Hash
Member level 1
Dear HFSS users,
A colleague of mine spotted that when exporting fields quantity on a surface or a volume using the field calculator (using the "Write" button), the ascii file generated by HFSS contains many duplicate points (same (x,y,z) points and more disturbing, different values associated!). According to the ANSYS support this is a normal behaviour. However, depending on the geometry, the dispersion between the values for a same point can be very large! So if you use these files as input of other processing, you may have surprises.
HFSS uses tetrahedrons, hence a point in a mesh is connected to either to 6 vertices, either to 2 on edges. And indeed, searching duplicates entries in the .fld file returns either 6 or 2 duplicates per point (but sometime other number of duplicate... I don't know why). The HFSS doc says that a quantity is stored from tangential values at vertices, so I understand that the given values in the .fld files correspond to these tangential values.
So, I would like to ask you:
1) are you aware of this behaviour?
2) how to you deal with it?
To illustrate the problem, here is a short Python script to analyze a HFSS export file (.fld file) :
For this example, I obtain 33% of points which are duplicates !
Analysis the data, most points are duplicated twice of 6 times, but not only :
The worst things is that the duplicated values are not the same !
Sometime the relative error between the mean of these duplicate and the duplicate points is negligible :
but sometime not, depending of the geometry !!
A colleague of mine spotted that when exporting fields quantity on a surface or a volume using the field calculator (using the "Write" button), the ascii file generated by HFSS contains many duplicate points (same (x,y,z) points and more disturbing, different values associated!). According to the ANSYS support this is a normal behaviour. However, depending on the geometry, the dispersion between the values for a same point can be very large! So if you use these files as input of other processing, you may have surprises.
HFSS uses tetrahedrons, hence a point in a mesh is connected to either to 6 vertices, either to 2 on edges. And indeed, searching duplicates entries in the .fld file returns either 6 or 2 duplicates per point (but sometime other number of duplicate... I don't know why). The HFSS doc says that a quantity is stored from tangential values at vertices, so I understand that the given values in the .fld files correspond to these tangential values.
So, I would like to ask you:
1) are you aware of this behaviour?
2) how to you deal with it?
To illustrate the problem, here is a short Python script to analyze a HFSS export file (.fld file) :
Python:
import pandas as pd
filename = 'E_inner_conductor.fld'
# import .fld file as a pandas dataframe
df = pd.read_csv(filename, skiprows=2, delimiter=' ', index_col=False, names=['x','y','z','Ex','Ey','Ez'])
# Find the number of duplicates (x,y,z) points
# keep only (x,y,z) points
df = df[['x','y','z']]
# remove duplicates, if any
df_unique_points = df.drop_duplicates()
# does it found some duplictated (x,y,z) elements?
print(f'Number of duplicate elements in {filename}: {len(df) - len(df_unique_points)} (ie. {len(df_unique_points)/len(df)*100}% of {len(df)} elements)')For this example, I obtain 33% of points which are duplicates !
Analysis the data, most points are duplicated twice of 6 times, but not only :
The worst things is that the duplicated values are not the same !
Sometime the relative error between the mean of these duplicate and the duplicate points is negligible :
but sometime not, depending of the geometry !!
