Regen limits & Non-locomotive Power Draw

[scaesare]

I touched on this in another thread, but with temporarily not being able to garage my S during this particularly cold weather we've been having, I've had opportunity to do some more observation.

It appears that the regen power limits for a cold battery do NOT correctly account for the non locomotive power draw (NLPD) of the car. This implies that your car will regen less than what the battery is capable of absorbing at a given temp by a margin equal to whatever "the rest of the car" is doing, be it pack heaters, cabin heaters, 12V DC battery charging, etc...

This is particularly irksome in very cold weather, as that's when you are most severely regen limited, and yet your NLPD is the greatest.

I draw these conclusions based on several recent mornings where my regen was very near zero, or disabled completely. At those times the pack heater and cabin heaters were cranked up. My resting power draw was ~10KW. At no time was I able to get the needle down in to regen as long as the regen limit was less than my 10KW NLPD.

Once the regen limit line was greater then my NLPD, I could trigger regen, but only by the difference of the two. In other words if my NLPD was still 10KW, and my regen limit line was at 15KW, I could still only manage 5KW worth of regen.

Now it's a bit tricky to estimate this, as the power and regen meters are both logarithmic, and at different scales. But given that the low end of the scales provide the greatest resolution on the displays, I'm pretty close I think.

As my NLPD draw reduced when my pack & cabin started to get up to temp, the regen needle would get closer to the limit line. By the time my NLPD was only 2-KW, I was very close to the regen limit, before it disappeared altogether.

I've always noticed that the regen meter would never quite hit the limit line, but I always attributed this to display imprecision, perhaps due to rounding error. Now I wonder if there's always 500-1000W of other stuff the car is doing that factors in to that.

I also don't believe it's just a display bug, as I can tell the difference between 0 and 10KW of regen, and the car's behavior is matching what the dash is telling me.

I've wondered if this was intentional in that he NLPD can vary pretty instantaneously, (i.e. turn cabin HVAC off), and they don't want to overshoot the regen power being pumped in to a cold battery. But based on the CANBUS snooping I've seen, the message rates between driveline and cabin are something like the hundreds to thousands per second. There's no reason why the car can't vary it's regen throttling instantly in reaction to a changed NLPD load.

Hence I think they may just be an oversight. It's unfortunate, because I'd love to have more effective regen in cold weather driving, and would also appreciate recovering as much energy as I can on those cold days where the pack heater is sucking juice from the battery.

I'm interested in other people's thoughts and experiences...

 

[MP3Mike]

There's no reason why the car can't vary it's regen throttling instantly in reaction to a changed NLPD load.

Maybe it can, but there are reasons that it shouldn't. If you are coasting and the cabin heater turns off you would suddenly be slowing down less, then the cabin heater turns back on and your regen would increase.

I assume Tesla wants regen to be predictable and consistent, not pulsing with changes to your NLPD load.

 

[Big Earl]

I will note that in all of my cold weather driving, I consistently find that the power meter will not display the full regen that the dashed line indicates is availble. One example is on a mild day that doesn’t require significant cabin heating. When the regen limit disappears, the car still falls short of full regen for a period of time. I believe that the regen limiter (dashed lines) is always an optimistic display of what’s actually available. I believe @David99 has noticed similar behavior in his car when monitoring CAN data.

 

[scaesare]

Maybe it can, but there are reasons that it shouldn't. If you are coasting and the cabin heater turns off you would suddenly be slowing down less, then the cabin heater turns back on and your regen would increase.

I assume Tesla wants regen to be predictable and consistent, not pulsing with changes to your NLPD load.

That's a good point.

But I'll note the car will cut regen completely and immediately mid-deceleration for instance when traction dictates. Of course that may be a more severe consideration.

But I'm not sure I can think of any sudden changes in NLPD that can be end-user initiated other than the cabin heater. Perhaps in that case it's possible to taper the cabin heater down and regen with it over the course of the next 5-10 seconds...

 

[David99]

Yes it's an oversight to not consider for example battery heater or cabin heater when limiting regen. I have seen 10 kW of power draw when taking off the foot off the pedal yet it would not do any regen because the battery was cold. It could have at least done those 10 kW.

At least in theory. In reality the cabin heater and battery heater draw power a little random. I recorded the power draw during preheating when the car was parked and not plugged in. The power draw went up and down for whatever reason. If the car would allow regen, there it would have to adjust in realtime to those changes in power consumption from those other things and that might be difficult and could feel very odd to the driver. But it might also be just technically difficult to do. They can't just let the rest go into the battery.

