Case Study: Using AI to work effectively with carriers and cut supply-chain costs

Artificial intelligence is growing up. As it matures, it is increasingly applied to industrial supply chains  to help optimize operations, improve efficiencies, and enhance decision-making. Often AI algorithms are used in demand forecasting to analyze trends and optimize inventory levels. AI can also optimize transportation routes and delivery schedules. It can power robots to streamline warehouse operations.

AI is promoted as a supply-chain panacea, but executives often ask for concrete examples of maturity. Where is AI actually being applied to complex problems?  

Below is one example of how extending supply-chain planning to include transportation carriers can advance the use of AI and cut millions of dollars from a business’ transportation spend.

Case study

The huge variability in freight volume from day-today drives significant inefficiencies for carriers, which ultimately affects the prices they charge. In many cases, the desired carrier simply refuses the shipments. As a result, shippers tender some loads to less-desirable carriers yielding a worse cost/service package.

A smarter solution is providing carriers with what they need to minimize deadhead miles and position their equipment to manage a smooth flow of consistent volume. George Lawrie from Forrester Research said, “Sales and operations planning now synchronizes materials, labor, and production capacity, period-by-period, with expected demand. Why not do the same with transportation?"

One large manufacturing company is successfully doing this by leveraging AI-enabled software to include transportation in the synchronization process. This synchronization uses technology that didn’t exist a few years ago. The technology employs a combination of AI and operations-research techniques to simultaneously manage flows on every lane across more than 300 lanes. The importance of being “simultaneous” is that trade-offs must be made for the whole network to be efficient.  For example, much-needed product must take precedence. It needs to be first to use limited truck capacity, space or labor.

Using data from systems such as supply planning, demand planning, enterprise-resource planning, transportation and warehouse management systems (TMS/WMS), this new technology:

·      Looks at all deployment transportation lanes, assessing each product's prioritization as well as network constraints, costs, and loading constraints for each truck.

·       Adjusts the optimized quantity of loads based on what must ship and the associated cost. (Historically, it was thought that you could "always" get a truck—they were an infinite resource—but at what price?).

·       Creates network rewards and penalties for transporting items during time periods other than the one the demand-planning system mandated. For example, shipping a high-priority demand item early is good—shipping it late could hurt customer service and satisfaction.

·       Simulates loading to determine what will ship each day for the next day for 30 days.

·       Tells the transportation-management system to reserve a vehicle—i.e., create the tender. This is important because most trucking companies lack the sophistication to meaningly use a forecast. They do, however, understand orders.

The load's contents, in this example, are defined as late as possible in the process and using the latest data from the supply-planning system. This, in itself, requires advanced logic. If the supply-planning system requires:

·       More product than can be accommodated on the earlier tendered loads; only the highest priority items should be sent.

·       Less product—i.e., the loads would not be full, wasting valuable trucking capacity, then the load contents must be augmented with future needs.

With this in place, gone are the days when:

·       Trucks queue up at a receiving site because there is minimal coordination among the lanes being planned into that location.  

·       A shipping plant is asked to load more vans than it can do or add overtime.

·       Trailers can’t be unloaded because there is no space in the building.

·       The shipping site has to shut down lines or move products off-site because there is no space to put what it produces.

What has this meant for the manufacturer:

·       Significant increase in first-tender acceptance, meaning they get the carrier they want at the right price /service.

·       A significant decrease in total transportation cost.

·       Customer-service measures such as "on-time" delivery hold steady or improve. The global VP of supply chain for the large manufacturing company mentioned above said, "One of the things we discovered is that the tool can maximize stock better than our planning system."

For the carriers, there is a steady, repeatable stream of volume, which enables them to position equipment with minimal deadhead miles. To be clear, the volume is not perfectly smooth. The company recognized that this would be a naïve approach as it is essential that it maintain a very high level of customer service. If there is a sudden rush of sales on one SKU from a DC, then, to avoid a stock out, that must be replenished. Similarly, other locations that serve the same distribution center may need to be reduced to handle and allow the shipment of the more critical SKU.

As this example shows, supply chains can leverage this newly grown-up AI to help themselves become more mature all while helping close the gap between planning and execution. AI can transform supply chains by reducing costs, improving efficiencies, and enhancing customer service and satisfaction.