Optimizing earthwork estimating for highway construction

The purpose of this research was to advance the state-of-the-art in earthwork estimating of highway projects. The thesis concentrated on the uncertainties associated with calculating the quantity of rock in cut areas, estimating the fleet production, and selecting the optimum distribution of material between cut and fill areas. Estimation of the quantity of rock was achieved by using chance-constrained linear programming--a technique that allows the user to transform stochastic quantity constraints into deterministic ones. The variability in fleet production was accounted for by using three-value, PERT-type estimates for each of the following components of earthwork: (1) excavation (including loading), (2) hauling, and (3) compaction (including unloading). The user further defines his estimates by selecting a "confidence factor" that represents the probability of not exceeding the target (or middle) value of the three-value estimate. A standard linear programming (LP) formulation, modified by the chance-constraints, was used to determine the optimum cut/fill distribution as well as the most efficient location for waste and borrow sites. The proposed system can be summarized in four steps. First, the user inputs cost data for each section (such as 1000-foot intervals) of roadway. Values (three-values for a probabilistic or a single-value for a deterministic estimate) are entered for excavation, haul, and compaction costs. The second step is the LP formulation including the chance-constrained rock quantities. The third step involves simulating the cost coefficients (determined in step one) resulting from the LP solution in order to produce a cost range. The fourth and final step is a comparison of the cost ranges determined in step three with the cost ranges provided by normal LP sensitivity analysis in step two. The system was applied to a highway project in Pennsylvania, and the proposed system was compared with other more traditional estimating methods. The conclusions reflect the fact that, although probabilistic estimating is in its infancy, it has great potential for reducing the risk and increasing the profits of earthwork contractors.