Dual Motor (and Triple) Torque Control for Maximum Efficiency

[kennybobby]

A Sept. 11, 2014 Tesla patent for controlling an AWD EV includes claims not just for the dual motor configuration of the 85D, but also for a three motor (85T ?) version.

CONTROL SYSTEM FOR AN ALL-WHEEL DRIVE ELECTRIC VEHICLE - Patent application


Not only is there a torque calculator (based upon look-up tables) to determine the split between front and rear, but a traction controller monitoring wheel slip.

The 9-11 patent doesn't go into details of any sort of sleep or idle mode, but refers to the motors as primary and assist. See the figure 2 for torque and power curves of example motors.

In a preferred embodiment of the invention, one of the two motors is the primary drive motor, e.g., motor 103, while the second motor, e.g., motor 109, is relegated to the role of an assisting motor. Preferably both motors 103 and 109 are AC induction motors. Additionally, in a preferred embodiment assist motor 109 is designed to have a relatively flat torque curve over a wide range of speeds, and therefore is capable of augmenting the output of primary motor 103 at high speeds, specifically in the range in which the torque of primary motor 103 is dropping off. FIGS. 2 and 3 illustrate torque and power curves, respectively, of exemplary motors. In particular, curves 201 and 301 represent the torque and power curves, respectively, of an exemplary primary motor; curves 203 and 303 represent the torque and power curves, respectively, of an exemplary assist motor; and curves 205 and 305 represent the torque and power curves, respectively, of the combination of the exemplary primary and assist motors.

See Figure 8 for the algorithm used to generate the look-up table utilized by the optimal torque split unit;

 

[wycolo]

I'm curious what design differences establish flattish torque curve for motor A as opposed to less-flat TC for motor B. Or would this only be from control circuitry?
--

 

[AudubonB]

Excellent find, as usual.

A question: line 205's torque curve gives approx. 500N-m, or 370 ft-lbs. The recent article from one of the on-line auto magazines discussing a showdown between a P85 and a P85D talked of the D having something like ??? 849? 948??? ft-lbs of torque shown on their dynometer to go with the 691 hp. Where am I misunderstanding? Perhaps it's in that these "exemplary" motors are not those used in our vehicles: the power curve maxing at around 380 kW corresponds to 510 hp, if I'm reading and converting these correctly.

 

[kennybobby]

I'm curious what design differences establish flattish torque curve for motor A as opposed to less-flat TC for motor B. Or would this only be from control circuitry?
--

Some simple motor design parameters that might help explain this include:
the physical geometry (diameter and length) of the motor; the number of magnetic poles which sets the operating speed range; the electrical characteristics of the motor (inductance and resistance) determined by the number of turns in the windings and the wire gage used, which in turn set the voltage and current limits.

For example,
A high pole count motor with large gage wire will have higher low-end torque (motor B) but fall off in speed sooner than a lower pole count high-speed motor wound with smaller gage wire (motor A). Motor B is wound to use higher currents, and motor A to use higher voltages.

For those in the Heart of Alaska, i don't think those curves were intended to necessarily represent any particular production motors.

 

[kennybobby]

Here is a an interesting paragraph from the patent that explains how it optimizes the torque split in order to reach minimum power usage:

The torque required from the combined drive trains calculated by unit 601, and the maximum available torque for the primary and assist motors, calculated by units 603 and 605 respectively, are input into the optimal torque split unit 607 as is the computed vehicle speed. Unit 607 optimizes the torque split between the two drive trains without taking into account wheel slip, thus splitting the desired combined torque, i.e., C_torque, into an optimal primary motor torque request and an optimal assist motor torque request, the split based solely on achieving maximum operating efficiency within the limits of the available torque for each motor.

 

[Johan]

Here is a an interesting paragraph from the patent that explains how it optimizes the torque split in order to reach minimum power usage:

The torque required from the combined drive trains calculated by unit 601, and the maximum available torque for the primary and assist motors, calculated by units 603 and 605 respectively, are input into the optimal torque split unit 607 as is the computed vehicle speed. Unit 607 optimizes the torque split between the two drive trains without taking into account wheel slip, thus splitting the desired combined torque, i.e., C_torque, into an optimal primary motor torque request and an optimal assist motor torque request, the split based solely on achieving maximum operating efficiency within the limits of the available torque for each motor.

So basically this is what's coming in the "end of January" firmware update, which will enable the P85D to have better efficiency as outlined in JB's blog post. (And later of course the S85D to have better range than the S85).