GM Volt Forum banner

1 - 20 of 46 Posts

·
Registered
Joined
·
743 Posts
Discussion Starter #1 (Edited)
During my dinner with Volt engineers testing in Colorado I asked them about variations in the KwH used number on the enegry usage screen. Usually it would show about 10.4 KwH used before the ICE came on but on occasion, especially when driving in the mountains, the number was as low as 9.6 KwH. They took data from my car and informed me that my battery was performing at peak level with no degradation, but that they'd get back to me with a specific explanation for this phenomena. Today I got that explanation.

Sorry, it's been a while getting back with you on your concern with the your kWh sometimes coming in at 9.6 to 9.8 kWh used, but most of the time coming in around 10.4 kWh used.

Let me start by introducing myself, my name is K**** H******. I work with L*** R**** at the Milford Proving Grounds in Milford, MI. I am the lead calibration engineer for the Hybrid Powertrain Control Module 2 - HPCM2 Controller, the Electronic Control Unit responsible for calculating the displayed kWh number.

The way this number is calculated is based on the change of the battery state of charge and it's total usable energy. From a displays standpoint we only update the display when the vehicle is running.

Where this may be influenced to a lower than expected value is when there is a battery state of charge correction. While energy is moving into or out of the battery, we use a Battery State Estimation algorithm to determine the State of Charge. Our battery conditioning capabilities in the Volt are engineered to consistently draw the most from our battery while protecting its long term performance capability. After the vehicle rests for a while, we can more accurately determine the state of charge. If that correction causes the state of charge to drop a little bit, that drop will not be accounted for in the kWhs used because the vehicle is not running. This doesn't mean that during that drive you used less of the battery, or the battery's capacity is less, it just shows that there's been corrections in that battery state of charge estimation. Many factors can cause this to happen such as temperature changes and the amount of energy that goes in and out of the pack.. ie mountainous driving.

Personally, I find that I can cause this number to change after many short ev drives with several opportunities for State of Charge corrections.
This really makes sense to me because the variations all seem to come with mountain driving or temperature changes. For instance the other day I drove on the flats in 50 degree weather for 45 miles. The screen showed 10.5 KwH consumed and the ICE came on a few miles from home and I consumed 0.1 gallons of gas. This morning I took the identical drive when the temperature was 27 degrees, and 10.1 KwH was the reading before the ICE came on, and I consumed 0.3 gallons of gas. So, as he said, temperature, as well as mountains, can affect the reading. In both situations I probably consumed the same wattage, about 10.4, but different conditions gave different readings.

It is nice to know the reason for this variation.
 

·
Registered
Joined
·
161 Posts
Thanks Gary - great info. Another good insight into the level of complexity under the hood & behind the dash. Even 15k miles into ownership, I continue to be impressed by the thoughtfulness that went into design, and the real leap the Volt represents in automotive technology.
 

·
Registered
Joined
·
70 Posts
This still leaves a bit of mystery for me.

I drive the same route to work and back everyday, and use up the battery every day. I used to get 10.3-10.5 kWh consistently, and then it dropped to 9.8-10.0 a few months ago, with maybe only one instance out of 50 getting 10.4 kWh. The car pretty much sits for the same length of time while I'm at work. I guess it could be temperature, but what I don't get is that there was a sudden break-over that dropped me down to the lower numbers. I pretty much went from getting ~10.4 to ~9.9 overnight, and it stayed that way, and did not correlate with any major temperature changes or driving changes. When it got much colder for a week, there was no further change in the total kWh used from the warmer week before!

If it goes back up next year, I guess I'll believe that it is due to temperature changes, but I think it has more to do with battery pack load leveling and the microcontrollers maybe picking different cells/ranges to use for the charge buffer and top and bottom SOC. I'm still wary to see if it will turn out to be actual pack degradation.
 

·
Registered
Joined
·
743 Posts
Discussion Starter #6
This still leaves a bit of mystery for me.

I drive the same route to work and back everyday, and use up the battery every day. I used to get 10.3-10.5 kWh consistently, and then it dropped to 9.8-10.0 a few months ago, with maybe only one instance out of 50 getting 10.4 kWh. The car pretty much sits for the same length of time while I'm at work. I guess it could be temperature, but what I don't get is that there was a sudden break-over that dropped me down to the lower numbers. I pretty much went from getting ~10.4 to ~9.9 overnight, and it stayed that way, and did not correlate with any major temperature changes or driving changes. When it got much colder for a week, there was no further change in the total kWh used from the warmer week before!

