Open william-swy opened 3 years ago
A significantly improved wind model has been added (HWM14
) using MATLAB's aerospace toolbox atmoshwm
. Major issue with runtime as it is ridiculously slow as seen in table below (takes 19 secs of the total 26 secs).
Going to need to find some solution for this as it means running the Monte Carlo simulation is going to take a super long time. Monte Carlo simulation for 1000 samples took more than 5 hrs using atmoshwm
while before a 10000 output took less than 1 hr.
Just for a reference, the original Fortran implementation of HWM14
takes 0.2 sec to run 20000 times while atmoshwm
takes 19 sec to run 18000 times. Obviously there are multiple factors in play, such as Fortran being compiled, but it is worth noting the huge difference in performance.
Currently the wind model being used is static, meaning the wind speed and direction is constant throughout the entire flight which is very unlikely to happen in reality. It would be really nice if a dynamic wind model was made and it would be even more nice if it was tuned to Truth and Consequences NM, USA for the month of December.
Look into implementing the von karman wind turbulence model.