The complete primary-frequency-control model combines the acausal hydropower plant with the signal-based governor.
The waterway sits inside the turbine-power path, so the plant seen by the governor is dynamic.
Here $x_p$ includes reservoir, penstock, surge-tank, turbine/shaft and generator dynamic states, while $z_p$ contains algebraic pressure, flow, torque and electrical-network variables.
A load increase is modeled by
$$P_L=P_{L0}+\Delta P_L H(t-t_d).$$Immediately after the disturbance, mechanical power has not changed, so
$$\Delta P_m-\Delta P_e<0,$$and frequency falls.
The governor sees $\Delta f<0$ and increases guide-vane opening. The resulting power increment can be represented as
$$\Delta P_m(s)=G_h(s)G_t(s)G_g(s)\Delta f(s),$$where $G_g$ is the governor/servo, $G_h$ the waterway dynamics, and $G_t$ the turbine conversion.
A reduced generator-grid model is
$$M\Delta\dot\omega+D\Delta\omega=\Delta P_m-\Delta P_L.$$Thus
$$\Delta\omega(s)=\frac{\Delta P_m(s)-\Delta P_L(s)}{Ms+D}.$$With permanent droop and no secondary control, steady-state frequency error is generally nonzero. If
$$\Delta P_m=-\frac{1}{R}\Delta f,$$then for a single-area approximation
$$\Delta f_{ss}=-\frac{\Delta P_L}{D+1/R}.$$Useful outputs are nadir, RoCoF, settling time, delivered power, gate travel, pressure excursion and surge-tank level. A hydropower-specific qualification study should therefore inspect both electrical and hydraulic constraints.
A textbook governor-turbine block can match low-frequency behavior, but OpenHPL exposes whether the same control action causes unacceptable pressure waves, surge oscillations, guide-vane saturation or turbine operating-point excursions.