My up peak round trip time calcuation is fine, but when I change to simulation I get long queues and long wait times. Why?
Up peak round trip time calculations and simulations can be shown to be consistent. However, there are some fundamental differences which need to be understood.
1. Round trip time calculations allow for a "Round Trip Time" inefficiencies or losses. In Elevate this value is an Analysis Data input. Some designers choose a low value, sometimes zero. Zero round trip time losses is the equivalent of saying that the elevators will operate in an ideal way without bunching or door reversal and with perfect dispatcher. So, if you want to demonstrate consistency, we recommend that you use at least 5% losses.
2. Sometimes designers forget to use an up peak mode for the selected dispatcher algorithm. Select an up peak mode in Analysis Data. Different up peak modes will work better depending on the intensity of the traffic - refer to the Elevate manual for a discussion of the different modes.
3. Round Trip Time calculations do not allow for door dwell times. To demonstrate consistency, you will need to reduce door dwell times. This requires that you enter your Elevator Data in advanced mode, rather than standard mode. Then you can reduce "Door Dwell 2" from the default (usually 2 s) to 0.5 s. For a discussion of door dwell times, please refer to the Elevate manual.
4. Round trip times allow for an average car loading which may be a fraction, e.g. 12.8. A simulation only allows for whole people, e.g. 12. One approach is to set the simulation Capacity Factor to 100%, and then to use Elevate's results graphs to determine the average and maximum loadings.
5. Round trip times are generally calculated around the saturation point of a system. If a system has an inherent handling capacity of 12%, at 12.1% queues will start to form in simulation. The longer you run a simulation, the worse the queues and waiting times will become. To demonstrate this, use the Elevate "step profile" (Passenger Data, Arrival Rate as option). Below saturation the waiting times are very low, then they rise rapidly as saturation is reached. Bearing in mind that traffic calculations are not exact, if your system is saturating at 12%, it may well be OK at 11.5% which is close enough. Running the step profile also demonstrates that the relationship between Waiting Time and Interval is complex. With low up peak traffic, there should always be an elevator ready to serve the call, hence the extremely low or zero average waiting time. During saturation, the interval may continue to be acceptable, but because elevators are full and passengers may have to wait two or more intervals to get into a car, the Waiting Time becomes large as queues form.
6. By default Elevate will simulate 15 minutes of continuous traffic at the peak level (e.g. 15% five minute handling capacity for 15 minutes). In theory the system should be able to cope with this if the RTT calculation says it can. Indeed if (1) to (5) have been taken into account, it normally will. However, some designers, consider that the system need only sustain the highest peak for 5 minutes. For example the "Barney up peak profile" rises and falls to the specified peak. Another approach, running at the peak just for 5 minutes without any traffic beforehand is considered too easy for the system as it begins with a "head start" - at the start of the simulation all the cars are empty and lined up at the home floor. When the simulation is run for longer (by using a profile or running for 15 minutes), the effect of this head start becomes negligible.
It is not unexpected and is perfectly reasonable for a RTT calculation to suggest a solution is 4 No 1000 Kg cars @ 1.6 m/s and the equivalent simulation to suggest 4 No 1250 kg cars @ 1.6 m/s. This is as close as you can expect given that one method is a mathematic model based on a simple round trip calculation, and the other is a detailed simulation.
Experienced traffic analysis practitioners tend to use RTT calculation as a rough guideline to determine a benchmark solution to benchmark design recommendations (e.g. 12% or 15% up peak handling capacity for an office building). Then they use simulation to determine a more realistic performance based on their knowledge of elevator traffic in different types of building at different times of the day. Trying to match the up peak RTT calculation with the simulation then becomes a moot point.