1 2 3 4
Product 1 680 0 680 0
Product 2 0 680 0 680
The net result of the second plan (one product per week at operation B) is that the output per week has increased from 560 units to 680 units, however additional inventory is accumulated between operations B to C. The relative cost of this additional inventory must be weighted against the relative increase in profits due to increased output in the selection of a final plan.
In a realistic example, the variability of production times, fluctuating sales requirement, and conflicts in scheduling multiple products must all be considered simultaneously. Our point in the simple example has been to demonstrate that the bottleneck operations must drive the entire planning and scheduling of manufacturing shop.
OPT produces production plans and detailed schedules using four basic modules.
1) Build net: This module crates a model of the manufacturing facility using data on work center capabilities (processing and set up times), routings that describe the flow of product through manufacturing, bills of materials, inventories, and sales forecasts. This model is in the form of a network.
2) Serve: The model of the shop is run through an iterative process to determine bottlenecks in the system. Serve resembles MRP in its workings, and one of its outputs is a load profile for each of the resources in the model. The most heavily utilized resource could produce a bottleneck in the system and must be examined carefully. Sometimes, rescheduling work from the heavily utilized machine to some other alternate machine may produce satisfactory results.
3) Split: The network model of the shop is divided into two parts: critical resources and non-critical resources. The bottleneck operation and all operations that follow it in the order of manufacturing (e.g. subassemblies and assembly) up to customer orders are included in the critical resource portion of the network. The remaining portion of the network includes non-critical resources.
4) Brain: The operations in the critical resource portion of the network are scheduled using a module called the Brain of OPT. the Brain of OPT determines production and transfer lot sizes and the timings of production for each product for the bottleneck operations. Its output is fed to serve module, which then produces the entire production plan.
There has been considerable discussion of the differences between MRP and OPT. We believe that, for dependent demand situations in a multiple fabrication/assembly operation, MRP provides a basic planning framework. Several ideas used by OPT can be useful for enhancing the workings of MRP. Particularly, the active identification of bottlenecks and the adjustment of schedules so that these bottlenecks are used judiciously (by transferring work to an alternate source, reducing set up times, or scheduling a long run) will improve the effectiveness of MRF. Similarly, the goal of reducing inventory in MRP can be further enhanced by distinguishing between production lots and transfer lots and by reducing the size of transfer lots. The basic files used in MRF are directly usable in OPT, so that system conversions can be made.
Mangers of process focused systems face extreme complexities in planning, scheduling, and control. These complexities are at a maximum when the productive system must cope with different levels of dependency of items in assembly, subassembly, and component manufacture. Managers who understand the dependency relationships are better able to create effective systems of planning and control.
The conceptual framework of requirements planning recognizes that the nature of demand for primary products is substantially different from that for the parts and components of the primary products. Demand for parts and components is dependent on the production schedules of the primary items. Often, parts and components are used in more than one primary product, so the requirements schedule for these items are determined by summing the needs indicated on the master schedule for all primary products. The demand dependence of parts and components has profound effects on the policies used to determine the timing of production orders and the size of production lots.
The demand for parts and components is dependent not only on the quantities needed but also on the timing of supply. Because we are dealing with an interlocking structure, the components must be ready for use at a precise time. If they are too late, production of primary items will be disrupted and delayed. If they are too early; in-process inventory costs will increase.
Because demand for components is dependent and is usually lumpy, some of the assumptions of traditional inventory models are not valid. For example, demand is, not constant, nor is it result of the aggregation of the independent demands of multiple sources. In short, demand variations for components are not due to random fluctuations. In addition, production and transfer lots are often not equal nor should they be equal. The result is that the economic order quantity policy sometimes belies its new.