# Having problems with solver settings and step size in simscape electrical

2 views (last 30 days)
K on 29 Jan 2024
Answered: Joel Van Sickel on 20 Feb 2024
Hello,
I problems with undertanding solver settings and step size in simscape electrical, I attached the model to make it easier.
To simulate a motor-generator unit I used the asynchronous motor with converter from the electrical engine dyno example (changed the frequency to 50Hz, Power to 250kW and PI gain of the speed loop) https://de.mathworks.com/help/sps/ug/electric-engine-dyno.html
with findamental sample time Ts=1e-5 and control sample time Tsc=1e-4
the solver is fixed step discerete with fixed step size = Ts.
The electrical load gets its active power from measurments data table (187 measurement points) each 2 measurement points are 30s apart from each other (measuring device measures 1 value every 30s). I interpolated a value every 10s so I got a total of 561 points. my question ist:
• with which step size should I input these 561 points in a look up table or signal editor so that my results make physically sense? I inputed them in a step size of 0.1 starting from time 3 (after motor acceleration phase) like shown bellow.
• what does a step size correspond in seconds and which model parameters sets the step size? does 0.1 in this case implies 10s or 1 seconds or how much?
I used the 0.1 step size because i previously inputed my value every natural number (1 2 3 ...) and as you see bellow the system response (velocity ASM) happens within 0.1 and the velocity remains constant the next 0.9 untill the following value is computed, which leads to these unwanted semi-Intervall
the fundamental sample Ts time is dependent of the switching frequency and has the be smaller or equal 1/(10*switching frequency) and the control sampe time has to be sammler Tsc or equal fundamental sample Ts.
with Ts=1e-5 and Tsc=1e-4 I thought that the step size 0.1 is their ratio Ts/Tsc but I can't seems to get the physical meaning behind what I implemented in the signal editor. Can you help me understand what step size is meaningful in this case and what is the physical meaning behind it? how do I get it right for next models
regards,

Joel Van Sickel on 20 Feb 2024
This may or may not be helpful, but does dive into solver details: https://www.youtube.com/watch?v=QWTQqJ2Akmg&t=26s
For your model, the rule of thumb would be if you are using fixed step solver (local solver) use at least t_switch/100 (100 simulation step sizes per switching cycle). Or you can switch to a variable step solver (ode23t or DAESSC) and not have to worry about it.
step size doesn't matter for importing time stamped data, as it will be interpolated according to the time step in the data file and the sample rate of the output signal.