On the roads in mountainous areas where traffic is low, most sections are very narrow, and it is difficult to pass oncoming vehicles. There is no big problem when ratio of large vehicles is low. However, during construction... [ view full abstract ]
On the roads in mountainous areas where traffic is low, most sections are very narrow, and it is difficult to pass oncoming vehicles. There is no big problem when ratio of large vehicles is low. However, during construction works such as tunnels and dams in mountainous area, the traffic of large-size vehicles temporarily increases on narrow carrying route for carry-in materials and carry-out sediment. In the case where much large-size vehicles more than predicted pass through the route, the improvement of the road is required. Since it is not a permanent increase in traffic, it is more effective to improve locally and flexibly than full 2-lanes improvement. Especially for roads where the traffic increases considerably, we decide appropriate passing places with using traffic simulations repeatedly, but it is very complicated. Therefore, we developed new evaluation indicators for passing and narrow sections, as an alternative to time-consuming traffic simulations, and reduced considerable time of iterative examinations. In addition, we tried optimizing placement of passing sections as minimization problems of the widening cost by combining proposed evaluation indicators and optimization method.
The proposed indicators can evaluate cases where both large-size and small-size vehicles are mixed and vehicle arrival distributions are with probability fluctuation, and it is highly versatile. By changing the combination of vehicle types that can pass each other in accordance with the road width, it is possible to evaluate a complex road environment in reality. Additionally, the probability variation in the arrival distribution is indispensable for reproducing the traffic phenomenon, which is an important factor for an alternative to traffic simulations. For a single narrow section, the comparison with a traffic simulation for the average waiting time and the necessary lengths of passing section at the both ends showed that the average absolute error between them was very small. In other words, we confirmed that the indicators can sufficiently substitute the traffic simulation.
For cases where multiple passing places must be arranged, we considered combination of widening patterns taking into account constraint conditions based on evaluation indicators and topographical restrictions as the minimization problem of widening cost. Then, we applied GA (Genetic Algorithm) which is one of optimizing methods by evolutionary computing. In expression of genotype, we configured a model that can set finely the variation of applicable construction methods and the sections where widening is impossible depending on the topography. When applied to a case which is based on the real problem and the result was checked by the traffic simulation, there was no problem in vehicle operation and the difference of the average waiting time from the evaluation indicator was as small as 3.4 seconds. Furthermore, in comparison with the conventional method by manual effort, it was possible to obtain an effective solution that drastically shortened the study time from 8 hours to 4 minutes, and also reduced both the widening cost and the average waiting time. In addition, characteristics of trade-off relationship between the widening cost and the average waiting time was clarified by using multi-objective optimization.
ICCCBE2018_fullpaper_Ver11.pdf 