BMW exec foresees the demise of the DCT from the M3

[Eau Rouge]

Originally Posted by captainaudio

What I was questioning was how that .5 seconds in faster shift times effects acceleration. In other words lets say that we could program a DCT to shift in either 50ms or 300ms and did a 0-100 run in both modes. How much faster would the time be with the faster shifts?

CA

Indeed. Faster by a meaningful margin IMHO.

The elapsed time would likely decrease by some 0.7s on three shifts which would be nice to have against an MT driver who only has to shift twice to eclipse 100mph.

[RandyB]

The big question in my mind has always been this - can the faster shifts overcome the added weight and increased parasitic losses compared to the 6MT? My guess is that it probably can. But by a meaningful margin? I'm not so sure. No question it is easier to drive fast and do it consistently.

I still prefer more driver involvement over less, and a tenth or two either way will not outweigh that for me. I also feel more comfortable owning a 6MT past the warranty period, but that wouldn't have stopped me from getting a M-DCT had I wanted one.

No doubt M3 owners are a passionate bunch, as is evidenced by how divided we are over so many details.

CF vs. Moonroof
18s vs 19s
M-DCT vs 6MT
EDC vs Non-EDC
Cast vs Forged

[JIO]

Originally Posted by _Jono_

AMGs trannys suck

I tested the C63 AMG, where the salesguy said it shifted at 100 ms. I was so disapointed in the shifts as to how slow thery were, even on the fasted "sport", manual mode. When I got back into my DCT, I was in heaven. Evvery torque converter automatic has to disrupt power to shift...that's not the case with DCT.

Notice that you no longer can get an automatic on the Porsche 911...only PDK.

DCT will get better and better, and smoother for people who still choose to drive in auto mode. For me, give me S4, S5, or S6...and the smile is worth an S7.

12 gear automatic? Gee, let's catch up to bicycles. They have like 30 gears...of which most bikers use about 5....he he.

[captainaudio]

Originally Posted by Eau Rouge

Indeed. Faster by a meaningful margin IMHO.

The elapsed time would likely decrease by some 0.7s on three shifts which would be nice to have against an MT driver who only has to shift twice to eclipse 100mph.

.7 seconds faster shift time does not equate to a .7 seconds faster ET.

CA

[Eau Rouge]

Originally Posted by captainaudio

.7 seconds faster shift time does not equate to a .7 seconds faster ET.

CA

I'd like to see the scenario you described be realized. Will faster shifts reduce ET at all? If not, then this whole speculative exercise is a moot point from a performance standpoint, no?

[double_j]

Originally Posted by Eau Rouge

I'd like to see the scenario you described be realized. Will faster shifts reduce ET at all? If not, then this whole speculative exercise is a moot point from a performance standpoint, no?

Can't you just look at an average difference between M3 DCT and 6MT trap or 0-60 times or would that be impacted too much by the different gearing, etc? I know that other factors come into play for each individual test but shouldn't it give an baseline indication about the approximate order of magnitude?

If a scan of DCT 0-60 results as a whole tend to be about a tenth or so faster to 60 than 6MT over two shifts then maybe that's a decent starting point for extrapolation.

[captainaudio]

Originally Posted by Eau Rouge

I'd like to see the scenario you described be realized. Will faster shifts reduce ET at all? If not, then this whole speculative exercise is a moot point from a performance standpoint, no?

It would seen logical that faster shift times would reduce the ET as the car stops accelerating while it is shifting and in fact begins to slow down as power is not being applied to the drive wheels.

I am confused as to whether a 200 millisecond difference in shift time will equate to a 200 miilisecond better ET (assuming one gear shift).

CA

[swamp2]

Originally Posted by captainaudio

What I was questioning was how that .5 seconds in faster shift times effects acceleration. In other words lets say that we could program a DCT to shift in either 50ms or 300ms and did a 0-100 run in both modes. How much faster would the time be with the faster shifts?

CA

You guys are both right. For the time to get to speed X you save basically the sum of the shift times saved. However, distance at a given time is more important in a race, curvy or straight. Here are some simulations done with CarTest, a physics based car simulator. It is ideal for answering questions like this because it performs perfect apples vs. apples tests completely eliminating driver, car and environmental factors which can override small mods.

This comparison is for the exact case you mentioned. A 7 speed M3 M-DCT shifting in 30 ms vs. one shifting in 400 ms (chose that to be slightly more extreme than 300 ms). 100 mph comes at about 10 seconds. At exactly 10 seconds the real DCT will have 3 car lengths on the other car. The times differences to 100 is dramatic - 1.1 s (3 shifts x .35 s works just fine). Do note I have maintained the number of gears and exact gear ratios for both cars - I did not substitute the actual gear ratios from the 6MT.

For a 1/4 mi drag race ET the shift times give you a little over 1 shift time difference of benefit. In term of car lengths it is about 4 1/4. Trap speed benefits by about 3 mph.

For the M3, the DCT is roughly equivalent to 20 hp! It is real. Totally real.

