[Eli]

If I take 20 new matched sticks and artificially induce extreme SoC imbalance and then assemble the pack and drive, what will happen?

Six sticks, each discharged and then charged to:

100mAh (just so it wasn't at 0% SoC)
1000mAh
2000mAh
3000mAh
4000mAh
5000mAh

All the other 14 sticks are charged to full. Here's what it looks like, input SoC wise:

Stick	Charge-based SoC
1	1.58% (100mAh)
2	15.45% (1000mAh)
3	30.82% (2000mAh)
4	46.20% (3000mAh)
5	61.57% (4000mAh)
6	76.97% (5000mAh)
7	99.34%
8	100.11%
9	100.12%
10	100.12%
11	100.14%
12	100.29%
13	100.42%
14	100.63%
15	100.71%
16	100.72%
17	100.80%
18	101.37%
19	101.40%
20	102.82%

Pack is assembled so that no two of these different SoC sicks are on any single tap.

This is being performed in my daily driver, which clocks a whopping 12 miles a day on average. Full disclosure: I've been driving a 2004 Honda Civic Hybrid. The same thing will happen in an Insight though. So to that point, there is no MIMA or anything being used here, totally stock.

I will be testing the battery at the time frames presented; 10 days, 30 days, 60 days and 90 days. After discharge to determine actual SoC, I will re-charge each stick to that nominal SoC and put it back into the car until the next interval.

Will all of the sticks ever reach the same SoC? Conventional wisdom says no, that the BCM "does not balance".

Discuss.

 

[Eli]

Ack.. didn't mean to post yet. I was still trying to edit the poll and everything, somehow clicked post. But that works....

 

[olrowdy01]

Can the car recover from artificial imbalance? How long will it take?

Please define "artificial imbalance". Normally we get the real thing (an unbalanced battery).

If you wait long enough the IMA battery will become -balanced- not imbalanced. As an example I found an Insight that sat outdoors unused for 10+ YEARS and the battery was [sort of] balanced.

The only bad thing was the -total- battery pack voltage was 0.816 of a volt after measuring the sticks and adding the voltages!

After two grid charges and discharges the battery was electrically balanced. Unfortunately the car couldn't be started or driven to see how well the pack would perform.

 

[retepsnikrep]

In daily use in the car it probably depends on the amount of forced charge that occurs before a positive recal after a negative one. It will likely be multiples of that figure. Each forced charge likely brings them slightly closer together.

I've never tried such an extreme imbalance. Be interesting to see your results/thoughts.

I don't think the answer is in terms of days with new sticks so haven't voted, but might be in terms of recal numbers.

If the sticks are just left to rest then they will eventually balance due to self discharge, but that will take a very long time with new sticks.

 

[Eli]

Sorry if I wasn't clear, I totally did not mean to post the thread when I did, I was still in the early stages of figuring out how to present the question and data. I was actually going to post after getting all of the data. I've been taking advantage of the new forum software to compose my posts rather than in Word. Somehow I clicked post, and then had to quickly put something up before I tended to a Prius install. I didn't do a good job of laying this out at all. We'll call this a test run. To Peter's point, maybe miles would be better than days?

So by imbalance, I mean SoC imbalance of the type that typically causes a pack to code. And by artificial imbalance, I mean that I've artificially induced a very wide SoC difference among the 20 sticks, from basically 0% to 100%, by stopping their charge at the above pre-defined intervals before assembling the pack. That's as opposed to this happening slowly over the course of the life of a battery.

By "Balance after X days", I mean that upon removal from the car and discharge of each stick, they will all be at the same SoC. Basically the question is, how well will they recover, if at all? Obviously a pack full of 0% SoC sticks is not functional, so I don't mean balance through self-discharge or sitting. I mean "high SoC balance"; the way a new pack comes.

This is being performed in my daily driver, which clocks a whopping 12 miles a day on average. Full disclosure: I've been driving a 2004 Honda Civic Hybrid 5 speed. The same thing will happen in an Insight though, BCM behavior is basically identical(to early Insight computers). So to that point, there is no MIMA or anything being used here, totally stock.

