Experimental and cfd technology for preventive reduction of diesel engine emissions caused by cavitation erosion (PREVERO)

An important project objective was to develop a multidimensional model for cavitation erosion. It was therefore necessary to develop statistical description of the bubble impact loads responsible for cavitation erosion. The procedure involved modelling of the material response taking into account its microstructure, computation of the erosion rate for the materials tested in WP1 and comparison with the measured mass-loss in order to validate the model. Finally, a CFD model was derived to simulate transient flows of bubbles that are generated at low-pressure regions and eventually collapse near the solid surface using the multi-fluid model. Predicting the probability for cavitation erosion was the main goal. In order to quantify the bubble collapses accurately the cavitation model implemented in the FIRE code had to be improved. Eulerian Multifluid Method was adopted to model the flow. In the scope of multi-fluid concept for multiphase flows, each phase is viewed as a continuous phase coexisting, in a statistical sense, with other fluid phases in time and space. Further it is assumed that the transport equations, derived from the conservation laws of mass, momentum and energy, are valid for each phase as they are for singlephase flows. However, it is necessary to apply an averaging procedure to smooth out otherwise intractable interfaces between the phases and to formulate the equations on the macroscopic basis which is solvable numerically. The averaging process introduces inevitably new variables and terms associated with the status of the multiphase mixture and the interactions between the phases. In addition, a bubble population balance equation is added to the family of conservation equations to account for bubble population evolutionary characteristics. These new terms have their respective physical implications and must be modelled, or closed, on the physical basis. Therefore, solution for multiphase flow problems ultimately entails the multiplication of the number of variables and governing equations and sub-models for phase-interactions.

Erosion cavitation is observed in many system, but the physic of the phenomena is not well understood. Therefore we have investigated what happen at the surface and in the materials when subjected to cavitation. Different materials in varying erosion cavitation conditions have been characterized using complementary techniques (optical microscopy, SEM, TEM, RX, Surface roughness) providing new evidences for the erosion fundamental understanding. The influence on mass loss and mass loss rate of initial specimens roughness has been studied. Specific deformation features near the eroded surface have been investigated. Phase and structural transformations have been observed after erosion. Finally, eroded surfaces have been observed to determine fracture mechanisms. The material undergo low cycle fatigue and is eroded when the implemented strain reach a critical value. Specimens were also characterized in terms of hardness, and hardness evolution to provide data for modeling.

CRF has performed an experimental campaign related to the code validation regarding the prediction of cavitation erosion in the A-throttle of the diffusor valve of the injector assembled in multijet diesel engine. The activity was organised as laboratory simulation test on hydraulic test rig. The experimental results were verified using the CFD code FIRE, the results were in good agreement. AVL used the experimental results of TU Graz, LFDT and LEGI to validate the advanced cavitation model and the erosion model. Compared were the velocity profiles, turbulence quantities, phase distributions and material removal rates. Good agreement between simulations was found.

The modelling of material response is a fundamental step in the prediction procedure. Impact loads are classified according to their amplitude with respect to material yield strength and ultimate strength. For the present application, mean impact load lies between both limits. Hence, the material surface is progressively hardened by successive impacts. The work hardening process was characterized by LTPCM from microhardness measurements on cross sections of eroded samples. A major parameter of the model is the thickness of the hardened layer. Using the impact load model, a relationship was derived between pit depth and impact load. It was systematically used to analyse pitting tests and determine the amplitude of the hydrodynamic load (typically in MPa) responsible for each pit. The distribution of impact loads is considered as the signature of the cavitating flow. In practice, the information on cavitating flow aggressiveness was reduced to three integral parameters: pitting rate, mean diameter and mean amplitude of impact loads. This basic description of flow aggressiveness was used to estimate mass loss. The erosion model developed in the framework of the PREVERO project allowed us to compute incubation time and mean depth of penetration rate MDPR. An equation has been derived to predict each of them as a function of flow aggressiveness and material properties measured by LTPCM. The model points out a characteristic time and a characteristic length for cavitation erosion. The characteristic time is the covering time i.e. the time required for the material surface to be entirely covered by impacts without overlapping. As for the characteristic length, it is the thickness of the hardened layer. The erosion rate MDPR under steady state conditions (measured typically in µm/h) is scaled by the ratio of this characteristic length to this characteristic time, with a multiplicative factor, which depends mainly upon the average amplitude of impact loads. The incubation time is proportional to the covering time with a coefficient which also depends upon load amplitude and which tends to unity when mean load approaches material ultimate strength. The values predicted by the model proved to be in satisfactory agreement with the experimental ones obtained from mass loss tests.

