Niels … this is not only a fantastic simulation .. but also a great explanation of the rationale for the development of each element of the hybrid drive. I never really understood why the planetary gearset was really required. Also I like how you have shown that an electric motor can serve as a clutch in an ideal world. Having watched this video I can see other applications such as diesel electric trains and ships where a planet gear could be employed to improve efficiency. Also do you know the efficiency of power transfer from ICE -> wheels .. versus ICE -> MG1 -> Inverter -> MG2 -> Wheels ? Well done.
Can someone please explain how the 'generator clutch' that replaces the millstone clutch works? The author describes it between 4:00 and 4:40 in the video. Are the coils that are connected in the video to the drive shaft supposed to be wrapped around another permanent magnet (different from the permanent magnet on the rotor connected to the engine's output shaft)? Are magnetic clutches that are fully contactless really feasible? I understand he's using it as a pedagogical device, but it just blew my mind because I'd never even considered that magnets without contact could be used as a clutch to transfer rotation.
Several of his pedagogical devices in the middle of the video are physically unbuildable, until he has added enough of the parts of the actual solution to make it actually work. His "generator clutch" shows up at a part of the video where the power-split planetary gearset hasn't been added yet. It just has a rotor that rotates and 'stator' coils that ... also rotate. Easy to do in an animation where they don't have to be mounted to anything or have any wires connected. After he mentions that later at 7:00, he adds in the planetary gearset, and now it's a thing you could build. Once you build it the real way, the MG1 stator really is stationary, and you probably wouldn't use the word 'clutch' as a description of MG1 anymore. In old-school automatic transmissions, there were things called clutches (which would use friction between two rotating parts, to transmit rotation) and other things called bands (which would use friction between the stationary case and a rotating part, to stop that rotating part and force some other part of a planetary gear train to rotate instead). That's closer to what's happening in MG1 in real life: the electromagnetic forces resist the rotation of MG1, forcing rotational power to pass between the engine and the output. Where the 'band' analogy breaks down is you never want a friction band to be partially applied. When the band is fully released and the part is spinning freely, that's fine. When the band is fully applied and the part is stopped, that's fine. Any intermediate stage where the part is making frictional heat against the band is only happening while the band is applying or releasing, which is a stage to be passed through as quickly as possible. When you replace a friction band with an MG, now you don't have to avoid those 'partial' stages anymore. They don't make damaging, wasted frictional heat; they just make electricity, which you can use. So in the Prius transmission that's what's actually happening most of the time. The MG1 rotor is spinning, but not spinning freely: there are electromagnetic forces restraining its rotation (and so making power flow mechanically through the gears), and the side-effect of electrical power out of or into MG1's coils also serves as a separate, electrical, power path. When you prevent any current flowing in MG1's coils, you allow MG1 to freewheel (like a band fully released) and you have the neutral function. The case where you hold MG1 to zero RPM (like a band fully applied) is possible, but doesn't really happen a lot. A three-phase motor is usually driven by sending changing currents into the three phase windings so the resulting magnetic field has an orientation that changes (and the permanent-magnet rotor follows it around). To just hold the rotor still, the ECU has to know the rotor's position—that's what the 'resolver' is for—and send, effectively, DC into the stator coils, proportioned among the three so the resulting stationary magnetic field is holding the rotor in place. Doing that isn't quite as efficient as holding something stopped with a friction band. The friction band wastes energy when it's partially applied and there's motion and friction and heat, but there's really no energy being lost when the part is really held at zero RPM. Holding the permanent-magnet rotor stopped using a steady winding current costs you some power. The winding current must be strong enough to resist whatever torque is on the rotor, and that current produces some waste heat according to the DC resistance of the windings. That may sound alarming if you are used to hearing the terms "inrush" or "locked-rotor" in connection with industrial motors. When you slam full voltage into a motor whose rotor isn't turning yet, you get a very large current and heat that would burn the motor right out if the rotor didn't come up to speed quickly. That's not what's going on with a Prius MG. To hold the rotor stationary only requires 'enough' current, and the inverter is able to supply just 'enough' voltage to do that. It isn't slamming full voltage into a stopped motor. We talk of these as being high-voltage motors (300-ish volts in gen 1, 500 in gen 2, 650 in gen 3, I don't know the later numbers), but those are the voltages needed at speed (you have to overcome the "back EMF" that the spinning motor generates back atcha). In the zero RPM case, with no back-EMF at all, you can get substantial torque out of them with more like eight to sixteen volts. At that rate you are producing some waste heat, but measured more in the hundreds of watts, not thousands. That stopped-rotor situation was considered in more detail in ➡this post⬅. Returning to the original question, if you really had a way to build the free-floaty-no-wires-attached stator from 4:00 in the video, you could indeed make it work like a clutch, again by using steady winding currents to hold the rotor and stator in a fixed position relative to each other. You wouldn't need extra permanent magnets or anything. But you would have the energy loss of the steady DC winding current just described, compared to a real clutch, which doesn't lose any energy when fully engaged.
