1. Automated object oriented simulation framework
Hello everyone and thanks for coming.
I have a pleasure to present today summary of my first quarter here at CERN.
I will describe automated object oriented simulation framework.
Michal Maciejewski
TE-MPE-PE

2. Acknowledgements
Many thanks to Emmanuele for his support, guidance and patience during design and validation of Simulink models.

3. Outline
Introduction
Architecture
OOP
Results
Summary

4. Standard Approach - motivation
Parameters are stored in Excel file
Blocks are placed and connected manually on schematic
Data can be accessed only manually
Tutaj wideo jak się robi symulację
Czy zgodzimy się, że wygodnie jest używać Excela do przechowywania parametrów?
Ideally we can build a model and run it. Ideally, but world is not ideal and we might have some bugs. We waste time and energy on debugging all the time.

5. Standard Approach - Drawbacks
Efficient only for relatively small models
Prone to copy&paste errors
Difficult to use models by others
Plenty of time spent on debugging and testing

6. Automated model design Framework for everyone…
Reusable libraries
Automated model design
Framework for everyone…
Main Assumptions
Pogrubienie, gdy omawiam cos.
Model defined only inside Excel File
Create framework reusable for others with clearly defined set of steps to follow in order to create different simulations.
Scalability: different set of components, variable number of blocks
User should focus on parameters and analysis of results
You can run the simulation without any knowledge of Simulink, Simscappe, OOP
…even without any Simulink skills
Validated once, used many times
Framework must be scalable

7. Framework Architecture
Black box
Model.netlist
Simulink Model
Pojawia się najpierw ten Excel, potem można kliknąć i pojawi się przykładowy plik Potem przesuwa się w lewo i z tego powstaje netlista i tworzone są obiekty
Thermal Mass
Helium Cooling
Heat Exchange
Quench Resistor
CLIQ …
Objects

8. Model Design Sequence Run simulation Create model Read Excel File
Create netlist
Initialise objects
Create model
Place blocks
Connect blocks
Set parameters
Run simulation
Plot results
Save results
It is pretty straight-forward

9. Object Oriented Programming
Creates new data type to entirely describe component
Provides a scheme to add new components
Simplifies libraries documentation process
Makes it easier to reuse software
Coherent and logical representation

10. Our library
Electrical
Thermal
Tutaj struktura drzewiasta z kolejnymi

11. R(T)=R(T0)*(1+œ*(T-T0))
Topic: Variable resistor - example
Temperature coefficient
R(T)=R(T0)*(1+œ*(T-T0))
2. Voltage across resistor
V =I*R(T)
3. Ohmic losses
Q=I*I* R(T)
Feel free to take notes during this short lecture 

12. R(T)=R(T0)*(1+œ*(T-T0))
Topic: Variable resistor - example
function setup
through( Q, [], OHM.Q );
across(T, n.T, p.T)
end
equations
let
R = R0*( 1 + alfa*(T-T0) ); in
v == i*R;
Q == i*i*R;
component resistorThermal < foundation.electrical.branch
% Resistor with thermal coefficient
nodes
OHM = foundation.thermal.thermal; % OHM:left
end
parameters
R0 = { 10, 'Ohm' }; % Reference Resistance
T0 = { 273, 'K' }; % Reference Temperature
alfa = {3.9e-3, '1/K'}; % Temperature coefficient
variables
Q = { 0, 'J/s' };
T = { 0, 'K' };
R(T)=R(T0)*(1+œ*(T-T0))
V =I*R(T)
Feel free to take notes during this short lecture 
Q=I*I* R(T)

13. Topic: Mutual Inductor- example
Ohm law for mutual inductor
𝑣 1 𝑣 2 ⋮ 𝑣 𝑛 = 𝐿 1 𝑀 21 ⋯ 𝑀 𝑛1 𝑀 12 𝐿 2 ⋱ ⋮ ⋮ 𝑀 1𝑛 ⋱ ⋯ ⋱ ⋯ ⋮ 𝐿 𝑛 𝑖 𝑖 2 ⋮ 𝑖 𝑛
Mutual inductor created automatically
Additionally the stability is verified

14. MQXC 2x18 simulation - demo
This is what you can see on 27” screen
Pokazac jakie sa czesci schematu na ekranie zaznaczyc co gdzie: Thermal

15. MQXC 2x36 simulation - demo

16. Comparison - Current
Opisać co na osiach jest

17. Comparison - Temperature
Opisać co na osiach jest

18. Next steps Design of Graphical User Interface (GUI)
Implementation of new components (Quench heaters, improved IFCL)
Modeling of new magnets (MQXF)

19. Summary Only one Excel file needed to define model
User can focus only on parameters calculation and analysis of results
Framework does not require any experience in Simulink/Simscape
This concept can be easily extended
This framework has clearly defined rules to add new features
Based on current results we can say that this idea is very promissing. Still we don’t know what our limitations are.

20. Summary - comparison Feature PSpice Matlab/Simulink
User friendly (support, model design) No
Yes
Multi-domain components
Ability to post-process results
Poor
Good
Automated model design
Hard
Simple
Reusable components
GUI
Automatic update of components on schematic
One of our motivations was to switch from Pspice to Simulink

21. Summary – time savings Automated framework Standard Approach
One of our motivations was to switch from Pspice to Simulink

22. Thank you for attention!