Improving Wind Farm Preconstruction and Short-Term Energy Production Forecasting Using Field Data, Large Eddy Simulation and Artificial Neural Networks

dc.contributor.advisorBhaganagar, Kiran
dc.contributor.authorNielson, Jordan
dc.contributor.committeeMemberKilger, Max
dc.contributor.committeeMemberAlaeddini, Adel
dc.contributor.committeeMemberFeng, Zhi-Gang
dc.descriptionThis item is available only to currently enrolled UTSA students, faculty or staff. To download, navigate to Log In in the top right-hand corner of this screen, then select Log in with my UTSA ID.
dc.description.abstractThe atmosphere can have dramatic effects on the power output of both individual wind turbines and large wind turbine arrays (farms). The current study uses field data, large eddy simulation, and artificial neural networks to better understand how the atmosphere affects wind turbines and implements innovative methods to improve forecasting. A novel and a robust high-fidelity numerical methodology has been developed to realistically estimate the net energy production of full-scale horizontal axis wind turbines in a convective atmospheric boundary layer, for both isolated and multiple wind turbine arrays by accounting for the wake effects between them. Large eddy simulation has been used to understand the role of atmospheric stability in net energy production (annual energy production) of full-scale horizontal axis wind turbines placed in the convective atmospheric boundary layer. The current study uses Artificial Neural Networks (ANN) to generate multi parameter input models to estimate the power produced by a single full-scale wind turbine operating in the atmospheric boundary layer. The study investigates the use of atmospheric metrics as input parameters into the ANN model. The atmospheric metrics include Richardson Number, turbulence intensity, and wind shear. A three parameter (wind speed, air density, and turbulence intensity) ANN model has been developed. Comparison of the ANN model to other power curve correction techniques demonstrated an improvement in the Mean Absolute Error (MAE) of 40% when compared to the density correction (the next closest). Errors in short-term wind power forecasting cause increased need for reserves and wind curtailment. Researchers are using the power of machine learning to improve short-term power forecasts. The study investigates the use of the Rapid Refresh (RAP) numerical weather prediction data to train Artificial Neural Networks (ANN) for short-term (one to eighteen hours ahead) wind power forecasts of a wind farm. In particular, the study investigates training ANNs with forecasted data compared to training with instantaneous data. The study showed that instantaneous power predictions could be improved by 10% by including RAP assimilation data as an input.
dc.description.departmentMechanical Engineering
dc.format.extent175 pages
dc.subjectArtificial Neural Networks
dc.subjectEnergy Production
dc.subjectField Data
dc.subjectLarge Eddy Simulations
dc.subjectWind Energy
dc.subject.classificationFluid mechanics
dc.subject.classificationMechanical engineering
dc.titleImproving Wind Farm Preconstruction and Short-Term Energy Production Forecasting Using Field Data, Large Eddy Simulation and Artificial Neural Networks
dcterms.accessRightspq_closed Engineering of Texas at San Antonio of Philosophy


Original bundle

Now showing 1 - 1 of 1
No Thumbnail Available
3.69 MB
Adobe Portable Document Format