Honda Power Unit Hardware & Software

[AJI]

33s has not been a real limit for some time.

At the beginning of last season, Ferrari stated 50s/lap deployment of their ERS. That's a total of 6MJ.

Mercedes is more.

Did they specify the output for that 50 seconds? The whole '33 seconds per lap' thing is kinda silly anyway because it implies that the drivers use the K like a light switch until the 4MJ SoC cycle is complete, which of course, they don't (or rather, the CE doesn't).
Direct H-K transfer is the truly mysterious bit.

[AJI]

There's a note on the diagram that says the ES State of Charge isn't allowed to deviate by more than 4MJ between maximum and minimum per lap.

And that's the kicker. The rules specify that you can only use 4MJ from the ES per lap, but with a 25kg ES you should be able to carry at least 12MJ of storage (I'm not sure if the weight is batteries only or everything associated including the case), so to my mind it's all about how you use that 4MJ in conjunction with direct H-K transfer. When we occasionally hear the drivers talk about de-rates I don't think the ES is empty, just that they have used the 4MJ limit for that lap.

[hurril]

I don't understand the recent thoughts of going completely haywire with transfers from the "k" to the "h"; where is that energy coming from do you think? It is not unlimited in the sense that there's an over-abundance in the source; it's unlimited in the very literal sense: you are free to transfer whichever amount you can produce. You cannot produce an unlimited amount of energy.

I think flywheel is a bad term when applied to the "h"; the amount of potential energy recoverable is probably way less than that which can be tapped into by "braking" against force/ torque applied either by the turbine or by this newfound theory of switched motoring.

But remember that this "k"-driven switched motoring is driven by physical fuel so it is an accounting trick more than one relating to efficiency. Min-maxing if you will.

[hurril]

There's a note on the diagram that says the ES State of Charge isn't allowed to deviate by more than 4MJ between maximum and minimum per lap.

And that's the kicker. The rules specify that you can only use 4MJ from the ES per lap, but with a 25kg ES you should be able to carry at least 12MJ of storage (I'm not sure if the weight is batteries only or everything associated including the case), so to my mind it's all about how you use that 4MJ in conjunction with direct H-K transfer. When we occasionally hear the drivers talk about de-rates I don't think the ES is empty, just that they have used the 4MJ limit for that lap.

It actually doesn't mean that. One lap could end at X, the next at X +/- 4Mj; what you do inside of that lap, however, is between you and God.

So at point P, you could be 8Mj over and at P+n, you're at -10Mj. So long as the delta lap-over-lap is <4Mj.

Also: you don't even have to go to the extremes I used there; you could just as well deposit and withdraw smaller amounts as the corners turn into straights and the straights into corners; the sum of deposits could be 10Mj and you'll be fine so long as the sum of the withdrawals are >6Mj (in this particular example.)

This rule "just" means that you are unable to charge up to a state where you're be able to run 100% electric compressor and "k" for several laps. It moderates the race pace.

[henry]

The wording of the SOC regulation refers to “any time the car is on the track” and not per lap. So you can cycle in the range between 0 and 4 (or 1 and 5) as much as you want.

[Hino]

Amazon's full resolution sample is online

https://abload.de/img/honda1dtuk2.jpg

https://abload.de/img/honda2gju8l.jpg

https://abload.de/img/honda31fubs.jpg

https://abload.de/img/honda4ffu02.jpg

I am currently doing a full translation of these diagrams, completely replacing the Japanese text with English, I'm almost done. Hopefully, if time permits, I will have everything uploaded within 24hrs.

[Craigy]

I don't understand the recent thoughts of going completely haywire with transfers from the "k" to the "h"; where is that energy coming from do you think? It is not unlimited in the sense that there's an over-abundance in the source; it's unlimited in the very literal sense: you are free to transfer whichever amount you can produce. You cannot produce an unlimited amount of energy.

