Cephalotus said:
Thanks a lot for the data on your "perfect day". Highly appreciated.
I hope that someday after the trip there will be data available for every day and every vehicle involved.
So I've not got all my raw data available for anyone who wants to look at it here:
https://www.dropbox.com/s/kt8in0sfmvdd5y9/Justin%20Suntrip%20Log.zip?dl=0
The start and stop point of the files isn't exactly at the start of each day depending on when I remembered to reset the databox logging device, but most of them are reasonably close. This is just raw log data, we're working on some web tools that will do automatic day by day summaries of all the stats and enhancements to the trip analyzer page so that it can handle and display way larger data files without chocking.
Can you show data from a day with rainy or overcast weather, too if you find some time to do so?
Still coming soon! But I'm going to start with another near perfect day which was the first day of the race. Conditions in the week before the Suntrip gathering were terrible (lots of people pedaled their rigs to Lyon in endless downpour), and then improved during the June 14th-19th preamble from Lyon to Chamonix, and finally on the race start day it couldn't have been better, with bright skies, clean air, and tailwinds. It was like everything was going to fall into place.
We left from France at 10am, went over two 1500m peaks, rode through a flat valley in Switzerland, and then climbed up and over a 2000m pass. Descended from there into Italy having covered 199km, pretty much spot on our initial daily target. If you look at the graph data, we were doing 30-35 kph on the flat valley, hovered above 20kph for most of the hill climbs while consuming 1400-1600 watts, and clamped our speeds to 40-50kph on the descents with regen.
On the solar, it shows a total of 4kWhr, but that's because I hadn't reset the logger from the previous day, so when we started it was already at 1200 Wh giving us a total of 2.8 KWhr. Since the battery was charged in the morning before the 10am departure we didn't collect anything from the sunrise to 10am point.
Because there was so much mountain climbing this was a perfect leg to showcase the benefits of DD hub with regen, though it also became apparent that we didn't have our setup quite as dialed as would be ideal. I had configured each Phaserunner controller to do a max of 12 amps of regen (so 900 watts net from both controllers) in order to keep the combined solar+regen charge current to just a little over 1C on the LiGo batteries to keep the cells happy.
However, what that meant is that if the hill was steep enough that 900 watts regen was not sufficient to keep the speed clamped, then the bike would continue to accelerate, and the faster down you went the LESS regen torque you would have on the motor because at a higher RPM fewer Nm are needed for a given braking power and regen wattage. We would only get more than 900 watts of regen once the speed of the bike had reached a point where the back-emf voltage of the motors was above the pack voltage which would be at +45 kph, and then suddenly the regen would increase again since the controller has no way to limit it.
This resulted in a situation where if our goal was to maximize the amount of battery charge recovered on the downhill, it was often advantageous to still use the mechanical brakes to keep our speed at say 30 kph. If we didn't use mechanical brakes, the speed might increase to 45 kph; we'd still be getting just 900 watts into the battery but the descent would be over that much sooner so the total amp-hours recovered would be less, with a much larger share of our potential energy would have gone into wind drag losses. It also meant that pedaling on the steep downhills like this was similarly counter productive, it would just mean going downhill faster less time spent getting this fixed regen into the battery.
Here's a link to just the data from the 2nd hill pass.
http://www.ebikes.ca/tools/trip-analyzer.html?trip=nbc8ZL
You can see that we get 42% regen range extension, with some sections where we were going above 50 kph and having more like 1200-1400 watts of regen from back-emf, but mostly we stayed slower using the brakes to keep it in the 900 watts regime.
Ideally, we would have had a setup with either higher C rated cells that can be charged at 2-3C, or a larger battery pack, that could take full advantage of absorbing the potential regen energy from the motors and Phaserunners. In that scenario with the phaserunners not having a max regen curren tlimit, we could have set a speed on the descent of say 30 kph, rowed and pedaled on the downhill sections too, probably had 1500+ watts going into the battery and likely doubled the amount of regen energy captured. Not that 42% regen is bad, but it doesn't represent the true potential in this scenario.
Another thing that this first day was very useful for was giving us an idea of how many Watt-hours we needed for a given elevation
climb. Here is the data on this pass at 100m elevation steps.
View attachment 1
As a good ballpark we could see that we were averaging about 100 watt-hours consumed for each 100m of elevation gain, and that became a really useful rule of thumb for budgeting our battery reserves on future hill climbs during the entire trip. If we knew that there was a 500m elevation climb ahead, we'd aim to ensure we had at least 500 watt-hours in the battery, or about half full, so that even if we lost sun while climbing there would still be enough in reserve to make it to the top. Once at the top, a flat battery is no issue since it's all regen going down, but if you run flat while climbing a hill then moving a behemoth of an ebike like this with no motor assist is a real grind.
You can see that on this 2000m pass, we dialed it pretty well. We took two breaks on the climb to allow some solar charging while the motors would cool, and only just hit the system's low voltage rollback right at the top of the mountain. The CA's LVC was set to 29V, but that includes the voltage drop along the long lead from the battery to the two motor controllers. The actual voltage as seen by the datalogger directly connected to the packs was a bit higher at 30.3V.