Lumped Capacitance and Three-Dimensional Computational Fluid Dynamics Conjugate Heat Transfer Modeling of an Automotive Turbocharger

One-dimensional wave-action engine models have become an essential tool within engine development including stages of component selection, understanding system interactions, and control strategy development. Simple turbocharger models are seen as a weak link in the accuracy of these simulation tools, and advanced models have been proposed to account for phenomena including heat transfer. In order to run within a full engine code, these models are necessarily simple in structure yet are required to describe a highly complex 3D problem. This paper aims to assess the validity of one of the key assumptions in simple heat transfer models, namely, that the heat transfer between the compressor casing and intake air occurs only after the compression process. Initially, a sensitivity study was conducted on a simple lumped capacity thermal model of a turbocharger. A new partition parameter was introduced α_A, which divides the internal wetted area of the compressor housing into pre- and postcompression. The sensitivity of heat fluxes to α_A was quantified with respect to the sensitivity to turbine inlet temperature (TIT). At low speeds, the TIT was the dominant effect on compressor efficiency, whereas at high speed α_A had a similar influence to TIT. However, modeling of the conduction within the compressor housing using an additional thermal resistance caused changes in heat flows of less than 10%. Three-dimensional computational fluid dynamics (CFD) analysis was undertaken using a number of cases approximating different values of α_A. It was seen that when considering a case similar to α_A = 0, meaning that heat transfer on the compressor side is considered to occur only after the compression process, significant temperature could build up in the impeller area of the compressor housing, indicating the importance of the precompression heat path. The 3D simulation was used to estimate a realistic value for α_A which was suggested to be between 0.15 and 0.3. Using a value of this magnitude in the lumped capacitance model showed that at low speed there would be less than 1% point effect on apparent efficiency which would be negligible compared to the 8% point seen as a result of TIT. In contrast, at high speeds, the impact of α_A was similar to that of TIT, both leading to approximately 1% point apparent efficiency error.