Acceleration with AC on

[Porsche]

Every now and then I turn on sport mode, turn off traction control, turn off the air conditioning, and stomp on the accelerator, just for a bit of fun. But then I thought to myself, this is an EV. Do I really have to turn off the AC? In an ICE car, the air conditioning compressor is mechanically connected to the engine and robs it of horsepower. But in an EV, the compressor has its own electric motor, not connected to the drive train at all. Yes, it does rob the battery of electricity, but there's no reason to expect that it subtracts from the available power to the drive train. Presumably, the drive motor's maximum power is set by a control system, and the accessories are all limited or fused individually. There may be a maximum allowable bus power, but it would be pretty poor design if the electrical bus could only provide exactly as much power as the drive system needed. Efficiency and range may be affected, but performance shouldn't be, right?

The AC really doesn't need all that much power, but what about the heat? Plus the degradation during winter? Is the electrical bus robust enough to supply full power to absolutely everything without limiting drive train power during the winter? Yes, range will certainly be reduced, but is performance limited as well? I haven't driven in the winter yet. Is acceleration limited based on temperature alone?

 

[Infinion]

The Li-ion batteries should be capable of some very high discharge rates, but it's not unlimited. Depending on construction, anywhere from 2C - 30C discharge rate can be expected.
For a 14-18kWh battery, 10C would allow 140kW - 180kW discharges depending on SoC. But is it really that cut and dry?

If you've ever requested the maximum 120kW to the motor inverter, you'll know there is a certain SoC where it will NOT deliver that power. This means that there is a certain point in the standard Li-ion range of 3V - 4.2V, or 288V - 403.2V in a 96S pack, that peak power will be less than 120kW for one reason or another.

Just like the compressor on a belt and pulley system loads the engine in an ICE, a compressor at a full-tilt of, say, 4kW would load down the Li-ion batteries and you'd see a non-negligible drop in battery voltage. For a Spark EV with 15kWh of capacity, 4kW constitutes a 0.267C discharge. This would be added on top of the 120kW discharge, or 8C discharge + 0.267C.

I don't have LG's pouch discharge curve characteristics, but for cylindrical cells, there are many examples.
Take a look at the discharge characteristics for this 5Ah 21700 lithium cell: https://lygte-info.dk/review/batteries2012/Vapcell%20INR21700%205000mAh%20P50%20%28Red%29%202021%20UK.html

In these discharge curves, we see that throughout various discharge rates, the cell voltage on the Y-axis gradually diminishes. We also see that as we move down to lower discharge curves (amps listed in the legend at bottom of chart), the voltage is sagging more and more for higher discharge rates. Finally, if we compare where the curves end for various discharge rates, we see that we're able to utilize less total energy from the cell shown in the X-axis' cumulative Ah and Wh, suggesting a drop in efficiency due to a number of reasons.


0.267C (1/4C) would represent a 1.3A discharge curve (closest to blue line) on this graph. Compared to a gentle 1/20C or 1/25C discharge (top red line), we are seeing a voltage sag of 0.02V. It's not much, but this sag would be multiplied by 96 for 96 cells connected serially (the same current is flowing through all serial cell groups.)

Now let's slam on a 6C discharge (yellow line above). We see a 0.7V difference between the gentle discharge and the aggressive one. In a 96C pack, this 0.7V * 96 would turn into a 67V drop in cell voltage. But 6C for a 15 kWh degraded Spark battery pack roughly accounts for 90kW of power. It would need an 8C discharge rate for 120kW. To make matters worse, this power pushing through the internal resistance of the cells increases temperature as seen in the below chart with temperature in the Y-axis on right-hand side, requiring more power usage to run thermal management.

In other words, depending on the discharge characteristics of pouch cells, a full-power acceleration with the AC on could pull upwards of 9-10C on the batteries.

