Automotive EMC : Numerical Simulation For Early EMC Design .

Page 3antenna (radius 3 mm.), connected to the ground planethrough 50 Ω loads. For this ElectroMagnetic Radiation(EMR) configuration, the validation criterion is the Transfer Function between the input voltage applied to theharness and the voltage drop induced on the 50 Ω load atthe bottom of the antenna, the frequencies of interestcovering the FM band, from DC up to 120MHz.For measurement accuracy, the whole experimental testset-up was located inside an anechoic chamber.The coupling methodology consists in building a MTLmodel of the harness (including ground planes), computing the currents generated along the harness path, andmaking these currents radiate as dipoles in the 3D simu lation. The coupling only requires a data format transferand a Fast Fourier Transform (FFT).In the reverse way, for the modeling of the emission froman equipment, the MTL code is used again to generate alumped model of the harness alone, still used in the circuitsimulator as a black-box component, able to represent therealistic signal propagation along the harness when excited by the associated non linear circuit. This has becomea standard feature of SABER-Harness .Figure 4 : Phase delay of current signalalong a CAN bus for MTL / Circuit coupling3 - REALISTIC APPLICATIONSFigure 3 : 3D/MTL versus Measurements for EMRThe figure 3 above shows the experimental results and thesimulated Transfer Function for two different models ofthe metallic connectors : short circuit wires into CRIPTE(simulation 1) and metallic surfaces within the threedimensional model (simulation 2). The agreement is excellent between experimental measurements and simulation results in the whole frequency range whatever themodeling of connectors is.However, this validation already exhibits an importantfeature for mastering the predictivity of numerical EMC :all relevant details (here, the precise modeling of thecontacts) are of importance.The second articulation to be validated stands between theMTL and the Circuit modeling. The two cases of emission and susceptibility led to different coupling approach,and to specific validations. In both cases, the objective ofthe coupling is to associate the harness (linear, with interference effects between wires) and the circuit (0D, butnon linear) in an integrated simulation model.For the susceptibility analysis, the MTL code is used togenerate an "active" (i.e. including a representation of thesources) Z-model of the harness inside the car, whichtakes into account all wires coupling and ground planes.This Z-model is then used in the circuit simulator as ablack-box component, which allows to represent the realistic EMC behavior of the aggressed harness when associated with its non linear terminal loads.Once the basic principles of the multi-level strategy havebeen validated on academic cases, the question of itspractical exploitation in realistic industrial context arises :this chapter will essentially focus on complementarypractical aspects of the methodology, essential from thepoint of view of industrial effectiveness, and which justified the work performed under Auto-EMC to connect allrequired solvers into an efficient integrated tool set.The first step to go through is the acquisition of the geometrical data, usually stored in CAD repositories.Figure 5 : Typical early car geometry definition(Crash Mesh)For 3D simulations, a mesh has to be generated from thegeometry of the car body : when addressing the earlystages of the car design, the 3D geometric modeling hasto face the problem of partial, approximate, and oftentopologically ill defined, geometrical descriptions, which,however, have never been built for EMC analysis.

Page 4To circumvent this difficulty, a specialized pre-processor(PRE-CEM) has been built, to offer convenient CADrepair and cleaning tools. To take also into account therarity of available geometry representations at this earlystage, PRE-CEM can input both CAD models in IGESformat, and pre-meshed models, typically in NASTRANformat.When it comes to the harness modeling, the MTL codeneeds to know the location of the reference structures tobe taken into account to get a realistic modeling of theLine parameters (capacitance and inductance). This led tothe development of a "parser", able to extract automatically partial cuts of the 3D structure acting as groundplanes, at user prescribed locations along the harnesspath. Finally, PRE-CEM is also able to input specificparts of the geometry files as representing wires, antennasor harness mean path : this information will be shared bythe 3D and MTL solvers in coupled simulations.a semi-structured mesh around them, approximating correctly the natural near-field symmetry of revolutionaround the thin wires.The discretization is performed in two steps, allowingindifferent use of FD or FE resolution algorithms : first, asurface mesh is built. In a FD context, this surface mesh isthus intercepted by the structured FD mesh, while in FEcontext, after semi-structured meshing of the emittingwires, the mesh generator will continue to produce acomplete tetrahedral unstructured 3D mesh.Figure 7 : Surface mesh with internal wiring.Figure 6 : Parsing of ground planesfor harness from 3D CAD dataThe next step is mesh generation for 3D simulations. Theproject used a commercial unstructured, automatic meshgenerator (PAM-GEN3D ) currently used for CFD(Computational Fluid Dynamics) applications. Its adaptation to EMC models was limited to the development oftwo specific features, dedicated to the meshing of thinsurfaces and of wires.Thin surfacesEMC models of cars have to deal with the thin metalsheets defining the car body; these metal sheets are physically perfectly conducting, thus avoiding the need tomodel explicitly their thickness. Therefore, for the sake ofCPU efficiency, a specific treatment corresponding toinfinitely thin surfaces was integrated into the advancingfront algorithm of the mesh generator.Wires3D field solvers usually use specific local models aroundthe wires to accurately describe their EM radiation inemission simulations; the numerical counterpart of thesespecific modeling of emitting wires is the requirement forThe last information needed to complete the computational model is the set of physical parameters. A specificdirect interface between CATIA/E3D and SABERHarness has been developed. However, the electric information provided by CAD systems is usually poorly convenient for direct acquisition into a simulation model, andhas to be recast in a convenient way by the user. Theseparameters can be defined for CRIPTE and PAM-CEM using their respective GUIs. Using PRE–CEM and allrelated tools described above, the complete modeling(CAD repair & cleaning, meshing, parsing, definition ofphysical parameters and simulation conditions) of a typical car can be performed within a couple of days.ElectricalComponentEquivalentCircuitGeometr. DataStructureHarnessAntenna2D-ParserCross SectionsTransmissionline analyserTransmission Line ParameterBehaviour ModelsNetworkSimulatorCrosstalk and Distributed Current3D-Field-SolverVoltage at the TransferBase of the AntennaFunctionFigure 8 : Multi-level approach for full car.

