Using intermediate components to exceed 4MJ per lap energy transfer between ES and MGUK

[aleks_ader]

@subcritacal71

I think that is already in use in sort of state of the art BMS management.

[Red Rock Mutley]

I just had a random thought.

Thinking out of the box is good

I think I follow your narrative. Can I walk through it and see what emerges. Hope you don't mind, I like using water analogies when talking about electricity, because most people can visualise flowing water better

... what if you could take a battery, virtually split it in two using the controller (say an ES-H and ES-K). This would be a dynamic allocation of the cells. For example 2MJ for ES-H and 2 MJ for ES-K (SOC).

That's no problem. In my ES there would be 4 buckets of water, and I can virtually split them by attaching post-it notes to each; I would label 2 buckets "H", and the other 2 "K"

Charge the ES-K from the K (2 MJ limit)

I can move one of the "K" buckets under the incoming pipe when harvesting from the MGU-K

and then under certain conditions, re-allocate the battery cells instead of transferring energy, either partially or fully (a sort of ES <> ES transfer but by way of re-allocation). I don't see a limit on ES to ES transfers.

That's ok too. There are no limits on moving the buckets around in the ES, and I can change the labels to read whatever I want

This would minimize losses and time incurred by cycling the H (for the K -> H -> ES route), for example. This would ensure that on tracks where the K can send its full 2 MJ of energy from the K straight to the battery, and then reused wherever they desire (either the H or the K). It could also be used the other way, H charges its 2MJ to ES-H, reallocates it to the ES-K, do this twice a lap, therefore 4MJ delivered to the K with minimal losses.

I need a bit of help on this bit. I think I understand what you mean; you're referring back to the original topic of transferring energy via intermediate components

If I understand you correctly, you're scheme uses the pipe between the ES and the MGU-K to transfer water to and from the ES in the conventional way. Unfortunately that's the pipe where the bottle neck is. It's limited to a maximum of 4 buckets of water out of the ES per lap, and maximum of 2 buckets in

The point with the MGU-H is it's connected to both the ES and the MGU-K by separate pipes, and crucially those separate pipes have no limits; as much water can through those pipes as you like. So, to exceed the 4/2 buckets of water per lap limits, there needs to be a way of connecting the MGU-H pipes together and then you can flow as much water as you like between the ES and the MGU-K

[subcritical71]

The flow diagram would look something like this

http://s50.photobucket.com/user/3874619 ... u.png.html

I just had a random thought. I don't know how battery construction works or how two virtual batteries might work. But, what if you could take a battery, virtually split it in two using the controller (say an ES-H and ES-K). This would be a dynamic allocation of the cells. For example 2MJ for ES-H and 2 MJ for ES-K (SOC). Charge the ES-K from the K (2 MJ limit) and then under certain conditions, re-allocate the battery cells instead of transferring energy, either partially or fully (a sort of ES <> ES transfer but by way of re-allocation). I don't see a limit on ES to ES transfers. This would minimize losses and time incurred by cycling the H (for the K -> H -> ES route), for example. This would ensure that on tracks where the K can send its full 2 MJ of energy from the K straight to the battery, and then reused wherever they desire (either the H or the K). It could also be used the other way, H charges its 2MJ to ES-H, reallocates it to the ES-K, do this twice a lap, therefore 4MJ delivered to the K with minimal losses.

What do you think? Has this been covered already? I'm sure there are other strategies that could be used that I haven't thought of.

This wouldn't make much sense.
The ES is made up of a multitude of individual cells anyway. With the voltage the MG-Units working in F1 it must be atleast several hundred, because every individual Cell will only give you a voltage of about 3.7 Volt. So you have to manage the whole Pack anyway and technically it doesn't make much difference for the energy managment software if you have one big pile of cells or two smaller piles.
One intresting aspect is the thermal managment though. If one pile alone is able to deliver the necessary voltage the MG-units need, you gain time to cool the other down which means you can get away with a smaller and lighter cooling system for the ES.

After I wrote the above I was having a look at a very interesting paper regarding dynamic allocation of battery cells. (http://citeseerx.ist.psu.edu/viewdoc/do ... 1&type=pdf) and came to the same thought as you did (bolded and italic'd above). If you are not allocating all of the cells at once then you could 'cool' the unused ones.

[subcritical71]

and then under certain conditions, re-allocate the battery cells instead of transferring energy, either partially or fully (a sort of ES <> ES transfer but by way of re-allocation). I don't see a limit on ES to ES transfers.

That's ok too. There are no limits on moving the buckets around in the ES, and I can change the labels to read whatever I want

And not all need to be allocated. There could be spare cells in case of failures and over-temperature conditions. I was using two cells (buckets) as an example but in real life it would be hundreds, maybe thousands of buckets). The paper I read after writing the above talks about dynamically allocating cells to deliver the optimized voltage for whatever operation you are performing at the time. For example, charging and discharging rates based on the consumer or producer. They used 5 and 10 second monitoring, but I could imagine that could be increased for finer control.

