Modeling and Simulation of Powertrains for Electric and Hybrid Vehicles Dr. Marco KLINGLER PSA Peugeot Citroën Vélizy-Villacoublay, FRANCE marco.klingler@mpsa.com FR-AM-5
Background The automotive context Overview of different simulation tools commonly used General simulation approach EMC analysis of a electric powertrain A generic typical electric powertrain EMC problems introduced by powertrains Specific EMC questions to address Modeling and simulating powertrains Components (Un-)shielded equipment near to chassis Chassis of the vehicle Cables and harnesses Interfacing between codes Summary and conclusions Overview 2
The automotive context The electric / electronic architecture (EEA) is developed separately Equipment are developed by suppliers Evolutions of the equipment during the commercial life of a vehicle Harnesses (schematic, routing, composition) are defined by the car manufacturer Several versions of harnesses for a same car model 3
Design of an EEA in the development process Time Car manufacturer Vehicle specifications EEA specifications Car body design Harness design Integration Corrections Validation Equipment suppliers New equipment Corrections EMC questions on the design of the EEA have to be answered even if the I/O loads of the equipment are not fully defined Generally, no physical prototype of the harness is available for measurements Design is achieved through 3D CAD models and electric schematics EMC design can only rely on basic EMC design rules, engineering experience and simulation 4
Representativeness of CAD data for EMC simulations 5
General characteristics of an EEA Harnesses are typically non uniform, composed of assembled cables twisted pairs coaxial cables (very few, mainly for RF applications or HV links) shielding (rarely, costly) Signals and power are transmitted and received by an equipment between 2 wires of a same harness (differential mode) or between 1 wire and the car body (common mode) or both alternatively Differential mode + Bat Zero volt conductor Common mode 6
Overview of different simulation tools commonly used Frequency-domain Time-domain 3D electromagnetic codes 2D magnetostatic or electrostatic codes Specific tools and codes 2D multi-conductor transmission line codes Electric circuit simulation codes 7
General simulation approach Draw a layout of the functional EEA respecting distances, and including ground wires and connections List or include the equipment and the harnesses that may affect the EEA (immunity) or be affected by it (emission) Include the parasitic EMC components Create an interaction matrix Identify for each cell of the matrix the type of coupling modes and evaluate the risk (experience, feed-back, ) For each coupling mode, identify all the components involved Define the goal of the simulation and be sure that the results that will be obtained will lead to a decision and/or a solution Verify that the configurations simulated are acceptable and respect other design rules and requirements (safety, ) List what data is available Introduce assumptions, approximations and simplifications Define what numerical tools can be used Evaluate the time needed (including iterations) to reach the goal Combine and optimize data, tools, methods to succeed before the deadline 8
High voltage battery A generic typical electric powertrain AC / DC converter P1 P2 P3 Electric motor Conducted emissions Switching noise Radiated emissions Switching noise up clock + harmonics Conducted emissions Switching noise Fundamental frequency Harmonics Shielding (equipment and cables), filtering, grounding, and connectivity are to be defined 9
Electrical features having an EMC impact Power, switching and Pulse Width Modulation Higher levels of current Low rise times (typ. 50μs to avoid power losses in heat) Broadband spectrum, mainly < 100 MHz Increase radiated and conducted emission levels Increase crosstalk Increase point to point voltage drops (common mode) Increase magnetic field levels Ground connections Safety requirements Currents should not use them as a return path!!! 10
EMC problems introduced by powertrains Immunity to external sources Normally none on the high voltage part Nothing specific to powertrains on the rest Emission Radio reception (AM, CB, FM, ) Audio systems Systems using low frequency communications (125 khz) Magnetic field sensors (Hall effect and LF) Health protection (Recommendation 1999/519/CE (*) : 6.28 μt between 80 Hz 150 khz) Environment (Directive 2006/28/CE above 30 MHz) (*) Directive 2004/40/CE allows higher levels 11
Origin and reasons of the EMC emission problems From the equipment Waveforms Components PCB layout Filtering Shielding Ground connections Location of the equipment From the harnesses Waveforms Impedances Type of cables Connections Bundling Height, length Routing 12
