Let's stop ICEheat, when cold...

[vtechtuning]

As you know some of my ideas/developments from "Let's go pure EV" thread, I promised to focus on gasoline engine heating function control.
I'm still not sure about exact algorythm of ICE start (and I cannot read AC-control ecu to check the algo just in code), I decided to check it step by step and look for a non-intrusive method of influence to it. I have some success (video in 2 hours, when it will be colder - now it is +4 and +7 in the car and I'm not sure, if it is cold enough to proove that partial success).

What I have now:

- gasoline engine is controlled not only by difference of temperatures (set and real) inside, nor ambient to internal difference. Algorythm is more sophisticated for sure (depends on actual load, delta of actual load (especially kick-down impulse prohibits gasoline engine to be used/started as heater - sounds ridiculous, because kick-down forces ICE start and driving with gasoline engine is a source of heat - but I have already found, that it is managed and handled totally separately). It depends on ambient to internal-set temperature, internal-set to internal-actual temperature and actual level of electric heating. I have no signs from my tests that it is dependent on battery temperature, or SOC.
- I have added such functionality to the PHEV box (pure EV mode box - for those who did not read "Let's go pure EV" thread) that box can simulate ultra short kick-down behaviour, that is not considered at all by PHEV ECU as kick down (as this function has a time delay and integration of signal hard coded) but PHEV ECU puts appropiate information to CAN (once) and this prohibits (or better to say - reduces importancy of gasoline engine start demand. Luckily such trigger has it's own delay in AC-control ecu (so for few seconds gasoline engine start by demand of heating is blocked or exactly - has lower priority).

Finally we can highly reduce gasoline engine start-for-heating reason and importancy and even started - it stops much faster. This is proven.

Box must have one additional line with reference temperature signal (internal temp signal) to know (estimate) what is actual probable demand due to internal temperature. To calculate total demand I need 2 more temperatures for sure, but signals are not available (at least - not easy available, without car alteration). I need to produce one plug for that, but hardware in the box is ready now. For tests I have soldered one signal line to temp sensor, so box knows internal temperature, and installed additional chinese thermometer with out and in temperatures, just for me, to monitor temps.

What do you think? Any more info about AC-ECU algo for heating?

https://youtu.be/xpp6i_ebVnM

I did two starts - first with box on, second - off. +4,5 C outside, similar cold inside (I first declared to do one test and second one after 1 hour, but I thought that during second start it is even warmer inside & gasoline engine is warmer - so for sure it should work shorter or max same time than in first try.
Observations:
- engine starts in both cases.
- with box on engine works 30 seconds, then stop
- then restarted (car off, car on) with box off - engine works for more than 3 minutes 15 seconds (then I switched it off manually as I want to cool down the car again and re-test).

I still do not understand fully why it starts it (for sure these complicated dependencies cause that, for instance outside temperature) but I have reduced it's working time more than 6 times in the first try (in both cases electric heating works same, 100% of heating power).

 

[Tipper]

My observation is that ICE starts on 'start-up' if the external temperature is less than 10C and the heater is selected on.

 

[vtechtuning]

My observation is that ICE starts on 'start-up' if the external temperature is less than 10C and the heater is selected on.

Ok, thx, I consider this to be a part of the "heating algo".

It seems to me basically such algo:

Treshold_of_ICE_start = - k1*(temp_outside - 10) + k2*(temp_inside_set - temp_inside_measured) - k3*temp_of_engine - k4*(other_power_demand) - k5*(electric_heater_level)

k1...k5 are various coefficients. If sum is higher than treshold, ICE starts. Some coefficients must have big value (like k3 - that is logical, or k4 - that I have proven by tests). But nothing is sure, these are my observations & prediction, how it may work. For sure it is not dependent on one temperature (or temperatures only).

5:30 in the morning, - 0,4 C outside (below 0 C). I tested my ICEheater stop plugin for my box switched on (so box can increase k4 coefficient influence), video uploading. Next test in 1-2 hours. But it looks not bad It seems problem of ICEheating is 90% solved

https://youtu.be/WHBbMcTxV98

ICE worked 1:01 (61 seconds).

Parameters: - 0,4 C outside, 2,1 C inside (chinese thermometer , 20 C set, fan set to 50%, battery full, car was whole night outside, connected to charger. 5:30 in the morning (dawn here).

 

[anko]

To me it also seems very strange that starting the engine for heating purposes depends on power demand for driving. Unless the "start engine" request from the A/C specifically tells the engine to run "low".

