EV Dynamics Physics Experiment

[larryh]

Regen Efficiency

 

Yesterday, I attempted to measure regen efficiency using Low to slow the car down from 40 mph to 15 mph.  The following chart shows the results of one of several attempts.  The blue line is the kinetic energy available for regen assuming the mass of the car is 1940 kg.  As the car slows down, the kinetic energy of the car decreases.  The blue line shows the amount of the decrease with time.  After 10 seconds, kinetic energy has decreased by about 0.07 kWh. 

 

The green line is the amount of kinetic energy that makes it to the motor.   The difference between the red and blue lines is the energy lost due to friction (aerodynamic drag, tire rolling resistance, and internal frictions).   The amount lost due to friction is estimated from coast down trials and computing power loss due to friction vs. speed (see post 4). 

 

The red line is the electrical energy generated by the motor.  The difference between the red and green lines is motor efficiency. 

 

From the trials, 82% of the kinetic energy was converted to electrical energy stored in the HVB.  Motor efficiency was 95%.  Motor efficiency depends on the actual mass of the car.  If the estimate of 1940 kg is 2% too low, then motor efficiency is actually 93%.

 

Edited May 25, 2015 at 02:44 PM by larryh

[larryh]

I measured the energy loss ascending hills at different speeds.  The energy loss is defined to be the difference between the energy output by the the HVB and the potential energy required to climb the hill.  The steeper the hill, the more the loss.  The faster you ascend the hill, the more the loss.  The energy loss (kWh/mile) are given in the table below:

             

              hill grade

speed    

(mph)     2%                8%

20          0.10              0.22

30          0.13              0.23

40          0.15              0.27

 

The loss for the 2% grade hill is similar to the loss on a level road.  As the grade of the hill increases, the loss increases significantly, mainly because the motor has to output a lot of power to climb the hill and the motor is generally only around 88% efficient.  Speed has less of an impact. 

Edited May 30, 2015 at 05:13 PM by larryh

[larryh]

I also measured the energy loss descending the same hill.  Unlike ascending the hill, the loss was independent of the grade of the hill.  The loss was almost the same when traveling on a level road, when ascending a 2% grade, and when descending up to an 8% grade hill (this indicates that regen efficiency is close to 100%).  As with ascending hills, the loss increases with increasing speed.

The measured power loss at 30 mph when descending the hill was 5.15 kW.  When ascending the 8% grade hill, it was 6.71.  The difference, 1.56 kW, is due to the efficiency of the motor being less than 100%.  The power required for the potential energy to ascend the hill at 30 mph is 19.03 kW.  This implies the motor power loss when ascending an 8% grade hill is 1.56 / 19.03 = 8% or motor efficiency is 92%.  At least the motor is slightly more efficient when climbing the hill than on a level road (88% efficiency) to help mitigate some of the motor losses when supplying the potential energy to climb the hill.

 

Note:  Actually there is a 3% increase in energy loss when descending the 8% grade hill vs. traveling on a level road.  That implies regen efficiency is around 97%. 

Edited May 30, 2015 at 07:20 PM by larryh

[larryh]

The following chart shows Energy Loss when descending an 8% grade hill at 30 mph.  The dark blue line shows potential energy assuming the car's mass is 1895 kg.  The dark blue line rises at first indicating the car is ascending a slight hill.  Then begins to fall slowly at time 0.007 hours as the descent begins.  At time 0.015 hours the dark blue line falls more rapidly at the start of the 8% grade descent. 

 

The purple line shows the energy supplied by the HVB.  It increases gradually when climbing the slight hill.  It levels off as the descent begins.  At this time, the potential energy supplied by the hill matches the energy required to overcome friction, so the purple line is level.  During the 8% grade, regen begins and the purple line begins to fall rapidly as the motor converts potential energy to electrical energy which is then supplied to the HVB.

 

The light blue line, shows the energy loss.  This is the difference between the dark blue and purple lines.  The light blue line is basically a straight line.  The slope of the line indicates the power loss. The slope of the line (power loss) increases by 3% during regen, indicating regen is approximately 97% efficient.

 

 

Edited May 30, 2015 at 07:46 PM by larryh

[larryh]

This chart is similar to the previous one except in the reverse direction.  The car is now ascending the same hill at 30 mph.

 

You can see the slope of the light blue line (energy loss) is steeper during the 8% grade portion than the rest of the hill, i.e. the slope decreases a little around time 0.018 hours.  Even though the motor is about 92% efficient in providing the potential energy required to climb the hill, it has to provide a lot of potential energy.  This additional energy loss increases the slope (power loss) during the 8% grade ascent prior to time 0.018 hours.  After 0.018 hours, the car no longer needs to supply much potential energy to ascend the remainder of the hill. 