Pinpointing the risks using an EMC interaction matrix EMISSION High voltage battery Equipment / Sensors / Antennas Cables / Signals DC link Coupling modes AC / DC converter AC link Electric motor Radiated Conducted Crosstalk Common impedance + Transfer impedance 13
Specific EMC questions to address What type of cables are needed? What transfer impedance (or shield effectiveness) is required? Should we group the 3-phase conductors in a single shielded cable? How should the 3-phase conductors be placed between each other? How should the interconnects be done? What should be grounded for EMC reasons? What needs a separate routing? What should be kept outside (beneath) the vehicle? Etc 14
Modeling passive linear components L = 10μΗ ( R = 100mΩ, C = 0.1pF)? dbω C = 470pF or 10nF (R = 80mΩ, L = 0.6nH)? dbω Frequency (MHz) Parasitic elements of passive components can often be neglected 15
Modeling non-linear or active components (diodes, IGBT, MOS) Most of the components have already a model in electric circuit codes, only the parameters have to be correctly chosen The use of a time-domain electric circuit code implicitly requires to import equivalent models of the entire EMC problem 16
Modeling shielded equipment near to vehicle chassis Cable bundle Ground wire Ground connection Vehicle chassis Equipment PCB ground plane Car body ground Harness model PCB Z int C ext Simple impedance and capacitance models can often be used 17
Modeling unshielded equipment near to vehicle chassis Cable bundle Ground wire Ground connection Vehicle chassis Equipment PCB ground plane Harness model PCB Car body ground C Simple capacitance model can often be used 18
100000 10000 Modeling unshielded equipment near to vehicle chassis (cont ) Measurement results Module impédance (Ω) Impedance (Ω) 1000 100 10 1 0.1 Partial ground plane Full ground plane 0 100 200 300 400 500 600 700 800 900 1000 Fréquence Frequency (MHz) 19
100000 Modeling unshielded equipment near to vehicle chassis (cont ) Simulation results 10000 Module impédance (Ω) Impedance (Ω) 1000 100 10 1 0 100 200 300 400 500 600 700 800 900 1000 Fréquence (MHz) Frequency (MHz) 20
Modeling the steel chassis of the vehicle The permeability of the steel chassis of the vehicle must be taken into account at low frequencies The values are not always given (μ r = 500 is used by default) μ r impacts on : Magnetic field distribution and shielding Skin depth Ground / cables global inductance and resistance Current return path EMC analysis requires to take also into account the thickness of the parts Difficulties to find a 3D code that is capable of Taking into account the permeability Taking into account simultaneously the thickness of the steel sheets of the vehicle (0.7 mm) and the size of the vehicle (4 x 1.5 m) MoM or BEM codes can be used with no thickness and μ r = 1 as first approximation 21
Effect of permeability on the magnetic field distribution Cable : Copper σ = 5.88 10 7 Ω -1.m 1 Plate : Steel σ = 5.88 10 6 Ω -1.m 1 T =1 mm R = 5 mm W = 2 m μ r = 1 F = 10 Hz μ r = 500 H = 2 cm Over and under estimations of the magnetic field if μr is not taken into account 22
Example of field distribution analysis Cable routing 23
Example of field distribution analysis Cable routing (cont ) N 1 N 2 N 3 Side Base N 4 N 5 N 6 Shaft Positions N 4 and 5 are the best solutions 24
Example of field distribution analysis Single or dual conductor H = 1 cm 1 conductor H = 1 cm D = 2 cm D = 4 cm H = 2 cm D = 2 cm Below Above Above : Same order or magnitude for 1 or 2 conductors Below : Magnetic fields below are approximately divided by 2 Both : Separation has a stronger impact than height 25
ρ μ r Copper = 1.68 10 = 1 8 Ω.m δ = Skin depth in the chassis ρ π f μ [m] Roughly 7 times less skin depth ρ μ r = = Steel 1.68 10 500 7 Ω.m Frequency Copper (mm) Steel (mm) 10 Hz 20.6 2.9 100 Hz 6.5 0.92 1 khz 2.06 0.29 10 khz 0.65 0.092 100 khz 0.206 0.029 1 MHz 0.065 0.0092 Skin effect starts already above 100 Hz for a thickness of 0.7 mm Requires an even finer mesh within the steel sheets 26
Modeling a single conductor above ground plane I(z,t) L dz R dz I(z+dz,t) V(z,t) C dz G dz V(z+dz,t) dz z ( z, t ) V z I z ( z, t ) = = R G I ( z, t ) V ( z, t ) ( z, t ) I L t V C t ( z, t ) V z I z () z = ( R + j ω L )() I z () z = ( G + j ω C ) V () z Lumped elements can be used if the length is small compared to wavelength (< λ / 10), otherwise TL models are required 27
Modeling the per-unit-length global inductance and resistance Cable : Copper σ = 5.88 10 7 Ω -1.m 1 Plate : Steel σ = 5.88 10 6 Ω -1.m 1 μ r = 500 T =1 mm R = 1 mm H = 2 cm W = 1.4 m 1.2 7.5 Inductance Per-unit-length inductance ( μh) 1.1 1 0.9 0.8 Resistance 7 6.5 6 5.5 Per-unit-length resistance (mω) 0.7 5 10 100 1000 10000 100000 Frequency (Hz) Inductance and resistance varies below a given frequency depending on the height 28
