Chapter 8 Conclusions and Future Work
8.1 Conclusions
Uncertainty is one of root causes of the production planning inaccuracy. This study has proved that the underlying causes of uncertainty such as varying processing time, machine failure, etc have considerable implication on the master production schedule (MPS). Ignoring them may lead to incorrect decision and ultimately make poor production and delivery performance. Through simulation approach, this study has been successfully able to estimate the effect of uncertainty and subsequently attempts to diminish its effect on the proposed MPS. Concerning to the development of valid and realistic MPS, this research has brought us to the valuable conclusions as follows.
1. The fuzzy multi-objective linear programming (FMOLP) can be surely employed to solve the MPS problem. Moreover, with certain parameter configurations, the FMOLP is able to yield better achievement level of objectives than non-fuzzy solution. The FMOLP approach is recommended, particularly, for a case where the objective functions are imprecise and can only be stated within certain aspiration level.
2. The FMOLP has ability to determine intelligently how much, when, and where the additional capacities (overtimes) are required such that the inventory can be reduced without affecting customer service level. This study has also shown that the overtime profile fluctuates in parallel with the customer demand variation. This highlights that the FMOLP is also suitable for handling the MPS creation with high varying customer demand.
3. In term of MPS development, the inventory objective contradicts with customer satisfaction and inventory target objective . The increasing of achievement level of inventory objective is balanced with the poor performance of customer satisfaction and inventory target objectives (Figure 16).
4. The genetic algorithm can solve efficiently the crisp single objective model equivalent to the FMOLP. This research reinforces the Soares’s finding (2008) that the GA can be applied to real industrial master planning problems.
5. Concerning to solving the MPS problem, the genetic algorithm is more efficient in term of processing time than the differential evolution (DE) algorithm. However, in term of overall degree achievement, the differential evolution is more reliable than genetic algorithm. 6. Through iteration process, the closed-loop MPS can bridge the drawback of approach
proposed by Kochhar (1998), Knowledge-based system approach, which does not take into account unpredictable event (uncertainty) in manufacturing system (execution level). The similar framework may also be employed to improve other production planning functions. 7. The proposed information system, which is integration of three systems (Dosimis-3, SAP R/3
and MATLAB), enables us to realize the closed-loop planning concept. SAP eases us to implement production planning logic such as SOP, demand management and MRP logic. MATLAB helps us to apply intelligence optimization technique and analyze easily the statistic data, while Dosimis-3 enable us to simulate the execution of planned orders. However, the IT expert skill is required to integrate those systems in professional and reliable manner if this methodology wants be implemented for real industrial case.
8. This study reveals that Dosimis-3 can be potentially used to enhance the functionality of Production Planning and Control of SAP because both of them have the similar mechanism for operation execution. The operation in SAP is triggered by a production order (PO), while the operation in Dosimis-3 can be driven by a work plan. If the PO and work plan can be mapped properly, then the production planning generated by SAP can be verified before the plan is released to real manufacturing system.
9. Using discrete event simulation, it can be proved that the underlying causes of uncertainty such as varying processing time and machine failure have considerable implication for the production lead time. In general, the uncertainty make the actual lead time be longer than the planned lead time. The higher the degree of uncertainty is, the larger discrepancy is found between the actual and planned lead time.
10. Through the simulation of POR execution, this study also reveals that the underlying cause of uncertainty should be taken into account as MPS development. Ignoring the uncertainty lead to the unrealistic MPS which is indicated by excessive unplanned overtime.
11. The load leveling technique using fuzzy multi-objective linear programming (FMOLP) seems working effectively to determine the quantity of material should be shifted from over- utilized resource to under-utilized resource. In addition, the approach for selection of under- and over-utilized resources, which is divided into four optimization levels, seems quite reliable to reduce the unplanned overtime without obstructing too much other criteria. 12. Using the propose load leveling technique, it is found that moving some amount of load quantities to previous period can keep the tentative MPS be valid and realistic under uncertain environment. It implies that if the manufacturing environment is quite volatile, it is suggested to produce material in advance as long as the capacity is available. However, one must note that producing materials in earlier period than they are consumed will increase inventory cost.
13. Last but not least, in general the closed-loop MPS, which incorporates fuzzy multi-objective linear programming (FMOLP) and simulation approach, is able definitely to develop valid and realistic MPS under uncertain environment. However, it is admitted that this methodology may be not suitable for stable environment in which uncertainty events are insignificant and ignorable.