In summary, when we view some discharge characteristic data, we see that high discharge rates reduce available power from a system. As other high and low voltage loads are added to an already aggressive discharge, the voltage sag will eventually reach a point where peak power to the motor will be limited to protect the cells from overcurrent discharge, or a loaded undervoltage cutoff. Or it could be that the working voltage simply results in a lower available power (power = voltage*current). These limits will occur sooner if the state of charge is already below nominal. But at what SoC I haven't paid close enough attention.

Maybe someone can can stomp on the throttle and see at how many battery bars they aren't getting 120kW, just so we all know, hah.

 

[Porsche]

Very interesting, quite a lot of information and good points, but I'm not sure I would accept all of your assumptions. or draw the same conclusions. First, at least to a first order approximation, Li-ion batteries put out pretty close to constant voltage over most of their charge range. All of your graphs pretty much demonstrate this (albeit these are for flashlight/camera batteries, not for cars, but the basic chemistry is the same). Next, the physical limit for discharge for an EV battery is nowhere near 120kW. It's more like a megawatt or more. OK, the Spark battery is pretty small, but must be at least half that and probably more. If this weren't true, not only would the car's efficiency be significantly worse than an ICE car, but would catch fire in minutes, or at least melt (literally, not virtually). That being said, you did pretty much answer my question. Of course, the Spark electronically limits the drive power to 120kW, so is independent of accessory power. Will it do so at some level of low charge? Maybe not. I don't know what kind of "limp home" limits are set in software. It does start to limit things at low charge (it is pretty funny when it offers to turn off the radio). The system will also not allow the battery to go to actual zero. I almost never let it get low, but I will see what happens. and experiment.

 

[Infinion]

at least to a first order approximation, Li-ion batteries put out pretty close to constant voltage over most of their charge range. All of your graphs pretty much demonstrate this (albeit these are for flashlight/camera batteries, not for cars, but the basic chemistry is the same).

It's actually quite the contrary. The graphs demonstrate a fully charged battery at 4.2 V and fully discharged at a safe 2.8V. With a 1st order approximation, you'd say the voltage varied by 1.4V using only 1 sig fig. 0th order approximation would be 1V, so based on this data, we can't consider the voltage to be constant. Multiply that difference 96 times and we have to work with a voltage range of 268.8V to 403.2V to a 1st order approximation, a difference of 134.4V between fully charged and fully discharged cells. Also keep in mind that EV's of the likes of Tesla that don't use pouch cells will use cells of this construction, so the chemistry and form factor should be a fairly representative comparison for EVs.

That being said, LIFEPO4 (LFP lithium-ion) has a much more constant voltage range:
https://lygte-info.dk/review/batteries2012/Vapcell IFR32700 6500mAh G65 (Blue) 2020 UK.html

This would be true of the 2014 Spark EV with the A123 LFP cells, so you could argue that, to a 1st order approximation and over roughly 80% of the discharge curve, the voltage remains constant. The low SoC performance should be quite good for the 2014 Spark EV, as well as not suffering as much voltage sag at high C discharges compared to NMC.

the physical limit for discharge for an EV battery is nowhere near 120kW. It's more like a megawatt or more.

There's a big difference between physical limit and safe physical limit. No EV discharges beyond a C rating specified by the cell manufacturer because going beyond that point irreversibly damages it (plating, destroyed graphite structure for ion intercalation, heat/boiling/vaporizing of electrolyte and pressure causing deformation or budging, ultimate loss of capacity and current capability) as you yourself alluded.

If you want repeatable results as is expected of a car, it must be designed within those safe limits. If you want to design a vehicle with the sole purpose of breaking a land speed record at peak power, you probably don't care for pack longevity.


It really depends on which EV we're talking about, too. Not all EV batteries or their packs are made the same, or with the same capacity and configuration. Max discharge power has a proportionality with capacity, among other limitations. I'd say the Spark's batteries would be hard pressed to release 0.5 Megawatts of stored electrical power safely, even though the batteries used are hybrid batteries which tend to support higher C ratings at low pack kWh sizes compared to higher capacity packs in BEV's.