I think flywheel is a bad term when applied to the "h"; the amount of potential energy recoverable is probably way less than that which can be tapped into by "braking" against force/ torque applied either by the turbine or by this newfound theory of switched motoring.

But remember that this "k"-driven switched motoring is driven by physical fuel so it is an accounting trick more than one relating to efficiency. Min-maxing if you will.

What I'm trying to describe isn't actually that easy to describe without an animation because by the nature of the topic we're talking about energy flow directions changing around. I feel like I'm doing a bad job of explaining it - sorry for that.

The idea that the MGU-H as a flywheel provides a route for the MGU-K to dump whatever energy is needed into the ES, within the limitations described above in my earlier post. Yes, this will ultimately be from fuel and the IC, of course. It's not generating energy, it's just permitting a workaround to get more into the ES on a lap by lap basis.
It's an energy transfer route which wasn't/isn't (to me) immediately obvious on the flow diagram that form the technical regulations.

In terms of accountancy for energy, if you could take energy from the end of every straight and deploy it at the start of the next straight, that isn't just accountancy with energy. It's a larger area under the speed graph, and a faster laptime. 500KJ at the end of the Kemmel isn't as helpful as 500KJ on the run down to Pouhon, in terms of laptime.

People mention "steady state" and that made me think that the flow on the diagram wasn't changing from K to H and back, but thanks to Honda's lovely end-of-2017 information we know that's a poor assumption: this is being done.

Regarding the terminology of what I've called the "flywheel" transfer technique - please bear in mind that it's not like we're spinning the MGU-H up once per lap to huge speed and then deploying it all on a straight or something like that, totally ignoring its role in the turbomachinery of the ICE. It wouldn't actually store very much power anyway unless it was really heavy. The MGU-H would instead be spun up and down many times per second, storing and releasing a smaller amount of energy each time, and doing so very frequently. It would essentially just be noise on the graph of the required ICE compressor speed - for example could be plus or minus less than 1% (~1000rpm either side) of what you'd otherwise want just for ICE usage, from the range of 0 to 125K RPM for the MGU-H (working range is likely 80K-120K overall). It's not like the compressor is directly attached to the crankshaft or something: the driver won't feel it spinning up and down by that much. I doubt that it'd even show up at the intake valves because the whole air intake system is full of compressed air that would absorb pulses of higher and lower pressure like a spring.

I offer the opinion that not only is this flywheel effect something that needs to be designed-in, the optimal way to get it working (staying at an optimal 4KG mass for the MGU-H) would be to make the disk (flywheel diameter) bigger than needed for the compressor or turbine needs. I mention this because it's been publicised that Honda moved their compressor to make it larger, and this is a potential reason for that, besides the needs of the ICE's compressor.
In Hondas case that would potentially require that the compressor be moved from the Vee of the engine to the front or the back in order for the diameter to grow so that the flywheel energy storage potential is larger to facilitate transfers per pulse.

Here are Some very rough flywheel storage figures - say we are trying to move energy from the K to the ES via the H, and we're doing that for 50s per lap (somewhere like Spa, which is 57% full throttle by time and 70% full throttle by distance on a 110s race lap). At no point are we trying to recover energy to the ES from the ERS-K via ERS-H route while the driver is at 100% throttle.

A few stated assumptions: 50s duration, flywheel of 4Kg (the minimum in the rules), 100mm diameter (seems sensible enough for the turbo sizing alone), 40 pulses per second, 2000rpm rev change (of the MGU-H) during a pulse. Energy per pulse depends on the mass distribution on the MGU-H (mass in the centre has less inertia than the edge), but lets say that it's something like 180J because the majority of the mass is on the outer ring of the MGU-H, due to some clever designer putting weight out there.
In this case, we have 50s per lap * 40 events per second * 180J per event = only an additional unmetered 360KJ per lap. That's only 3s of full deployment and hardly worth the trouble. You could go for a larger RPM change, at cost of the ICE seeing more and more variable levels of manifold pressure.