[captainaudio]

I had a feeling that if anyone had an answer to my question it would be Swamp2.

Thank you

It would seem to me that the differnece in Et would be slightly more that the difference in shift times as the car will be slowing down while the shift is being performed and there is no power being applied to the wheels. Obviously the car is not going to slow down very much in 200 milliseconds but it will slow down,

CA

[AlphaBeta]

Leading transmission supplier ZF has hinted that nine-, 10-, 11- and 12-speed transmissions are a possibility in the future, but Kretschmer says BMW is not rushing to be the first to have a nine-speed auto.

That is hilarious & THANK GOD!

[bruce.augenstein@comcast.]

Originally Posted by captainaudio

It would seen logical that faster shift times would reduce the ET as the car stops accelerating while it is shifting and in fact begins to slow down as power is not being applied to the drive wheels.

I am confused as to whether a 200 millisecond difference in shift time will equate to a 200 miilisecond better ET (assuming one gear shift).

CA

As you say, quicker shifts lead to better acceleration - but it's complicated.

First of all, benefit varies by gear - but you will not gain a reduction in ET equal to the difference in shift times, on average - unless you are racing for very long distances. The first shift is the most important, ET and speed-wise, because you have quite a lot of time for the benefit of the quicker shift to actually be realized. By contrast, the three-four shift is relatively unimportant, because you'll be over the line before meaningful results can occur.

Let's take a SWAG at this. Work with me here, since I'm trying to figure it out as I go.

In a quarter mile, most cars will need three shifts before they hit the traps. Let's assume the three shifts take one second in total, and your car blazes the traps in 12.60 seconds at 112 mph.

Now let's say you take a caffeine hit in the pits, and run again, except this time you're doing the dab-pull, dab-push, dab-pull routine, and the three shifts take you a total of six-tenths of a second. Two tenths per shift is a reasonable number if you're just dabbing the clutch, and not going to the floor with it.

So, you've gained four tenths of a second in acceleration overall - meaning you're actually accelerating for about a three point two percent (3.2%) longer duration on that run.

This is pretty much the same as saying you've got a three point two percent better power-to-weight, and since we know that ET and speed vary by the cube root of the power-to-weight delta, we can say that, based on time-of-shift alone, the car will gain about one percent in trap speed, and the ET will drop by one percent.

So now you're at 12.48, ET-wise, at just over 113 mph - based only on time delivering power.

There's more, though.

Speeding up the shift means that the engine will have less time to lose rpm between shifts, so when the clutch (from whatever gearbox) re-engages, the engine will be at a higher rpm.

Let's say second gear demands 5000 rpm after a one-two shift at 8400, but since the engine only loses rpm at a rate of 4000 rpm per second with closed throttle (definitely a guess - feel free), a .33 shift duration means the engine will lose around 1300 rpm, and will be at about 7100 rpm when the clutch hits.

Result? A 2100 rpm torque spike due to rotational inertia will be thrown at the transmission and drive train, because all they need at that speed is 5000 rpm. This is why you "get rubber" on shifts.

Now let's speed the shifts up to two-tenths of a second. Second gear still only needs 5000 rpm at the shift point, but now, after a one-two shift at 8400, the engine only loses 800 rpm, and is at 7600 rpm after the clutch hits.

Result? We get a 2600 rpm torque spike, and the car jumps around 24% harder due to the transfer of inertia energy, compared to the 2100 rpm spike when we were being lazy.

OK, what if we powershift?

Now, we're at the 8400 rpm cutoff when the clutch hits, so we get a 3400 rpm rotational inertia torque spike. Car jumps even harder after the shift.

How does this translate to, say, quarter mile ET and speed numbers?

I have no way to precisely quantify this, not knowing the rotational inertia involved with any given vehicle, but the drag racing rule of thumb is that one might expect around three tenths, with a corresponding increase in trap speed.

Potential? 12.30 @ 114.7 mph, on a 12.60 @ 112 "standard'.

Just speeding up the shift from .33 seconds to .20 seconds, without powershifting, might get you a low 12.40, at just below 114, accounting for the punchier rotational inertia spike from the quicker shift.

The numbers are made up, but probably not very far off. Note that I've not accounted for the loss of velocity during shifts, nor made any speed allowance for reduction in acceleration time due to an improved ET.

Did I mention it gets complicated?

Bruce

Edit: Swamp, just saw your reply. If you have the time and inclination, would you throw in .33 second shift times against .20 seconds?

[swamp2]

Originally Posted by captainaudio

I had a feeling that if anyone had an answer to my question it would be Swamp2.

Thank you

It would seem to me that the differnece in Et would be slightly more that the difference in shift times as the car will be slowing down while the shift is being performed and there is no power being applied to the wheels. Obviously the car is not going to slow down very much in 200 milliseconds but it will slow down,

CA

Sure.

