Forecasting the Construction Expenditure Cash Flow for Transportation Design-Build Projects with a Case-Based Reasoning Model

Abstract

Accurately forecasting the construction expenditure cash flow of transportation projects is critical for state departments of transportation (state DOTs) to secure sufficient funding to cover their fiscal obligations throughout the project development timeline. However, there is no quantitative model to assist state DOTs in accurately forecasting expenditure cash flows for design-build projects. At the outset of awarding a typical design-build contract, the design has not been finalized to enable exact quantities for a detailed cost estimate. This issue represents a big difference between design-build and design-bid-build projects that makes estimating the project payouts more difficult for design-build projects. This research for the first time creates an expenditure cash-flow forecasting model for transportation design-build projects based on case-based reasoning and a genetic algorithm. The model utilizes information about project-specific characteristics and external market factors. The applicability of the proposed model is examined on a data set of 33 transportation design-build projects delivered by Georgia Department of Transportation (GDOT) from April 2007 to January 2020. The results show great accuracy in forecasting expenditure cash flows of these projects. The major contribution of this study lies on the creation of a new forecasting model, which enables reasonably accurate prediction of expenditure cash flow of transportation design-build projects. This research identifies that even early at the procurement phase of a design-build project when exact quantities and detailed cost estimates have not been fully developed, the combination of conceptual project information and local construction market indicators offers the capability to predict the future expenditure cash flow of the project through establishing similarities between the project to be awarded and historical design-build projects. This research provides a novel approach to quantify the similarities that will be used as critical inputs into a case-based reasoning algorithm for predicting the expenditure cash flow of the project using expenditure records of most similar historical projects in the design-build data set. It is anticipated that transportation agencies can benefit from the forecasting model presented in this study by enhancing their processes of estimating their financial obligations on the onset of letting design-build contracts. The forecasting model will help transportation agencies to avoid underestimating and overestimating the capital needed to build a design-build project during the contract duration. Therefore, limited financial resources of transportation government agencies will be utilized more efficiently and effectively, and the likelihood of running into disputes for fund unavailability and cost overruns will be reduced.