Hi @Jingfan Liu. Which solver did you use in the original model?

It seems that one of the constraints uses a geometric function containing at least one variable. That makes the model nonlinear; either NLP or MINLP.

BARON is a global solver that can handle NLP and MINLP but it cannot handle geometric functions with variables. Octeract is another global solver for NLP and MINLP which can handle geometric functions with variables. You can also use a local nonlinear solver: IPOPT, CONOPT, Knitro or SNOPT for NLP, and Knitro or the GMP-OA module for MINLP.

Thank you for your answer. I built the OPF model based on a tutorial in “How-to”, and I don't get any errors when I run the tutorial's model of OPF. Is this due to a difference in our solver settings? And how to set the solver manually?

The OPF model from the tutorial uses a geometric function containing a variable, so you must have used a different solver for that model. You can change the (default) solver in the solver configuration. You can also specify the solver in the solve-statement, e.g.,

Solve OPF where solver := 'CONOPT 4.1';