Turbocharger modeling

The flow in engine turbocharger compressors and turbines is highly unsteady in nature, as it responds to the intake and exhaust manifolds of the internal combustion engine. The optimization of the turbocharger system is therefore a very difficult task, since the devices operate at off-design conditions for most of the engine cycle. Experimental studies allow for improving the understanding on the behavior of the engine components, in particular when tests are performed under real engine operating conditions; however, the experimental tests can be more efficient if they are combined with theoretical simulation tools, which help to select significant engine operating conditions.

Novel one-dimensional approaches to model turbocharger turbines working under unsteady pulsating flows have been implemented. Models have been validated against data coming from experimental investigations performed on a test rig at the ICE Laboratory of the University of Genoa (ICEG-DIMSET, Prof. M. Capobianco).


Related publications and documents:
1 - F. Piscaglia, A. Onorati, S. Marelli, M. Capobianco, "Unsteady behavior in turbocharger compressors and turbines: Experimental analysis and numerical simulation", 8th International Conference on "Engines for Automobile" - ICE2007, SAE paper n. 2007-24-0081

Turbocharger maps fitting and extrapolation

 The basic assumption in the map fitting method is, for a given turbine geometry, that all normalized experimental points may be fitted by one function, if they are plotted in the normalized mass flow rate-normalized BSR plane; similarly, all normalized efficiency points will be fitted by a single curve in the normalized BSR-normalized efficiency plane. In both cases, the fitting procedure is based on a non-linear least-squares method. A restricted step method is used to approximate only a certain region (trust region) of the objective function; when an adequate model of the objective function is found within the trust region, the region is then expanded. Conversely, if the approximation is poor, then the region is contracted. The trust region algorithm performs iterations to seek the best coefficients to fit the curve in the normalized mass flow rate-normalized BSR plane and in the normalized BSR-normalized efficiency plane. Finally, turbine maps are built by scaling non-dimensional curves for an arbitrary range of turbocharger speeds of practical interests. For each turbocharger shaft speed, the optimum mass flow ratio, the optimum Blade Speed Ratio and the optimum efficiency are derived; values will be used to to compute actual mass flow rates and actual BSRs from non-dimensional plots. The algorithm automatically detectes choked flow conditions. The procedure is repeated for each VGT rack position and for each rotational speed in the range considered. Sometimes user data may not follow some expected theoretical behavior or may cover only a very small range along the speed lines. This often happens especially for data points at the lowest pressure ratios, because they are difficult to measure or because sometimes they are subsequently added to the turbine map. Because of this, those data points do not follow the pattern of the real data and the algorithm might produce poor fit and extrapolation of the raw data. When data scattering occurs, an algorithm to skip scattered data during the fitting procedure has been employed. The method has proved to be sufficiently robust even if the quality of experimental data is poor.

Related publications and documents:
1 - G. D'Errico, G. Montenegro, A. Onorati, F. Piscaglia. "Integrated 1D-multiD Fluid Dynamic Simulation of a Turbocharged Diesel Engine with Complete Intake and Exhaust Systems", SAE Int. Congress & Exposition 2010, paper n. 2010-01-0194, April 13-15, 2010, Detroit, Michigan, USA
2 - F. Piscaglia, A. Montorfano. Private notes.