Probabilistic Life-Cycle Cost Analysis of Pavements

Abstract

Significant research has been conducted over the past decade to enhance probabilistic pavement life-cycle cost analysis (LCCA) models, yet the drawing of broad conclusions from different studies is difficult because of the significant variation in scope and sources of uncertainty. Specifically, these two issues make it difficult to infer from existing research (a) which parameters are significant contributors to uncertainty for a specific context and (b) how context affects the analysis. The goal of this research was to address this problem by implementing an LCCA model in a range of scenarios that vary in location, traffic conditions, design life (e.g., year to first rehabilitation), analysis period, maintenance schedule, and discount rate. Results from the analysis indicated that, in relation to the drivers of variation, uncertainty about initial cost was the principal driver of variation across the case studies. Other parameters, such as the predicted performance of pavement over time, could also be important drivers of variation and in particular a matter for lower-volume roads, for which thinner pavement designs are used. In terms of contextual decisions, some decisions, such as whether to use mechanistic–empirical pavement designs instead of the paving design manual of a state department of transportation to determine future maintenance events, seem to have a larger impact than do others. For example, analysis period and design life, though important, affect the final results significantly less, although in some instances they can play a role in differentiating between alternative designs.

Probabilistic Characterization of Life-Cycle Agency and User Costs: Case Study of Minnesota

“…The probabilistic LCCA model used in this study follows that of Swei et al (7,13) except that the scope of the analysis now incorporates user costs. The approach accounts for uncertainty not only as it relates to construction and user costs described pre viously, but also other relevant inputs that underlie the model.…”

“…Previous studies that have focused on the cost for pay items in pavement projects include Sanders et al (16), who developed a series of bivariate regression models between bid unit price and quantity, and Shrestha et al (17), where a similar analysis was conducted to characterize the degree of dependence between the two. Furthermore, an even smaller subset of the LCCA literature has considered the evolution of the cost for construction activities over time, with the only known example being the work of Swei et al (7). Several studies have accounted for user costs associated with alternative investment strategies (6,18).…”

“…Several studies have demonstrated that estimates of the unit price of pay items for an LCCA influence the accompanying results (7). For the most part, existing research typically develops representative parametric estimates of the unit price of relevant pay items by fitting available data with a best-fit distribution (14).…”

“…Despite this progress, several opportunities remain to augment the current LCCA frameworks utilized by planners. In particular, a previous study by Swei et al demonstrated that estimates of the unit price of pay items within an LCCA are a major contributor to both total agency expected life-cycle costs (LCC) and variation across a series of case studies (7). Therefore, it is of paramount importance that estimates of expected construction costs as well as associated variation that enter an LCCA appropriately represent available empirical data.…”

Probabilistic life cycle cash flow forecasting with price uncertainty following a geometric Brownian motion

“…Other researchers indeed observe the lag in attention for price uncertainty in construction and engineering. Swei, Gregory, and Kirchain (2015) and Swei, Gregory, and Kirchain (2017) explicitly state that price variations for asphalt have been ignored in scholarly research. As a reason the authors put forward a lack of empirical data and as such unfamiliarity with the application of econometric forecasting models.…”