PSQ Newsletter

Just how robust are DDMRP Buffers? In an exercise at the 2015 SAPICS Conference, we used completely random data supplied by 57 different sources to test the resiliency of the DDMRP method in both long and short lead time environments. The results speak for themselves.

Demand Driven Technologies was an exhibitor at the 2015 SAPICS Convention in Sun City, South Africa. As a demonstration of the resilience of Demand Driven MRP buffering, DD Tech collected random demand values from visitors to its booth. The goal of the study was to illustrate how a DDMRP buffer could achieve very high customer service and strong inventory turnover without the use of a forecast.

Each participant in the study was handed a card and asked to provide 10 days of demand records which could be of any value from 0 to 500. There were no other restrictions placed on the input provided by the study participants. The completed cards were sequenced as they were submitted by the visitors and then inputted to DD Tech’s DDMRP simulation tool. On the right is an example of a completed card.

With 62 participants in the study we were able to model Part 1 for a full year. We modelled part 2 using the 260 days of demand not used in Part 1 and re-used the first 105 days of Part 2 demand to complete a full year analysis. Details of key parameters for each part were as follows:

In addition, we set the buffers to have the ability to Qualify Order Spikes only 3 days in advance. This limited the benefit that Order Spike Qualification would have on the results achieved by the buffer. The Order Spike Threshold was set at 10% of the Red Zone.

The simulation of the demand for Part 1 – Widget resulted in 100% customer service and 5.74 inventory turns for a part with a 90 day lead time. The buffers rapidly increased in size during the first few months of the simulation as the demand provided by the participants was much higher than the starting assumption for ADU of 30 units per day.

The buffer trend graph demonstrates how it quickly increased in size and adjusted to the greater rate of demand of 183 units per day versus the starting assumption of 80. The actual ADU rate was more than 2 times the opening assumption and pressured on hand position which bottomed out at 4566 units. However, the incoming supply from orders generated early in the year put the buffer back into a strong position to maintain service the balance of the year.

During the year there were 19 supply orders generated with an average order size of 3534 which equates to 19 days supply.

We then turned off Order Spike Qualification and found that the simulation generated identical results. This indicates that the level of variability in this sample was moderate and didn’t require the additional protection that Order Spike Qualification provides.

The simulation for Part 2 resulted in 100% customer service and inventory turns of 15.56. The minimum on hand balance was 380 units and suggests that increased red zone safety coverage would be appropriate to further reduce the risk of stock outs.

In the simulation for Part 2 the buffer size flexed upward to address a greater rate of demand than in the Part 1 scenario. Average daily demand for the year for Part 2 was 204 units per day and was 2.5 times as large as the opening assumption of 80 units per day. On hand inventory levels were under pressure in this initial ramp up period and reached a minimum on hand of 276 units.

Inventory Turns were at 16.69 per year which indicates an excellent flow for this item while maintaining 100% customer service level.

54 Supply Orders were generated during the year supporting the rapid turnover rate for the inventory. Average Supply Order size was 1381 which represented roughly 7 days supply.

As in the Part 1 example we turned off Order Spike Qualification. In this case we experienced 3 days of stock out and a service level of 99.2%.

While 99.2% customer service would be considered excellent in most clients we then addressed the stock out days by increasing the Safety Percentage for the Red Zone from 120% to 200% with the following results:

Service improved to 100%. Minimum on hand increased to 437 units while inventory turnover rate declined slightly to 13.78 annual turns. Again, this result was achieved without any forward visibility to sales order demand.

Building on the Part 2 simulation we then tested the impact of a large Minimum Order Quantity (MoQ) on buffer performance. We simulated a MoQ of 5000 for Part 2. This represented roughly 33 days consumption for a part with a 21 day lead time with the following results.

Inventory turnover declined roughly 35% to 8.90 turns. Supply orders declined to 13 from 54. Average on hand became 8370 which is roughly 41 days of supply. Minimum on Hand increased to 2614 as a result of the larger and less frequent order size.

In our work with clients we consistently find items with Minimum Order Quantities representing substantial multiples of usage over the part’s lead time. For purchased items, this often represents an ineffective trade off as the resulting discount rarely justifies the impact the MoQ has on the flow of materials. The same can be said for minimum batch sizes in production where efficiency metrics cause large ‘artificial batches’ which impede flow.

The simulation of buffer performance using random demand values provided by visitors to our booth was a very real and interesting test of the Demand Driven MRP methodology. Other than the upper limit of 500 we had no idea what demand input we’d be getting from the study participants.

Both parts that were modelled in the simulation achieved service levels of 100% while also driving very solid inventory turn-over rates. It’s critically important to understand that this performance was achieved with at most three days of forward visibility to demand.

We applied a high variability safety threshold due to the unknown rate of demand which drove the perfect service levels achieved in the simulation. We also used the simulation to demonstrate how adjusting buffer parameters such as minimum order quantity affects buffer performance.

The core concept of Demand Driven MRP buffers is that they are designed to achieve constant material availability. The resilience of the buffers was proven in the examples above. Supply orders were triggered based on actual sales and the penetration of the buffers. High inventory turn rates were achieved without the prevalent inventory distortions seen in forecast driven methodologies.

DDMRP also provides users with a very easy to follow signaling system for planning and supply chain execution.

DD Tech provides free simulation analysis to companies interested in gaining a better understanding of the impact that DDMRP tactics and technology can have on their supply chain performance. If you’re interested please feel free to contact us at:

There is a rapidly growing body of knowledge regarding Demand Driven MRP. Please refer to the following links for more information

We’d like to thank the large group of SAPICS attendees who visited us in the Expo and participated in the simulation. We greatly appreciate their time and interest without which this study would have been impossible.

We’d also like to thank our partners at PSQ who assisted with the simulation. PSQ provide Demand Driven MRP training and implementation support in South Africa. Learn more about PSQ at www.psq.co.za