If it goes back up next year, I guess I'll believe that it is due to temperature changes, but I think it has more to do with battery pack load leveling and the microcontrollers maybe picking different cells/ranges to use for the charge buffer and top and bottom SOC. I'm still wary to see if it will turn out to be actual pack degradation.
I had the same concern which is why the question came up. They did a data dump from my car and when they got back to Detroit wrote me saying,

What I can tell you from the data is that your battery capacity is the same as when the car was new, and your vehicle looks very healthy, even with the mountainous terrain in Durango.
I trust them.
 

·
Registered
Joined
·
218 Posts
Gary; my question is where the computer reading the KWH is located, in battery or car? The reason I ask is my battery was replaced a few months ago, before replacement I would see 9.8 to 10.5KWH after replacement I would see 9.0 to 9.9 KWH. If the computer is in the battery then the variation could be in calibration, otherwise it looks like I may have a lessor battery.
 

·
Registered
Joined
·
743 Posts
Discussion Starter #8
Gary; my question is where the computer reading the KWH is located, in battery or car? The reason I ask is my battery was replaced a few months ago, before replacement I would see 9.8 to 10.5KWH after replacement I would see 9.0 to 9.9 KWH. If the computer is in the battery then the variation could be in calibration, otherwise it looks like I may have a lessor battery.
Good question. Don't know the answer but just assumed it was in the car.
 

·
Registered
Joined
·
1,415 Posts
While energy is moving into or out of the battery, we use a Battery State Estimation algorithm to determine the State of Charge. Our battery conditioning capabilities in the Volt are engineered to consistently draw the most from our battery while protecting its long term performance capability. After the vehicle rests for a while, we can more accurately determine the state of charge.

Great Post with implications. Based on the above statement, it appears absolute SOC is based at least in part on a battery parameter that changes during regen/draw. Very likely HV Volts. The kWh is likely a display of SOC below charged max in 1/10 kWh increments. The OBDII port PID 015b has SOC at one byte or .063 kWh resolution. Net, 2011 Volt owners should be able to use that for a Torque App phone display of kWh used. (219-A)*.063 where 219 is the presumed fully charged SOC byte that is obtained by observation when fully charged, no drains.. Alternatively, the SOC in kWh is A*.063 for 2011's

The other implication is this is going on in the HPCM2 controller which is addressed at $7E4. Not the ECU where SOC is apparently mirrored at 015b.
 

·
Premium Member
Joined
·
14,156 Posts
Very interesting post!

Nice post.
Not sure I'm understanding what he's saying. I think he's saying that when you're driving the car the Volt uses an algorithm not a measurement to estimate the usage. That estimate, which appears on the display, isn't as accurate as a measurement and hence the display can be off. The measurement will occur when you turn the car is OFF and it sits for a bit but that measurement won't show up on the display because only the estimate gets displayed.

If this is the case I'm left wondering what happens if you restart the car after it's been sitting for long enough or the Volt to measure the SOC. Does the display now reflect the measurement or does it just continue to reflect the estimates? (I'm thinking that it just uses the estimate).

In this case I'm not sure why you'd bring up the measurement. The answer would just be that the display represents an estimate based on accounting which can be off in one direction or another.

Or maybe I'm just dense.
 

·
Registered
Joined
·
1,579 Posts
These things are sufficiently convoluted for even the brightest of us to appear dense :)

An accurate SoC cannot be performed while the battery is charged/discharged. The internal resistance of the cells varies with temperature and the temperature of the individual cells probably varies somewhat within the pack. A quiescent point at some reference temperature is needed. This is probably why when parking the car there is one range estimate, and when coming back after a while there is another indicating an actual measurement, often higher, indicating the estimate is rather conservative.
 

·
Registered
Joined
·
743 Posts
Discussion Starter #12
Very interesting post!

Not sure I'm understanding what he's saying. I think he's saying that when you're driving the car the Volt uses an algorithm not a measurement to estimate the usage. That estimate, which appears on the display, isn't as accurate as a measurement and hence the display can be off. The measurement will occur when you turn the car is OFF and it sits for a bit but that measurement won't show up on the display because only the estimate gets displayed.

If this is the case I'm left wondering what happens if you restart the car after it's been sitting for long enough or the Volt to measure the SOC. Does the display now reflect the measurement or does it just continue to reflect the estimates? (I'm thinking that it just uses the estimate).

In this case I'm not sure why you'd bring up the measurement. The answer would just be that the display represents an estimate based on accounting which can be off in one direction or another.

