[bwilson4web]

The Toyota transaxle is radically different from traditional transmissions. Instead of clutches and shifting gear sets, there are two powerful motor generators, a planetary gear, and computer software. This article discusses how it works with new data showing it in operation.

NORMAL POWER SPLIT DEVICE

At low speeds and high power settings, 28% of the engine torque spins motor generator 1 (MG1) that under computer control, generates electrical power. The counter torque, 72% pushes through the power split device towards the drive wheels:

But having taken 28% of the engine torque to generate electricity, this electrical power goes to a second motor generator (MG2) on the far side of the power split device to drive the wheels. Simple, yes? . . . not quite.

The power split device (PSD) is a set of planetary gears:

The engine drives the spider gear carrier with 28% of the torque going to MG1 to generate electricity. Thanks to the gear ratio, 72% of the torque goes to the ring gear and on to the drive wheels. Then the MG1 power powers MG2 as a motor to help turn the drive wheels.

Now this happens at slow speeds and high power settings. The following graph shows this during a hill climb between seconds 1050 and 1150:

The orange line is MG1 working as a generator, a negative power draw, taking 28% of the engine torque. This power then goes to MG2 which runs as a motor to add this power back with the 72% of engine torque that goes directly to the ring gear. Now you'll notice there were several similar cases when the red-line, engine power peaked and MG1 power became a generator. During these times, the engine spins much faster as it makes more power needed to handle little rises in the road. But something is happening at the lower power times, energy recirculate.

ENERGY RECIRCULATE ("HERETICAL MODE")

Very little power is needed to sustain power on flat terrain so it helps to have a higher gear ratio, an overdrive gear. The Prius does this by reversing the roles of MG1 and MG2. MG2 runs as a generator and sends the power to MG1 to drive against the engine, to make it turn slower:

The engine working against the MG1 now turns slower, in a lower power but much more fuel efficient region. This set of data shows the two modes:

At startup, MG1 works as a generator and the electrical power goes to MG2 that runs as a motor. Once at speed, the roles reverse and MG2 becomes the generator passing power to MG1 that slows down the engine to put it in a lower power but very high efficiency mode. But this begs the question about the efficiency of the electrical path between MG1 and MG2 regardless of direction.

Using a Graham Miniscanner, I collected data climbing a hill to generate this graph:

As the engine power goes up, MG1 runs as a generator passing power to MG2 that works as a motor. After taking away the vehicle electrical overhead, about 445W, the efficiency is in the 85-98% range. Due to sampling errors, we don't have a single hard efficiency number but a collection of points. The hill has a brief reduction in climbing grade where the power demand slacked off and the car briefly shifted into energy recirculate mode.

Ordinarily we don't like to see energy converted from mechanical-to-electrical and back, electrical-to-mechanical. Yet the most efficient vehicle mode is in energy recirculate mode:

So taking one point at the graph at ~9kW ICE power (~12 hp) the MG2->MG1 efficiency is in the 75-92% range. Notice the MG2 power is in the 5-10kW region. Normally a single gear stage is treated at 98% efficient so this is enough to roughly estimate the Prius transaxle efficiency:

• 7kW * 83% => 5.8 kW :: power flows toward the engine from MG2, 1.2 kW lost
• 9+5.8 kW * 98% => 14.5 kW :: positive that includes the MG1 and ICE power, .3 kW lost
• -7 kW + 14.5 kW => 7.5 kW :: net power output to the drive wheels
• 7.5 kW / 9 kW => 83% efficiency :: net transaxle efficiency, 1.5 kW lost

A more accurate estimate can be achieved by doing a proper statistical analysis of the data but for back-of-the-envelope, this is good enough. Still, I'm surprised at the low efficiency. But as the engine power increases, the ratio of MG2->MG1 power, the lossy path, decreases improving transaxle efficiency. This begins to explain the 'flat' efficiency measured in the field:

This flat efficiency from 50-65 mph has always bothered me. Now I understand that the NHW11 transaxle efficiency improves at higher power settings at a rate that compensates for the higher vehicle drag. But this is not the whole picture as the engine efficiency is part of the system:

Loss of power in the transmission does not mean poor vehicle performance if the improved engine efficiency more than compensates for this loss. To see the total vehicle efficiency requires another set that incorporates the engine efficiency, the brake specific fuel consumption (BSFC.) Fortunately, data from this set has revealed a way to make these measurements.

