### Abstract

Original language | English |
---|

Number of pages | 18 |
---|---|

Publication status | Published - 2018 |

### Bibliographical note

IRPWIND 609795. This is an internal report and therefore not available in full text. Please contact author's or director of author's department for further information### Cite this

}

**TOPFARM Examples. Work Package 62, Deliverable number D62.9.** / Bedon, Gabriele; Rinker, Jennifer; Verelst, David Robert.

Research output: Book/Report › Report › Research

TY - RPRT

T1 - TOPFARM Examples. Work Package 62, Deliverable number D62.9

AU - Bedon, Gabriele

AU - Rinker, Jennifer

AU - Verelst, David Robert

N1 - IRPWIND 609795. This is an internal report and therefore not available in full text. Please contact author's or director of author's department for further information

PY - 2018

Y1 - 2018

N2 - This deliverable summarizes the current state of TOPFARM and presents two examples to demonstrate TOPFARM’s utility in benchmarking different wake models for different wind farms. Thanks in part to efforts in this deliverable, TOPFARM has been recently redesigned to not only be compatible with the newest version of OpenMDAO but also to facilitate the definition of user-defined cost and wake models. It is hoped that these development efforts will maximize the utility of TOPFARM into the future and also encourage collaborative research efforts between different institutions. The first example presented within this deliverable is an AEP calculation for Horns Rev with two different wake models: the GCL model (open-source) and the Fuga model (proprietary). The resulting AEP values indicate that, for this wind farm, the GCL wake model produces a substantially lower AEP. The second presented example optimizes the layout for a three-turbine wind farm with wind coming from a single inflow direction. The three turbines are initially oriented parallel to the wind flow, and the expectation is that the optimized layout will be arranged such that turbines are perpendicular to the flow. The example is simple, but the expected output is intuitive and it therefore allows a “sanity check” that the optimization procedure produces expected results. The resulting layouts were optimized with both the GCL wake model and with the Fuga wake model, and the results from both cases were as expected. The final AEP for the optimized wind farm layouts were similar. In both examples, the AEP calculation for Horns Rev and the three-turbine optimization, the GCL model was found to run substantially slower than the Fuga model (13x slower for the three-turbine optimization, 200x slower for the Horns Rev AEP calculation). This difference is due in part to a suboptimal implementation of the GCL model, but is likely primarily due to differences in the wake models themselves. Future work is underway to update the GCL wake model in the FUSED-Wake interface, as well as to modify the Fuga software to allow the extraction of turbine-specific information during AEP calculations. This will allow for better benchmarking in the future.

AB - This deliverable summarizes the current state of TOPFARM and presents two examples to demonstrate TOPFARM’s utility in benchmarking different wake models for different wind farms. Thanks in part to efforts in this deliverable, TOPFARM has been recently redesigned to not only be compatible with the newest version of OpenMDAO but also to facilitate the definition of user-defined cost and wake models. It is hoped that these development efforts will maximize the utility of TOPFARM into the future and also encourage collaborative research efforts between different institutions. The first example presented within this deliverable is an AEP calculation for Horns Rev with two different wake models: the GCL model (open-source) and the Fuga model (proprietary). The resulting AEP values indicate that, for this wind farm, the GCL wake model produces a substantially lower AEP. The second presented example optimizes the layout for a three-turbine wind farm with wind coming from a single inflow direction. The three turbines are initially oriented parallel to the wind flow, and the expectation is that the optimized layout will be arranged such that turbines are perpendicular to the flow. The example is simple, but the expected output is intuitive and it therefore allows a “sanity check” that the optimization procedure produces expected results. The resulting layouts were optimized with both the GCL wake model and with the Fuga wake model, and the results from both cases were as expected. The final AEP for the optimized wind farm layouts were similar. In both examples, the AEP calculation for Horns Rev and the three-turbine optimization, the GCL model was found to run substantially slower than the Fuga model (13x slower for the three-turbine optimization, 200x slower for the Horns Rev AEP calculation). This difference is due in part to a suboptimal implementation of the GCL model, but is likely primarily due to differences in the wake models themselves. Future work is underway to update the GCL wake model in the FUSED-Wake interface, as well as to modify the Fuga software to allow the extraction of turbine-specific information during AEP calculations. This will allow for better benchmarking in the future.

M3 - Report

BT - TOPFARM Examples. Work Package 62, Deliverable number D62.9

ER -