Or maybe I'm just dense.
Dense? Certainly not as dense as I am. But your question is easy enough to test, both with it sitting a while or sitting after a partial charge, which does not zero the numbers. I'll be doing that.
 

·
Registered
Joined
·
1,415 Posts
The display kWh reading most often commented on is that read when the ICE comes on. The ICE trigger on is based on SOC....<21% or whatever. Consequently the kWh estimate seen at that point can wander due to the measurement of SOC (under heavy load) or because the display is estimated. Obviously, without data this is more theology than fact.

A key point here is that the display is not an integration of an actual amps*volts measures. Given a "settling time" exists for SOC, I'll bet that the integrated real value would be all over the map given the variability of draw and regen currents and temperature.

Anybody with a Dash Daq know?
 

·
Registered
Joined
·
4,728 Posts
Not sure I'm understanding what he's saying. <SNIP>Or maybe I'm just dense.
DonC, we’ll have to leave the subject of your density for another thread. lol ;)
This topic has been discussed numerous times, in numerous threads and the GM engineering response here does provide some additional clarity most notably that active auto-corrections in existing SOC estimates is primarily responsible for any observed variations in the kWh display. In short these variations are due to a combination of factors, but most notably how and when pack SOC is being established, the accuracy of pack voltage-current measurement and the method/s of integration used.

The state of charge of a Li-Ion cell cannot really be directly “measured” by any non-chemical means, (by some sort of “sensor” for instance) it is therefore always an estimation. An estimation based on different sets of factors that are dependent on a wide range of conditions that might exist when that battery is being utilized by a plug-in electric vehicle. An additional complication is the physical construction of the pack, the number of cells and their parallel and series connections to each other as well as the limitations of the BMS itself (most notably processing power and the precision and shear number of the necessary input channels for collecting the required data such as temperature, voltage, and current)

There are in fact many hundreds of research papers, and many dozens of texts where various methods and formula have been created in an attempt to accurately estimate existing SOC in a multi-celled BEV or HEV battery pack.
GM uses their own patented methodologies and algorithms that were/are developed in-house, usually in cooperation with the battery cell OEM (in this case LG) that will utilize and actively introduce numerous algorithmic variables, (with and without various Kalman filtering methods applied) in order to be as accurate as required for the desired use. (more on that in a bit, but obviously end-user instrumentation can typically be many times less accurate/precise than what might be necessary internally within the BMS)

The simplest non-chemical method to estimate SOC is to measure cell open circuit voltage-Voc (i.e. when no current is flowing in or out of the cell) as there is typically a known direct relationship to SOC that can be derived specifically from this voltage measurement ( and usually established via chemical based corroboration). But this relationship however is not linear and complicated by pack structure and therefore may be limited the necessary number of voltage input provided. For instance since the Volt uses 96 cell “triplets” that are connected in parallel, (and assuming a good state of health) then in 96 direct measurements must be made and an average Voc established that once compared to the model (that also factors in existing cell temperatures present at the time of the measurement- an additional variable) a very accurate estimation of SOC can be made. Most importantly this estimation must be made to the high precision necessary for the powertrain controls to properly utilize it as an accurate “starting point” of SOC (post charge) at the commencement of a trip cycle.

Once the car is “ON” however (and more notably providing significant current necessary for propulsion, thermal management, and cabin conditioning) it becomes near impossible to accurately estimate actual state of charge merely through observed voltages due the variation in electrical loads placed on the battery pack. Thus the existing depth of discharge (DoD) present during CD mode will significantly affect the measured cell/triplet voltages, so at that point a different method if estimating SOC must be employed. This method combines the observed triplet voltages with a measurement of current flow both in and out (as most everyone knows the center stack display ignores the "ins") of the pack via an inductive current probe/sensor. By simply observing amps in vs out this input permits the HPCM2 to accurately estimate existing SOC while the car is active by essentially “counting coulombs” and calculating the net effect relative to that previously established SOC starting point.

So yes, a much more complex (and very active) set of algorithms is utilized to estimate the SOC while the battery pack is actively delivering or receiving power throughout a typical drive cycle. Which in turn is matched against and modeled after various Kalman-type filters are applied in order to achieve the desired precision necessary for controlling primary SOC dependent behavior such as the switch from CD to CS or the ICE run set points for mountain mode. However, even while ON there will be times where a more accurate detection of existing SOC can be made, and if significant enough, a correction CAN be made that would therefore be reflected in the displayed kWh readout. So in short the displayed value may vary depending on the number and levels of these corrections that took place uring the trip cycle.

