| In the design and planning process for multilane highways, the two-way combined design hour volume (DHV) is converted to a directional design hour volume (DDHV) by multiplying the DHV with the directional distribution (D). Lack of precision in the use of a DDHV value could compromise the accuracy of the design and operational analysis. This study is concerned mainly with the estimation of the DDHV from sample traffic counts.; The methodology of DDHV prediction in this thesis treats the two directional flows at a site as two different streams. The DDHV models involving such independent variables as used in the traditional DHV models indicates that they are unable to provide good prediction results. However, a more successful DDHV model is proposed in this thesis. It consists of ranking the highest hourly volumes in a selected month for the sample traffic counter and comparing them with those of all the available permanent traffic counters (PTCs) in the same month. A mean squared error (MSE) is obtained between the sample counter and each PTC. The PTC with the least MSE value is assumed to represent the traffic variation pattern of the sample counter, and its traffic parameters are thus used to estimate the required DDHV. The application of the proposed approach to the study sites in the thesis indicates that it can produce the same statistical accuracy of the DDHV estimates as obtained for the traditional DHV estimates. By using the method, an analyst can avoid the application of directional split, and a major source of error is thus eliminated from DDHV estimation. (Abstract shortened by UMI.)... |
