DAS: Intelligent Scheduling Systems for Shipbuildiug

Daewoo Shipbuilding Company, one of the largest shipbuilders in the world, has been much troubled with planning and scheduling its production process for a long time since its establishment. To solve the problems, Korea Advanced Institute of Science and Technology (KAIST) and Daewoo have been jointly performing the DAS (DAewoo Shipbuilding Scheduling) Project for three years from 1991 to 1993. For integrating the scheduling expert systems for shipbuilding, we employed the hierarchical scheduling architecture. To automate dynamic spatial layout of objects in various areas of the shipyard, we developed spatial scheduling expert systems. For reliable estimation of man-hour requirements, we implemented the neural network based man-hour estimator. In addition, we developed the Panelled Block Assembly Shop Scheduler and the Long Range Production Planner. For this large-scale project, we devised the phased development strategy consisting of three phases such as vision revelation, data dependent realization, and prospective enhancement. The DAS systems successfully launched in January 1994 and are being actively used as indispensable systems in the shipyard resulting in signiticant improvement in productivity and reporting visible and positive impacts in many ways. Problem Description: The Company’s History Struggling with Scheduling Daewoo Shipbuilding Company is one of the largest shipbuilders in the world. Its main products are VLCC (Very Large Crude Oil Carrier) and container ship. It also involves heavy machinery, plant and automobiles that make minor contribution to the company sales. Its shipbuilding yard has three docks. One is the largest dock in the world capable of manufacturing one million ton VLCC and has a Goliath crane which can hold up to 900 tons. Employees are twelve thousand and the sales in 1993 were 2 billion dollars. Shipbuilding is the make-to-order manufacturing and takes long time to complete the products. It includes so many complicated processes from an order to the final delivery and requires large amount of working capital. It is very difficult to manage the resources including material, human, facilities and information. Therefore, the scheduling and control of shipbuilding plant is a very complex task and a nightmare. The company has struggled for an effective scheduling to achieve total optimization since its establishment. Poor scheduling keeps workers waiting for the prerequisite subassemblies, causes fluctuation of work loads resulting in expensive overtime works, and causes delay in delivery. The company has attempted various project management software such as PROJACS, VISION and X-PERT as well as in-house development with conventional programs, but all these efforts failed because they couldn’t reflect scheduler’s knowledge flexibly considering the complicated interrelated factors. In addition, no software could support the dynamic spatial layout even though the spatial resources with material handling equipment like cranes are bottlenecked resources. To develop the integrated scheduling expert systems overcoming these problems, KAIST and Daewoo have been jointly performing the DAS (DAewoo Shipbuilding Scheduling Expert Systems) Project for three years from 1991 to 1993. The DAS systems successfully launched in January 1994 and have been used as indispensable systems in the shipyard resulting in the significant improvement in productivity and reporting the visible and positive impact in many ways. Application Description The DAS project adopted various expert systems technologies to cope with the scheduling task. Key sub-systems in DAS project are as follows. DAS-ERECT: Erection Scheduler at Docks DAS-CURVE: Curved Block Assembly Shop Scheduler DAS-PANEL: Panelled Block Assembly Shop Scheduler DAS-MH: Neural Network based Man-hour Estimator 90 IAAI-95 From: IAAI-95 Proceedings. Copyright © 1995, AAAI (www.aaai.org). All rights reserved. DAS-LPP: Long-term Production Planner DAS-ERECT: Erection Scheduler at Docks The initial shipbuilding process begins with block division. Usually, a large tanker is divided into hundreds of sized blocks as shown in Figure 1. These can be categorized into two types: flat-bottomed, which are assembled in the Panelled Block Assembly Shop(PBS), and curved-bottomed, which are assembled in the Curved Block Assembly Shop(CBS). Multiple blocks may be assembled into larger ones (called super-blocks) in the Pre-Erection Shop(PES) to reduce the assembly time needed at the main erection dock. At the main docks a number of blocks and super-blocks are constructed (‘erected’) using a very large Goliath crane. In a shipbuilding plant, there are usually one or two dry docks for erection and corresponding pre-erection shops as shown in Figure 2. Figure 1. A Ship and its Partition into Blocks The important characteristics of erection scheduling include: Sequential Erection at Each Dock: Blocks and super-blocks in a dock have to be erected one at a time because there is only one Goliath crane at each dock. However, there may be multiple docks in some plants. Large Search Space: Since the schedulers have to consider multiple ships, each composed of 400 500 blocks, a manual search for the best schedule is beyond mental processing capacity. Thus, there is a need to develop a proper search method. Technical Knowledge for Erection Sequencing: The technical knowledge which restricts the erection sequences is available. Utilization of Resources: Key resources in shipbuilding comprise human workforces, cranes and space. To utilize these resources effectively, a schedule should be established to balance the loads among PBS, CBS, Pre-erection shops and docks. Hierarchical System Architecture for the Shipbuilding Scheduling For erection scheduling at the dock, we employed the hierarchical scheduling architecture (Lee, Suh, & Fox 1993) and developed the constraint directed graph search technique (Lee et al. 1994). In the hierarchical architecture, detailed schedules for the lower level assembly plants are delegated to the individual plant’s schedulers (DAS-CURVE and DAS-PANEL in this case) as far as the requirements from the higher level scheduler (DAS-ERECT) are satisfied(Figure 3). However, if the lower level scheduling is impossible, the higher level scheduler attempts to adjust the original requirements. In the constraint directed graph search, we amalgamated the notions of graph expansion and constraint directed pruning into an algorithm.