DDMRP: 'We use forecasts where they are the least wrong'

The original vision was that the DDMRP – a flow and stock management methodology – did not need forecasts to function. The reality is a bit more nuanced. Insights come from Jérémy Catteloin, an expert engineer in supply chain management at BEVOLTA.

« In a DDMRP tool, the demand calculation is based on the orders on hand. However, the stock level objectives – that is, the buffer level – can potentially be calculated on the basis of a smoothed forecast over a certain horizon (or not, depending on how the buffer in question is configured).

Thus, you can rely on or completely dismiss forecasts, and this point is generally decided in the project phase. It is possible to revisit this decision afterwards if necessary. The most important thing is that forecasts are used where they are least incorrect, especially to size stock objectives, layer by layer (in the buffer), over a given time horizon, without going to exact precision, while being fairly robust against small forecasting errors that may occur from week to week or from day to day. And remain generally right.

In DDMRP, today's demand calculation is based on today's orders on hand, whereas in classic MRP, today's demand calculation is based on the forecast that was made some time ago. One immediately understands the greater margin for error in the MRP system!

In DDMRP, the Average Daily Usage (ADU) is calculated more smoothly since it takes into account the reality of the week's orders to calculate that week's needs, whereas in MRP there is stronger variability, which becomes more and more incorrect as you go back up the bill of materials. What works on a finished product thus turns out completely wrong upstream of the bill of materials.

By smoothing, the DDMRP methodology ensures more stability. Yet this is key for an industrial system. It helps to avoid the Christmas tree effect with a permanent 'stop & go' and a stronger 'bullwhip effect' (effet coup de fouet).

With buffers we mitigate, absorb, in short we better control the variability of the supply chain. »

Comments collected by Ghislain Journé