Behaviour of engine start caused by A/C (compared to engine start by pressing Save or Charge button) are:
- The engine will generate as much power as needed for driving but no access power will be produced for charging the battery. When parked, the engine effectively idles (until you hit Save or Charge)
- The engine will not engage in parallel drive at 40+ MPH (until you hit Save or Charge)
- The engine will start when fuel level is very low (which it will not do when you hit Save or Charge)
- The engine will start at very high SOC (which it for most of us will not do when you hit Save or Charge)

When you start the car with the heater turned up, the engine will start. When you immediately turn off the heater, the engine will keep running for a little while, I assume to prevent excessive wear and such. 30 seconds as a minimal sounds about right. I can imagine that the duration of this will depend on the ambient temperature and more parameters and could indeed be 90 seconds in colder environments.

It appears as if the result you got in your tests are consistent with blocking the "engine start" request from the A/C just a little too late: the engine was started and as soon as your device kicks in, the engine is allowed to stop again. But it will only do softer 30 - 90 seconds. I think the same may have happened when you had turned down the heater, directly after starting the car.

I wonder, what happens when you start the car with the device active and the heater turned off and then, after a few seconds, turn on the heater? Will the engine run for 30 - 90 seconds? Or not at all?

Altogether, this is a good result, as it allows youth use the electrical heater rather than the engine for heating (after 30 - 90 seconds) on short trips. But I am still hoping to get rid of that first burst. A couple of times this winter, the engine of my car started when I had to move my car from a non charging parking bay into a charging parking bay or vice versa, a distance of 20 meters or so. Only because I forgot to switch off the heater when I parked the car.

Oh and another question: why would you need ambient temperature feed to your device? Why not let the device do it's thing with the "ultra short kick down" always? Until you switch it off?

 

[ClasseClas]

My observation is that ICE starts on 'start-up' if the external temperature is less than 10C and the heater is selected on.

I have noted that also. If no preheating then ICE will start if ambient is below 10C.

 

[ClasseClas]

I am very glad that you vtechtuning work on cracking the algorithm for this!

Your assumption on algorithm looks reasonable. It is a bit hard to give you exact information about own experiences as they are based on own views and conclusions and not based on exact measurements.

However, my view is that the algorithm is dead stupid. Here is the use case;
-Temp below 10C (rather below 0C). I preheat the car for 30 minutes. Thus coupe temperature is so high that ICE will not start. Fully charged battery minus preheating.
-I drive 10 minutes to the store.
-I go shopping for say 10 minutes.
-When I return to my car and start it ICE will engage. Most likely the trigger for this is only 1) low temp and 2) low coupe temp. It seems that it does not know anything about possible residual heat in the electrical heater that would inhibit start of ICE. I guess that there is no temp sensor in the electrical heater (similar to an engine heater temp sensor). Given this situation at idle when ICE starts there is relatively low power demand. Car stands still.

Please also look into this thread with some observations,
http://www.myoutlanderphev.com/forum/viewtopic.php?p=24251#p24251
1) It is possible to set AC in manual mode in MMCS (turn off "snowflake") resulting in lower power consumption
2) It is possible to set "economy" ACC mode in MMCS (engaged by the ECO mode button) resulting in lower power consumption
3) Battery temperature may have an impact on ICE start

 

[anko]

It seems that it does not know anything about possible residual heat in the electrical heater that would inhibit start of ICE. I guess that there is no temp sensor in the electrical heater (similar to an engine heater temp sensor).

Not quite. There are two temp sensor, one for the inlet and one for the outlet of the electrical heater. But do not forget, the water temperature in the electric heater will drop much, much faster than that of the engine coolant:
- there is very little fluid involved
- there is no large heavy massive solid body that can hold the heat (engine)

 

[anko]

My observation is that ICE starts on 'start-up' if the external temperature is less than 10C and the heater is selected on.

May I suggest to rephrase this into:

ICE does not start on 'start-up' if the external temperature is more than 10C

Also I think the 10 deg C border is not solid. I have been able to drive away at 7 deg C and use the heater without the engine starting up. But not always.

 

[ClasseClas]

It seems that it does not know anything about possible residual heat in the electrical heater that would inhibit start of ICE. I guess that there is no temp sensor in the electrical heater (similar to an engine heater temp sensor).

Not quite. There are two temp sensor, one for the inlet and one for the outlet of the electrical heater. But do not forget, the water temperature in the electric heater will drop much, much faster than that of the engine coolant:
- there is very little fluid involved
- there is no large heavy massive solid body that can hold the heat (engine)

Thanks for the info! As I wrote earlier it is easy to draw conclusions based on assumptions which are not fact based.

