CAPÍTULO II.2: REVISIÓN DE PROGRAMAS DE DEPORTE ESCOLAR
FASE 6 Diseño de instrumentos de seguimiento y evaluación
8. Deporte escolar en la Comunidad Autónoma de Cataluña
In order to develop the decision-analysis framework (Figure 1 . 1 ) , the topics of wind energy resource modelling, renewable energy based distributed generation system modelling, and multiple criteria decision analysis modelling software were reviewed to assess the most suitable. Reviews of software capability were undertaken using both evaluation titles where available, and reports of software use in the literature. Where there were many software titles to chose from, as was the case with decision analysis software, a set of capability-criteria identified software that would not meet these. Therefore, the choice of software was based on capability, cost, and suitability for the decision-analysis framework.
3. 1 Wind Resource Modelling Software
The identification and subsequent assessment of potential wind turbine generator (WTG) sites normally requires some extent of feasibility study based around a wind-monitoring programme at the subject site. Wind energy resource monitoring programmes of necessity can be both long in duration and expensive relative to the cost of a small to medium sized WTG5 installation project, with much of the expense being incurred in the installation and maintenance of monitoring equipment and on the subsequent analysis of the data generated. For this reason, many potential investors sometimes either postpone or drop altogether the project (Simoes et al., 1 999).
To ensure optimal site choice while also retaining a reasonable level of cost when assessing a number of potential WTG sites, especially in complex terrain, models need to be used that can estimate the wind energy potential of the respective sites. Van Lieshout (2000), Rohatgi & Nelson (1 994), Beljaars et al. ( 1 987), Tammelin & Hyvonen ( 1 999), Watson & Landberg ( 1 999), Ayotte (1 997), Focken et al. ( 1 999) , Heinemann et al. ( 1 999), and Reid (1 997) all indicated that several computer models have been developed specifically to calculate and predict wind flows over a given area and that one or two strategically placed anemometers within the area of interest may be all that is necessary to provide the required input (Van Lieshout, 2000).
Computer modelling has rapidly evolved from the early mathematical models and much written regarding this evolution, especially for wind flow over complex terrain. Rohatgi & Nelson, ( 1 994) , Walmsley & Taylor, (1 996), and Wood (2000) provided good reviews of the chronological development of wind modelling. There are two main categories of calculation theory used in current modelling practices, the mass-consistent model, and the Jackson - Hunt model (Rohatgi & Nelson, 1 994). The operating principle of the mass-consistent model is quite simple. Wind data was used from within the modelled area to develop an initial estimate of the wind climate. This initial estimate, adjusted by way of coefficients, achieved a modelled wind
5 Small to medium sized in this context is from 0.5 kW to 50 kW. 32
Review and Selection of the Modelling Software
field that departed from the original measured wind field only enough to satisfy the conservation of mass. A key feature of the adjustment coefficients was the minimisation of the amount required to conserve the mass flow, hence they are known as mass-consistent models. The most recent application of this theory has been the codes called numerical objective analysis of boundary layer (NOABL) and the NOABL* code, which accounts for thermal stratification (ibid.) .
Mass-consistent models will not be considered further due to the lack of suitable software appropriate for use i n this study.
The second of the two models, based on the Jackson-Hunt theory (Figure 3 . 1 ), differed markedly from and has superseded the mass-consistent model (ibid. ). It attempts to solve a set of equations based around the conservation of both momentum6 and mass. The momentum-conservation calculations include a representation of the incompressible, time dependent and neutrally stratified airflow (ibid. ).
Figure 3.1 The chronological development of wind models based on the Jackson-Hunt theory. Figure 3.1 was adapted from Rohatgi & Nelson (1 994), Walmsley & Taylor (1 996), and Wood (2000).
Various site research and benchmark field measurements exercises has seen these two distinct model theories evolve in two directions of development. Data from Askervein Hill (Scotland), Kettles Hill (Canada), Blasheval (Scotland), and several other locations were utilised either to validate th� models or to develop them further (Walmsley & Taylor, 1 996; Bowen &
Mortensen, 1 996; Beljaars et al., 1 987; Wood, 2000) . As an example of such development Beljaars et al. (1 987) introduced a new linear model for neutral surface-layer flow over complex
terrain called 'mixed spectral finite-difference' (MSFD). This model was a successor to the 'Mason and Sykes 3D Jackson and Hunt' model (MS3DJH) developed by Walmsley in 1 982 which was in turn based on the '2D Jackson and Hunt' theory of 1 975 and its extension to 3D by Mason and Sykes i n 1 979. The contiguous development of the Jackson and Hunt based models (Figure 3. 1 ) indicated the various branches that the research has taken over time and 6 This is based on the Navier-Stokes equation (Rohatgi & Nelson, 1 994).
h ow the theory was now a component in many of the numerical models developed in the last 30 years.
