Dual-clutch gearboxes -- torque rating and tuned engines

[ijd]

I posted this recently on another forum (S-max owners club) with no response -- can anyone help? What we need to answer this is someone who knows the nuts and bolts of what goes on inside a DCT box (preferably the Ford/Getrag one), especially about how it communicates with and co-operates with the engine management system.

Since Ford have publicly stated that they want to replace conventional autos with the Powershift DCT box across the entire car range, I guess it won't be too long before the choice is DCT or manual -- and at this point I expect many people will opt for the DCT given that it doesn't penalise either power or performance like a conventional auto.

This then raises some interesting questions about what happens with tuned engines, especially turbodiesels with massively increased torque...

To get smooth shifts most DCT boxes include torque sensors (I don't know about the Ford/Getrag one, can anyone confirm this?), and the gearbox control system has to be closely integrated with the engine management system, so that fuelling is changed momentarily on up/down shifts.

What happens if/when the engine is tuned? Does this screw up the shift quality unless the gearbox control system is also reprogrammed? How do the tuners know how to change the mapping (and do they?) to keep the gearbox and engine working together smoothly? (I gather this is one of the biggest problems with DCT, integration of the two systems not just adding the new box in).

An even bigger issue could be what happens if the gearbox limits the allowable engine torque (using the torque sensors) to protect itself, or at least keep torque within the specified limits? Start with the 2.2TTDCi at 400nm, tune it up to 500-550nm or more, then connect to a DCT box with a 450nm torque rating which is determined to keep it that way by telling the engine to cut the juice...

Can someone with experience of tuned engines (especially diesels) and knowledge of DCT boxes shed some light on these issues? (no problem, small problem, big problem...)

Ian

[GavinPearson]

I don't know who uses a torque sensor, to calculate slip you need speed sensors. But regardless it's not as big a problem as you might think, as the engine will generate a lot of torque in cold ambient eg Sweden and not much hot eg a Mediterranean summer's day.

The key to all this is that the engine looks at the things going on in the engine and calculates a torque output. When the transmission is going to shift the engine torque output is cut, the power shifts from one clutch to another at a calibrated rate, and then the torque is allowed to increase to the previous level.

If the engine is tuned by increasing intercooling then the signal will remain accurate. If the ECM tables are altered and the torque signal is inaccurate then the shifts will be rough and inconsistent, potentially with way too much slip and causing premature failure. Bottom line - the method of tuning needs to be very well thought out to match the engine and transmission and not overload the capacity of the clutches.

[ijd]

GavinPearson said:

I don't know who uses a torque sensor, to calculate slip you need speed sensors. But regardless it's not as big a problem as you might think, as the engine will generate a lot of torque in cold ambient eg Sweden and not much hot eg a Mediterranean summer's day.

The key to all this is that the engine looks at the things going on in the engine and calculates a torque output. When the transmission is going to shift the engine torque output is cut, the power shifts from one clutch to another at a calibrated rate, and then the torque is allowed to increase to the previous level.

If the engine is tuned by increasing intercooling then the signal will remain accurate. If the ECM tables are altered and the torque signal is inaccurate then the shifts will be rough and inconsistent, potentially with way too much slip and causing premature failure. Bottom line - the method of tuning needs to be very well thought out to match the engine and transmission and not overload the capacity of the clutches.

So it's an open-loop system based on estimated torque output from the engine, not a closed-loop one? No wonder the manufacturers find it difficult to get just right if they're guessing what's coming out of the engine instead of measuring it!

I also thought that the speed of change (rate of clutch changeover) was also varied depending on how fast the change needs to be (throttle opening etc) not just a fixed-speed change? Also some manufacturers claim the changeover is adaptive to compensate for wear and variation and give the smoothest change, but this could be bull**it...

Ian

Ian

[GavinPearson]

ijd said:

So it's an open-loop system based on estimated torque output from the engine, not a closed-loop one? No wonder the manufacturers find it difficult to get just right if they're guessing what's coming out of the engine instead of measuring it!