Keep in mind, the instrumented kWh value displayed on 2012+ center stack, was merely an attempt to quantify specific “cherry picked” aspects of these algorithmic calculations into something that might be interesting or prove useful to the more tech-savvy Volt owner and was actually implemented in direct response to numerous requests by many of the Volt’s early adopters. However I doubt anyone at GM expected it to come under such scrutiny by the Volt community. I guess they should have anticipated the obvious response would be an attempt to establish an actual practical USE for this data (over and above the singular energy event it was meant to portrait) But as discussed by the GM Powertrain engineer in the OPs post, this displayed kWh value is subject to specific data collection variables and potential correction factor adjustments that are very much usage dependent.
Therefore (and this echoes what I have stated in other threads on this topic) this kWh value should NOT be used as a method of comparison between trip cycles or between different cars, nor can it be used to estimate or predict such items such as pack degradation or End-Of-Life (EOL).

It is after all merely an approximation of total CD energy delivered by the battery during the last drive cycle (as a function of the last charge event) and meant to give the Volt owner some idea what amount of energy was used during the last trip cycle - that is all.

HTH
WopOnTour

EDIT: hmm after pounding this out and posting this as a response to DonC, THEN reading it back I have come to realize this might be the worst case of verbal diarrhea I've ever created! :mad: Oh well hopefully it will somehow makes some sort of sense... lol
 

·
Registered
Joined
·
186 Posts
So if I get this right.....My 2012 gets only 9.7 Kw before the ICE starts. The reason from what I am reading here is that I don't get the higher numbers because I drive almost entirely at highway speed and that type of draw affects the reading, but my battery is still getting the most available? If I drove at city speeds for a while( a few charges?) would my number go back up?
 

·
Premium Member
Joined
·
14,156 Posts
However, even while ON there will be times where a more accurate detection of existing SOC can be made, and if significant enough, a correction CAN be made that would therefore be reflected in the displayed kWh readout. So in short the displayed value may vary depending on the number and levels of these corrections that took place uring the trip cycle.
...
But as discussed by the GM Powertrain engineer in the OPs post, this displayed kWh value is subject to specific data collection variables and potential correction factor adjustments that are very much usage dependent.
So ... long story short, depending on the drive cycle the center stack number for kWh used may be more accurate or less accurate and we shouldn't obsess about the exact number because it's not designed to be that precise.
 

·
Registered
Joined
·
6,464 Posts
The state of charge of a Li-Ion cell cannot really be directly “measured” by any non-chemical means, (by some sort of “sensor” for instance) it is therefore always an estimation.
Bingo. I think you could have summed it up there. The response from GM was spot on and it explains everything. Frankly, I'm surprised none of the forum members here thought of this (maybe they did?) as the reason for the variation in kWh displayed. I remember in some GM literature, somewhere, that they stated that they could not measure the SOC accurately because there is so little sag in battery-cell voltage over the standard operating SOC, therefore, they had to create an algorithm to "estimate" the battery's state of charge.
 

·
Registered
Joined
·
1,916 Posts
Very interesting post and confirms my observations. I originally assumed the KWh used was a directly measured value but observations led me to believe otherwise. It is estimated from battery SOC measurements and correlated to the usage bars. This aligns with observations that the middle bars and middle KWh's are often "fatter" than the first and last.

Personally, I'ld rather see actually measured KWh usage and have the estimate range and bars come from the algorithms but GM is leaps and bounds above the competition in the SOC algorithms and range estimates. I can live with their KWh usage estimate.
 

·
Registered
Joined
·
172 Posts
The energy used (kWh) is measurable, not an estimate. The SOC is the value that the car is estimating. That's what the GM engineers were saying.

...Personally, I'ld rather see actually measured KWh usage and have the estimate range and bars come from the algorithms but GM is leaps and bounds above the competition in the SOC algorithms and range estimates. I can live with their KWh usage estimate.
 

·
Registered
Joined
·
1,916 Posts
The energy used (kWh) is measurable, not an estimate. The SOC is the value that the car is estimating. That's what the GM engineers were saying.
Perhaps I'm misreading but I disagree. There certainly a lot of measurements being taken and used in their calculations but they are using the SOC in determining the KWh used, or at least that is how interpreted this comment that was in response to the used KWh amount displayed:

The way this number is calculated is based on the change of the battery state of charge and it's total usable energy. From a displays standpoint we only update the display when the vehicle is running.
 
1 - 20 of 46 Posts
Top