Replacement battery cost

[MrDRMorgan]

...After an easy 8 amp L1 EVSE fully charge overnight, my 2014 Spark EV's guess-o-meter read 105 miles this morning. Then, I drove 12.2 miles around town in stop and go traffic and averaged 6.4 mi / kWh....
... Today, my wife and I took the 2015 Spark EV for a run and we drove to Oakland, CA and back for 132 miles round trip. Road speed was 60-70 mph most of the entire trip and I still achieved 5.7 mi/kWh for the entire trip.

These numbers are hard to believe. Are these 'round trips' to negate elevation change and wind?

How slow do you go? Do you stay with traffic or are you guys like those hyper-miling Prius drivers that no one loves?

In cold weather and at interstate speeds I can be in the low 3.x miles/kWh.

Do you guys believe your charge rate has any bearing on your driving results? 1kW, 1.4kW, 3.3kW or 48kW. I only use the two highest rates.
Maybe it's the quality of California electrons?!?

These numbers are being achieved in 65 - 70 deg. F weather with no special driving other than I drive in L and not D to get maximum regeneration. Yep, I drive with the traffic too. The trip to Oakland and back included driving over and back across two passes with about 5% grades. Again, this is being done in "perfect" weather with no winds. I have seen my range get hammered in strong head winds and/or cold (30-45 deg. F) temperatures. I do not think the charge rate has any impact on the range. I use L1, L2 and DCFC when it is available. On my trip to Oakland and back, I used DCFC three times - going, returning and at the end of my trip. The last DCFC station is about 10 miles from my home. I recharged to 100%, reset the trip meter and returned home on the freeway. The trip meter showed 7.3 mi/kWh when I arrived at my home. My freeway speed was 55-60 mph until I got to my off ramp and then I drove 1 mile to my home.
Perhaps, unlike Ponce De Leon, I have found the fountain of "super electrons" here in California.

 

[SparkevBlogspot]

I'm trying to understand how I can start plotting the way you are doing it.

I don't know how he's doing it since SparkEV doesn't show battery %. But when you use DCFC, it shows you % and kWh to reach that. For example, if you start with 15% and end at 90% and it took 14 kWh, then the battery capacity would be

13 kWh / (90-15) * 100 = 17.3 kWh

averaged 6.4 mi / kWh.... These numbers are hard to believe.

Not really hard if you drive in town or in traffic. At 24 MPH, it would be 7.2 mi/kWh, and you can achieve 6.4 mi/kWh with careful throttle control at speeds of 30-40 MPH. I got 51 mi/kWh (not 5.1) going downhill even at 60 MPH and few traffic lights! :mrgreen:

In fact the 'Pack' does not NEED to be replaced. Only the aged cells within.

That's technically true, but if you want to have optimum, you have to find similar cells (ie, balance them). While BMS will try, best would be to have them matched in the first place, and only sure way to do that is to get new cells. Otherwise, you risk having to redo the pack more often.

Another problem is finding the cells. There will only be so many "used" in market as there aren't many SparkEV in existence. In addition, A123 cells will be even harder to find, and LG cells might be hard, too, if Volt cells are different than 2015+ cells.

 

[kevin]

One way to measure the battery capacity is to use the Energy Screen which gives the energy consumed and percentage used then extrapolate from those measurments.

I only use discharges that are >50%. I collect the data by taking a photo of the various screens just before I charge and then enter it into the spread sheet later.

This is the capacity of my 2014 Spark over the last two and a half years. It has degraded about 10% and is now about 17kWh from 19.5kWh initially. The max range has dropped somewhat from ~111 to 104 miles (with some seasonal variation).

kevin

Kevin,
I trying to understand what you are plotting. My bold (above) must be the main part of your data gathering, correct?

I know the GOM can not be used for anything other then 'Guess Trending'.

I'm trying to understand how I can start plotting the way you are doing it.

It's just my commutes and weekend blasts are all over the spectrum of efficiency.

Especially this time of year. It can be a cold morning, using comfortable heat, blasting along with all my pals in the left lane of an interstate at 70-80 MPH.
Then my evening commute home can be 35* warmer, no climate control, taking the back roads at a max of 50 MPH and use a fraction of the energy compared to the morning commute.
In fact my evening commute (after a full charge) can start with 72 miles on the GOM and I arrive home with 62 miles displayed after a +29 mile drive.

