I know this is an old thread. It was suggested that I post my answer to this same question from bicycles stackexchange. Link:
Here is my previous answer:
I recall this thread, and thought I'd add a link to a post describing an impromptu experiment I conducted this week, which tested the impact on the power demand of a test rider (172cm 60kg female on a track pursuit bike riding at a quasi-steady state velocity on an indoor wooden velodrome) of another rider (185cm 80kg male on mass start track bike riding in close proximity), and to compare this with the test rider's solo ride power demand.
The tests examined the following locations of the other rider relative to the test rider:
- immediately in front of the test rider
- riding next to the test rider (on their outside)
- immediately behind the test rider
- completely away from the test rider and not riding on the track (to
provide data on the solo power demand for the test rider).
I use specialist technology to assess rider aerodynamics in real time and had the chance to perform this experiment at an indoor velodrome (Dunc Gray Velodrome, Sydney), so that we could at least conduct such an experiment in well controlled, no wind, low yaw angle conditions.
Test runs were repeated for validation and confirmation of results. The testing protocols and analysis of data provide values for the apparent CdA (coefficient of drag x frontal area, units: m^2) value for each of the test conditions. I then use the apparent CdA data to show the power demand for the test rider to maintain a 40km/h average speed.
This is the link to my write up, which includes links to other experiments and published science on the topic.
Here are the summary of the data in chart and table form, which show the power required for the test rider to maintain 40km/h while riding solo, and with the other rider in various relative positions:
In summary, compared with the power required (195W) for her to maintain 40km/h (lap average speed) on the velodrome:
Drafting immediately behind the other rider gives massive benefit (-76W, -39%). No surprises there.
Having a rider immediately behind (~1/2 wheel gap) provided ~ -7W (-3%) benefit for the lead rider.
Having a rider ride right next to her (~0.8m - 1.0m lateral gap between wheels) created an additional power demand of ~ +10W (+5%).
The result of a 7W (3%) benefit to the lead rider of having a rider immediately behind is in line with previous experimental results and published studies. So while the effect is small, and would be difficult to feel while riding, it is a real effect, at least in low wind conditions.
The side by side ride result showing an additional power demand of 10W (5%) in low yaw conditions is more novel, and has interesting implications for team formation events (e.g. team pursuit and team time trial) and rider changeovers.
Of course different rider morphologies, individual aerodynamic properties, riding alignment configurations and wind conditions will yield different results to this impromptu experiment, but I thought it interesting none the less.