DNV GL were recent winners of an E.ON-organised competition to test accuracy in wind power computer modelling.
The achievement in predictive analysis could lead to a direct improvement in the financing conditions for wind project developers, thereby reducing the cost of electricity delivered to the grid and improving returns.
Participants in the competition performed a series of wind flow analyses on real-word wind farm sites. They were asked to make their best predictions of the wind conditions at locations corresponding to existing measurement masts. The predictions were then compared to the actual measurements; the smaller the difference, the higher the ranking in the blind test.
Jean Francois Corbett, Head of CFD wind flow modelling at DNV GL, told Power Engineering International, “The better we model the flow, the more we know exactly how much each turbine can produce. Knowing that precisely reduces uncertainty and investment risk, and this is how this innovation brings costs down.”
“That’s the main goal of flow modelling, as it leads to better financing and reducing the costs.”
Advances in Computational Fluid Dynamics have greatly aided DNV GL in their work, and that technology along with the experience of the company and its workforce, combined to provide the forecasting excellence that saw them recognised in E.ON’s competition.
Corbett said, “Computational Fluid Dynamics (CFD) models represent a more complete set of physics compared to current industry standard models. This allows us to understand more of how the air behaves than had previously been the case, and to better predict how the wind flows over a site.”
However, CFD models are “too early to release into the wild just yet”, said Corbett. “You need a good deal of technical expertise not only to operate them properly, but also to correctly interpret the vast amounts of information they produce. Placed in the wrong hands, CFD can give misleading results potentially leading to costly mistakes.”
“We had regional specialists working on the on eight different blind test sites in four different countries. Much of the time involved reworking the input data, testing for irregularities and anomalies and getting them fixed; and then pouring over the CFD model outputs to ensure the end result represented our best estimate of the flow field.”
Corbett added that the modelling the company performs aims to not only reduce uncertainty but also to design better wind farm layouts for optimal production, as well as to identify locations that could cause turbines unnecessary damage.
“It’s also about putting the turbines at the right places to produce as much as possible; producing more with fewer turbines. That increases revenue and reduces capital expenditure.”
The most attractive wind farm sites are often found at locations where the wind conditions are difficult to quantify and those sites suitability must also be factored in, in terms of the potential damage to equipment.
“Our flow modelling helps keep turbines away from areas where there are adverse flow conditions which would shake them and break them. This reduces operational and maintenance costs. ”
Earlier this year, DNV GL published a validation of their CFD model on 115 commercial sites.
[bc_video account_id=”” player_id=”” video_id=””]