Overview of Circuit Simulation Programs ECE 546 DIGITAL COMPUTATIONAL TECHNIQUES FOR ELECTRONIC...

Overview of Circuit Simulation Programs

ECE 546 DIGITAL COMPUTATIONAL TECHNIQUES FOR ELECTRONIC CIRCUITS

January 8, 2008

Oleg Wasynczuk

Need for System-of-Subsystems Approach

Complex engineered systems such as aircraft, modern automobiles, or the terrestrial electric power grid involve a broad spectrum of technologies and interactive subsystems that must work synergistically in order to operate properly

Inter-dependencies between subsystems are becoming more and more prominent

More-Electric Aircraft Power System

Modeling Approaches

Synchronous Machine Subsystem Models

Distributed Parameter Coupled Circuit Steady State

)( uxfx

,dtd

IZEeV j ~~

jEe

jXRZ

I~

V~

Power Electronic Subsystem ModelsAverage ValueDetailed

)( uxfx

,dtd

)()(

)()( );(

iiiif

if

iiiiii

sgst

ttsdtd

,,

,

1

110

x

Txxxfx

Simulation Approaches

Circuit-Based Approaches (Spice, EMTP, Saber, PSIM, Simplorer)

System-Based Approaches (Simulink, ACSL, Dymola) Block-diagram and/or differential equation oriented Extensive set of tool boxes including

ASMG (Simulink, ACSL) Power System Blockset (Simulink) …

Finite-Element-Based Approaches (Ansys, Maxwell, …)

Circuit-Based Approaches

Circuit-Based Approaches

Example Subsystem

(Motor Controller)

Circuit-Based Approaches

Circuit-Based Approaches

Resistor-Companion Circuit

15

2

1

654

321321

9

8

7

k

SS

S

S

v

v

v

ggg

gggggggg

i

i

i

i

i

Update Formula

O(n3) computational complexity where n = number of non-datum nodes

Circuit-Based Approaches

Simulation Approaches

Circuit-Based Approaches (Spice, Saber, PSIM, Simplorer)

System-Based Approaches (Simulink, ACSL, Dymola) Block-diagram and/or differential equation oriented Extensive set of tool boxes including

ASMG (Simlink, ACSL) Power System Blockset (Simulink) …

Finite-Element-Based Approaches (Ansys, Maxwell, …)

System-Based Approaches

Hierarchical system definition

System-Based Approaches

Common Simulink Component Models

System-Based Approaches

System-Based Approaches

1

1

1

0

1p

i

ikiki

p

i

iki

k th ,xfxx

When user starts model, Simulink applies selected integration algorithm to approximate solution at discrete but not necessarily uniform instants of time

General Multi-step Formula

Implicit algorithms require solution of nonlinear equation (dimension = number of states) at each time step. Newton-Raphson iteration generally used.

Explicit if 01

System-Based Approaches

Stiff System: A system with both fast and slow dynamics

Stiffly Stable Integration Algorithm: the ability to increase the time step after fast transients subside

Stiffly Stable Algorithms are implicit!

System-Based Approaches

Computational Complexity

System-Based Approaches

Dilemma

Simulink Algorithms

Shampine and Reichelt, The MATLAB ODE Suite, SIAM J. Sci. Comput.,Vol. 18, No. 1, pp. 1-22, January 1997.

System-Based Approaches

Simulation Approaches

Circuit-Based Approaches (Spice, Saber, PSIM, Simplorer)

System-Based Approaches (Simulink, ACSL, Dymola) Block-diagram and/or differential equation oriented Extensive set of tool boxes including

ASMG (Simulink, ACSL) Power System Blockset (Simulink) …

Finite-Element-Based Approaches (Ansys, Maxwell, …)

Finite-Element Based Approaches

4000-10000 Nodes

uSaa

M dtd

FEA

Conventional Parallel Computing Paradigm

Conventional Parallel Computing Paradigm

Conventional Parallel Computing Paradigm

Conventional Parallel Computing Paradigm

At best m-fold reduction in computation time assuming zero communication latency

Computational gain further bounded by Amdahl’s Law

p

T)α1(αTT 1

1 p where serial portion 1,0α

therefore

α

1S

)α1(α

1

T

TS 1

pp

p

Distributed Heterogeneous Simulation (DHS)

DHS Definition

Synchronized interconnection of any number of dynamic subsystem simulations

Developed using any combination of programs/languages

Implemented on:• Single computer/workstation/supercomputer• Local area network (Intranet)• Wide area network (Internet)

Sample DHS Computer Setup

DHS Concept

Much better than M-fold (potentially M3) improvement in speed

DHS Links Environment

Flexibility of DHS

Heterogeneous platforms (Windows, Unix, Linux, ...) Heterogeneous languages (ACSL, MATLAB/Simulink,

Saber, EASY5, C, C++, FORTRAN, Java,…) Heterogeneous simulation approaches (single-rate, multi-

rate, state model based, resistor-companion, finite difference/element,...)

Heterogeneous networks (Ethernet, SCI, ScramnetTM, MyrinetTM,...)

Use “best” language for each component/subsystem Proprietary information protected Super-linear increase in computational speed across a network

of desktop computers No need to translate models into common language Legacy code can be used directly Conducive to team design/analysis Remote interconnection Eliminate need to develop average-value models for system

stability assessment Real-time (hardware-in-the-loop) capability for some systems System Integrator(s) do not have to be familiar with the

language(s) used to create subsystem simulation(s)

Key Advantages of DHS

More-Electric Aircraft Power System

Optimum Allocation

Optimum Allocation

Optimum Allocation

18.5 speedup with 4 computers