Virtual Calibration Case Study – OBD Calibration (Diagnostic)

In the previous blog, I discussed closed-loop virtual calibration that incorporates controls into the model. I also discussed the various types of work and analysis that are performed with closed-loop virtual calibration. In this blog I will go deeper into a case study of how virtual calibration is used to save money on OBD calibration.

OBD is a diagnostic system that is used to protect the powertrain or maintain compliance of various emissions systems. Regulatory agencies have direct visibility of a company’s diagnostic systems, so compliance is crucial. The testing required to develop and calibrate the diagnostics costs upwards of multi-millions of dollars per engine program, and there are very strict deadlines in place for these tasks to be completed.

The consequence of getting the diagnostic calibration wrong is severe and impacts reliability, warranty, and product perception; all very critical areas downstream of powertrain development. It can even delay production which is an expensive outcome for any engineering organization.

Overboost Diagnostic

An example of a high-risk diagnostic is the overboost diagnostic, which is high-risk due to the chance of encountering mechanical limits (cylinder, turbo, etc.) that cause prototype damage or unintended performance degradation during calibration. The ability to perform this test is not difficult but the risk and certainty of robustness are the challenge.

This makes the overboost diagnostic a great example to demonstrate the power of virtual calibration.

There is a significant cost associated with attempting to re-create the failed condition, and there are a few questions that incur a large cost to answer:

  • How much should the component be failed? – Repeated machining then testing
  • Is the diagnostic robust (to ambient/component variability)? – Iterative testing
  • How will the rest of the control system react? – Iterative testing

Using Simulation for Overboost Diagnostics

Simulation helps answer the questions above and allows engineers to front-load overboost diagnostics to decrease the likelihood of repeated prototype damage.

With simulation, the powertrain model can be modified to represent the overboost condition. The variable geometry rack position signal sent to the turbocharger can have a reduced upper limit threshold or the signal can have a reduced gain/offset applied to it. It is up to the model developer to determine which strategy adequately represents their failure mode in conjunction with their OBD demo agreement/requirements. Below are pictures showing how easy it is to implement this in the GT-SUITE model map.

With open-loop virtual calibration, the calibrator runs the model from a nominal (non-failed) powertrain’s cycle and varies the offset/gain/limiter in order to influence the boost pressure. They will be able to determine how close to the mechanical limit they will be able to take the physical powertrain through this method. They can also feed the powertrain output signals into the diagnostic controls to see the impact (obviously not accounting for controller interaction). This provides great cost and time savings.

The diagnostic strategy and threshold are tuned in this virtual environment and refined/validated in a physical test. In a closed-loop simulation the calibrator will be able to treat the powertrain model as a close representation of a physical powertrain. Other portions of the powertrain controls (de-rate, exhaust gas recirculation, etc.) can be observed to see if there is interaction (positive and negative).

The benefit in using virtual calibration in a diagnostic development program is strong. The existing process is quite iterative and risky, and performing most of this work in a virtual environment decreases the time and risk associated with diagnostic development. A few of the benefits of incorporating virtual calibration into the existing process are:

  • Parallelization of the tuning in virtual which allows engineers to receive quick results
  • Flexibility of powertrain model which allows the incorporation of failure modes at will and allows operations at “strange” conditions
  • Zero risk of damage or degradation

Conclusion

Diagnostic development is a very challenging and critical area of powertrain development. The process is made more efficient by incorporating virtual calibration into the current process and by finalizing/refining the work with testing.

This finishes up the series of blogs on virtual calibration for powertrain development for now.

If you are interested in discussing virtual calibration or would like more details please reach out and contact us.

Written By: Michael Bambula