This has been the subject of a great deal of research over the past few decades. Computation techniques including artificial intelligence, genetic algorithms, etc. are being developed (GA load flow techniques was the subject of my Master's degree project 10 years ago.)
Stoll, Harry - Least Cost Electric Utility Planning
A Genetic Algorithm (GA) is an optimization method motivated by natural selection and genetics. Parameters of the problem are encoded into a finite length string utilizing a user-defined alphabet (a common approach is to use binary). Once the problem is encoded, the solution begins with a population of solutions (sometimes just random numbers). The GA moves from one generation to the next by evaluating the fitness of each member of the population (i.e., how well it solves the initial problem), selecting the members to parent the next generation (based on fitness) then creating the next generation of solutions.
To get back to the original question, there are numerous possible objectives for load flow calculations. Generally the goal is to computationally model the steady state condition of a power system. Normally you specify real and reactive power flow in and out of each bus and designate one “swing” bus to provide one degree of freedom for the system. The load flow computation then calculates the voltage at each bus and complex power flows and losses.
Programs can adjust various system parameters to optimize certain results. For example you can attempt to mimize system losses, minimize cost of generation while maintaining required reserves, optimize system voltages, reactive power flow or stability margins, etc. These are very computationally intensive calculations for large systems so there is a lot of research on methods to improve speed and accuracy. Many of these give rise to non-linear conditions. GA's are good at handling non-linear systems.
Comment: The steady-state operation of power systems as a topic for the study is named power flow or load flow.
There is a need to derive system equations describing the power system. The system equations are nonlinear. Two trial and error iterative methods, Gauss-Seidel and Newton-Raphson, are used to provide approximate solutions. A simplified linearized power flow called DC power flow can also be used.
Thanks for the useful info Ahleman, I have come across the Gauss Seidel shall and Newton Raphson (iterative) methods but am intrigued by the GA technique you mention - I will follow the threads.
Some more fodder for the grey matter to process
_______________________________________
Regards -
Colin J Flatters
Consulting Engineer & Project Manager
