---
title: Coupled Aero-Hydro-Mooring-Servo Model
url: https://www.emergentmind.com/topics/coupled-aero-hydro-mooring-servo-model
type: topic
---

# Coupled Aero-Hydro-Mooring-Servo Model

A coupled aero-hydro-mooring-servo model is an integrated dynamic formulation in which aerodynamic loading, hydrodynamic forcing, mooring reactions, and control actions are solved as mutually dependent processes rather than as separate subsystems. In floating offshore wind turbines, this coupling is expressed through the feedback loop by which platform motion alters rotor-relative inflow and wake transport, aerodynamic thrust and torque excite rigid-body motion, moorings set restoring characteristics and sometimes add damping, and the servo system modifies both rotor dynamics and platform stability through blade-pitch and generator-torque actuation [2509.11707]. In experimental realizations, the same concept extends to hardware-in-the-loop configurations in which physical rotors generate the wake and aerodynamic loads while a real-time numerical model closes the platform-motion and mooring loop [2601.06964].

## 1. Definition and scope

The term denotes a class of multi-physics models rather than a single canonical formulation. In the most complete floating-wind form, the platform is treated as a rigid body with six degrees of freedom, the rotor/drivetrain is represented by at least one rotational degree of freedom, mooring forces are obtained from either quasi-static or dynamic line models, hydrodynamics are represented by restoring, radiation, excitation, and viscous terms, and the servo layer updates generator torque and collective blade pitch at each time step [2509.11707]. Reduced-order variants are also common. A hardware-in-the-loop wake-interaction study specialized the platform to surge and pitch in order to satisfy real-time constraints [2601.06964], while FOWFSim-Dyn represented a floating farm as particles translating only in surge and sway, with wake transport resolved along wake centerlines [2009.02585].

A recurring misconception is that “coupled” merely means that loads from several domains are added in one equation. The cited work uses the term in a stricter sense. Platform motion changes aerodynamic inflow, wake position, and controller feedback signals; aerodynamic loading changes platform motion; mooring and hydrostatic stiffness shift the natural frequencies that determine whether wake or control disturbances are amplified; and the controller can either suppress or inject effective damping [2509.11707]. In wake-interaction experiments this distinction is explicit: most prior studies prescribed platform motions, whereas the hardware-in-the-loop method was introduced specifically to preserve two-way coupling between wake-induced aerodynamic loading and floating-platform dynamics [2601.06964].

The concept is not limited to wind turbines. A tethered UAV-buoy system was also formulated as a coupled aero-hydro-mooring-servo problem, with aerodynamic thrust, buoy hydrodynamics, cable tension, and a polar-coordinate controller solved in one closed-loop model [2107.14662]. This suggests that the defining property of the class is not the application sector but the explicit co-evolution of environmental loads, tether or mooring reactions, and servo actions.

## 2. Governing equations and state representations

The common structural form is a rigid-body balance augmented by aerodynamic, hydrodynamic, mooring, and control terms. A general floating-wind representation is written as
$$
\big(\boldsymbol{M} + \boldsymbol{A}(\omega)\big)\,\ddot{\boldsymbol{x}}
+ \boldsymbol{B}(\omega)\,\dot{\boldsymbol{x}}
+ \boldsymbol{K}\,\boldsymbol{x}
=
\boldsymbol{F}_{\mathrm{aero}}(\boldsymbol{x},U_{\mathrm{eff}})
+ \boldsymbol{F}_{\mathrm{wave}}(t)
+ \boldsymbol{F}_{\mathrm{moor}}(\boldsymbol{x},\dot{\boldsymbol{x}})
+ \boldsymbol{F}_{\mathrm{ctrl}}(t),
$$
with $\boldsymbol{x}$ containing the six rigid-body degrees of freedom and the hydrodynamic operators collecting added mass, damping, and restoring terms [2601.06964]. A full 6-DOF time-domain implementation writes the platform equation in body coordinates as
$$
M \ddot{\xi} = F_{\mathrm{hs}} + F_{\mathrm{exci}} + F_{\mathrm{rad}} + F_{\mathrm{drag}} + F_{\mathrm{moor}} + F_{\mathrm{aero}} - mg\,e_z,
$$
where $\xi = [x,y,z,\phi,\theta,\psi]^\top$ and the radiation term follows the Cummins form
$$
F_{\mathrm{rad}}(t) = -A_\infty \ddot{\xi}(t) - \int_0^t K_r(\tau)\,\dot{\xi}(t-\tau)\,d\tau .
$$
Rotor dynamics are typically appended as
$$
J_r \dot{\omega} = T_a(\omega,V_{\mathrm{rel}},\beta) - T_g,
$$
with $\omega$ the rotor speed, $\beta$ the collective blade pitch, and $T_g$ the generator-torque command [2509.11707].