 

There is a small amount of regen during the slight decline at the end. 

 

 

Edited May 30, 2015 at 08:01 PM by larryh

[larryh]

The two previous charts demonstrate why you will get worse mileage on a hilly road vs. a level road.  The light blue lines show the energy output from the HVB less the kinetic/potential energy of the car.   This represents energy that you can never recover to do useful work, i.e. energy that is lost forever to heat, sound, and various other non-useful forms of energy.  This is unlike the potential/kinetic energy of the car which is always 100% available to do useful work and thus excluded in the energy losses calculation. 

 

If you make a round trip, the change in kinetic/potential energy of the car is 0 and the final value of the blue line will be the total energy consumed from the HVB.

 

The blue line will always remain the same or increase.  It can never decrease.  By definition, an energy loss is not recoverable to do useful work (we would be violating several laws of Physics if the blue line decreased). 

 

If we travel at constant speed on a level road, then energy loss will increase at a constant rate, i.e. the blue line will be a straight line with constant slope (the slope representing a constant power loss).  If we travel up a hill, the slope of the line will increase.  The car needs to supply the potential energy to climb the hill.  Since the motor is not 100% efficient, some of the energy from the HVB used to provide the potential energy will be lost during the conversion to potential energy.  When we travel down the hill and there is regen, then again the slope of the line will increase.  The motor is not 100% efficient converting kinetic/potential energy to electrical energy.  In either case, energy loss will be greater than it would be on a level road.  The car will have consumed more energy from the HVB for the hilly trip vs. the same trip on a level road. 

Edited May 30, 2015 at 10:18 PM by larryh

[larryh]

The following chart shows energy loss during a short trip on a level road (see post 38).  I accelerate to 50 mph, maintain the speed for a short while, and then come to a stop.  Energy loss is the green line.  It gradually increases with increasing speed (friction increases with speed).  The slope is greater than what one would expect from friction alone since the motor is not 100% efficient in providing the kinetic energy to accelerate the car.  Once 50 mph is attained, energy loss continues at a constant rate due to friction, i.e. the green line is straight with constant slope.  When stopping, the slope of the green line decreases as friction decreases with decreasing speed.  The slope of the line is slightly greater than from friction alone since regen is not 100% efficient.

 

The green and blue lines coincide at the end of the trip.  The road is level, so there is no change in potential energy.  The trip began with 0 kinetic energy (the car was stopped) and ended with 0 kinetic energy (again the car was stopped)--the red line.  The green line is the difference between the blue (energy consumed from the HVB) and red lines. 

 

The majority of the energy loss for the trip is determined by friction associated with the speed of the car.  Only a small portion of the loss is due to motor inefficiency in providing kinetic energy to accelerate the car or during regen. 

 

Note that energy loss when accelerating the car to 50 mph (prior to time 0.007) was 0.073 kWh.  The energy loss when decelerating the car from 50 mph (after time 0.010) was 0.033 kWh.  Adjusting for the different time intervals, the average power loss during acceleration was 10.8 kW and the average power loss during braking was 6.9 kW.  More energy was lost during acceleration than braking.  The average power loss while maintaining 50 mph was 11.7 kW--this is the power consumed from the HVB to maintain constant speed. 

 

Edited May 30, 2015 at 11:13 PM by larryh

[larryh]

The following chart is the same as in the previous post except I added the purple line which shows energy loss due exclusively to friction and the light blue line which is the difference between the green and purple lines, i.e. the energy loss resulting from all other sources than friction (mainly from the motor being less than 100% efficient).

 

The purple line showing energy loss from friction (aerodynamic drag, tire rolling resistance, and other sources of internal friction) depends mainly on the car's speed.  The only way to reduce this energy loss is to drive slower.

 

The light blue line shows energy loss resulting from motor inefficiency when providing the kinetic energy for acceleration and when converting kinetic energy to electricity during regenerative braking.  Most of this energy loss occurs during acceleration.  The kinetic energy of the car when going 50 mph is about 0.13 kWh.  The energy loss during acceleration was about 0.024 kWh (the height of the light blue line when 50 mph is reached and the light blue line levels off).  So 18% of the kinetic energy was lost, i.e. the motor was 82% efficient in providing the kinetic energy to accelerate the car.

 

During regenerative braking, 0.005 kWh of energy was lost (the height the light blue rises during regenerative braking which starts just before time 0.01 hours).   So 4% of the kinetic energy is lost by the motor, i.e. the motor was 96% efficient.