Wow @ChapmanF . The physics/electronics of controlling MG1 (sun gear) RPM fascinates me and is what got me digging deeper. Assuming the engine(planet carrier) always spins at its ideal RPM, I understand that for any given vehicle speed (which fixes the ring gear RPM) there is an RPM for MG1 which it should be held at so that all of the engine's torque reaches the wheels. It's impressive how you intuited my underlying quest even though I was asking about the simplified drive train presented by OP. I loved that OP simplified the hybrid drive train and I wanted to be 100% sure I understood the physics of his simplified drive train before I go deeper down the prius e-cvt rabbit hole. Questions below At 9:51 in the video OP does mention that sliding contacts would allow his simplified hybrid drive train to work. Assuming such contacts do their job, how would the DC winding currents you mention work to make the drive shaft RPM equal to the RPM of the rotor (which is driven by the engine)? Will the DC winding currents create a magnetic field which causes the windings to keep chasing the rotor? Do you have a sense for how huge these currents have to be to create a magnetic field strong enough to not have any slippage? If we start the DC winding currents and engage the magnetic clutch before the engine is started (should be easier at 0 rpm), will the clutch slip when the engine starts? If the answer is yes, is it because there's a limit to how much torque the magnetic clutch can pass through? How does the energy usage of the magnetic clutch scale with the torque it's rated for? I think I also understand now why the electric motor on the drive shaft (4:58 in the video) is almost a requirement in his simplified drive train (OP made it sound like it's only added to save energy, but i think it's a requirement). Without the motor, it would be much harder to synchronize RPM of the magnetic clutch windings with the magnetic clutch rotor, especially at startup. I'd think it's best to start the DC winding currents only after the motor syncs the clutch rotor and the clutch windings. Am I wrong?
You can simplify the picture by imagining yourself in a reference frame that moves with the (floating, rotating, sliding-electrical-contact coils). Now to you it just looks like the scenario from my earlier post where the coils are a stator and you're just holding the rotor locked to it in one unchanging relative position. That change of reference frame, by the way, isn't just a trick I'm using here to make something easier to see; it's the exact basis of the Park and Clarke transforms the car's software uses to simplify calculating the three-phase currents to use (and not just for the locked-together case, but for the case of relative rotation too). Tip of hat here to Edith Clarke, first woman professionally employed as an EE in the United States, first female EE professor in the country, and first woman to present a paper at the American Institute of Electrical Engineers. Not as huge as you might think. It helps to remember you don't need to make a magnetic field strong enough to resist any conceivable torque. You need it to resist the largest torque it's going to see. In a Prius transmission, MG1 always sees 30/108 of the torque at the engine crankshaft. The engine torque rating depends on the Prius generation. We could use 142 N⋅m, which is the figure for the gen 3 engine. 30/108 of that is about 39 N⋅m of MG1 torque. Feed in DC currents adequate to make that much torque and the rotor is going to stay put. I'm not sure that we have a graph of MG1 winding current to torque. In the old post I linked above, we do have that graph for MG2, where 39 N⋅m of torque would call for somewhere in the neighborhood of 50 amps. This is where you have to remember not to think always in terms of the full motor rated voltage (650 V for gen 3). If you blindly multiplied 50 A ✕ 650 V, you'd think yikes! 