If we adjust the shape of the flywheel (make it 200mm diameter rather than 100mm above) and leave everything else the same:
Then in this case we have 50s per lap * 40 events per second * 700J per event = 1.4MJ per lap. Quite worthwhile. Perhaps worth moving the flywheel disk out of the Vee of the engine so make it larger without being heavier, permitting more flywheel inertial storage without a weight penalty.

In terms of the effect on the ICE of doing all of this, it would see some variations in manifold pressure and in exhaust pressure due to the speeding up and slowing down of the MGU-H, compressor and turbine, but the variations would be so fast that it really wouldn't make the ICE feel different to drive than if the system wasn't in place.

In terms of mapping the ES deployment/charge, the big plus point of this system is to work around the 2MJ transfer limit from MGU-K to ES. You'd essentially have this system pulling power from the IC via the MGU-K, dumping it into the MGU-H and on to the ES whenever the driver isn't requesting 100% from the PU - it could be recovering any time on part throttle or zero throttle, as required. You would still use the ERS-K to ES direct route for the 2MJ permitted per lap of course, but the rest of the energy would be most effectively recovered using the flywheel method.

In terms of redeploying ES energy, there's a flow limit of 4MJ from the ES to the MGU-K using the direct route per lap, but again this could be worked around because there's nothing stopping the route from the ES to the K via the H in flywheel-routed "deployment" mode.

The 4MJ SoC ES limits don't appear to stop the car from actually receiving and deploying more than 4MJ per lap because of how that rule is worded.
It appears that the single biggest deployment event you can have within a lap is 4MJ (taking the ES from "full" to "empty"), but so long as you discharge and recharge you can cumulatively have more than 4 MJ entering and leaving the ES in a given lap. The ES to K *direct* route is limited to 4MJ out per lap, and 2MJ back to ES. There's no limit on ES to MGU-H transfers or rate, remember.

The route via the flywheel system suggested would raise those 2MJ and 4MJ limits significantly, potentially to the point where the MGU-K is deploying 120kW the whole time the PU is at 100% output.

Can I make a hypothetical scenario to explain this? (might not be that hypothetical, as it may correlate to Honda's first pass at the PU a few years ago):

If you were limited to (for example) 2MJ of energy into the ES per lap because you'd designed a system where only the MGU-K charged the ES, then you'd be limited to 2MJ of deployment on the next lap. You'd hear drivers complain a lot about a lack of deployment, because your 2MJ is only enough for 16.6s of deployment round a given lap. Does this sound familiar?

You would need to figure out how to add to the energy input into the ES, over the 2MJ you'd started off with.
You're potentially already at the 2MJ direct limit from the MGU-K from braking and from charging towards the end of straights, so you need to add whatever you could get from MGU-H using compounding. This is essentially limited by what the ICE's operating parameters are: the higher the amount of energy remaining in the exhaust gases, the worse the basic ICE performance was at driving the rear wheels directly, and the more fuel-hungry the ICE was.

If instead of this, you designed a system to pull energy from the K via the H into the ES, you have the ability to charge the ES at an abritrary amount per lap, and deploy 4MJ directly via the MGU-K on every lap as you please.

You can do that deployment wherever you want (start of straights, of course, but actually on quite a few circuits, most of the lap because you have 33s to play with).
You also have the ability to supplement this direct ERS-K drive route with energy taken from the ES to the MGU-K, then dumped in small pulses into the MGU-K using the "unmetered" ES->H->K route.
Your 120kW deployment is in fact made of pulses of power direct from the ES to the MGU-K, interleaved with pulses of power from the MGU-H to the MGU-K.
Your deployment per lap consists of 4MJ direct from ES to K plus the additional energy using the unmetered ES to MGU-H to MGU-K route, within the constraint that the ES State of Charge can't go above 4MJ or below 0MJ from their starting values.

