TDS Convergence is not likely.

[xiaojiangsi]

Based on kundur_full.xlsx, I set the fault start from 1.0s to 1.1s at bus 1, and when I run the tds, it shows the information as follow:

Time step reduced to zero. Convergence is not likely. Simulation terminated at t=1.1000 s.

I wonder why this would happen, how I get the result under the above situation (Based on kundur_full.xlsx, I set the fault start from 1.0s to 1.1s at bus 1).

[cuihantao]

I was able to replicate this isse. It is due to the non-convergence of post-fault equations using the pre-fault values as the initial values.

One temporary solution is to scale the initial values to the post-fault equations by setting ss.Fault.config.scale.

import andes

andes.config_logger(20)

%matplotlib inline

ss = andes.run("kundur_full_fault.xlsx")

xy = ss.TDS.init()

ss.Fault.config.scale = 1.8
ss.TDS.run()

ss.TDS.plt.plot(ss.Bus.v)

kundur_full_fault.xlsx

[xiaojiangsi]

I tried the above and it works under the situation of fault start from 1.0s to 1.1s at bus 1. But in other situations which don't need change the scale can be converged, after set the scale at 1.8, it is not work. Could you tell me how to adjust the scale in different scenarios? What is the usage of scale? for example when I alter the tc to 1.05, the scale change doesn't work:

[cuihantao]

When a fault gets cleared, algebraic variables change drastically. E.g., voltages can go from nearly zero back to 1.0. As we are using Newton's method for solving the DAE, the initial values are crucial for the immediate step after fault clearance.

What's done in ANDES is the restoration of pre-fault algebraic variables. ss.Fault.config.scale is the scaling factor for the pre-fault algebraic variables for adjusting the initial values. There is no theory about adjusting the initial value; some trial and error are expected for severe disturbances, like the one on Bus 1 where there is a generator. See here: https://github.com/CURENT/andes/blob/e4d1c274129d2e1ff0287ade895625f633b9c640/andes/models/timer.py#L208.

[xiaojiangsi]

Thanks for your answer.