Note: I try to only charge at work. So would I only take a data point in the mornings when arriving at the free public L2?
If I use the fast way to and from work my commute may be less than 50 miles. In nice weather this may be less than 50% battery usage daily. Will this not be good data?

How does your method take in to account climate control use and ambient temp differences
Thanks for any help you can give us on your system!

Norton,

All you need is the energy used (E) and the % used (P) values form the energy screen. The extrapolated usable battery capacity would be 100*E/P. The result is in kWh.

For example if the kWh used is 12.8 and the percentage used was 75%, this would give a result of 100*12/75 = 17.1kWh. Because of the limited number of decimal places on the display
it is best to only use measurements where you have done a fairly deep discharge (I use at least 50%), also you must not do a partial charge, only complete charges.

I find that a convenient way to record the measurements is to take a photo of the screen rathe than writing them down in the car.

This does rely on the battery management system (BMS) having a good measurement of the battery capacity. Some of the capacity changes may errors in the BMS capacity estimation.

The energy screen seems to take into account the energy used by all consumers such as heating/cooling and battery conditioning - I haven't done enough measurements to see if the battery capacity varies significantly with temperature, I'm in the bay area so the temperature is pretty benign (although it does get chilly in the winter when I'm trying to conserve energy!

My overall energy consumption is ~5.7mi/kWh as measured on the energy screen or ~4.7mi/kWh including charger losses. Onstar displays lifetime 160mpge - 20kWh/100miles on my latest monthly report although I have seen 19mi/kWh in the summer. Frequently I achieve 6.4mi/kWh when the weather is mild.

kevin

 

[SparkevBlogspot]

averaged 6.4 mi / kWh.... These numbers are hard to believe.

New blog post to describe mi/kWh. According to calculations (in Mr. Spock's voice), 6.4 mi/kWh would be about 45 MPH without using heat/AC/radio/lights.

Image is from my blog. You can read the gory details here.

http://sparkev.blogspot.com/2016/03/range-polynomial.html

All you need is the energy used (E) and the % used (P)

I think the question is how to get the % and energy. Only easy way is to use DCFC. If not, one would need some sort of OBD device. Do you use DCFC or some custom made tool from solder (sorry to hear about solder).

 

[kevin]

All you need is the energy used (E) and the % used (P)

I think the question is how to get the % and energy. Only easy way is to use DCFC. If not, one would need some sort of OBD device. Do you use DCFC or some custom made tool from solder (sorry to hear about solder).[/quote]

You get them straight from the energy screen in the car:

kevin

 

[MrDRMorgan]

All you need is the energy used (E) and the % used (P)

I think the question is how to get the % and energy. Only easy way is to use DCFC. If not, one would need some sort of OBD device. Do you use DCFC or some custom made tool from solder (sorry to hear about solder).

You get them straight from the energy screen in the car:

kevin

[/quote]

I ran this today. Using an NRG eVgo DCFC station 10 miles from my home, I charged my 2015 Spark EV to 100% from 55% until the DCFC stopped charging. Results were 7.50 kWh consumed for 45% of the battery capacity used. This equals 16.67 kWh battery capacity. I drove home and, upon arriving at my home, noted the number of kWh used and what % of the battery's capacity it represented. 1.6 kWh / 9% = 17.78 kWh battery capacity per the car’s info. I am going to use the results of this test as my baseline for future tests to determine what capacity-degradation I can expect. The blue ball guess-o-meter read 102 miles at the end of the full charge.

Added note: I stood in front of the car the entire time it was charging and did not hear any pump or fan come on. Air temperature was about 70 deg. F. at 4pm. What I do not know is what the charging efficiency of the DCFC is. Did I actually put 7.50 kWh into the battery as the DCFC display showed or was that the power put into the AC to DC converter? If it is the converter input, what is the charging efficiency? To make my numbers equal, the DCFC charging efficiency would have to be ~94%. What about the contribution of regeneration back into the battery?

 

[NORTON]

...This does rely on the battery management system (BMS) having a good measurement of the battery capacity. Some of the capacity changes may [be] errors in the BMS capacity estimation.
kevin

Kevin,
First, thanks for bringing this to the forum!

