charging, discharging data and usage

Having now done a few explorations with my scoot I can say that the Mercane does indeed (at least in my case) come with good batteries. However what I wanted to talk about here was the interesting observations borne out of yesterdays run.

What I did and why

I decided that a nice run up my favorite (and nearby) mountain road would be in order on a quiet boxing day. I anticipated from experience that I could make it from my house all the way to the top of the mountain and back on a single charge. While I've done similar distance runs before it has not been with such a gruelling non stop slope as this. As you see its steep countryside.

The trip out to the mountain, up and down the mountain and then back home is about 20km and while its steep up there, it means that I need no power at all to get back (and because I enjoy the speed on the way down, I wouldn't really benefit from regen brakes for this or 99% of my trips). Just to be safe I used the "Economy" setting for the trip out, but satisfied when I got to the bottom I'd have enough to get home I put it back on "Power" setting. As it turns out for the part up the hill speeds were not dissimilar to my previous runs with "Power", just about 2kmh lower (which benefits power consumption).

I was satisfied at the top of the hill where it showed 48.7V, which works out to be 3.7V per cell, which is not much below the rating for the batterys "normal" level.

So on return I charged it back up and noted that it (after a minute or so to allow it to settle) was showing a battery level of 47.6V

How much of my energy did I use?

One thing for sure, is that because of the very shallow discharge curve of Lithium batteries, and slight variences in chemistry and even perhaps discharge levels within my 13S6P pack (bundles of 6 cells in Parallel in a string of 13 bundles in Series) it is hard to know the state of discharge. Like did I discharge it to half, nearly empty ... what?

I figured that the best way to determine this was to again use my 150A power monitor (featured in other articles such as here) to actually know this.

However I decided to go a step further and look to see if 1) how long the change took before the Amps going in backed off, 2) what was the exact relationship to Volts duing that constant current phase and 3) look to see how long the balancing phase took to complete.

Here is a graph of my data, you'll notice that at first I was not sampling every 10 minutes.

so basically the voltage crept up a a small amount (starting at 48.08 as soon as I plugged it in) and moving finally to 54.9 (which dropped to 54.6 shown on the bikes volt meter as soon as I unplugged the charger). Finally the charge indicator went Green just after the 4:30 interval (I happend to be there writing down the data) ... so the BMS is doing a great job IMO.

The next thing you may notice (if you read graphs not just look at the pretty picture) is that it took a little over FIVE HOURS to charge. Now this is unsurprising with the piddly little 2A charger provided. Why "piddly" well the cells in the parallel bundles are 2.2Ah, and so since there are 6 cells in a bundle that means 13.2Ah, most makers suggest that charging at 0.5C (where C is capacity) is quite ok. Thus 0.5 x 13.2 = 6.6A thus being fine and which would have charged it in 2.5 hours.

Two guesses what I'm buying next ...

Measured Values

Now what's also interesting in the data presented by the 150A (which saves me doing the maths) is the Wh supplied to the battery. This was 410Wh.

Now this is interesting because as the battery is rated to 633Wh that's actually quite a bit over half discharge. Indeed its probably the ideal discharge to take it to to maximize battery life. As previously discussed (here) discharging the battery to about 60% (regularly, not as a one off) still results in a good 600 charge cycles before the battery is degraded to any significance.

which means that the battery should last me a good couple of years (all things being equal). That's very encouraging.

What can I do with this?

Well knowing that charge and discharge of Lithium batterys is in the high 90's efficient (nearly 99%) I thought it would be interesting to see if I can (like a Tesla onboard system does) use this to estimate range. As I've already established good models of how my energy demands are on my scooter with this bicycle computational tool (my parameters) over on this blog post, and that I have access to some good data on the steepness of things (via google as well as my own measurements) I can do some estimations if my calculations are right.

So having this data I used it to feed into my GPS log of the trip and make some basic ball park figures. This is the trip data from my phone GPS app;

Note, there is a flat spot on elevation because I got a phone call mid test and stopped to answer it. Then using that simulator (mentioned) I was able to roughly (ball park) divide the trip power demands up thus:

into different power level demands based on plugging that data for slope and speed into that simulator. Making some simple assumptions and measuring scale off the graph I got:

60 min = 532 pixels
1 min = 9pixels

High power period @650W = 148px + 90px = 238 = 26min = 281Wh
Standard power period @300W = 72px + 84 px = 156px = 17min = 85Wh
Only 43min in the hour was powered (because "gravity")
281+85 = 366Wh

Which when you consider that as reported above that charging required 410Wh that's pretty darn accurate.

Next, when you factor in the distance covered (as a challenging drive condition) I used 410Wh to cover 20.83km. Alternatively this works out to be 1.96kWh / 100km ... This is something we can now use to compare to many EV's such as (say) a Nissan Leaf which is something over 13kWh/100km

That's pretty darn good.

Outcomes?

this serves to verify a few things to me:

1. the battery pack rating is good for real world expectation (I don't think I'll do a "till fail" test, because walking in the heat is not attractive)
2. the battery pack is good quality
3. my empirical observation derived Coefficient of Rolling Resistance (Crr) seems to hold well in many circumstances

Pretty darn good I'd say