Electricity Consumption Prediction during Ship Construction: A Machine Learning Approach

Abstract

In the maritime industry, estimating electricity consumption during ship construction is pivotal for establishing budgets and ensuring cost-effective project management and resource allocation. The research aims to develop a machine learning model that enables shipbuilders to estimate electricity consumption during the preliminary design phase of various vessel types, including container ships, bulk carriers, passenger vessels and oil tankers. The methodology involved calculating the welded area and volume of different sections of ship structure using specific drawings for each vessel type. Subsequently, the weight of welded metal was determined from the volume, considering the density of the metal. This was followed by calculating the required electrode consumption based on the weight of welded metal and the deposition rate. Finally, electricity consumption was calculated based on the electrode requirements for each ship. Leveraging advanced machine learning techniques, a linear regression model is constructed, utilizing multiple variables such as ship length, breadth, depth, and ship’s type to establish a predictive relationship with electricity consumption. The model is cross-checked with field data and found consistent result with industry standard. It is hoped that this model will empower shipbuilders with a predictive tool during the preliminary design phase, aiding in estimating electricity consumption costs. By integrating this model into the shipbuilding process, stakeholders can proactively devise budgets and allocate resources more efficiently, thus optimizing the overall construction endeavor.