This simulator models a small power system where two coherent synchronous machines share a single frequency, alongside grid-following renewable injection, a fast-responding battery and frequency-sensitive demand. Add or remove load, trip a generator, and watch inertia, governor droop, secondary dispatch, battery support and load shedding work together to keep generation and load in balance.
• A 3D view of the shared-frequency power system — two synchronous generators, renewable source and battery — with orbit, auto-rotate and expand controls, plus a component legend. • Dispatch and demand controls: run time, step or advance the clock, add/remove 20 MW of load, trip generator A or B, and restore previously shed load. • A resource-response panel and charts for system frequency, supply vs. demand, generator-by-generator sharing (each with a 100 MW model limit), and battery/imbalance response. • Live metrics, an event log, a plain-language explanation of what changed and why, guided experiments, a verification bench, a diagnostic challenge, and a knowledge-check quiz.
The model tracks Ebase × d(ω²)/dt = ΔP, where ΔP is mechanical input plus renewable plus battery power minus demand and losses, and Ebase is the combined H×S kinetic energy of all connected machines (50 Hz corresponds to ω = 1, so f = 50ω). When load changes suddenly, frequency first moves according to that combined inertia — more connected machines and higher inertia constants mean a slower initial frequency slide.
Each governor then adds power proportional to frequency error (100 × (50−f)/(R×50) MW per unit), which arrests the slide but leaves a droop offset, similar to the governor-frequency lab. Demand itself is frequency-sensitive (demand = (nominal − shed) × [1 + D(ω−1)]), and integral secondary control — an AGC-like dispatch signal — is shared among online units to drive frequency back toward 50 Hz over a longer time horizon, removing the droop offset the way isochronous control does in a single-machine system.
Renewable output in this model is an imposed scenario input — it contributes MW but adds no rotational inertia, since it's grid-following rather than grid-forming. The battery is energy-limited (±20 MW, 5 MWh, 95% round-trip efficiency in each direction) and reacts with a 150 ms response lag to provide fast, frequency-sensitive support — again not grid-forming, so it needs synchronous machines already established on the system.
Under-frequency load shedding sheds 10% of current nominal demand at each of three thresholds — 48.8 Hz, 48.3 Hz and 47.8 Hz — after 200 ms spent below the threshold, and each shedding stage latches until manually restored. If frequency stays outside 45–55 Hz for two seconds, the teaching fixture de-energizes entirely, and without any synchronous machines connected, the grid-following renewable and battery sources cannot establish frequency on their own. These are illustrative values for teaching purposes, not utility operating settings.
The combined kinetic energy (inertia) of all connected synchronous machines. The model applies Ebase × d(ω²)/dt = ΔP, so a larger connected inertia (more machines, or machines with higher H) means the same power imbalance produces a slower, shallower frequency dip, giving governors more time to respond.
That is the expected droop offset from proportional governor control, which restores balance but not the original frequency setpoint. Removing that offset requires secondary control — an integral, AGC-like dispatch signal shared among the online units that gradually adjusts output to drive frequency back to exactly 50 Hz.
Both are modeled as grid-following rather than grid-forming resources. The renewable source injects a scenario-defined amount of power but contributes no rotational inertia, and the battery, while it can react quickly (a 150 ms lag) within its ±20 MW, 5 MWh limits, provides frequency-sensitive support rather than establishing the system frequency itself. Without at least one synchronous machine connected, the model cannot establish a stable frequency at all.
Under-frequency load shedding trips in three 10%-of-demand stages at 48.8 Hz, 48.3 Hz and 47.8 Hz, each after frequency has spent 200 ms below that threshold. Each stage latches — it stays shed until the load is manually restored using the restore control, reflecting how real load-shedding schemes require deliberate reconnection rather than automatically picking load back up.