The Quantitative Supply Chain

Statistical demand forecasting was pioneered in the 20th century, however it can be revisited with better, more modern statistical approaches. Yet, these methods aren't necessarily more difficult to implement than their older counterparts, because software tooling has also significantly progressed. We are joined by Nicolas Vandeput, a Supply Chain Scientist specialized in demand forecasting and inventory optimization. We discuss his new book: 'Data Science for Supply Chain Forecast'.

The lead time is the total amount of time, typically counted in days, associated with the inventory replenishment cycle. The amount of stocks that a supply chain needs to operate tends to be roughly proportional to its lead times. Accurately estimating future lead times is critical for accurately estimating the amount of inventory needed to fulfill future demand. However, it is a fundamental factor that is often overlooked by companies, with a far greater importance being placed on forecasting.

In supply chain, the service level defines the probability of not hitting a stock-out during the next ordering cycle. However, the fill rate defines the fraction of the customer demand that will be properly served. Service levels and fill rates are distinct, and should not be confused.

The Min/Max inventory method defines two stock levels: first, a replenishment threshold, referred to as the 'min', and second, a replenishment target, referred to as the 'max'. Yet, despite its popularity, this method is not suitable for most modern supply chains.

Supply chains have to undergo profound changes to benefit from software innovations such as predictive optimization technologies. However, transforming complex, many-decade old organizations is difficult. We are joined by Hervé Hillion, one of the founding members of SAY Partners. Hervé explains how change can be effectively conducted across an entire supply chain.

An aerospace MRO needs to optimize its inventory in order to both minimize the number of AOG incidents and minimize the amount of stocks, typically held in the form of rotables. We are joined by Antony Nardozza, the Head of Inventory Solutions at OEM Services, who previously worked for both Airbus and Spairliners.

Forecasting demand in aerospace is difficult because demand is both highly erratic and intermittent. Also, TAT (turnaround times) need to be forecasted as well. In practice, probabilistic forecasts are required to achieve any meaningful results in the aerospace industry.

Growth, and more generally trends, needs to be taken into account in order to deliver accurate demand forecasts. However, growth; as a statistical pattern, proves to be more difficult and more elusive to capture than other well-known patterns such as seasonality.

3PL - or third-party logistics - represent a way for companies to externalize their warehousing, distribution and fullfillment. Often when ordering a product on Amazon or eBay for example, the merchant is not actually going to ship the product directly to you themselves and instead uses a specialized company called a 3PL who ships the product on their behalf.

Seasonality is one of the major cyclical patterns that can be used to improve forecasting accuracy. Most supply chain processes tend to be seasonal to some degree. Not only because of demand, but also lead times.