Operations Planning and Scheduling

Multidimensional Planning

Manufacturers are seeing early success with AI in operations planning and scheduling because of AI’s ability to analyze multiple dimensions at once. As EY’s Americas Data and AI Leader for Manufacturing Farooque Munshi pointed out, “With traditional models, you were optimizing either for growth or capacity or cost. But with AI, manufacturers can look across different factors, run scenarios, discover bottlenecks, and accordingly make decisions. We are definitely seeing early wins and success stories that are driving productivity and growth.”

AI for demand forecasting is helping one communications technology provider address unpredictable demand patterns. “We’re able to leverage an AI platform to use master data going back 10 or 20 years to develop predictive analytics for SKU-level forecasting on a monthly and quarterly basis,” an operations leader told us. “As a result, we’re much more efficient at production planning. Before we started this AI initiative, our demand forecasting was about 65-70% accurate. Now we’re anywhere from 88-92% accurate by leveraging our master data, market intelligence, input from our sales and marketing team. It has paid some serious dividends by reducing a lot of manual inputs,” he added.

For production planning and optimization, one pumps manufacturer is turning to AI to better handle large spikes in orders from its high-volume gas and water meter businesses. A senior executive shared, “When a large order from a municipality or a utility comes in, it can cause major disruptions on the factory floor. As those orders come in, we use AI to analyze multiple sources of information, such as our order pipeline, potential orders including likelihood, and matching all of that with what we have in inventory, as well as what would need to happen to be able to manufacture those specific orders. As part of the analysis, we can identify orders that are high-margin or urgent and prioritize those.” Before AI, it would have been necessary for a planner to “flip through five different Excel sheets and hundreds of lines of data,” she explained. The results were impressive in the first few weeks of the project. “What used to take one person several days can now be done in five or 10 minutes. Based on the first few weeks of the program, we also expect to see productivity savings on the production line by increasing accuracy and avoiding last minute order changes,” she said.

GALLO is planning to deploy AI to reduce changeovers. “Changeovers in production take time,” Erich Kaepp, Chief Operations and Supply Chain Officer explained. “We want to avoid setting up the machines for a particular product and then discovering at the last minute we have no labels. With AI we can make sure we actually have all of the products necessary.”

One component manufacturer is using AI for order intake optimization (OIO). “We have been production constrained for a long time, so it is very important that we're properly allocating the capacity we have to the demand that's coming in. With our OIO system, orders come in and we dynamically optimize how much product goes to each customer based on our inventory and what we can produce within the lead time. Afterwards, we communicate with our customers to let them know what will arrive in their shipment and when. In the past, if we couldn't fulfill an order, I didn't know it until the time of shipping. This meant in some cases our trucks were going out half full – I was just shipping air. With this new process, we're doing all that math up front, and I'm able to optimize what will ship at the end of the day,” a distribution executive shared. 

The results have been impressive: “We've saved about $5-6 million per year in transportation alone. Our service penalties have dropped 75% from $20-30 million per year to the low single digits,” the executive said. “We’re also doing a better job of communicating with our customers. Since we are talking to them earlier, it has softened a lot of the relationships and eliminated a lot of the noise in the system. Since the order intake optimization system went live, we have executed better than we have in a decade.”

Load Shifting for Energy Management 

Energy costs are a growing concern. Philipp Leutiger, Senior Partner at Roland Berger talked about working with manufacturers to use AI for energy load shifting to avoid peak pricing. “There are different demand patterns depending on the hour of the day, so you can take demand numbers and use AI to drill down into how your operations are going to be set up. Demand forecasting is a super relevant topic right now,” Leutiger said.

One energy-intensive manufacturer we interviewed uses AI for weather-responsive energy control. Through a combination of traditional machine learning and AI, the company is able to determine how temperature and humidity affect energy consumption and costs. “We saw a material impact from simply understanding the correlation. Now, if we expect a certain type of weather, we can plan around it and schedule higher consumption for different days,” the manufacturer told us.

In the operational planning and scheduling space, AI is helping manufacturers get more out of their shop floor operations in terms of existing assets and staff. Manufacturers are able to avoid being surprised by spikes in demand causing bottlenecks in production and material as well as staff shortages. This puts them in a better position to meet deadlines for critical and high margin orders.