A Stochastic Three-Dimensional Cost Estimation System for Hot Mix Asphalt in the State of Alabama

Abstract

Asphalt is one of the most critical commodities for the American infrastructure. It is used as a paving material on the majority of the U.S. roadways. The National Asphalt Paving Association (NAPA) estimates that about 94 percent of all paved roads in the U.S. are surfaced with asphalt (NAPA 2018). The relevance of this paving material is not different in the state of Alabama, where this material is used in about 98 percent of the paved roads (AAPA 2016). It has resulted in the purchase of Hot Mix Asphalt (HMA) consuming over 40 percent of the Alabama Department of Transportation (ALDOT) annual construction budget, becoming an item that requires special attention during resource allocation and cost estimating procedures. Therefore, any improvements in ALDOT’s HMA cost estimating procedures is expected to be reflected in better budget control and a more effective use of ALDOT’s limited available resources. ALDOT’s available resources have become increasingly scarce in recent decades while there is an increasing demand for transportation infrastructure rehabilitation and improvements. This is what motivated the research presented in this thesis. To improve the effectiveness of ALDOT’s HMA cost estimating practices, this research proposes a three-dimensional HMA cost estimating system based on historical bid data from previous projects awarded by ALDOT. This system was the result of the assessment of five factors and the modeling of their impacts on HMA prices in Alabama. The five factors are: 1) project scale, 2) time, 3) geographic location, 4) estimating uncertainty, and 5) level of competition. The first three factors correspond to the “three dimensions” of the system, while the fourth factor was considered by designing the system to generate stochastic HMA cost estimates in lieu of the traditional deterministic estimates. The impact of level of competition on HMA prices was evaluated in this study, but no significant variations on this factor were found across the state, so this factor was not incorporated into the proposed cost estimating system.

The proposed HMA cost estimating system was developed, and its effectiveness demonstrated, through its application to a case study item and using historical bid data from 3,661 projects awarded by ALDOT between 2006 and 2016. The system integrates a number of elements and quantitative procedures including non-linear regression, a construction cost indexing system (CCIS), a location cost index (LCI), and the stochastic analysis of the expected performance of the system. Finally, the effectiveness of the system is demonstrated using an innovative Moving-Window Cross Validation (MWCV) approach and via statistical testing to assess the level of significance of the estimating improvement offered by each of the three dimensions. The results of the validation process revealed that each dimension would significantly improve cost estimating effectiveness for the case study item, with an overall improvement of 47.7% and 43.6% in accuracy and reliability, respectively.