Integrated regression and fuzzy model to determine remaining project duration

• Main Author: Scully, Clement Cornelius
• Format: Thesis
• Language: English
• Published: 2022
• Subjects: TJ Mechanical engineering and machinery
• Online Access: http://eprints.utm.my/id/eprint/102137/1/ClementCorneliusScullyPSKM2022.pdf.pdf; http://eprints.utm.my/id/eprint/102137/; http://dms.library.utm.my:8080/vital/access/manager/Repository/vital:149092

Many attempts have been made by previous researchers to improve the monitoring process in engineering phase of a Floating Production Storage and Offloading (FPSO) conversion project. Earned Value Method (EVM) has been used to monitor the engineering phase progress towards the completion of project with the emphasis on cost monitoring, but it lacks the ability to capture the changes of expected project remaining duration based on changes in engineering deliverables and manhours performance. This study aimed to analyse the correlation of engineering phase monitoring variables to provide a visual projection of expected project duration. A model was developed using fuzzy method to produce a surface plot of engineering phase remaining duration. Stepwise regression selection was done on sets of variables combinations of Master Deliverable Register, manhours and progress by comparing the adjusted R2 values for each set. Regression coefficients from the selected set were extracted to form an algorithm to be implemented into fuzzy membership rule. Triangular membership is selected with each variables having seven linguistic terms ranging from ‘Minimum’ to ‘Maximum’. Sensitivity analysis was conducted on the remaining duration plot to determine the impact of linguistic terms towards the extent changes of remaining duration. The model was validated by comparing it with the remaining duration determined using the EVM. It was found that the contour on fuzzy surface plot provides a visual forecast on expected project delays when a plateau is formed where events such as correction of documentations, manhours redundancies, re-work, restriction of manhours, resource levelling, review cycles, delay of Vendor Data Information and clashes with construction site are predicted. The remaining duration from the fuzzy model showed an overall improvement of accuracy when compared with calculated remaining duration from EVM method with 9% lower Mean Absolute Percentage Error and an average of Root Mean Square Error of 18 days as compared to EVM error of 31 days. It can be concluded that fuzzy surface plot enables prediction of the remaining duration and project stagnations from analysis on the surface contours and plateaus. Thus, the proposed model in this study serves as an alternative technique of top-down method for determining the remaining duration.