My idea is not to predict average power of recumbent. I have the power meter and I trust it and for every one that has a power meter, I assume it is accurate.
My aim is to compare an "ideal" traditional bike to different kind of recumbents. The Ambrosini formula simulates very well a standard upright bike. In all my tests my Assioma power meter values are very close to Ambrosini values, especially with my Trek Emonda upright. There was a max of 2-3% difference between real and expected data.
So, every time I make a strava segment, I can compare my performance to the one I could reach in a standard bike, even if I have not one, using my excel table. And every now and then I directly compare, see for example:
The excel table works like that. For exaple, if Larry had ridden a 7 kg upright on the first day race course, how much power would he had to put more (or less) on the pedals, instead of using his Vendetta? The answer is: he would have put 7% more power on the pedals in that 5 km sector (0.93 W/W Ambrosini).
In the study case, me and Larry have the same values (0.93 in the 5 km sector and 0,85/0.86 in the first sector of the climb,which confirms the data above) . That means that we both have a similar bikes and similar performance. And that power meters were correctly working.
What is the reason I make all these comparisons? Because I want to see where and when recumbents are better than traditional bikes and, in parallel, this let me compare different types of recumbents.
For example I want answer the question: on the championship 3 hour course, it was better to use a velomobile or a recumbent to set the fastest time or to spend less energy during the whole competition?
This is possible even without "Ambrosini" values. See for example this:
These are some data of a single lap of the 3 hour championship.
I have taken the ones that are near to 33 km/h average speed. If you have the same speed, you can see the average power needed. The less, the better bike.
This let me compare bike performance (and not rider). Because you can generally compare two bikes that ride with the same speed or with the same power. In this case it is easier to take data with the same speed.
What I see is that the average power used by riders are: 179W for Larry, 153W for me, 190 for Peter in the velombile, 174 for Geoffry in the Zokra. Unfortunatly Jo has not set a 33 km/h average speed lap, so I have to compare a 35,3 km/h lap, with 240W (of course he had an higher power but how much higher?) .
So, what I see is that, for that course, the HTH is a little more efficient because I only needed 153W. Larry needed 179W (15% more).
And this was not the performance of a single lap. Infact at the end of the race me and Larry had very similar average power (and also heart rate and infact we have nearly the same FTP at the same heart rate, and that confirms that power meter data are right). But my average speed was 34 km/h and Larry's 31 km/h. If you see the W/W ambrosini data, you see the same values. The HTH has a 0,64 performance, while the Vendetta 0,75. The DF XL was worse, meaning that, on that course, probably, it was better to ride a recumbent than a velombile. Jo Stein effort is even better, because his valute is the worst of all (0,87). So we can say that he had probably the worst bike but the best legs.... Probably, if he had set a 33 km/h lap, the value could be 0,80 but not less.
I repeat, this is my way of comparing bikes because I want to know how bad or good is my bike compared the others, and so I can try to improve its performance in the condition where it is worse.
I think it can be useful in some way but I also undestand it is complicated and maybe not to intuitive (or maybe there may be better ways to analyse bike performances).