90D Range slowly declining

[Max*]

MAX*, the 90's have a totally different traction pack called the Flexpack (there is a thread that shows the 90 part number and a part's list which gives it that name). The 90D and P90D and P90DL (and possibly even the X 90's) each have the same pack part number. It is possible that this new traction pack or battery management system differs from the 85 packs in form and perhaps even features. Of course we do not know this to be a fact.

I understand that, and the whole silicon anode, but that doesn't explain why (based on MarcG's speculation) that the external weather be used in the rated range algorithm ONLY for the 90kwh model.

I would assume [if his speculation is accurate] that the rated range algorithm would change for ALL battery types (60-90kwh).

And what I'm saying is that I don't think post-v7 algorithm takes weather into account for rated range based on my non-90kwh battery.

 

[msnow]

Thanks! Some data is better than none..

I've been thinking a little bit about this recent rated range drop, and had a thought while taking a shower this morning. It's somewhat playing devil's advocate, but please hear (or read) me out of a minute:

I know that *historically* the rated range has NOT been impacted by ambient temperature. HOWEVER seeing that the weather HAS gotten colder, and in my personal experience over the last couple of weeks it's been in the low-mid teens in ºC (50's in ºF) whereas it's been in the 20's ºC (70's ºF) before, AND that most of us are now on some version of Firmware 7.0 - could it be that Tesla has updated the displayed rated range algorithm to now take into account recent ambient temperatures??

The weather here in SoCal got down to the high 50's but that just started a few days ago. Prior to that it was in the 85-100 degree range and my biggest range loss was during the very hot weeks. Based on what I'm reading, the SC techs are all over the place on this. Some say it's normal to lose 2%-3% in a month so don't worry about it, some say it's not really happening (just a little algorithm thing), some say weather, driving habit or a combination of all of these. Doesn't make sense to me.

 

[Canuck]

As I see it, there are only two things can answer the questions of whether the reduced range people are seeing is actual degradation or miscalculations by the algorithm: (1) Time -- for Tesla to play around with, and get right, the 90 algorithm; or (2) Someone with a 90 battery and reduced range at 90% to charge to 100%, then drain to about 5 or less miles, charge to 100% again, drain to about 5 or less miles again, then charge to 90% and see if the lost range is back.

If anyone can think of anything else to determine this issue, please let us know.

 

[BertL]

The weather here in SoCal got down to the high 50's but that just started a few days ago. Prior to that it was in the 85-100 degree range and my biggest range loss was during the very hot weeks. Based on what I'm reading, the SC techs are all over the place on this. Some say it's normal to lose 2%-3% in a month so don't worry about it, some say it's not really happening (just a little algorithm thing), some say weather, driving habit or combination of all of these. Doesn't make sense to me.

I agree. Does not make a lot of sense. Plenty of well meaning folks here are trying to help with speculation, but IMHO none of us have sufficient facts to create a decent hypothesis of what is going on with our 90Ds quite yet. I'm keeping my own rated range log, including mean temps each day I record an entry, and trying to check about the same time each morning. With only 5 weeks of ownership, there is nothing that adds to this discussion. When I have a few more weeks of data points, or some pattern begins to perhaps emerge, I'll post here.

My one ah ha data point (at least to me) this morning that gave me confidence rated range is just an approximation as many people keep saying, and/or there is rounding involved in how the number is presented: at a 90% charge my MS did starting at midnight (from over 100mi remaining rated range), I had 253 rated miles when it was complete and I checked at 4:30am this morning. At 8am when I backed my MS not 15' out into the driveway for a wash, it showed 254 when I got out and closed the door. With only opening/closing doors a few times, my MS was back to 253 on the IC when I pulled it back into the garage a little after 9am. Temp in my backyard went from 49 at 4:30am to 66 at 9:30am during all that time. ...nothing conclusive, I know -- just more tinder for the smoldering fire.

---

...and I know I'm not going crazy, and it's too early for a glass of wine, but I just happened to open up the Tesla App on my iPad (2 hours after writing the above...) Rated range showed 254 with charging scheduled for 12M as it should. Not 5 seconds later, it changed to 253 (OK, I know the app may not have yet talked to the MS), then - few seconds later went down again to 252. While I was looking at the screen in a bit of amazement, it went back up to 254 and has stayed there for a full 5 minutes. I just went out, opened the MS door, and rated range is 254. It's now 73F in the garage. ...so more tinder for the fire with a rated range that evolves across a 30 sec period, while MS is off, parked, and sitting in the garage plugged into its HPWC.