Your deployment strategy now revolves around maximising the factors you control: ES SoC and MGU-H sizing, along with where on the lap to charge and where to deploy, using each system.
The more power the MGU-H has been designed to transfer, the physically bigger it will be
There's a compromise in terms of volume versus weight versus power rating.
You likely size the MGU-H to cope with the harshest duty-cycle scenarios on the calendar (Spa?).

Question: What would people prefer we call this technique if not "flywheel"?

[Craigy]

Amazon's full resolution sample is online

https://abload.de/img/honda1dtuk2.jpg

https://abload.de/img/honda2gju8l.jpg

https://abload.de/img/honda31fubs.jpg

https://abload.de/img/honda4ffu02.jpg

I am currently doing a full translation of these diagrams, completely replacing the Japanese text with English, I'm almost done. Hopefully, if time permits, I will have everything uploaded within 24hrs.

That would be most helpful.

[Jejking]

There's a note on the diagram that says the ES State of Charge isn't allowed to deviate by more than 4MJ between maximum and minimum per lap.

And that's the kicker. The rules specify that you can only use 4MJ from the ES per lap, but with a 25kg ES you should be able to carry at least 12MJ of storage (I'm not sure if the weight is batteries only or everything associated including the case), so to my mind it's all about how you use that 4MJ in conjunction with direct H-K transfer. When we occasionally hear the drivers talk about de-rates I don't think the ES is empty, just that they have used the 4MJ limit for that lap.

With Craigy ^^.

Thought it was due to the flowing nature of the Sepang circuit and many corners where you need boost to get maximum acceleration, that caused track-specific problems for the teams. You can't spend more than 4MJ at all times, with what you manage to recover. I'm taking Hamilton's problems in Baku and Malaysia for example, because he's driving the best systems that are out there. Basically they spend time in the sim and free practice to get data to know where to harvest and deploy best for best overall laptime. Then the system learns where to cut in, probably based on GPS data. Certain tracks are easier than others. "Canada, 4.5km, is easy for us to do (multi-lap continuous deployment).

My point is: this optimization process makes sure they already know the amount of regen and they will manage the energy flow around those problems. If the demand is greater than the input, they must be aware. It implies that it's the ES going out of oomph too soon. The drivers ran into de-rates, but cured this with settings in Baku. Hamilton flipped a switch too many and so did Rosberg. because the system is already optimized for 4 MJ deployment and the problems vanished later on. Laptimes came back down. You can't magically change an optimized system I believe, only deploy less and save battery.

As I am typing this, I'm starting to doubt if the drivers can do more than only toy around with harvesting and deployment? With all the menus and submenus, can they influence car-part specifically, f.e. how much power is used from MGU-H to power MGU-K?

Hope my answer makes sense to you, finding it hard to get my head around the topic

[Tommy Cookers]

well I commend Craigy on his persistence

and tend to agree regarding the MGU-H rpm suitable for the energy transfer flywheel (as being in the normal range)
because MG torque will be limited by permissible current, MG power will be good only at reasonably high rpm
though of course there will be competing factors tending to limit rates of change of MG load

[MrPotatoHead]

Craigy is 100% right on the fact that without some nice animations it's very hard to describe the energy flow over time in just words. It is incredibly complex for sure.

Also - you have to take it with a grain of salt when you hear someone like Ferrari say "50 seconds per lap" because without knowing the duty cycle and thus power output for those 50 seconds it's impossible to know the energy total actually released.

[henry]

Craigy’s posts are very thought provoking.

They reminded me that with the advent of this formula I tried to understand the power requirements for the MGU-H to defeat lag. I think the numbers are also relevant to the “flywheel” discussion

Rotrex advertise an electrically driven centrifugal supercharger, C38-61, which has similar sort of flow/pressure profiles to the F1 numbers. It has a rotational inertia of 9.0E-3 kgm2. I doubled that to allow for the turbine and MGU-H itself giving 1.8E-2 to be accelerated.

