An Investigation of Preliminary Engineering Funding Variability and a Model to Forecast Project Level Preliminary Engineering Expenditures

Turner, Bethany, Civil Engineering - School of Engineering and Applied Science, University of Virginia
Smith, Brian, Department of Civil Engineering, University of Virginia
Miller, John, Department of Civil Engineering, University of Virginia
Parkany, A, Department of Civil Engineering, University of Virginia

For the Virginia Department of Transportation (VDOT), preliminary engineering (PE) includes most activities that occur prior to construction such as scoping, detailed design, environmental review, and advertisement. Because this definition is broader than that used by other organizations, their methodologies for forecasting PE costs are not necessarily transferrable to Virginia. This project takes a deeper look at VDOT project development and outlines the definition of PE, addresses availability issues locating PE expenditure data and develops a model to better forecast the PE cost of a construction project.
Currently, VDOT forecasts PE costs based solely on the construction estimate, and the percentage of total project costs devoted to PE is inversely proportional to this estimate. Accordingly, forecasting PE costs for smaller projects, i.e., under $5 million, merits attention.
Based on 124 projects reviewed by the researchers and DOT experts, this research develops an approach for forecasting PE costs as a function of statistically significant characteristics typically known at the project’s inception: length, duration, level of required environmental review, locally administered status and Right-Of-Way (RW). Twenty-seven projects were used to test the model and the new approach reduced the mean absolute error from about $200,000 to $110,000. This error reduction was evaluated as statistically significant (p = 0.02). Additionally, compared to the original approach, the new approach nominally reduced the mean percentage error from 135% to 47%. Although an immediate benefit is more accurate PE forecasts, these results also demonstrate the importance of providing forecasts as a range based on a statistical or empirical confidence interval rather than solely as a point estimate.

MS (Master of Science)
All rights reserved (no additional license for public reuse)
Issued Date: