View Full Version : How to improve HCH2 battery management
10-09-2006, 01:52 PM
Does anybody else find themselves getting frustrated at the HCH2's choices for forced charging? I mean, for example, I have been climbing this fairly extended moderate incline for 4-5 miles, using some assist, but not a lot. I understand it will need to recharge, but why does it kick in 1-2 bars of recharge when I am at 6-7 bars, and not even finished my climb?
I find myself wanting to use more throttle, just to keep the car from charging! Why, you ask...? Because I have too often run into that pesky situation where the car has been auto-charging itself too high, and then when I do finally get to the slow, long off-ramp, or a couple of decent downhills, I end up using some friction brakes since the battery is full! :mad: ANybody else running into this, and what are you doing about it?
I have thought about trying to pulse a bit, and use some EV mode at 65-70mph. This would be frustrating, though, to me and morning traffic. I have been reading the threads on the MIMA project, and saying, 'Yeah, a little bit of that would be nice', since I am chastising my car more often than I'd like for it's difficult-to-predict charge management algorithms. And she's sooo nice, otherwise. Can anybody else provide some insight into this issue?
10-09-2006, 02:05 PM
I know that is one of Wayne's big beefs too. Forced regen usually occurs when your battery is too low (in the case of the hch I below 1/2 charge) Maybe try to stay off the battery as much as possible so there is no regen possible. Of course this means you will have more of the friction brake happening.
Thats my opinion anyway. I'm sure Wayne and the others have a better understanding of what to do.
10-09-2006, 02:49 PM
I understand Wayne's beef, certainly. Clearly, though it is an important feature of the transmission, and if you can't take advantage to use the charge that you are collecting by regen braking, then what is the point of the IMA? Yesterday while driving into PA for a short trip, I found myself noting all the situations where lights or stop signs are politely planted at the bottom of valleys. Those little guys are the poster children for regen braking, so every time I run into one and am using friction brakes I know I am essentially losing FE under those stops. ALthough, perhaps you are right, and if I used less assist overall during those long hills, it wouldn't go into forced charging? I dunno.
That's why mostly I am curious about what we know about the algorithm used by the software for forced charging, and then adjust my behavior to take advantage of it. Or perhaps my car is just a bit odd, forced-charging more than it should be what at Soc=5-7 bars?
10-09-2006, 02:52 PM
This is somewhat related. Once you get below 4 bars and start to get the forced regeneration, the car wants to stay in the regeneration mode even after 5 or more bars have accumulated. This behavior seems to stay for the remainder of the drive. One way around this is to do a FAS. This seems to reset things.
___I most certainly hear you loud and clear :( If it were me, I would find a spot in the far right lane while RR to keep out of assist at all costs during the climb. That forced charging scenario is an absolute killer and one I would avoid assist on any climb at all costs.
___If Mike finds the time, I am sure he could set you up with a relatively simple Assist/no-Assist switch. Having the ability to dial it in is a nice solution but if he knows what leads and voltages trigger an assist request, a simple relay with that leads running through it and the control wire back to a switch would solve this problem immediately. If I owned an HCH-II, I would have assist thing toggle out 98% of the time myself in fact!
___Someone was experimenting with the HCH-II’s Assist modes here at CleanMPG IIRC but I do not remember who or what exactly they were doing. I do not think it was a FAS switch (it may have been)? It was not that long ago but it went right along with Mike’s knowledge of the Insight’s IMA - inner workings. I tried searching for it and it did not pop up like I would have thought.
10-10-2006, 10:47 AM
Thanks for the insight, Wayne. It is indeed a difficult stretch near the end of my commute on the Garden State Parkway, where the speed limit jumps from 55 to 65, road shrinks to 4 lanes, and there is a (guessing) 500-700 ft climb over the course of 4-5 miles. I don't like to drop below 60, out of politeness in busy traffic, and desire to shave precious minutes from the commute.
Today I essentially tried what you recommended, paying careful attention to pull back if even one bar of assist showed up during the last 3-4 miles of the climb. But still it tried to force regen with 6 bars near the end. And I ended up using forced-friction-braking near a second long downhill coming to my work lot. ANd that was with trying to EV extensively after the first long off ramp. It just feels like avoiding assist is exacerbating the issue. Or maybe I should just be glad to start off the commute home with high SoC?
I found some info here, especially post #30:
On another greenhybrid forum, I found some conspiratorial whisperings about stealth assist/regen? Do you think this happens in all the IMA systems, where you may be undergoing very mild regen / assist that is below the threshold of 1 bar, but is still affecting the SoC and the inclinations of the software to forced regen?
Thanks for the info on a possible toggle, and I'll keep an eye out for that.
10-10-2006, 11:07 AM
In a way, there always has to be some kind of "stealth regen" in order to run the car's electrical systems. That's just the motor/generator working like the alternator on my car, which always spins with the engine to produce some amount of juice.
___Tarabell must have known exactly what we were talking about as she knew the exact post. This might be something worth following up on with MSantos.
Re: 103 miles per gallon for 47.8 miles. (http://www.cleanmpg.com/forums/showthread.php?p=11927#post11927)
___As Tim said, there is some hidden charging under the hood and you will see this when cruising down the highway in some cases. You will not see any forced charging but your SoC will creep up to 7 of 8 bars. The Insight did this anytime you were less then 19 and she would work her way back up to 19 each and every time without fail. There are times when the Inverter to the 12V is not live and at that point it is really smart … And in some cases, really really dumb :(
___Where did this “Forced Regen” nomenclature start anyway? There is Charge and Assist. Regen is a type of charge brought about by the brakes. Forced and hidden charging is brought about by the ICE and BCM.
10-10-2006, 10:14 PM
Thanks for putting me straight, Wayne. I've been a bit loose with my vocab :rolleyes: Forced charging it is. ANd know I know about the hidden charging.
That's a good link. I've also gotten a reply from MSantos in a GH thread:
So I see that it is probably an endemic IMA 'feature'. I am curious what else the experts around her know about the fine details of the BCM (Battery Cycle Management?) that I can judo to my advantage. Knowing... half the battle... and all that.
___I am sure you have already read Tarabell’s HCH-II article, Adapting Basic Hypermiling Techniques to the HCH-II (http://www.cleanmpg.com/forums/showthread.php?t=1306) as it is about the most succinct HCH-II specific article on the net wrt blowing out the EPA. Once you get into the forced stuff, there is a whole other realm but that one is for those of us that think hypermiling is a sport to play every time we turn the key :)
___The BCM is the battery control or condition module which supplies info to the MCM (Motor control module). It gets worse as you keep tracing down the line too ;)
___Great reading if you can stand it … BCM (http://www.**************.net/encyclopedia/enbcm.html). Follow the links for the rest of the chain and you will have a very good basic understanding of what IMA is and what it does behind the scenes.
10-11-2006, 02:35 AM
My observations from driving this car over the past 9mos or so...
You do, indeed, have hidden charge and hidden assist modes, as well as 'minor' and 'major' charge and assist modes.
On the assist side, you can have 0-7 bars. Most of the time, you'll see 3 on a moderate acceleration cycle (engine RPM < 2200 or so). At ~ 2500rpm, you'll see a 4th bar. Kick the engine into high valvelift mode (i-VTEC, generally at 3400rpm and above), and you'll start seeing max assist. I've seen this on one occasion - when I climbed a 5800' elevation (a mountain ?) in upstate NY on my way to Cooperstown over Labor Day. Not having experience at climbing significant grades, and having a 55mph limit on that grade from a dead stop at the base, I got to see what max torque at 3900 rpm and max hp at 4800rpm felt like - not a pleasant experience far from home.