This would minimize losses and time incurred by cycling the H (for the K -> H -> ES route), for example. This would ensure that on tracks where the K can send its full 2 MJ of energy from the K straight to the battery, and then reused wherever they desire (either the H or the K). It could also be used the other way, H charges its 2MJ to ES-H, reallocates it to the ES-K, do this twice a lap, therefore 4MJ delivered to the K with minimal losses.

I need a bit of help on this bit. I think I understand what you mean; you're referring back to the original topic of transferring energy via intermediate components

If I understand you correctly, you're scheme uses the pipe between the ES and the MGU-K to transfer water to and from the ES in the conventional way. Unfortunately that's the pipe where the bottle neck is. It's limited to a maximum of 4 buckets of water out of the ES per lap, and maximum of 2 buckets in

Maybe a better way of explaining it is from the MGU-H side. You could use the MGU-H to charge the ES-H battery until fully charged. We'll assume the ES-K was fully charged. After the MGU-K deploys those 2 MJ from the ES-K the ES-H now becomes the ES-K, therefore there are 2 MJ more the MGU-K can use. While the MGU-K is consuming the 2 MJ that originally started in the ES-H, the MGU-H is now charging the empty ES-H (which used to be the ES-K)... wow, there is no way your going to follow what is going on in this head of mine! This switching and optimization of the number of cells 'connected' can put the respective batteries at their peak efficiencies(?)

The point with the MGU-H is it's connected to both the ES and the MGU-K by separate pipes, and crucially those separate pipes have no limits; as much water can through those pipes as you like. So, to exceed the 4/2 buckets of water per lap limits, there needs to be a way of connecting the MGU-H pipes together and then you can flow as much water as you like between the ES and the MGU-K

This is already solved with the switching of the MGU-H between harvest and generating like Honda use, isn't it?

[henry]

It’s interesting that cooling etc. has come up. I thought that this might be a reason for Ferrari to implement a dual battery scheme from the outset.

I think there might be a benefit when looking at the “extra harvest” scenario.

When in extra harvest mode the duty cycle means that while the MGU-K is spinning up the MGU-H with 120kW the Turbine is also spinning it up with, say, 60kW. So if the cycle is symmetrical the H will harvest at 240kW. That’s 180kW to recover the kinetic energy plus 60kW because the H will also be recovering the turbine energy stream. And the ES will be charged at 240kW. The extra charge will only be at an average 60kW I.e. the K for half the cycle.

That’s a high price to pay. They could, of course, change the duty cycle and reduce the K harvest time and extend the H time. But of course this reduces the benefit cells 60kW.

If you have an ES scheme that allows better management of the stresses associated with this mechanism then you can probably use more advantageous duty cycles and harvest more energy, and then on the next straight that will look like a power boost.

[Red Rock Mutley]

You could use the MGU-H to charge the ES-H battery until fully charged

It’s interesting that cooling etc. has come up. I thought that this might be a reason for Ferrari to implement a dual battery scheme from the outset.

That's a good point, batteries are fundamentally a chemical process, they deliver more energy when warm and conversely, accept more energy when cold (when the internal resistance is lower). There's a mismatch in efficiency between charge and discharge. I can see there's a potential gain by segmenting the battery, allocating one half to charge and the other to discharge. Actively cool the charge segment, while letting the discharge segment heat up. Then swap the segments

[PlatinumZealot]

Remember guys. Reverse Extra Harvesting.

After 4MJ is sent to the MGUK.. You can still continue discharging the battery through the MGUH to the MGUK using the oposite of the extra harvest technique.

[Red Rock Mutley]

Do we think all the engine manufacturers are using extra harvest & extra deployment?

We know from the Christmas 2017 magazine interview that Honda are at least performing the extra harvest. Skip forward a couple of design and qualification cycles and that's pretty much around the time the Ferrari engine showed signs of extra deployment. Presumably Renault now know the trick via their association with McLaren. That leaves Mercedes. I can't remember, was it them who estimated Ferrari needed 4.5MJ deployment to match their speed profile

While I’m here, I need to credit muramasa. It took ages of scrolling back through the Honda engine thread to find the source so I’ll post the link here viewtopic.php?f=4&t=18874&start=13440#p749725

[henry]

Do we think all the engine manufacturers are using extra harvest & extra deployment?

