Physical Modelling with Simscape™

Rick Hyde

1
© 2013 The MathWorks, Inc.
Outline
 Part 1: Introduction to Simscape
– Review approaches to modelling
– Overview of Simscape-based libraries
– Introduction to physical networks
– Simscape tutorial: DC motor
 Part 2: Application examples
– PMSM and drive
– Landing gear extension/retraction
– Engine cooling system
2
 Designing physical systems
 Multi-domain
– Mechanical
R 6 5
– Electrical
– Hydraulic/pneumatic
– Thermal 2 43 1

 Modelling environment
– Multi-domain
– Mix with algorithms/control
– Multiple levels of fidelity
– Optimization tools
– Code generation
3
Data-driven versus physics-based modelling
Data-driven Physics-based
 Data sources  Mathematical paradigms
– Experimental data – ODEs & DAEs
– External tool (complexity- – State charts
reduction exercise) – Physical networks
 Complex behaviour with many  Behaviour is adequately
unknowns approximated by limited set of
 Examples equations
– Combustion engine  Examples
– Aerodynamics – Algorithms/control
– Gears & clutches

E ( x,  ) x
  f ( x, u,  )
4
Extending Simulink® using Simscape™
Simulink Simscape extension
 Equation set  Equation set
– x  f e ( x, u,  ) – E ( x,  ) x  f ( x, u,  )
– Explicit equation – Implicit equation
 Relevance  Relevance
– Single body motion – 1-D multi-body systems e.g.
– Multiple-body motion when drivelines
there is compliance – Electrical networks
– Most algorithms (control) – Hydraulic/pneumatic networks

s1 s2
s3
5
What does this model represent?

6
What does this model represent?

7
 Modelling an electrical circuit in Simulink
Step 1: figure out the equations
Step 2: build the model

8
Modelling an electrical circuit in Simscape

Build the model

9
Modelling an electrical circuit in Simscape

Component equations
from library blocks

Additional constraints
from circuit topology
11
Algebraic loops in Simscape

Simscape converts entire network to equations and solves

them simultaneously, so it intrinsically solves
algebraic loops.

12
Signal-based modelling methods
pose challenges
 Very flexible, but can be difficult
to use efficiently
 Requires expertise in
several areas
– Physics, math, programming
 Deriving system level equations
is difficult and error prone
 Resulting model can be
difficult to read and maintain
13
Physical modelling methods
ideal for plant models
 Build accurate models quickly
– System-level equations
derived automatically
 Model is easier to read
– Reflects structure of system
 Easier to update model
– New technologies or designs
can be easily incorporated
into the model
14
 SimPowerSystems

SimMechanics

SimHydraulics
Simscape libraries

Simscape

SimDriveline
MATLAB, Simulink

15 SimElectronics
Extensive Component
Libraries
 SimElectronics > 90
component models
– Actuators, drivers
– Sensors
– Semiconductors
– Integrated circuits

 Models look like

schematics
– Easy to read
and interpret 16
Generate Code
s1 s2
 Use Simulink Coder to u s3

convert models into C code Controller Plant

– Share with other users
(Model Reference) s1 s2 .DLL
u s3
(Simulink)
– Protect intellectual property
 Model Reference Protected Mode
 Alternative: ssc_protect
s1 s2

– Run hardware-in-the-loop
u s3

simulations
17
Outline
 Part 1: Introduction to Simscape
– Review approaches to modelling
– Overview of Simscape-based libraries
– Introduction to physical networks
– Simscape tutorial: DC motor
 Part 2: Application examples
– PMSM and drive
– Landing gear extension/retraction
– Engine cooling system
18
 Through & across variables

q p3
p1 p2 p1 p2

p4

 Abstract to a physical network

 All nodes have the same pressure (across variable)
 Sum of flows (through variables) at a node is zero
 Each component must specify an equation involving the through
and/or across variables at its boundary
19
Through & across variables by domain
Domain Through Across
Electrical Current Voltage
Hydraulic Flow rate Pressure
Rotational mechanical Torque Angular speed

Translational mechanical Force Translational speed

Pneumatic Mass flow Pressure

Heat flow Temperature
Magnetic Magnetic flux Magneto motive force
Thermal Heat flow Temperature
20
 Components
Domain (across, through) Resistance Inductance Capacitance
Electrical (v, i) Resistor Inductor Capacitor
Rotational mechanical ?

Translational mechanical

Hydraulic

• Resistance (R) dissipates power.

