Wsd aw wad aad

Shires, N.; , "MRP-more resources please!," Increased Production Through Discrete Event Simulation, IEE Colloquium on , vol., no., pp.5/1-5/3, 24 Jun 1993
URL: http://ieeexplore.ieee.org/stamp/stamp.jsp?tp=&arnumber=247517&isnumber=6335

MRP - More Resources Please !
Dr Nigel Shires

High ratios of lead time to production time and non-adherences to schedule which are evident inmany MRP-controlled factories today. One study among many to observe this has been carried out as part of the Eureka FORCAST'') project, carried out by a number of collaborating companies and Higher Educational Institutions in the UK, France and Austria.
An outcome of one part of the FORCAST project which has been aimed at reducing the no adherences to schedules and reducing the lead time /production time ratio is a discret event simulation – based scheduling toolkit called Preactor.

To a production planner carrying out day-to-day scheduling in a factory, it often appears at first sight that in order to reduce the lead times and achieve adherence to schedules more resources (machines and people) are required. The production planner sees the MRP system as something which is loading thefactory with work which can never be completed by the delivery date, and hence probably thinks that MRP stands for More Resources Please !

Production scheduling is concerned with balancing the production capacity (the resources) required and the capacity available. The measurement of capacity available should be measured with an accuracy appropriate to that which the scheduling functionrequires. For example, at the level of MRP, one of the scheduling functions is to determine when material is required for purchasing. The capacity of the resources is assumed to be infinite. Production operations are loaded into time buckets (typically week by week) and the predicted load of operations on the resources is then 'levelled' to balance the capacity required to that available. The daterequired for material to be purchased can then be predicted.

The measurement of production capacity available within a time bucket at the MRP level has some assumptions built in (i.e. there will be no effect from sequence-dependent set-up times, jobs can be transported at any time within the factory). This measurement is normally an overoptimistic value, and hence a technique is used to take someaccount of these assumptions in the capacity measurement. This technique is to impose one or more overall constraints such as allowing only one operation per job per week to take place, or allowing a maximum load of
80% within any single time bucket. These overall constraints improve the prediction of when the jobs will actually take place but at the expense of reducing the effective capacityand resulting in the well known large ratio of lead time to actual production time.
The prediction of when the operations and jobs will take place is normally sufficiently accurate for purchasing requirements (where there are medium to long term lead times for purchasing raw material stock, for example). However, the prediction is not sufficiently accurate for day-to-day scheduling, in that thereis a level of confidence associated with the prediction that any particular operation will actually occur in its predicted time bucket. This confidence level is related to the accuracy of the resource capacity prediction.

If the resource capacity prediction was totally accurate (no assumptions) and the factory was
totally deterministic (all operation durations could be predicted with 100%accuracy), then it
would be possible to be 100% confident in the prediction that a particular operation would take place in a particular time bucket. However, in the real world, one is more likely to be 70% confident that a particular operation would take place in a particular time bucket. If an operation does not then take place in the predicted time bucket (normally due to the resource being...