How Vintage Linear Systems Controllers Have Become Inadequate In Renewables-Heavy Power Systems: Limitations and New Solutions

Abstract

In power systems, various control engineering algorithms are instrumented to regulate voltages and frequencies. One of these algorithms is load frequency control (LFC). Thanks to developments in control theory and convex optimization, numerous LFC strategies have been proposed. However, many rely on a reduced, linearized model of power systems where: (i) load and renewables buses are eliminated, (ii) only synchronous machines are modeled, (iii) nonlinear network transients are replaced with their linear approximations around tight operating regions, and (iv) algebraic power flows are eliminated. In contrast, the increased penetration of renewables from wind and solar introduces significant uncertainty which can only be captured explicitly via a nonlinear differential algebraic equation (NDAE) model that incorporates loads, renewables, and nonlinear machine and power flow transients. As it is demonstrated herein, such vintage, linearization-based approaches may be inadequate. They can be easily overwhelmed by high degree of uncertainty, thereby failing to stabilize power systems post-disturbance. To that end, we showcase the limitations of some widely utilized linearization-based LFC methods and offer a new solution which manifests through a simple, linearization-free approach that can handle large disturbances in power systems originating from a sudden reduction of renewables generation and increase of demands.