Calibration Of International Capacity Models For Dual Lane Roundabouts During Saturation Flow Condition
Free (open access)
469 - 480
H. M. N. Al-Madani
Varying differences in the prediction of roundabout capacities were observed between various international models and field data. This study attempts to calibrate 7 international models for forced flow conditions. These include UK, aaSIDRA, Swiss, US FHWA, HCM, French and German models. These are compared with field data collected from 12 large roundabouts in Bahrain. They are compared considering scattered plots, Root Mean Square Error (RMSE) and t-tests. Each model is calibrated by a trial multiplicative factor which is then adjusted in an iterative manner until best match with actual data is achieved. The calibration adjusts the Y-intercept and the slope of each model. The calibration factors, based on the initial results, showed high variations in the adjustment factors between one model and another. The calibrated models showed substantial improvements and were fairly comparable with actual data. The UK, aaSIDRA and French models, with over 40% reduction, required the greatest adjustments to suite the field data. The FHWA model required 35% reduction and the German model required 15% increase to reasonably fit the field data. The HCM method required no adjustments and the Swiss model required a reduction of 5%. The calibration factors were as follows in respective order: 0.5225, 0.515, 0.585, 0.65, 1.16, 1.0 and 0.95. While differences in capacity prediction between one international model and another confuse the practitioners; calibrated models assist them to have just predictions regarding roundabout capacities during forced flow conditions. The developed calibrations require further investigation to reduce the deviation between the calibrated models and the actual data. Keywords: capacity, maximum entry flow, circulating flow, exiting flow, calibration, adjustment factor.
capacity, maximum entry flow, circulating flow, exiting flow, calibration, adjustment factor.