Traction battery power is the product of the voltage and current. But traction battery voltage is also a function of state of charge and current. Since the Prius does a good job of keeping the state of charge fairly constant after the car is warmed up, the traction battery voltage then becomes a stronger function of current:

V = 0.0063 * (A**2) + 0.8973 * A + 300.85
P = 0.0063 * (A**3) + 0.8973 * (A**2) + 300.85 * A

What this means is traction battery voltage can be replaced by mass air flow to calculate the fuel burn. Combined with the ICE power, we have the data needed to calculate total vehicle energy outside of true velocity and altitude which can be obtained via GPS.

Bob Wilson

[GeorgiaHybrid]

To simplify...press the skinny pedal to go faster, press the fat pedal to slow down.....

If you don't press the skinny pedal hard, you won't burn as much gas.










Bob, does your head ever hurt????

[bwilson4web]

GeorgiaHybrid said:

. . .
Bob, does your head ever hurt????

Well I have been married for over 30 years . . .

Bob Wilson

[GeorgiaHybrid]

That sounds about right....28 here. I like the charts although it did take a bit of study to get the relationships you were making.

[bwilson4web]

GeorgiaHybrid said:

. . . I like the charts although it did take a bit of study to get the relationships you were making.

The Prius transmission is not easy to understand and I'm looking at the 2001-03 version. Toyota made subsequent improvements that they have not really shared beyond the technical elements. Their modeling data is a little light but it is proprietary and I can not fault them for keeping their trade secrets and more than my curiosity about 'how did they do that?' But this recent operational data begins to raise the cover over what they've done and Toyota's accomplishment is awesome.

I've read and re-read multiple reports about the Prius transaxle and performance:

• "Evaluation of 2004 Toyota Prius Hybrid Electric Drive System", Staunton, R.H., Ayers, C.W., Marlino, L.D., Chiasson, J.N., Burress, T.A., ORNL/TM-2006/423, Oak Ridge National Laboratory, UT- Battelle, May 2006 - this easily Google'd report is one of the most detailed analysis whose Appendix B, Table B-1, pp. 70-84, has the same data I measured in our NHW11 but in a massive table on the bench. This very precise and detailed table lacks one thing, field data from a working Prius. Every possible operational mode of an NHW20 Prius is in that table but without operational field data, it might as well be a random collection of points that cover every possible operational state of an NHW20 transaxle. Even though Appendix B is a lot to chew, the rest of this paper is well worth downloading and reading. It was until my recent experiments the only credible report about how the NHW20 transaxle and by extention, the other Toyota hybrid transaxles work. Sad to say, I've not found an equivalent Ford paper.
• Muller 2001/08/25 data - saveded in Prius Technical Stuff files. This was the first data set recorded from the NHW11 and first introduced me to 'heretical' mode or what Toyota calls "energy recirculate mode." Even now, I use "heretical mode" to communicate with the pioneers.
• "Developmen of New-Generation Hybrid System THS II - Drastic Improvement of Power Performance and Fuel Economy", Muta, Koichiro, Yamazaki, Makoto, Tokieda, Junji, SAE 2004-01-0064, March 8, 2004 - is the first paper that discusses the Toyota term, "energy recirculate" which until then it had been called "heretical mode."
• "An Analytic Foundation for the Two-Mode Hybrid-Electric Powertrain with a Comparison to the Single-Mode Toyota Prius THS-II Powertrain", Meisel, Jerome, SAE 2009-01-1321, Georgia Tech, March 3, 2009 - a theoretical paper, I was not ready for the claims when I first read it. Now that I have a way to measure power loss in the field, I'm in a better position to apply experiment to this model.