I also thought that the speed of change (rate of clutch changeover) was also varied depending on how fast the change needs to be (throttle opening etc) not just a fixed-speed change? Also some manufacturers claim the changeover is adaptive to compensate for wear and variation and give the smoothest change, but this could be bull**it...

Ian

The difficulty in answering is trying to explain one of the most complicated systems on a vehicle and getting it into a paragraph is less than easy.

The torque signal is very reliable - the models used to predict it are now very refined. In addition the clutches are pulled in at a pre-calibrated rate and adapt values are used to keep shift refinement at the optimum level. Speed sensors are used to assist the closed loop control elements.

But back to the issue of calibrating the system to suit an increased output engine - if the tuner increases output to levels within the capacity of the clutches, and the torque signal is reliable, there won't be a problem. But the tuner also needs to be cognaisant of the fact that many manufacturers limit torque in a number of gears and so unless that is adjusted the engine will behave much the same as it currently does.

[ijd]

GavinPearson said:

The difficulty in answering is trying to explain one of the most complicated systems on a vehicle and getting it into a paragraph is less than easy.

The torque signal is very reliable - the models used to predict it are now very refined. In addition the clutches are pulled in at a pre-calibrated rate and adapt values are used to keep shift refinement at the optimum level. Speed sensors are used to assist the closed loop control elements.

But back to the issue of calibrating the system to suit an increased output engine - if the tuner increases output to levels within the capacity of the clutches, and the torque signal is reliable, there won't be a problem. But the tuner also needs to be cognaisant of the fact that many manufacturers limit torque in a number of gears and so unless that is adjusted the engine will behave much the same as it currently does.

I understand complicated, I'm an engineer :-)

I assume the estimated torque is based on fuel flow? Th.is should be accurate for a diesel with enough air for complete combustion (which ought to be the case or particle emissions will go through the roof) because efficiency is almost constant, so even after tuning this will be correct -- so long as the tuning isn't done with (for example) an add-on box that tells lies about fuel flow to persuade the engine management system (EMS) to squirt more in an an attempts to produce more power (and smoke). It might be a lot harder to ensure that it stays correct for a tuned petrol engine which has a much less direct fuel-torque link.

I assume that tuners must know about torque limits in gears, but probably ignore or increase them to get increased performance -- though this must present some risk to transmission parts downstream of the gearbox like halfshafts.

So what happens if the torque exceeds the capacity of the clutches, and is this the same thing as the official torque rating of the gearbox? Is limiting the torque the sole responsibility of the EMS, or does the control unit in the gearbox tell it "whoa, too much torque, turn the juice down" -- and what happens if the EMS ignores this?

I also assume that the detailed control of the gearchange is done by the controller inside the box, getting a torque signal as input from the EMS and outputting some request to the EMS to momentarily decrease or increase fuelling during up and down changes to allow for the inertia of the engine/flywheel -- is this correct?

If so, then the EMS code also needs to take this into account to get smooth changes out of the gearbox, and again this is something I'm not sure engine tuners know about or have the equipment or knowledge to deal with (or would spend the hours to get right).

Basically, I'm trying to find out if tuning an engine attached to a DCT gearbox is a much harder job to get right than one attached to a manual gearbox, and if there are likely to be any unforeseen issues like the gearbox actively limiting the torque and/or the change quality being screwed up -- I did read that designing a DCT gearbox was the easy bit, optimising the changes and getting it to work correctly and smoothly with the EMS was much more difficult and time-consuming (and therefore the bit that engine tuners are unlikely to do properly).

Gavin, it sounds as if you know the answer to this...

Ian

P.S. I'm not an engine tuner, just one of many interested potential customers :-)

[GavinPearson]

ijd said:

I assume the estimated torque is based on fuel flow? Th.is should be accurate for a diesel with enough air for complete combustion (which ought to be the case or particle emissions will go through the roof) because efficiency is almost constant, so even after tuning this will be correct -- so long as the tuning isn't done with (for example) an add-on box that tells lies about fuel flow to persuade the engine management system (EMS) to squirt more in an an attempts to produce more power (and smoke). It might be a lot harder to ensure that it stays correct for a tuned petrol engine which has a much less direct fuel-torque link.