If you accelerate lightly (under 1800rpm), you can drop to two bars of assist by the time you hit 25mph, and eliminate it altogether by 30-35mph. During this phase, you'll see 30mpg on the iFCD as you get past 20mph, 40mpg as you pass 30mph, and by the time you're going 42mph, you'll be at / above 50mpg on the iFCD. This is on fairly flat ground, of course. If you increase load between 30-40mph, the assist will pop back on, one bar at a time. After 42mph, you can increase load just a bit, and not see any assist bars. However, I do believe you will receive 'hidden assist' until you raise the iFCD above 50mpg.
After 45mph, visible assist under slight to moderate acceleration is rare. Rather, IMA is starting to look at forced charging - more on this in a bit.
On the charge side, you can have 0, 1, or 3 bars.
0 bars charge (or, hidden charge) occurs:
- When you're travelling at > 50mph, for more than 2minutes or so. The MCM assumes that you're going to be maintaining this speed for a bit, and tells the DC/DC converter "hey...you and I are not going to be working hard at all right now. Let's trickle charge the IMA battery to 80%". And so, it does so. Get on the expressway for > 10min, and you'll see a charged up traction battery in short order. If you're at a 'low' 5 bars of charge, you may even see one green bar for a couple of minutes - especially right after your acceleration cycle.
- When you put your headlamps on. For some reason, Honda has programmed IMA to force charge the IMA battery whenever your headlamps are on - they must be afraid that the extra electrical load will drain the 12v (and the 156v during auto-stops and the like), and so the DC/DC converter works to put charge in both batteries and maintain that charge at a higher level.
- After you finish with a moderate to long acceleration phase, running 3 bars of assist for 10-15 sec and then decrease load. You won't see the SoC increase, but there's some hidden charging going on.
You'll see 1 bar of charge under the following conditions:
- You have a longer acceleration phase of 25sec or more, and then decrease load. However, you'll see it *more* frequently with a *higher* SoC than a *lower* one. My thinking is that the charging algorithm is programmed to 'top off' if possible - so that with 6 or 7 bars on the SoC meter, it knows it just needs 'a little bit' to bring things back to 'optimum'. In a lower SoC of 'low 6' or '5 bars', you'll see the 'minor' charging less frequently. Almost as if the system says...I'll save a charge cycle (wear&tear) and let the SoC drop off a bit before I start charging again.
The SoC meter is approximate, as we all know - but it seems to be on a non-linear scale, like the gas gauge - the meter can stay on 5 bars of charge for a much longer time than it does on 6 or 7 bars (and at 8/8, the car becomes even more assist-happy than it usually is, because it's been 'overcharged' - sometimes intentionally - and the MCM is trying to get the charge back to 'optimal').
At a certain point, you'll not see visible charging until the meter drops to 4 bars. Then the fun begins.
At 4 bars of charge, the BCM reports to the MCM, "Hey...you've starved me long enough. I need to be topped off, and keep charging the batteries until I tell you to stop". MCM kicks the DC/DC into overdrive, and says "I'll kick into generator-only mode and supply you power". You say..."hey...where'd my assist and FE go?" Because you'll notice it takes a lot longer to get to speed, as there's no real torque from the tiny ICE alone, and you struggle to get higher than 40-45mpg on the iFCD - even at light load. At this point, just grit your teeth for 5-10minutes of lowered FE, depending on your speed / stop-n-go traffic. This is the 'major' forced charge (3 green bars) that we all dread seeing.
If you get to a 'low 4' bar scenario on the SoC, it'll get worse - you'll lose auto-stop, and you'll see 1 green bar of charge at anything higher than 10mph - but no visible bars below that speed or at idle. IMA does use a tiny bit of hidden assist to get you past 10mph, and then right back to charging once the RPMs are up a bit and the CVT can shift into it's next lower ratio (yes, I know it doesn't technically 'shift' like an A/T). You won't get auto-stop back until the SoC shows 5 bars, and you won't get traditional '3 bar' assist until you get to 6/8 on the meter. Rather, only one meager assist bar will show up between 5 and 6 bars SoC, during this charge phase.