We know from the Christmas 2017 magazine interview that Honda are at least performing the extra harvest. Skip forward a couple of design and qualification cycles and that's pretty much around the time the Ferrari engine showed signs of extra deployment. Presumably Renault now know the trick via their association with McLaren. That leaves Mercedes. I can't remember, was it them who estimated Ferrari needed 4.5MJ deployment to match their speed profile

While I’m here, I need to credit muramasa. It took ages of scrolling back through the Honda engine thread to find the source so I’ll post the link here viewtopic.php?f=4&t=18874&start=13440#p749725

I can’t know whether all the manufacturers use the “extra” routes.

From my reasoning I have posted earlier I think the extra routes place high duty cycles on the H and the ES. And the harvest and deploy rates are lower than the direct, 120kW rate. So their use might be conditioned by reliability, cooling or other gating factors. So perhaps used in qualifying and attack/defend situations.

Thanks for reminding me of @Muramasa’s post. It’s an enormous pity that we no longer get his contributions.

I don’t know if you are aware of an earlier post on page 845 of that thread which has a full translation of the magazine article.

[Red Rock Mutley]

I don’t know if you are aware of an earlier post on page 845 of that thread which has a full translation of the magazine article.

Those MGU deployment graphs make fascinating reading, particularly the MGU-K trace tending towards the throttle trace. Don't know why it hadn't dawned on me before, but I've always read "full deployment" to read "using all available battery power" as opposed to "operating the MGU-K at full power all the time"

[Dr. Acula]

I also have to thank Red Rock Mutley for posting a link to the respective page in the Honda engine thread. I tried to find it my self a few days ago, but simply gave up after about 20 pages...

But one thing to consider, this indirect way of harvest energy for the ES is very track specific. On tracks like the Yas Marina Circuit it's quite effective. On tracks like Silverstone or Monza on the other hand, where you will have problems to even reach the 2MJ limit, it's more or less useless.

[Red Rock Mutley]

But one thing to consider, this indirect way of harvest energy for the ES is very track specific.

Yes, presumably it is track specific, although I don't know with Silverstone, there's a lot of part-throttle time. Looking at the published figures it's conceivable they exceed 2MJ MGU-K harvest

[With the proviso of the exact reference lap time used by each source is unknown, and as such there is a high level of uncertainty associated with their figures]

Using this year's pole time as a reference figure, 85.892s

Brembo give a brake time of 11.5s per lap. So that's 11.5s x 120W = 1.38 MJ under braking

Mercedes give a lap time on full-throttle of 58% = 49.82s. So that leaves a part-throttle time of 24.58s

The Honda article indicates approximately 60W MGU-K harvest under part acceleration: 24.58s x 60W = 1.47 MJ

Putting those together gives an upper estimate of 2.85 MJ per lap from the MGU-K alone (potential)

So, taking things at face value, there appears to be a potential of using Extra Harvest at Silverstone, albeit with a high level of uncertainty

-------
Turning to the deployment side, MGU-K deployment very approximately equals the full-throttle time, 49.82s x 120W = 5.98 MJ

MGU-H harvest time also approximates the full-throttle time, 49.82s x 60W = 2.99 MJ

Subtracting the MGU-H energy leaves 2.99 MJ per lap to come from the MGU-K to enable full deployment

So using those very rough, back of an envelope type calculations, we are in the ballpark of being able to achieve full deployment using Extra Harvest. Obviously that's using a qualifying lap time, while full deployment is more relevant to race laps

[Red Rock Mutley]

I need to correct something I said earlier. It's clear from the Honda article that they don't harvest from the MGU-K at the end of the straight. They motor the MGU-K at full power until deployment runs out, then continue on ICE alone, and it's that latter phase that is associated with the rain light flashing - the lack of MGU-K assistance while WOT

[wuzak]

Brembo give a brake time of 11.5s per lap. So that's 11.5s x 120W = 1.38 MJ under braking

How much of that 11s is heavy enough to allow 120kW harvest from the rear wheels?

[Red Rock Mutley]

That's a good question. 120KW seems very light braking. I can't see anywhere on the Honda graphs where MGU-K power dips below maximum under braking, except that is for momentary interruptions under downshifts

Brembo imply sub 3g is the minimum deceleration, with a total dissipation of 103 KWh (av. 10.4 MJ per lap by my back of the envelope maths)

With every lap, brakes are used 8 times, but in 3 of the bends drivers rely on their brakes for less than 7 tenths of a second.

All told, the braking system is used for 11 and a half seconds in each lap, amounting to 14% of the entire duration of the race.
Even maximum deceleration is affected by the reduced need to slam on the brakes: the average value per lap is 3.6 g due to turn 10 and the curve right after, since deceleration does not even get to 3 g for either of these bends.

All this means that only a moderate amount of energy is dissipated in braking compared to other tracks: 103 kWh, half as much as the Budapest track and ​​Singapore.

www.brembo.com/en/company/news/formula ... bo-brakes