Notation is not uniform across formulations. In floating-platform models, $\theta$ or $\Phi$ often denotes platform pitch, whereas in rotor-control equations $\beta$ denotes blade pitch. A reduced analytical model of rotor-speed/platform-pitch coupling uses the state $x=(\theta,\dot{\theta},\phi,\dot{\phi})^\top$, where $\dot{\theta}=\omega$ is rotor-speed perturbation and $\phi$ is platform-pitch perturbation [2211.10362].

Farm-scale coupled models introduce an additional wake state. FOWFSim-Dyn defines a nonlinear state-space system
$$
\dot{\mathbf{x}}_{\mathrm{farm}}(t)
=
f\!\big(\mathbf{x}_{\mathrm{farm}}(t),\mathbf{u}_{\mathrm{farm}}(t),\mathbf{V}_\infty(t)\big),
$$
with turbine states containing positions and velocities and wake states containing centerline displacement, wake velocities, and wake diameter at discrete downstream points [2009.02585]. The wake transport obeys advection-type partial differential equations for centerline position, average wake velocity, and wake expansion, which are then discretized by upwind finite differences.

The same coupled structure can be cast in Euler-Lagrange form. In the tethered UAV-buoy example, the generalized coordinates include buoy surge and heave, tether angle, and vehicle attitude, and the governing equation is written as
$$
\mathbf{M}(\mathbf{q})\,\ddot{\mathbf{q}}
+
\mathbf{C}(\mathbf{q},\dot{\mathbf{q}})\,\dot{\mathbf{q}}
+
\mathbf{D}\,\tilde{\dot{\mathbf{q}}}
+
\mathbf{g}(\mathbf{q})
=
\boldsymbol{\tau}_{\mathrm{servo}}
+
\mathbf{F}_{\mathrm{hydro}}
+
\mathbf{F}_{\mathrm{aero}}
+
\mathbf{F}_{\mathrm{moor}},
$$
which makes explicit that the coupled-model concept is compatible with both rigid-body state-space and variational formulations [2107.14662].

## 3. Aerodynamic, hydrodynamic, mooring, and servo submodels

The aerodynamic block ranges from measured thrust feedback to blade-element formulations. In the hardware-in-the-loop wake experiments, aerodynamic loads were not computed by software; they were measured directly at the tower top, and the physical rotors were redesigned to reproduce the full-scale thrust coefficient $C_T$ and its sensitivity to blade inflow angle [2601.06964]. In simulation-oriented floating-wind models, aerodynamics are commonly handled by BEMT. OREGEN_BEMT computes local relative inflow, solves the induction factors iteratively with tip/hub-loss and modified Glauert corrections, and integrates elemental normal and tangential loads to total thrust and aerodynamic torque [2509.11707]. FOWFSim-Dyn instead uses yawed actuator-disk relations for $C_T$ and $C_P$, which makes the wake model computationally inexpensive at farm scale [2009.02585].

Wake treatment is one of the principal axes along which coupled models differ. In the hardware-in-the-loop framework, the wake is generated physically by an upstream rotor and measured by hot-wire anemometry; it is therefore not a numerical submodel at all [2601.06964]. In OREGEN-based simulations the aerodynamic solver receives turbulent inflow, and controller inputs are based on the rotor-disk-averaged wind speed
$$
V_d(t)=\frac{1}{A}\int_A u(x,t)\,dA,
$$
which smooths spatially incoherent turbulence and reduces control-input noise [2509.11707]. In FOWFSim-Dyn the wake is a dynamic parametric object governed by one-dimensional momentum conservation, wake-centerline advection, and a constant temporal expansion rate, with downstream rotor inflow obtained through Gaussian wake profiles and root-sum-square superposition [2009.02585].

