The impact of spanwise surface temperature heterogeneity on steady stably stratified Ekman layers is systematically studied using large-eddy simulation (LES). Spanwise varying strips of high and low surface temperature are imposed in idealised LES of stable boundary layers (SBLs), in which a steady state results from a balance between cooling at the ground and heating due to imposed synoptic subsidence. Consistent with previous studies on channel flows with streamwise-aligned surface heterogeneity (e.g. Bon & Meyers, J. Fluid Mech. 2022, pp. 1-38), large-scale secondary circulations develop and extend deep into the stable Ekman layer. Coriolis effects enhance counterclockwise circulations while reducing clockwise ones, thereby tilting the mean secondary flow structures towards the left (in the northern hemisphere). Nevertheless, for the considered surface temperature contrasts of 1.5-12 K and spanwise wavelengths of 100-800 m, the impact on mean SBL structure is substantial. As the surface temperature difference or strip width increases, secondary flows and dispersive fluxes strengthen, eventually reaching the top of the SBL. This augmentation further enhances near-surface gradients, elevates SBL depths and low-level jets, and reduces mean surface heat fluxes. Novel correlations between characteristics of the surface heterogeneity and their impact on the mean SBL structure are proposed. Moreover, the local surface fluxes are shown to significantly deviate from the mean, highlighting that horizontally averaged SBL properties do not capture all important physical processes in a heterogeneous flow. Overall, this work affirms that thermal surface heterogeneity is a crucial factor in governing transport processes within the atmospheric stable boundary layer.

Accurate modeling of wind farm-atmosphere interactions is critical for reliable energy yield assessments and flow control strategies. However, formal model comparison methodologies that quantify model form uncertainty by also accounting for parameter uncertainty are still lacking. This study presents an enhanced Bayesian uncertainty quantification framework for the calibration and validation of engineering wind farm flow models. Building on previous work, the framework explicitly incorporates model inadequacy through a parametrized model error distribution, enabling the separation of model and measurement uncertainties. The improved framework is demonstrated using a large-eddy simulation dataset for wind farm blockage and atmospheric gravity waves in conventionally neutral boundary layers. Two models of differing fidelity - a standard Gaussian wake model and an atmospheric perturbation model (APM) - are calibrated and compared. The posterior distribution of the model parameters reveals insights into model behavior and highlights areas for further improvement, for instance, when estimated parameter values are inconsistent across the model chain. In addition, it is shown that not explicitly incorporating model inadequacy results in an overly confident posterior distribution and renders derived stochastic flow models incapable of representing model uncertainty. A comparison of the quantified model uncertainty shows that the APM has significantly lower uncertainty than a standard wake model for this dataset, as the wake model is unable to represent wind farm blockage effects. This demonstrates the utility of the framework for objective model comparison with quantified parameter and model uncertainty given a reference dataset. Both the framework and the parallelized sequential Monte Carlo algorithm for accelerated posterior sampling are made available through the open-source Python package UMBRA.

Abstract This paper compares four-dimensional variational data assimilation (4D-Var) and ensemble Kalman filtering (EnKF) based on large-eddy simulation (LES) for turbulence reconstruction in atmospheric boundary layers. For 4D-Var, the reconstruction is cast into a PDE constrained optimization problem following a classical maximum a posteriori approach, where LES arises as a strong constraint. The problem is regularized based on the two-point spatial correlation tensor that is approximated by the analytical Hunt-Graham-Wilson model ( J. Fluid Mech ., vol. 981, 2024, A28). The 4D-Var is then incorporated in a receding-horizon framework to assimilate over longer horizons. For EnKF, an ensemble of LES realizations is generated to represent uncertainties, and updated based on the measurements and the associated covariance matrices. Additionally, covariance localization is used to decrease the required ensemble size. For a realistic lidar-based reconstruction case, it is shown that – when tuned properly – EnKF may achieve similar reconstruction accuracy and overall computational cost as 4D-Var. However, in terms of compute time, EnKF outperforms 4D-Var by 1–2 orders of magnitude due to the embarassingly parallel ensemble evaluation, making it a more suitable candidate for real-time applications.

