Welcome to Task 31 Summary Page


Benchmarking of Wind Farm Flow Models

National Renewable Energy Center (Cener), Spain
National Renewable Energy Laboratory, United States

OA Representatives:

Javier Sanz RODRIGO
National Renewable Energy Centre (CENER)
Calle Ciudad de la Innovación 7
Tel +34 948 25 28 00
Fax +34 948 27 07 74
Email jsrodrigo@cener.com

National Renewable Energy Laboratory (NREL)
1617 Cole Blvd.
Golden CO 80401-3393
Tel +1 303 384 7081
Email Patrick.Moriarty@nrel.gov

1.0 Introduction

Since the late 1980’s with the appearance of the European Wind Atlas (1), the standard model for wind resource assessment has been Wind Atlas Analysis and Application Program (WAsP) with its Wind Atlas Methodology. The model, based on a linearization of the Navier Stokes equations originally introduced by Jackson and Hunt (1975), is meant to be used reliably in neutral atmospheric conditions over mild terrain, with sufficiently gentle slopes in order to ensure fully attached flows. Nevertheless, due to its simple usage and the increasing experience of the users with the model, WAsP has also been used in situations out of its range of applicability. 

The alternative to linear models, such as WAsP, is to retain the non-linearity of the Navier Stokes equations and simulate both momentum and turbulence with computational fluid dynamics (CFD) models adapted to atmospheric flows. Even though the computational cost is significantly higher compared to linear models, it is currently affordable for conventional personal computers. The application of CFD in wind resource assessment is still largely based on Reynolds average Navier Stokes (RANS) (2) turbulence models since large-eddy simulation (LES) (3) still remains far more costly and few academic simulations have been made in small sites. CFD models based on steady RANS simulations are being developed for wind resource assessment in order to complement linear models in complex terrain and other complex flow situations (wakes, forests, obstacles, etc).

Using CFD in operational wind resource assessment is less than 10 years old and there are currently a large variety of commercial and research models in the market. Yet, the transition from traditional linear models requires significant training and experience from the user due to the extended degrees of freedom of the CFD solver, compared with the linear model, which is more user-dependent. To overcome this difficulty, commercial CFD software developers are designing user-friendly interfaces that can emulate to some extend the traditional way of working with linear models. Research CFD models in contrast are either based on generic commercial CFD solvers or on in-house or open-source codes and are used by researchers due to their flexibility to adapt to site-specific topographic and atmospheric conditions.

As with wind modeling, wake modeling for wind turbines originated in the 1980’s with work by Ainslie (1988) (4). These algebraic models, which are still widely used for wind farm layout today, are based on simple momentum and fluid dynamic similarity theories or simplified solutions to the Navier Stokes equations. The problem with these models is that they lack many of the required physical processes needed to predict wind turbine wake behavior, which results in unpredicted wake losses by 10% in many operational wind farms.

The turbine models embedded in an atmospheric model come in many different varieties and ranges of complexity and they are used for different scales of calculations. The simplest is a drag element that extracts momentum and injects turbulence over a few simulation grid points. These models often use mesoscale models with larger domains to determine macro influences of large wind farms. The next level of complexity is blade element momentum-based models that calculate blade forces and the wake influence using a global momentum balance. The forces in these models are then distributed around a disk and the influence of axial and rotational momentum is then propagated into the wake. Such a model can also be coupled to a wake meandering model that predicts the unsteady oscillation of the wake as it moves downstream. As turbine models get more complicated, the details of the blade aerodynamics become more prevalent. Recent calculations of multiple turbine interactions have used actuator line methods, where the blades are treated as airfoils distributed along rotating lines. Various other inviscid calculations of blade aerodynamics can also be used, including panel methods and boundary element methods that directly calculate the blade forces instead of using airfoil lookup tables.

With the need to calculate viscous aerodynamics of the blades, researchers have moved into CFD modeling. As with wind models, researchers have used RANS, unsteady RANS, detached eddy simulations (DES) (which is a hybrid between RANS and LES), and even full LES of rotating blades. Researchers have also created computational domains where the rotor plane is treated as a viscous area and the downstream region treated as inviscid, which can lead to significant computational time savings. Although, typically the more detail contained in the turbine model, the smaller the simulation due to constraints of computing resources.

Common to both wind and wake modeling the model developer has to design a model evaluation strategy that proves that the model is correctly formulated (verification) and provides an accurate representation of the real world from the perspective of the intended uses of the model (validation).