Torque will be based on not only fuel flow but whether the engine is regenerating, the timing of the individual quantities of fuel, plus boost, egr rates, back pressure and that's just the start.

ijd said:

I assume that tuners must know about torque limits in gears, but probably ignore or increase them to get increased performance -- though this must present some risk to transmission parts downstream of the gearbox like halfshafts.

In many cases they just turn them off and yes, it does cause issues downstream.

ijd said:

So what happens if the torque exceeds the capacity of the clutches, and is this the same thing as the official torque rating of the gearbox? Is limiting the torque the sole responsibility of the EMS, or does the control unit in the gearbox tell it "whoa, too much torque, turn the juice down" -- and what happens if the EMS ignores this?

If torque exceeds cluctch capacity you get slip. Torque limitation is a joint responsibility. The engine provides a torque signal, during a change the trans commands a torque reduction, if that is not met it commands more. What happens next is dependent on the engine & transmission management strategies.

ijd said:

I also assume that the detailed control of the gearchange is done by the controller inside the box, getting a torque signal as input from the EMS and outputting some request to the EMS to momentarily decrease or increase fuelling during up and down changes to allow for the inertia of the engine/flywheel -- is this correct?

Broadly yes, but also accounts for vehicle speed loss / gain on inclines, and aerodynamic drag.

ijd said:

If so, then the EMS code also needs to take this into account to get smooth changes out of the gearbox, and again this is something I'm not sure engine tuners know about or have the equipment or knowledge to deal with (or would spend the hours to get right).

It really is an awful lot of work to do properly. In many cases the work might have already been done. I know of a supercharged petrol engined vehicle where the torque signal is accurate to 30% over nominal and the transmission can take it, so the tuner looks like a genius for making the blower spin faster and putting a very big intercooler on the car. But if the work wasn't done it would be a total disaster.

ijd said:

Basically, I'm trying to find out if tuning an engine attached to a DCT gearbox is a much harder job to get right than one attached to a manual gearbox, and if there are likely to be any unforeseen issues like the gearbox actively limiting the torque and/or the change quality being screwed up -- I did read that designing a DCT gearbox was the easy bit, optimising the changes and getting it to work correctly and smoothly with the EMS was much more difficult and time-consuming (and therefore the bit that engine tuners are unlikely to do properly).

I'd say it's a very hard job to get right and I would avoid like the plague until it is proven that those selling the chips have a decent reliability record. But there is stuff that can be done that will have an immediate benefit that is already calibrated into the chip and down to people to take advantage of (better intercooling, less restrictive air cleaner) and it will be far gentler on the engine components.

[ijd]

GavinPearson said:

I'd say it's a very hard job to get right and I would avoid like the plague until it is proven that those selling the chips have a decent reliability record. But there is stuff that can be done that will have an immediate benefit that is already calibrated into the chip and down to people to take advantage of (better intercooling, less restrictive air cleaner) and it will be far gentler on the engine components.

Thanks Gavin, that's all very helpful, it's a pleasure to get a reply from someone who obviously knows the subject :-)

Of course with modern turbodiesels (which is where this question started off) by far the most common approach is just chipping or reprogramming the ECU -- basically by turning up the boost pressure and fuelling -- because this needs no hardware changes, so nosy garages and insurers don't spot it and do unpleasant things like voiding warranties or insurance policies...

It seems fairly normal to get about 20% increase in power and torque this way (range seems to be 15-25% looking at published figures). The engine we're talking about is the PSA twin-turbo 2.2HDI which Ford have just introduced, which as standard is rated at 175ps and 400nm (maybe 420nm with overboost), the DCT box is the Ford-Getrag Powershift rated at 450nm which should appear paired with the 2.2TTDCi later this year. So with just ECU remapping I expect at least 500nm is likely, which is 50nm above the box rating.

