Distributed Wind Resource Assessment for Small, Kilowatt-Sized Wind Turbines using Computational Flow Modeling Software

T. L. Acker, B. Bhattarai, R. Shrestha

Research output: Contribution to journalConference articlepeer-review

Abstract

A major challenge in deciding to invest in a wind energy system as part of an off-grid, small-scale renewable energy system is accurately estimating the annual energy production (AEP). Computational models hold promise to provide useful distributed wind resource assessment information at a reasonable cost. This paper describes the methods employed and results obtained from using wind flow modeling software, in this case Meteodyn WT, combined with wind speed data to predict the AEP of a 2.4 kW Skystream 3.7 wind turbine, and compare the AEP to measurements. Results showed AEP prediction errors ranging from <5% to ∼80% depending on the nature of the wind speed data used. Using a single wind speed data source could lead to an acceptable AEP (<10% error), but could well lead to much higher errors. Two methods of addressing this problem were demonstrated: 1) average several AEP predictions made using single wind speed data sources; or, 2) use multiple data sources simultaneously when making an AEP prediction. The latter of these two appears the most promising with lower errors in AEP. Another significant result of this work was demonstrating that using NREL Wind Toolkit wind speed data can produce good results in predicting AEP.

Original languageEnglish (US)
Article number012013
JournalJournal of Physics: Conference Series
Volume1452
Issue number1
DOIs
StatePublished - Mar 3 2020
EventNorth American Wind Energy Academy, NAWEA 2019 and the International Conference on Future Technologies in Wind Energy 2019, WindTech 2019 - Amherst, United States
Duration: Oct 14 2019Oct 16 2019

ASJC Scopus subject areas

  • Physics and Astronomy(all)

Fingerprint Dive into the research topics of 'Distributed Wind Resource Assessment for Small, Kilowatt-Sized Wind Turbines using Computational Flow Modeling Software'. Together they form a unique fingerprint.

Cite this