[Joshua Bretz]

This is a good analysis. I did a stupid try at this earlier just assuming constant power. But there is another dimension to the numbers: the inverter may have different higher power/torque ratings for acceleration as opposed to constant load. That is, the capacity of the magnetics to support a given flux may be different than the capacity to remove heat.

The inverter may be able to do a 0-60 run without even heating up much. The trick is embedding an accurate dynamic model of the inverter semiconductor junction temperature. Then, you run at constant junction temperature instead of (an artifically low) constant power/current.

 

[Tom]

Need more parameters to do this

What is needed is a peak junction temperature, say 200C. Assume you start at zero current, then the junction is at engine ambient. The heat sinks/cooling have a large time constant ~ 1 minute, so it takes some time to get to peak junction temperature. Note: This requires a transient 3D analysis to get temp peak at center of die. If we have some info/guesses on heat sinks, cooling, and number, die dimensions and type of drivers, then given power waveforms, we can make a reasonable estimate of transient junction temp vs power waveforms. We can then use this to throttle back the drive to maintain peak junction temp during a power cycle.

Also have to do a reliability analysis to see if drivers can last 150,000 miles if run for high duty cycle at peak temp.

 

[Cybereye]

The Tile of the picture says "GM Volt Estimated Peformance curves". I'm wonder is that base on the simulator or the real thing on the Malibu test car? I'm guessing it a simulator, not a real thing. I can't see a test Malibu being a Drag Coefficient = 0.25. The info may be base on the wind tunnel with the clay volt model. Hey Tom, where did you get the info from?

 

[Texas]

Tom, Nice analysis. I hope you keep tweaking your model's parameters until you can use the Volt to verify everything. Your model can then be used for subsequent design changes as well as to critique other manufacturer's claims on soon-to-be-released models. Keep up the good work!

 

[BillR]

Different Gear Ratio?

Tom,

Your analysis seems pretty good. Do you have more accurate information on the traction motor?

In the following thread, I looked at the drivetrain possibilities.

http://www.gm-volt.com/forum/showthread.php?t=252

In my final post, I used a traction motor similar to the AC Propulsion design. I assumed the peak output was increased to ~150 kW, and that the max motor speed was 12,000 rpm. This equated to a 7.595 reduction gear, which should make a big difference in the acceleration from your 4.73 ratio.

I wasn't sure how to do the integration for speed as you have done, so I'm glad to see someone complete that.

Keep up the good work!

 

[Tom]

Source of Data

The curves are simulations for vehicle dynamics I developed essentially using dv/dt =acceleration = Net Force/mass and then integrating for v. (This approach gave results that agreed with performance curves for the GM EV1.) Initially, I tried using the GM Volt "Specifications" (e.g. http://www.autobloggreen.com/2007/01/07/detroit-auto-show-full-specifications-on-the-chevy-volt/), but could not get the target performance results of 60mph at 8.5 seconds. I then did a web search on the parameters for the GM Volt and found other higher performance estimates. I then guessed a conservative estimate for drag coefficient, based on statements that it is lower than 0.29. This matched the Volt target performance.

 

[Tom]

Higher Gear Ratio

Increasing the gear ratio will increase the force to the tires. A gear ratio of 7.595 causes the torque to fall off at 37 mph because it reaches max power. The torque above 37 mph = PowerMax/velocity, i.e. torque falls off with increasing velocity. End result: the higher gear ratio will get to 37 mph about 2 seconds quicker, but above about 50 mph, the acceleration is the same for the two gear ratios.

You're right. The higher gear ratio will get you faster off the line (below 50 mph). My gear criterion of max torque at 60 mph was not the best choice. Thanks for the feedback.

 

[Joshua Bretz]

But there is another dimension to the numbers: the inverter may have different higher power/torque ratings for acceleration as opposed to constant load. That is, the capacity of the magnetics to support a given flux may be different than the capacity to remove heat.

Here's some patents that highlight how much effort GM is putting into cooling the inverter and motor:

http://patft.uspto.gov/netahtml/PTO/srchnum.htm
7210304

http://appft1.uspto.gov/netahtml/PTO/srchnum.html
20080101013
20070216236
20070047209
20050035678
20050035672

 

[Tom]

Fmea

Joshua,

Thanks for the info. It appears to be wonderful new technology. Unfortunately, the more complicated something is, the greater the probability of failure. I hope these folks have done a thorough Failure Modes Effects Analysis and environmental testing. The laws of the universe dictate Murphy's law, especially when building high volume products. At high volumes, I have sometimes seen systems fail as a result of the convergence of three highly improbable/impossible events.

I would have a few concerns. This system must be hermetically sealed. Any moisture getting into the dielectric fluid with a trace of electrolyte can cause the metallization on the power device die to corrode over time. Failure or loss of power to the fluid pump could cause the power devices to overheat. The devices probably have self protecting thermal monitoring and would shutdown. The vehicle would then shutdown. I wonder if there could be vapor lock of the heat pump?

 

[KevinC]

Overheated die? What designer worth his salt pushes a device to critical limits? Parallel drivers until worse case environment nets 50% capacity per device. This isn't new.

 

[Tom]

Undersize die to minimize silicon costs.

Overheated die? What designer worth his salt pushes a device to critical limits? Parallel drivers until worse case environment nets 50% capacity per device. This isn't new.

The implication of the patents is that GM will use undersized die with fluid "refridgeration" to extract heat. If you lose the fluid flow, devices would be beyond max ratings.

 

[hvacman]

Fluid-cooled electronics are nothing new. Cray's super computers were water-cooled. If things failed there, you toasted $$$$$. Server farms are now getting so energy-dense that some are migrating to water-cooled electronics.

Most high-power, high-voltage transformers are oil-cooled, though many rely upon natural convection to circulate the fluid.

The apparently-natural convection phase-change aspect of one of the inverter cooling loops intrigues me. Sounds a lot like the refrigerant-based heat pipes that keep the tundra frozen under the Alaska pipeline.