Summary:

The primary objectives of this research were to: 1) develop an efficient and accurate method for bridge repair and replacement cost estimating; and 2) implement the developed method in a cost estimation worksheet; 3) provide an automated system that allows periodic updating of costs. To accomplish these objectives, several modeling approaches were investigated, including simple linear regression, multiple linear regression, lasso regression, and two versions of case-based reasoning (CBR). Overall, it was found that all approaches could produce more accurate cost predictions for individual pay items than the existing MDOT engineering estimate, and that lasso regression was most effective. Next, cost estimation models were developed for the bridge work types currently used by MDOT. An approach was developed to identify work types within the cost database using primary pay items, and the total work type cost within a project was determined by summing all pay item costs associated with that work type. Several different work type cost approaches were developed: base, supplemental, substitute, and cost average approaches. Both average unit costs and regressed costs were provided. A time-series based MHCCI projection method was developed to consider future cost projections. A cost estimation spreadsheet and automated updating program were developed.

Publications: