Post on 10-Mar-2020
Model Based Design andSystem Simulation withSimulationX and Tool Integration with optiSLang
Dr. Andreas Uhlig
Uwe Grätz
ITI GmbH
© 2012 ITI GmbH | www.itisim.com
2Introduction
9th Annual Weimar Optimization and Stochastic DaysNovember 29–30, 2012
Who we are ⋅⋅⋅⋅ Where we areout ITI
• Multi-faceted high-technology
• Leader in virtual system engineering
• Headquarters located in Dresden downtown
Berlin
Dresden
Munich
Frankfurt
© 2012 ITI GmbH | www.itisim.com
3
9th Annual Weimar Optimization and Stochastic DaysNovember 29–30, 2012
Core Business
Introduction
• Offering complete solutions for system modeling, simulation, analysis and testing
• development of simulation software
• engineering services
• product distribution
• product integration
© 2012 ITI GmbH | www.itisim.com
4
9th Annual Weimar Optimization and Stochastic DaysNovember 29–30, 2012
Content
Content
• Model Based Design
• Equation Based Modeling
• Design of a Membrane Cylinder –an Example for Sensitivity Analysis
• Interface to optiSLang
• Summary and Outlook
© 2012 ITI GmbH | www.itisim.com
5
9th Annual Weimar Optimization and Stochastic DaysNovember 29–30, 2012
System Simulation / Model Based Design
• New challenges in systems engineering
• Time, cost,
• Quality, safety,
• Energy efficiency
• Dependency of subsystems
• Early assessment of designs
• Virtual prototyping - of system!
• Use models and simulations
• Re-use models
Model Based Design
© 2012 ITI GmbH | www.itisim.com
6
9th Annual Weimar Optimization and Stochastic DaysNovember 29–30, 2012
Requirements for Model Based System Design
• Multidomain modeling
• Hierarchical modeling
• Replaceable models
• One model for multiple analyses
Model Based Design
© 2012 ITI GmbH | www.itisim.com
7
9th Annual Weimar Optimization and Stochastic DaysNovember 29–30, 2012
Multidomain Modeling (SimulationX Libraries)
Model Based Design | Requirements | Multidomain Modeling
Signal Blocks
Mechanics
Powertrain
Electro-
Mechanics
Magnetics
PneumaticsHydraulics
Thermo
© 2012 ITI GmbH | www.itisim.com
8
9th Annual Weimar Optimization and Stochastic DaysNovember 29–30, 2012
Multi-body Systems
Signal Blocks, ControlsStatecharts
Network Elements
Domain Specific Workspaces for the User
Model Based Design | Requirements | Multidomain Modeling
© 2012 ITI GmbH | www.itisim.com
9
9th Annual Weimar Optimization and Stochastic DaysNovember 29–30, 2012
Level 1 – Basic Model
Model Based Design | Requirements | Hierarchical Modeling
• Simple rigid powertrain model including main components• Analysis of feasibility and general physical relations
Load
EngineGearbox
WheelCar
DrivingResistance
Tire-Road Contact
Gearbox Output Shaft
Flywheel
1D2D
3D
© 2012 ITI GmbH | www.itisim.com
10
9th Annual Weimar Optimization and Stochastic DaysNovember 29–30, 2012
• Case studies
• Proof of concept simulations
• Detailing of interesting elements
• Regard of elastic forces
Driving Resistance
Gearbox Output Shaft
Flywheel
Engine
Car
• DoF: < 10
• Parameter: 20…50
• Frequency: < 100 Hz
Gearbox
1D2D
3D
• Simple engine model
• Merge elements in compounds
Level 2 – Detailed Model
Model Based Design | Requirements | Hierarchical Modeling
© 2012 ITI GmbH | www.itisim.com
11
9th Annual Weimar Optimization and Stochastic DaysNovember 29–30, 2012
Level 3 – Complex Model
Cylinders, Crank, PTOs
Turbocharger
• Multi-domain compounds
• DoF: > 10
• Parameter: > 50
• Frequency: > 100 Hz
DrivingResistance
GearboxOutput ShaftFlywheel
Engine Car
Gearbox
1D2D
3D
• Detailed engine & gearbox model
• StateCharts
