Preliminary engineering (PE) for a highway project encompasses two efforts: planning to minimize the physical, social, and human environmental impacts of projects and engineering design to deliver the best alternative. PE efforts begin years in advance of the project’s construction letting, often five years or more. An efficient and accurate method to estimate PE costs would benefit transportation departments. Typically, departments estimate PE costs as a percentage of construction costs disregarding other project-specific parameters. By analyzing 461 North Carolina Department of Transportation bridge projects and 188 roadway projects let between 2001 through 2009, the research team developed statistical models linking variation in PE costs and PE duration with distinctive project parameters. The development of a user interface application aids agency users in executing the models to predict a project's PE cost ratio. Modeling strategies included multiple linear regression, hierarchical linear models, Dirichlet process linear models, and multilevel Dirichlet process linear models (MDPLM). The 461 bridge projects exhibited a mean PE cost ratio of 27.8% (ratio of PE cost over estimated construction cost) and a mean PE duration of 66.1 months. Mean PE cost ratio for the 188 roadway projects was 11.7% with a mean PE duration of 55.1 months. Project parameters utilized in the predictive models included project scope classification such as widening or new location, dimensional variables (project length, structure length, detour length, and number of spans); geographical region; and estimated costs for construction and right of way. The MDPLM minimized the mean absolute prediction error for bridges’ PE cost ratio, but interpretation of variable effects and sensitivity is difficult because of the multilevel structure. Regression modeling results are also reported since sensitivity interpretation from them is more direct.