Verification, validation and uncertainty quantification (VV&UQ) are fundamental problems in the development of any engineering model. This process allows a comprehensive transition from experience and test-based design to simulation-based design, producing more efficient and cost-effective design solutions (5). The adoption of VV&UQ procedures is an unresolved issue in wind resource assessment due to the inherent complexity of the system to model. The main difficulties are threefold: the domain size requires large wind tunnels and computer clusters, the wind conditions are the result of the interaction of a wide range of spatial and temporal scales, the simulation of open flow fields produces ill-defined boundary conditions.  

As stated in the COST 732 Action (2009) report on microscale model evaluation (6), there is not a distinct definition of the requirements of a validation test case dataset and the procedure to use it in a consistent and systematic way. A basic requirement for any validation exercise is that the model and the validation dataset share the same or a very similar hypothesis. This basic rule is already difficult to fulfil since most of the microscale wind assessment models are based on steady-state simulations and field measurements are intrinsically transient and modulated by mesoscale effects. Intensive filtering of the field data and ensemble averaging is often necessary in order to match the desired flow conditions. A complementary solution to this “limitation” of the field data is to conduct wind tunnel measurements at a reduced scale. The controlled environment of the wind tunnel has been a fundamental tool for validation of CFD models even if, for atmospheric flows, all the similarity criteria cannot be met at the same time.

A clever strategy for VV&UQ that combines field and laboratory measurements will be developed in this IEA Task. To this end, a set of verification and validation test cases will be selected for benchmarking of models with increasing levels of complexity. Some test cases are readily available from the literature and some others will come from experimental facilities of the partners of the project. These intercomparison case studies will produce enough background information for the discussion of the VV&UQ strategies.

2.0 Objectives and Strategy

The Task aims at providing a forum for industrial, governmental and academic partners to develop and define quality-check procedures, as well as to improve atmospheric boundary layer and wind turbine wake models for use in wind energy.  The working methodology will be based on the benchmarking of different wind and wake modeling techniques in order to identify and quantify best practices for using these models under a range of conditions, both onshore and offshore, from flat to very complex terrain. These benchmarks will involve model intercomparison versus experimental data. The best practices will cover the wide range of tools currently used by the industry and will attempt to quantify the uncertainty bounds for each types of model.

Most of the work will be organized around benchmark exercises on validation test cases. In order to facilitate the management of these exercises, the “WINDBENCH” model validation web platform will be made available by CENER, which will act as Administrator. This tool is designed such that the test case can be managed by the owner of the data, with standardized procedures on how to define a test case, schedule the benchmark exercise and administer access to the data. A set of questionnaires will compile relevant information and guide the benchmark exercises. An evaluation protocol will be agreed to by the participants and a scientific committee will be designated to supervise the correct implementation of each test case.

Wind Turbines at sea

Figure 1 : Sketch of the “WINDBENCH” web-portal for management of test cases

3.0 Progress in 2010

Task 31 was approved by the IEA Wind ExCo in October 2010. Since then, the OAs have been collecting expressions of interest from potential participants in the Task. In two months 30 organizations from 16 IEAWind member countries, research institutes, and academia have declared interest. Negotiations are under way to consolidate the operational budget of the Task, which should start effectively in the second half of 2011.

4.0 Plans for 2011 and Beyond

Besides the consolidation of the participants in the Task, 2011 will be devoted to the design of a detailed work plan. To this end an inventory of test cases will be elaborated and a schedule of basic simulations will be designed in order to get acquainted with the models and the evaluation protocol.

(1) Troen I., Petersen E.L., 1989, European Wind Atlas, Risø National Laboratory, Roskilde. ISBN 87-550-1482-8. 656 pp
(2) Silva Lopes A., Palma J.M.L.M., Castro F.A., 2007, Simulation of the Askervein Flow. Part 2: Large-Eddy Simulations, Boundary-Layer Meteorol. 125: 85-108
(3) Bechmann A., Sørensen N.N., 2010, Hybrid RANS/LES Applied to Complex Terrain, Wind Energy 13: 36-50
(4) Ainslie, J. F., 1988, Calculating the flowfield in the wake of wind turbines, Journal of Wind Engineering and Industrial Aerodynamics vol. 27; 213-224
(5) Oberkampf W.L., 2010, Verification, Validation and Uncertainty Quantification of Simulation Results, NAFEMS WWW Virtual Conference, November 15-16
(6) Britter R. and Schatzmann M., 2007, Model Evaluation Guidance and Protocol Document, COST Action 732, © COST Office, distributed by University of Hamburg, ISBN: 3-00-018312-4