Preface to the Special Issue on Simulation Technologies and Computational Design

V=~ v-V 3 ~~1fjCJ~~-$rmiddotff~~~It~ttL

Johann Kolar

A1 A~IszligINF (ETH) ~~

The development of power electronics converters is driven by demands for higher efficiency lower functionshyal volume and lower costs However due to continuous progress over the past decades power density and effishyciency barriers will be reached in near future Furthershymore high peak-to- average load ratios extreme opershyating temperatures or reliability specifications have to be considered more and more frequently in the course of converter designs Such multi- objective requirements can no longer be fulfi11ed with classical design methods but only on the basis of a holistic view with a high deshygree of detail Therefore apart from the basic convertshyer functions considered so far special emphasis must be placed on the high frequency characteristics of magnetshyic components EMC (electromagnetic compatibility) filtering interconnection techniques thermal manageshyment of active and passive components and thermoshymechanical properties of power semiconductor packagshyes Due to the high complexity of a simultaneous conshysideration of a11 such aspects the best possible tradeshyoff between different requirements can only be found by means of digital simulation and a final multi- objective optimization In this way apart from the clarification of technical details also a comprehensive comparison of competing concepts can be carried out and hardware prototyping can be transferred to the end of the develshyopment process This allows to significantly cut develshyopment costs and time-to- market

The challenge of a comprehensive description of power electronics converters consists in the acquisition and combination of models of functional elements of difshyferent kinds and in the coverage of extremely wide time length and frequency ranges For example switching events must be represented in the ns range whereas for the analysis of mission profiles hours must be covered Here a conventional approach via coupling of discipline-specific FEM (finite element method)shybased simulation tools would lead to extreme computashytion times and a huge amount of data and therefore certainly is not a feasible path Thus new methods must be found that creatively exploit the heterogeneous structure and different physical dimensions and time

constants of the functional elements of power electronshyics converters in order to arrive at reduced- order modshyels that can be handled by the computer

As an example the geometry of apower semiconshyductor module may be analyzed with FEM- simulations and translated into a reduced-order thermal equivalent circuit that is limited to a few temperatures of intermeshydiate layers This equivalent circuit may be integrated in a natural way into a circuit simulation of the conshyverter structure In a similar way starting from an electromagnetic FEM or PEEC (partial element equivashylent circuit) analysis the internal power module layout may be represented in a simplified equivalent network and the behavior of the power semiconductors could be approximated by behavioral circuit models Fina11y also the electromagnetic interface to the environment could be represented in the same manner In this way eg the losses of the power semiconductors could be simulated for the actual circuit layout and directly coushypled with the thermal model Subsequently the resultshying profiles of the intermediate layer temperatures could be integrated into a lifetime analysis employing physics-based thermo- mechanical models parametershyized via cycling tests

In general a representation based on equivalent circuits allows a simple coupling of models of different disciplines within a subsystem but also a direct conshynection of various subsystem models to an entire conshyverter The research at the Power Electronics Systems Laboratory of ETH Zurich has focused now for over 5 years on the automated generation of reduced-order subsystem models and their connection to overall conshyverter models Here important areas are thermo-elecshytric modeling the extraction of equivalent models of inshyterconnection systems and EMC filter stages including couplings between filter elements by means of PEEC and the physics- based modeling of the lifetime of metshyal- metal interfaces on the basis of accumulated deforshymation energy which leads to excellent agreement with experimental cycling tests The design platform arisshying in this way is complemented by design tools for EMC filters and high frequency magnetics with adshy

150 ( 2)

I

vanced thermal management as weIl as analytical deshyscriptions of advanced cooling systems In parallel an optimization platform is being developed that allows a simultaneous optimization of all converter subsystems where also the physical dimensions of individual comshyponents appear as variables The effectiveness of this research is weIl illustrated in the realization of power electronics converters with extreme power densities or efficiencies such as a non-isolated automotive DC- DC converter with apower density of 30 kWdm3 an isolatshyed DC-DC converter with apower density of 10 kWdm3

and a single- phase PFC (power factor correction) sysshytem with an efficiency of over 99

In future multi- disciplinary simulation tools will gain enormously in importance and will be expanded to virtual prototyping platforms A converter design can then be translated largely automatically into a reshyduced- order model and a complete picture of the loadshying of components at selected operating points or misshysion profiles can be obtained Here one could also imagine the integration of component data basis and cost models or the iterative optimization of component values or operating parameters via higher level optimishyzation routines In the analysis of the subsequently reshyalized hardware multi- disciplinary simulation could advantageously serve for augmenting the display space ie the simulation model could be adapted via calibrashytion routines to the measurement results so that nonshyaccessible physical values such as internal temperashytures or electromagnetic fields could be determined and displayed via parallel simulation A further application could be the analysis of the sensitivity of system perforshymance indices in respect of component characteristics which would enable the derivation of roadmaps for an effective further development of component technoloshygies Finally the transition from virtual prototyping to virtual manufacturing would be possible where quesshytions of eg manufacturability and logistics could be considered in an early stage of the design