If this weren't true, not only would the car's efficiency be significantly worse than an ICE car, but would catch fire in minutes, or at least melt (literally, not virtually).

That's a bit of a non sequitur. Take the 2019-2021 Tesla Model 3's as tested and logged on Bjorn Nyland's youtube channel and google drive spreadsheets.
https://docs.google.com/spreadsheets/d/16mGOOveEcxO85bVkp51YG7-DGWijpI7zY8yh6p7VnEw/edit?usp=sharing
https://youtu.be/QGjwEAWTcrw?t=171
A model 3 performance with a hot battery (peak performance ready) has <400kW discharge power and it is easily several times more efficient than an ICE car. SR+ is even less at >240kW. Why would discharge power have to be >500kW for the Spark EV's pack to be more efficient than an ICE either?


Anyways, I'm interested in this peak power claim. My position is that power is going to plateau as the voltage sags to nothing.
Let's use the data from the earlier INR21700 cell in my last reply and take a snapshot of all the voltages and currents at say 90% SoC and extrapolate where peak power would be if everything was ideal.

The manufacturer specified 25A max discharge, or 5C, which is just around 80W or so. Extrapolating with a perfect fit curve, peak power occurs at 90A or 18C. This is going to damage the cell for sure but we see the expected plateau.

Now let's set the cell up in a 96S9P configuration so we get roughly 45 Ah comparable to a 2015/16 Spark EV and find that idealized plateau again.

at 190V and 800A it falls short of 160kW. The max discharge for these cells would hit around 70-75kW, so we'd probably want a larger configuration of 96S15P to hit the 120kW power figure we want safely. In GM's case, they opted for the hybrid pouch batteries, so they could, reach higher powers for lower capacity and physical size.

Of course, the Spark electronically limits the drive power to 120kW, so is independent of accessory power.

I wouldn't be so sure about this. HVAC is not a part of accessory power, which is fused at 20A in the low voltage fuse block. The HVAC runs on the 300+ VDC battery pack and is fused at 30A along with the coolant heater at 30A. The Auxiliary power control module will feed 1.5kW to the 12-14V automotive systems and is fused on the high voltage battery side at 12A.

On a high level, if you look at the instrument cluster, you see HVAC power as well as battery heater metered to the same total power readout. If drive power is independent from all other power, why is it indicated here? I believe that 120kW is a hard limit set for the common busbar, and power delivered to the traction motor will be throttled first before all other loads. But I'd love to see some data from TorquePro whether it will ever exceed this.

The traction motor is another interesting consideration for why keeping HVAC and other drains off will have an improvement on acceleration. At the heart of every Spark EV is a permanent magnet synchronous motor (PMSM). As RPM rises in a PMSM, back-emf will rise with it, ideally until the back-emf voltages equals the supply voltage and no current flows.

The inverter is responsible for vector control, and creates the 3-phases of rotating magnetic field, as well as current regulation based on the throttle. If voltage is lower due to low SoC, or loads are active that are sagging down the voltage of the batteries, the maximum RPM and torque will be reduced. This is due to the intrinsic back-emf of the motor, as well as the fact that it takes time for current to rise in a motor phase's windings. A higher voltage produces a sharper current angle, resulting in a higher average torque if current can build up fast enough during a firing angle of that particular phase.

Will it do so at some level of low charge? Maybe not. I don't know what kind of "limp home" limits are set in software. It does start to limit things at low charge (it is pretty funny when it offers to turn off the radio). The system will also not allow the battery to go to actual zero. I almost never let it get low, but I will see what happens. and experiment.

I drove up a steep mountain until the battery indicated "LOW". Got as close to zero as I ever got, but then gained 10 kilometers back on the way down. Not sure how much power it was pulling to the traction motor, but car simply got slower and slower. The radio being off probably would've made some difference, but not much. In fact, depending on the song, it could probably feel a little faster :lol: .

 

[ayebah]

84mph on a up hill. While the gauge was showing 0. 2014 lfp batteries can for sure discharge at a high c rating at the lower end of capacity.