Optimizing inventory with sporadic demand
Our manufacturing client had a particular challenge with sporadic demand for a high number of items.
nVentic were asked to analyse the demand for these items and to identify an approach to manage them that would deliver an improvement in performance and also be pragmatic - compatible with existing technology and with clear criteria for planners to follow.
3 approaches were considered:
nVentic were asked to analyse the demand for these items and to identify an approach to manage them that would deliver an improvement in performance and also be pragmatic - compatible with existing technology and with clear criteria for planners to follow.
3 approaches were considered:
- Deterministic planning, allowing planners to fix orders based on their experience and knowledge
- Plan to forecast, tying orders to the forecast
- Replenishment, setting re-order points
Approach 3 was tested using 6 different statistical approaches.
A range of other levers were also considered, including reducing lead times, reducing minimum order quantities, reducing batch sizes, increasing review frequency, shaping usage, varying service levels, and updating policies.
A dozen criteria were identified to segment items appropriately.
nVentic identified for the client which items should follow each of the 3 approaches, and then, through testing, identified appropriate statistical distributions to support the replenishment approach. An assumption of normal distribution underlies a lot of inventory management technology but this is particularly problematic with highly sporadic demand.
A range of other levers were also considered, including reducing lead times, reducing minimum order quantities, reducing batch sizes, increasing review frequency, shaping usage, varying service levels, and updating policies.
A dozen criteria were identified to segment items appropriately.
nVentic identified for the client which items should follow each of the 3 approaches, and then, through testing, identified appropriate statistical distributions to support the replenishment approach. An assumption of normal distribution underlies a lot of inventory management technology but this is particularly problematic with highly sporadic demand.
Results
Our analysis found that two thirds of the sporadics considered would benefit from a different approach, which could be managed using their existing technology. The high service levels could be maintained while reducing inventory levels by 17% overall.
The remaining third had extremely sporadic demand (no demand at all for 9 consecutive months) and for these items alternative strategies, such as increasing customer lead times or finding alternative sources of supply, were the only realistic option.
We found that Forecast Value Added (FVA) was too low for most items to make plan to forecast a viable alternative. Replenishment was the best option for a significant percentage of items, but it was essential to factor the high variability and sporadicity into considerations of safety stock and re-order level. It is also essential to understand that the segmentation is not static and that the client needs to refresh the analysis regularly to take account of changes in demand.
Our analysis found that two thirds of the sporadics considered would benefit from a different approach, which could be managed using their existing technology. The high service levels could be maintained while reducing inventory levels by 17% overall.
The remaining third had extremely sporadic demand (no demand at all for 9 consecutive months) and for these items alternative strategies, such as increasing customer lead times or finding alternative sources of supply, were the only realistic option.
We found that Forecast Value Added (FVA) was too low for most items to make plan to forecast a viable alternative. Replenishment was the best option for a significant percentage of items, but it was essential to factor the high variability and sporadicity into considerations of safety stock and re-order level. It is also essential to understand that the segmentation is not static and that the client needs to refresh the analysis regularly to take account of changes in demand.
Would you like to talk to one of our experts? Contact us