In summary the design process of power electronshyics systems which is currently based to a large extent upon human knowledge and experience will be extendshy

ed into a new dimension Virtual design platforms will integrate know- how from various disciplines and beshycome an important strategie factor Here special chalshylenges arise in the training of engineers in the utilizashytion of multi-disciplinary tools and in ensuring the coshyordinated extension and reliable availability of the simshyulation programs However such simulation systems alshyways will only model and optimize existing concepts the creation of fundamentally new concepts creativity will remain the inherent task of the human being the engineer

fOSltYJgt~~ CtplusmniAJ I L Fmi)

1iIt1J ~~~ jl[ 0) ~1J $ - 1iIt1J Woc Ici il $-~-t- if -C je2 lti1t ~ u JiLIqJ1UL IciszligJUiLIL~9 0 -cas -3 j 0 pfi3lt~d1O) ~

mJiBUttt EMI 7 -( )[9 ilfuOCsectlfI1 -~1Y 7 - )0) ~m~~ ~ltO)WM~~~~~~~~~~O)~~M

tt1jtJ L -~ T j Q~ e ~ 0 U I ci TgtfftO)ouml Jl-rl9~1I1-C0

0)~MIcil3lnlL~0 eIszligbn-s o L n~~FJl9 0Cf-Re u TT ) 9 )[Y l l - Y 3 ytJ~f]~ -cas 0 0 u tJ u ~~Ilt

~ i ~ ffl L ~1ft0) 11i c Ici iJt~~ r1 -pT - y it tJtlj je 9 0 ~~ ffr~~Cf-i~~ ffl L ~jJnxy l l - Y 3 ytJQ ~cas0o LnlLcl I) ~9euroJA r-- e~r10)ifrj)tJS1n0o

1iIt~ M~clij1iltm3lt~~0)~~~~~~M~2~

~ T a-E-9 0 L e -C Y AT b~1~E-T)H~9 0 L e tJgJJJIL~0o A-1 A~~If-tje+ (ETH Zurich) O) 7 -r l ~ r-- 0 = Y ~ Y ATb 7pJf1E~ (PES) -C lci 5 if LLtlLb~ -J T 1szligXj[E T )[~ sectjlij]Inx u Y A T b~1

~9euroM~tt~Yll - Y3YO)~1E~ff-JT~-s o ~-

1iIt~0)E-T)[1~ ~Y A T b F1-P EMI 7 -( )[91L ~lt0 ~~ET )[O)trillplusmn t - r-- iJ-1 ~ )[rr)[t-IL cl 0ftf ifri0)~ioFFjl_rl9~ ET)[1~~ c ~1I~~pJf1EM~ e u T L 0 30kWQ 0)3 ~m DC-DC J y ) - 9 lOkWQ O)~g

~w DC-DC J Y- 9~clijk1J$ 99 Rt_O)lILfsect PFC

~~m u L O)pJf1EO)~k1Jt1~~~iE u T L 0 1qS91L lci 1amp~S91L~rt11t~~JW119 0 c1L~M~tt0o

0

fftIcifI~-PJ r) r) IL1amp(f u T L ~~jtMIciXO)fi~-WILi1t eh 1amp~~IWt5 Y r--7t - b1ci~ilJllLb~-sWr1~~0)J

r)r)~Ijj(I)An l-I~ljJGIIj~e~0 o utJu Yl l-Y 3 YY-)[cET)[1~-P~1~1ci c2T u ffruLJ yt -r r-- O)(gIj~IciAO)fjtWIj e u lt 4-1~u1l~-Cas 0 0

151 (3)

0

I

vanced thermal management as weIl as analytical deshyscriptions of advanced cooling systems In parallel an optimization platform is being developed that allows a simultaneous optimization of all converter subsystems where also the physical dimensions of individual comshyponents appear as variables The effectiveness of this research is weIl illustrated in the realization of power electronics converters with extreme power densities or efficiencies such as a non-isolated automotive DC- DC converter with apower density of 30 kWdm3 an isolatshyed DC-DC converter with apower density of 10 kWdm3

and a single- phase PFC (power factor correction) sysshytem with an efficiency of over 99

