الفهرس 
INTRODUCTIONOnly 14 pages are availabe for public view
 |  | from | 207 |  |  |
| | | from | 207 | | |
Abstract
Unconventional median U-turn intersections have been extensively implemented along
major corridors in Cairo, Egypt. These unconventional intersections do not involve
signalization at any point. They utilize a non-traversable median with a U-turn crossover at
the downstream to manage all crossing movements and thus, creating two-sided weaving
sections between the minor approach and the U-turn crossover. In this paper, VISSIM was
used to model and simulate these weaving sections through an experimental analysis with
4 influential factors namely; major demand, minor demand, weaving length, and the minor
through traffic split (% Mi THR).
The experimental design resulted in 960 scenario runs, which were automated through
an external Visual Basic program developed specifically for this study. The first stage of
the analysis was dedicated to estimate the capacities of the weaving sections and the minor
entrance, which were found negatively correlated. Increasing the major demand caused a
decrease in the minor entrance capacity and an increase in the capacities of the weaving
sections. It was also found that capacities increase with the increase in weaving length;
however, increasing the length beyond 200 meters was not beneficial. Increasing the minor
through split caused an increase in volume ratio and a decrease in capacities. Furthermore,
regression analysis was used to develop various simulation based capacity prediction
models that resulted in a relatively high R2 values.
The second stage of the analysis was dedicated to test the appropriateness of the HCM
2010 weaving methodology when applied to the urban weaving sections to predict
capacities, lane change rates, and speeds. Comparisons between the predicted and the
simulated estimates showed that the HCM 2010 methodology provided higher capacity
predictions up to 1.6 times the simulated capacities. On the other hand, the developed
regression models produced capacity estimations that were more realistic. This provided
evidence that the structure of the developed models is more suited to represent capacities
of similar weaving configurations.
Further comparisons using paired t-tests and parity plots showed that, the HCM 2010
methodology also underpredicted lane change rates; therefore, speed predictions were
higher than the simulated speeds at each weaving section. Finally, an effort was carried out
to calibrate and modify the speed prediction algorithms of the HCM 2010; however, the
effort did not yield any significant results.