In the driving part of the cycle this inertia will be driven by the MGU-K, 120 kw, plus the excess power from the turbine, that is the power in excess of that to drive the compressor. Obviously the latter is very variable, I’ll use 30kw for calcs. So the driving phase is 150 kw. Assuming 20 hz and symmetrical drive and recovery, and a starting speed of 110,000 rpm the speed will go up only 180 rpm. So the fluctuations will be relatively small.

Assuming symmetry the recovery will be 150 x .025 = 3.75 kJ. This gives 75 kJ/sec.

This would mean the MGU-H outputting 150kw in this mode vs the much lower, maybe 60 kw, in sustain mode.

Craigy made the good point that the cycle does not need to be symmetrical. If the MGU-H could generate at 300kw the output per cycle would double.

[AJI]

I don't understand the recent thoughts of going completely haywire with transfers from the "k" to the "h"; where is that energy coming from do you think? It is not unlimited in the sense that there's an over-abundance in the source; it's unlimited in the very literal sense: you are free to transfer whichever amount you can produce. You cannot produce an unlimited amount of energy.

I think flywheel is a bad term when applied to the "h"; the amount of potential energy recoverable is probably way less than that which can be tapped into by "braking" against force/ torque applied either by the turbine or by this newfound theory of switched motoring.

But remember that this "k"-driven switched motoring is driven by physical fuel so it is an accounting trick more than one relating to efficiency. Min-maxing if you will.

What I'm trying to describe isn't actually that easy to describe without an animation because by the nature of the topic we're talking about energy flow directions changing around. I feel like I'm doing a bad job of explaining it - sorry for that.

The idea that the MGU-H as a flywheel provides a route for the MGU-K to dump whatever energy is needed into the ES, within the limitations described above in my earlier post. Yes, this will ultimately be from fuel and the IC, of course. It's not generating energy, it's just permitting a workaround to get more into the ES on a lap by lap basis.
It's an energy transfer route which wasn't/isn't (to me) immediately obvious on the flow diagram that form the technical regulations.

In terms of accountancy for energy, if you could take energy from the end of every straight and deploy it at the start of the next straight, that isn't just accountancy with energy. It's a larger area under the speed graph, and a faster laptime. 500KJ at the end of the Kemmel isn't as helpful as 500KJ on the run down to Pouhon, in terms of laptime.

People mention "steady state" and that made me think that the flow on the diagram wasn't changing from K to H and back, but thanks to Honda's lovely end-of-2017 information we know that's a poor assumption: this is being done.

Regarding the terminology of what I've called the "flywheel" transfer technique - please bear in mind that it's not like we're spinning the MGU-H up once per lap to huge speed and then deploying it all on a straight or something like that, totally ignoring its role in the turbomachinery of the ICE. It wouldn't actually store very much power anyway unless it was really heavy. The MGU-H would instead be spun up and down many times per second, storing and releasing a smaller amount of energy each time, and doing so very frequently. It would essentially just be noise on the graph of the required ICE compressor speed - for example could be plus or minus less than 1% (~1000rpm either side) of what you'd otherwise want just for ICE usage, from the range of 0 to 125K RPM for the MGU-H (working range is likely 80K-120K overall). It's not like the compressor is directly attached to the crankshaft or something: the driver won't feel it spinning up and down by that much. I doubt that it'd even show up at the intake valves because the whole air intake system is full of compressed air that would absorb pulses of higher and lower pressure like a spring.

I offer the opinion that not only is this flywheel effect something that needs to be designed-in, the optimal way to get it working (staying at an optimal 4KG mass for the MGU-H) would be to make the disk (flywheel diameter) bigger than needed for the compressor or turbine needs. I mention this because it's been publicised that Honda moved their compressor to make it larger, and this is a potential reason for that, besides the needs of the ICE's compressor.
In Hondas case that would potentially require that the compressor be moved from the Vee of the engine to the front or the back in order for the diameter to grow so that the flywheel energy storage potential is larger to facilitate transfers per pulse.

