Numerical site assessment with CFD

Measurement is costly and usually limited to a small local area, whereas numerical methods for the analysis of wind conditions fully map a site. In simple terrain, the wind atlas method can be used, but in complex terrain ­– wooded, hilly, or mountainous – this method displays local weaknesses. This is where computational fluid dynamics comes in.

A key factor in the success of a wind farm is the correct assessment of the site. If there are insufficient measurements or if the terrain is too complex, numerical simulations are ideal for predicting the potential yield. Long-term data with temporal resolution can be obtained from mesoscale simulations based on analyses of global weather models.

Wake effect of a wind farm in front of a forested hill

To determine local effects, however, computational fluid dynamics is used. This allows for the details of the terrain and the local conditions to be taken into account. For these simulations, Fraunhofer IWES uses the open code OpenFOAM. It has the advantage of being parallelizable, allowing even large areas to be simulated with ease; even transient simulations are possible.

The open code enables simulation conditions to be precisely adapted to the requirements of the atmospheric simulation. For example, in addition to a mesh generator, Fraunhofer IWES has developed tools to generate wake models for wind farms (generation of wooded areas from image files, consideration of stable and unstable stratifications) and implemented them in the code. Linking the mesoscale simulations with the local (microscale) simulations is currently under development.

TerrainMesher

TerrainMesher-generated structure for the simulation in complex terrain

A number of add-ons have been developed for site assessment: for example, a mesh generator for complex terrain and the simulation of entire wind farms. The meshes, using which numerical simulations discretize the space, are vastly important to the result of the simulations. Fraunhofer IWES offers a structured mesh generator for site simulations, which is freely available to download. In the meantime, many functionalities have been added to the tool, one important one being that any 2D meshes can be used as the base mesh. They do not necessarily have to be structured, as the mesh generator is no longer based on blockMesh – hence its name “terrainMesher”. One part of the base mesh is first projected onto the terrain in STL format. Fraunhofer IWES offers a terrainMesher for site simulation:  

https://github.com/jonasIWES/terrainBlockMesher

 

Then the rest of the mesh is dynamically relaxed, which prevents excessive discontinuities and thus improves the resulting cell quality. The resulting 2D mesh is expanded to a 3D mesh of a given height. The user can set the height of the first cell via either the base or the desired grading. This is outwardly adjusted so that the maximum aspect ratio is not exceeded. The vertical splines are also dynamically optimized so that they stand orthogonally on the terrain and maintain maximum distance from one another.

Based on the wind farm optimization code from Oldenburg University “FLaP” and the open field equation solver “OpenFOAM”, Fraunhofer IWES has developed the wind farm optimization software “flapFoam”. It projects the wakes of individual turbines into a simulated or modelled wind field. The wind farm is then laid out according to the best possible result on the basis of various optimization criteria and methods.

In addition to various wake models, the program also has the ability to compute the wakes of the individual wind turbines using CFD. The code is at an advanced stage of development.

 

Efficient power monitoring with dynamic power curve

Stochastic methods provide a broad range of analysis for an environment characterized by an incoming turbulence. In collaboration with ForWind at the university of Oldenburg, Fraunhofer IWES promotes these methods, e.g. CTRW wind field model, continuous time random walk model as well as the dynamic power curve for power monitoring. A main method is the analyses of noise profiles. Different sources of noise and deterministic dynamics can be separated in a signal. This process can be applied on all kinds of data, which are influenced by deterministic and random parts.

The dynamic power curve is a quick and cost effective method to monitor the power output of wind turbines and whole wind farms. The software is based on stochastic examination method, which enables the user to determine a turbine´s power curve in only a few days by using hub anemometer and power output data. The generated data provide an overview of the wind turbine functionality for manufacturers, operators, service companies and public utilities. Deviating behavior of same type turbines is reliably detected, so that improvements of turbines and wind farms result in a reduction of yield losses.

