A novel CFD approach for an improved prediction of

AVL Advanced Simulation Technologies International User Conference 2015 Abstract

A novel CFD approach for an improved prediction of particulate emissions in GDI engines by considering the varying piston surface

temperature Authors: Fabian Köpple, Paul Jochmann, Alexander Hettinger, Andreas Kufferath

Company/ University, Department: Robert Bosch GmbH

Keywords: GDI, Spray-Wall-Interaction, Wall film, Particulate Emission, Piston Surface Temperature

Abstract: The emission of particulate matter from future gasoline direct-injection (GDI) engines has to be optimized, to comply with more stringent emission standards such as EU6. Therefore, the mechanisms responsible for the formation of particles have to be analyzed in detail. From earlier investigations it is well-known that the deposition of liquid fuel wall films in the combustion chamber is a significant source of particle formation in GDI engines. Capturing the detailed dynamics of the deposited liquid fuel is therefore a key feature for the correct prediction of soot generated in a GDI engine.

In a previous study particularly the wall surface temperature and the temperature drop due to the interaction with liquid fuel spray were identified as important parameters influencing the spray-wall interaction and thus the wall film formation. Nevertheless, in conventional CFD models, the cooling of wall surfaces due to interaction with liquid fuel spray cannot be captured because of an assumed constant wall temperature over the whole engine cycle. In this case, the calculated wall film mass is considerably too small. For this reason, the previously developed approach of calculating heat conduction in thin walls was extended to moving meshes, enabling the CFD calculation to consider the temperature drop due to the spray-cooling in engine calculations.

In this work the complete CFD approach, incorporating a detailed description of the entire underlying model chain, starting with the internal nozzle flow and the spray modeling, the modeling of spray-wall interaction and the wall film formation and evaporation, as well as the modeling of combustion and emissions will be described in detail. Additionally the validation of the aforementioned novel CFD approach with surface temperature measurements on the piston of a fired, optically accessible, single-cylinder engine will be shown.

In summary, this study will reveal that with this improved and validated CFD approach the impacts of different parameters influencing the mixture formation and combustion in GDI engines, as e.g. the start of injection or the rail pressure, on the particle emissions can be evaluated. Thus the prediction of particulate emissions by numerical simulation in a GDI engine could be improved considerably, gainfully supporting the combustion system development of GDI engines.


Knocking Optimization in Gasoline Engine by Means of CFD Method

Authors Marco Poli

Company/ University, Department: Piaggio

Authors Andrej Poredos, Primoz Gorensek

Company/ University, Department: AVL AST d.o.o.

Keywords: PFI engine, CFD Fire, Knocking, Combustion

Abstract: Spark ignited engines are today operated more and more under high load conditions. However, the increased loading capabilities and more and more compact design, the need for efficiency in terms of fuel consumption and reducing specific CO2 emissions also increases. Since the gasoline engine operation is inherently limited by knocking at high loads, strategies must be defined to efficiently identify and simulate knocking appearance.

The current study presents optimization of the existing engine design in terms of reducing abnormal combustion behaviour by means of CFD method only. Baseline and partially optimized port fuel injection engine designs were investigated in details by comprehensive CFD study including fuel injection, evaporation, wall interaction, wall-film, ignition and combustion. Different spark timings were investigated for both engine designs on knocking appearance. Simulation results were compared to the experimental results thus, simulation model was well validated. Optimization work was done by CFD simulation method and new engine design was defined based on simulation results only. Simulation of optimized design was repeated and final design version was validated on the test bed. Despite the emphasis is put on the auto-ignition effects, it is explained how CFD simulation can support engine development process in all combustion related aspects.


Thermal Analysis of a 4 Cylinder GDI Engine

Authors Cristiano Pecollo

Company/ University, Department: Fiat Chrysler Autombiles

Authors Andrej Poredos, Simon Urbas


Keywords: GDI engine, CFD Fire, Thermal analysis, Cooling, Combustion

Abstract: Significant effort in developing today’s gasoline engines is being spent to improve the power performance and fuel economy. One of the most important fields related to this objective is heat transfer analysis. From the heat transfer perspective, it is of interest to reduce the heat losses in the engine in an attempt to achieve higher mechanical work output.