Here are Some very rough flywheel storage figures - say we are trying to move energy from the K to the ES via the H, and we're doing that for 50s per lap (somewhere like Spa, which is 57% full throttle by time and 70% full throttle by distance on a 110s race lap). At no point are we trying to recover energy to the ES from the ERS-K via ERS-H route while the driver is at 100% throttle.

A few stated assumptions: 50s duration, flywheel of 4Kg (the minimum in the rules), 100mm diameter (seems sensible enough for the turbo sizing alone), 40 pulses per second, 2000rpm rev change (of the MGU-H) during a pulse. Energy per pulse depends on the mass distribution on the MGU-H (mass in the centre has less inertia than the edge), but lets say that it's something like 180J because the majority of the mass is on the outer ring of the MGU-H, due to some clever designer putting weight out there.
In this case, we have 50s per lap * 40 events per second * 180J per event = only an additional unmetered 360KJ per lap. That's only 3s of full deployment and hardly worth the trouble. You could go for a larger RPM change, at cost of the ICE seeing more and more variable levels of manifold pressure.

If we adjust the shape of the flywheel (make it 200mm diameter rather than 100mm above) and leave everything else the same:
Then in this case we have 50s per lap * 40 events per second * 700J per event = 1.4MJ per lap. Quite worthwhile. Perhaps worth moving the flywheel disk out of the Vee of the engine so make it larger without being heavier, permitting more flywheel inertial storage without a weight penalty.

In terms of the effect on the ICE of doing all of this, it would see some variations in manifold pressure and in exhaust pressure due to the speeding up and slowing down of the MGU-H, compressor and turbine, but the variations would be so fast that it really wouldn't make the ICE feel different to drive than if the system wasn't in place.

In terms of mapping the ES deployment/charge, the big plus point of this system is to work around the 2MJ transfer limit from MGU-K to ES. You'd essentially have this system pulling power from the IC via the MGU-K, dumping it into the MGU-H and on to the ES whenever the driver isn't requesting 100% from the PU - it could be recovering any time on part throttle or zero throttle, as required. You would still use the ERS-K to ES direct route for the 2MJ permitted per lap of course, but the rest of the energy would be most effectively recovered using the flywheel method.

In terms of redeploying ES energy, there's a flow limit of 4MJ from the ES to the MGU-K using the direct route per lap, but again this could be worked around because there's nothing stopping the route from the ES to the K via the H in flywheel-routed "deployment" mode.

The 4MJ SoC ES limits don't appear to stop the car from actually receiving and deploying more than 4MJ per lap because of how that rule is worded.
It appears that the single biggest deployment event you can have within a lap is 4MJ (taking the ES from "full" to "empty"), but so long as you discharge and recharge you can cumulatively have more than 4 MJ entering and leaving the ES in a given lap. The ES to K *direct* route is limited to 4MJ out per lap, and 2MJ back to ES. There's no limit on ES to MGU-H transfers or rate, remember.

The route via the flywheel system suggested would raise those 2MJ and 4MJ limits significantly, potentially to the point where the MGU-K is deploying 120kW the whole time the PU is at 100% output.

Can I make a hypothetical scenario to explain this? (might not be that hypothetical, as it may correlate to Honda's first pass at the PU a few years ago):

If you were limited to (for example) 2MJ of energy into the ES per lap because you'd designed a system where only the MGU-K charged the ES, then you'd be limited to 2MJ of deployment on the next lap. You'd hear drivers complain a lot about a lack of deployment, because your 2MJ is only enough for 16.6s of deployment round a given lap. Does this sound familiar?

You would need to figure out how to add to the energy input into the ES, over the 2MJ you'd started off with.
You're potentially already at the 2MJ direct limit from the MGU-K from braking and from charging towards the end of straights, so you need to add whatever you could get from MGU-H using compounding. This is essentially limited by what the ICE's operating parameters are: the higher the amount of energy remaining in the exhaust gases, the worse the basic ICE performance was at driving the rear wheels directly, and the more fuel-hungry the ICE was.

