Technical Library



Simple Models of Engine Noise
by Glenn Pietila

This is a review of an interesting paper I attended while at the SAE NVH conference in Traverse City last month. The authors wanted to refine the simple parametric model that they use to predict engine noise in the early stages of vehicle development. Furthermore, they wanted to use the legacy data, gathered from years of testing engines according to the J1074 standard, to verify engine performance trends and correct if necessary the corresponding parametric models.

According to the J1074 standard, the noise radiated by the engine is characterized with a four microphone average during an rpm sweep at three operating conditions: full load, no load and no load with accessory belt removed. Using these measurements, combustion noise is defined by the difference between full load and no load and accessory noise as the difference between no load with accessory and no load with the accessory belt removed. Mechanical load noise is assumed to be the noise measured at no load without the accessory belt. With these assumptions, the total engine noise is decomposed into its main contributions: combustion, mechanical and accessories.

This technique was applied to three engines of similar layout and structure, but different displacement (i.e. different bore and stroke). A new parametric model was derived to describe the differences of the mechanical noise among the three engines. According to this model, mechanical noise is most effectively reduced by reducing the stroke. As for the combustion noise, a simple regression equation based on engine power density (horsepower/displacement) provided the best fit to the data from the three engines. By using this model, the effect of increased engine power density on total radiated noise can be accounted for when positioning a new design engine relative to current competitive engines.

After watching the paper and reading the written version, I have to admit that I was surprised to see that simple, parametric models can be used to predict the noise radiated by the engine.