The height of the atmospheric boundary layer (ABL) exerts a significant influence on flow behavior within wind farms and directly impacts their performance. This study investigates how variations in ABL height and capping inversion layer thickness affect the efficiency and power output of a gigawatt-scale wind farm. Five advanced numerical approaches, ranging from high-fidelity large-eddy simulations (LESs) to Reynolds-averaged Navier-Stokes (RANS), are used to model farm-scale flow dynamics under shallow (similar to 150 m) and deep (similar to 500 m) ABL conditions. The results consistently show that shallow ABLs increase flow blockage and turbine wake interactions, leading to reduced power production. In contrast, deeper ABLs promote enhanced wake recovery and increased overall energy yield. These trends were observed across all solvers, demonstrating the robustness of the findings. Notably, while some quantitative differences emerged depending on modeling fidelity and computational domain size, the overarching trends remained consistent among the participating research institutions and industry partners. The simulation cases performed are complex, and the results of the different methods show a variation of up to 10 %, and further research is needed to limit this gap. Based on these results, it is not clear to what extent the variation depends on the fidelity level of the models used. The study concludes that ABL height and stability are critical parameters to consider in wind energy siting and turbine layout design to optimize performance across varying atmospheric conditions.

A longstanding challenge in CO_2 flux measurements above vegetation is the unclosure of the flux balance in night-time stable boundary layers. In recent years, the impact of surface temperature heterogeneity in stable boundary layers on momentum and heat-flux balances as a result of secondary motions has received increased attention. In the current work, we set up a series of idealized large-eddy simulations in stable boundary layers and look at the effect of such surface temperature heterogeneity on the CO_2 flux balance problem, while keeping the surface Rossby number and the background-flow stability fixed. To reflect differences in crops and vegetation, heterogeneous boundary conditions for potential temperature and CO_2 flux are prescribed by introducing a patch in the centre of the domain, having higher temperature or CO_2 flux than the surroundings. In the classical homogeneous temperature setting, increased CO_2 flux in the patch leads to the development of a (shallow) internal CO_2 boundary layer (IBL) over the patch, with a classical decoupling between the ground flux and the flux above the IBL. The introduction of locally higher temperature in the patch leads to increased CO_2 fluxes. Even in case of a homogeneous CO_2 flux distribution, the vertical turbulent flux increases by up to 50 CO_2 profile by improved turbulent mixing. When both heterogeneous temperature and CO_2 fluxes are combined, we find that both effects compete. We further find that the introduction of surface temperature heterogeneity leads to the emergence of strong secondary motions at the spanwise patch edges. However, a detailed CO_2 budget analysis reveals that these motions are only important for flux balances that include the patch edges. Closer to the centre of the patch the dominant mechanism relates to the development of an internal temperature and CO_2 boundary layer over the patch.

Although a complete characterisation of the probability distribution in the phase space of turbulent flows remains elusive, accurately sampling this distribution is essential for both synthetic turbulence generation and turbulent flow reconstruction. Motivated by these applications, we examine to what extent a machine-learned distribution can approximate the physical invariant distribution of turbulent channel flow at Re_τ=180. We assess three important properties of the approximation: physical ensemble statistics, consistent conditional sampling, and dynamical invariance. To this end, a flow-based generative model is trained on a minimal conditional flow unit, which we define as the smallest domain outside which conditional fields, given a single observation at the domain centre, are indistinguishable from unconditional fields in terms of mean-square discrepancy to other conditional fields. We also introduce a consistent procedure for sampling from the conditional learned distribution. Comparisons with direct numerical simulation show that synthetic turbulent fields reproduce key statistical and dynamical features of turbulence, including intermittency and nonlinear energy transfer. The consistency of conditional sampling is demonstrated in a flow reconstruction problem, and subsequently used to generate synthetic turbulent velocity fields on a large domain. When adopted as initial conditions in direct numerical simulations, these fields yield physical and statistically stationary ensemble statistics, indicating that the learned distribution provides a good approximation to the natural distribution of the turbulent dynamical system.

We propose a data assimilation algorithm based on large-eddy simulation (LES) and a weakly constrained four-dimensional variational (4D-Var) approach, applied to the reconstruction of turbulence in a neutral atmospheric boundary layer. In contrast to the more common strong 4D-Var formulation, the state equations arise as a weak constraint, introducing a state-space modelling error that also needs to be estimated. This results in an optimization problem defined over both space and time, where model errors are represented by a space-time forcing term added to the evolution equations. The 4D-Var framework follows from a Bayesian maximum a posteriori formulation, in which the prior information leads to a regularization of the resulting optimization problem. Here, we leverage turbulence theory to approximate the prior statistics of the error sources, particularly focusing on the state-space modelling error introduced by the weak formulation. We assume that these model errors are primarily due to inaccuracies in the subgrid-scale term of the LES. Moreover, we show that it is possible to estimate the sum of subgrid-scale model and error term together, avoiding the need for a subgrid-scale model in the LES reconstruction model. The 4D-Var optimization problem is preconditioned using the Hessians of the regularization terms, and solved with a classical L-BFGS method with line search. We demonstrate the approach based on virtual lidar measurements that are obtained from a fine reference simulation of a pressure-driven boundary layer, and use these in the 4D-Var formulation in which a coarse-grid LES model is used for the reconstruction (also using a smaller computational domain). For the cases investigated, we find that the algoritm yields super-linear convergence, with improved reconstruction accuracy (up to 40 %) compared to the strong formulation.

