By Laura Oriol
Some models cover large and complex systems that may require different levels of simulation. Smaller subareas typically require detailed simulation approaches, such as microscopic simulation, whereas it is far faster to use simplified models, such as mesoscopic simulation, to model the wider areas.
On the basis that microscopic and mesoscopic models are never going to produce the same results, it is important to get a good match in behavior wherever possible. Assuming that a microsimulation approach provides more accurate outcomes, in some circumstances it is necessary to calibrate the default meso parameters to achieve analogous results in both models.
The following guidelines show how you can alter existing parameters in meso to match the microscopic model behavior in certain situations.
Lane changes and interaction between vehicles due to the crossing of traffic streams is accurately modeled in microscopic simulation, but might be less accurate and too optimistic in mesoscopic simulation.
In view of the results (Figure 2), we recommend adopting a Reaction Time Factor value between 1.15 and 1.25 in mesoscopic simulation to provide similar flows at weaving areas as in microscopic simulation. Note that for very short weaving sections (<150m), mesoscopic simulation still provides significantly higher values, suggesting interaction between vehicles that try not to miss their exit or to get lost. Therefore, if your model includes a short weaving section, you’ll need to set greater Reaction Time Factor values to take this effect into account. In any case, for design reasons, it is highly unlikely you'll find such short weaving areas.
In certain cases, it may be necessary to implement stop lines in turnings to increase their capacity or to reproduce the actual behavior of drivers. This is the case in the following intersection (Figure 3) where vehicles can turn left but must give way to vehicles on the main road.
Stop lines are available for detailed microscopic simulation models, but not in simpler models such as mesoscopic simulation. This reduces by default the capacity of these turnings in a significant way for mesoscopic simulation processes. To obtain similar outcomes in both models, it is possible to add an internal section or to build different control plans for each model, but these solutions can be difficult to maintain. A more elegant solution to the problem consists in calibrating the existing meso give way parameters of the turning. The following example tries to provide a generalized procedure for this.
If we take the previous intersection and the following control plan:
The stop line provides space for about two vehicles. To achieve this storage capacity, meso give way parameters need to be calibrated to allow at least two vehicles take the turn in each cycle for dense traffic conditions. In Turnings with give way, use the gap-acceptance model parameters (Figure 6).
For this example,
– Total Green Time = Green Time + Red Percentage · Yellow Time = 40s + 0.5·3s = 41.5s
– Total Green Time / Number of vehicles = 41.5s/2vehicles = 20.75s/veh
Each vehicle has 20.75s to take the turn. The general idea is to set a very low Final Safety Margin (See Figure 6) to ensure that the vehicle goes through and a Give Way Time Factor that corresponds to the desired maximum time the vehicle will wait (20.75s). The default Initial Safety Margin is 10 seconds, so for this example the Give Way Time Factor must be set to 2.1 (20.75s/10s=2.1).
By following this procedure, you should achieve the desired turning flows (Figure 7).