• R=voltage/current (ratio of across to through)

21
 Components
Domain Resistance Inductance Capacitance
Electrical (v, i) Resistor Inductor Capacitor
Rotational mechanical (w, trq) Rotational damper

Translational mechanical

Hydraulic

22
 Components
Domain Resistance Inductance Capacitance
Electrical (v, i) Resistor Inductor Capacitor
Rotational mechanical (w, trq) Rotational damper

Translational mechanical (v, F) Translational

damper
Hydraulic

23
 Components
Domain Resistance Inductance Capacitance
Electrical (v, i) Resistor Inductor Capacitor
Rotational mechanical (w, trq) Rotational damper

Translational mechanical (v, F) Translational damper

Hydraulic (p, Q) ?

24
 Components
Domain Resistance Inductance Capacitance
Electrical (V, i) Resistor Inductor Capacitor
Rotational mechanical (w, trq) Rotational damper

Translational mechanical (v, F) Translational damper

Hydraulic (p, Q) Hydraulic orifice

p=kQn

25
 Components
Domain Resistance Inductance Capacitance
Electrical (v, i) Resistor Inductor Capacitor
Rotational mechanical (w, trq) Rotational damper ?

Translational mechanical (v, F) Translational damper

Hydraulic (p, Q) Hydraulic orifice

p=kQn

• Inductance keeps through variable flowing (current)

26
 Components
Domain Resistance Inductance Capacitance
Electrical (v, i) Resistor Inductor Capacitor
Rotational mechanical (w, trq) Rotational damper Rotational Spring

Translational mechanical (v, F) Translational damper

Hydraulic (p, Q) Hydraulic orifice

p=kQn

• Inductance keeps through variable flowing (current)

27
 Components
Domain Resistance Inductance Capacitance
Electrical (v, i) Resistor Inductor Capacitor
Rotational mechanical (w, trq) Rotational damper Rotational Spring ?

Translational mechanical (v, F) Translational damper Translational spring

Hydraulic (p, Q) Hydraulic orifice Accumulator

p=kQn

• Capacitor maintains across variable (voltage)

28
 Components
Domain Resistance Inductance Capacitance
Electrical (v, i) Resistor Inductor Capacitor
Rotational mechanical (w, trq) Rotational damper Rotational Spring Inertia

Translational mechanical (v, F) Translational damper Translational spring Mass

Hydraulic (p, Q) Hydraulic orifice Accumulator Fluid inertia

p=kQn

• Capacitor maintains across variable (voltage)

29
The inerter

• Invented by Prof Malcolm Smith, Cambridge University.

• Force is proportional to relative acceleration.

30
Outline
 Part 1: Introduction to Simscape
– Review approaches to modelling
– Overview of Simscape-based libraries
– Introduction to physical networks
– Simscape tutorial: DC motor
 Part 2: Application examples
– PMSM and drive
– Landing gear extension/retraction
– Engine cooling system
31
DC Motor: Modelling options
1. Use SimElectronics DC Motor
block
2. Build an equivalent circuit
using Simscape foundation
library
3. Define a custom component
using Simscape language
– Equation-based
– Composite
32
SimElectronics® pre-built component

33
SimElectronics DC motor parameterization options
 Parameterized according
to datasheet standards
 Multiple methods for
assigning parameters
 Description,
functionality and
formulation provided
in Help dialog

34
SimElectronics pre-built component example

35
Equivalent circuit based on Simscape foundation
library components

36
Equation-based Simscape language component
1. Define equations
2. Define ports/connections
3. Define variables 𝑑𝑖
𝑣 = 𝑅𝑖 + 𝐿 + 𝑘𝜔
4. Define parameters 𝑑𝑡
5. Write block help
6. Add icon [optional]
7. Publish

37
Equation-based Simscape language example

38
Equation-based Simscape language component

𝑑𝑖
𝑣 = 𝑅𝑖 + 𝐿 + 𝑘𝜔
𝑑𝑡
𝜏 = 𝑘𝑖
39
Composite Simscape language component

 Connect components
in Simscape language file
– Refer to existing components
– Overwrite parameter values if necessary

– Connect components

40
Outline
 Part 1: Introduction to Simscape
– Review approaches to modelling
– Overview of Simscape-based libraries
– Introduction to physical networks
– Simscape tutorial: DC motor
 Part 2: Application examples
– PMSM and drive
– Landing gear extension/retraction
– Engine cooling system
41
PMSM and drive modelling
 Components
– PMSM
– Switching devices (IGBTs)
– Controller
 Tools
– SimPowerSystems
 System-level to support circuit and controller design
– SimElectronics
 Detailed switching device models – fine-tuning and losses prediction

42
 IGBT modelling
I-V characteristic Dynamics

Complexity
 Piecewise linear  No dynamics
 Lookup table  Behavioural
 Physics-based  Constant capacitance
 Composite/complex  Nonlinear charge model

