Estimation of the flywheel torque from the combustion torque

[pietro82]

Hi all,

for a customer I have to perform some durability analysis of their transmission. This transmission is installed on a vehicle with direct injection diesel engine. The customer has provided me some CAN traces with the estimated combustion torque, that is the torque estimated from injection timing and flue flow rate. Unfortunately it is not the flywheel torque and it's evident from a quick analysis of the data. Does anyone have any suggestion for estimating the flywheel torque from the combustion torque?
I don't need a precise value, but it must be realistic and not too high.

I'm sorry if my question is not clear, I'm a not a native English speaker

Thanks

 

[ivymike]

You haven't been totally clear in describing what data you have received. It sounds like you have received a calculated "indicated" torque trace for a single cylinder (perhaps?) and you want to create a forcing input for an analytical model of the drivetrain.

If I have that right, then my suggestion would be the following:
- set up your model to include a mass-elastic representation of the engine crankshaft and attachments including the flywheel and reciprocating components
- apply your single-cylinder inputs as excitations to the engine submodel
- find a way to represent frictional losses at each cylinder (use damping elements at the pistons, etc)
- verify that the predicted steady-state output torque and power of the submodel matches those for the real engine

After the above, you have a couple of choices
- Use the flywheel torque predicted by the model above as an input to a separate model of the drivetrain for simplified computation
- Couple the engine submodel directly to the drivetrain model so that they can excite each other

 

[pietro82]

Hi ivymike,

thanks for your reply. Well the torque is the total engine torque I think computed from the injected fuel with a formula like: T(s)=Mf(s)*Fsh/eff, where T(s) is the torque signal, Mf(s) is injected fuel signal, Fsh is the fuel specific heat and eff is total combustion efficiency. I don't have the cylinder pressure and I don't have any transmission model yet. I'm performing some fatigue analysis using the rotating rainflow approach of the before mentioned torque data. Moreover they haven't provided me the predicted steady-state output torque to compare the data I have with real ones. I hoped to use this kind of formula: T_load=T_comb-T_loss, where T_load is the flywheel torque, the T_comb is the combustion torque and T_loss are the losses such as, friction, cooling fan, pumps and so on.

Anyway after perforing this analysis, I'll do what you suggest: develop a transmission model excited by the engine one.

 

[murpia]

Sounds like the main data missing is the friction of the engine:

Crank torque = (Combustion torque) - (Friction torque)​

Regards, Ian

 

[pietro82]

Hi murpia,

you're right, I need a to estimate the friction and I don't know how to do it. In my case:

T_transmission=T_combustion-T_friction-T_coolingfan

Where T_coolingfan is the lost torque due to the fan. The issue is getting T_friction and T_coolingfan. Is there any empirical formula for T_friction? Is there any emperical formula for T_coolingfan from the fan data? I just need an estimation of these values.


Thanks

 

[GregLocock]

You won't be awfully far adrift if you ignore them, but I doubt you'll see much more than 80% of indicated power at the flywheel.

Cheers

Greg Locock


New here? Try reading these, they might help FAQ731-376

http://eng-tips.com/market.cfm?

 

[tbuelna]

The loads transmitted thru the transmission components are limited to the dynamic torque forces produced at the flywheel versus the opposing traction forces at the drive tires contact with the road. The peak instantaneous input torques to the transmission are of great concern when calculating the fatigue life of gears, bearings, shafts, etc. And the mass inertias of the engine components (including the flywheel) can have a significant impact on the peak instantaneous torque forces the transmission components are subjected to versus the individual cylinder combustion pressures.

 

[SomptingGuy]

I get the impression that the OP is somehow hoping to use mean values in a fatigue calc.

- Steve

 

[tbuelna]

Gear fatigue calcs are often based on a composite load/cycle value calculated using something like a cubic mean of several load/speed conditions. With a device like a multi-speed auto transmission, the huge potential number of combinations of engine and transmission operating configurations/conditions makes calculating fatigue life of a specific transmission component a challenge.

