Search Results

In the present study, ice nucleation in sessile water drops during continuous cool down is studied experimentally under the impact of a constant electric field, to determine its influence on heterogeneous nucleation. The experimental setup enables simultaneous observation of multiple drops under well-defined conditions with and without an electric field and at temperatures down to -40 °C. A single experimental run contains 40 drops exposed to the same conditions. Drops with a well-defined size are produced employing a drop-on-demand drop generator. Based on multiple experimental runs using the same drops, the nucleation behavior is analyzed using statistical methods to determine the drop survival curves and nucleation site densities for varying conditions. Besides the influence of the electric field, the influence of different drop ensembles is investigated for a constant cooling rate of 5 K/min.

Accretion and shedding of ice layers is a serious problem for various engineering applications. In particular, ice layers growing due to ice crystal impingement on warm parts of an aircraft jet engine pose a severe hazard since they seriously affect safe operation of an aircraft. The material properties, and in the first place the strength of an ice layer, are crucial for the mechanisms leading to, and taking place during, both accretion and shedding of an ice layer. In the present study, the apparent yield strength of dry granular ice layers is examined employing a novel rigid-body-penetration approach. Dynamic projectile penetration into granular ice layers of varying porosity and ice grain size is experimentally investigated for different projectile impact velocities using a high-speed video system and post-processing of the captured video data.

Icing is a major hazard for aviation safety. Over the last decades an additional risk has been identified when flying in clouds with high concentrations of ice-crystals where ice accretion may occur on warm parts of the engine core, resulting in engine incidents such as loss of engine thrust, strong vibrations, blade damage, or even the inability to restart engines. Performing physical engine tests in icing wind tunnels is extremely challenging, therefore, the need for numerical simulation tools able to accurately predict ICI (Ice Crystal Icing) is urgent and paramount for the aeronautics industry, especially regarding the development of new generation engines (UHBR = Ultra High Bypass Ratio, CROR = Counter rotating Open Rotor, ATP = Advanced Turboprop) for which analysis methods largely based on previous engines experience may be less and less applicable. The European research project MUSIC-haic has been conceived to fill this gap and has started in September 2018.

Aerodynamic properties of a BMW car model taking over a truck model are studied computationally by applying the scale-resolving PANS (Partially-averaged Navier-Stokes) approach. Both vehicles represent down-scaled (1:2.5), geometrically-similar models of realistic vehicle configurations for which on-road measurements have been performed by Schrefl (2008). The operating conditions of the modelled ‘on-road’ overtaking maneuver are determined by applying the dynamic similarity concept in terms of Reynolds number consistency. The simulated vehicle configuration constitutes of a non-moving truck model and a car model moving against the air flow, the velocity of which corresponds to the car velocity.

Premature and uncontrolled flame initiation, called pre-ignition (PI), is a prominent issue in the development of spark-ignited engines. It is commonly assumed that this abnormal combustion mode hinders progress in engine downsizing, thus inhibiting development of more efficient engines. The phenomenon is primarily observed in highly turbocharged spark ignited (SI) engines in the full load regime at low engine speeds. Subsequent engine knock induces extremely high peak pressures, potentially causing severe engine damage. The mechanisms leading to this phenomenon are not completely understood; however, it is quite plausible that a multiphase process is responsible for the pre-ignition. One effect could be the interaction between injected fuel drops and the oil film on the cylinder liner. Under certain conditions, droplets of oil or oil/fuel mixture can detach or splash from the film, leading to pre-ignition at the droplet surface towards the end of the compression phase.

Some widely-used scale-resolving turbulence models are comparatively assessed in simulating the aerodynamic behavior of a full-scale AUDI-A1 car configuration. The presently considered hybrid RANS/LES (RANS – Reynolds-Averaged Navier-Stokes; LES – Large-Eddy Simulation) models include the well-known DDES (Delayed Detached-Eddy Simulation) scheme and two further variable-resolution formulations denoted by PANS (Partially-Averaged Navier-Stokes; Basara, 2011) and VLES (Very LES; Chang et al., 2014). Whereas the DDES method represents the originally proposed formulation based on the one-equation Spalart-Almaras model (Spalart et al. 2006), whose RANS/LES interface position is directly correlated to the underlying grid resolution, the other two models represent ‘true’ seamless formulations, providing a smooth transition from Unsteady RANS to LES in terms of a dynamic “resolution parameter” variation.

The present work deals with a computational study of a ‘DrivAer’ car model, the rear-end shape of which corresponds to the Notchback configuration (Heft et al. [1] and Heft [2]). The study investigates the effects of the underbody geometry and wheel rotation on the aerodynamic performance. The configurations with detailed and smooth underbody as well as with stationary and rotating wheels are considered. The computational model applied relies on a VLES (Very Large Eddy Simulation) formulation, Chang et al. [3]. The residual turbulence related to the VLES framework is presently modelled by a RANS-based (Reynolds-Averaged Navier-Stokes), four-equation (D(k,ɛ,ζ, f)/Dt) near-wall eddy-viscosity model, Hanjalic et al. [4].