But as you stated there could be errors. This is the BMS 'self-reporting' on itself, after all.

The idea of using a DCFC as a secondary way of gathering data on battery capacity is also not ideal, and may have even more errors.

When I'm at a DCFC,in the heat of the summer, I hear the cooling fans and the AC compressor running strong.
When at one in the dead of winter, I honestly haven't stood in front of the car and listened. I'm in the car with it powered up and using climate control to keep me warm, but the AC compressor system should not be needed to cool the battery.

Therefore, taking a reading from the DCFC display is NOT telling what went into just the battery. TMS is doing its job at a DCFC!

With the Spark EV, I assume during cold weather DCFC the AC compressor does not need to assist with cooling the battery. The battery coolant would only have to run through its heat exchanger/radiator with the cooling fan on a slow speed.
(I don't know if there is actually a 'radiator' in the nose just for the battery coolant loop. Has anyone seen a technical report on the various cooling systems in the Spark EV?)
(The Volt's cooling system is totally explained on that forum. It has 5 different heat exchangers in the nose.)

(I work with mid to large aircraft. Their batteries have 'Cap Checks' yearly. They are discharged on a controlled load and measured to see if they meet the spec. 85% is airworthy.
There is nothing like an actual capacity test to measure a battery's performance. Auto shop load tests do not tell the health of a battery. I had a flaky car lead acid battery that I tested both ways. There is no comparison.)

 

[oilerlord]

Newbie question...did all 2014 Spark EV's with thd 2LT package come with DCFS - or did that have to be ordered as an option?

Edit: Found the answer. It's a $750 option.

http://insideevs.com/chevrolet-presents-dc-fast-charging-for-2014-spark-ev-wvideo/

 

[NORTON]

oiler,
I know when I went shopping for a used '14 I had to insist the car dealership send a picture of the open charge door.
I'm almost certain it was an option.

Word: Don't tell the salesdogs what you are looking for. Don't mention "The $750 optional DCFC". They know diddly. And you don't want to give them anything they can use....
Keep it DL and just ask them to send a pic of the charge port. I had one send a google image link to hundreds of them. I had to slyly say "I want to see the one on the car you have for sale."

Good Luck !!

 

[SparkevBlogspot]

Kevin, thanks for the photo. I never charge to 100% (or try not to), so I never paid attention to "since full" display. Regarding your plot, why did you choose second order polynomial instead of exponential?

But as you stated there could be errors.

Yeah, simple measurements we can make will have large errors. That's reflected in kevin's plot (R^2 is not 1).

But I think what we're after is the trend, not absolute accuracy. Even relative accuracy would suffer if done at different temp and AC on/off, but you can minimize the errors by controlling what you can (ie, no AC or always AC). Using AC/heat (assume 3kW) with DCFC for 20 min would be 1 kWh out of 13 kWh, less than 8%. Even when using AC, etc. at random, enough data points would average them out, though it would point to lower battery capacity than actual.

I think far better is to tap into car's OBD and query power / energy / miles as it's running. Then you know how much energy was taken from the battery for given miles and percentage. For best test, one could start with high % and drive to low % to minimize errors. This is what I want to do with large SD card serving as on-going storage. So far, progress is slow.

 

[oilerlord]

oiler,
I know when I went shopping for a used '14 I had to insist the car dealership send a picture of the open charge door.
I'm almost certain it was an option.

Word: Don't tell the salesdogs what you are looking for. Don't mention "The $750 optional DCFC". They know diddly. And you don't want to give them anything they can use....
Keep it DL and just ask them to send a pic of the charge port. I had one send a google image link to hundreds of them. I had to slyly say "I want to see the one on the car you have for sale."

Good Luck !!

Totally correct, In my experience so far, dealers are absolutely clueless about EV's. I called a dealer yesterday about a 2013 Volt that looked like a good deal - he told me "It's too bad that GM doesn't make the Volt anymore". Huh?

It's almost like this dealer in California really doesn't want to sell me the Spark EV. I initially asked for the cost of the replacement battery...he said he'd get his service manager to get me that info...never happened. I then asked him to text me a photo of the charging port...that never happened either. Perhaps they don't care if they make the sale or not because they are already losing their shirt on the deal.

It's an absolute certainty that if/when I buy an EV; I'll be dependent on forums like this one for support.