So that is my 'like' reading. <GRINS> Understanding satiates my curiosity and experimentation teaches what is true.

Bob Wilson

[bwilson4web]

I captured a boatload of data Saturday, July 30 with ambient temperatures ~77-80F and no wind, with combined GPS and Graham Miniscanner data. The GPS provides location and altitude to calculate the vehicle kinetic and potential energy change in Watts. The Graham Miniscanner collected MG1 and MG2 torque and rpm as well as traction battery current and mass air flow for fuel consumption. With this data collection, I can calculate the total vehicle energy flows. This chart shows my first effort to visualize this data:

This graph shows total energy states starting from a dead stop at Martin Road and Whitesburg Drive and making a left turn, going west on Martin Road. Initially, I accelerated on EV power and then slowed down due to traffic. Then I accelerated to 48 mph indicated, 50 mph GPS to the gate at Redstone. Then I disengaged cruise control and let the car coast with the built-in regenerative drag to the gate and a complete stop. Some observations:

• Initial acceleration, 18:28:28-18:28:58- the transaxle is in "normal" mode where MG1 is working as a generator and MG2 is just a motor. Because of the acceleration used, the traction battery energy (green line,) after starting the engine pretty much stayed near zero. The bulk of the acceleration energy came from the engine but the ratio of energy flowing between MG1->MG2 became a smaller portion as the car accelerated. Also, I'm surprised that the total kinetic and potential energy, Newtonian energy state, remained nearly constant. This was not by design but something captured during this event. I was expecting to see the Newtonian energy track the ICE energy and am significantly disappointed that it did not. The chart suggests roughly a 50% loss of ICE energy versus the expected Newtonian energy change. NOTE: this is experimental data so I need to go back and make sure I have calculated kinetic and potential energy correctly.
• Traffic and cruise control engagement, 18:28:58-18:29:04 - this is just 'driving the car' and getting through some merge lanes and getting cruise control engaged.
• Cruise at 50 mph true, 18:29:04-18:31:04 - with the exception of some traffic, the speed was constant so the potential energy changed following the contour of the road. All of this was in "energy recirculate" mode where MG2 is the generator and MG1 is the motor driving the engine to turn slower. Roughly half of the ICE power flows through the MG2->MG1 path and the remainder must take the mechanical path, the gears. Also, as the vehicle speed increased, the aerodynamic drag, an unmeasured effect, would have taken a larger percentage of the energy which would have kept the Newtonian energy level constant.
• Coast down, 18:31:04-18:32:04 - while above the hybrid transition speed, 42 mph, the engine provides some braking force. Then the engine stops turning over and the remaining kinetic energy is absorbed by MG2 charging the traction battery. Again, the ratio of the Newtonian energy to traction battery energy is the efficiency of regenerative braking.

I'm sorry if this seems complex but there are multiple energy flows going on. To the best of my knowledge, this is the first time a chart like this has ever been published or shared.

Bob Wilson

ps. The minor divisions are 6 seconds so the series looks like:
18:28:28 18:28:28
18:28:34
18:28:40
18:28:46
18:28:52
18:28:58
18:29:04
18:29:10
18:29:16
18:29:22 18:29:28 18:29:28
18:29:34
18:29:40
18:29:46
18:29:52
18:29:58
18:30:04
18:30:10
18:30:16
18:30:22

[bwilson4web]

This effort is also aimed at comparing transaxle efficiency, the electrical path around the Power Split Device (PSD) which is a fraction of the total ICE power:


This shows transaxle efficiency for the electrical power path and combined electrical and mechanical path. The total efficiency of this stage is closer to 90% at this speed and load. Not shown are the subsequent reduction gears, differential and CVT

Bob Wilson