Cost Analysis of Construction Using the Cost Significant Model and Artificial Neural Network or Feasibility Study Cost Estimation in Process Industry (Case Study: Feasibility Study Projects in the Process Industry Sector in Indonesia)

Abstract

The process industry plays a vital role in the global economy, with increasing complexities requiring significant capital investment for projects. Accurate cost estimation in the early stages is critical to avoid cost overruns, delays, and poor quality. During the feasibility study phase, accurate cost estimation is crucial for project success. This study employs the cost significant model (CSM) to identify key cost elements influencing the total construction cost and compares the model's accuracy with actual costs in 50 process industry projects in Indonesia from 2010 to 2024. Artificial Neural Networks (ANN) with backpropagation are also used to validate the model and calculate the Mean Absolute Percentage Error (MAPE) for accuracy assessment. Analysis identifies significant cost components such as Mechanical ISBL (55.89%), Mechanical OSBL (7.65%), Electrical (6.06%), and Superstructure (15.07%), collectively contributing 84.66% of total costs. ANN-ALL achieves the highest accuracy (L: -4.76%, H: 0.55%), followed by CSM Model 3, ANN-CSM, and others, as per AACE International guidelines. Models exceeding Class 4 benchmarks are unsuitable for feasibility studies. Data analysis combines regression using SPSS and ANN modeling with MATLAB, highlighting the ANN-ALL model's superiority in cost estimation accuracy.