There are 3 possible problems I can see, in ascending order of severity:

1. Jerky changes due to increased torque/remapping
2. Torque being limited by box to 450nm by instructing ECU to back off
3. Clutch packs wearing out prematurely and having to be (expensively) replaced

1) is just a driveability issue, 2) is the driver only getting half the torque increase they've paid for, but 3) is the killer because it won't show up for some time and until there are a fair number of remapped cars out there -- at which time it suddenly starts getting very expensive for the owners, who blame the tuner, who points out that they don't guarantee that the transmission won't be damaged by their remap :-)

I know it's asking a difficult question, but for these torque figures do you have any guess whether 1) 2) or 3) is likely? (no guarantees here, just an engineering opinion :-)

As far as avoiding like the plague is concerned, this is a chicken-and-egg situation -- you can't prove it's reliable (or not) until there are enough remapped cars out there, but if the owners wait until reliability has been proved before getting it done there won't be the cars out there to prove it either way...

Cheers

Ian

[ijd]

By the way -- Gavin, have you worked on DCT development? It certainly sounds that way...

If so please do give some details, I'm being plagued by a bull**itter who is slagging off DCTs in spite of knowing little about them and never having driven one, and who's basically said that nothing posted on the Pistonheads forum is worth taking any notice of :-)

Cheers

Ian

[GavinPearson]

ijd said:

There are 3 possible problems I can see, in ascending order of severity:

1. Jerky changes due to increased torque/remapping
2. Torque being limited by box to 450nm by instructing ECU to back off
3. Clutch packs wearing out prematurely and having to be (expensively) replaced

1) is just a driveability issue, 2) is the driver only getting half the torque increase they've paid for, but 3) is the killer because it won't show up for some time and until there are a fair number of remapped cars out there -- at which time it suddenly starts getting very expensive for the owners, who blame the tuner, who points out that they don't guarantee that the transmission won't be damaged by their remap :-)

I know it's asking a difficult question, but for these torque figures do you have any guess whether 1) 2) or 3) is likely? (no guarantees here, just an engineering opinion :-){/quote]
Realistically all three are going to be affected and that is why I wouldn't bother.

It says in the owner's handbook that once you modify the car that you are on your own for warranty and if there was a problem - head gaskets, burnt pistons or the transmission then the bill would be expensive.

A lot of the electronics firms are putting software in to detect "chipping" for this very reason.

ijd said:

By the way -- Gavin, have you worked on DCT development? It certainly sounds that way...

If so please do give some details, I'm being plagued by a bull**itter who is slagging off DCTs in spite of knowing little about them and never having driven one, and who's basically said that nothing posted on the Pistonheads forum is worth taking any notice of :-)

Cheers

Ian

Only in a very small way (in OEM engineering terms), but I like to stay informed. But the control principles for clutch pack changeover are very similar to automatics which I have a fair amount of experience with.

[ijd]

GavinPearson said:

ijd said:

I know it's asking a difficult question, but for these torque figures do you have any guess whether 1) 2) or 3) is likely? (no guarantees here, just an engineering opinion :-){/quote]
Realistically all three are going to be affected and that is why I wouldn't bother.

It says in the owner's handbook that once you modify the car that you are on your own for warranty and if there was a problem - head gaskets, burnt pistons or the transmission then the bill would be expensive.

A lot of the electronics firms are putting software in to detect "chipping" for this very reason.

Indeed they are, but isn't this a bit like the continuing arms race between DRM proponents and free-everything hackers? A lot of the tuning firms claim their ECU reprogramming is either not easily detectable (by most garages, anyway), reversible without trace, or both. As soon as the manufacturers come up with a new method of protecting their code the hackers -- sorry, tuners -- will be working on a way to crack this.

With turbodiesels in particular there doesn't seem to be much risk in getting a reasonable increase in power/torque via remapping without reliability problems, given the safety margins that manufacturers have to build in. This is why quite a lot of people go down this route -- I guess if you then drove at maximum power for a long time up a hill at 50C ambient something in the engine might die, but most cars never get driven in a worst-case situation like this.