Abstract Large offshore wind farms represent significant obstacles to airflow, leading to mesoscale interactions with the atmosphere, in addition to the traditional wake effects. In earlier works, an Atmospheric Perturbation Model (APM) consisting of three layers was developed to account for both effects. Recently, we have developed a more modular N -layer generalization of the APM. In the present work, we evaluate the performance of this new model under realistic conditions by conducting a case study of the Belgian offshore zone. The model is run based on weather forecast inflow conditions for the year 2021 and the results are compared against operational data. While the original Three-layer model fails in 14% of the cases, due to atmospheric states that do not meet the model requirements, the new model provides power predictions in 99.5% of the cases considered. Moreover, the monthly mean absolute error is consistently reduced throughout the year compared to a simple wake model and to the Three-layer model. Finally, we show that the model generally captures detrimental mesoscale effects, but misrepresents the favorable ones.

The Science of Making Torque from Wind, commonly known as the Torque conference, is one of the two flagship conferences of the European Academy of Wind Energy, organized biannually, alternating with the Wind Energy Science Conference. The first ever Torque conference was organized in 2004 at TU Delft, making the 2026 event the 11 th Torque conference in the series. Torque2026 is organized June 3–5 in Bruges, Belgium, and is taking place at the renowned concert hall. This year’s conference is jointly organized by the University of Leuven, Vrije Universiteit Brussel, and Ghent University. Historically, the Torque conference started as an aero-mechanical conference with focus on the turbine – the choice of the moniker Torque over the more logically sounding “making power from wind” originally tried to reflect that sentiment. Nowadays, the Torque conference has grown in size but also in scope, reflecting the range of disciplines and scales that are of importance, while keeping the core focus on the more technical wind-energy disciplines. To reflect this scientific diversity, the conference is organized into eight scientific themes: Wind resource, metocean & extreme conditions (Theme I); Wind farms and wakes (Theme II); Aerodynamics, aeroelasticity, and aeroacoustics (Theme III); Turbine technology, energy conversion, drive train (Theme IV); Reliability, monitoring and sensing, O&M (Theme V); Structures, structural integrity, materials (Theme VI); Floating Wind (Theme VII); and Emerging technologies, VAWTs, small wind turbine, airborne wind energy (Theme VIII). For the Torque2026 edition, a call for 4-page abstracts was launched in Summer 2025. In total, 627 abstracts were submitted and reviewed by at least two anonymous reviewers. Based on this review process, oral and poster presentations were selected, and authors were invited to submit a full 10-page proceedings paper, taking into account the prior review feedback. These papers were reviewed again by at least two anonymous reviewers with an additional round for major or minor changes, with the option to extend the paper up to a maximum of 11 pages in total to be able to take into account comments and suggestions from the reviewers. Finally 494 papers were accepted for publication in the current conference proceedings, of which 67 in Theme I, 142 in Theme II, 69 in Theme III, 40 in Theme IV, 76 in Theme V, 23 in Theme VI, 47 in Theme VII, and 30 in Theme VIII.

Abstract This paper applies four-dimensional variational data assimilation (4D-Var) and large-eddy simulation (LES) for turbulence reconstruction based on scanning lidar measurements from the Belgian wind farm clusters in the North Sea. Using a Bayesian approach, the 4D-Var problem is formulated as a PDE constrained optimization problem regularized based on the Hunt-Graham-Wilson model ( J. Fluid Mech ., vol. 981, 2024, A28) for the two-point correlation tensor, where the LES is imposed as a strong or weak constraint. To enable assimilation over longer periods, 4D-Var problem is incorporated in a receding-horizon framework, where a logarithmic profile is subsequently estimated and imposed as background flow. For the data acquisition, 3D Scanning Doppler lidars are considered, focusing on the upstream flow (no wind turbines or wakes) and considering neutral conditions. The reconstructed flow displays coherent turbulent structures and achieves satisfactory reconstruction accuracy compared to the reference line-of-sight wind velocity measurements, where weak 4D-Var generally achieves lower reconstruction errors. However, the predictive capabilities of the framework do not extend beyond the assimilation windows due to the limited data availability and simplifying modeling assumptions. Nevertheless, the current work can be considered as a first step towards LES-based 4D-Var in practice.

