Mining transportation logs for understanding the after-assembly block manufacturing process in the shipbuilding industry

In the after-assembly block manufacturing process in the shipbuilding industry, domain experts or industrial managers have the following questions regarding the first step in terms of reducing the overhead transportation cost due to irregularities not defined in a process design: “What tasks are bottlenecks?” and “How long do the blocks remain waiting in stockyards?” We provide the answers to these two questions. In the process mining framework, we propose a method automatically extracting the most frequent task flows from transport usage histories. Considering characteristics of our application, we use a clustering technique to identify heterogeneous groups of process instances, and then derive a process model independently by group. Process models extracted from real-world transportation logs, are verified by domain experts and labelled based on their interpretations. Consequently, we conceptualize the “standard process” from one global process model. Moreover, local models derived from groups of process instances reflect unknown context regarding characteristics of blocks. Our proposed method can provide conceptualized process models and process (or waiting in stockyards) times as a performance indicator. Providing reasonable answers to their questions, it helps domain experts better understand and manage the actual process. With the extension of the conventional methodology for our application problem, the main contributions of this research are that our proposed approach provides insight into the after-assembly block manufacturing process, and describes the first step for reducing transportation costs.

► The overhead cost due to irregularity is critical in the shipbuilding industry.
► We propose a method to extract a process model from transport usage histories.
► The method is based on clustering and association rule techniques.
► It captures irregular workflows, commonly occurring but not planned.
► It helps domain experts better understand and manage the process.