Analizing a signalized intersection with shared (thru/left thru/right) lanes. 1

[TrafficDad]

I am reviewing a traffic impact study which includes the LOS analisis of a signalized intersection on which the main road has two lanes in each direction and the cross-street has one lane in each directions. None of the four approaches have turning lanes and the signal is a two-phase signal. The left turning volumes from the main street are relatively high, and since there is no left turn lane or left turning phase, one of the two lanes on the main road gets blocked be left turning vehicles and congestion ensues.
However, the HCS (or synchro) analysis does not show this and rather comes up with a LOS C for the intersection as a whole as well as the main road approaches. But I know this cannot be true given the left turning volumes and the two-phase signal.

The volumes are as follows:
Main Road (E-W)
EBL:255
EBT:255
EBR:20
WBL:125
WBT:500
WBR:30
Side Street (N-S)
NBL:55
NBT:375
NBL:80
SBL:10
SBT:195
SBL:95

Evidently a volume of 255 vehicles turning left (EB)are enough to warrant an exclusive left turn lane and given the 500 vehicles in the opposing direction (WB) I would also include a protected-permitted phase. However, the analyisis results with the current lane configuration show a delay of only 21.7 seconds (LOS C) for the EB direction. So the submitter of the reports claims that no improvements are needed and that the intersection should work fine without turning lanes.

So, my question is: Why doesn't the analysis results show that there would be a problem caused by left turners not being able to turn left in the EB direction?

Thanks.

 

[ACtrafficengr]

Without seeing the analysis report, it's hard to say. The intersection may have enough capacity to operate acceptably as is. It's also possible the analyst tweaked some parameters to get unrealistic results.

My first thought you should consider more than just delays. Four-lane undivided roadways are notorious for poor safety performance. As a profession, we spend too much effort on capacity, and not enough on safety and livability. Is the intersection operating safely? Will it continue to do so with the added traffic?

Which leads to my second thought. This site may be a good road diet candidate. A peak hour volume of 1185 vph correlates with an AADT of 8000 (rural) to 11,000 vpd (urban). That's well within the doable range of a 4-lane to 3 lane conversion. Above 15k, you'd have to do more analysis, and possibly some intersection improvements. You'd probably have similar operations and considerably improved safety. Your left lanes are probably de facto left turn lanes anyway. FHWA says reductions in overall crashes of 19-47% can be expected, and most locations report improved or unchanged retail sales in the corridor. It's worth thinking about.

 

[teeman]

Synchro is a mathematical model. In most instances it provides a reasonable approximation of operations. Sometimes the analysis bears no resemblance to what is occurring in the real world. If the model is not accurate you have to make adjustments to the inputs. In this case it may be the permissive left turn factor in the lane geometry page. On some problems I've adjusted the lane width to reduce the saturation flow to better mimic what is actually happening in the field.

 

[johf]

This could be a model calibration issue. A traffic model is intended to reflect real world behaviour and so must be calibrated to do this. The parameters used by the model may not reflect real world behaviour. This could be saturation flow, headway, flows/volumes, etc.

In my experience, a congested intersection that when modelled doesn't reflect the congestion often occurs when the model is not supplied with the true traffic demand. A congested intersection will usually mean that traffic demand exceeds capacity. Traffic counts taken at the intersection will be a measure of the capacity, not the demand.