Torque Model Calibration

[sprasad6]

If anyone has worked as a calibrator with the Bosch s/w I'd like to understand a few things:

1. How is the torque model created?
2. During the base calibration if the vvt sweep for a intake vvt engine is done for the best CA50 ignition then why do a ignition sweep after that, considering you would get the best results with CA50 ignition

Thank guys.

[anonymous-user]

sprasad6 said:

If anyone has worked as a calibrator with the Bosch s/w I'd like to understand a few things:

1. How is the torque model created?
2. During the base calibration if the vvt sweep for a intake vvt engine is done for the best CA50 ignition then why do a ignition sweep after that, considering you would get the best results with CA50 ignition

Thank guys.

Modern EMS are calibrated with a Design Of Experiments (DOE) simultaneous multi-dimensional model with a a Degrees of Freedom (DOF) approach, typically, something like 4 to 8 variables will be "mapped" in one pass, and the model based optimiser will then use a stastical approach to curve fit the surfaces to each DOF. In this way, we can adjust igntion, cam timing, ignition ange, injection angle, fuel mass, boost pressure, throttle position, EGR ratio, simultaneously to minimise the data collection mandate that would otherwise be logiscially burdensome! Those parameters will typically be driven during this process by automated "automappers" using a "control in the loop" type architecture.

Once the surfaces are calculated, individual parameterisations can then be drawn from the model to suit the particular calibration tables used by an EMS. The inverse torque model is also constructed to use as the torque feed forward estimator.

These days in the quest for ultimate efficiency, real time cylinder pressure data will be used to calculate the thermal efficiency and to drive the optimisation of the final calibration characteristics.


Simples. ;-)

[catman]

Max_Torque said:

Modern EMS are calibrated with a Design Of Experiments (DOE) simultaneous multi-dimensional model with a a Degrees of Freedom (DOF) approach, typically, something like 4 to 8 variables will be "mapped" in one pass, and the model based optimiser will then use a stastical approach to curve fit the surfaces to each DOF. In this way, we can adjust igntion, cam timing, ignition ange, injection angle, fuel mass, boost pressure, throttle position, EGR ratio, simultaneously to minimise the data collection mandate that would otherwise be logiscially burdensome! Those parameters will typically be driven during this process by automated "automappers" using a "control in the loop" type architecture.

Once the surfaces are calculated, individual parameterisations can then be drawn from the model to suit the particular calibration tables used by an EMS. The inverse torque model is also constructed to use as the torque feed forward estimator.

These days in the quest for ultimate efficiency, real time cylinder pressure data will be used to calculate the thermal efficiency and to drive the optimisation of the final calibration characteristics.


Simples. ;-)

If you don't know, just say so. The guy needs help.

[GreenV8S]

catman said:

If you don't know, just say so. The guy needs help.

[AER]

sprasad6 said:

If anyone has worked as a calibrator with the Bosch s/w I'd like to understand a few things:

1. How is the torque model created?
2. During the base calibration if the vvt sweep for a intake vvt engine is done for the best CA50 ignition then why do a ignition sweep after that, considering you would get the best results with CA50 ignition

Thank guys.

The whole idea of a torque model is that you can choose to operate the engine away from MBT ignition and best VE for various control reasons (i.e idle control, gearshifting, traction control etc.) so you're not only calibrating for the optimum point but all the possible operating points and generating a model that estimates the torque this will deliver - hence "torque model"

Yes, Max explained pretty much how its done. It's a bit complicated but not too much so. Just statistical models regressed from tonnes of data.

Edited by AER on Wednesday 9th December 10:15