In future multi- disciplinary simulation tools will gain enormously in importance and will be expanded to virtual prototyping platforms A converter design can then be translated largely automatically into a reshyduced- order model and a complete picture of the loadshying of components at selected operating points or misshysion profiles can be obtained Here one could also imagine the integration of component data basis and cost models or the iterative optimization of component values or operating parameters via higher level optimishyzation routines In the analysis of the subsequently reshyalized hardware multi- disciplinary simulation could advantageously serve for augmenting the display space ie the simulation model could be adapted via calibrashytion routines to the measurement results so that nonshyaccessible physical values such as internal temperashytures or electromagnetic fields could be determined and displayed via parallel simulation A further application could be the analysis of the sensitivity of system perforshymance indices in respect of component characteristics which would enable the derivation of roadmaps for an effective further development of component technoloshygies Finally the transition from virtual prototyping to virtual manufacturing would be possible where quesshytions of eg manufacturability and logistics could be considered in an early stage of the design

In summary the design process of power electronshyics systems which is currently based to a large extent upon human knowledge and experience will be extendshy

ed into a new dimension Virtual design platforms will integrate know- how from various disciplines and beshycome an important strategie factor Here special chalshylenges arise in the training of engineers in the utilizashytion of multi-disciplinary tools and in ensuring the coshyordinated extension and reliable availability of the simshyulation programs However such simulation systems alshyways will only model and optimize existing concepts the creation of fundamentally new concepts creativity will remain the inherent task of the human being the engineer

fOSltYJgt~~ CtplusmniAJ I L Fmi)

1iIt1J ~~~ jl[ 0) ~1J $ - 1iIt1J Woc Ici il $-~-t- if -C je2 lti1t ~ u JiLIqJ1UL IciszligJUiLIL~9 0 -cas -3 j 0 pfi3lt~d1O) ~

mJiBUttt EMI 7 -( )[9 ilfuOCsectlfI1 -~1Y 7 - )0) ~m~~ ~ltO)WM~~~~~~~~~~O)~~M

tt1jtJ L -~ T j Q~ e ~ 0 U I ci TgtfftO)ouml Jl-rl9~1I1-C0

0)~MIcil3lnlL~0 eIszligbn-s o L n~~FJl9 0Cf-Re u TT ) 9 )[Y l l - Y 3 ytJ~f]~ -cas 0 0 u tJ u ~~Ilt

~ i ~ ffl L ~1ft0) 11i c Ici iJt~~ r1 -pT - y it tJtlj je 9 0 ~~ ffr~~Cf-i~~ ffl L ~jJnxy l l - Y 3 ytJQ ~cas0o LnlLcl I) ~9euroJA r-- e~r10)ifrj)tJS1n0o

1iIt~ M~clij1iltm3lt~~0)~~~~~~M~2~

~ T a-E-9 0 L e -C Y AT b~1~E-T)H~9 0 L e tJgJJJIL~0o A-1 A~~If-tje+ (ETH Zurich) O) 7 -r l ~ r-- 0 = Y ~ Y ATb 7pJf1E~ (PES) -C lci 5 if LLtlLb~ -J T 1szligXj[E T )[~ sectjlij]Inx u Y A T b~1

~9euroM~tt~Yll - Y3YO)~1E~ff-JT~-s o ~-

1iIt~0)E-T)[1~ ~Y A T b F1-P EMI 7 -( )[91L ~lt0 ~~ET )[O)trillplusmn t - r-- iJ-1 ~ )[rr)[t-IL cl 0ftf ifri0)~ioFFjl_rl9~ ET)[1~~ c ~1I~~pJf1EM~ e u T L 0 30kWQ 0)3 ~m DC-DC J y ) - 9 lOkWQ O)~g

~w DC-DC J Y- 9~clijk1J$ 99 Rt_O)lILfsect PFC

~~m u L O)pJf1EO)~k1Jt1~~~iE u T L 0 1qS91L lci 1amp~S91L~rt11t~~JW119 0 c1L~M~tt0o

0

fftIcifI~-PJ r) r) IL1amp(f u T L ~~jtMIciXO)fi~-WILi1t eh 1amp~~IWt5 Y r--7t - b1ci~ilJllLb~-sWr1~~0)J

r)r)~Ijj(I)An l-I~ljJGIIj~e~0 o utJu Yl l-Y 3 YY-)[cET)[1~-P~1~1ci c2T u ffruLJ yt -r r-- O)(gIj~IciAO)fjtWIj e u lt 4-1~u1l~-Cas 0 0

151 (3)

0