Drafting experiment

I've posted a couple of times about drafting to increase fuel economy. And the Mythbusters have covered this as well. There seems to be no question that it's effective, witness the Tour de France and Nascar. So, what is the nature of the improvement?

I can only get better fuel economy if I have to overcome less force to go forward. Drafting certainly doesn't change rolling resistance or drivetrain losses, hence it can only affect aerodynamic drag. As I've discussed on previous occasions, aerodynamic drag, D, can be modeled by 

where rho is density, v is speed, Cd is drag coefficient, and A is frontal area. At a fixed speed (where Cd will also be constant) the only thing that can change is rho, the density. Thus, my working hypothesis is that drafting is effective due to a zone of lowered pressure behind the front vehicle.

I've purchased the Vernier LabQuest and a variety of sensors, one of which is the Gas Pressure Sensor which measures absolute (not gauge) pressure. I determined to see if I could measure this effect. I attempted to do so by cracking my window and extending the inlet tube of the sensor through the crack. My idea was to approach to and withdraw from the rear of the truck, recording my estimate of distance from the truck and the time as shown on the LabQuest display on my iPhone. I'd be looking for a reduction in measured pressure as I approached the truck.

So how did it go? Not so well. After multiple attempts, if anything, I recorded a trend to lower pressures as I moved away from trucks, higher as I moved in. And yet that can't be right. Or can it? This was a poorly designed experiment in that all factors that might affect the measured pressure were not controlled. My speculation is the confounding factor is the fact that I am on a freeway with hills and a general upward trend (I was eastbound, that is, away from the ocean toward the inland area). As I climb, the pressure will decline by about 0.012 kilopascals/meter and the range measured in the entire experiment was about 0.3 kilopascals, which would represent a climb of something like 25 meters. Now, the experiment was over a period of about 250 seconds and a driven distance of about 6 kilometers so such a variation in elevation is certainly not out of the question.

How can I get a better result? Since I don't have maps that indicate elevation and location with sufficient accuracy to adjust the pressure data to remove elevation changes, I can only think of two options. First would be to repeat the experiment in a controlled location. This is very difficult in that I don't have access to a truck and a suitable location. What else? My plan is to run the experiment and note the location, then to run it at the same location with no drafting. This should give me the ability to subtract out changes caused by elevation changes.  A negative result would cause major head scratching, stay tuned.