Factory flow is non-linear so don't use Master Production Schedules

The use of forecasts and Master Production Schedules to drive ex-stock replenishment via DRP/MRP/ERP/APS is, incontrovertibly, a big mistake.

That's because supply chains through factories are non-linear which, in this context, means that the timing of factory output has a highly unpredictable relationship with that of the input (1). ERP/MRP/APS Master Production Schedules assume that this lead-time parameter is known and stable, the fact that it isn't is a serious weakness and means that providing high service levels generates significant costs and inventory as efforts are made to compensate.

The degree to which the lead-time (ie. the input to output interval) is volatile and unpredictable is related to two factors...........

1. Capacity Utilisation - there is a non-linear relationship between the wait time of materials in front of a work centre and its level of capacity utilisation; as utilisation grows from low levels so the lead-time also grows, slowly at first but increasingly quickly at higher levels of utilisation, as demonstrated below

Lead-times also become more unpredictable at higher levels of utilisation because, as can be seen, they become far more sensitive to processing delays and demand upticks. If the forecasts were accurate and capacity utilisation not too high this non-linearity wouldn't matter because the right amount of material would arrive roughly on time to meet demand. Unfortunately the forecasts aren't accurate because, even with world class forecast mix accuracy of 80%, c80% of items will be found to have >40% error. In consequence, Supply Planners spend a great deal of time interrupting the supply schedules to expedite those items for which there is an apparent service threat and this results in an increase in the other factor which contributes to unpredictable lead-times...........

2. Variability - is the degree to which the materials' speed through the supply chain varies and fails to align with demand, including not moving at all. When materials aren't moving they're queuing and this occurs either because the processing work centre has had problems running at the expected speed (eg. due to a breakdown/quality problem etc) or more has arrived to be processed than the work centre can currently cope with (eg. due to the arrival of batches). The interruption of schedules by Planners to prevent a backorder, caused by an under-forecast, contributes significantly to variability by stopping the flow of the interrupted items, thereby increasing their lead-time and, because of MRP's dependent demand network, this ripples upstream affecting all other items on the same routing - unless spare capacity can be found to catch up. The relationship between lead-time, inventory, variability and capacity utilisation at a single work centre is demonstrated graphically below, is described by this formula and can be seen occurring at Are all Supply Planners driving red cars?

To summarise

1. Lead-times are assumed to be known, fixed and stable by ERP/MRP/APS master production schedules, if they're not then the timing of inventory arrivals cannot be relied upon and service levels suffer
2. Increases in variability and capacity utilisation lead to longer and more unpredictable lead-times
3. The growth and unpredictability of lead-time is particularly severe at higher levels of utilisation
4. Lead-time growth and unpredictably becomes severe at lower levels of utilisation if variability is high

These non-linear impacts upon lead-time, caused by variability and capacity utilisation, mean that using inaccurate forecast-driven master production schedules through a factory leads to a self-inflicted vicious circle in behaviour that destroys performance: the inaccuracy of the forecasts and master production schedules leads Supply Planners to increase flow variability by interrupting the schedules to expedite backorder threatened items, this increase in variability increases the length and volatility of the lead-times which further impacts service levels. Efforts to improve service through further micro-management of the schedules (S&OE anyone?) or by increasing the MRP lead-time parameters makes the problem worse by further increasing variability and capacity utilisation respectively.

As a consequence, manufacturing supply chains, driven by master production schedules, perform very ineffectively - they operate with lead-times that are longer, and inventories that are greater (2) than they need be, they have difficulty providing reliable service levels and shop floor stability is frequently disrupted by schedule interruptions - all of which also has a detrimental impact upon cost of goods because the factory isn't stable and has to periodically use unplanned capacity to "clear the backorders".

You can gauge the degree to which your supply chain is under-performing by tracking the frequency of expedites, use of unplanned over-time and comparing your actual inventory value with that of its theoretical level (calculated by aggregating, for all items, their 1/2 average order quantity plus safety stock -3).

How should supply through factories be driven if not with an MPS? An important criterion is that any new methodology should involve as few schedule interruptions as possible because it is they that generate the performance destroying flow variability.

The alternative to inaccurate forecast-driven Master Production Schedules is Enterprise-wide Pull (often known as Demand Driven MRP). This methodology uses one or more re-order point/cycle decoupling mechanisms within the supply chain to enable materials to flow in line with demand, thereby eliminating the variability generated by schedule interventions/expedites (and multiple decoupling mechanisms also prevent process variability being propagated/amplified). Master Production Schedules are no longer necessary for driving execution as replenishment by the de-coupled work centres simply follow a stable repeating sequence and cycle of the quantities needed to reach each items stock target - without any deviations and interruptions!??

It uses a robust process for managing significant and exceptional demand / supply events needing advanced stock builds and requires a periodic calibration process. The latter involves a number of Master Settings being periodically updated in line with future average demand trend and variability, lead-time, desired service level and others, see Forget the Master Production Schedule

The best way to determine the degree to which your manufacturing supply chain would benefit from adopting Enterprise-wide Pull, in place of master production schedules, is to run a simulation using historical demand and your parameters and compare your actual performance, and follow that up with a software supported pilot (4).

Typical benefits are consistent achievement of planned service levels from c40% less inventory with significant reductions in lead-time, use of catch-up capacity, expediting and factory floor instability (5).

For further information, see The SC Replenishment Problem (and how to solve it) and/or SC Variability - what it is, why its bad and how it can be minimised

If you need any help or advice, just contact me through LinkedIn.

(1) This non-linearity applies in any supply chain in which there is a processing capacity constraint, a factory work centre is an obvious example but it also applies at warehouses, to new product development pipelines, NHS waiting lists, unsolved crime backlogs etc etc

(2) Increased lead-times lead to increased inventory through Little's Law: inventory=lead-time (days) x throughput (daily)

(3) You should also check that your safety stocks are sized correctly, for instance using x weeks of the forecast provides you with no control of the service level provided which might be far too high or low. At the very least you should use a calculation that recognises demand variability such as the cycle service calculation: z factor for x% cycle service x SQRT replenishment cycle x SD demand variability. For further detail see Supply Chain Service Calculations

(4) Software is mostly SaaS/Cloud and operates through your ERP transaction system, the providers often provide such a simulation and low cost pilots can be set up very quickly. See DDI compliant software

(5) See DDI case studies