I will be testing the battery at the time frames presented; 10 days, 30 days, 60 days and 90 days. After discharge to determine SoC, I will re-charge each stick to that nominal SoC and put it back into the car until the next interval.

Will all of the sticks ever reach the same SoC? Conventional wisdom says no, that the BCM "does not balance".

I've edited my OP with some of these details.

 

[eq1]

My bet is that you get a P1449-78 code -- that's what should happen, right? Does that end the test or do you then over ride that and continue in some fashion?

Once you get beyond the trouble-coding -- so once you get at least about 10-20% in the lowest stick -- I'd guess it'd take no less than 43 days/drives... But 12 mile drives per day isn't a lot, that 43 day minimum would require that you max-out the charge state per drive... Not sure how/how long it'd take you to get that 10-20% in the lowest stick though...

I'd guess you'd lose about 300mAh per day in the high sticks, in the 75%+ SoC range... That would mean you'd be down to a true 75% or so after (25 X 65)/300mAh= 5.4 days...

And let's say you'd then be able to recoup about 5% per drive/day in the low sticks before pos recal. So it'd take another ~ 4 or 5 days to get the lowest stick to 20%...

So, at 10 days you might reduce the disparity between the lowest 0% and highest 100% to 20% and 75%. I think that'd be the best case scenario. This assumes you've done something to circumvent the trouble coding, like pulling the fuse or battery neg cable...

Then you might be able to achieve say 150mAh per day/drive balancing. So 20 days X 150mAh=3000mAh. By day 30 you've reduced the disparity between low and high (20% and 75%) to 66% and 75%... That seems really optimistic...

I think I might want to double all these times, so it'd take like 20 days to get to 20/75 instead of 10, and it'd take another 40 days to get to 66/75...

I don't know, complicated problem...

 

[Eli]

Fascinating analysis. You were definitely one of those I was thinking of when I set out to perform the experiment. I'm actually going to perform the 30 day test next weekend. Again, did not plan this post out very well........ lol

When I initially put the battery in, I experienced a negative recal and was careful to baby things before the positive recal so as to avoid an error light. But yeah, I'd have just reset and carried on, noting them.

Had I put two of the lower SoC sticks on any single tap, it would have been virtually impossible to avoid and I would have likely experienced a light or three. Have not experienced another negative recal since. If I was draining the battery down low and keeping it there, this would not work.

Peter and eq1's comments make me realize that I should comment on my driving style. I am definitely driving "for the battery", as opposed to "for fuel conomy". Basically, driving like I have a bad battery.

My route into work is just typical city traffic, I usually park at the shop around 75% BCM SoC and 170V. I've found a route that has some decent hills for the commute home; I'm able to get the BCM SoC down below 65.0% before getting in some good regen. If you don't get it down to 65.0% or below, it won't hang at 72.0% on the way back up. So I have been doing one of these 72.0% hang cycles per day. Typically, I've been getting home with the BCM SoC at 80% and the battery at ~180V. It's been so cold here that I only ever see 35-40A of regen, peaking out around 190V. Lowest I've had the BCM SoC is 51%, but usually just below 65% before starting aggressive regen back up to 80%.

In this regard, I am definitely "gaming the system", to make it happen as quickly as possible. But only in the sense that I know how to drive to maximize the passive balancing that naturally occurs with NiMH. Other people do too, whether they know it or not.

Here's the 10 day data, as a teaser:

Stick	10 Day Disch	SoC	SoC Gain
1	1685	25.92%	24.34%
2	2476	38.09%	22.65%
3	3365	51.77%	20.95%
4	4249	65.37%	19.17%
5	4934	75.91%	14.34%
6	5392	82.95%	5.98%
7	5594	86.06%	-13.28%
8	5626	86.55%	-13.55%
9	5583	85.89%	-14.23%
10	5567	85.65%	-14.48%
11	5590	86.00%	-14.14%
12	5518	84.89%	-15.40%
13	5601	86.17%	-14.25%
14	5617	86.42%	-14.22%
15	5659	87.06%	-13.65%
16	5618	86.43%	-14.29%
17	5592	86.03%	-14.77%
18	5612	86.34%	-15.03%
19	5615	86.38%	-15.02%
20	5645	86.85%	-15.97%

 

[eq1]

^ Hey, I was kind of close - 24/86 is close to the 20/75 I guessed for day 10... I was thinking that one tap was at zero though, it didn't register that you were pairing a zero stick with a full stick. And other things...