Hydrodynamic fidelity likewise spans a hierarchy. The 6-DOF OREGEN_X implementation uses linear Cummins radiation-diffraction with nonlinear wave excitation and Morison viscous drag [2509.11707]. The hardware-in-the-loop platform model adopts a lighter real-time approximation: hydrostatics and infinite-frequency added mass are obtained from a panel-based potential-flow model, while radiation memory and higher-order nonlinear hydrodynamics are neglected in favor of a constant linearized damping matrix [2601.06964]. FOWFSim-Dyn reduces hydrodynamics further to added mass and Morison-type drag in planar motion, without radiation, diffraction, or hydrostatic rotational restoring [2009.02585].

Mooring formulations are especially diverse. For real-time wind-tunnel coupling, mooring restoring is represented by linear stiffness embedded in $\boldsymbol{K}_{\mathrm{hs,moor}}$, with any mooring-related damping absorbed into the aggregated hydrodynamic damping [2601.06964]. In the 6-DOF efficient time-domain simulator, the default mooring model is quasi-static catenary, with MoorDyn available as an alternative but not used in the main scenarios [2509.11707]. At higher fidelity, ARMoor models the line as a torsion- and shear-free Kirchhoff rod with axial stretch, bending curvature, per-unit-length normal and tangential drag, added mass, and a penalty-based barrier for seabed contact [2502.10256]. This formulation returns fairlead reactions to the platform solver and can act as a drop-in replacement for MoorDyn.

The servo layer includes both conventional region-based turbine control and platform-motion compensation. In the efficient FOWT model, below-rated operation uses torque PI with the common alternative law $T_g=k\omega^2$, while above-rated operation uses a gain-scheduled pitch PI loop referenced to rated rotor speed [2509.11707]. The same framework superimposes platform-feedback terms,
$$
\Delta \beta_{\mathrm{fb}}(t)=K_\beta(V_d)\,x_{\mathrm{hub}}(t), \qquad
\Delta T_{g,\mathrm{fb}}(t)=K_\tau(V_d)\,x_{\mathrm{hub}}(t),
$$
where $x_{\mathrm{hub}}=\dot{x}+h_t\dot{\theta}$ is the nacelle or hub fore-aft velocity. In the hardware-in-the-loop wake tests, by contrast, no active blade-pitch or torque control was engaged; the servo function was instead embodied in force filtering, inertia/gravity compensation, and real-time robotic motion tracking [2601.06964].

## 4. Coupling strategies and numerical realization

A coupled model is defined not only by its submodels but by the way they exchange states and loads. In the efficient 6-DOF FOWT solver, the data flow at each time step is explicit: platform states are sent to aerodynamics and control, aerodynamics and moorings return loads to the platform equation, and the control system updates $T_g$ and $\beta$ before the next step [2509.11707]. The solver advances platform degrees of freedom and rotor speed in a single integrated march, and the paper reports that full one-hour DLC runs complete in about 10 minutes on a laptop.

Hardware-in-the-loop realization makes the coupling architecture physically visible. Two 1.2 m diameter rotors at geometric scale 1:150 are mounted on robotic platforms that reproduce surge and pitch, while a real-time numerical model computes the motion response from measured aerodynamic forces plus hydrostatic, mooring, and hydrodynamic terms [2601.06964]. The model runs at a 1 ms time step at model scale, corresponding to approximately 60 ms at full scale because the chosen length and velocity scales yield a time scale of 1:60. The loop is explicit: measure tower-top forces and platform motion, remove inertia and gravity of the rotor-nacelle assembly, integrate the rigid-body equations, command the robots, and let the resulting motion feed back into the aerodynamics and wake.