The present study uses a novel approach of using multi-material heat transfer analysis for the prediction of temperature distribution in the solid structure of four-cylinder GDI engine. This is an alternative method to the conventional fluid-solid (CFD-FE) coupling, which is usually used for temperature distribution in solid structure and stress analysis. This approach represents the simulation of the heat transfer within the cylinder head structure and its parts by considering the solid and fluid parts of the engine as a multi-domain. Heat exchange is determined depending on the engine operating conditions by considering the complete engine cycle on the gas side. This includes induction, compression and the expansion stroke. With this method the three domains approach is reduced to only two CFD domains where the finite element (FE) tool is no longer required. The CFD tool is used instead and the simulation workflow is significantly simplified. A nucleate boiling effect on water side as well as contact resistance effect in solid part are considered. The simulation workflow is shown as well as the structure temperature distribution as a result. The simulation results are compared to the experimental results (thermo-couples temperatures).


Fast Optimization of Spray Pattern and Injection Strategies in GDI Engine by using CFD simulation

Authors C. Forte1, A. Siliato2, G.M. Bianchi1

Company: 1University of Bologna, 2NAIS srl

Keywords: CFD Simulation, Spray Pattern, GDI engine

Abstract: Next generation of GDI engines design to fulfill upcoming regulations on CO2 and, in general, on gaseous engine out emissions will require the development of advanced combustion systems based on tailored mixture distribution targets with a unique match between injection and intake flow.

Injector spray pattern and injection strategies are thus mandatory goals which are faced with a huge complexity to deal with. Multiple design options and multiple final targets would result in unaffordable Design of Experiment matrix for both experimental testing at bench and CFD simulation approaches. In order to combine the need of optimized spray pattern design with engine targets and the need to shorten time-to-results, a fast and robust multi-objective optimization technique has been developed in order to drive engineer toward the best solution in a short time.

The method is based on a three steps methodology whose core is the so called Spray Design Tool which allows to run a CFD simulation of multiple combination of spray patterns, then providing the best solution in term of spray pattern, Injection pressure, Injection phase and pulsing according to the give targets in terms of mixture distribution wall impingement and in-cylinder turbulence.

An application on a GDI series engine system will be presented.


CFD Simulation of Flow Field inside the Wankel Rotary Engine between Intake and Compression Stroke

Authors (please underline main author): Thanapol Poojitganont, Heinz Peter Berg

Company/ University, Department: Brandenburg University of Technology Cottbus - Senftenberg, Chair for Combustion Engines and Flight Propulsion

Keywords: Wankel rotary engine, Flow simulation, CFD simulation

Abstract: The rotary engine can be treated and evaluated with respect to performance characteristics as a displacement type internal combustion engine of four-stoke, one-cycle similar to the reciprocating engine. For any combustion engines, to reach the maximum power output, the mixture formation inside the engine is one factor, which should be considered. The spray characteristic from the injector and the flow field inside the combustion chamber are two main important components which involve in the mixture formation mechanism. In this research the AVL-FIRE has been used as a simulation tool. The continuing calculation from the starting rotor position at 150 deg BTDC to the ending rotor position at 30 deg ATDC has been developed together with the mesh moving function. The numerical simulation from the intake stroke to the end of the compression stroke has been successively performed to investigate the flow field inside the engine which is governed by many parameters, for example, engine RPM, recess shape and position, intake pressure, etc.

Finally the results from this research can be used as a design tool, for example, for recess shape and position design, etc. In addition together with the spray investigation, the mixture formation mechanism could be comprehensible. It could be eventually used as a tool for improving the Wankel rotary engine to get the appropriate combustion and also the maximum power output as well.