Abstract As the size of the wind farm increases, the blockage effect associated with self-induced atmospheric gravity waves becomes increasingly important. To model this efficiently, the Atmospheric Perturbation Model (APM) simulates mesoscale flow in the atmospheric boundary layer and is coupled with an engineering wake model to capture turbine-scale effects. In this study, we employ a blockage-corrected wake model, a faster version of predicting farm power by combining two-scale momentum (2SM) theory with the APM. Model predictions are evaluated against operational power data from the Belgian transmission system operator, considering the actual layout of the Belgian–Dutch offshore wind farm cluster in the North Sea. Five years of ERA5 reanalysis data are utilized to evaluate the impact of wind-farm-induced gravity waves on annual energy production. Results show that the blockage-corrected wake model reproduces similar trends to those observed in the operational data, based on the difference in mean power between the low and high boundary layer cases across wind speeds.

Integrating floating photovoltaic (FPV) installations into offshore wind farms has been proposed as a major opportunity to scale up offshore renewable energy generation. The interaction between these hybrid wind-solar farms and the atmospheric boundary layer (ABL) is addressed in the present study. Idealized large-eddy simulations (LESs) are used to investigate the flow through both isolated wind and solar farms, as well as combined wind-solar farms, in varying configurations and under different atmospheric conditions. When the FPV modules are arranged in long strips parallel to the flow direction, secondary motions arise due to the temperature difference between the warm FPV modules and the colder sea surface, significantly affecting the horizontal distribution of mean wind speed in the ABL. Associated downdrafts between the strips increase entrainment of high-speed momentum from above, thereby increasing the wind speed at these locations. LESs hybrid wind-solar farms reveal that this is beneficial for wind turbines when they are located between the FPV arrays, leading, for the cases that we considered, to a farm averaged power increase of up to 30% compared to an isolated wind farm. It is shown that the ratio between inertial and buoyancy forces, quantified by a heterogeneity Richardson number Rih, plays a crucial role in the formation of secondary flows and the resulting wind farm power enhancement. We further show that no significant power gains, but also losses emerge in situations when the solar panels are aligned perpendicular to the flow. Although the present results suggest a large potential for wind-solar hybridization, future research should focus on a wider range of scenarios, obtained from a realistic distribution of wind directions and speeds to quantify the potential beneficial impact on real annual wind power production.

Large secondary motions are generated in boundary layers if a surface heterogeneity of wavelengths comparable to the boundary-layer thickness exists. Examples of surface introduced a simple model for these motions in the case of neutral stratification. These dynamics can significantly change in thermally stratified flows. An extension to the model considering buoyancy effects is presented, and the impact of different simplifying assumptions is investigated (namely, background gradient versus no background gradient, and constant eddy viscosity versus height-dependent eddy viscosity). All variants of the model, including the original, are validated against the direct numerical simulations (DNSs) of Bon and Meyers [J. Fluid Mech. 933, A57 (2022)] and Bon et al. [J. Fluid Mech. 970, A20 (2023)]. The DNS data concern a stably stratified channel flow with smooth walls and heterogeneous temperature on the bottom and top boundaries. For strong secondary motions, the model predictions are improved when the effect of background velocity and temperature gradients is included along with the buoyancy term. However, adding gradients alone or buoyancy alone only drives the results far from the DNSs. On the other hand, in the case of weak secondary motions, buoyancy alone shows limited improvement when compared to the original (neutral without gradients) formulation. These background gradients add physically important coupling terms in the equations that relate to vortex stretching and transport of mean-flow temperature by the secondary motions. Finally, adding a height-dependent eddy viscosity to the model improves the representation of secondary motions close to the wall.