Farm-scale dynamic wake models use a different numerical structure. FOWFSim-Dyn discretizes wake-centerline PDEs with upwind finite differences and advances them with either fourth-order Runge-Kutta or forward Euler, subject to a CFL-like condition $c_i\Delta t/\Delta x \le 1$ [2009.02585]. The platform positions and velocities alter wake advection speed, boundary conditions at the rotor, and the relative placement of downstream machines inside upstream wakes. This is a weaker but still explicit form of aero-hydro-mooring coupling, tailored to control and engineering studies rather than turbine-level load reconstruction.

High-fidelity mooring co-simulation can introduce additional substepping. ARMoor uses B-spline spatial discretization and a second-order implicit midpoint-trapezoidal hybrid time integrator, with Newton-Raphson iterations and consistent tangents [2502.10256]. For coupled operation, the recommended exchange rate with the platform solver is 50–100 Hz, with mooring substeps under high-frequency excitation. This reflects a general numerical principle across the literature: strong coupling can be retained with explicit exchange when time scales are mild, but line dynamics, added-mass effects, and seabed contact may require either finer integration or predictor-corrector iteration.

## 5. Stability, damping, and control-induced phenomena

One of the central findings in this literature is that the servo layer is not dynamically neutral. In above-rated floating-wind operation, baseline pitch control can generate negative aerodynamic damping. In the 6-DOF efficient model this is analyzed through the pitch equation
$$
I_\theta \ddot{\theta} + (c_\theta + c_{\mathrm{aero}})\dot{\theta} + k_\theta \theta = M_{\mathrm{exc}}(t),
$$
where, near rated wind speed, baseline control can produce $\partial F_T/\partial V_{\mathrm{rel}}<0$, implying $c_{\mathrm{aero}}<0$ and amplification of platform pitch [2509.11707]. The proposed remedy chooses
$$
K_\beta(V_d)=\frac{\partial F_T/\partial V_{\mathrm{rel}}}{\partial F_T/\partial \beta},
\qquad
K_\tau(V_d)= -\Big[\frac{\partial T_a}{\partial V_{\mathrm{rel}}} - K_\beta(V_d)\frac{\partial T_a}{\partial \beta}\Big],
$$
so that thrust sensitivity to hub velocity is canceled and rotor torque is approximately decoupled from platform motion.

A reduced analytical model makes the damping decomposition explicit. For coupled rotor-speed and platform-pitch dynamics, the pitch equation can be written as
$$
J_t\ddot{\phi} + \bigl(D_t + h_t^2 - k_\beta h_t\bigr)\dot{\phi} + K_t\phi = \tau_{\mathrm{ext}},
$$
so the effective pitch damping is
$$
C_{\mathrm{pitch}} = D_t + h_t^2 - k_\beta h_t .
$$
Here $D_t$ collects hydrodynamic and structural damping, $h_t^2$ is the aerodynamic-convective contribution arising from the relative wind $v_r=v-h_t\dot{\phi}$, and $-k_\beta h_t$ is the direct control-induced term [2211.10362]. On this basis, the “$\zeta_{\mathrm{plt}}$-fixed” strategy selects
$$
k_\beta = \frac{1}{h_t}\Bigl(D_t + h_t^2 - 2\sqrt{K_tJ_t}\,\zeta_{\mathrm{plt}}\Bigr),
$$
which changes the pitch damping ratio without changing the natural pitch period.

The same reduced model identifies two non-minimum-phase zeros: one in the $\beta \rightarrow \phi$ channel and one in the $\beta \rightarrow \omega$ channel [2211.10362]. Their physical interpretation is that a blade-pitch action intended to regulate rotor speed can transiently move the platform or rotor in the wrong direction because thrust, torque, and platform-velocity feedback interact through the relative-wind term. Generator-torque compensation of the form
$$
k_{\tau_g} = -\,m_{\tau_g}\,\frac{h_t}{N_g}, \qquad m_{\tau_g}\in[0,1],
$$
is proposed to avoid the rotor-speed non-minimum-phase behavior.

