A deeper understanding of the physical nature of cycle-to-cycle variations (CCV) in internal combustion engines (ICE) as well as reliable simulation strategies to predict these CCV are indispensable for the development of modern highly efficient combustion engines. Since the combustion process in ICE strongly depends on the turbulent flow field in the cylinder and, for spark-ignited engines, especially around the spark plug, the prediction of CCV using computational fluid dynamics (CFD) is limited to the modeling of turbulent flows. One possible way to determine CCV is by applying large eddy simulation (LES), whose potential in this field has already been shown despite its drawback of requiring considerable computational time and resources. This paper presents a novel strategy based on unsteady Reynolds-averaged Navier–Stokes (uRANS) CFD in combination with variational autoencoders (VAEs). A VAE is trained with flow field data from presimulated cycles at a specific crank angle. Then, the VAE can be used to generate artificial flow fields that serve to initialize new CFD simulations of the combustion process. With this novel approach, a high number of individual cycles can be simulated in a fraction of the time that LES needs for the same amount of cycles. Since the VAE is trained on data from presimulated cycles, the physical information of the cycles is transferred to the generated artificial cycles.

Cycle-to-cycle variations (CCV) in spark-ignited (SI) engines impose performance limitations and in the extreme limit can lead to very strong, potentially damaging cycles. Thus, CCV force sub-optimal engine operating conditions. A deeper understanding of CCV is key to enabling control strategies, improving engine design and reducing the negative impact of CCV on engine operation. This paper presents a new simulation strategy which allows investigation of the impact of individual physical quantities (e.g., flow field or turbulence quantities) on CCV separately. As a first step, multi-cycle unsteady Reynolds-averaged Navier–Stokes (uRANS) computational fluid dynamics (CFD) simulations of a spark-ignited natural gas engine are performed. For each cycle, simulation results just prior to each spark timing are taken. Next, simulation results from different cycles are combined: one quantity, e.g., the flow field, is extracted from a snapshot of one given cycle, and all other quantities are taken from a snapshot from a different cycle. Such a combination yields a new snapshot. With the combined snapshot, the simulation is continued until the end of combustion. The results obtained with combined snapshots show that the velocity field seems to have the highest impact on CCV. Turbulence intensity, quantified by the turbulent kinetic energy and turbulent kinetic energy dissipation rate, has a similar value for all snapshots. Thus, their impact on CCV is small compared to the flow field. This novel methodology is very flexible and allows investigation of the sources of CCV which have been difficult to investigate in the past.

This paper empirically studies commonly observed training difficulties of Physics-Informed Neural Networks (PINNs) on dynamical systems. Our results indicate that fixed points which are inherent to these systems play a key role in the optimization of the in PINNs embedded physics loss function. We observe that the loss landscape exhibits local optima that are shaped by the presence of fixed points. We find that these local optima contribute to the complexity of the physics loss optimization which can explain common training difficulties and resulting nonphysical predictions. Under certain settings, e.g., initial conditions close to fixed points or long simulations times, we show that those optima can even become better than that of the desired solution.

This paper presents a method for transferring a prechamber from a single-cylinder engine (SCE) to a rapid compression machine (RCM) as a pre-step for carrying out optical experiments. The overall objective of the transfer procedure is to achieve similar flow field conditions in the prechamber which is mounted on an RCM as compared to the same prechamber mounted on an SCE. Computational fluid dynamics (CFD) simulations of a large-bore natural gas SCE were carried out up to ignition timing which serve as the basis for the transfer procedure. The mass-flow rate in the prechamber neck, pressure difference between the main chamber and prechamber, prechamber pressure and mass in the prechamber were defined as the governing thermodynamic quantities which affect the evolution of the flow field in the prechamber during the compression stroke. All of these quantities in an RCM are largely influenced by operating piston speed and compression ratio for given initial conditions. The RCM operation parameters for each governing quantity were estimated with the help of 0D simulations. CFD simulations of RCM were performed for the determined RCM operation parameters, and the evolution of the velocity field and turbulent quantities in the prechamber were evaluated. The procedure has been applied to two prechambers of different shape and size and a good agreement between the RCM and corresponding SCE CFD simulations was achieved for the flow field conditions in both of the prechambers near the ignition timing. It was found that mass-flow rate in the prechamber neck is the most important quantity which influences the prechamber flow field. Overall this numerical procedure provided a flexible and computationally feasible approach to adjusting RCM operation parameters for transferring a prechamber from the SCE to RCM.

Understanding turbulent premixed flame propagation under engine-relevant conditions is crucial for advancing efficient combustion technologies. This study applies high-fidelity Large Eddy Simulation (LES) to support the design and optimization of a fan-stirred constant volume combustion chamber intended for future experiments on spherically propagating turbulent premixed flames. The design is representative of large-bore engine conditions: moderate to high turbulence intensities with integral length scales significantly larger than the characteristic flame thickness. The numerical analysis focuses on this flow regime by accurately resolving the incompressible, isothermal, transient flow field generated by six rotating fans operating at two different rotation speeds of 2000 and 10,000 rpm, aiming to achieve high turbulence intensities with minimal mean flow in the optical measurement region. The evaluation of the LES-predicted turbulence intensities and the mean velocities indicates that the flow field generated in the targeted central region is statistically stationary and spatially homogeneous. The analysis of the Reynolds stress tensor invariants in the Lumley triangle proves a nearly isotropic turbulence. Moreover, the computed turbulence intensities and integral length scales evaluated with velocity autocorrelation functions are in close agreement with target values representative of large-bore internal combustion engines. These findings confirm the viability of the proposed design for future experimental investigations of flame–turbulence interactions of different fuels like ammonia and ammonia-hydrogen mixtures under engine-like operating conditions.