Back to the programme printer.gif Print

Delegates are invited to meet and discuss with the poster presenters in this topic directly after the session 'The model chain: First steps towards tomorrow's technology' taking place on Thursday, 13 March 2014 at 09:00-10:30. The meet-the-authors will take place in the poster area.

Nikola Karadza Energy Institute Hrvoje Pozar, Croatia
Nikola Karadza (1) F P
(1) Energy Institute Hrvoje Pozar, Zagreb, Croatia

Printer friendly version: printer.gif Print


Meso to micro model results imperfections and solutions


Wind resource model results obtained by case-by-case meso to micro modelling at complex terrain can lead to divergence with insufficient stationary solutions for coherent conclusion.
Divergence and accurate wind flow simulation around boundaries of the CFD model are in our experience two most prominent problems in resolving the mesoscale wind fields to local scale by use of CFD.
Therefore, in this work we will present different testing results of several approaches to overcome those problems. All model results were compared to wind resource measurements at selected sites, and relevant Weibull distribution parameters were calculated by using a specially crafted evolutionary algorithm.


Wind mapping with resolution (200 m) acceptable for wind farm sitting, in complex terrain areas using meso to micro modelling performed on a territory of a Croatian coastal county raised some questions on reliability and validation of modeled data. Mesoscale data at some cases could not be dowscaled to micro using CFD model due to large number of simulation cases leading to divergent solutions despite reasonable number of iterations.
Another important issue addressed in this work is necessary overlapping width around smaller computational domains when dowscaling mesoscale model results using microscale CFD model.

Main body of abstract

The selected area is modelled using meso scale model MASS data and refined to micro scale resolution by CFD model WindSim.
The Initial step of the applied method was to prepare meso-scale data of MASS model for the entire area of interest and set it to be boundary and initial input data for CFD model WindSim. The custom codes for preparation and post processing of the simulation data have been developed using programming language Python.
Visualisation of the data and validation against measurements on several control points across entire domain, based on at least two year measurement period and long-term scaled accordingly, pointed out some further issues that needed to be resolved in post processing of the raw model results, like discontinuities along calculation domain borders.
Validation against measurements also pointed some raw model results imperfections, especially on complex terrains due to the fact that some meso scale model initial condition cases in dowscaling by CFD model WindSim failed to reach convergence on 200 m resolution. In these cases a practical solution was developed to maintain the original mesoscale data representative distribution, while introducing local orographic effects. Therefore the final model results include the assimilated replacements for non-convergent cases by adapting the initial meso scale conditions and scaling them according to microscale grid orography.
After assembling the final model results: filtered cases with convergent simulation results and the assimilated cases, the mapping process required calculation of Weibull distribution parameters for all the microscale points using a custom evolutionary computation algorithm.


In this work several weak points of meso to micro modelling in complex Croatian orography were recognized. Small overlapping areas around computational domains, when compiling the final CFD results, can lead into discontinuities and false border results (i.e. so-called bounding box effects). The solution for this problem is to select a large enough buffer zone around the results inside domain, and to disregard the overlapping areas when compiling the final results.
Another issue we addressed is the assimilation of divergent cases of meso to micro modelling by performing an orographic adaptation on micro scale according to grid point characteristics.

Learning objectives
Based on the above experiences, the paper points out various important points crucial for a successful and applicable CFD-based simulation, and also indicates several possibly problematic points related to scalability of such CFD simulation process for large areas with complex orography. Also it gives a practical guidelines on resolving two main problems, misleading border results and divergence at micro scale.