Having a power meter is too much fun. I was wanted to quantify the impact of a heavy rim on an all-out sprint. This sprint had a 5s mean maximal power of 1527W.
The short story:
Average cost of a heavy rim (Velocity Deep-V) vs. a light rim (Zipp 280/303), during the entire sprint is 0.2% of my power, or 2.6W out of 1301W available.
The MAXIMUM cost was 0.25%, or 4.0W out of 1604W available.
This is just for the rim, not counting anything else on the wheel.
That's not much difference. At all. There are extremely rare circumstances where 0.2% would be the difference between winning and losing a sprint. Still, for the same price as a Deep-V, an Aerohead will chew up half of the difference, so you're down to 1.3W.
One interesting (?) point is that my max acceleration was 1.56 m/sē (0.16 G )
The long story:
I took my recent 5s best sprint power file (1527W), and exported it as a .csv.
In Excel, I worked next to the sprint data to calculate angular acceleration of the wheels during the sprint. I calculated a moment of inertia for the rim (550g Deep V, assumed a thin ring concentrated at the bead seat diameter).
From this, I was able to calculate the angular acceleration for each line of data during the sprint. Then I calculated the torque required to generate that angular acceleration on a 311mm radius ring of 0.55kg.
Looking at the torque required, I compared that to the recorded torque at the hub, and came up with a %
The calculations actually show a greater effect since I assumed the weight was at the bead seat diameter, rather than somewhere in the middle. I also assumed the high weight for the Deep-V and the low weight for the Zipp.
Since the recorded speed is affected about one row below the torque hits, I did some offsetting for my equations.
The spreadsheet is available for review here (check around line 1000). Please let me know if you see any errors -- it's been a long time since physics 101...
If you've never seen a power file, it might be interesting to view anyway (you can graph Watts yourself).