تکامل یک سیستم برنامه ریزی تولید: مطالعه موردی 10 ساله

This paper describes the evolution of a production planning system (PPS) from a simple work sequence generation tool to a useful, sustained scheduling system. Three stages of evolution are described. In the first stage, a Gantt chart sequencing tool was converted to a scheduler's information system. This change was driven by the need to support the scheduler's daily task. The second stage of evolution was caused by an MRP–ERP conversion. The integration and conversion increased overhead and complexity in the job task and hence the tool, including the transformation of the previously integrated dispatching/scheduling task into separate dispatching and scheduling activities. The third stage of evolution has been small continuous improvements driven by management reporting requirements. PPS was developed in 1996 and has been fully operational since January 1997. Two major insights are discussed in this paper: the implications of supporting the scheduling task versus work sequence generation, and the software design requirements for evolutionary change as the software is used in an ever-changing situation.

Production planning systems (PPS) are a specialized form of decision support system (DSS). There are over 100 commercial systems available [1]. The purpose of a PPS is to take manufacturing requirements, match them with a model of the factory or the supply chain and, using various algorithms and technology, and craft a work sequence either automatically or with manual intervention. In this paper, the term scheduling will be reserved for what the human scheduler does as their daily job task—part of which is the creation of detailed work sequences. This differentiation is important since one of the major drivers of the software evolution described in this paper was the explicit recognition and support of the scheduling task, above and beyond the creation of detailed sequences. McKay and Wiers [2] and [7] describe the production control tasks of planning, scheduling, and dispatching. In this case study the individuals with a title of scheduler performed the tasks of scheduling and dispatching. If detailed work sequencing is focused upon, the problem appears relatively simple and many sequencing tools have been quickly built with a Gantt chart interface and some form of sequencing engine. The concentration has been on the data representation, database, Gantt chart visualization and manipulation, and on the sophisticated mathematics used to create a recommended sequence. Unfortunately, there are few success stories and there are still many challenges that remain when creating planning and logistics systems that have to work in a real factory [3], [4], [5] and [6]. While there is not one sole issue or reason for PPS success or failure [7], one potential issue relating to the failure of planning systems is the distinction between generating a sequence and scheduling and the lack of information system support for the latter. This was the first major insight obtained from this longitudinal case study. Another possible reason for system failure is the rigid structure or inability to easily adapt to changing requirements in the factory environment—both information system and organization changes. Learning what aspects of the software should be easily changed was the second major insight. This case study will describe three stages of evolution are described: • moving from a basic work sequence generation tool to a scheduler's information system—driven by the need to support the scheduler's task; • moving from a legacy MRP system to ERP (SAP)—driven by changes in the information system infrastructure; • continuous evolution—driven by management and organizational requirements. The first evolution occurred during a 4-month period following the initial live test of the software in the Fall of 1996. This evolution saw the code triple in size and the functionality more than double. This basic system was then used for approximately 1.5 years before the ERP migration. When the plant converted to the ERP, the scheduler's tool doubled in complexity to deal with the added functionality required by the ERP interface and associated task changes. The decision support system continues to be used and has been fully operational and in daily use for over 10 years. The factory using the software has been participating in ongoing research on production control throughout this time period. This relationship has presented a relatively unique research opportunity to research, develop, deploy, and evolve a live decision support system over an extended time horizon. The first phase of evolution is perhaps of the most interest to researchers who are contemplating the creation of decision support tools for a scheduler or dispatcher task. The lessons and insights gained relate to the understanding of the task and software architecture. The second phase of evolution is more interesting to individuals integrating existing production control tools with legacy or ERP systems. The insights in this context relate more to the changes and impositions created by the larger and more rigid systems. The third phase of evolution provides lessons to both audiences—the need for change and adaptation to the organizational demands as a plant itself evolves. This paper is organized as follows. Section 2 will present a review of DSS evolutionary literature followed by specific discussion of case studies addressing evolution in the production planning and scheduling field. Section 3 will provide background information for the case study. Section 4 will discuss the system development processes and underlying architecture. Sections 5, 6 and 7 will discuss in-depth each of the three evolutionary stages. Section 8 will discuss and summarize the results. Section 9 will present some concluding thoughts.

Decision support tools for planning and scheduling are complicated entities. They need to ‘fit’ the scheduler's tasks and support information processing beyond simple sequence generation. Understanding the impact of frequency, accuracy and scope changes on the task and the resulting disruptions in the current scheduling process appears to be important in the development and deployment of the decision support systems. The introduction of added functionality can also force a task to outgrow a single individual. However, since the tasks of planning and control cannot be totally independent among individuals, a task sharing structure might also be required and this should be considered in any system design. It is not obvious from the literature if the current generation of scheduling and planning tools (finite capacity schedulers and advanced planning and scheduling systems) address these issues. From reviews and experiences with a number of the commercial systems, it still appears that the focus is on the sequencing and creating the initial plan and not on the actual tasks of planning, scheduling, and dispatching [2]. These are clearly areas for future research. We have been able to build and probe in one situation, but the lessons are not necessarily generic or prescriptive. A flow-shop tool was created for the assembly area of the factory and the user-interface and functionality suite are almost 100% different from the job shop tool. An agile development approach was also used and similar results obtained—excellent user adoption, robust tool, and rapid delivery of a production quality system. A significant difference is that the flow-shop area functions in a focused-factory fashion with one individual planning, scheduling, and dispatching. While not discussed in this paper, the contrasts between flow and job shops, and hierarchical and focused factory schemas are also suggested as areas for fruitful research. Solid taxonomies and guiding principles are required. Finally, the impact of an ERP implementation on production planning and control was also observed in this case study. It was observed that although ERP systems can yield substantial benefits in many parts of the organization, it is possible in certain situations that the implementation can lead to additional personnel requirements within production control.