Clustering risk assessment method for shipbuilding industry

Abstract

Purpose

The purpose of this paper is to develop a risk assessment method for production processes of large-size steel ship hulls.

Design/methodology/approach

This study uses a quantitative-probabilistic approach with involvement of clustering technique in order to analyse the database of accidents and predict the process risk. The case-based reasoning is used in here. A set of technological hazard classes as a basis for analysing the similarities between the production processes is proposed. The method has been explained using a case study on large-size shipyard.

Findings

Statistical and clustering approach ensures effective risk managing in shipbuilding process designing. Results show that by selection of adequate number of clusters in the database, the quality of predictions can be controlled.

Research limitations/implications

The suggested k-means method using the Euclidean distance measure is initial approach. Testing the other distance measures and consideration of fuzzy clustering method is desirable in the future. The analysis in the case study is simplified. The use of the method according to prediction of risk related to loss of health or life among people exposed to the hazards is presented.

Practical implications

The risk index allows to compare the processes in terms of security, as well as provide significant information at the technology design stage of production task.

Originality/value

There are no studies on quantitative methods developed specifically for managing risks in shipbuilding processes. Proposed list of technological hazard classes allows to utilize database of past processes accidents in risk prediction. The clustering method of analysing the database is agile thanks to the number of clusters parameter. The case study basing on actual data from the real shipyard constitutes additional value of the paper.

Citation

Romuald Iwańkowicz, R. and Rosochacki, W. (2014), "Clustering risk assessment method for shipbuilding industry", Industrial Management & Data Systems, Vol. 114 No. 9, pp. 1499-1518. https://doi.org/10.1108/IMDS-06-2014-0193