 

[kevin]

Kevin, thanks for the photo. I never charge to 100% (or try not to), so I never paid attention to "since full" display. Regarding your plot, why did you choose second order polynomial instead of exponential?

But as you stated there could be errors.

Yeah, simple measurements we can make will have large errors. That's reflected in kevin's plot (R^2 is not 1).

But I think what we're after is the trend, not absolute accuracy. Even relative accuracy would suffer if done at different temp and AC on/off, but you can minimize the errors by controlling what you can (ie, no AC or always AC). Using AC/heat (assume 3kW) with DCFC for 20 min would be 1 kWh out of 13 kWh, less than 8%. Even when using AC, etc. at random, enough data points would average them out, though it would point to lower battery capacity than actual.

I think far better is to tap into car's OBD and query power / energy / miles as it's running. Then you know how much energy was taken from the battery for given miles and percentage. For best test, one could start with high % and drive to low % to minimize errors. This is what I want to do with large SD card serving as on-going storage. So far, progress is slow.

"Regarding your plot, why did you choose second order polynomial instead of exponential?"

The polynomial seemed to give a slightly better fit than an exponential. R squared is about 0.96 with the polynomial and about 0.94 with an exponential.

"I think far better is to tap into car's OBD and query power / energy / miles as it's running. Then you know how much energy was taken from the battery for given miles and percentage. For best test, one could start with high % and drive to low % to minimize errors. This is what I want to do with large SD card serving as on-going storage. So far, progress is slow"

I don't see why that would give any different results - that is what the car is doing anyway.

In addition to using the information from the energy details screen I also reset the trip counter when I do a full charge to give the energy per mile for that charge.

I agree that a larger discharge will give better accuracy not just from the reduction in quantization errors but will also give the BMS an opportunity to update its estimation model to match the battery. I assume that BMS uses a combination of Coulomb counting and battery voltage to perform the SOC estimation, the battery voltage is needed at the SOC extremes to know when the battery is getting full or empty.

kevin

 

[oilerlord]

I've given more thought to advantage of having DCFC installed or not. Though we don't have the infrastructure now, it may someday be available as our new provincial government is embracing renewable energy and may offer incentives for municipalities to install DCFC stations. We do have one Tesla supercharger in Alberta, and it is possible that Tesla may license other manufacturers to use them some day. Being able to drive virtually everywhere without worrying about getting stranded with a dead battery is a big deal.

For that reason, if I do buy a Spark EV (or anything else); DCFC is a requirement.

 

[SparkevBlogspot]

Tesla may license other manufacturers to use them some day.

Unfortunately, they are not compatible with SparkEV. There are at least 3 DCFC standards: Tesla uses Superchargers, Asian (Nissan, Mitsubishi, Kia, Hyundai) cars use Chademo, other American and Europeans use CCS (BMW, Chevy, VW). SparkEV has CCS. I wouldn't look to Tesla for SparkEV compatibility. In Europe, they have AC fast charger (~43kW) used by Renault.

In terms of raw numbers, there are more Chademo than CCS, though CCS is growing. Many are upgrading Chademo to dual head Chademo + CCS, and I see some having CCS only. CCS is cheapest at 25kW from BMW costing about $7.5K while 50kW Chademo only from Nissan costs about $20K (was $17K few years ago). This is probably why some are installing CCS only charger.

In theory, there could be Chademo to CCS converter (or vice versa). As of now, no such converter exists. However, there is converter for Tesla to use Chademo (I think $500), so DCFC converter is a possibility, though probably never for use with Superchargers.

 

[NORTON]

... it is possible that Tesla may license other manufacturers to use them some day. ...

I'm hoping the same thing.
Tesla currently sells a 'Chademo-to-Tesla' adapter cord for $450.
Some Teslas in my metro area are using this adapter cord and can now charge at all the dual-cord DCFC units, and suck up Free Electrons at a 48kW rate.
This is half the power of a Tesla Supercharger.

It is possible that someday that there will be 'Tesla-to-CCS' adapter cords with an internal ID code for billing or Tesla could add card readers on their Superchargers.
There would also need to be a communication protocol to the Supercharger to only put out 48kW when a CCS EV is using it.

