I'm trying to model the delay caused by closing one lane on an interstate (2 lanes to 1). Based on TRB articles I found, this configuration would have a capacity ranging from 1,200-1,600 vph, with a variety of factors affecting capacity.

I did a test simulation of a lane closure of approx. 1,000 ft, link speeds of 55mph before the closure, 45 mph during closure and 65 mph afterwards. I set PHF at .80. I also changed the headway factors to 1.05 for the segment before the closure and 1.1 during the closure; I didn't change it for the segment after the closure. Truck % was set at 10%.

With the above variables, I found that volumes of 1,600 or less were able to travel with very little delay. At 1,700 vph, I started to get delay with the merging traffic. At 1,800 vph, delays really started to mount and at 1,900 vph, queues would be half mile or greater.

For simulating an actual work zone, I could collect the variables that would affect capacity (e.g. truck percentage, length of closure, positioning of advance warning signs, work zone speed limits, nearby on-ramps and volumes, grades, etc.) but the one variable that I'm not sure about is the headway. I figure headways would increase in a work zone as drivers become more cautious as a whole, but I'm not sure by what degree. I figure a higher headway factor would apply for rural areas vs urban areas and active or intense work zones vs. idle or light work zones.
Does anybody have experience with choosing an accurate headway factor and other things to consider in modeling a situation like this?
Thanks.

Disclaimer:
I have never worked on a work zone delay.

Software Limiation:
You are aware that synchro/simtraffic is not best suited for freeway environment (Trafficware may or may not agree with me)

Confirmation:
I think you are talking about travel time delay, not intersection delay.

My 2 cents:
After the work one is there any other issues (merge/diverge/ramp)?
People would allow other people to merge in a work zone. changing a lane may be easier in work zone than a regular day. people may slow down to let other people move in their lane (probably even in NYC)

Is there a lot of tourists/unfamiliar drivers?
Make sure your taper is correct.

Drivers would in general act more aggresively as the volume increases in terms of lane changes.

Hello Jahid,
Thanks for your reply. Regarding your comments, you're correct about the travel time delay. Queue lengths are also a good measure and SimTraffic is useful here since it graphically shows the queues.

There is an offramp before the work zone and an onramp afterwards--I don't know enough details on how the work zone will be placed around them but the ramps do not have much traffic and I'm not expecting them to have a large effect on mainline operations.

For your last point, it will be summer and it's a rural area, so I think there will be a good number of tourists and unfamiliar drivers. We have no data on percentages though.
This should increase the headway compared to urban commuter traffic though I'm not sure by how much; I guessed with the factors of 1.05 and 1.10 above.

SimTraffic is the only tool I have to try to model this scenario. I might not accept the results as accurate, but I figure it's worth a try. Capacity-wise, the simulations do seem to agree with numbers in the HCM and other research so it looks good so far.

I remember reading something about workzone capacity studies in college. I recall a TRB study done in North Carolina. I don't know if that's enough information to help you or not...

Found it -
TRB 1529

Full text of the article

http://trb.metapress.com/content/d5673774qk6j4752/fulltext.pdf

doesn't have headway data though, says (available in future publication). This was published in 1996 so hopefully its out there by now.

also found this one from illinois

http://ntl.bts.gov/lib/24000/24700/24734/Wk-Zone-fin_rep2-1.pdf

Eric,
Thanks for those links. From those articles and some others I've found, you can get a pretty good idea of what a reasonable capacity for any workzone will be (1,200 vph to 1,900 seems to be the range).

As for actually modeling it with SimTraffic, it seems there are two variables you have to experiment with: headway factor and the lane positioning distances. The geometry, % trucks, volumes and speeds are comparatively much easier to find via recorded or field data. By adjusting the headway factor and positioning distances, you can model a very wide range in capacities.

Advice for anybody else trying this:
-Don't set the mandatory distance too low. You can get trucks blocking each other right at the merge point and the logjam never clears.
-If there's an onramp just before the lane closure, modeling will be much more difficult. You have to get good volumes for the ramp and also the geometry of the merge into the through lanes. If there are no ramps, model one anyway and put a ramp volume of zero--you have to have at least one intersection in order to get a SimTraffic simulation.
-If the lane closure is very long, work is very intense or with lots of distractions, and travel speeds are much slower than unrestricted speeds, the bottleneck will be the actual lane closure area. Otherwise, the bottleneck will be due to the merging. Just my personal experiences, I think the former situation is much more common in urban areas and the latter common in rural areas, but I have no data about that.

since you're modeling high speed links, you might want to check out this topic I started under simtraffic 6 forum

http://trafficware.infopop.cc/eve/forums/a/tpc/f/5786041442/m/9111094961

I'm not sure how much it applies to synchro 7 though, I haven't gone back and seen how the model has changed in this regard.