DEVELOPMENT AND APPLICATION OF LOOKING FORWARD SCHEDULING ALGORITHM IN SHIPBUILDING MANAGEMENT

Abstract

The purpose of the dissertation was two fold: 1. To implement and test the method of looking forward scheduling algorithm (LFSA, hereafter) in a practical context; and as a result of testing and implementation. 2. Extend the algorithm into spatial arrangement problem and making relevant comparisons with the existing methodologies such as genetic algorithms. The methodology for the study was our own indigenously developed LFSA. This was used to assist in the analysis and providing a solution to the bottleneck-forming situation between pre-erection area (PEA) and dock for the application at the Daewoo Shipbuilding and Marine Engineering Company. In the course of the design and testing with the real time block erection network and the related data several questions came up; some had a direct bearing on the applicability of LFSA and others did not. In order or to defend our claim and competence of the algorithm a computer program in the VC++ platform is developed to perform extensive tests. The issues were theoretical in nature and we have addressed them in this dissertation. They are: i) Expanding the Scope of Usage of LFSA ii) Analyzing the sensitivity and effects of LFSA on the developed schedules under various test conditions A brief discussion of each topic is as follows. Looking Forward Scheduling Algorithm In the shipbuilding industry various problems of erection are merged due to formation of bottlenecks in the block erection flow pattern. The problem gets accumulated in real-time erection at the PE area. When such a problem is encountered, a support data of the entire erection sequence should be available. Here planning is done by reasoning about the future events in order to verify the existence of a reasonable series of openings to accomplish a goal. This technique helps in achieving benefits like handling search complications, in resolving goal conflicts and anticipation of bottleneck formation well in advance so as to take necessary countermeasures and boosts the decision support system. The data is being evaluated and an anticipatory function is quite relevant in day-to-day planning operation. The system updates database with rearrangement of off-critical blocks in the erection sequence diagram. As a result of such a system, planners can foresee months ahead and can effectively make decisions regarding the control of loads on the man, machine and work flow path. Such a forecasting efficiency helps us in eliminating conventionally used backtracking related techniques. A computer program to update the database of block arrangement pattern based on this heuristic formulation is performed and its competence is argued. Spatial Scheduling A genetic algorithm (GA) is a search algorithm by [5] that attempts to emulate the evolutionary mechanism of natural biological systems where the best gene is selected for the next generation (i.e. the survival of the fittest). In the optimization problems, GA approach has been successful for generating global solutions for which traditional search techniques have not been effective and to build robust search strategy, GAs employs 'evolutionary mechanism' as described using (Fig. A). There are many severe constraint conditions in the shipyard. The primary attention in this paper is focused is into the problem domain, which is limited to the pre-erection area. A pre- erection area is defined as the area which is adjacent to the dockyard and over which the giant goliath crane runs and this area is used to place the grand blocks those are ready for the erection of the ship in the dock. The addressed major constraints become a monstrous problem for the schedulers in the shipyard namely, space based and time based. ◁그림 ▷(원문을 참조하세요) A team of many schedulers using their personal expertise using a computer graphic program without any intelligent or decisional support handles the space situation. Here various pre-defined geometrical shapes are being matched manually in the graphical interface of the program. This takes a lot of manipulation and manpower as the individual cases had to be studied. The time factor is dealt with the conventional erection sequence diagram where there exist a lot of uncertainties since in this portion only the blocks on the critical path have a fixed date of erection otherwise the remaining blocks remain uncertain. This again is handled by a team of planners to resolve using their expertise and still poses a very tedious effort on the part of the schedulers.