We would like to see how energy required for heating correlates with the wind acting on a building. This requires a wind metric that can be projected onto different orientations of building faces and that can be easily aggregated to days, weeks and months.
Right now we have hourly data: average wind speed and direction. It ought to be obvious that you cannot just take a monthly average: "this month we had an average wind from the west" is insufficient.
The discussion is twofold:
Can you treat the wind as a vector that you can split into components? As in: 10 m/s from the northwest can be split into 7 m/s from the north and 7 m/s from the west.
If you can split the wind into components, how can you aggregate them?
The sum of the (hourly) speeds
The sum of the square of the (hourly) speeds (you might see this as proportional to the "force" or "pressure" the wind exerts on the building)
The sum of the cubes of the (hourly) speeds (the cube of the daily average wind speed is seen as proportional to the yield of wind turbines, we use this metric at EnergyID to build prediction models for those)
We would like to see how energy required for heating correlates with the wind acting on a building. This requires a wind metric that can be projected onto different orientations of building faces and that can be easily aggregated to days, weeks and months.
Right now we have hourly data: average wind speed and direction. It ought to be obvious that you cannot just take a monthly average: "this month we had an average wind from the west" is insufficient.
The discussion is twofold:
Can you treat the wind as a vector that you can split into components? As in: 10 m/s from the northwest can be split into 7 m/s from the north and 7 m/s from the west.
If you can split the wind into components, how can you aggregate them?
Thoughts?
(I personally think the squares are what we need)