 

[apacheguy]

As I see it, there are only two things can answer the questions of whether the reduced range people are seeing is actual degradation or miscalculations by the algorithm: (1) Time -- for Tesla to play around with, and get right, the 90 algorithm; or (2) Someone with a 90 battery and reduced range at 90% to charge to 100%, then drain to about 5 or less miles, charge to 100% again, drain to about 5 or less miles again, then charge to 90% and see if the lost range is back.

Agree. Someone with a 90 needs to do (2). But one cycle down to < 10% and then back up to 90% should be sufficient to trigger an algo recalculation. I don't see the need for the second one.

 

[Cottonwood]

Agree. Someone with a 90 needs to do (2). But one cycle down to < 10% and then back up to 90% should be sufficient to trigger an algo recalculation. I don't see the need for the second one.

Also, after the 100% charge, wait a couple of days for the pack to balance. On a 100% charge, it checks how out of balance the pack is, then starts balancing, but the little resistors that are used to do the balancing can only drain 0.1 Amps on each parallel set of batteries, that are over 200 Amp-Hr. Even a 0.5% out of balance is over 1 Amp-hr that can take over 10 hours to balance.

You don't have to, and should not leave the pack at 100%; just wait the couple of days after the 100% event for the pack to balance itself.

 

[sorka]

I don't think you need to charge to 100%. The bleeders start balancing at 93%. I wouldn't charge up to 100% and leave it for days. Not good for the battery. Charge to 95 or 96% and leave it for a day and then discharge to below 90% and then recharge back to 90%. If that still doesn't do it, run it down to 10% and recharge to 90%.

 

[Cottonwood]

I don't think you need to charge to 100%. The bleeders start balancing at 93%. I wouldn't charge up to 100% and leave it for days. Not good for the battery. Charge to 95 or 96% and leave it for a day and then discharge to below 90% and then recharge back to 90%. If that still doesn't do it, run it down to 10% and recharge to 90%.

I'm sorry if I was not clear. You don't have to leave the car at 100%, 93%, or somewhere in between for a couple of days. You only have to take the car above 93% (100% is above 93%) and then back down to a regular charge state so that the car can make the imbalance measurement. Then you need to wait a couple of days (worst case) for the bleeder resistors to have time to use those imbalance measurements to do the balancing.

 

[apacheguy]

Yeah, the balancing thing is worth a shot but I guess I just don't see how all of these 90 kWh can be out of balance with only a few cycles on them.

 

[sorka]

I'm sorry if I was not clear. You don't have to leave the car at 100%, 93%, or somewhere in between for a couple of days. You only have to take the car above 93% (100% is above 93%) and then back down to a regular charge state so that the car can make the imbalance measurement. Then you need to wait a couple of days (worst case) for the bleeder resistors to have time to use those imbalance measurements to do the balancing.

Referencing wk057's post on balancing:

Basically there are a couple of things at work causing the low range display on some cars.

1) Energy usage calibration.

This is probably the biggest offender. The car will keep track of energy usage and cell voltage, along with estimating losses due to internal resistance, wiring, etc. Eventually this will get thrown off if you never/rarely allow the pack to charge in the constant-voltage (100%) charge area and/or you never/rarely drain the pack close to 0%. So the solution is to give the algorithm new data by charging to 100% and draining as low as comfortable, but somewhere near 10% or lower is best. It won't hurt anything as long as you charge immediately.

2) Cell balancing

There are 96 cell groups in the Model S 85/90 kWh packs, and 84 in the 60/70 packs. Eventually due to slight variations these cell groups will end up at slightly different voltages. However, the Model S uses the highest voltage out of all 96 to determine when to cut the charge. So, even if 95 of the cell groups are at say, 97% if one of them is at 100% it can't safely charge anymore.

So the BMS system has balancing bleeder circuits. Once the BMS is aware of which cell groups are higher than the rest it activates a low power load specific to that cell group to drain the excess capacity they have and bring them back in line with the other cells. However, the BMS does not appear to activate any balancers until it gets a voltage reading during the constant-voltage phase of a charge, which is > ~93% SoC. So, charging to 100% also has the benefit of getting the BMS to activate the balancing circuits. These will stay active for hours/days/weeks until they bleed off sufficient power from the affected cell groups, and will happen continuously in the background even after driving and all and even if the pack is not at 100% since it knows how long it will need to bleed for based on the voltage difference during the 100% charge.

While the cells balance, as you charge over the coming days the cells will be able to charge more fully.

So essentially you want to get the car as much data about the pack as possible. This way it has more information to go by and thus have a more optimistic range prediction.

I take from this that it could take at least hours to completely balance out.