The continuous time random walk (CTRW) model


The base of a realistic load calculation for wind turbines is a good model of the incoming wind field. However, wind field generators based on Gaussian distributions have problems to reproduce the correct distribution of changes in wind speed compared to real wind fields. In contradiction to the model, a realistic wind field comprises a large amount of incidences and its peak wind speeds outstrip the calculated Gaussian prediction by far. The “continuous time random walk” (CTRW) model is able to generate correlated wind fields with realistic conditions. In collaboration with ForWind at the university of Oldenburg, Fraunhofer IWES offers a wind field generator based on the CTRW method.

Fraunhofer IWES uses primarily the open source code OpenFOAM or the derivative FOAMextended for CFD simulations. Even though the code is open, it is quite challenging to understand the program and apply it efficiently. Fraunhofer IWES offers OpenFOAM courses for users with different levels of knowledge:

  • Introductions to OpenFOAM
  • Courses for the use of OpenFOAM in site assessment or aerodynamics
  • Courses on programing in OpenFOAM

Estimation of inter-annual power density variations

The generation of wind energy depends to a significant extent on the changing weather conditions. This is why the yield from wind energy plants and farms fluctuates year-on-year. To ensure economic operation, it must be clarified whether the full potential output over the course of the year was achieved or reduced as a result of disruptive influences such as shadowing and sub-optimal plant operation. For a number of years now indexes have been used for onshore plants in Germany which allow the achieved wind farm output to be compared against the expected output. On the basis of this, operators can identify and remedy the causes of performance losses and thus increase profitability. The “Offshore Wind Energy Index” (FROENIX) uses this enhanced methodology at offshore wind farms in the German Bight.

The Fraunhofer IWES Offshore Wind Energy Index (FROENIX) provides an opportunity to estimate the inter-annual power density variation at offshore wind farm sites. It reveals the percentage deviation of the average wind power density during one year compared to the average wind power density of the past 5 and 10 years. As such, it allows wind farm operators and owners as well as grid operators to assess their wind farms’ power output in comparison with the long-term average and to identify the causes of power losses.

The index is computed separately for each of the 13 offshore wind farm clusters (as defined in Bundesfachplan Nordsee 2012 (2012 Federal Plan for the North Sea)). It is calculated from more than a decade of mesoscale simulations of the wind conditions over the German Bight with a resolution of 30 minutes in time. The horizontal resolution of the simulations is 2.1 km. These data (time series, wind fields and statistics) can be made available on request. Furthermore, Fraunhofer IWES can also provide simulations which take the follow-on effects of large offshore wind clusters into account. The simulations did not take any wind farm effects within the German Bight into consideration. Consequently, the impact of the expansion of surrounding wind farms can be estimated by comparing the actual production data with the wind index data.

Fraunhofer IWES offers a wide range of services for the offshore wind energy industry. This includes energy yield and wind farm power performance assessments, offshore wind measurements, and numerical simulations of individual wind turbines, entire wind farms and even wind farm clusters, as well as large-scale tests of support structures and offshore construction methods.

Cluster evaluations for the years 2015 and 2014

In order to see the results, please move your curser over the cluster

Bokeh Plot

The software FOXES is a modular wind farm simulation and wake modelling toolbox which is based on engineering wake models. It has many applications, for example

  • Wind farm optimization, e.g. layout optimization or wake steering,
  • Wind farm post-construction analysis,
  • Wake model studies, comparison and validation,
  • Wind farm simulations invoking complex model chains.

FOXES is build upon many years of experience with wake model code development at IWES, starting with the C++ based in-house code "flapFOAM" (2011-2019) and the Python based direct predecessor "flappy" (2019-2022).

Documentation: https://fraunhoferiwes.github.io/foxes.docs/index.html
Source code: https://github.com/FraunhoferIWES/foxes

Optimization with the new iWOPY software: iWOPY is a meta package that provides one single interface to many different existing open-source Python optimization packages. Once you describe your optimization within the very general framework of iWOPY you can switch between the supported external libraries and their solvers by simply replacing the optimizer object in your script. Hence, iWOPY aims at simplifying complex optimization processes and offers you an easy tool for accessing a  variety of efficient optimizers.

 

 

You find all the information here: 

Documentation: https://fraunhoferiwes.github.io/iwopy.docs/index.html
Source code: https://github.com/FraunhoferIWES/iwopy
PyPi reference: https://pypi.org/project/iwopy/