 

By definition, the light blue line remains constant when traveling at a constant speed of 50 mph.

 

Edited May 31, 2015 at 01:09 PM by larryh

[larryh]

Doing several analyses similar to the chart in the previous post for different acceleration rates, I come up with the following chart.  It shows motor efficiency during acceleration vs. average power consumed from the HVB during acceleration.  The faster you accelerate, the more power required from the HVB.  Two bars on the empower screen corresponds to about 25 kW of power from the HVB. 

 

The chart shows the ratio of the kinetic energy of the car, supplied by the motor, divided by the energy supplied by the HVB.  Efficiency ranges from about 85% with slow acceleration to 72% with fast acceleration.  The kinetic energy of the car at 50 mph is approximately 0.135 kWh of energy.  If you want to accelerate to that speed, that is how much mechanical energy the motor needs to supply.  If you accelerate slowly, the HVB will have to supply 0.135 / 0.85 = 0.159 kWh of energy.  If you accelerate quickly, it will have to supply 0.135 / 0.72 = 0.188 kWh.  It will have to supply an additional 0.03 kWh of electricity to provide the kinetic energy for 50 mph. Note that energy to overcome friction is not being considering in these computations.  Much more additional energy than 0.03 kWh will be required to overcome friction during faster acceleration.

 

 

Edited June 1, 2015 at 10:52 PM by larryh

[Automate]

I suspect that motor torque (used to compute motor output power), is estimated based upon the power from the HVB.

When I first heard of the motor / generator torque values I assumed the same thing but the more I look into it the more I'm convinced there is actually a torque sensor, probably some kind of strain gauge inside the transmission.

 

First there are OBDII fault codes that get turned on when the ECU detects invalid values from the sensors:

P0A18 – Motor Torque Sensor Circuit Range/Performance (non-MIL, Hazard, wrench light)

P0A23 – Generator Torque Sensor Circuit Range/Performance (non-MIL, Hazard, wrench light)

If actual torque was calculated from motor / generator output current then you would not have these and instead have something like motor / generator current sensor fault

 

Second, at least on my 2010 FFH the torque values don't go to exactly zero when the torque commands and motor / generator currents go to zero.  On my car the torque values float around plus or minus 0.2 or 0.4 ft-lbs when the motor / generators are off which is what you would expect when a real strain gauge does not go to exactly zero when the force is zero.  

Edited June 5, 2015 at 07:35 PM by Automate

[larryh]

The names of the PIDs are:

 

GTQ_OUT Generator Torque from AC Source
MTQ_OUT Motor Torque from AC Source

 

That seems to imply that that the are calculated from the AC input to the motor/generator.

 

When the motor/generator RPMs are zero, the torque is 0.0 to -0.2 N-m.

 

If you plot torque vs. rpm for the generator in EV mode, you get a piecewise linear function.  You get a similar plot for the motor when in neutral.  That would indicate that the values are computed rather than measured. 

 

Edited June 5, 2015 at 03:07 PM by larryh

[Automate]

The reason you are getting a "piecewise" graph is your data only has a resolution of 0.1 and only a small range of 0 to 2.5 N-m.  You need to collect the data with at least 0.01 resolution to get a smother curve.

 

Take a look at this FORScan screenshot with my FFH in park

The GTO_CMD and MTO_CMD are the torque commands from the control module to the inverter (variable frequency drive) that controls the motors.  This is a calculated value.  The GTO_OUT and MTO_OUT is the sensor measured actual output of the corresponding motors.  Notice the CMDs are at zero but the OUT (actual) are not.  If OUT was a calculation it would match CMD.'

 

Look on the Hybrid Synergy Drive Wiki page http://en.wikipedia.org/wiki/Hybrid_Synergy_Drive

And you will see the statement:

"Special couplings and sensors monitor rotation speed of each shaft and the total torque on the drive shafts, for feedback to the control computer."

The Ford eCVT uses the same design as the Toyota Synergy and at one time both Ford and Toyota hybrid transmissions were made by the same company.