Now, as the electric heater is faster to heat the cabin than the ICE given above reasoning, it would be the first choice to deploy rather than the slower ICE heating. Still, this is not the case. Then the question is why? Maybe it is about the max power available from the batteries which set the limits here. The algorithm of vtechtuning with "k4*(other_power_demand)" may come in strongly as the main limiting factor.

A side comment: On Toyota Prius gen 4 there is now a PTC heater available which will provide even faster heating and also reduce the ICE usage for heating.

 

[anko]

The electric heater throws in 4 kW max. The engine roughly 70% (heat losses) of the energy in the fuel burned during heating. Or even more when the engine is not contributing to driving. That is an awful lot of heat.

 

[vtechtuning]

Anko - thx for the heat graph (PW). This explains a lot to me.
It is so nice that when we work all together, we can improve our cars so much.
I have updated my algo of heating, it seems to be much more complete now.

 

[aaa3]

Does this mean that the ICEheaterStop option for the PHEV BOX will be available soon?

 

[anko]

Anko - thx for the heat graph (PW). This explains a lot to me.

For those who wonder:

White trace shows use of electric heater elements:

00 = none
10 = small element
20 = large element
30 = small + large element
40 = two large elements
50 = all elements

The rest should be self explanatory ...

During this run, it was 7 deg and the engine did not start when I turned the heater on. So, I used Charge mode. This explains why there is no "Engine Start Request" from the AC.

 

[Douglas]

I use the following method to avoid the ICE starting.
Turn off heater before I turn the car off. (I have also used the tip regarding defaulting the A/C to off).
If I am not sure if I turned the heater off at the end of the previous journey I fire up in Acc mode and turn the heater off if its on.
Turn the car on, select ECO mode
While driving on electric turn the heater on.
The heater then seems to run on battery alone and until the battery is very low the ICE doesn't fire up.

So far tested ok down to about 3 deg C. PS I leave the heater setting on 21 deg C.

Just my observations

 

[anko]

As Vtechtuning said, it is all about understanding the decision making algorithm. The below is pretty much copied from email I sent to the maker of Real Time Charts plugin for Torque Pro (https://play.google.com/store/apps/details?id=com.pjt.realtimecharts_v1):

It shows:

Engine Coolant Temp
Ambient Temp
Electrical Heater Inlet Temp
Electrical Heater Outlet Temp
Engine Start Request (not functional)
Interior Temperature
Heater Element Activation (as explained in previous email)
Engine RPM (just to see whether the engine is running)

The Mitsubishi Outlander PHEV is a plug-in hybrid vehicle. Not all, but many owners make an effort to use as little fossil fuels as possible. In their attempts, they are frustrated by a design decision by Mitsubishi to start the engine for heating purposes under some conditions. They claim that, under these circumstances, it is more efficient and faster to use engine heat. The first is certainly true from a kWh perspective, but not necessarily what you want. Especially when making a trip well within EV range.

There is an ongoing search for driving / operating strategies or even tools that help minimise or better eliminate engine use for heating on short trips. It all starts with understanding the decision making algorithm (parameters can be ambient temp, interior temp, requested temp, heater and cooland fluid temps, etc.). Creating this chart is part of that quest.

Interestingly enough, it demonstrates that, even when the engine is started to provide heat, the electrical heater is still blasting away. Coolant circuit and heater circuit (which are two parts of a single circuit with a thermostatic valve) are heated up separately. Only when the coolant reaches about 70 deg C, the valve is opened and the circuits connected. At any time, the coolant is hotter than the heater fluid, so why the valve is not opened from the beginning, I don’t know. Perhaps to allow the engine to get to the right temperature faster? But it does not seem very efficient from a kWh perspective, as Mitsubishi suggests.

Also, Mitsubishi says, the engine is only helping ‘to get the heating process started’, but this diagram shows that the engine keeps running even after the heater has been turned off. So, apparently by ‘get the heating process started’ they mean ‘meet the thresholds set in the decision making process’, which involves more than have sufficiently hot heater fluid.

At the beginning of this trip, it was about 7 deg C and I had (purposely) not pre heated. But when I set the A/C unit to 20 deg C, the electric heater started but not the engine. I had to go as high as 24 deg C before the engine would start. But this kept my engine running (what seemed like) forever. At 6::32 and a bit I turned down the A/C unit to 20 deg C and then the engine stopped and the heater took over from there.