Unfortunately a Spark EV would still not be a cross-country car, but this would allow a Bolt to 'Head out on the highway' !!!!

 

[SparkevBlogspot]

I don't see why that would give any different results - that is what the car is doing anyway.

I would think car's sensors via OBD would have better resolution than display. Also, having on-going measurement device that records all the time would result in more data than "full charge" events, which I rarely achieve. Assuming Gaussian-type errors, more the merrier for accuracy.

There would also need to be a communication protocol to the Supercharger to only put out 48kW when a CCS EV is using it.

Comm may not need anything special. Tesla tapers very steeply. At about 80%, Tesla is charging at 45kW. Because SparkEV is more efficient, SparkEV around 80% (also about 45kW) would be adding more miles per charging time than Tesla. Scroll down to Tesla discussion in my charging blog post (again!) about how fantastic SparkEV is; for a cheap little car to spank $80K Tesla in charging speed, even for just a little bit, is just WOW!

http://sparkev.blogspot.com/2015/12/sparkev-is-quickest-charging-ev-in-world.html

 

[kevin]

I don't see why that would give any different results - that is what the car is doing anyway.

I would think car's sensors via OBD would have better resolution than display. Also, having on-going measurement device that records all the time would result in more data than "full charge" events, which I rarely achieve. Assuming Gaussian-type errors, more the merrier for accuracy.

I still don't see it although with your own code thereby be fewer unknowns.

"I would think car's sensors via OBD would have better resolution than display."

The display has the same data (probably from CAN bus) - unless the data is intentionally truncated it will be limited to the same resolution as an external OBD.

"Also, having on-going measurement device that records all the time would result in more data than "full charge" events, which I rarely achieve. Assuming Gaussian-type errors, more the merrier for accuracy"

The Battery Monitoring System will use all the information it has - it will be be continually integrating the load current and power product to determine the used energy - not just at "full charge", and combining that with the voltage measurements to determine SOC. The integration will, as you say, allow random errors to average out but that is occurring within the BMS not just an external OBD

The "Full charge" event is just to provide a reference point, using an external monitor you will still need the same reference point from the same BMS.

I do agree that it would be useful to have an external monitor as you can then record the data and format how it is displayed but I think you will be disappointed if you are expecting significantly better accuracy.

good luck

kevin

 

[SparkevBlogspot]

not just at "full charge", and combining that with the voltage measurements to determine SOC.

Unless you charge to full, LCD percentage would be meaningless. That would make it impossible to determine battery capacity. As for LCD sig figures, I suspect it's more than 2 digits if one measures using OBD. With more samples (ie, higher sampling), averaging would be quicker. Anyway, my goal is to have something there to record all the time with time stamp so I can pull it out once in a while to do some analysis. One of these days...

 

[agdodgerfan]

I have almost 70k miles on my 2014. I usually fully charge at night. I now get between 60 and 70 miles per charge. I will start tracking used/% and report back when i get 10 to 15 measurements. I recently noticed it drop available range i live in LA so winter shouldn't be too much of a factor for lower range.

 

[MrDRMorgan]

I have almost 70k miles on my 2014. I usually fully charge at night. I now get between 60 and 70 miles per charge. I will start tracking used/% and report back when i get 10 to 15 measurements. I recently noticed it drop available range i live in LA so winter shouldn't be too much of a factor for lower range.

Before you charge your Spark EV, use the Energy Info screen on the infotainment system and calculate your battery capacity by dividing the number of kWh used by the total percent used. Then fully charge your Spark EV until the Energy Info resets to zero. This will give you a good estimate of your battery capacity. I have been doing this since March 2016 and I average the readings for each month and plot the results. In March 2016 my average was 18.94 Kwh. In December 2016 my average was 18.07 kWh. So far in January 2017 my average for 3 charging sessions is 17.83 kWh. I will continue doing this throughout 2017 to see if the drop is due to cold weather or real battery degradation. Total driving distance for the period Mar 16 to Jan 17 is only 4500 miles and I charged [L2 EVSE] the vehicle 81 times during this time period. Note: This information is for my 2014 Spark EV 2LT w/o DCFC. Last summer this car's GOM was running 105 to 112 miles after a full charge. The current cold weather - 30 deg. F to 40 deg. F - and heater use has dropped the full charge GOM to 70 - 80 miles as expected.