Model Based Design | Requirements | Hierarchical Modeling
© 2012 ITI GmbH | www.itisim.com
12
9th Annual Weimar Optimization and Stochastic DaysNovember 29–30, 2012
Replaceable Models
Easily switchto compatible type
� Efficient modeling
Model Based Design | Requirements
© 2012 ITI GmbH | www.itisim.com
13
9th Annual Weimar Optimization and Stochastic DaysNovember 29–30, 2012
One model for multiple analyses
Model Based Design | Requirements
• Transient in time domain
• Static equilibrium (DC analyses)
• Stationary simulation (non-linear, frequency domain)
• Linear system analysis of the entire system:
• Eigenfrequencies, Eigenmodes
• Energy analyses
• Frequency response
• Poles/Zeros
© 2012 ITI GmbH | www.itisim.com
14
9th Annual Weimar Optimization and Stochastic DaysNovember 29–30, 2012
• Physical laws are describedin text books in terms offormulas like
• In programs this isimplemented dependent on the goal of the simulation
Equation Based Modeling
Equation Based Modeling
F = m a
Equation
F := m a
or
a := F / m
or
m := F / a
Assigments
© 2012 ITI GmbH | www.itisim.com
15
9th Annual Weimar Optimization and Stochastic DaysNovember 29–30, 2012
Modeling Concept – Lumped (Network) Elements
• Definition of potential and flow quantities for each physical domain
• Models consist of Elements and Connections
• Connectionscalculate potential quantities
and define conservation
equations
(e.g. ΣF = 0
for mechanical nodes)
• Elementsdefine relations between
flow- und potential variables within elements
(e.g. F = k * ∆x in a mechanical spring model)
element
FF
element
FF
element
FF ∑ = 0F
Node
...,x,x &
element
FF
element
FF
element
FF ∑ = 0F
Node
...,x,x &
Equation Based Modeling
© 2012 ITI GmbH | www.itisim.com
16
9th Annual Weimar Optimization and Stochastic DaysNovember 29–30, 2012
Electronics
Mechanicstranslational
Potential Quantity
FlowQuantity
SOURCES BASIC ELEMENT TYPES
CapacitiveElement
InductiveElement
DissipativeElement
Voltage V Current I Capacitor
UCI &×= UCI &×=
Inductor
∫×= dtUL
1I ∫×= dtU
L
1I
Resistor
R
UI =
R
UI =
Velocity v Force F Mass
vmF &×= vmF &×=
Spring
∫×= dtvcF ∫×= dtvcF
Damper
vdF ×= vdF ×=
MechanicsAngularVelocity ω
Torque T Inertia
ω×= &JT ω×= &JT
RotationalSpring
∫ω×= dtcT ∫ω×= dtcT
RotationalDamper
ω×= dT ω×= dT
Hydraulics
Thermics
Pressure p Volume Flow Q Volume
pVQ &×= pVQ &×=
Line (without loss)
∫×= dtpL
1Q
H∫×= dtp
L
1Q
H
Throttle
HRp
Q =
HRp
Q =
Temperature T Heat Flow P Heat Capacity
TCP &×= TCP &×=
-Heat Resistance
T
RP th=
T
RP th=
rotational
Analogies in Physical Domains
Equation Based Modeling
© 2012 ITI GmbH | www.itisim.com
17
9th Annual Weimar Optimization and Stochastic DaysNovember 29–30, 2012
17
• Allows Acausal and causal modeling
• Distinguishes Through and Across variables
• Multi-domain modeling
• Variables carry Attributes like units, documentation, min value, max value and nominal value
• Modeling based on a modeling language Standard
Modelica is a registered trademark of Modelica Association
Modelica Language
Equation Based Modeling
© 2012 ITI GmbH | www.itisim.com
18
9th Annual Weimar Optimization and Stochastic DaysNovember 29–30, 2012
18
• Standardized by Modelica Association
• Formed in September 1996
• An independent, international, non-profit modeling standards organization
• A clearing house for public domain Modelica libraries and documentation
• Organizer of Modelica design meetings and conferences
• www.modelica.org
• SimulationX is based on Modelica
Modelica is a registered trademark of Modelica Association
Who is Modelica?
