Design-Bid-Build (DBB) projects are procured by government agencies typically through thecompetitive bidding process. The decision of the contracting authorities regarding which projectsproceed to the bidding stage depends, in part, upon the early estimates of probable cost. Effortsare made to make this estimate as realistic as possible. Irrespective of the estimate of probablecost, the actual project cost is established by the amount of the winning bid and the cost ofchange orders during construction phase. The change order costs generally are known during theconstruction phase of the projects. However, the bid cost of projects can be estimated byanalyzing the bid data of historical projects. This study will develop a method to predict thefuture projects’ bid cost of unit price items based on quantities of items. The regression modelsfor various unit price items will be developed by analyzing historical bid data of 151 DBB roadprojects undertaken by the Clark County Department of Public Works in southern Nevada from1991 through 2008. The total value of construction was equivalent to $841 million whenconverted into a June, 2011 base cost. Statistical models were developed to improve themethodologies for estimating bid-item unit pricing and to reduce variances that result in largediscrepancies between engineers’ estimates and actual bid-award amounts. These regressionmodels can be used to predict the actual bid cost of the unit price items based on quantity of theunit items.
展开▼