Interesting that all your full sticks end up at ~86%. I recall that's what I was estimating the car's '75%' SoC was at.

So, what exactly is going on here? Is the balancing happening more because 1) the high sticks lose more charge (high charge state self discharge) or because 2) they charge slower/less than the others?

In other words, at the end of your daily drive you leave the pack at a high charge state and the high sticks lose more to self discharge than the low sticks, so the next day you're able to charge the low ones more simply because the high ones are no longer butting-up against the pos recal point, and the low ones gain ground... That's idea 1). Idea 2) is that you're able to charge the low ones more simply because they have lower resistance at a lower charge state... My own sense is that 1) should be the more powerful balancing force, that 2) doesn't do much. But I don't really know...

 

[retepsnikrep]

Interesting 10 day data.

I assume now they will get closer together again in the next ten days.

 

[Eli]

It's a combination of both cell efficiency and self-discharge that causes the balancing effect. Efficiency goes down and SD goes up with increasing SoC (especially above ~80%). Efficiency shouldn't be confused with IR. The IR difference measured between sticks at 10 days is fairly negligible (0.25 milliohms) and doesn't really have much to do with SoC. Efficiency would be around 96% for the lowest SoC sticks and 88% for the highest SoC sticks, if you were to measure it directly.

When I start driving in the morning, BCM SoC is still at 80% and the battery is down to around 166V. I have to get the BCM SoC down below 75% before regen is allowed again. During regular city driving on my way in, the BCM SoC basically oscillates between 75 and 80% between each stop light. At night, there is no traffic and I can play around a little more...

 

[Jonny69]

Interesting. So did it not throw a P1449 (74?) at the beginning based on the fact that the sticks were quite obviously out of balance?

 

[eq1]

^ In general I think it would have been a P1449-78 -- not able to charge the pack from empty more than about 10% because some taps reach full too soon. Not sure if this can happen directly after installing an imbalanced pack from a 'reset' state (like fuse pull or neg cable pull) or if you need to get a neg recal first. But in any event, Eli was able to charge enough, gingerly, to avoid the trouble coding...

It's a combination of both cell efficiency and self-discharge that causes the balancing effect. Efficiency goes down and SD goes up with increasing SoC (especially above ~80%). Efficiency shouldn't be confused with IR. The IR difference measured between sticks at 10 days is fairly negligible (0.25 milliohms) and doesn't really have much to do with SoC....

So, the 'efficiency going down' at the top of the charge state range doesn't dovetail with higher resistance? Don't the two go hand in hand, like two sides of the same coin? For example, if you apply a current to a cell charged to the high charge state range, voltage will go up more and faster, right?, than with the same current applied to a different cell at a lower charge state. Doesn't the higher, faster voltage increase necessarily mean high resistance? Or 'impedance'?

Or maybe I'm/we're being too general with our use of quote "IR," that "IR" is a catchall phrase we tend to use that really encompasses a few different things... Like "activation overvoltage," "concentration overvoltage," and "internal resistance" proper, where latter means just the resistance of the hard-wired stuff in the cell, like plate welds and the like...

edit: Actually, let me just add one thing here. In the 'Civic Battery paper', they basically reduce 'IR' to two things: "polarization resistance" and "resistance internal," where resistance internal is the hard-wired stuff. Polarization resistance depends on only a few things -- a couple constants, temp, exchange current density, and electrode surface area. All these except electrode surface area and maybe current density are more or less irrelevant in this high charge state/balancing test context. So, it seems like what's relevant at the high charge state is that the electrode surface area effectively decreases. And/or the current density increases. I.e. what's happening is that the polarization resistance increases...

