[Ken Fry]

Quote:

1. How does efficiency change with load and rpm for the generator?

2. How does efficiency change with load and rpm for the motor?

3. How about the controller?

4. What is the overall efficiency in terms of vehicle speed and load?

5. And is energy moving "through" the batteries? What conditions affect the charging/discharging efficiency? (not counting regen)


6. Like we don't consider ICE engines to be really 35% efficient because of operating conditions, we should not make the same mistake with electrical drivetrains, no?

Not sure if the quoting will make sense, but these q's were asked in another thread. They are great questions... in fact, spot on. I find it irksome when I talk to a motor manufacturer and they cannot give me anything other that a full load dyno run for an electric motor. Even more irksome, when based on this run, I calculate efficiency, and find that it peaks at 80 or 85% (including controller losses) but the marketing spin claims the motor/controller combo is generally over 90% efficient.

Specifics:
1. The motor generator unit in the POC runs at constant speed and near constant load. If load drops, it shuts off (in the POC I do this manually). So in practice, the generator only runs when the battery bank can absorb all the power it can produce at its peak efficiency load and rpm. In the POC the amperage is not allowed to go outside the range of 50-60 amps. In that range as speed the generator is within 1% of 89% efficient. (I use 85% in my spreadsheets.) There is nothing but wires (and a contactor) between the generator and batteries in the POC. (In the production version, there are diodes.) So the POC generator produces 58V x 55A continuously: 3190 watts. The engine output must then be 3190/.85 = 3752 watts. That's 5.03 hp. At that load, the POC engine is about 24% efficient. If the batteries are somewhere in the range of 30% to 70% charged, and I go up a hill, the load on the engine increases, and its efficiency (in the POC goes up to 26%, which is as good as it can get (without modifying it for closed loop, Atkinson cycle etc.) The batteries then have to take over most of the load, and because they are relatively big (huge in comparison to a Prius-like hybrid) they do not sag much. They have no difficulty in putting out 600 amps (10C) briefly, whereas the generator can only put out 60 amps. In the POC, the batteries cannot put out more than 5C (300 amps) because the motor controller sets a limit than keeps the motor from glowing red hot.

2. This is the killer. Motors are remarkably level in efficiency (in other words the curve is pretty flat) but vendors rarely seem to show the very light load part of the curve, where it has to drop off to 0% when the motor is just starting to move, and functioning only as a heater. So the only way that you can asses efficiency is to plug the actual values (which you can get from some manufacturers with prodding... or can generate in a shop with a dyno). Azure Dynamics publishes pretty good data that covers most of the operating range without driving themselves crazy. Others just assume you don't care or that you are willing to do your own dyno testing. I guess the situation is just like it is for commercial engines, where some vendors provide actual BSFC maps, but most just provide a single BSFC curve based on full load dyno testing. For my application, the latter is not a problem, because I operate very near the middle of the peak. But there are times when the drive motors operate at much less than peak efficiency. So in modeling for Urban (especially) and Highway (to a lesser extent) You have to insert actual figures for particular loads and RPMs.

Failing to do this gives you the equivalent of the mpg calculator on this site which is completely useless for anything other than a single speed where you are certain of the actual BSFC, at that load. If you put in some average, then there is one speed for which the calculated mpg is correct, and it is incorrect for every other speed. This makes mpg as low speeds (where engine efficiency might be 8%) completely bogus, if you use 20% as an average. In my view, the calculator should produce only one line, for a particular speed and load.

Short of paying the $25,000 for EPA style dyno runs, modeling with actual efficiencies at actual loads and rpms is the only way to get meaningful MPG figures. On-road testing is just too variable for a vehicle developer's use. (Read about the trials and tribulations of coast down testing on this site to get a feel for variability.) Once you've got 100 cars out there and people are responding to the EPA site, then averages of many people and many conditions closes in on the EPA numbers.

3. The controller: these days, it might as well be considered wire: 99% is often quoted as "typical most of the time". (With a 150kW motor 1% is 1500 watts -- about like a stove burner. Don't touch the controller.)

4. The overall efficiency in terms of vehicle speed and load: This is all over the place for a plug-in hybrid. As a simple rule of thumb, you can take a steady state condition for 45 mph, going up a 2% grade, and in a minute of two get an MPG figure which is close to the combined EPA figure... for many vehicles. But otherwise there is not single efficiency number that works -- it changes a lot with speed and load. The guesstimate I mention is only useful for deciding to go ahead with a project, at which point you have to gather up all the relevant data or do the testing of components. If you a hobbiest, then none of this matters all that much, other, than, I suppose, as background. Call Honda for the BSFC maps for a CBR 250 , and you probably won't get much help.

5. Is energy moving through the batteries? There is a level-road, still-wind cruise speed at which the engine has to provide all the energy to more the car. At that condition, all the energy effectively bypasses the battery. In the more typical case, in which the engine comes on for 20 minutes and then shuts off for perhaps a similar time, the the losses in and out of the battery matter. With lead acid, these can be a deal killer. With Lithium X they tend to be small in both directions, often 95% in and 95% out, but worse if you get up to 80% charge... at which point you should not be charging.

Conditions: Mainly temperature, which is why most systems are actively heated and cooled. Secondarily, charge rate, with either way high or way low being not so good.

6. Yes, we should not make those "generality" mistakes for either vehicle type. If you model with a particular "average" efficiency, the model does not work. After you have done real modelling, then you can work backward and say that the overall drive train efficiency is X, (if you wanted to do so) for a particular stated condition... but then some journalist would pick up that figure and use it completely inappropriately.

Sorry for the long post. This stuff is hard to simplify without conveying the wrong info. You clearly "get it", but there are others who may be reading who need a little more detail. Performance prediction spreadsheets can go on for many pages, and the guesstimates or calculators at electric motor vendor sites rarely work right -- in practice they are not useful even for comparison shopping because, for example, the efficiency curve of a PMAC motor and an AC induction motor are quite different. At a target speed and load, the difference in efficiency can be 50% vs 80%, even if the motors are "appropriately" sized. If one is too big and another too small, then the differences can be greater.

Really good questions.
Ken