Detector networks that measure environmental radiation serve as radiological surveillance and early warning networks in many countries across Europe and beyond. Their goal is to detect anomalous radioactive signatures that indicate the release of radionuclides into the environment. Often, the background ambient dose equivalent rate H˙*(10) is predicted using meteorological information. However, in dense detector networks, the correlation between different detectors is expected to contain markedly more information. In this work, we investigate how the joint observations by neighbouring detectors can be leveraged to predict the background H˙*(10). Treating it as a stochastic vector, we show that its distribution can be approximated as multivariate normal. We reframe the question of background prediction as a Bayesian inference problem including priors and likelihood. Finally, we show that the conditional distribution can be used to make predictions. To perform the inferences we use PyMC. All inferences are performed using real data for the nuclear sites in Doel and Mol, Belgium. We validate our calibrated model on previously unseen data. Application of the model to a case with known anomalous behaviour – observations during the operation of Belgian Reactor 1 (BR1) in Mol – highlights the relevance of our method for anomaly detection and quantification.

Over the past few years, numerous studies have shown the detrimental impact of flow blockage on wind farm power production. In the present work, we investigate the benefits of a simple collective axial-induction set point strategy for power maximization and load reduction in the presence of blockage. To this end, we perform a series of large-eddy simulations (LESs) over a wind farm consisting of 100 IEA 15 MW turbines and build wind farm power and thrust coefficient curves under three different conventionally neutral boundary layers and one truly neutral boundary layer. As a result of the large-scale effects, we show that the wind farm power and thrust coefficient curves deviate significantly from those of an isolated turbine. We carry out a trade-off analysis and determine that, while the optimal thrust set point is still correctly predicted by the Betz limit under wake-only conditions, it shifts towards lower operating regimes under strong blockage conditions. In such cases, we observe a minor power increase with respect to the Betz thrust set point, accompanied by a load reduction of about 5 %. More interestingly, we show that for some conditions the loads can be reduced by up to 19 %, at the expense of a power decrease of only 1 %.

The growing importance of the offshore wind energy sector emphasizes the need for projections of the long-term energy yield for existing and planned wind farm installations. In the North Sea, where wind farms are already pivotal to the electricity mix of the surrounding countries, the production capacity is set to increase tenfold by 2050. Studies suggest that, by 2050, the wind climate over the North Sea basin may differ significantly from the historical climate (Carvalho et al., 2021; Hahmann et al., 2022). Here, we combine an analysis of CMIP6 projections with an ERA5-driven, mesoscale wind farm simulation to further explore the impact of near-future wind climate changes over the North Sea on the energy production. First, an ensemble of 17 GCMs is reduced to 12 GCMs based on an analysis of the ability to represent the historical wind rose at 100 m MSL (1985-2014). Next, we identify future decades for each season where the wind rose exceeds the range of the historical decadal variability. Based on these extreme wind roses, we then apply a sub-sampling to a 30-year, ERA5-driven COSMO-CLM simulation covering the North Sea and incorporating a projected, 250 GW wind farm layout. Based on the sub-sampled datasets, we then quantify the impact of these extreme 10-year wind roses on the energy production of different wind farm clusters and compare this against an historical baseline.

Recent work by has shown that wind-farm blockage introduces an unfavourable pressure gradient in front of the farm and a favourable pressure gradient in the farm, which are strongly correlated with the nonlocal efficiency and wake efficiency, respectively. In particular, the favourable pressure gradient in the farm increases the farm wake efficiency, defined as the average farm power normalized by the average front-row power. Here, we investigate the impact of blockage on wake development and the power of wind turbines using an idealized large-eddy simulation setup in which blockage conditions are artificially introduced using a rigid lid, in addition to using neutral stratification and no wind veer. We simulate both infinite and finite single turbine rows, as well as a setup with two staggered rows. Blockage strength is adjusted by varying the boundary layer height (H) and turbine spacing (S). We find that blockage strongly affects near-wake behaviour, altering Froude momentum theory, by introducing a favourable pressure difference (Delta pNW) across the turbine row. The same setup also leads to an unfavourable pressure difference (Delta pFW) in the far wake, which simply follows from the rigid-lid conditions and the change in momentum flux due to wake recovery. A strong positive correlation of -Delta pNW with both the power coefficient (CP) and thrust coefficient (CT) is observed. Specifically, as S and H decrease, -Delta pNW, CP, and CT increase. At the same time, a lower induction is observed at the rotor disc, and a lower wake deficit, in the near wake. The reduction of near-wake velocity deficit as a result of blockage also translates into lower deficits and wake widths in the far wake. When scaling the far-wake development with the initial far-wake deficit and width, we do not see a direct effect of the adverse pressure gradient on the wake recovery. However, we do see a profound effect of H on the wake spreading, with higher boundary layers leading to faster spreading. This relates to the fact that the wake can more freely expand vertically in high-boundary layer cases into a larger region of high-speed flow than for shallow boundary layers. Finally, we introduce a simplified Froude momentum balance to parameterize the relation between blockage, pressure drop, and near-wake properties and compare it to the large-eddy simulation results.

