Periodic Reporting for period 3 - TotalControl (Advanced integrated supervisory and wind turbine control for optimal operation of large Wind Power Plants)
Reporting period: 2020-09-01 to 2022-05-31
Cost of Energy (COE) is the most important single factor in deployment of renewables in the energy system. Reduction of COE is, among other things, directly related to operational control of Wind Power Plant (WPP) as a whole and the individual wind turbines (WT) within them. In the TotalControl project (TC), the COE reduction is being pursued by developing and validating advanced integrated WPP/WT control schemes, where all essential interactions between the WPP WTs are accounted for including both production and load aspects. Optimal WPP control is traditionally formulated as a one-parameter optimization problem focusing on the WPP production only. However, ultimately the optimal WPP performance should result from a multi-objective optimization problem, where the optimal economic performance of a WPP is pursued over the WPP lifetime, conditioned on external grid demands. This is what TC is all about. The suggested integrated WPP/WT control approach seeks the optimal economical WPP revenue - i.e. the optimal economic balance between WPP power production and WPP operational costs. This is done by developing hierarchically coupled WPP and WT control schemes conditioned on a set of superior grid operator demands. In the WPP control design phase, information is only fed from the WPP controller to the individual WT controllers, whereas in online operational control available WT and WPP flow field information will be assimilated into the WPP control for optimal system performance. Furthermore, the WPP controller will also make use of current market information (e.g. energy price, demand for ancillary services etc.) as well as information about the state of individual turbines (e.g. current operational state, maintenance requirements and component lifetime comsumption) to allow COE objectives to be optimised dynamically.
Work in TC to 31-08-2020:
WP1: A reference WPP layout is designed as a reference throughout the project, and associated high-fidelity CFD flow fields produced and stored. The reference WPP also include an electro-mechanical model. Two medium fidelity models - the linear CFD RANS solver Fuga and the Dynamic Wake Meandering (DWM) model - have been updated to include non-neutral stratification and turbine yaw. Further, two simple and fast dynamic engineering WPP model has been developed and validated. Finally, machine learning has been investigated as an alternative to first-principles modelling. As for full-scale validation, 3 long-range lidars have been installed at the Lillgrund WPP - one scanning the inflow field, and the two others resolving the waked flow field inside the WPP. The measuring campaign are successfully concluded. This dataset is facilitating validation studies of both the high-fidelity CFD models and the lower fidelity models.
WP2: An open-loop WPP control optimization platform are developed giving wind farm control schedules conditioned on mean wind speed and mean wind direction. This platform uses WPP production as the objective function and optimizes individual WT de-rate. For quantification of optimal WPP control on OPEX, production, lifetime cost models and load surrogate models are needed. A cost model is developed and complemented with the surrogate model. The surrogate model is based on a huge number of aeroelastic simulations combined with the unsteady DWM flow field model. Preliminary results show an AEP increase of 2% and an estimated 5-years increase in lifetime. Moreover, techniques to measure the condition of a WT is developed. Finally, a WPP reactive power control algorithm - optimizing the reactive power dispatch between the wind turbines in a farm – is developed.
WP3: The Levenmouth WT (LWT) is used at the demonstration case. New wind turbine controller functionalities needed for WPP control of this WT is developed based on aeroelastic simulations. A set of reference loads was simulated, and the controller design finalized and prepared for field implementation on this 7MW WT - maximum power de-rating, delta control, IPC, 2P-IPC, a model predictive control scheme and algorithms for Lidar-assisted control, respectively. For load alleviation, specifications for tower-top sensor requirements for IPC is completed. For ancillary active power control, operation of an inverter as a Virtual Synchronous Machine has been analyzed. Moreover, an investigation of methods for active damping of tower vibration is completed. Experimental vise hardware needed for the Lidar installation on the LWT is produced, and installation of two Lidar’s - forward and backward facing - is completed. The yaw tests are running, and preparations for the various controller tests are largely completed. Finally, a selection of Lidar measurement data sets is ready for the CFD simulations investigating the induction zone wind field model.
WP4: Work on grid modelling is initiated. Regarding modelling, implementation of the LongSim code has been completed and coupled to the grid simulation tool KERMIT, providing a holistic model for the study of grid frequency support. A case study on the Irish electric grid is completed. A novel clustering method to differentiate local turbulence from larger-scale weather effects is developed. Finally, a fast method is developed for computing fatigue cycles based on turbulence spectra. Experimental vise, the VSM scheme is ready in the lab at SINTEF and measurements ongoing.
