Evaluation of Buildings Structure Alternatives Using Life-Cycle Cost Prediction Model

Nouh, Ahmed; El‐Dash, K. M.; Basiouny, M.; S. El Hadididi, Omia

doi:10.21608/bfemu.2022.235797

Evaluation of Buildings Structure Alternatives Using Life-Cycle Cost Prediction Model

Document Type : Research Studies

Authors

¹ Doctor, Department of Project Management, Civil Engineer, Benha Faculty of Engineering, Benha University, Egypt

² Department of Project Management, Civil Engineer, Shubra Faculity of Engineering, Benha University, Egypt

³ Dean Sinai Faculty of Engineering, Professor, Department of Environmental and Sanitary, Civil Engineer, Sinai University, Egypt

⁴ PhD researcher of Project Management, Civil Engineer, Benha Faculity of Engineering, Benha University, Egypt

10.21608/bfemu.2022.235797

Abstract

A life-cycle-cost (LCC) is a powerful tool used to make economic decisions for construction building. LCC is a practice of accounting for all expenditures incurred over the lifetime of a particular structure. Costs at any given time are discounted back to a fixed date, based on assumed rates of inflation and the time-value of money. This study investigates the feasibility of obtaining an accurate deep learning prediction model of building LCC by applying historical data of similar projects. The applied LCC input and output criteria are gathered from previous literature studies. The input criteria are building area, floor height, no. of floors, structure & envelope type, building age, and year of built. The output categories include the relevant costs initial cost, operating and maintenance cost, environmental impact cost, and the end of life, each of them have its criteria. An electronic questionnaire of analytical hierarchy process (AHP) is developed to weight the selected criteria to be ready for the prediction model. Only 37 responses were received from Egypt and from outside Egypt and we excluded five of them to achieve the consistency. The Deep Belief network is developed with Restricted Boltzmann machine hidden layers based on 312 training data set of input and output criteria. Three case studies are devoted to validating on the assumption modelling procedures. The probability distributions of each case study predicted outputs are investigated by using statistical regression methodology.

Keywords