 

[Eli]

Hmm. That's a good question and an interesting perspective. It's not IR that causes the steeper voltage slope at higher SoC though, it's chemical stuff.. that I wish I knew more about. Like I said, the IR spread between the sticks at their 10 day state of charge level was a nominal 0.25 milliohms, and there is no correlation to SoC.

It's true that cell IR is bathtub shaped and goes up at low and high SoC, but it isn't that apparent when you are measuring IR in a static state. Except at the very extremes, a given cell should measure it's nominal IR at all times.

It will probably be faster for me to illustrate with Paint than search for actual examples...

So a rather poor and extreme illustration of the voltage curve of two different sticks, with wildly different IR. Try and imagine both lines are the same, just offset. The (arbitrary) numbers are what you might expect if you measured IR at those points.

Edit: And yeah, if we're not all talking about the "same" IR, then that can be problematic too. In this context I mean either AC 1000Hz impedance or current step method. Current step method tends to emphasize high and low SoC IR differences compared to the AC 1000Hz method. Both have their merits.

 

[eq1]

My guess is that, whatever or however you make your measurements probably doesn't or wouldn't capture the 'polarization resistance' increases at the high charge state... In general it sounds like you have a specific definition/conception of 'IR' that doesn't include the 'stuff' happening at the top end. Rather, we're calling that stuff "inefficiency." Which makes sense.

After a little skimming of that paper and thinking about it a bit, I think I kind of get what's going on. I mean, it's really pretty simple in a gross-conceptual way. The more charged the positive electrode is, the harder* it is to pull protons out of it (i.e. to charge it). I'm pretty sure storing those protons in the negative electrode isn't the limiting factor... Transferring them through the electrolyte might play a role. But simply not being able to pull those protons out of the pos electrode as easily is like having the faucet turned down -- and the pressure, i.e. voltage goes up. My guess is that quick pulse-like 'IR' measurements wouldn't capture much of this process, it probably takes too long, is cumulative, etc...

* I don't actually know that it's harder per se, but rather, there's just a lot fewer protons to pull, a lot fewer locations/opportunities at which charging can take place. So, given a fixed current, these fuller cells will obviously charge less/slower than the lesser-charged ones -- and the lesser-charged ones can catch up...

 

[eq1]

Eli, just curious: How many sticks can you discharge/charge individually but at the same time? I was thinking that your testing is cool - but that it'd be a ton of work if you couldn't 'do' more than a single stick at a time... I'm often deterred from doing as much testing as I'd 'like' simply because it's such a chore to break the pack down and work with sticks.

 

[Eli]

Around 160...

I wouldn't be able to do stuff like this otherwise, it would take an eternity and a half. Not to mention rebuilding batteries....

 

[eq1]

^ Very nice... It's still some work, so, glad you're doing this. Looking forward to the 30 day numbers...

 

[Harborman]

How long does it take for the hybrid battery to fully discharge? My car is stored all Winter. Can I assume it will be completely discharged - with all the sticks at an equal (very minimal) voltage? Thanks

 

[Eli]

That's a very tricky question. Ideally, it takes a very long time - years. Such is the case with new cells. Sitting for a winter doesn't do the pack any favors, but it shouldn't be catastrophic.

That goes out the window once you begin using the battery. Even after only a year of use, some cells can start to be different from others. One of those differences can be self-discharge.

So to answer your question - no, you should not assume that it will be completely discharged after sitting for a winter. It very likely won't be. At the worst, you will have sticks that have self-discharged to a very low voltage while others have barely self-discharged at all, which will result in error codes. I suppose it would be marginally better if they all self-discharge down to a very low voltage, but would still result in codes. At best, they will have all self-discharged exactly the same (small) amount, and you can pick up right where you left off.

 

[Harborman]

OK thanks... I thought I’ve read other threads that have indicated a complete discharge occurs after sitting for several months... I better not assume that! I’ll plan to do a full grid charge before driving next spring. (Do you recommend discharging prior to charging?) Thanks