In the framework of the PREVERO project, LFDT was in charge of providing experimental and theoretical support for the physical analysis and modeling of cavitation process in a confined geometry. The goal was achieved by slowing down the cavitation process and expanding the bubbles in a specially designed vacuum chamber in order to be able to obtain experimental data on meso- (bubble clouds) as well as on micro-scale (single bubbles). The experimental results showed strong evolutionary characteristics of cavitating water flow which served for implementation of FIRE code which is now capable to cope with multiscale nature of cavitation process. Two principle experiments were carried out to study cavitation in a confined geometry similar to valve cavitation. The first one has been designed to study a single bubble induced cavitation, while the second experiment enabled to study massive cavitation in the slot region. Here, the following flow regimes were observed and analysed in terms of structural function, void fraction and bubble number density: so called detachment region where bubble breakup was observed in case of gas cavitation, large scale cavitation region where macroscopic bubbles formed clusters and bubble collapse region where individual bubbles collapsed due to the subcooled conditions. The experimental data served also as a benchmark validation of bulk liquid turbulence CFD model.

CFD simulation studies in the first stage included basic cavitating flow calculations on the geometry from the University of Ljubljana to test the basic capability of the applied CFD tool AVL FIRE. The results were compared to the measured data. The geometries investigated at other partners' (LEGI, TU Graz, BOSCH) have been investigated in preliminary studies. In all of the simulations it was important to explore the limits of the code to define the software requirements of the new numerical cavitation model, developed and implemented into FIRE later on by AVL. Cavitation phenomena itself was predicted successfully; the limit was the quantification of the collapsing bubbles which are the driving force of the erosion process. An advanced cavitation model was implemented in the FIRE code in order to quantify the number of collapsing bubbles, which cause erosion damage. An advanced turbulence model (k-zeta-f turb. model previously available in FIRE for single-phase applications) was adopted for cavitating flows and validated against experimental results obtained by University of Ljubljana LFDT. The experimental results on A-throttle geometry were verified by CRF using the CFD code FIRE, the results were in good agreement. BOSCH used 2 CFD codes to validate their own experimental results. Physical flow chracteristics were identified. It can be concluded that distribution and the position of the predicted and the measured volume fraction was in acceptable agreement. Shear layer induced cavitation behind the throttle is difficult or even impossible to catch with RANS based two-phase models. Small variations of the experimental geometry lead to significant changes of the flow field and can not be represented in the simulation model.

The objectives of the presented research were experimental measurements of the effects in high pressure cavitating diesel flows similar to those in real diesel engines with common rail injectors. Accurate experimental information about the local flow parameters, cavitation onset and surface erosion is necessary for further development of the numerical models for prediction of cavitation and consequent erosive effects.

In the framework of the PREVERO project, LEGI was in charge of providing experimental and theoretical support for the physical analysis and modelling of cavitation erosion. A new experimental facility has been built to support the development of the erosion model. The objective was to provide basic experimental data for the validation of the modelling from both hydrodynamic and material viewpoints. The erosion model allows the prediction of the incubation time and the erosion rate (MDPR). AVL implemented the erosion model into the FIRE program. Erosion model validation was performed by AVL with the LEGI geometry. Numerical results showed that cavitation is expected to occur on the place where cavitation region ends. This was in agreement with the findings reported by LEGI, TU Graz and BOSCH. The maximum velocity around the cavitation reported by LEGI was 65 m/s, calculated velocity around the vapor region was 65 - 75m/s. The size of the cavitation region was fluctuating, which was observed in experiments as well. The size of the cavitation region corresponds to the erosion location, as reported previously. The erosion is predicted in the bubble collapse region in agreement with the measurement as well. The steady-state material removel rates reported by LEGI were 3-5e-10m/s, AVL FIRE predicted 3.33e-10m/s material removal.

AVL implemeted the erosion model provided by LEGI into the FIRE program. Erosion model validation was performed by AVL with the LEGI geometry. Numerical results showed that cavitation is expected to occur on the place where cavitation region ends. This was in agreement with the findings reported by LEGI, TU Graz and BOSCH. The maximum velocity around the cavitation reported by LEGI was 65 m/s, calculated velocity around the vapour region was 65 - 75 m/s. The size of the cavitation region was fluctuating, which was observed in experiments as well. The size of the cavitation region corresponds to the erosion location, as reported previously. The erosion is predicted in the bubble collapse region in agreement with the measurement as well. The steady-state material removal rates reported by LEGI were 3-5e-10m/s, AVL FIRE predicted 3.33e-10m/s material removal.

Leistungen

Diese Seite teilen

Herunterladen