Equation Based Modeling
© 2012 ITI GmbH | www.itisim.com
19
9th Annual Weimar Optimization and Stochastic DaysNovember 29–30, 2012
Mathematical Models represented by Modelica
• Differential Algebraic Equation(DAE)
• DAE (semi-explizit)
),,,,,(0 trdpxxf &=
),,,,,( trdpyxfx =&
Equation Based Modeling
),,,,,(0 trdpyxf=
x… continuous states
y… algebraic variables
p… parameters
d… discrete variable
R…root functions
t… time
© 2012 ITI GmbH | www.itisim.com
20
9th Annual Weimar Optimization and Stochastic DaysNovember 29–30, 2012
Event iteration
Global symbolic analysis
Time iteration
Iterate discrete states
Equation system preparation
Time step
Calc. consistent initial values
Prepare newTime step
Success?No
Yes
Finished?No
YesEvent?No
Yes
Start
Message
Message
STOP
Equation Based Modeling
© 2012 ITI GmbH | www.itisim.com
21
9th Annual Weimar Optimization and Stochastic DaysNovember 29–30, 2012
Symbolic Equation Handling
• No need for the user to decide which variables to solve for
• Reduces the number of different components in a library
• Redundant states can be eliminated in models
• Parts of models can be solved symbolically rather than numerically
• Symbolic initialization of model variables
• Increased performance of the simulation is gained at translation time
21
Modelica is a registered trademark of Modelica Association
Equation Based Modeling
© 2012 ITI GmbH | www.itisim.com
22
• Steady state tests and oscillating loads with higher frequency
• Membrane cylinder:
• Low friction losses
• Low sticking forces
• Low tolerance required -> low cost
• Small stroke
• Application with membrane cylinders
• Fatigue tests up to 50Hz and ±0,5 mm
• Testing machines (e.g. dental implants )
Pneumatic actuation of testing machines
Zahnimplantat
Quelle: QZM e.V.
Application | Membrane Cylinder
Fiedler, M.: Modellbildung und numerische Optimierung am Beispiel eines servopneumatischen Membranzylinderantriebs,TU Dresden, Dissertation 2010
© 2012 ITI GmbH | www.itisim.com
23
• Enlargement of amplitude of the membrane cylinder wrt. the whole drive concept
• Modification of the cylinder design – especially of the inner geometry
• Decreasing package dimensions, too
• Complex relation between stroke x, pressure p, chamber volume V, force F
F= f(p, A)
A=f(p, x)
Development Objectives
Application | Membrane Cylinder
© 2012 ITI GmbH | www.itisim.com
24
Workflow of optimization prozess
Optimization of membrane cylinder
Simulation modell ofmembrane cylinder
Elasticity of membrane
FEM – Simulation of the membrane
Membrane behavior(effective area, volume)
Geometry data
Application | Membrane Cylinder
© 2012 ITI GmbH | www.itisim.com
25
• Definition of limits of considered design parameters
• Control of the cylinder by a pneumatic valve
• Calculation of the influence of design parameters to the steady state membrane behavior
Simulation Model
Application | Membrane Cylinder
© 2012 ITI GmbH | www.itisim.com
26
Simulation Model
Ideal Valve Behavior
• no dynamics
• flow like VP60 1200 l/min
P-Controler(only option, for valve control)
Membrane Cylinder MechanicsMembrane Cylinder - Membrane behavoir
• no friction and damping
• Implementation of maps for effective area and chamber volumes
Signal Processing
endStop1
ctr1
ctr2
source1
ctr1
ctr2
in1
V_A
portA
portB
pinTh
V_B
portA
portB
pinTh
mass1
ctr1
ctr2
v_ab
x_k
dp
F_zyl
x_k
x
y
p_0
port
in1
Auslass_Bport
dirValve53
portA
portB
portP
portR
portS
y
Auslass_A
port
F_Last
x
y
dp_bar
x
y
x_w
x
y
sum1
x1
x2
x3
y
K
x
y
lever1
ctr1
ctr2
spring1
ctr1
ctr2
preset1ctr1
in1
Ventil_soll_V
x
y
Ventil_Signal
x
y
Anschluss_A
portA
portB Anschluss_B
portA
portB
vB
portA
portB
pinTh
vA
portA
portB
pinTh
pA
portP
p
pB
portP
p
pA_minus_pB
x1
x2
x3
y
sensor1
ctr1x
v
a
singlePass1
x
s
y
minimum
x
y
maximum
x
y
ampl
x1
x2
x3
y
Application | Membrane Cylinder