However if exceeding the box torque rating on a DCT makes the clutches slip at high engine speed (as opposed to startoff or momentarily during gear changes) then the clutch pack would destroy itself in very short order due to massive heat dissipation. If the remapped ECU ignores (or doesn't understand) requests from the DCT to turn down the juice this could be an immediate and very expensive consequence of remapping.

Of course the box torque rating must also have some safety margin built in so this *probably* won't happen so long as the torque doesn't exceed this by too much. But it's not comfortable to have a failure mechanism which could destroy the clutch packs in a few seconds, maybe without much or any warning -- a bit like the onset of high-speed detonation which can destroy an engine in a similar time.

The difference is that tuners are able to spot and prevent high-speed detonation, it's quite possible they wouldn't spot clutch pack slip -- or more likely it doesn't happen when they do the remapping when the car is new, but after several thousand miles when the clutches have worn a bit they spontaneously self-combust when you put your foot down and a gearchange triggers clutch pack slip... :-(

Ian

[GavinPearson]

ijd said:

With turbodiesels in particular there doesn't seem to be much risk in getting a reasonable increase in power/torque via remapping without reliability problems, given the safety margins that manufacturers have to build in. This is why quite a lot of people go down this route -- I guess if you then drove at maximum power for a long time up a hill at 50C ambient something in the engine might die, but most cars never get driven in a worst-case situation like this.

I wouldn't go assuming anything. When you get high boost and cold ambient temperatures the cylinder pressures go sky high, enough to blow a head gasket.

ijd said:

However if exceeding the box torque rating on a DCT makes the clutches slip at high engine speed (as opposed to startoff or momentarily during gear changes) then the clutch pack would destroy itself in very short order due to massive heat dissipation. If the remapped ECU ignores (or doesn't understand) requests from the DCT to turn down the juice this could be an immediate and very expensive consequence of remapping.

It will slip at peak torque, which could be quite a low speed in the case of a turbodiesel.

ijd said:

Of course the box torque rating must also have some safety margin built in so this *probably* won't happen so long as the torque doesn't exceed this by too much. But it's not comfortable to have a failure mechanism which could destroy the clutch packs in a few seconds, maybe without much or any warning -- a bit like the onset of high-speed detonation which can destroy an engine in a similar time.

Again, don't go betting on how high the safety margin is. In the past 15 years the automotive industry has become increasingly able to add electronic monitors to mechanical systems to downsize them for improved economy yet still make them durable. But in most cases there is very little extra margin. The rules of old do not apply any more.

ijd said:

The difference is that tuners are able to spot and prevent high-speed detonation, it's quite possible they wouldn't spot clutch pack slip -- or more likely it doesn't happen when they do the remapping when the car is new, but after several thousand miles when the clutches have worn a bit they spontaneously self-combust when you put your foot down and a gearchange triggers clutch pack slip... :-(
Ian

I wouldn't even go that far, a tuner does their work on a sample of one vehicle, and OEMs do their work on a statistically valid sample verified on fleets of vehicles. In the case of dealing with the effects of tuning plus exuberant use the most likely scenario is an OBD monitor detecting excessive slip and putting the engine into some sort of significantly reduced power mode.

[ijd]

I agree that safety margins for this kind of thing are less than they used to be, because more accurate design and analysis (and more sophisticated protection) allows this -- after all, if something is just strong enough for its purpose that's fine, and it's lighter and cheaper than something which is heavily over-engineered.

None of which will stop people tuning engines, especially turbodiesels where it's apparently so easy with fewer potential problems (like detonation) than petrol engines. In this case peak torque not only increases but always moves up the rev range to 3000rpm or so because the boost limit has been increased, and if slip happens at these revs the question is whether the gearbox will be able to save itself or not -- if the engine heeds the request to turn the torque down then no harm will be done, if not the clutch pack will die very rapidly because people are much less likely to spot slip with an automatic box where they expect the engine speed to change anyway.

I'm not saying that the engine tuners are especially careful or clever, after all they're using up most or all of the safety margin built in by the manufacturers. What I'm trying to spot is whether DCT adds a new (and potentially rather expensive) risk when tuning an engine, and the answer seems to be yes.

Ian