Mooring dynamics can also alter the effective damping and inertia pathways. ARMoor reports that under normal pulsating fairlead loads the line transitions from a drag-dominated regime at low frequencies to an added-mass-dominated regime at higher frequencies, with a transition near $f \approx 3$ Hz under the studied conditions [2502.10256]. Under tangential forcing, axial and bending dynamics are strongly coupled, and comparable normal and tangential amplitudes can emerge. This indicates that mooring models are not merely restoring-force providers; at sufficient fidelity they become dynamical subsystems with their own regime changes and coupling pathways.

## 6. Validation, applications, and limitations

Validation evidence spans wind-tunnel experiments, reduced-order wake benchmarks, time-domain controller studies, and mooring-module comparisons. In the hardware-in-the-loop wind-tunnel study, two aligned turbines at 5.75D spacing showed that the downstream machine, fully immersed in the upstream wake, produced $T_2=627$ kN, approximately 34% of the upstream $T_1=1841$ kN, with mean surge and pitch of approximately $6.81$ m and $1.3^\circ$ compared with $20.1$ m and $3.8^\circ$ upstream [2601.06964]. At the same time, the downstream platform exhibited amplified low-frequency motion peaks because wake turbulence increased energy near the platform natural surge and pitch frequencies of approximately $0.005$ Hz and $0.040$ Hz at full scale.

The efficient coupled 6-DOF model was validated against bottom-fixed IEA 15 MW steady curves and then used for turbulent-wind and irregular-sea studies [2509.11707]. Its modified platform-feedback controller eliminated the negative damping effect on platform pitch while maintaining small rotor-speed variations. Constant-gain pitch feedback could also stabilize the platform, but at large wind speeds it produced overspeed exceeding the safety threshold of 20%, whereas the gain-scheduled modified controller stayed below that threshold across all DLC 1.6 cases. The same study found that turbulent wind increased the maximum rotor speed by 7.9% to 23.7% over the steady-wind cases considered.

At farm scale, FOWFSim-Dyn reported steady-state wake-centerline discrepancies of at most 8.19% of rotor diameter and far-wake velocity-profile errors below 3.87% of the free-stream wind speed, indicating what the paper terms a satisfactory level of fidelity for engineering applications [2009.02585]. For mooring dynamics, ARMoor reproduced OpenFAST/MoorDyn fairlead positions within 0.74% and anchor tensions within 1.01% in the reported comparisons, while also resolving seabed contact and axial-bending coupling unavailable in simpler quasi-static catenary models [2502.10256]. In reduced-order damping analysis, the $\zeta_{\mathrm{plt}}$-fixed strategy applied to the UMaine VolturnUS-S IEA 15 MW model reduced platform pitch and tower-base fatigue, with the paper reporting an average DEL reduction on the order of 15% and power changes typically within $\pm 1\%$ [2211.10362].

The limitations are equally clear. The hardware-in-the-loop methodology deliberately breaks strict Froude and Reynolds similitude in order to preserve inflow quality and low-frequency thrust modulation, so absolute loads and motions are not fully scaled [2601.06964]. The efficient 6-DOF solver assumes rigid blades and tower, relies mainly on quasi-static moorings, and derives its feedback gains from BEMT sensitivities [2509.11707]. FOWFSim-Dyn omits pitch, roll, heave, yaw, unsteady aerodynamics, and dynamic mooring effects [2009.02585]. ARMoor excludes torsion and shear and models seabed interaction through a smooth unilateral barrier without frictional tangential contact [2502.10256]. The reduced damping model condenses hydrodynamics and mooring physics into lumped parameters and therefore cannot replace aero-hydro-servo-elastic simulation when higher-order platform or structural modes matter [2211.10362].

Taken together, these studies show that a coupled aero-hydro-mooring-servo model is best understood as a fidelity-scalable framework. Its minimal form may be a planar state-space model for wind-farm control; its real-time form may be a two-degree-of-freedom hardware-in-the-loop platform model; its control-design form may be a reduced rotor-speed/platform-pitch system with explicit damping decomposition; and its high-fidelity form may include dynamic mooring rods, radiation-memory hydrodynamics, and gain-scheduled servo feedback. What remains invariant across these implementations is the insistence that wake, motion, mooring, and control be treated as a closed dynamical loop rather than as sequential corrections.

Source: https://www.emergentmind.com/topics/coupled-aero-hydro-mooring-servo-model