Reduction of aerodynamic drag of vehicles using flow control and

AVL FIRE

S. Krajnović1, G. Minelli1, M. Mirzaei1 and J. Östh 1Department of Applied Mechanics, Chalmers University of Technology

Keywords: vehicle aerodynamics, LES, PANS, URANS, flow control

Most vehicle manufacturers have research programs aiming to reduce vehicle’s aerodynamic drag. In addition to aerodynamic shape optimization (styling), two new techniques have received great attention in recent years: flow control and platooning. The advantage of flow control is that greater freedom in vehicle styling can be obtained. Furthermore, ones the aerodynamic shape optimization has reached its limitations due e.g. loading or manufacturing constraints, flow control can help further reduction of aerodynamic drag.

The aim of this paper is to give a brief overview of the research performed at Chalmers in the research group led by the first author. This research is performed using time-dependent numerical simulations with the AVL FIRE with the aim to develop flow control techniques. Both passive and active flow control techniques are studied. Add–on devices such as impinging devices used in [1] were found to be efficient in drag reduction of so called Ahmed body. A row of short infinite cylinders was used as impinging devices in [1] for formation of streaks in the streamwise direction leading to the separation delay. Another example of passive flow control is presented in [2] where rear flaps forming a wake cavity were studied using Partially-Aveaged Navier-Stokes (PANS) simulations. Although around 10% of drag reduction was obtained using such simple solutions, the passive flow control devices can per definition not adapt to the change in flow conditions (such as change in vehicle speed, overtaking maneuvers or crosswinds). Active flow control was studied using large-eddy simulations (LES) or PANS in [3-10]. The active actuation strategies are aiming to influence either boundary layers at the front of the vehicle or shear layers formed behind the vehicle. Moving surfaces in form of rotating cylinders were used in [3] with the aim to prevent/delay separation of the boundary layer formed on a simple bluff body. Suction of the flow at the tilted A-pillars of a simplified vehicle was applied with the aim to control formation of the trailing vortices in [4]. This study showed that the trailing vortices can be removed using continuous suction at a price of high energy requirement for the actuation.

Most of the investigations of the active flow control aim to manipulate shear layers formed by separation of the boundary layers at the rear of the vehicle. This has been done either using continuous blowing or suction [5] or periodic blowing and suction [6-10]. Also here, application of synthetic jets for periodic blowing and suction is better choice than constant blowing or suction from the point of energy efficiency. Results presented in [6] showed that flow control using periodic blowing and suction can be efficiently studied using LES technique although at rather low Reynolds number. LES in this work was used to study the flow actuation and explain how it works. There are of course no guaranties that the actuation strategies optimized for low Reynolds number flow is efficient at the high Reynolds number. Therefore, the same flow was studied using PANS in [7] and it was shown that the PANS predicts the flow control accurately at least at the low experimental Reynolds number. The remaining step is to show that the PANS techniques works also at the high Reynolds numbers of real vehicles.

mailto:[email protected]

Our most recent work is aiming to influence the separation at the A-pillar of trucks by using synthetic jets. The change of the target for the actuation from the rear of the truck to the front part is a result of the fact that truck manufacturers normally produce only the truck itself and not the trailer. Another drag reduction strategy is so called platooning where vehicles are placed one after another at relatively short distances. The result of such placement is change in the flow field of the individual vehicles leading normally to their drag reduction. First results of the LES of platooning were presented in [8]. REFERENCES