 

 

 

Edited June 6, 2015 at 03:49 AM by Automate

[larryh]

I updated the chart below to show the three piece-wise sections of torque vs. rpm with three colored lines.  The generator torque as a function of rpm is as follows, where y is the generator torque in N-m (positive) and x is the generator rpm (negative):

y = -1.3x/2000, -2000 <= x;
y = 2.5e-8x^2 – 0.0001x + 1, -2000 <= x <= -6000;
y = x/6000 + 3.5, -6000 <= x

 

It seems strange that torque would be such a perfect piece-wise linear/parabolic function with round and non-arbitrary constants. In addition, the colored lines in the plot have "corners". I don't think a plot of actual torque would have corners.  What would suddenly be changing to create the corners at -2000 and -6000 rpm?  It looks to me like someone took the actual torque vs. rpm curve of the generator and created a piece-wise approximation, and the car is simply reporting this approximation.

 

 

Notice the CMDs are at zero but the OUT (actual) are not.  If OUT was a calculation it would match CMD.'

 

The sensors that measure current to the motor are not perfect.  So if the car is computing torque based on the current/voltage input to the motor, the result is not necessarily going to be match the commanded value.

 

 

Edited June 6, 2015 at 11:18 AM by larryh

[Automate]

Not sure about the "corner" at 2000 RPM but the one at 6000 RPM is easy to explain.  The motor/battery combination has reached its kW (or HP) limit.  kW is a product of torque times RPM.  Above 6000 RPMs in order to not go over the kW limit the torque has to reduce as the RPMs increase.

 

Take the numbers from your graphs and calculate

kW = Torque (N.m) x Speed (RPM) / 9.5488

The graph should be a flat line above 6000 RPM at the kW limit.

[larryh]

See post 21:  http://www.fordfusionenergiforum.com/topic/3179-ev-dynamics-physics-experiment/?p=20696  That contains the torque, rpm, and power data for the motor.

 

The plot above is for the generator.  The generator performs no useful function when in EV mode.  It just spins--no power is being supplied to the generator and the generator is not producing any power.  The only reason there would be any torque is from friction.  This would tend to slow the generator down. 

Edited June 6, 2015 at 12:43 PM by larryh

[larryh]

This is a plot of generator power vs. rpm in EV mode.  I'm not exactly sure how to interpret this plot.  My best guess is that this is the amount of power that is being provided to the generator to keep it spinning at the given rpm.  If that power were not provided, the generator would slow down to due to internal friction. Internal friction from the generator is consuming almost 2 kW of power when rpm exceeds 8000.  I assume that the electric motor has to provide this much power to maintain speed to overcome friction inside the generator, plus it has to provide additional power to overcome other friction sources. 

 

Edited June 6, 2015 at 02:06 PM by larryh

[Automate]

It just spins--no power is being supplied to the generator and the generator is not producing any power. 

Not true.  Hybrids use permanent magnet motors.  The spinning magnetic field in the rotor will induce a current in the windings of the stator.  What Ford engineers decide to do with this current is a different matter.  For efficiency reasons they may not send it back to the batteries but the current/voltage still exist from the physics of the way a motor works.

 

Get a motor up to speed and then remove all power.  Grab hold of the motor leads while its spinning down and you WILL get shocked.

 

This induced current/voltage will result in a torque on the generator.

Edited June 6, 2015 at 02:07 PM by Automate

[larryh]

The generator circuit is most likely open in EV mode, so no power is consumed by the generator to generate electricity.  I can only guess how it works since I don't have enough insight into what is going on inside the car.   When the car is in neutral, the motor is disconnected from the HVB so no power is consumed from or sent to the HVB.  I see similar results for the motor when the car is in neutral. 

Edited June 6, 2015 at 02:17 PM by larryh

[Automate]

Your assumptions are contradictory.

 

You are saying the GTQ_OUT is not measured with a torque sensor but rather calculated from the measured current.

But then your also saying the generator circuit is open and therefore there should be no current.  No current would result in no calculated torque. 

 

I don't disagree that the generator may not be sending current back to the battery.  But even if the circuit is disconnected, the physics of the spinning generator creates current/voltage.  If it has nowhere to go it will be converted into wasted heat in the generator.

 

But if you follow my wikipedia link above or do a search for "hybrid cvt torque sensor" you will see that hybid transmissions do have torque sensors.

Edited June 6, 2015 at 05:06 PM by Automate

[larryh]

Things are not perfect.  There may be stray current or noise that is affecting the current sensor measurements. 

 

From the plot in post 66, it appears the generator is wasting almost 2 kW of power when rpm exceeds 6000.  I'm not familiar enough with motors/generators to state all the sources of wasted energy, so I just lump them all under the term "friction". 

 

I don't know exactly what the generator/motor torque measurements are.  When it is named torque measured from AC source, that would imply there are other sources for measuring torque.  The car may well have actual torque sensors, but no communication path exists to report them to the SOBDMC.  Or alternatively, Ford has decided there is no reason to report the actual torque sensor measurements to the outside world.  Without a manual explaining exactly what the GTQ_OUT or MTQ_OUT measurements are, we can only speculate what they are.  From my experience with the torque measurements reported by the car, they don't seem to be consistent with what I would expect. 

