Engine simulation software has become synonymous with automotive design and component development. An integral part of any engine simulation is the correct modeling of the air flow at the intake. This air flow, which is highly compressible and unsteady, has a first order influence on the trapped air mass inside the cylinder and therefore on the behavior of the engine (torque response and emissions). The non-linear modeling of the air paths at the intake is done using a space-time meshing and by solving the 1D equations with a proper time scheme. Such methods are the bases of today’s engine simulation codes [1]. The main constraint with these methods is the time needed to model complex geometries, whether being the simulation time or the time spent on calibrating the said models with experimental measurements. These complicated geometries become problematic to accurately predict, particularly the charge air cooler (CAC) which is responsible for cooling the air flow on a turbocharged engine.
Another approach is to use frequency domain models to describe the fluctuating pressure and mass flow [2]. Although this approach is simpler, faster in terms of computing time and offers many experimental techniques to characterize complicated geometries; important limitations can appear when it is confronted to the effects of high pressure levels and pulsating mass flow. Furthermore, the models behind such methods are designed to be used in the frequency domain, contrary to an engine simulation that works solely with time domain variables.
In this article, a linear frequency domain model known as a transfer matrix is used. This concept is nothing new in acoustics; however here it was developed by experimentally measuring the transfer matrix [3] for a simple tube on a dedicated bench under conditions similar to those encountered on an engine [4]. The approach is then extended to measure the transfer matrix of a charge air cooler (CAC) on a real engine test bench. The measured discrete transfer matrix, defined in terms of fluctuating pressure and mass flow, is then transformed to a continuous frequency model and coded in Simulink®. The latter is coupled to the non-linear engine simulation software GT-Power®.
The objective is to accurately model the pressure and mass flow of a complicated geometry using experimental measurements and a linear frequency model then to couple the transfer matrix to an engine simulation code, thus replacing the need for a meshed model.