 

[pietro82]

Hi all,

thanks for your reply.We need to update the durability target of the transmission using the engine CANBus data. So I have several CAN traces for the aformentioned torque and other data (i.e. engine rpm, engaged gear, vehicle speed). I've firstly used the aforementioned torque to perform the fatigue analysis but I noticed the liftime using that torque is in my opinion too short so I think that is better to investigate more about this torque. So for this reason I want to calculate the real torque on the trasmission.

@tbuelna: you wrote: "Gear fatigue calcs are often based on a composite load/cycle value calculated using something like a cubic mean of several load/speed conditions." I know the rotating rainflow. Which cubic mean do you mean?

thanks

cheers

 

[GregLocock]

If my 80% is right then your fatigue lifes will be out by a factor of 2 short, roughly.

Cheers

Greg Locock


New here? Try reading these, they might help FAQ731-376

http://eng-tips.com/market.cfm?

 

[tbuelna]

pietro82,

Say you are calculating the tooth bending fatigue life of a specific gear in the transmission. You will need to establish what the mean loads and cycles the gear tooth is subjected to. These values are often established using a cubic mean of the gear loads/speeds at percentage of operating life for each. For example, the input gear of the 1st ratio will operate at X rpm/at X load/for X percent of its life, and at Y rpm/at Y load/for Y percent of its life, and at Z rpm/at Z load/for Z percent of its life, etc.

There are published standards for fatigue properties (S-N) of common gear materials from AGMA or ISO. Once you establish what the cubic mean tooth bending stresses are, you can select the appropriate material and heat treatment. Just remember that when it comes to calculating fatigue life, the results are statistical. And thus can vary greatly depending upon factors such as reliability levels.

 

[GregLocock]

I am 99% certain that your engine people can give you a very good estimate of flywheel torque from rpm and fuelling, if they won't do so then keep pestering.


Cheers

Greg Locock


New here? Try reading these, they might help FAQ731-376

http://eng-tips.com/market.cfm?

 

[pietro82]

Hi tbuelna,

I got it, you mean the ISO 6336-6 I use the same approach, but I didn't know the cubic mean load. In few papers I read that it is more used for estimating the lifetime of bearings than of spur gears.

 

[pietro82]

Hi Greg,

I hope so

 

[Tmoose]

I think the full throttle torque applied to the flywheel by the crankshaft varies at least as radically as shown in Figs 2, 3, etc here -

http://www.epi-eng.com/piston_engine_technology/torsional_excitation_from_piston_engines.htm

How much downstream softening/averaging occurs due flywheel inertia/weight, any clutch disk dampers, a long skinny trans input shaft, etc, would need to be pondered upon.

Is "flywheel" being used for the vestigial, not very fly-wheely sheet metal component in the nearly prehistoric auto transmissions? (The only ones of which I have slight knowledge)

http://www.secondchancegarage.com/articles/images/transadapter/photo4.jpg

http://www.jacksautoparts.com/catalog/transmission/flexplate.jpg

If so, despite the light flywheel, I speculate an auto trans torque converter would probably bring the instantaneous output torque closer to an average value, perhaps related to fuel/air use at least when not in lock-up mode.

 

[tbuelna]

pietro82,

When it comes to determining the composite load/cycle value required for a fatigue analysis, the approach used for gears and rolling element bearings is fairly similar. The main difference is that rolling element bearings are usually only concerned with contact stresses, while gears are concerned with both contact stresses and bending stresses.

Of course, with an IC engine drivetrain things get a bit more complicated due to the dynamic variation in torque over a given rotation of the engine output shaft/flywheel. All of the drivetrain components between the flywheel and drive tire contacts will be subject to the relative torque forces existing at their respective shafts. If the peak instantaneous torque created at the flywheel is xx ft-lbs, then the gear mesh in the transmission being used will see the same instantaneous torque forces. Ideally, you would not want to have any gear tooth numbers or meshing frequencies that coincide with the engine's firing frequencies.