Ferrari Power Unit Hardware & Software

[turbof1]

Enough speculation about legality in the past. It's a hardware topic; unless someone can specifically point out the illegality of a part, this is not having a home here.

Removed a whole lot of posts.

[sosic2121]

Assuming there is truth to Ferrari's battery exploit, it's inherently difficult to prove. Obviously, there is a sensor there by the FIA that all teams use that measures the amount of energy that passes through the sensor to make sure no more than 4MJ is used per lap. If there is truth to what the rumors say, there's a secondary circuit connecting to the battery that does not pass through the FIA's sensor.

If I understand correctly what you are saying you suggest that ferrari is using more than 4MJ from battery. This is not illegal .
State of charge of battery is limited to 4MJ, but you can charge and empty battery as many time you wish during one lap.
What is limited to 4MJ/lap is ES>MGU-K.
However, you energy can be sent energy via H.
ES>MGU-H>MGU-K this is also legal.
IF Ferrari is cheating, it's something else.

[turbof1]

Assuming there is truth to Ferrari's battery exploit, it's inherently difficult to prove. Obviously, there is a sensor there by the FIA that all teams use that measures the amount of energy that passes through the sensor to make sure no more than 4MJ is used per lap. If there is truth to what the rumors say, there's a secondary circuit connecting to the battery that does not pass through the FIA's sensor.

If I understand correctly what you are saying you suggest that ferrari is using more than 4MJ from battery. This is not illegal .
State of charge of battery is limited to 4MJ, but you can charge and empty battery as many time you wish during one lap.
What is limited to 4MJ/lap is ES>MGU-K.
However, you energy can be sent energy via H.
ES>MGU-H>MGU-K this is also legal.
IF Ferrari is cheating, it's something else.

The question that begs here: can one circumvent the 4MJ pull from the battery by getting the energy flow from the ES to the MGU-H to the MGU-K? Energy exchange between the battery and the MGU-H on one hand is unlimited, and energy exchange between MGU-H and MGU-K is also unlimited on the other hand. Only the energy exchange between the MGU-K and the ES is capped at 4MJ.

Even if not allowed: the output of the MGU-K is 120kW. That is 0,12MJ/s. 4MJ will give you 33.33s of 120kW power, but you can also have 40s/50s/60s of 120 kW power, depending how much energy you can generate AND consume a the same time, from the mgu-h suplemented by the 4MJ from the battery. The energy generated does not have to be stored on ES, but can directly feed into the MGU-K deploy mode.

Usually, you are storing the MGU-K harvesting mode on the ES. Naturally, because you are braking so in no need for 'motion' energy. However, the MGU-H is generating the power when on the throttle, so you are going to spent that energy immediately instead of storing it on the ES because then you want enough 'motion' energy to sustain 120kW as long as possible.

[henry]

Assuming there is truth to Ferrari's battery exploit, it's inherently difficult to prove. Obviously, there is a sensor there by the FIA that all teams use that measures the amount of energy that passes through the sensor to make sure no more than 4MJ is used per lap. If there is truth to what the rumors say, there's a secondary circuit connecting to the battery that does not pass through the FIA's sensor.

If I understand correctly what you are saying you suggest that ferrari is using more than 4MJ from battery. This is not illegal .
State of charge of battery is limited to 4MJ, but you can charge and empty battery as many time you wish during one lap.
What is limited to 4MJ/lap is ES>MGU-K.
However, you energy can be sent energy via H.
ES>MGU-H>MGU-K this is also legal.
IF Ferrari is cheating, it's something else.

The question that begs here: can one circumvent the 4MJ pull from the battery by getting the energy flow from the ES to the MGU-H to the MGU-K? Energy exchange between the battery and the MGU-H on one hand is unlimited, and energy exchange between MGU-H and MGU-K is also unlimited on the other hand. Only the energy exchange between the MGU-K and the ES is capped at 4MJ.

Even if not allowed: the output of the MGU-K is 120kW. That is 0,12MJ/s. 4MJ will give you 33.33s of 120kW power, but you can also have 40s/50s/60s of 120 kW power, depending how much energy you can generate AND consume a the same time, from the mgu-h suplemented by the 4MJ from the battery. The energy generated does not have to be stored on ES, but can directly feed into the MGU-K deploy mode.

Usually, you are storing the MGU-K harvesting mode on the ES. Naturally, because you are braking so in no need for 'motion' energy. However, the MGU-H is generating the power when on the throttle, so you are going to spent that energy immediately instead of storing it on the ES because then you want enough 'motion' energy to sustain 120kW as long as possible.

To be pedantic the output of the MGU-K is specified as 120 kW but it is controlled/monitored by the output of the control unit that drives it. The FIA predict the conversion efficiency at 95% so the output of the CU-K is allowed to be 126.3 kW. With the prescribed 4 MJ from the ES to the MGU-K the time is 31.66 seconds.

The path ES> H> K would be less efficient than direct ES >K, so if you could find a way to send more energy directly you would. It is in relation to the latter that I think the FIA investigated. The implementation of Ferrari’s ES apparently made this difficult.

The measurement of the energy flows is complicated because there are only two sensors and three sources/sinks for the energy. I don’t know how they resolve this issue. I’m guessing it involves some software and some assumptions about how you can decide where energy leaving one source is destined. Maybe the switching between batteries in Ferrari’s ES brought the assumptions, and their implementation in software, into question.

I wonder how the FIA monitor and approve such software code? I assume each manufacturer writes their own. Seems at odds with the approach to the monitoring of fuel with a “fit this and don’t moan” approach.