Now with the Model 3 they might have been fixing this already. The M3 has no battery heater. Instead it uses the motor/inverter to generate heat if needed. Those units are already part of the coolant loop and naturally get hot with normal use. So it is perfectly safe to 'abuse' them to generate heat. I'm not sure how it works technically. They somehow can send electricity through there without creating any torque. Now what if they can create negative torque (what regen does) without generating any energy. The energy would be dissipated in the motor/inverter were it becomes heat. In other words they could create the regen effect of slowing the car down but instead of charging the battery it would all be converted to heat. The coolant carries the heat to wherever it can be used (warm up the cabin and battery).

Now that's a theory based on the ability of the Model 3 to use the motor and inverter as a heating unit. If it were true they would make winter driving much more efficient. Any M3 owners here in cold weather? Does regen gets limited when the battery is very cold?

 

[MP3Mike]

Does regen gets limited when the battery is very cold?

Yes it does. (From the videos have seen.)

 

[Big Earl]

Regarding regen on Model 3:

How limited though? It can only pump so much heat into the battery coolant loop at once. I suspect that when the battery is cold, the car can produce maybe 10 kW worth of regen-as-heat into the battery until the battery temperature increases and electrical regen is allowed. It would be interesting to observe regen behavior on a cold day.

 

[mongo]

I'm wondering if what we want to happen may not be feasible.
In order for the DU to create regen, it needs to push energy back into the HV bus. If it does that, the voltage will rise. There is a very fine line between a voltage that only takes over from the battery, and a voltage that is charging the battery. (The pack has a low equivilent resistance so there is not much droop to work with)

If the DU regen is current based with low ripple, then it may be feasible. But with the AC motor, the ripple may cause battery charge events at high frequency that would degrade the cells if their temperature is too low. To prevent this, they may be exercising an abundance of caution in regen operation since the degradation is cumulative and not reversible.

 

[David99]

I'm wondering if what we want to happen may not be feasible.
In order for the DU to create regen, it needs to push energy back into the HV bus. If it does that, the voltage will rise. There is a very fine line between a voltage that only takes over from the battery, and a voltage that is charging the battery. (The pack has a low equivilent resistance so there is not much droop to work with)

If the DU regen is current based with low ripple, then it may be feasible. But with the AC motor, the ripple may cause battery charge events at high frequency that would degrade the cells if their temperature is too low. To prevent this, they may be exercising an abundance of caution in regen operation since the degradation is cumulative and not reversible.

What if it isn't regen. Rather it is putting the inverter in reverse while the car still moves forward. If you now press the accelerator it would use energy but it would slow down the car. You are basically using power to slow down the car. The motor would obviously get warm doing that 'tug of war'.
@Big Earl made an excellent point. This method might be limited to a certain power level to not burn out the motor but also not to produce too much heat that would bring up the coolant temperature too high running into the battery. It is a bad idea to send coolant into the battery that is much warmer. It would create a temperature difference within the battery which would be bad.

 

[mongo]

What if it isn't regen. Rather it is putting the inverter in reverse while the car still moves forward. If you now press the accelerator it would use energy but it would slow down the car. You are basically using power to slow down the car. The motor would obviously get warm doing that 'tug of war'.
@Big Earl made an excellent point. This method might be limited to a certain power level to not burn out the motor but also not to produce too much heat that would bring up the coolant temperature too high running into the battery. It is a bad idea to send coolant into the battery that is much warmer. It would create a temperature difference within the battery which would be bad.

What you are describing is regen. The kinectic energy if the car has to go somewhere. The inverter drives the windings slower than the rotor speed (or in the reverse direction, both look the same to the rotor) and the mechanical energy is converted back to electrical.

If it were a brushed PM DC motor, you could short the windings and get braking that way (and heat the motor). The 3 is a brushless PM motor, so it could do a waste dump of energy without backfeeding the battery via shorting the coils electrically (all low side switches on). On an AC motor, the stator windings create the rotor field, so inducing a field and then shorting it may be possible, but much more difficult.

 

[scaesare]

So along the lines of pawer draw varations perhaps being the reason for this, I observed the following while stopeed this morning (ignore audio please):

I was doing no manipulation of any vehicle controls, so perhaps this is the pack heater modulation being described earlier. This indeed might be annoying if it affected regen...