Page 5As an application, we shall now briefly describe thesimulation of the radiation of internal harnesses of a realcar towards one receiving antenna located on the backhood of the vehicle. A single prototype cabling (CD audionetwork) was considered, running through the whole carfrom the front end to the back hood, with some additionalparts around the front window (connected to the receivingFM antenna).In this case, the generic multi-level coupling procedurereads as illustrated in Figure 8. The CRIPTE analysis ofthe harness is performed in a way similar to what hasbeen described in chapter 2 above, as is the couplingbetween the equipment and the harness.At the end of this simulation process, the entire electro magnetic environment of the car is available for visualization of radiated fields around the vehicle (as depicted inFigure 9 below, using isolines display in a cutting plane),as well as far field radiation pattern, induced voltages andcurrents on the receiving antenna (leading directly to theevaluation of the Transfer Function), and currents alongthe harness. The relevance of those various outputs,which can be used for building transient movies, dependsof course on the type of problem under investigation(emission, interference, etc ).4 - COMPLEMENTARY INVESTIGATIONSAll the validation tests sketched in the previous chaptersdemonstrated that the predictivity of the proposed approach to numerical EMC enjoys a fairly satisfactorylevel of formal predictivity. But it is of importance tonotice that, for obtaining a precise fitting between numerical and experimental results, the latter had to be obtained in strictly controlled conditions.Indeed, the industrial effectiveness of this formal predictivity is not straightforward, since cars, as real industrialproducts, are subject to a lot of local uncertainties, both atthe level of individual components design (e.g. detailedwiring of bundles) and at the level of system integration(e.g. exact path of harness).Starting from this fact, the AutoEMC project developedalso two specific actions, focusing the issue of making,through numerical simulation, early EMC design decisions which would not be over-sensitive to the uncertaintyoccurring in the real-life process of the product buildingand integration. The first task addressed the problem ofthe sensitivity of EMC behavior of harnesses to the dispersion of their main geometric and physical parameters,and their statistical handling. The second task focused onbest practice rules able to ensure that early numericalEMC design decisions would still hold at the level of thecomplete, deliverable, car.A very simple device, made of a pair of parallel wiresrunning over a ground plane to which they are connected,was used to numerically analyze the dependency ofcrosstalk and transmission properties on geometric parameters (wires diameter, distance to ground) and on wirepermittivity. The main result is that the major factor ofinfluence is the distance to the ground, the other effects,while not totally negligible, decreasing quickly with thisdistance. Using the same device, and allowing the wiresto stand at a wavy distance from the ground, numericalexperiments were performed, using a random-wire modeldeveloped by Politecnico di Torino [6].Figure 9 : 3D field and harness currents.In this example, simulated and experimental TransferFunctions were compared at the level of the back-hoodreceiving antenna. This comparison evidenced the frequent necessity to take into account geometric details. Toobtain a good agreement between simulation and measurements, the slots located above the back hood had to betaken into account (while the initially given mesh definedfor crashworthiness did not represent them).For early EMC design, when the exact geometry is stillnot totally fixed, this requirement clearly brings somepractical limitation to the theoretical predictivity of numerical analysis : an approximate geometric descriptionmay be fairly satisfactory for the bulk car body, whilesome specific parts (slots, shelters, ) have to be precisely known to ensure the relia

Automotive EMC : Numerical Simulation for Early EMC Design of Cars Flavio CANAVERO 1, Jean-Claude KEDZIA 2, Philippe RAVIER 3 and Bernhard SCHOLL 4 1 Politecnico di Torino - Corso Duca degli .