To touch on the point that the original poster brought up - when you're at speed, and decrease the load on the engine, you can often drop the RPM back to 1100 or so, thanks to the CVT. The MCM sees this and says..."hey...there's so little load right now, I can use this occasion to charge the battery a little bit". And, thus, you see the 1 green bar of charge for up to 30sec or so, and the iFCD 5-10mpg less than you know where you can usually keep it. This is what he's referring to - he's worked to drive with as little load as possible, and the IMA 'took advantage' of this and robbed him of FE while slowing him down as well. So he feels he's been 'penalized' for driving as efficiently as possible, and he can't control the car to behave differently. Not yet, anyway.
Speaking of the CVT and the optimal FE ratio - if you can keep the car in the 38-44mph range, you can drive with the ICE in the 100mpg range (pegged). There may be some hidden assist going on here; I'm not sure. But the car is definitely not in the 'EV-glide' mode - you can feel when the ICE shuts off / back on under the EV-mode, and you'll notice that the ICE does not do this in this speed range under careful throttle control. I used to try and keep in this mode whenever possible. But I've learned over the last few weeks as I optimize my new work commute that the EV-glide (0 assist, 0 charge, ICE-off) is actually less of a chore to maintain, and the rewards can be greater, if you're smart with the opportunities (ie, downhill grades, 20-30mph zones, etc).
If I got the control sequence wrong in some of these phases between the components, my apologies. I also realize that this may have rambled a bit. But it's 2:30am as I write this, so I'll ask for leniency in this case... :)
One last thing - while the Civic's charging algorithms may leave something to be desired, they work better than in the Accord - there, you have an IMA battery that's basically used for auto-stop, if you drive the car for economy. But a heavy sequence of stop-n-go can drain that car's SoC down to 2/6 bars in no time (actually, ~ 16 stops in a short period would do it). You can initiate a lot more auto-stops w/o draining the battery in the Civic. It's a much more useful system in the Civic, both from an assist and an electrical systems retention / ICE restarting (auto-stop) standpoint.
10-11-2006, 05:23 AM
^^ terffic reading! Thank you for describing the way your hybrid works in detail, I enjoyed reading it.
10-11-2006, 08:13 AM
Wow! Very keen and useful observations. It certainly puts things into perspective for me, especially about the various levels of charge and assist that I was otherwise struggling to make sense of. It also makes me believe there's really not much else I can do about it. Except moving on and focusing on FE during other segments.
I imagine a number of others also have been paying attention during long declines, when they watch the battery get full quickly and perhaps they feel 'what a waste!' like I do.
You've captured some of my frustration, and I can certainly appreciate that the Civic is loads better than the accord in its algorithms. To an extent, I suspect that the algorithm is meant for the heavy-footed, and that's why it could use some improvement for us hypermilers. Uh errr, oops, not to boldly claim that I am a hypermiler, yet. Still have about 1k miles to go :)
Allrighty, well, I'll keep working at it! Hopin' to break 55mpg average by the end of the year! Let's hope for a mild winter...
___Excellent write up as usual!
10-11-2006, 10:27 AM
xcel, I would pillage small towns for an assist cutout switch. When I'm driving at a constant speed, I can get better MPG keeping my foot on the gas than using cruise control. That is completely illogical. :confused:
10-11-2006, 12:29 PM
This is a complicated problem! It sounds like the assist kill switch would be one good option for low load steady-state driving with a few hills. But I also think that the car would benefit from a much larger (Li-ion?) battery.
At the root of lost FE and "try again next time" segments is the fact that any time you move the car with electricity, you want it to have been stored during prior regenerative braking. Any time it comes from a force charge, you have to account for several stages of conversion loss that have taken place between the engine and the wheels. In essence, losing FE during said force charge is paying that cost up-front. A larger battery and more aggressive regen algorithm would allow for more efficient and more complete storage of the kinetic energy that you don't need anymore. No more riding the friction brakes down that long hill, and no more loss of assist when you need it most. And you can count on camping at 80mpg+ during a well-managed cruise!