Synthetic-aperture radar images and mesoscale models show that wind-farm wakes differ from single-turbine wakes. For instance, wind-farm wakes often narrow and do not disperse over long distances, contrasting the broader and more dissipating wakes of individual turbines. In this work, we aim to better understand the mechanisms that govern wind-farm wake behaviour and recovery. Hence we study the wake properties of a $1.6$ GW wind farm operating in conventionally neutral boundary layers with capping-inversion heights $203$ , $319$ , $507$ and $1001$ m. In shallow boundary layers, we find strong flow decelerations that reduce the Coriolis force magnitude, leading to an anticlockwise wake deflection in the Northern Hemisphere. In deep boundary layers, the vertical turbulent entrainment of momentum adds clockwise-turning flow from aloft into the wake region, leading to a faster recovery rate and a clockwise wake deflection. To estimate the wake properties, we propose a simple function to fit the velocity magnitude profiles along the spanwise direction. In the vertical direction, the wake spreads up to the capping-inversion height, which significantly limits vertical wake development in shallow-boundary-layer cases. In the horizontal direction and for shallow boundary layers, the wake behaves as two distinct mixing layers located at the lateral wake edges, which expand and turn towards their low-velocity side, causing the wake to narrow along the streamwise direction. A detailed analysis of the momentum budget reveals that in deep boundary layers, the wake is predominantly replenished through turbulent vertical entrainment. Conversely, in shallow boundary layers, wakes are mostly replenished by mean flow advection in the spanwise direction.

Turbine–wake and farm–atmosphere interactions can reduce wind farm power production. To model farm performance, it is important to understand the impact of different flow effects on the farm efficiency (i.e. farm power normalised by the power of the same number of isolated turbines). In this study we analyse the results of 43 large-eddy simulations (LESs) of wind farms in a range of conventionally neutral boundary layers (CNBLs). First, we show that the farm efficiency ηf is not well correlated with the wake efficiency ηw (i.e. farm power normalised by the power of front-row turbines). This suggests that existing metrics, classifying the loss of farm power into wake loss and farm blockage loss, are not best suited for understanding large wind farm performance. We then evaluate the assumption of scale separation in the two-scale momentum theory (Nishino and Dunstan, 2020) using the LES results. Building upon this theory, we propose two new metrics for wind farm performance: turbine-scale efficiency ηTS, reflecting the losses due to turbine–wake interactions, and farm-scale efficiency ηFS, indicating the losses due to farm–atmosphere interactions. The LES results show that ηTS is insensitive to the atmospheric condition, whereas ηFS is insensitive to the turbine layout. Finally, we show that a recently developed analytical wind farm model predicts ηFS with an average error of 5.7 % from the LES results.

ABSTRACTTo counteract detrimental turbine–turbine aerodynamic interactions within large farms and increase overall power production, closed‐loop wind‐farm control strategies such as wake steering have emerged as a popular means to facilitate real‐time wind‐farm flow control. The optimal wake steering set points to maximize farm power production for a given inflow condition are generally determined using fast engineering models. However, due to a lack of fast structural models, influence of wind‐farm flow control on turbine structural fatigue and loading is generally not considered. In this work, we develop a methodology for combined power and loads optimization by coupling a surrogate loads model with an analytical quasi‐static Gaussian wake merging model. The look‐up table‐based fatigue model is developed offline through a series of OpenFast simulations, covering different operational states of a DTU 10‐MW reference wind turbine, and verified against large eddy simulations with aeroelastic coupling. Subsequently, optimal control set points for the TotalControl reference wind power plant are obtained using the analytical model and tested in a wind‐farm emulator that is based on large eddy simulations. The wake model is calibrated online using in a quasi‐static closed‐loop manner. Benefits of the closed‐loop controller are exhibited via comparison of farm performance against greedy operation and against open‐loop control results obtained without feedback or calibration. Results show that the closed‐loop control out performs open‐loop control, with farm configurations with deep turbine arrays showcasing the highest gains. Inclusion of fatigue in the cost function through the developed LUT also leads to interesting insights, with reduced blade root fatigue loading without significant decrease in power production when compared to open‐loop control. A case study is also performed, which showcases real‐world applications for the developed closed‐loop controller and the load LUT, in which the closed‐loop controller is shown to react to scenarios with turbine operation shutdown, optimizing the yaw angles online to maximize performance for the new layout.