Honda’s F1 KERS motor: 60 kW, 21,000 RPM, >7 kg
25 November 2010
by Jack Rosebro
|2009 Honda Formula One chassis configuration, showing KERS component location. Click to enlarge.|
During the EVS 25 electric vehicle expo in Shanghai earlier this months, Honda engineers presented a glimpse into the development of an advanced high-performance electric motor that Honda had designed as part of a KERS (Kinetic Energy Recovery System) intended for its 2009 Formula One race car.
KERS (Kinetic Energy Recovery System) regenerative braking systems are limited to storing up to 400 kilojoules of energy per lap, which can then be released at up to 60 kW (80 hp) for up to 6.67 seconds as part of a “push-to-pass” strategy. Depending on track design, this can increase vehicle acceleration by up to 15 kilometers per hour, gaining up to 20 meters of distance per lap.
However, KERS systems present significant design challenges in that they must be implemented without adversely affecting vehicle aerodynamics, weight, weight distribution, collision safety, fuel tank capacity, or center of gravity. Honda engineers decided to implement an electric motor/battery pack solution, rather than a flywheel solution, mounting the motor on the engine’s left front side with its power control unit (PCU) ahead of it, inside the monocoque chassis.
The motor was cooled by engine oil and the PCU was cooled by a dedicated coolant loop. The 106-cell lithium-ion battery pack was mounted in the forward section of the vehicle’s keel to preserve the vehicle’s center of gravity and take advantage of draft air cooling.
Given the light weight (minimum 620 kg/1,367 pounds) and compact design of Formula One cars, Honda calculated that the motor would have to be no more than 100 mm (4 inches) in diameter, 200 mm in length, and produce approximately 8 kilowatts (kW) per kilogram (almost 5 hp per pound) of motor weight. By comparison, a typical mass production hybrid or electric vehicle motor produces between 1.0 and 2.5 kW per kilogram.
Stator Core. A three-phase, four-pole, twelve-tooth design with a double lap-wound stator and permanent magnet rotor was selected. Operating motor speed would be roughly equivalent to engine speed, with a range of 13,000 to 21,000 rpm (engine speed was limited to 18,000 rpm during the 2009 F1 season).
An initial concern, particularly given the high motor speed, was the effect of iron losses in the stator core, which is typically an AC motor’s heaviest component. Conventional grain-oriented silicon steel was not efficient enough to be used in a motor that would meet the project’s motor size and weight targets, and an iron-cobalt alloy (49Fe-49Co-2V) was used to produce the motor’s stator core laminations. This yielded a 30% increase in flux density and a 15% increase in torque.
|Relative magnetic flux densities of 49Fe-49Co-2V iron-cobalt alloy, as compared to more common silicon steel. Click to enlarge.|
The iron-cobalt alloy’s iron losses were further reduced via a combination of technologies. A post-rolling heat treatment reduced core hysteresis losses, and an ultra-thin oxidized insulation coating was developed, which allowed the engineers to reduce stator core lamination thickness to a tenth of a millimeter per lamination while preserving the desired iron-to-insulation ratio. These refinements reduced the alloy’s iron losses by a further 60%.
|The magnetic flux density of Honda’s iron- cobalt alloy was increased through multiple refinements. Click to enlarge.|
Although Honda has not released details of the motor control strategy, the maximum current frequency for a 21,000 rpm four-pole motor, assuming an equal number of poles in both rotor and stator, is 700Hz, and pulse width modulated voltages driving a motor at such a speed must inherently use extremely fast transients, creating voltage stresses within the winding. Not surprisingly, the stator winding was wound with scratch-resistant inverter duty wire to reduce the possibility of pinholes created during the manufacturing and/or winding process, and therefore reduce the possibility of winding failure.
Rotor Design. Honda developed a high-coercivity magnet with an intrinsic coercivity of at least 1.1 Ma/m at 160ºC (320ºF), and tuned the magnetization angles for maximum torque. To minimize temperature increases and resultant eddy current losses, 448 magnets (28 axially, 16 circumferentially) were used in the rotor’s interior PM configuration.
Rotor diameter was reduced by employing the rotor’s shaft as part of the rotor flux circuits. A high tensile filament winding made of organic high-strength fibers encloses the rotor, preventing magnet burst at the centrifugal forces produced at the rotor’s 21,000 rpm redline. The rotor’s ceramic ball bearings were lubricated by high-temperature grease rather than oil, to simplify the oil circuit and reduce losses.
|Section view of motor, showing rotor and stator coolant flow paths; engine oil is used as coolant. Click to enlarge.|
A water-cooled motor was initially developed, but could not meet Honda’s size/performance targets. The final motor was cooled by engine oil, with stator cooling passages around the circumference of the rotor and the rotor cooled by oil passing through its hollow shaft. A thin cylindrical resin sleeve structure was mounted in the motor’s rotor-stator air gap to isolate stator cooling oil from the rotor and eliminate windage losses.
|Front-mounted gearbox, which transfers motor torque to the engine crankshaft. Click to enlarge.|
Implementation and testing. Motor torque was delivered to the engine’s crankshaft via a front-mounted five-gear gearbox. Vehicle testing commenced in April 2008 with straight course accelerations, and progressed into circuit tests at Silverstone the following month. Additional circuit tests were conducted at full load in September, at the Jerez circuit in Spain, yielding the following results:
- Lap times were reduced by ~0.4 seconds per lap at full assist;
- Speed was increased by 7 km/h, resulting in a gain of 7.8 meters (1.6 car lengths) on a straightaway with a continuous 324 kJ of assist.