WP5: The goal of the WP is to raise awareness of the project results. This is done by setting up a website, designing a project visual identity and releasing a project video explaining what the TC project is about. Lastly, several publications on conferences and in journals. A newsletter is circulated on LinkedIn, and TC co-organised a Mini-Symposium on Wind Farm Control at WESC2019.
WP1: A reference WPP layout is designed as a reference throughout the project, and associated high-fidelity CFD flow fields produced and stored. The reference WPP also include an electro-mechanical model. Two medium fidelity models - the linear CFD RANS solver Fuga and the Dynamic Wake Meandering (DWM) model - have been updated to include non-neutral stratification and turbine yaw. Further, two simple and fast dynamic engineering WPP model has been developed and validated. Finally, machine learning has been investigated as an alternative to first-principles modelling. As for full-scale validation, 3 long-range lidars have been installed at the Lillgrund WPP - one scanning the inflow field, and the two others resolving the waked flow field inside the WPP. The measuring campaign are successfully concluded. This dataset is facilitating validation studies of both the high-fidelity CFD models and the lower fidelity models.
WP2: An open-loop WPP control optimization platform are developed giving wind farm control schedules conditioned on mean wind speed and mean wind direction. This platform uses WPP production as the objective function and optimizes individual WT de-rate. For quantification of optimal WPP control on OPEX, production, lifetime cost models and load surrogate models are needed. A cost model is developed and complemented with the surrogate model. The surrogate model is based on a huge number of aeroelastic simulations combined with the unsteady DWM flow field model. Preliminary results show an AEP increase of 2% and an estimated 5-years increase in lifetime. Moreover, techniques to measure the condition of a WT is developed. Finally, a WPP reactive power control algorithm - optimizing the reactive power dispatch between the wind turbines in a farm – is developed.
WP3: The Levenmouth WT (LWT) is used at the demonstration case. New wind turbine controller functionalities needed for WPP control of this WT is developed based on aeroelastic simulations. A set of reference loads was simulated, and the controller design finalized and prepared for field implementation on this 7MW WT - maximum power de-rating, delta control, IPC, 2P-IPC, a model predictive control scheme and algorithms for Lidar-assisted control, respectively. For load alleviation, specifications for tower-top sensor requirements for IPC is completed. For ancillary active power control, operation of an inverter as a Virtual Synchronous Machine has been analyzed. Moreover, an investigation of methods for active damping of tower vibration is completed. Experimental vise hardware needed for the Lidar installation on the LWT is produced, and installation of two Lidar’s - forward and backward facing - is completed. The yaw tests are running, and preparations for the various controller tests are largely completed. Finally, a selection of Lidar measurement data sets is ready for the CFD simulations investigating the induction zone wind field model.
WP4: Work on grid modelling is initiated. Regarding modelling, implementation of the LongSim code has been completed and coupled to the grid simulation tool KERMIT, providing a holistic model for the study of grid frequency support. A case study on the Irish electric grid is completed. A novel clustering method to differentiate local turbulence from larger-scale weather effects is developed. Finally, a fast method is developed for computing fatigue cycles based on turbulence spectra. Experimental vise, the VSM scheme is ready in the lab at SINTEF and measurements ongoing.
WP5: The goal of the WP is to raise awareness of the project results. This is done by setting up a website, designing a project visual identity and releasing a project video explaining what the TC project is about. Lastly, several publications on conferences and in journals. A newsletter is circulated on LinkedIn, and TC co-organised a Mini-Symposium on Wind Farm Control at WESC2019.
In TC advanced WPP controllers increasing WPP profitability and enlarging operational versatility is developed. This is done by moving WPP controller design from optimized individual WT operation to a coordinated, cooperative optimization of the overall WPP performance. Full-scale data sets resolving the waked flow field inside a WPP is recorded. This dataset promises to become a new standard in wind-farm field data. Also the flow field recordings around a large 7MW is monitored. Regarding modeling, a platform for design of optimized WPP control schedules maximizing WPP power is developed. Moreover, work to include load aspects, using a surrogate model, is ongoing. Regarding WT control, a set of novel control schedules are prepared for full-scale implementation in the LWT. Moreover, the wind farm simulator LongSim has been completed and coupled to the grid simulation tool KERMIT. Finally, a novel clustering method to differentiate local turbulence from larger-scale weather effects is developed. TC aims to develop advanced WPP control optimization for power production, fatigue loading, O&M and grid integration aspects. The multi-objective optimization of WPP energy extraction, WT loading, taking into account estimated O&M costs and timing, WPP grid losses, and electricity market prices will lead to an increase in WPP profits. Moreover, extending the WPP’s capability to provide ancillary services and actively participate in balancing opens up new operational opportunities for WPP owners, which may significantly increase future revenues in electricity markets