After that, other options open up. What if you could run down that big battery to 20 or 30% during the day and top it back up to ~60 or 70% on grid power at night? Such a "mild PHEV" would be great for people who run a driving cycle that would typically demolish the current iteration of IMA. I envision a system that charges the battery at night but leaves enough room to capture energy from regen should you run into it early in the driving cycle. It would pretty much keep the same charge/assist habits as the current HCH, but would take much longer to hit that wall where the juice runs out. As it approaches the lower limits of battery capacity it would probably have to start acting more like the small battery, perhaps trickle charging early in order to make hard force charging a rare or non-existent event. Accomplishing that would be part software, part battery sizing.
If only we had the time and resources to try it...anybody at Honda listening?
10-12-2006, 12:48 AM
To clarify something... 'major' forced charging (where my SoC has dropped down to 4 bars) *is* a fairly rare event for me (avg 1x/ week). It happens to me as often as it does because:
- I drive efficiently enought to not have to use the footbrake very much, especially at the onset of slowing down. A 'more normal' driver would gain full regen much more than I do, and top off their battery easier.
- I live in a fairly flat area, and thus don't have many opportunities to have gravity provide regen (or, rather, low/no load charging).
- I can & do actually control my coast regen - through throttle manipulation, I can choose to reduce or eliminate it as my needs warrant. Again, much easier to do in the Civic than in the Accord.
Because of these things, forced charging becomes a 'failsafe'. Honda expected the 'usual' driver to make more use of the footbrake than most of us do, and a 'major' forced charge was not something they intended to make happen often. Those of us who drive efficiently & have topography advantages have learned to exploit the system; forced charging is the inevitable (?) result.
I actually think the current Civic's charging algorithms are pretty good, compared to the earlier iterations. This is actually "IMA-4", when you think about it - each of the hybrids Honda has brought to market has a different set. A lot of learning has occured on their end since the 2000 Insight.
Be happy the car is as efficient as it is, for what it is. It's a car with a very small engine for it's size with an electric motor for a torque boost. It's not an electric car with a gas engine to supplement charging of a huge traction battery. A Honda executive was quoted as saying he thought you'd get about 42mpg in real-world conditions in it. As we know, 42mpg is pretty hard to get in the car - unless you drive it inefficiently, as most 'regular' drivers do. Thus, his prediction.
FWIW, I get close to 54mpg on a tank-over-tank year-round basis in northern Illinois (and in the high 60's on certain trips, fairly frequently on my outbound work commute, which is downhill of my home). I don't use the 'advanced' techniques like FAS coasting, etc. I also use A/C when it gets hot in the summer. And there are people here who run rings around my comparatively humble accomplishments. I just drive the car efficiently as it was designed to be, and probably not as efficiently as some others here if they limited themselves to the same techniques as myself.
I'm off to bed for now, but we should discuss two other things at a later time that affect your SoC - the CVT, and your A/C system when it's in use. We've touched on the CVT a bit; the A/C is something I should have mentioned last night but forgot about. Hint - engaging the A/C compressor drops your SoC *in addition to* placing more load on the engine (because the electric half is the 'primary' end for the system). So you take a *double* hit on your FE by using it - immediate with the ICE, and later with the MCM force charging your battery. But it's worth it to me, not to sweat all day. There are a couple of things that do help - one is setting your temp to 80F - the car still keeps you cool at that temperature. The other is keeping your fan speed to 4/7 or below when your A/C is operating. Speed 5 and above kick the compressor on more - if you have a garage, try it out sometime - you'll feel / hear it turn on when you raise the fan speed (dependent on ambient temperature).
Copyright 2006 Clean MPG, LLC. All Rights Reserved.
vBulletin® v3.6.7, Copyright ©2000-2013, Jelsoft Enterprises Ltd.