 

[apacheguy]

I take from this that it could take at least hours to completely balance out.

Indeed. But there is no need to let the car sit > 93% to complete the balancing. That is all Cottonwood is saying.

 

[Canuck]

Yeah, the balancing thing is worth a shot but I guess I just don't see how all of these 90 kWh can be out of balance with only a few cycles on them.

I agree. I don't think it has to do with balancing.

 

[rawn77]

Once again this is the OP of this thread.
Yesterday morning, as typical after a week or so I reached a new 90% low of 248 (down from 258 when delivered end of September)
Not a happy camper!!

Ron

 

[Max*]

Once again this is the OP of this thread.
Yesterday morning, as typical after a week or so I reached a new 90% low of 248 (down from 258 when delivered end of September)
Not a happy camper!!

Ron

Have you tried re-balancing as was described on the previous page?

 

[FlasherZ]

In the past few weeks, temperature has played a big role in calculated range, moreso than previous software releases. My car (Feb '14 remanufactured battery) charged full to 266 miles / 90% to 238 miles in September, now charges full to 262 / 90% to 234 miles. Meanwhile, the average temperature in my garage is 20-30 degrees lower than in September. That doesn't mean the range drop is entirely attributable to temperature shift for winter, but I think it might play some role.

 

[apacheguy]

@Flasher - Isn't the snowflake over the range display supposed to indicate when range is being "hidden" due to cold temps?

 

[rawn77]

In the past few weeks, temperature has played a big role in calculated range, moreso than previous software releases. My car (Feb '14 remanufactured battery) charged full to 266 miles / 90% to 238 miles in September, now charges full to 262 / 90% to 234 miles. Meanwhile, the average temperature in my garage is 20-30 degrees lower than in September. That doesn't mean the range drop is entirely attributable to temperature shift for winter, but I think it might play some role.

A friend of mine got his 70D delivered the same week as mine; we live in the same city; so same temperatures. His 90% range has not budged since he got it.
From what's been explained to me, the range is calculated by multiplying the SOC (state of charge) by an EPA algorithm; and not anything to do with temperature.
So either my battery's ability to charge up to a certain level (SOC) has changed or the computer is unable to accurately determine the true SOC.

I will let my battery deplete to less than 50 miles and then do 100% charges over the next 2 weeks. I'll report back on the 90% range.

Ron

 

[apacheguy]

So either my battery's ability to charge up to a certain level (SOC) has changed or the computer is unable to accurately determine the true SOC.

As I understand it, there is physically less capacity that can be drawn out of a cold pack compared to a warm one. However, when this is the case it should be indicated:

Battery Meter and Cold Weather

In cold weather part of the battery meter may appear blue to show the portion of the battery that will not be available until the battery reaches operating temperature. If the car is plugged in, you can heat the battery by turning on climate control via the mobile app.

 

[liuping]

I will let my battery deplete to less than 50 miles and then do 100% charges over the next 2 weeks. I'll report back on the 90% range.

I'm trying to find a time to do this as well, mine is down to 251 from 257 when I picked it up in the beginning of October. It was an inventory car so it already had 800 miles on it, must of already lost several miles before I got it.

I'm going to try to do a 100% charge at a local super charger (off hours, so not to block anyone visiting) so I can also log the charge rate for this thread: P90D Supercharging Data

 

[FlasherZ]

@Flasher - Isn't the snowflake over the range display supposed to indicate when range is being "hidden" due to cold temps?

On the extreme end of things, yes, but that's independent from what I'm seeing. The snowflake is when the car "reserves" some range - and I've only seen it twice, for 1-2 miles. This appears to be range calculation sensitivity based on the pack temps.

I went through some of my data from charging last year, as well. Here are some sample 90% charge numbers on days from late 2014 to today, in the middle of each month. I pulled the maximum 90% charge over a span of 2-3 days. I don't have time to correlate the temperature over those days (or look at software updates, or looked at in proximity to the last range charge for rebalancing), but you can see how the number varies seasonally. Note that the peak occurs in the spring, around Mar-Apr. This is the range data reported by the car via the API.

EDIT: I pulled some more data back to January, 2014 and here's what it looks like... keep in mind these are based on single samples from each month, so the data is very "noisy". If I were better skilled in MongoDB, I could pull every sample that was at 90% and graph it. But I'm horrible with it, and especially the nested document work that has to be done from TeslaMS.

I suspect I had a major software upgrade occur in Feb, 2014, for example, when the range jumped from 225.9 to 240.1 from Feb-Mar. In addition, the high June, 2014 number was immediately after a week of range charging every day while on vacation, probably at the best balance the pack's had in a while.