The consumption of fossil fuels has recently experienced stringent mitigation measures to reduce the emission of greenhouse gases, global warming, and adverse climate changes. Whereas development of renewable energy sources has been prioritized, it is worth noting that the consumption of fossil fuels cannot be stopped once. This calls for cleaner and environmentally friendly technological solutions. In particular, the current internal combustion engine design strategies are driven by environmental and energy concerns regarding fuel consumption and carbon dioxide emission requiring an increase in the thermal efficiency and specific power. This is yet to be achieved in spark-ignition units, primarily owing to the onset of knock that hinders the engine operation under optimal combustion phasing.
Research findings have shown that the cycle-to-cycle variability of mixture preparation and turbulent combustion significantly contributes to the sporadic occurrence of the knocking events. The complexity of its prediction compels conservative operating conditions, thus losing efficiency in the quest for durability.
Cumulative density function (CDF) models have been mainly used in predicting knock onset and understanding knock processes. Regarding the stochastic nature of knock and the turbulence combustion variability, large-eddy simulations (LES) have been used as a suitable technique for computational fluid dynamics (CFD) analysis. This technique is, however, complicated and time demanding and thus not preferred in most industrial design. The development of Reynolds Averaged Navier Stokes (RANS) models representing average flows was a promising solution but only if its limitations such as inability to account for the knock-dependent cycles are addressed.
To address these challenges, a combined effort by Dr. Alessandro d’Adamo, Dr. Sebastiano Breda, Dr. Fabio Berni and Prof. Stefano Fontanesi from the University of Modena and Reggio Emilia simulated the full power high-speed operation of a currently made turbocharged GDI engine under knock-safe, knock-limited and light knocking conditions using the LES, RANS and RANS-PDF techniques. The primary objective was to conduct a critical comparison of the results from the three modeling techniques. This work is currently published in the journal Applied Energy.
Despite a reduction in the number of the engine cycles, large-eddy simulations accurately predicted the edge-of-knock operations thus reproducing the experimental outcomes regarding the variability and average values of both combustion and knock. Besides, these results can be used in cases where the engine cannot be tested for durability. The same engine conditions were simulated using the RANS framework. Unlike large-eddy simulations, it suffered from a systematic underestimation of knock intensity due to the difference between the average knock intensity and the knock intensity of the average cycle. This meant that it could not be used in providing any information regarding knock probability, thus confirming the widely experienced difficulties in correlating knock tendency from RANS simulations with real world engine testing.
An improved statistical RANS-PDF model developed in this study was used to overcome the limitation of the RANS technique by reconstructing the log-normal distributions of the knock intensity. It uses equations for mixture fractions and enthalpy variances. It exhibited a good agreement with the large-eddy simulations results and thus can be directly compared to the test-bench outcomes. This proved the feasibility of the model in dealing with knock statistics, for ease integration with the Reynolds Averaged Navier Stokes-based industrial design workflow.
In a statement to Advances in Engineering, Dr. Alessandro d’Adamo and Prof. Stefano Fontanesi commented “The RANS-PDF model originated after years of struggles in correlating the (minimum) knock intensity from industrial RANS CFD simulations against the evidence of heavily knocking tested engines. Fuel-related chemistry aspects were implemented and did not solve the issue, which appeared linked to some more fundamental model assumption. They further added “Epiphany came when we got a wide distant picture of everything we were doing: the “A” of RANS stands for “Average”. We were treating a stochastic process through averages! With the addition of transported PDFs, the meaningfulness of RANS simulations immediately improved.”
Overall, the presented innovative analysis provides a better understanding of the knock-interceptor phenomena by allowing a coherent comparison of the simulations and experimental results. Dr. Alessandro d’Adamo, highlighted this as a high boost in the design and development of spark ignition engines capable of meeting desirable environmental and energy consumption targets.
Reference
d’Adamo, A., Breda, S., Berni, F., & Fontanesi, S. (2019). The potential of statistical RANS to predict knock tendency: Comparison with LES and experiments on a spark-ignition engine. Applied Energy, 249, 126-142.