The final design achieved 7.8 kW (10.46 hp) of power per kilogram, close to Honda's design goal of 8 kilowatts per kilogram. Peak motor efficiency was 99% and peak generator efficiency during regen was 93%. Motor weight was 6.9 kilograms.
|Estimated lap time reductions on FI circuits, using Honda's KERS motor drive. Click to enlarge.|
KERS systems were introduced to Formula One during the 2009 racing season, and were key to several race wins, but were subsequently withdrawn due to development issues. KERS is expected to return to F1 in 2013, concurrent with the introduction of new engine regulations. Honda sold its Formula One team to Brawn GP Limited in 2009.
Development of F1 KERS motor, November 2010 Tamotsu Kawamura, Hirofumi Atarashi, and Takehiro Miyoshi, Automobile R & D Center, Honda R & D Co., Ltd.
For a 4 pole motor to do 21,000 rpm requires 700 Hz. The 42 kHz being referred to is the PWM carrier frequency inside the inverter.
Posted by: Paul | 25 November 2010 at 12:55 PM
Good correction, Paul.
Lots of good info here (at least for us uninformed) not including typos like vehicle acceleration is measured in km/hr.
Posted by: ToppaTom | 25 November 2010 at 02:59 PM
Wonder how long it will be until some very smart person applies these motor efficiencies (Peak motor efficiency was 99% and peak generator efficiency during regen was 93%.) to a small standard EV and we start solving some of our acceptance problems.
Posted by: Lucas | 25 November 2010 at 03:01 PM
I hope, more successfully;
"Lap times were reduced by ~0.4 seconds per lap at full assist"
Posted by: ToppaTom | 25 November 2010 at 03:06 PM
Great info. I wish they gave us more specs on the battery pack as well. I was very curious as to it's size, weight, voltage, etc but never found any team giving out that info.
I was also curious if they were able to make it through a season with the same batteries or if they had to be replaced with the incredible power/kg they were probably putting into it during regen braking.
Posted by: DaveD | 25 November 2010 at 05:39 PM
Certainly are excellent efficiencies to put to use in a PHEV range extender.
Posted by: Reel$$ | 25 November 2010 at 06:00 PM
Yes, if the same efficiencies were applied to current e-drive and ICE genset, future PHEVs and BEVs could be under one tonne and perform much better then current 3-tonne ICE monsters.
Super reinforced plastics + advanced next generation batteries will also contribute to make extended range (500+ Km) post 2015 BEVs.
Posted by: HarveyD | 25 November 2010 at 07:23 PM
A nearly century old dodge automobile had an integrated starter generator. Do they still have a starter motor for this engine as well. ..HG..
Posted by: Henry Gibson | 25 November 2010 at 07:29 PM
It took ICE over 120 years to reach 30% efficiency and tank to wheel efficiency of about 17%. BEVs can do much better (up to 5x) in the next 10 years or so. ICE vehicles have a very long way to go and may never catch up to BEVs.
If we could recover all the energy wasted with ICE vehicles in the last 120 years, we could operate equivalent BEV fleet (free) for the next five or six centuries.
Much the same could be done with domestic and industrial HVAC. Advanced electrified systems using COP 4.0 to 6.0 Heat pumps instead of very low efficiency wood, coal, oil, NG based systems could reduce energy used by five times and more.
The technologies are there for both. Obstructions to change are numerous. Interested parties are fighting it with success.
Posted by: HarveyD | 26 November 2010 at 08:29 AM
@ Paul: Thanks for the correction. I forgot to convert from minutes (RPM) to seconds (Hz). The Honda paper, however, does not give any details about inverter frequency.
@ ToppaTom: The SI unit for "hour" is h, so we use km/h.
Posted by: Jack Rosebro | 26 November 2010 at 10:20 AM
What I meant was - there must be a typo -
". . increase vehicle acceleration by up to 15 kilometers per hour"
". . accelerate the vehicle by up to 15 kilometers per hour [on the straight?]" A common (and acceptable) use of the term acceleration, used for an increase in speed.
Technically, SI acceleration is in meters per second per second (m/s2).
Posted by: ToppaTom | 28 November 2010 at 10:36 AM
I don't believe Formula 1 cars have starter motors at all, they rely on an external system. So once a race starts the driver has to keep the engine running otherwise it is game over.
Posted by: Patrick | 29 November 2010 at 05:26 AM
Posted by: Henry Gibson | 05 December 2010 at 11:40 AM