Modeling uniform multi-conductor harnesses above a ground plane L 22 Lossless model example I 2 (t,z) L 12 e C 12 I 1 (t,z) L 11 I 1 (t,z+dz) V 1 (t,z) e C 11 V 1 (t,z+dz) z Lossless time-domain equations z+dz ( ) V 1 I I = L 1 + L 2 I 11 12 z t t 1 e V V V = C 1 e + C 1 2 11 12 z t t V I I 2 = L 1 + L 2 I ( V V ) V 21 22 2 e = C 2 1 e + C z t t 21 22 z t t Lumped elements allows modeling lossy MCTL in time-domain and frequency domain 2 29
30 Modeling uniform multi-conductor harnesses above a ground plane (cont ) + = + = t I L t I L z V t I L t I L z V 2 22 1 21 2 2 12 1 11 1 V 1 (t,z b ) V 1 (t,z a ) I 1 (t,z a ) I 2 (t,z a ) z b z a Frequency-depend lossy MCTL only in frequency domain Lossless time-domain equations + = + = t V C t V C z I t V C t V C z I 2 22 1 21 2 2 12 1 11 1 I 1 (t,z b ) Gray box model
Modeling the per-unit-length parameters of uniform multi-conductor harnesses Per-unit-length inductances and resistances must be obtained using a 2D LF magneto dynamic code Per-unit-length capacitances and conductances can be obtained using a 2D electrostatic code or a multi-conductor transmission line code Per-unit-length inductances cannot be derived from the values of the p-u-l capacitances and the propagation velocity 31
Modeling non-uniform wire bundles and harnesses Power cable Non uniform harness Power cable Uniform Uniform Uniform Uniform [L1],[R1] [L2],[R2] [L3],[R3] [L4],[R4] [C1],[G1] [C2],[G2] [C3],[G3] [C4],[G4] 32
Modeling non-uniform wire bundles and harnesses (cont ) -70 30 different configurations (runs) Common mode current (dba) -80-90 -100-110 4 m harness network 15 conductors 2 cm segmentation 0 50 100 150 200 250 300 350 400 Frequency (MHz) One configuration is usually sufficient for low frequencies 33
Modeling transfer impedance (and admittance) of shielded cables Ic Vc Vp Ip Transfer impedance Transfer admittance V c = Z z c I c V p = Z z p I p Z t I p Z t I c I c = Y + z c V c I p = + Y z p V p Y t V p Y t V c 34
V c = Z z c I c I c = + Y z c V c Modeling transfer impedance (and admittance) of shielded cables Split into to sub-models by introducing current-controlled voltage sources and voltage-controlled current sources V p = Z z p I p I p = + Y z p V p Z t I p Y t V p Z t I c Y t V c V c V p I c I p Z c Z c Z I t Y V t Z I t Y V t p c p c Δz Y c Y c 35
Transfer impedance of some coaxial cables The transfer impedance is generally constant below 200 khz 36
Modeling the current return path in the vehicle chassis L = 3 m R = 1 mm 17.5 cm H = 2 cm 1 V 17.5 cm W = 1.4 m 37
Modeling the current return path in the vehicle chassis (cont ) Cable : σ = 5.88 10 7 Ω -1.m 1 Plate : σ = 5.88 10 6 Ω -1.m 1 μ r = 1 No thickness (BEM) 10 Hz 100 Hz 1 khz 10 khz 100 khz Currents concentrate under the cable as frequency increases 38
Modeling the current return path in the vehicle chassis (cont ) 17.5 cm 2 m R = 1 mm H = 2 cm 17.5 cm 1 V W = 1.4 m 17.5 cm 39
Modeling the current return path in the vehicle chassis (cont ) Cable : σ = 5.88 10 7 Ω -1.m 1 Plate : σ = 5.88 10 6 Ω -1.m 1 μ r = 1 No thickness (BEM) 10 Hz 100 Hz 1 khz 10 khz 100 khz Low frequency limit of the transmission line theory depends mainly on the height and ground impedance 40
Compacting in the frequency-domain for interfacing with other codes In the frequency-domain, an entire 1D, 2D or 3D linear model can be compressed into a single admittance coupling matrix Y, obtained by short-circuiting all the terminals of the model and applying successively a voltage at each terminal V i I i Linear system I n I 1 V n I1... In = Y... Yn 11 1......... Y Y 1 n... nn V1... Vn V j I j V 1 I = Y V Often requires special techniques (ex : vector fitting) when imported into time-domain codes 41
The easy aspects Most of the EMC problems are in low frequencies Electromagnetic quasi-static behaviors Simplifications are possible The difficulties Summary and conclusions Presence of non-linear components (diodes, MOS, IGBT, ) which require time-domain models and simulations PWM with very short rise- and fall-times compared to long functional periods which can lead to consequent simulation durations Skin effect which is obtained in the frequency-domain is difficult to model in time-domain Converting frequency-domain models for time-domain code requires special techniques Permeability should be taken into account and requires then to mesh the thickness of the chassis of the vehicle Many major parameters are unknown (characteristics of MOS and IGBTs, permeability of the steel, transfer impedances of shielded cables, ) Requires several different codes and interfacing between them 42