Modelling Innovation - PC MOVA Hold Messages

At Multimodal, as well as looking to provide modelling efficiency tools and processes (see our Technical Series), we also strive to innovate and come up with solutions to get more out of our microsimulation software.

As always with our technical posts, we acknowledge that there may be other methodologies out there, but we want to share our current methodologies for future development and assist those experiencing the same problems. This blog does not apply to the recently released PC MOVA version 3 - documentation suggests that hold messages are now included, although we have not yet been able to test this new functionality. However, this should apply for any older version of PC MOVA.

 

The Problem…

Two recent projects which Multimodal have undertaken identified an issue with the traffic signal add-on – PC MOVA – being unable to accurately model hold messages within the external simulator (in this case, VISSIM).

A hold message requires a signal to be constantly sent, which cannot be set-up when configuring the PC MOVA connection file on its own.

As such Multimodal, in collaboration with Dan Preece at Integrated Traffic Services Ltd, have identified a workaround within PTV VISSIM.

 

The Concept…

This workaround uses ‘dummy links’ to send and maintain hold messages, with detectors and signal heads on the dummy link set to the appropriate phases on the junction.

The idea of the dummy link set-up is that when a signal turns green (which is linked to a phase on the junction), the traffic passes over the stop-line and runs over the MOVA detector. The detector, which has been linked within the PC MOVA Connection File, generates a demand and triggers a hold message. The constant flow of traffic keeps this hold message active until the phase turns red (in line with the associated phase at the signal controlled junction).

 

The Set-Up…

An example of the set-up is shown in Figure 1.

Figure 1 – Example Dummy Link Set-Up for Hold Messages

For each dummy link used to enable ‘Hold’ messages, the following network elements are required:

  • A signal head which is linked to the appropriate phase;
  • The associated MOVA detector placed after the stop-line;
  • A vehicle input of 3600 vehicles/hour to model a constant demand; and
  • A custom speed distribution (31mph) and driving behaviour (based on an Autonomous Vehicle Study carried out by PTV) to ensure constant link speeds, flows and close vehicle proximity.

To account for delayed hold messages, the detector on the dummy link is located a sufficient distance away from the signal head (see Figure 2), with the datasets amended to achieve the required clearance.

Figure 2 – Example Dummy Link Set-Up for Delayed Hold Messages
 

The Result…

For a project that involved the update and revalidation of a single motorway junction, the main change was the VISSIM model and PC MOVA set-up to link together three separate signal controlled nodes (which were previously modelled separately under PC MOVA control). This utilised the hold message approach detailed above and the result of the journey time validation before and after are shown below.

Fig003.PNG

From the table above, it can be seen that in the AM peak, the level of journey time validation remained at the high level as in the previous VISSIM version. However, in the PM peak, we see an improved level of validation in the updated model. This suggests that the linking of the signals through the hold methodology had a positive impact on the network and produced a more representative base model.

Traffic flow validation took the form of comparing the exit flows at the junction, which in both model versions (previous and updated) were 100% against GEH and WebTAG Individual Flow Criteria.

 

Final Thoughts…

The methodology of linking signal controllers through hold messages using dummy links appears to be a workaround which is worth considering in producing more representative signal operation.

It is acknowledged that there may still be a slight lag and subsequent offset of the signals using this method (~two seconds), but the efficiency in the model is likely to counteract this and provides a better linking facility overall.

Daniel BentComment