[dans79]

I noticed this earlier today!

http://www.grandprix.com/news/fia-to-ke ... anada.html

But race director Charlie Whiting said more monitoring in Canada may now be necessary.

"I'm not saying we'll definitely do it, because I assume we will come to the same conclusion," he said.

"But Ferrari will make changes to the software in Canada. I'm not an expert, but we're trying to understand the difference of the two parts of the battery.

[henry]

I noticed this earlier today!

http://www.grandprix.com/news/fia-to-ke ... anada.html

But race director Charlie Whiting said more monitoring in Canada may now be necessary.

"I'm not saying we'll definitely do it, because I assume we will come to the same conclusion," he said.

"But Ferrari will make changes to the software in Canada. I'm not an expert, but we're trying to understand the difference of the two parts of the battery.

I found this interesting

There is one battery, but the system perceives it as two," Whiting adding. "I don't think I'm giving anything away."

So not two physical batteries at all. Or maybe it’s just that there is confusion about what “battery” means. I think of it as a fixed collection of connected cells. Maybe Ferrari’s take is more virtual selecting the collection on the fly. I can see how that might be difficult to understand.

[dans79]

I found this interesting

There is one battery, but the system perceives it as two," Whiting adding. "I don't think I'm giving anything away."

I saw that as well, but given that it's a Charlie quote I have not confidence that he has any idea what he is talking about.

[GrandAxe]

I noticed this earlier today!

http://www.grandprix.com/news/fia-to-ke ... anada.html

But race director Charlie Whiting said more monitoring in Canada may now be necessary.

"I'm not saying we'll definitely do it, because I assume we will come to the same conclusion," he said.

"But Ferrari will make changes to the software in Canada. I'm not an expert, but we're trying to understand the difference of the two parts of the battery.

I found this interesting

There is one battery, but the system perceives it as two," Whiting adding. "I don't think I'm giving anything away."

So not two physical batteries at all. Or maybe it’s just that there is confusion about what “battery” means. I think of it as a fixed collection of connected cells. Maybe Ferrari’s take is more virtual selecting the collection on the fly. I can see how that might be difficult to understand.

Interesting indeed.

If the story that the Ferrari system perceives the single battery as two has come from an expert, then the straight forward conclusion is that there are at least two modes by which energy can be delivered from the battery.

This raises further questions:

• Can one mode mask the other?

• What governs switching between modes?

• Do both modes run consecutively?

• Is their power output constant or variable?

If switching between modes is through software whose internals are a tangle of something like Bayesian statistics or some other AI method, then resolution of the problem might be well beyond the FIA's capacity. Instead we will see other teams quickly follow suit and a Pandora's box opened, because there are many other areas of the car where such can be applied (rear tyre heating, "legal" ABS etc). 2019 might be very interesting.

[sosic2121]

I think we are missing something here.

AFAIK Honda showed extra harvesting is possible, and then of course extra deployed is also possible. (someone even suggested MGU-H with dual stator windings).

Also i think there is enough energy in turbo compound mode to power MGU-K 100% of the time during the lap. (ES+MGU-H >MGU-K)

So what are they doing then? I don't think they cheat 120kW rule.
Extra power has to come from ICE(20kw is mentioned).
I think during Q Ferrari(and others?) is using electric charger mode. "Small" power gains, huge energy drain.

I think that on beginning of straightsthey go to max power mode, and at the end of straight they go to max harvest mode. No turbocompound mode during Q.
This is possible in Q since they can have little more fuel and can finish the lap with flat ES.

So this year Ferrari made big gains in Q, but made no gains in race trim.

During breaking can the throttle body be open? Is there a rule forbidding it?

[dans79]

So what are they doing then? I don't think they cheat 120kW rule.

That depends on how the FIA is regulating the 120kW limit. They are most likely just monitoring the the power sent to the MGU-K. Depending on the type of sensor the FIA used it could easily be intentionally or unintentionally fooled.

[roon]

AFAIK Honda showed extra harvesting is possible, and then of course extra deployed is also possible. (someone even suggested MGU-H with dual stator windings).

...

During breaking can the throttle body be open? Is there a rule forbidding it?

Might have been me. I brought it up in the Honda thread. viewtopic.php?p=734068#p734068 The H is an unlimited pathway between the ES and the K; only catch is there's a conversion loss. Electrical>kinetic>electrical. The Japanese article which was translated by Hino in the Honda thread suggested Honda switched back and forth between generating and motoring in order to treat the H as a secondary feed cable supplying the K. I was thinking it might be more elegant to have a dual stator to avoid switching and simply design half the H as an electrical-kinetic-electrical link.

No rules around throttles or application thereof. If throttles are even used.

dans79 brings up a good point. It would be good to know how the sensor operates.

[63l8qrrfy6]

It's an IVT Isabellenhutte sensor

[Big Tea]

It's an IVT Isabellenhutte sensor

So were they suspecting some trickery with PWM?

[63l8qrrfy6]

It's an IVT Isabellenhutte sensor

So were they suspecting some trickery with PWM?

That's the standard FIA sensor everyone has to use, I don't know what the additional Ferrari hardware consists of.

[sosic2121]

It's an IVT Isabellenhutte sensor

I assume MGU-K is connected with 2 leads.
So sensor messures voltage drop between MGU-K those 2 leads and current in one of them.
Could engine mass used to bypass the current measuring part of the sensor?
That lead could be grounded to the engine somewhere in CE and also in MGU-K.
Ratio of electrical resistance of leads and of engine would determine how much current would go through lead and how much trough engine.