High-Tech Supply Chains: Harnessing AI for Efficiency

    We are all at the whim of a great beast that lives in the heart of our society. It lurks, just beneath the surface, waiting to rise up and snarl us without warning. At least, that’s one interpretation of supply chains.

    The aerospace industry is no different, and is arguably more susceptible to shifts in the global supply chain. After all, base materials like aluminum start out as bauxite mined in Australia and spend a full year in transit or storage before it arrives as sheets for fabrication. And that’s just aluminum. Every level of the aerospace industry is dependent on long and complex supply chains.

    The average aerospace vehicle contains a myriad of systems, from physical flight controls, chemical propellants, batteries, and a truly impressive amount of computing power. That, combined with a massive ecosystem of companies and sub-contractors, means that the aerospace industry is inherently complex. Not just complicated, but truly complex.

    Complex vs Complicated

    To understand the nature of the beast, we have to first spend a few minutes diving into the idea of complexity. And here, we don’t mean how a heating and cooling system on a jet engine works. That’s undeniably complicated, but it’s not complex.

    Complex systems are those in which the same inputs don’t necessarily yield the same results. Take farming, for example. There are so many different variables and probabilities at work determining if a field will yield a bumper crop, that the farmer can control everything exactly the same, year after year, but there’s no way to predict rainfall with certainty, no way to predict storms and blight and flooding. So farming is a complex system by nature, just like aerospace.

    So is the global supply chain. Sure if you aggregate enough data and a high enough level, you can get some amount of predictability. This is called forecasting, and it is a critical part of how many cutting-edge industries operate. But these forecasts aren’t certainties. And that’s a problem for the human part of the equation.

    The Human Factor in Supply Chains

    People don’t do well with uncertainty. Even the cutting-edge technicians who work in aerospace are still, at their core, human. As humans, we all want to know things for sure. So the tendency is to take high-level forecasts that are probabilities and look at them as omens of the future. Oracles, if you will, are applied to the lower levels of a supply chain. This leads to many problems.

    Chronic over or under-ordering applies a bullwhip effect back up the supply chain. Each level has a little bit longer tail on it, to the point where manufacturing sometimes takes a year or more to order raw materials. Remember the aluminum example? It all has to come from somewhere.

    Heavy hitters in the tech industry, from Lockheed Martin to IAI, rely on these forecasts to manage their supply chains. Historically, they’ve optimized for each level independently, and let the market forces do what they will. But there’s a new idea in town.

    Genetive algorithms, machine learning, all forms of artificial intelligence, allow a revolutionary take to supply chain management. The buffer is the key.

    At each level, there is a buffer of goods or materials waiting to be either sold or processed. This buffer represents waste, as these materials aren’t being actively used, and have to be kept somewhere. The automotive lot is the classic example; acres of cars built in the hopes that someday, someone will want them.

    Solutions for the Future

    AI allows cascading analysis of the supply chain demand in real-time, which means that the buffers are no long analyzed on a quarterly or even annual basis. They can be adjusted minute to minute, based on real market demand signals and the constraints of the specific supply chain.

    In addition to AI, the innovative use of microsatellites and maritime shipping allows the raw data capture necessary to fuel real-time algorithmic analysis.

    This application of AI may seem like science fiction, but these algorithms are already in use by some companies with incredible results. Manufacturing giants like Toshiba have seen a nearly two-fold increase in their supply chain efficiency with this technique.

    So when you go out and buy that next thing you need, just remember that someday, you might place an order, and it’s made the same day and delivered. After all, most tech is on-demand; why shouldn’t the future of aerospace be on-demand?


    Latest articles

    Related articles