You might need a bit of basic calculus to help answer your question. When you look at a speed vs. time curve the MT basically flat lines during the shift (technically it is worse than flat lining, as you mentioned, it will actually have a negative slope, but in a vacuum and with no parasitic rolling losses it flat lines and a flat line is a close enough approximation to make the point here ). A flat line in the speed curve simply adds up the shift time benefits sequentially for time to a given speed. However a flat line in the speed curve still gives you a distance gain during that time, the distance vs. time curve thus does not flat line it simply maintains a constant slope! During that same interval the DCTs distance vs. time curve will get only slightly ahead because its slope will keep getting steeper as it is basically constantly accelerating (this is so if we call the shift instant, which is again a totally fine approximation to understand the essence). The differences are thus a muted in distance vs. time due to this.

You can see this in the real world as well. When watching such a race from the grandstand the DCT car will basically appear to constantly and steadily moving ahead (see the curve above I just posted), however when filmed from the DCT car (both a moving and accelerating reference frame) looking at the MT car you can clearly see the MT cars shifts very clearly and the abrupt change in relative position is also very apparent from this perspective.

Hope that is clear and helps.

[swamp2]

Originally Posted by bruce.augenstein@comcast.

Edit: Swamp, just saw your reply. If you have the time and inclination, would you throw in .33 second shift times against .20 seconds?

It is tough to get the engage time and shift times to match what you see exactly on the acceleration curves. There is a bit of judgement as well as to when the shift ends. When it begins is pretty obvious/precise but the end is more "fuzzy". Anyway between 0.20 and 0.33 seconds I get the following:

12.63 vs 12.79: .16s difference
112.5 vs. 111.2: 1.3 mph difference

That makes your estimates right in the ball park but a much better estimate on trap than time.

Unfortunately there is no control over powershifting vs. not in CarTest.

[bruce.augenstein@comcast.]

Originally Posted by swamp2

It is tough to get the engage time and shift times to match what you see exactly on the acceleration curves. There is a bit of judgement as well as to when the shift ends. When it begins is pretty obvious/precise but the end is more "fuzzy". Anyway between 0.20 and 0.33 seconds I get the following:

12.63 vs 12.79: .16s difference
112.5 vs. 111.2: 1.3 mph difference

That makes your estimates right in the ball park but a much better estimate on trap than time.

Unfortunately there is no control over powershifting vs. not in CarTest.

Close enough. Thanks.

[e92_m3]

Originally Posted by GOLFFRR

interesting.

+1! Very interesting read indeed!

[swamp2]

Originally Posted by bruce.augenstein@comcast.

So, you've gained four tenths of a second in acceleration overall - meaning you're actually accelerating for about a three point two percent (3.2%) longer duration on that run.

This is pretty much the same as saying you've got a three point two percent better power-to-weight, and since we know that ET and speed vary by the cube root of the power-to-weight delta, we can say that, based on time-of-shift alone, the car will gain about one percent in trap speed, and the ET will drop by one percent.

Although this might be very roughly true for typical ICE cars running in the 1/4 mi, the general principal of equivalence better a longer time spend accelerating and a correspondingly higher power to weight ratio is far from a general truth for accelerating bodies.

[bruce.augenstein@comcast.]

Originally Posted by swamp2

Although this might be very roughly true for typical ICE cars running in the 1/4 mi, the general principal of equivalence better a longer time spend accelerating and a correspondingly higher power to weight ratio is far from a general truth for accelerating bodies.

Yes.

The only reason this SWAG is even in the ballpark is because the shifts are spaced...

Bruce

[JCtx]

Originally Posted by JIO

Notice that you no longer can get an automatic on the Porsche 911...only PDK.

PDK IS an automatic (shifts by itself); just a different kind .

[Xelloss]

I do not see the DCT ever being replaced by a slushbox in most high performance vehicles. Even with the major advances in transmission technology, and the near parity the slushbox has attained, the negative connotation it brings will never be attractive to the enthusiast. I believe that this article is to highlight that from a commercial perspective, the DCT will not be able to supplant the slushbox in widespread application as it originally intended. With Ferrari dropping the traditional manual from its current and future lineup and other manufacturers adopting similar practices, I believe the DCT will be here for the foreseeable future till a better transmission technology in the general principles of manual actuation/control will be developed.


Not to mention BMW M decides what they use for their M cars as this executive is gladly not from M division. I trust BMW M to maintain that sporty characteristic and identity in their M cars (M "SAV's" are excluded and are not "cars"). If they dare use a slushbox in any M car, consider my loyalty to BMW M and BMW as a whole null. It's already trying with these new turbo engines...

[SenorFunkyPants]

Originally Posted by GhostRideTheWhip

PDK is Porsche's answer to BMW's M-DCT. It's a duel clutch transmission that I personally think bests BMW's.

Hmmm I'm no expert but ISTR that Porsche were the first with a working dual clutch transmission and were using PDKs in their race cars in the 1980's.
VW were next with their road car DSGs and then theres BMW who don't actually make the DCT but licence them from Getrag.

This clip gives a idea of how the PDK (DCT) works: