The current cost of construction for new roads is enormous – as an example, the new Asia Pacific Gateway and Corridor Initiative, the 40-km four-lane road connecting B.C.’s Deltaport to Highway 1, is targeted at $1.5-billion. New research coming out of the Centre for Optimization, Convex Analysis, and Nonsmooth Analysis (COCANA) at UBC’s Okanagan campus combines mathematics, computer science, and civil engineering to transform the process of designing new roads and potentially vastly reduce their construction costs.
Fifty years ago, road design consisted of engineers surveying the land, drawing an accurate map, and selecting a route. The onset of personal computers led to techniques that separated road design into three interconnected levels: horizontal alignment, vertical alignment, and earthwork. Horizontal alignment explores a satellite view of the potential site and selects a path for the new road. Having fixed the horizontal alignment, vertical alignment takes a slice through the actual landscape and seeks to generate a target height at each location on the future road. Finally, earthwork examines the current state of the terrain to determine a method to rearrange the earth into a smooth surface with the desired vertical alignment. Once all three steps are addressed, an approximate construction cost for the new road can be calculated.
The goal is to determine a design that will minimize cost while taking safety and environmental constraints into account. This is a highly repetitive process requiring many hours of skilled labour and does not provide any guarantee on the quality of the final design.
New research from UBC is poised to revolutionize road design. Working in collaboration with Softree Technical Systems Inc. (www.softree.com), researchers at COCANA have developed new software to quickly and accurately determine the optimal vertical alignment for a given horizontal alignment. The first step was to solve the earthwork problem more accurately by taking into account natural obstacles such as rivers or mountains. Then better approximations were introduced at the vertical alignment by taking into account the side slopes of the road. Finally, state-of-the-art mathematical modeling and computer optimization algorithms were used to simultaneously solve the earthwork and vertical alignment problem.
The first version of the software, RoadEng-Optimal, will automatically determine the lowest-cost vertical alignment based on geographical data, horizontal alignment, and a variety of engineering safety constraints (such as stopping sight distance). Where it once took engineers dozens of hours to create a single approximate solution, RoadEng-Optimal will determine the mathematically proved best solution in a matter of minutes. This will allow engineers to explore hundreds of possible horizontal alignments, leading to better road designs. Preliminary results show savings of up to 25 per cent. Considering the cost of a new road, even a one per cent saving could go a long way.
Acknowledgements: Research towards this new technology has been ongoing for more than three years and is funded from a variety of sources, including Natural Sciences and Engineering Research Council of Canada (NSERC) and the MITACS research network. UBC researchers include Dr. Warren Hare (Ph.D. Mathematics), Dr. Yves Lucet (Ph.D. Computer Science), Dr. Solomon Tesfamariam (Ph.D. Civil Engineering), Valentin Koch (M.Sc. Optimization), Faisal Rahman (M.Sc. Optimization), and Shahadat Hossain (M.Sc. candidate Optimization). Softree Technical Systems Inc. has filed a US Provisional Patent.