32,500 watts of power just to hold this thing still?! But those high rated voltages are only needed at speed, to overcome the back-EMF that's being generated as the rotor spins. When there's no relative motion, there's no back-EMF, and when the currents are effectively DC, the amps-to-volts relationship depends just on the windings' actual resistance, which is measured in milliohms. It probably takes no more than ten volts to make 50 amps flow in those windings at DC, so that's just 500 watts of power: not to be sneezed at, but not off the scale of other possible power losses in a drivetrain. And that's a maximum, of course: to hold MG1's rotor still, do you even need to make a strong-enough magnetic field for the maximum torque that could be delivered? No, you just need to resist whatever torque it is experiencing at the moment. Much of the time, that will be a lot less. Still, whatever electrical power you do use to hold a rotor stationary is essentially all wasted. Sure, you've ensured all the engine torque goes through to the output, but the electrical power you're using to do that is contributing no additional mechanical motion on top of that. It is only making MG1 warm and being lost. That means there's a difference between the most-efficient way to use a real clutch, and the most-efficient way to use MG-as-a-clutch. When you have a real clutch, your most-efficient state is when it is locked together with no relative motion and no frictional loss. With MG-as-clutch, that condition of no-relative-motion ends up being a local dip in your efficiency curve; you have to supply a modest amount of electrical power to do it, and of that power you're supplying, none is contributing to motion and all is going to heat. Shifting the operating point in either direction from that local dip would be an improvement. You could allow some relative motion so the MG is generating some electrical power for you, which you can ship to the other MG to add to the engine's contribution, or ship to the battery for use later. Or you could electrically drive the MG in the other direction, using electrical power to do it, but getting some motion to show for it, not exclusively heat. So if you were writing the firmware to operate a Prius drivetrain as efficiently as you could, chances are you would not program it to seek out that zero-MG1-RPM state. You would, instead, tend to program it to skip over that region when possible, and prefer operating on either side of it.
Once again @ChapmanF you answered questions before I could even ask them well. I'd added a few questions to my previous post a couple of minutes before you posted, so you couldn't have read them, but you answered them anyway. It all makes sense now. I tip my hat not just to Edith Clarke, but to countless others who make her work accessible, including you. Also glad to see some numbers (500 watts peak seems higher than I was imagining, but i'm glad you shared some rough numbers) Once again this is so fascinating. The fact that allowing some slipping (and generating electricity to use elsewhere) is more efficient than no-relative-motion is so counter intuitive. Are you 100% sure this is true, specifically in the context of OPs simplified hybrid drivetrain?
My opinion would count less than that of a credentialed EE who worked on the programming for a PSD drivetrain, if you can find one of those. But for what my opinion is worth, I'm pretty sure of it. There has to be current flow to achieve any electromagnetic effect in an MG. In almost all cases, that accomplishes some amount of mechanical work, and creates some waste heat. The exception is when you're holding RPM at zero: the mechanical work is torque ✕ RPM, so at zero RPM you're accomplishing none of that, but you're still making the waste heat. It may, at that point, be reduced to just a few hundred watts of waste heat, but when that's 100% of what you're accomplishing, it means your efficiency—mechanical work out ÷ electrical work in—is genuinely zero. So zero RPM is the lowest point of an MG's efficiency-vs.-RPM graph, considered in isolation. I'm not saying that's the lowest point of the whole drivetrain's efficiency graph, as it may be a point where mechanical power is passing along the other path pretty efficiently. But I would expect it to represent at least a local dip in the graph of efficiency for the whole drivetrain, that you could profit by operating to one or the other side of. Making an actual graph of efficiency for the whole drivetrain, or saying just how far away from zero MG RPM the sweet spots would be, would call for detailed modeling that I definitely haven't done.
