Posters

Abstract

The Data-driven Modified Spectral Correction approach is described herein as a potential alternative to the prevalent Extreme Wind (in this document Vref or V50 are alternatively used) Estimation methodologies in the industry. The method targets several challenges in the existing technologies, like Gumbel Method, Method of Independent storms, and classical Spectral Correction method [Larsén et.al.[1]]. As the name suggests, the Data-driven Modified Spectral Correction is a different V50 estimation approach to the traditional Spectral Correction Method described in the Larsén et.al. [1].  The following challenges are targeted and achieved through incorporating the novelty to the process: 1. The traditional V50 estimation approaches like Gumbel V50 Method and Method of Independent Storms potentially requires a long duration of data with a significantly high availability, which may not be the case for several sites assessed.   2. The method should exhibit a low degree of uncertainty on the estimations. This would also include that expectation that for N different time durations within a timeseries, from T1, T2 … TN , the estimations V50-1, V50-2 … V50-N should be in actuality very close to each other.  3. Any error in the data, viz. one sudden high value which exists in the data due to possible low quality data cleaning, should not cause an error in estimation or large uncertainty. 4. If an Extreme wind value, for example, 200-year wind, is already in the measurements, the algorithm still should be able to predict the extreme wind correctly and avoid over-estimation.  5. In the traditional Spectral correction method, the modelled spectrum is known to follow the general slope of -5/3, however the actual measured signal may not exactly be so. Thus, there might be a necessity to model the hybrid spectrum such that it is more representative of the site. 6. It is seen from Larsén et.al. that the primary source of Uncertainties related to the traditional Spectral Correction arises from the Mesoscale data and the Cross-over Frequency [which is the frequency at which the mesoscale data and modelled data overlaps in the spectral domain]. So, ideally the estimate should not be significantly sensitive to Mesoscale data and the cross-over frequency, for most practical reasons and ease of implementation. Lower uncertainty in Vref estimation drives optimum turbine and tower design, Vref being such crucial design driver, which is also aided by this novel patent-pending approach.   BIBLIOGRAPHY   1. Hansen, B. O., Larsén, X. G., Kelly, M. C., Rathmann, O. S., Berg, J., Bechmann, A., ... Ejsing Jørgensen, H. (2016). Extreme Wind Calculation Applying Spectral Correction Method – Test and Validation. DTU Wind Energy. (DTU Wind Energy E, Vol. 0098). 2. Karman T Von. Progress in the statistical theory of turbulence. Proc. Nat. Acad. Sci. USA, 1948, 34(11): 530-539