Open Collections

Iannantuono, Alexander Francis

University of British Columbia

Roads continue to be an important part of global infrastructure. Mathematical models have been used to solve the vertical alignment problem in road design. In particular, the Quasi-Network Flow (QNF) model is formulated as an optimization problem, and yields a vertical alignment of least cost. However, the resulting road profiles often show signs of excessive ‘waviness’ that are unideal from a driving perspective. In this work, we propose the use of L¹ and Total Variation (TV) regularization within the QNF model as a means of providing control over such waviness. With a dataset of nineteen roads, we present a heuristic for choosing regularization parameters, and determine an appropriate choice for its use. Further, we show that regularization reduces waviness on three case studies of real roads, provided by our industrial partner. Moreover, we provide some timing results to empirically show that the regularized models take at most 40% longer to run than the original model.

Text

2022-12-25

Attribution-NonCommercial-NoDerivatives 4.0 International

http://hdl.handle.net/2429/82219

Mathematics

University of British Columbia

UBCO

http://creativecommons.org/licenses/by-nc-nd/4.0/