Complexity:
 Slower simulations, detailed parameterization required
 Higher-fidelity/better predictions
43
IGBT modelling  SimPowerSystems
– Piecewise linear
– No dynamics or constant capacitance
 SimElectronics

Complexity
– Nonlinear, physics-based, simple
– Datasheet parameterization
– Easily tuneable by optimization
 SPICE
– Composite from complex physics-based
components
– Manufacturer NETLIST
44
pmsm_drive

45
System-level equivalent of PMSM and drive
 Example pmsm_drive suitable for
– Component design
– Electrical system design
 What about the system designer e.g. for a hybrid
vehicle?
– Model the whole system and predict efficiency/performance
– Multiple drive cycles
– HIL test

46
Abstracted PMSM drive
 Energy-based
– Mechanical power out = power in minus losses
– Losses (physics-based analysis)
 Copper losses proportional to the torque
 Iron losses dependent on voltage and switching frequency
– Losses (empirical approach)
 Tabulate as function of torque and speed (and possibly DC volts)
 Table derived from detailed PMSM and drive model

 Assume torque is closed-loop controlled

47
Abstracted PMSM drive

48
sm_landing_gear

49
 Landing Gear Design
Model:

Problem: Determine the force

required of the hydraulic cylinder
to meet system specifications
Solution: Use Simulink,
SimMechanics, SimHydraulics, and
SimElectronics to design the system 50
Modeling Thermal Fluid Systems
Thermal Liquid Library
 New domain and library
for thermal liquid systems
– Single-phase liquids
– Fluid properties vary
with temperature

 Applications
– Heat exchangers
– Pipelines
51
Engine cooling example

52
 Challenges in Thermal Fluid Simulation Fluid
Volume

 Restrictions on block combinations Pneumatic Domain

– Some combinations require
fluid volumes at nodes
– Blocks sharing differential variables
makes initialization difficult Volume usually required at node

 Poor behavior at regime changes,

Tube 1 Tube 2 Tube 3
low flow rates, and flow reversals Flow Direction

– Zero crossing difficulties, solver issues Properties in Tube 2

depend on flow direction
53
Addressing challenges
with full-flux method
 Each element contains
a volume of fluid
– No fluid volumes at nodes

 Control volume approach Heat Mech.

Fluid Volume
used in each element
BoundaryA BoundaryB
– Internal nodes defined Thermal Fluid Thermal Fluid
Internal Node
– Network nodes have
no differential variables
Node Temp.
Temperature Convection
 Power flux includes dominates
Temp. A Conduction
convection+conduction dominates
– Smooth temperature variation Temp. B mdot
of node temperature at zero flow 54
Simscape language best practice
 Build incrementally
 Write test scripts/harnesses
 Avoid discontinuities
 Use appropriate level of fidelity
 Write out equations for custom blocks before building

55
Conclusions
 Simscape extends Simulink to
– Support a network approach
– Create models that topologically match the physical system
 Create component models by
– Using ready-made blocks from MathWorks’ libraries
– Constructing composite components from Simscape
foundation library blocks
– Writing custom equations in Simscape language
 All the benefits of Simulink apply e.g.
– Code generation
– Tight integration with MATLAB® 56
LINKS
Physical modelling central web-page:
http://www.mathworks.co.uk/physical-modeling/

Essentials of physical modelling:

http://www.mathworks.co.uk/help/physmod/simscape/ug/essential-physical-
modeling-techniques.html

57
 Viewing Simscape Simulations Results
ssc_explore
 Explore simulation results
from entire physical network
– Select multiple signals
– Overlay or separate plots
– Arrange plots
– Extract plot to separate window
 Spend more time analyzing,
less time simulating
 Download from
MATLAB Central
http://www.mathworks.com/matlabcentral/fileexchange/28184-simscape-simulation-results-explorer
58
 Zero-Crossing Statistics
 Log zero-crossing statistics
for Simscape networks
– Shows when ZCs occur
– Can help indicate location
of simulation bottlenecks

59
Local Solvers Improve Variable step, implicit (ODE15s)
Simulation Performance Controller
Physical
Network
Physical
 Use fixed-step, implicit solvers Network
locally on Simscape networks
– Use implicit solvers only
Fixed step, explicit (ODE1)
where necessary
Local fixed step
Fixed-step, implicit (ODE14x)
 Configure solver per implicit solver #1
Physical
Network
physical network Controller
Physical
– Run different physical networks Network
at different sample rates Local fixed step
 Electrical (higher), mechanical (lower) implicit solver #2

 Primary benefit is to speed-up

simulations where fixed-step solvers Generated
Code
are required, like HIL 60