Transportation project cost estimation is the process of forecasting the total cost of a project. Decision makers need accurate cost estimates, to allocate resources across projects efficiently. However, the transportation sector is notorious for project costs exceeding original estimates. Improved transportation cost estimation procedures allow for a more efficient and sustainable allocation of financial resources.
AASHTOWare Project Estimation is the primary cost estimation software used by several state Departments of Transportation. AASHTOWare Project Estimator allows users to enter the items used in a transportation project and the quantity of each item used. AASHTOWare then estimates a unit price for each item based upon historic data derived from previous transportation projects in the state that have used those items. AASHTOWare also generates 94.5% confidence intervals for each item unit cost estimate. However, previous research has demonstrated that these confidence intervals are derived under the assumption that the underlying costs are normally distributed, rather than using the actual cost distributions present in the underlying data (Reichard, et al., 2021). AASHTOWare also does not currently report uncertainty for the total project cost estimate. Decision makers would benefit from having access to uncertainty for individual cost components, as well as the total project cost estimate. Knowledge of total project cost uncertainty would allow planners to hold some funding in reserve, to insure against potential cost overruns, or release funds that are not needed on reserve.
This research proposes to use bootstrap analysis to derive confidence intervals for each item in the AASHTOWare database (Efron and Tibshirani, 1994; Efron and Tibshirani, 1986). Then, Monte Carlo simulation will be employed to estimate the probability distribution of total expected project costs, based upon the variability of item costs (Guensler and Leonard, 1995). With this addition, AASHTOWare would be able to report the 5th and 95th (or 10th and 90th) percentile cost estimates, along with the expected cost of a project to decision makers with resource allocation. The use of asymmetrical bootstrap confidence intervals for item costs within the model would better reflect the actual distribution of item costs, as well as the combined effect for the entire project.
Previous research has also shown that a relatively small subset of project items can be responsible for the vast majority of total project uncertainty (Reichard, et al., 2021). Items that have a large degree of component cost uncertainty and that are used in large quantities may be responsible for several million dollars in total project cost uncertainty. This research will also assess the potential impact of high-use, high-cost-volatility items on total costs across project types. The analyses will reveal which items are the greatest contributors to total project cost uncertainty and should assist in future policy efforts designed to control costs (perhaps through long-term contracts designed to reduce item cost volatility), thereby reducing transportation project cost uncertainty. The findings generated by this research will improve sustainable transportation lifecycle cost estimation, allowing for funding to be used more efficiently and to maximize benefits per capital dollar spent.