Engine activity at the end has to do with empty drive battery, not heating demand. After all, it is a hybrid ;-)

 

[vtechtuning]

As for now all boxes use two way ICE stop - by inducted pikes of load and by surging current from internal temp thermistor to control a gap between internal temp measured and demanded.
With the help of Anko I found following rules:
1. Start of ICE is demanded, when sum of influencing parameters from various sensors/ecus goes over threshold for start. That treshold is constant.
2. ICE is started when cold at first because of there is no part of data comming from AC controller, and they will be reducing demand, if gap between internal temp and demanded temp is low.
3. Let's have 5C outside. ICE starts immediately, then even if I leave only "pulses method" on, it will work about 30 seconds, instead of 60 or 90, when pulsing off. It looks it is only pre-heating of engine before it will be loaded with heavier work, and not for heating purposes (at least not solely for heating purposes).
4. If we switch off AC and then start system, and then (immediately) start AC at any demanded temperature (of course not oven-type) with temp sensor to internal temp gap controlled (reduced) - it won't start at all.
So both methods working. Pulsing reduces time of work if accidentaly started, and controlling the gap to be lower than 9,5C adds negative value to heat demand, thus stopping ICE from start.
This works for me. Let me know what do you think?

 

[anko]

2. ICE is started when cold at first because of there is no part of data comming from AC controller, and they will be reducing demand, if gap between internal temp and demanded temp is low.

And yet, if the gap is already small due to higher ambient temperatures, the engine will not start. So, simply 'missing data from the A/C' can IMHO not be the only factor.

3. Let's have 5C outside. ICE starts immediately, then even if I leave only "pulses method" on, it will work about 30 seconds, instead of 60 or 90, when pulsing off. It looks it is only pre-heating of engine before it will be loaded with heavier work, and not for heating purposes (at least not solely for heating purposes).

Not sure about that. If this was the case, it should also do this when the heater is turned off. And it never does.

4. If we switch off AC and then start system, and then (immediately) start AC at any demanded temperature (of course not oven-type) with temp sensor to internal temp gap controlled (reduced) - it won't start at all.

That is my observation as well. At 5 - 6 deg C ambient, the car thought it was 13.x interior and did not start the engine until requested temp was as high as 22.5 deg C.

So both methods working. Pulsing reduces time of work if accidentaly started, and controlling the gap to be lower than 9,5C adds negative value to heat demand, thus stopping ICE from start.
This works for me. Let me know what do you think?

Agreed. But still, I am hoping for a final solution that prevents ICE startup when I have to move my car 11 meters from a normal parking bay to a parking bay where I can charge, and I have forgotten to turn of the A/C

 

[DerekT]

I have the same problem, but in the opposite sense, ie moving my car a few feet away from the charging point (outside the garage) to let my wife take her car from garage. Using the method outlined earlier by 'Douglas' works for me. My routine is:
-> ACC -> ON, check the heater is OFF, then depress footbrake, press POWER again, EV starts, ICE doesn't. Car can then be moved.
Should I wish to drive somewhere at this point, press ECO, and turn heater ON, then heater runs, but ICE remains OFF.

 

[anko]

Should I wish to drive somewhere at this point, press ECO, and turn heater ON, then heater runs, but ICE remains OFF.

Believe me, when is cold enough (and it does not even have to be that cold, it will start the ICE, even with ECO mode engaged.

The parameter that controls this is a threshold between actual cabin temp and requested cabin tem. ECO mode adds just a few degrees to the threshold. No more.

I am still hoping / waiting for an update from vtechtuning for the PHEVbox, which artificially lifts the cabine temperature sensor reading to about 17 deg C minimum, instead of the current 13 deg C minimum. I am pretty sure this would allow me to leave the heater on and set at 20 or 21 deg C under all conditions, without risking the ICE being started up.

 

[Steepndeep]

Another starting sequence which works and is fast is to use two hands while starting. One pressing the ventilation "OFF" button and the other the "START" button simultaneously. By chear magic (or probably a complete oversight by Mitsu engineers) the boot sequence actually turns the ventilation off before the "check if ICE needs to start" sequence starts. But as I stated in other posts, the car is clearly not engineered for a driver who can think.

On a similar note I just received the Swedish Mitsu Magazine. On the backside there is an invitation to testdrive the new PHEV, now with EV prio button. No delivery dates stated and no info on upgrade of existing cars. Being quite apalled by how little attention Mitsu pays to existing customers I fear they have introduced the new button to clearly show that old cars cannot be updated because "they do not have the button".