TMODEL has a great advantage over most other transportation planning models in that it models delay at nodes where the delay actually occurs. This is especially true when modeling an arterial street system where the nodes typically represent signalized intersections. Although this capability is very logical and should provide better assignments, little has been done to determine appropriate node delay parameters to be used with the model.
In TMODEL, the node delay is calculated with the formula D=K(v/c)E + B. During some recent assignment runs with the default values for the node delay parameters, I noticed that several nodes had v/c ratios of 2.0 or greater, and the model was still assigning traffic through these intersections. Now we all know that once the v/c ratio exceeds 1.0 at a signalized intersection the delay becomes excessive and traffic is diverted to alternative routes. Therefore, I began to investigate just how the node delays are affected by increases in the v/c ratio.
Using the default values from TMODEL, when the v/c ratio is calculated to be 1.0, the node delay is equal to 0.68 minutes or about 41 seconds. This would be about level of service E which seems appropriate. But as the v/c ratio increases the delay does not increase as much as would be expected. In order to better illustrate this, a graph was prepared showing the calculated delays with the default delay parameters from TMODEL, estimated parameters using the Highway Capacity Manual methodology, and a modified estimate for the TMODEL factors.
As can be seen in the figure, the Default TMODEL factors do
not project delays which would be
expected from The Highway Capacity Manual Methodology, especially when the v/c ratio
exceeds 1.0. The Highway Capacity Manual delay calculations shown in the figure were based
on the assumption that the average cycle length is 100 seconds, the weighted average approach
g/c ratio is 0.4, and the weighted average approach capacity is 3,000 vph.
In order to match the delays projected from the Highway Capacity Manual, the following delay parameters were used: K = 0.40, E = 8.0, and B = 0.25. The use of these parameters at signalized intersections modeled in TMODEL should provide more realistic intersection delays and therefore a more realistic assignment.