If instead of this, you designed a system to pull energy from the K via the H into the ES, you have the ability to charge the ES at an abritrary amount per lap, and deploy 4MJ directly via the MGU-K on every lap as you please.

You can do that deployment wherever you want (start of straights, of course, but actually on quite a few circuits, most of the lap because you have 33s to play with).
You also have the ability to supplement this direct ERS-K drive route with energy taken from the ES to the MGU-K, then dumped in small pulses into the MGU-K using the "unmetered" ES->H->K route.
Your 120kW deployment is in fact made of pulses of power direct from the ES to the MGU-K, interleaved with pulses of power from the MGU-H to the MGU-K.
Your deployment per lap consists of 4MJ direct from ES to K plus the additional energy using the unmetered ES to MGU-H to MGU-K route, within the constraint that the ES State of Charge can't go above 4MJ or below 0MJ from their starting values.

Your deployment strategy now revolves around maximising the factors you control: ES SoC and MGU-H sizing, along with where on the lap to charge and where to deploy, using each system.
The more power the MGU-H has been designed to transfer, the physically bigger it will be
There's a compromise in terms of volume versus weight versus power rating.
You likely size the MGU-H to cope with the harshest duty-cycle scenarios on the calendar (Spa?).

Question: What would people prefer we call this technique if not "flywheel"?

I absolutely love where this topic is going. Thanks Craigy and godlameroso for kicking it off, but as this is the Honda PU thread and this discussion applies to all manufacturers (and teams for that matter) can I please ask one of the mods to move it to its own thread?
There's so much more to discuss.

[Craigy]

I absolutely love where this topic is going. Thanks Craigy and godlameroso for kicking it off, but as this is the Honda PU thread and this discussion applies to all manufacturers (and teams for that matter) can I please ask one of the mods to move it to its own thread?
There's so much more to discuss.

Thank you all for the kind words. I agree this should probably be in the main engine formula thread.

*reports own post*

[PlatinumZealot]

I don't understand the recent thoughts of going completely haywire with transfers from the "k" to the "h"; where is that energy coming from do you think? It is not unlimited in the sense that there's an over-abundance in the source; it's unlimited in the very literal sense: you are free to transfer whichever amount you can produce. You cannot produce an unlimited amount of energy.

I think flywheel is a bad term when applied to the "h"; the amount of potential energy recoverable is probably way less than that which can be tapped into by "braking" against force/ torque applied either by the turbine or by this newfound theory of switched motoring.

But remember that this "k"-driven switched motoring is driven by physical fuel so it is an accounting trick more than one relating to efficiency. Min-maxing if you will.

Mguk energy is harvested from the rear axle primarily during braking but less so durnig coasting at top speed. Of course some tracks give less harvesting than others. And in the same way, some tracks are less sensitive to ERS deployment... Such as monaco.

You missunderstand when we say flywheel energy. We are not talking about energy from the turbine wheel here. We sre talking about residual kinetic energy from using the MGUK to spin the MGUH in 20 to 40 Hz cycles.

The K drive motoring is not intentionally driven by burning fuel. That would be inefficient. It is primarily driven by braking. The only time fuel energy may be used to the K would be for like cornering events and high speed coasting. But this is just a benefficial side effect of using wasted energy from the driver controlling the car with the throttle in corners and high speed coasting.

EDIT:
AJI said it better than me!
On the end of straights where alot of energy is burnt for less laptime you can dump energy from the ICE into the energy store using the same mguh flywheel loophole for use at another part of the track where you need that acceleration to make a big improvement in laptime. The idea is not to burn fuel wantonly. But there are places like end of straights and in some corners where fuel is wasted anway.. So better to use it for "extra harvest"