Details

Autor(en) / Beteiligte

• Grullon Varela, Rodolfo Antonio

Titel

Particle image velocimetry sensitivity analysis using automatic differentiation

Ort / Verlag

ProQuest Dissertations & Theses

Erscheinungsjahr

2016

Quelle

ProQuest Dissertations & Theses A&I

Beschreibungen/Notizen

• A particle image velocimetry (PIV) computer software is analyzed in this work by applying automatic differentiation on it. We create two artificial images that contained particles that where moved with a known velocity field over time. These artificial images were created with parameters that we would have on real PIV experiments. Then we applied a PIV software to find the velocity output vectors. As we mentioned before, we applied automatic differentiation through all the algorithm to track the derivatives of the output vectors regarding interesting parameters declared as inputs. By analyzing these derivatives we analyze the sensitivity of the output vectors to changes on each one of the parameters analyzed. One of the most important derivatives calculated in this project was the derivative of the output regarding the image intensity. In future work we plan to use this derivative combined with the intensity probability distribution of each image pixel, to find PIV uncertainties. If we achieve this goal we will find an uncertainty method that will save computational power and will give uncertainty values with computer accuracy.