1. S. Krajnovic, LES investigation of passive flow control around an Ahmed body, Journal of Fluids Engineering, 136(12), doi: 10.1115/1.4027221, 2014. 2. J. Östh and S. Krajnovic, ”Simulations of flow around a simplified train model with a drag reducing device” Proceedings from Conference on Modelling fluid flow CMFF’12 (edited by J. Vad), September 4-7, 2012, Budapest 3. X. Han and S. Krajnovic, Large Eddy Simulation of Flow Control Around a Cube Subjected to Momentum Injection. Flow Turbulence and Combustion, 92 (1-2) pp. 527-542, 2014. 4. S. Krajnovic, J. Östh and B. Basara ”LES study of breakdown control of A-pillar vortex”, International Journal of Flow Control, Vol. 2, No. 4, December 2010. 5. S. Krajnovic, J. Östh and B. Basara, ”LES of active flow control around an Ahmed body with active flow control. ” In Proceedings of the Conference on Modelling Fluid Flow (CMFF’09), The 14th International Conference on Fluid Flow Technologies, Budapest, Hungary, September 9-12, 2009. 6. S. Krajnovic and J. Fernandes ”Numerical simulation of the flow around a simplified vehicle model with active flow control”, International Journal of Heat and Fluid Flow, 32 (1), pages 192-200, 2011. 7. X. Han and S. Krajnovic, Study of active flow control for a simplified vehicle model using the PANS method, International Journal of Heat and Fluid Flow, Vol. 42, p. 139-150, 2013. 8. D. Uystepruyst and S. Krajnovic, LES of the row around several cuboids in a row, International Journal of Heat and Fluid Flow, Vol. 44, p. 414-424, 2013. 9. G. Minelli, S. Krajnović, “Numerical Investigation of Active Flow Control of Boundary Layer Separation on a Bluff Body”, Int. Conf. on Jets, Wakes and Separeted Flows, KTH Mechanics, Stockholm, Sweden, 2015. 10. G. Minelli, S Krajnović , “Actuation of the Flow Field Around a Frontstep with a Rounded Leading Edge”, Turbulence, Heat and Mass Transfer 8, Sarajevo, Bosnia and Herzegovina, 2015. 11. M. Mirzaei and S. Krajnovic, “Large-Eddy Simulations of flow around two generic vehicles in a platoon”, Int. Conf. on Jets, Wakes and Separeted Flows, KTH Mechanics, Stockholm, Sweden, 2015. 12. M. Mirzaei and S. Krajnovic, “Numerical study of aerodynamic interactions in a homogeneous multi-vehicle formation”, Int. Conf. on Jets, Wakes and Separeted Flows, KTH Mechanics, Stockholm, Sweden, 2015. 13. J. Östh, B. Noack, S. Krajnovic, D. Barros and J. Boree, On the need for a nonlinear subscale turbulence term in POD models as exemplified for a high-Reynolds-number flow over an Ahmed body. Journal of Fluid Mechanics, 747 pp. 518-544, 2014.

Detailed Heat Transfer Modelling & Optimisation of

a Forced-Convection Kiln using CFD methods

H. Maier1*, W. Berger2 and R. Breyner2 1 gridlab GmbH

2voestalpine Special Wire GmbH

Keywords: wire, patenting, kiln, heat transfer, flow conditions, CFD

Abstract. The efficiency of many industrial processes is strongly governed by the 3D-flow

conditions in the fluid domain. An important step in the production of special wires is the patenting

in forced convection kilns. An unfavourable deviation in the distribution of key parameters (e.g.

velocity and temperature) can be found in the material properties of the product later.

The usage of CFD methods (Computational Fluid Dynamics) is an alternative capable approach to

analyse the 3D-heat transfer process in the kiln in detail with less effort & costs. In the current

improvement study a detailed CFD-modelling of two different forced convection kilns was done for

a typical load case. Considering the detailed kiln geometries and important heat transport

phenomena the initial situation in the heat transfer zone was analysed for each kiln.

In a second step an improvement concept was developed by increasing the heat transfer into the

wire as well as smoothing the flow distribution in the heating zone of the kiln. After revamping of

the kilns a number of different improvements could be realized. Despite of the plant’s higher

flexibility at different load cases a significant improvement of the wire’s product quality could be

found. With respect to important key parameters the deviation of material properties could be

reduced to >90%. In a next step a solid-fluid-coupling will be done in the CFD-approach which

allows examining the developing temperature profile of each wire line in the kiln.

http://dx.doi.org/10.4028/www.scientific.net/KEM.651-653.1525

http://www.ttp.net


Motorcycle Heat Transfer Analysis

Authors Daniele Suzzi

Company/ University, Department: qpunkt GmbH

Authors Simon Urbas


Keywords: Motorcycle, CFD Fire, Thermal analysis, Exhaust Cooling

Abstract: In the development of the sport motorcycle an important factor is relation between good aerodynamic and sufficient cooling of all relevant parts (radiators, engine, and exhaust system). These two demands are in trade-off relationship what requires balancing and compromises. The heat transfer rate for specific component depends upon the velocity of the vehicle, geometry of related parts and the ambient temperature.