Airframe icing caused by supercooled large droplets (SLD) has been identified as a severe hazard in aviation. This study presents an investigation of impact of a supercooled drop onto superhydrophobic and partially wettable substrates. Drop impact, spreading and rebound were observed using a high-speed video system. The maximum spreading diameter of an impacting drop on partially wettable surfaces was measured. The temperature effect on this parameter was only minor for a wide range of the drop and substrate temperatures. However solidification hindered receding when both the drop and substrate temperatures were below 0°C. The minimum receding diameter and the speed of ice accretion on the substrate were measured for various wall and drop temperatures. The two parameters increased almost linearly with the decrease of the wall temperature, but eventually leveled off beyond a certain substrate temperature.

Numerical experiments have been presently conducted aiming at studying the influence of the surface energy on the crystallization process of supercooled water in terms of the supercooling degrees. The mathematical model consists primarily of the equation governing the thermal energy field solved independently in both phases in accordance with the two-scalar approach by utilizing the Stefan condition at the interface to couple both temperature fields. The computational algorithm relying on the level-set method for solid-liquid interface capturing has been appropriately upgraded aiming at accuracy level increase with respect to the discretization of the thermal energy equation and the normal-to-interface derivative of the temperature field. The model describes the freezing mechanism under supercooled conditions, relying on the physical and mathematical description of the two-phase moving-boundary approach.

The aerodynamic properties of a BMW car model, representing a 40%-scaled model of a relevant car configuration, are studied computationally by means of the Unsteady RANS (Reynolds-Averaged Navier-Stokes) and Hybrid RANS/LES (Large-Eddy Simulation) approaches. The reference database (geometry, operating parameters and surface pressure distribution) are adopted from an experimental investigation carried out in the wind tunnel of the BMW Group in Munich (Schrefl, 2008). The present computational study focuses on validation of some recently developed turbulence models for unsteady flow computations in conjunction with the universal wall treatment combining integration up to the wall and high Reynolds number wall functions in such complex flow situations. The turbulence model adopted in both Unsteady RANS and PANS (Partially-Averaged Navier Stokes) frameworks is the four-equation ζ − f formulation of Hanjalic et al. (2004) based on the Elliptic Relaxation Concept (Durbin, 1991).

Recent aircraft incidents and accidents have highlighted the existence of icing cloud characteristics beyond the actual certification envelope defined by the JAR/FAR Appendix C, which accounts for an icing envelope comprising water droplets up to a diameter of 50 μm. The main concern is the presence of SLD (Supercooled Large Droplets), with droplet diameters well beyond 50 microns. In a previous European-funded project, EURICE, in-flight icing conditions and theoretical studies were performed to demonstrate the existence of SLD and to help characterize SLD clouds. Within the EXTICE project the problem of SLD simulation is addressed with both numerical and experimental tools is being addressed. In this paper the objectives and main achievements of the EXTICE project will be described.

In the present work we investigated experimentally and computationally the unsteady flow around a BMW car model including wheels*. This simulation yields mean flow and turbulence fields, enabling the study aerodynamic coefficients (drag and lift coefficients, three-dimensional/spatial wall-pressure distribution) as well as some unsteady flow phenomena in the car wake (analysis of the vortex shedding frequency). Comparisons with experimental findings are presented. The computational approach used is based on solving the complete transient Reynolds-Averaged Navier-Stokes (TRANS) equations. Special attention is devoted to turbulence modelling and the near-wall treatment of turbulence. The flow calculations were performed using a robust, eddy-viscosity-based ζ - ƒ turbulence model in the framework of the elliptic relaxation concept and in conjunction with the universal wall treatment, combining integration up to the wall and wall functions.

The potential of single-point turbulence closure models for predicting the flow aerodynamics and turbulence in internal combustion engines (IC) was investigated by computational study of idealized valveless piston/cylinder configurations. The main flow cases considered are the swirling flow in a single stroke rapid compression machine (RCM) with flat and bowl-shaped cylinder head, as well as cyclic compression. Although still remote from a real engine, these configurations enable to analyse joint effects of major phenomena governing the aerodynamics in IC engines: shear, separation, swirl and compression/expansion. Prior to the computation of these engine-like flows, an extensive validation of applied turbulence models was performed in homogeneous and wall bounded shear flows, each featuring separately rotation, swirl and mean flow compression effects.

The accuracy of laser Doppler and phase Doppler measurements in a very dense spray can be affected by the fact that the signals are very noisy and thus the achievable data rate is very low. Indeed, in some cases only a small fraction of the drops passing through the measurement volume will be detected and validated and furthermore, those drops which are validated may also be non-representative of the total drop population, for instance larger drops may be preferentially validated because of their higher scattering intensity. To systematically investigate some of the influencing parameters in such situations, an experimental set-up comprising a common rail Diesel system, a single hole nozzle injector and a commercial Dual-PD system (Dantec) has been assembled. The injection duration was purposely kept very long, up to 8 milliseconds, to better analyze the main injection period.