Work Plan

The project structure is composed of four work packages:

  •     WP0: Management and coordination (CENER)
  •     WP1: Benchmarking from mesoscale to microscale “wind” models (CENER)
  •     WP2: Benchmarking from near-wake to wind farm “wake” models (NREL)
  •     WP3: VV&UQ framework and user guidelines (DTU)

The majority of the technical work will be around the setting-up and running of benchmarks for model intercomparison following the procedures established in the Model Evaluation Protocol (MEP). Similarly to Task 31 Phase 1 the benchmarking work is divided into two model categories: “wind” and “wake” models. Now “wind” is related to atmospheric boundary layer modeling from mesoscale to microscale and “wake” is related to models from near wake to wind farm scales. Management and reporting on these benchmarks will be done online at the windbench.net portal. The activity will result in an update of the MEP with the definition of a methodology for VV&UQ that can be generally applied to mesoscale and microscale models. A 2nd edition of the best-practice guidelines document that accompanies the MEP will be also released.



WP1: Benchmarking from mesoscale to microscale models
This WP deals with benchmarking activities for the development of atmospheric boundary layer models dealing with atmospheric scales from mesoscale to microscale, either as separate models or as a downscaling model-chain.
Participants working with microscale models will continue the activity initiated in Task 31 Phase 1. Topics like complex terrain, forest canopies and atmospheric stability will be considered. Modelling and characterizing atmospheric stability will be an important focus of this group, to produce a more realistic description of the wind conditions beyond the surface layer. Validation will be based mostly on mean quantities from steady and unsteady (for example a diurnal cycle) simulations.
Mesoscale models will be integrated in windbench with a dedicated catalog. Benchmarks will be established to verify model sensitivities to set-up options like: initial and boundary conditions (reanalysis data, elevation and land-use databases), boundary-layer parameterization, nesting procedures and spatial resolution. Validation will be fundamentally based on: 1) episodes of short duration on well-defined synoptic and surface conditions that allow detailed insight into model performance under specific conditions 2) long-term (various months to one year) integrations of the models to retrieve wind climate statistics and produce wind maps, as the primary application of
these models.

WP2: Benchmarking from near-wake to wind farm array models
This WP deals with wake modelling at turbine to wind farm scales. Farm to farm wake modelling, notably offshore, is also a possible extension of the scope but the availability of suitable test cases for validation is much more unlikely.
Wind farm wake models will continue activities initiated in Task 31 where the focus was mainly on steady-state, neutrally-stratified surface-layer models in flat/offshore conditions. Following the development of ABL models in WP1 we shall expect benchmarks from industry data on stability effects in offshore and complex terrain conditions. Progress on the statistical blending of simulations to mimic array efficiency statistics will be also an important area of discussion.
The near and transition wake region will deserve special attention during this second phase of Wakebench. Wind farm models often struggle to simulate the power deficit at the second row of a wind farm as the incoming flow develops to a wind farm “canopy” boundary layer. Turning these “far-wake” models to “full-wake” models implies better representation of near-wake flow induced by a more realistic modelling of rotor aerodynamics. Detailed experiments from wind tunnels and lidar campaigns can be very useful for this area of research provided quality-check procedures are established for the generation of validation data from these sources. IEA Task 32 is already developing a lidar use case formalism for the selection and implementation of well-documented and repeatable lidar methods addressing complex flow data requirements. This formalism will be implemented here considering validation data requirements so as to determine which flow situations can be measured with lidar methods. A benchmark of wind tunnel experiments on a set of wind turbine configurations will be established in order to assess the dependencies of the validation data on the experimental facility and operating conditions. Comparison with similarity theory and other wind tunnel and full-scale experiments will help determining how different variables of interest relate operating conditions from the laboratory to the real world.

WP3: VV&UQ framework and user guidelines
The model evaluation protocol defined in Task 31 will be updated to cover the wider scientific scope of this new Task. Many aspects of the validation framework are still applicable but there are a number of elements that are specific to the new models, for instance, more variables of interest, new metrics, etc.
The objective of the validation process is to evaluate model performance based on error metrics and input uncertainty and build an estimate of the predicting accuracy of the model. Uncertainty quantification determines how likely a variable of interest is when certain aspects of the modelling system are not exactly known. The UQ process
determines how the input and parameter uncertainty propagates through the model to the output variable of interest and are combined with the model inadequacy. This new Task will gather experts working on UQ for wind assessment from microscale and mesoscale models to define a common integrated framework.
The results of the benchmark exercises of WP1 and WP2 will be compiled and analyzed thoroughly. The aim of this work package is also to reach consensus about best practice guidelines and procedures for modelling wind farms and reporting the wind farm energy performance prediction accuracy.

Content Visibility

Public