The present paper shows the cooling of a sport motorcycle with the driver where the focus is set to exhaust system cooling. The whole exhaust system was considered as multi-material setup with the different materials considering also heat transfer contact resistance. Two different scenarios were considered – a driving mode and a standing mode. In both cases the flow around the exhaust system is significantly different what provides different heat transfer rates. Both situations are important scenarios for sufficient cooling of exhaust system which is one of thermally most exposed parts and positioned usually very close to the driver.

Enhancements of a high-performance CFD-DEM coupled code towards heat and mass transfer in pharmaceutical application

Georg Scharrer1, Charles Radeke2, Dalibor Jajcevic3

1 CATRA GmbH

2 Research Center Pharmaceutical Engineering GmbH3 SES-Tec OG

Abstract

Production processes in pharmaceutical industry were investigated and optimized utilizing

experimental approaches and statistical methods for decades. Since simulation methods

and models, as well as computer hardware had vast development and enhancement in

recent years, there is a lot of interest in applying established simulation techniques from

other industries to pharmaceutical production processes and single unit operations.

Especially Computational Fluid Dynamics (CFD) coupled with the Discrete Element Method

(DEM) introduced by Cundall and Strack 1979 is able to model and simulate processes that

deal with particles/powder/granules/pellets in a fluid phase (gaseous or liquid). Even

tablet movement in coating processes is possible to simulate utilizing a glued sphere

approach (rigid body coupling) or a direct shape approach.

Current DEM software capabilities are limited to a few hundreds of thousands of single

particles. Also CFD-DEM coupling is established in quite a few simulation codes now.

According to the above mentioned processes in Pharma industry it is necessary to model

and simulate processes with fluid-particle interaction taking into account at least tens of

millions of particles.

To overcome this issue, RCPE developed a Graphic Processing Unit (GPU) based DEM

software prototype called eXtended Particle Simulation (XPS), which is able to currently

deal with up to 70 million particles. This software prototype was coupled to AVL FIRE ® via

AVL Code Coupling Interface (ACCI), see Jajcevic et al. 2013, with respect to necessary drag

and lift force conditions which are defined in Beetstra 2007 and Koch and Hill 2001.

Recently heat transfer models were also integrated in XPS and furthermore fully coupled to

AVL FIRE ® via ACCI. First simulations show proper and valid results and the work to

attach mass transfer models to the coupling algorithms is ongoing.

This talk provides brief information on the pharmaceutical requirements of the described

CFD-DEM coupling and shows how the coupling is established for drag force and heat

transfer. Furthermore validation examples are shown to proof, that the coupling and the

methodologies developed are correct. As application of the established XPS-AVL FIRE ®

coupling, modeling and simulation of fluidized bed processes, drying of particle based

polymer pastes and the optimization approach for a tablet pan coater including spray and

drying behavior is shown.

Finally an outlook is given to show further plans in development of the established coupling

and availability of the entire software bundle.


GPU solvers and beyond in applications

Authors: DI Dr. Gundolf Haase, DI Dr. Manfred Liebmann, DI Aurel Neic, DI Dr. Gernot Plank

Company/ University, Department: University of Graz, Institute for Mathematics and Scientific Computing

Keywords: GPU programming, multigrid solver, non-linear solvers, cardiovascular simulations

Abstract: Simulations always demand faster hardware and fast algorithms for solving the problems of interest. A quite common acceleration method is an MPI parallelization based on some domain decomposition approach that utilizes many CPU cores. This sort of parallelization is no longer applicable for accelerator cards as GPUs or Intel's Xeon Phi.