Edited June 6, 2015 at 06:22 PM by larryh

[jdbob]

"Special couplings and sensors monitor rotation speed of each shaft and the total torque on the drive shafts, for feedback to the control computer."

 

The System Diagram for the HF35 shows TFT (Transmission Fluid Temperature) sensor, Motor/Generator coil temperature sensors, and Motor/Generator speed sensors. No sign of any torque sensors.

[Automate]

 

I don't know exactly what the generator/motor torque measurements are.  When it is named torque measured from AC source, that would imply there are other sources for measuring torque.  The car may well have actual torque sensors, but no communication path exists to report them to the SOBDMC.  Or alternatively, Ford has decided there is no reason to report the actual torque sensor measurements to the outside world.  Without a manual explaining exactly what the GTQ_OUT or MTQ_OUT measurements are, we can only speculate what they are.  From my experience with the torque measurements reported by the car, they don't seem to be consistent with what I would expect. 

 

The signals actually come from the TCM. (Transmission Control Module)  FORScan shows this.  Ford does not provide a lot of info on the sensors because they are internal to the transmission possibly even internal to the motors.  Not even dealers are allowed to take the eCVT transmission apart and repair it.  If anything goes wrong they have to replace the whole transmission.

 

FORScan also shows that there is a Engine Torque measurement.  Obviously this can't be derived from any battery current sensor.  I would agree that the description " Generator Torque from AC Source" is not clear.  It could mean the torque sensor generates an AC signal that must be conditioned into an torque value.  Or it could be they want to make sure mechanics know its for the AC motor / generator and not the engine torque sensor.

 

And back to my first post.  

 

The fault codes are:

P0A18 – Motor Torque Sensor

P0A23 – Generator Torque Sensor

 

Why would Ford call it a Torque Sensor if it is really a Current/Wattage Sensor?

Edited June 6, 2015 at 07:39 PM by Automate

[larryh]

 Is there any chance you could use your data to put such a map together for the Energi?

 

The following figure shows the efficiency map for a typical permanent-magnet synchronous motor similar to the one used in the Energi.  The solid lines are the efficiency contour lines.  The maximum efficiency contour line is the closed path at the lower left. 

 

The dashed line shows the path taken during maximum acceleration.  That path is not particularly efficient--it is, in fact, probably the least efficient path.  The straight section of the path is the constant torque section below 15 mph--the motor is outputting maximum torque, but less than maximum power.  The curved section is the constant power section above 15 mph--the motor is outputting maximum power, but less than maximum torque.

 

Below 30 mph, you want to supply maybe 1/4 maximum torque to operate the motor at maximum efficiency (i.e. the path taken goes through the maximum efficiency region within the closed loop contour).  Once you get above about 30 mph, you want less torque/power.  That means slow acceleration is most efficient.

 

It appears at lower speed, i.e. less than 15 mph, operating the motor at any torque other than about 1/4 maximum torque, the motor is less efficient.  In other words, at about 1/4 maximum torque, the motor operates most efficiently. 

 

At higher speeds, i.e. above about 30 mph, the behavior is different.  At constant speed the car is more efficient with increasing power/torque.  With increasing speed, the car is more efficient with decreasing power/torque (efficiency decreases with increasing speed).  That would explain why the motor is more efficient climbing hills.  It requires more torque/power at the same rpm when climbing a hill versus traveling on a level road.  During acceleration, you want to keep the power/torque low. 

 

 

 

Edited June 7, 2015 at 04:17 PM by larryh

[Automate]

@larryh

 

Do you have a list of the PIDs you used for these experiments?

 

The best list I could find (this one http://i-fix.us/Torque-Pro-Settings-for-Ford-Fusion-and-C-MAX-Energi.php) do not include motor and generator current or KW

 

Thanks,

Edited June 9, 2015 at 06:43 PM by Automate

[larryh]

I use the PIDs listed above.  I don't know of any PIDs that provide motor or generator current or power.   I simply compute motor/generator output power as 2*pi*torque*rpm/60/1000 in kW.  In EV mode you can estimate the electrical power supplied to the motor by computing the power output by the HVB (current x voltage) minus power consumed by the DC to DC converter (used to power accessories).  Note the header values for the PIDs in that listing are wrong. 

Edited June 9, 2015 at 06:37 PM by larryh