Fatal error when any pencil thickness = 1 and turbines are used

[karimali5]

I added a fatal error message to igrid.F90 when the domain decomposition (either manual or automatic) result in a single-cell pencil in any direction. This will cause the simulation to fail if wind turbines are used. The reason of faliure is in the actuatorDisk_filtered.F90 file, specifically here:

! link grids and read inputs 
this%dx=xG(2,1,1)-xG(1,1,1)
this%dy=yG(1,2,1)-yG(1,1,1)
this%dz=zG(1,1,2)-zG(1,1,1)

If the y or z pencils have a single cell, the second or third lines should actually give an error because it would be accessing a location out of bound, but intel compiler returns a zero instead (for yG(1,2,1) or zG(1,1,2)). This yields a zero dy or dz. If, for example, dy=0 (which was in my case), the grid points marked as within the rotor's disk will be selected from points generated along a line rather than selecting from a rectangle.

When the simulation starts with this setup, all of the turbine's thrust is distributed along a few number of points along a line which breaks the simulation (NANs appear). I am not sure if it breaks because of high force/point or during the Gaussian kernel mapping onto the 1D line points. Either way the simulation breaks.

I added this fatal error message to stop the simulation early on if turbines are being used with a single-cell pencil. If this happens, maybe select a specific decomposition (set prow & pcol manually) rather than using the auto mode (prow=pcol=0).

[kirbyh]

I believe this was patched in this PR, could you try pulling recent changes from igridSGS-KSH into your branch and seeing if that fixes the problem?