make the ‘best’ use of the available ingredients and select the optimal processing techniques to maximize total demand satisfied? Or optimal distribution – what’s the best way to maintain balance through disruption and shortages while eliminating delivery bias across the network? These are critical and expensive problems in need of better solutions.
Creativity drives better solutions
As digital transformation increases the adoption of automation and algorithms, it’s important to highlight that replacing the human is not the goal. The best use of fancier math is to automate the obvious and calculate the complexity that is beyond the cognitive capacity of our brains. But we must remember that math lacks context, collaboration and conscience. It cannot derive meaning from context, build relationships, or be held accountable. As Gary Kasparov, the famous chess grandmaster, said: “It’s about our creativity, our intuition, our human qualities that machines will always lack. So, we have to define the territory where machines should concentrate their efforts.”
Fusing heuristics, machine learning, and optimization greatly increases machine precision and can complement human ingenuity in better ways to address those vexing supply chain trade-offs. And we can minimize the perceived trade-off to calculate trade-offs – we get more accurate answers without overly compromising on speed. Malcolm McLean spurred the global shipping revolution when he had to wait too long for his cotton bales to be unloaded. You don’t have to wait too long for more intelligent decisions – it is possible to have both speed and accuracy through better supply chain optimization.
Polly Mitchell-Guthrie is the VP of Industry Outreach and Thought Leadership at Kinaxis, the leader in empowering people to make confident supply chain decisions. Previously she served in roles as director of Analytical Consulting Services at the University of North Carolina Health Care System, senior manager of the Advanced Analytics Customer Liaison Group in SAS’ Research and Development Division, and Director of the SAS Global Academic Program.
Mitchell-Guthrie has an MBA from the Kenan-Flagler Business School of the University of North Carolina at Chapel Hill, where she also received her BA in political science as a Morehead Scholar. She has been active in many roles within INFORMS (the Institute for Operations Research and Management Sciences), including serving as the chair and vice chair of the Analytics Certification Board and secretary of the Analytics Society.