A virtual heard simulation combines cardioelectrical phenomina described via the bidomain equations (a system of elliptic and parabolic PDEs together with an ODE system) with the mechanical behaviour of the stimulated heart muscles (large deformations). Both systems have non-linear, anisotropic materials and non-linear coupling terms between the two phenomena. We use Krylov iteration schemes with algebraic multigrid preconditioning for solving the resulting sparse systems of linear equations in the most inner non-linear loop. The use of GPUs accelerated these computation by a factor of nine for solving a single system of equations. In order to achieve also a good speedup for the overall non-linear solver the calculation as well as the accumulation of the finite element matrix entries has to be performed on the accelerator card too. We present our way combining flexibility with performance gains for the overall non-linear iteration on clusters of GPUs.

While using CUDA for programming GPUs requires an additional implementation of the algorithms several new approaches as OpenACC and OpenMP 4.0 give hope for maintaining only one code with additional pragma directives but acceptable performance on GPU/Xeon Phi. We will present first results when using these specialized pragmas. Additionally, some promising third party code generation tools for ODEs, stencil computations and matrix generation that target a wide range of hardware will be mentioned.


Technical Adaptivity for cloud-enabled CAE Solutions with AVL Fire: Requirements, Concepts and Application in Engineering

Authors (please underline main author): Alexander Heine

Company/ University, Department: CPU 24/7 GmbH

Keywords: High Performance Computing, HPC, CAE, cloud, HPC On Demand, data security

Abstract:

The whole field of outsourcing High Performance Computing (HPC) resources or “HPC-in-the-cloud” still meets with considerable skepticism and their usage is still correspondingly limited. This is not surprising given the sensitivity of data that can have a decisive effect on competitiveness. Potential users in the B2B sector are inhibited in particular by various issues and concerns. But these concerns can be handled or even removed if stakeholders understand what finally characterises a CAE cloud from a technical point of view and what happens with processed data. This presentation describes the progress in HPC On Demand business solutions for cloud-enabled CAE applications. More precisely, it covers their requirements, the basic cloud concepts and their state-of-the-art utilisation in engineering. Cloud-enabled CAE solutions allow users renting high performance computing hardware along with licenses to compute resource-intensive applications in order to increase feasibility, accuracy and performance as well as efficiency of CAE simulations. Important criteria are always the underlying cloud infrastructure, scalability, bottlenecks (data transfer, network interconnection, licenses, etc.), security policies and offered support models. The following technical issues will be addressed in detail:

• Data security – Where is the data located and what security measures are being taken? • Performance – What performance can be expected when running CAE simulation on

an HPC cloud infrastructure? What about their scalability? • Access – How can data be transferred? How are compute resources accessed and

applications started? Is it possible to monitor a running calculation or to even analyse calculations results in the cloud?

• Support – What about licenses, consulting and customising? What service and support can be expected?

Examples with AVL Fire serve as evidence of all these aspects. Related benchmarks using CPU 24/7 dedicated bare metal servers indicate reasonable performance and scalability. Besides, through the cooperation between ISV and resource provider an AVL optimised ready-to-run


HPC platform has been created - adapted to the AVL user’s demands, with optimised hardware configurations including comprehensive support.


Leveraging Cloud HPC for AVL Fire Simulations

Authors (please underline main author): I. Graedel

Company/ University, Department: Rescale

Keywords: FIRE, CFD, Cloud,

Abstract: Rescale provides a cloud simulation platform designed to accelerate CAE simulations by providing instant, scalable high performance computing resources. Engineers and scientists can customize globally-located HPC hardware and optimize computing resources to fit every AVL Fire simulation. Through a strategic partnership, AVL and Rescale offer on-demand hardware and software licenses securely through Rescale's elastic, dynamic platform. Learn how to run AVL Fire simulations on Rescale's platform that was designed for enterprises and is backed by industry-leading security practices.

Documents

A novel CFD approach for an improved prediction of