Are You A Gearhead In A Control Engineer’s Clothing?

By Raymond Turin

You don’t mind getting your hands dirty if that’s what it takes to squeeze every last drop of performance out of your engine. But, of course, you also like to get the job done. So, what if your job is to develop the type of engine controls that set the bar for the rest of the competition? The type of controls that embody the proverbial “out of the box” thinking.


If that is you, then every trip to your experimental set up, the engine dyno, the vehicle dyno, the test vehicle, etc., as much as it may be in line with your inclinations, becomes an impediment to progress.


The proper configuration, instrumentation, and utilization of the test infrastructure is a nasty kind of business. So nasty, in fact, that one could easily conclude that it was the breeding ground for Murphy’s law. Hail thee, Sisyphus!

Nevertheless, in order to validate and debug any new control algorithm, the proper experimental setup is an absolute prerequisite. Or is it?


… that you only needed tools within arm’s reach. And that you could develop, debug, and validate sophisticated engine controls within the confines of your desk. Experimental validation is now a mere afterthought. What would it take to accomplish this?

Clearly, at the center of your needs would be an appropriate surrogate for your engine. Namely, an engine model that:

  • Can be quickly configured in native Simulink to represent your engine
  • Is accurate across the entire engine operating envelope
  • Enables interactive control algorithm development and debugging
  • Is supported by a modular control strategy
  • Is available before actual hardware
  • Can be automatically tuned to precisely match your engine once actual engine data is available
  • Simulates fast enough to be useful for interactive control design

There are a handful of commercial engine models that match some of these criteria, but there is only one that matches all of them.

The widespread mean-value models, which treat the engine as a continuous pump, lack the features to emulate important nonlinear effects, e.g., effects that influence operational stability around idle speed. They are also poor at predicting the effect of changing engine geometry parameters. The same is true for any type of purely regression-based models.

Wave propagation models are widely used to support engine design. However, due to the need for iterative numeric integration, they simulate far too slowly to be useful for interactive engine control design. Real-time or faster simulation is a crucial aspect of efficiently debugging new control algorithms under various engine operating scenarios.

The only class of models with the potentialCrank angle to meet the above requirements are zero-dimensional, physics-based models that accurately emulate the event-driven combustion characteristics of an engine. Zero-dimensional, in this context, means that spatial effects of mass and energy storage/flow are neglected. This permits the use of ordinary differential equations to describe the overall engine behavior. To further limit computational expense, the complex effects of reacting gas species during combustion are captured in the terms of standard regression-based heat-release models. These simplifications are important prerequisites for real-time execution.

As an expert in the field, you may ask: “Using this type of model, how can you possibly achieve the necessary accuracy across the entire engine operating envelope?” This is precisely where Enginuity distinguishes itself from other physics-based, real-time models.

Enginuity hinges on a unique set of models that accurately capture the combustion characteristics in a single engine operating point, complemented with models that capture the influence of changes in other operating conditions. Enginuity also provides a feature to automatically tune the combustion parameters based on standard experimental engine mapping data. Using these tools, you are ready to go within a short period of time.

A properly tuned Enginuity model puts you in a position to conduct all critical control design and validation steps right at your desk. Use your time to generate IP rather than wrestling with loose connections, crossed wires, or with the almost hopeless attempt to rapidly manufacture repeatable experimental setups for critical dynamic transients.

Design and debug your controls rapidly at your desk. Once validated, download them right onto your test vehicle with confidence.

Next Steps:

  • Want more great insights?  Subscribe to our blog.
  • Generate C code automatically for your driver controls with QuantiPhi.
  • Data dependency analysis in UniPhi reduces debugging to a fraction the industry tolerates.