3. 1. 1 WAsP
The Wind Atlas Analysis and Application Programme (WAsP), known initially as the BZ-WAsP model normalises wind-speed and direction data relative to the roughness and site obstacles at the reference wind-monitoring site. This normalised wind climate was then used to estimate the wind climate at other sites using their site-specific roughness and obstacle inputs and assumptions. WAsP has been shown to give accurate wind predictions over low, smooth hills of small to moderate slope and length that ensure attached flows (Bowen & Mortensen, 1 996). It has a zooming grid coordinate system, which is one major difference from other Jackson - Hunt models (Rohatgi & Nelson, 1 994). The Bessel expansion on a zooming grid (BZ) component was developed and added into the WAsP computer code in 1 987. This radially zooming grid has the advantage of allowing an increasingly finer spatial resolution of calculated data and terrain details in the region of interest as the radial origin coincides with the site of the data collection e.g. anemometer location or modelled site.
The accuracy of WAsP was limited where the terrain was very steep and separating flows occur, as these flows were treated incorrectly by the linear calculations. This deficiency in the WAsP model was described in detail by Bowen & Mortensen (1 996) (Appendix B - 1 4. 1 ). In addition, WAsP does not consider any potential large-scale stratification due to thermally driven wind flow systems. In acknowledging this limitation, Farrugia & Scerri (1 999) considered that the use of WAsP would still save money and avoid time-consuming monitoring programmes in a site prospecting exercise. This was confirmed by Hansen & Mortensen (1 999) who used WAsP modelling for micro-siting and wind fClrm layout optimisation after a five month period of measurements using calibrated site parameters and reference data. The cost of WAsP software was NZ$7000 in 2003.
Known Limitations of WAsP
As previously indicated, WAsP was designed to model the wind climate over relatively flat or gently hilly terrain but in many situations, the terrain was not as described and therefore WAsP was often used outside of its design performance envelope (Bowen & Mortensen, 1 996; Bowen & Mortensen, 2004; Frank, 1 999). Using the WAsP model this way could result in errors, and this was especially so when predicting the wind climate from one terrain type into another terrain type. Errors can also be introduced into WAsP by endeavouring to model a wind climate affected by atmospheric and terrain induced instability and stratification, diurnal sea breezes or land breezes, down slope and fohn winds in mountainous terrain and the channelling of wind in valleys.
Bowen & Mortensen (2004) and Frank (1 999) indicated that a WAsP utility programme called the Ruggedness IndeX (RIX) produced a good measure of site terrain differences and, based on research by Bowen & Mortensen (1 996), was reported as a good measure of the proportionality of any error present in a predicted wind climate.
Review and Selection of the Modelling Software
Most modelling errors in wind resource modelling have their origins as either a data measurement and analysis or physical model error. Ayotte et al. (200 1 ) indicated that this
divides the error into that which is attributable to the wind model and associated parameters being used, and that which is due to the analysis method within which the measurements are examined and processed. As far as physical model error, Bowen & Mortensen ( 1 996) have
clearly outlined the origin of the accumulated error in WAsP to be dependent on the degree to which the operational performance limits of WAsP were exceeded by the atmospheric conditions at the time of data collection and the terrain over which the wind climate was modelled. A detailed analysis of the origin of such errors has been done (Appendix B - 1 4 . 1 ) ,
and Bowen & Mortensen (1 996) concluded that the "magnitudes of the individual procedure
errors depend on the degree that each site contravenes the orographic limits of the WAsP prediction modeL" Also, that the "relative sizes of the two procedure errors may be assumed to be roughly proportional to the individual site ruggedness, thus determining the accuracy and bias of the overall prediction by the WAsP modeL" This led to the development of the RIX number as an indication of the magnitude and sign of the error in any WAsP modelling (ibid. ).
Thus, to some extent, the RIX number mitigates some of the inadequacies of the model in steep or rugged terrain.
3. 1 . 2 WindScape Raptor
Ayotte & Taylor ( 1 995) describe another model based closely on the mixed spectral finite difference (MSFD) model, where in this development; a more complex and technically complete turbulence closure scheme was introduced. Steggel et al. (200 1 ) reported that several models were integrated as part of this model. The basis of the method was the regional-scale model, The Air Pollution Model (TAPM), developed by the Atmospheric Research department of the Australian Commonwealth Scientific and Industrial Research Organisation (CSIRO). Ayotte & Taylor ( 1 995) developed a fine scale model called Raptor, which in conjunction with the meteorological components of TAPM formed a suite of programmes called WindScape. Raptor is a MSFD three-dimensional model for the boundary layer flow over moderate terrain and assumes neutral stratification within the boundary layer. Being a model based on linear equations, it is restricted to flows over terrain of moderate slopes. Steggel et al. (2001 ) indicated that a non-linear version of the model (RaptorNd was under test and expected to be able to be used to model wind flow over steep slopes. The WindScape model continues to evolve as new modelling methodologies develop and validation sites become available, but "was not yet available in shrink wrapped form" (Ayotte, 2001 ). However, since April 2001 , the WindScape system has been used by the model developers to map more than 30 areas on behalf of 10 clients (Steggel et al., 2001 ).
The mechanisms of WindScape were described in some detail including the
meteorological components of TAPM with the wind flow model, Raptor:
The meteorological component of TAP M employs a terrain-following vertical coordinate system for three-dimensional simulations. The model solves the momentum equations for horizontal wind components, the continuity equation for vertical velocity, and the