© 2012 ITI GmbH | www.itisim.com
27
Simulation Model for Optimization
Ideal Valve Behavior
• no dynamics
• flow like VP60 1200 l/min
P-Controler(only option, for valve control)
Membrane Cylinder MechanicsMembrane Cylinder - Membrane behavoir
• no friction and damping
• Implementation of maps for effective area and chamber volumes
Signal Processing
endStop1
ctr1
ctr2
source1
ctr1
ctr2
in1
V_A
portA
portB
pinTh
V_B
portA
portB
pinTh
mass1
ctr1
ctr2
v_ab
x_k
dp
F_zyl
x_k
x
y
p_0
port
in1
Auslass_Bport
dirValve53
portA
portB
portP
portR
portS
y
Auslass_A
port
F_Last
x
y
dp_bar
x
y
x_w
x
y
sum1
x1
x2
x3
y
K
x
y
lever1
ctr1
ctr2
spring1
ctr1
ctr2
preset1ctr1
in1
Ventil_soll_V
x
y
Ventil_Signal
x
y
Anschluss_A
portA
portB Anschluss_B
portA
portB
vB
portA
portB
pinTh
vA
portA
portB
pinTh
pA
portP
p
pB
portP
p
pA_minus_pB
x1
x2
x3
y
sensor1
ctr1x
v
a
singlePass1
x
s
y
minimum
x
y
maximum
x
y
ampl
x1
x2
x3
y
Application | Membrane Cylinder
© 2012 ITI GmbH | www.itisim.com
28
9th Annual Weimar Optimization and Stochastic DaysNovember 29–30, 2012
Optimization with optiSLang
optiSLang
Optimization with optiSLang
© 2012 ITI GmbH | www.itisim.com
29
9th Annual Weimar Optimization and Stochastic DaysNovember 29–30, 2012
Optimization with optiSLang
optiSLang
Optimization with optiSLang
© 2012 ITI GmbH | www.itisim.com
30
9th Annual Weimar Optimization and Stochastic DaysNovember 29–30, 2012
Sensitivity Analysis with optiSLang
First Attempt:
• 4 Parameters included
• Sensitivity Analyses executed
• CoP 1% ���� results not usable
• …
• Check model again
• Include more parameters
• …
Optimization with optiSLang
© 2012 ITI GmbH | www.itisim.com
31
9th Annual Weimar Optimization and Stochastic DaysNovember 29–30, 2012
Sensitivity Analysis with optiSLang
• 6 Parameters included
• Sensitivity Analyses executed
• CoP 83%
• Continue with optimization….
Optimization with optiSLang
© 2012 ITI GmbH | www.itisim.com
32
9th Annual Weimar Optimization and Stochastic DaysNovember 29–30, 2012
Interfaces to optiSLang
Benefit of Integration
• optiSLang 4 offers a range of direct integration nodes.
• In the case of SimulationX, the direct interface allows an easy and user-friendly parameter and response definition.
• In the optimization or calibration analysis, the specified properties are modified directly in the SimulationX model according to the defined ranges and the response values calculated for each design.
• Using the SimulationX API (COM based),the SimulationX model components, including their properties, can be directly accessed in the parametrization process of optiSLang.
© 2012 ITI GmbH | www.itisim.com
33
9th Annual Weimar Optimization and Stochastic DaysNovember 29–30, 2012
(Component) Object Model
Objects & Collections
• Application
• Simulation models
• Model components
• Parameters
• Result variables
• Element Types
• Result windows
• SelectionConnections
Connection
SimObjects
SimObjects
Applic
ation
Active Document
Docum
ent
Document
Types
Type
Entities
Entity
Result Window
Result Window
Parameters
Parameter
Results
Results
Parameters
Parameter
Parameters
Parameter
Results
Results
Results
Results
Selection
SimObjects
SimObjects
Connections
Connection
API of SimulationX
© 2012 ITI GmbH | www.itisim.com
34
9th Annual Weimar Optimization and Stochastic DaysNovember 29–30, 2012
• Model based design is core of future systems engineering
• Use noncausal models for different tasks (Modelica)
• SimulationX as powerful platform for systems engineering
• A pneumatic application used as test case for sensitivityanalysis
• Productiv optimization workflow with optiSLang 4
Next Steps
� Further increase efficiency
� Jump start guide (SimulationX + optiSLang)
Summary
Summary and Outlook
ITI GmbH
Headquarters
Webergasse 1 · Haus C2
01067 Dresden · Germany
T + 49 (0) 351.260 50 - 0
www.itisim.com
Supporting your vision