Problems with 6 dof model

[jguarato]

Hi @raphaeltimbo,

As I mentioned in Issue #1034, I decided to conduct the same analyses performed in the ROSS tutorial part 2 🔗 for the 6 dof model. The objective was to identify where errors could arise within the code. As some problems were indeed found, I opened this new issue to report them.

[jguarato]

Upon executing part 2 of the tutorial, I encountered the same error when invoking run_static() and run_ucs():

in Rotor.__init__(self, shaft_elements, disk_elements, bearing_elements, point_mass_elements, min_w, max_w, rated_w, tag)
    209 if len(tags_list) != len(set(tags_list)):
    210     raise ValueError("Tags should be unique.")
--> 212 self.number_dof = self._check_number_dof()
    214 ####################################################
    215 # Rotor summary
    216 ####################################################
    217 columns = [
    218     "type",
    219     "n",
   (...)
    237     "tag",
    238 ]
...
    562         "The number of degrees o freedom of all elements must be the same! There are BEARING elements with discrepant DoFs."
    563     )
    565 return int(number_dof)

Exception: The number of degrees o freedom of all elements must be the same! There are BEARING elements with discrepant DoFs.

which can be fixed by replacing the call of BearingElement()class with elm.__class__(). After fixing that, I compared the results obtained from the 4 dof model with the 6 dof model, and observed the following disparities:

• Running modal analysis:
  
  

• Plotting Campbell diagram:
  

• Plotting the deflected shape for the unbalance response:
  

• Plotting the deflected shape for the unbalance response:
  

[jguarato]

So, I started debugging the result classes in results.py to check if there was any coding that fixed the dof number to 4. Therefore, I needed to make some changes mainly in Shape and ForcedResponseResults. After that, I got the following results for the 6 dof model:

[jguarato]

Investigating the code I realized that there was still a problem related to the 6 dof model. I started debugging the matrices of the shaft elements by comparing them with the matrices of the 4 dof model. Consequently, I noticed that some signs in K matrix and M matrix did not coincide:

After correcting the corresponding signs, the following results were obtained:

[jguarato]

Finally, I would like to highlight two points:

• Apparently, most of the identified issues have been fixed. However, there is still an issue related to the UCS map. When running the UCS map for the 6 dof model passing different numbers of modes to the run_ucs function, the curves that appear on the map are not always smooth as for the 4 dof model. I will continue to investigate this.
  

• When I compared the M matrix of the shaft element of both models, I noticed another difference between them. In the 4 dof model, the shear effect is taken into account with the consideration of phi variable, however, in the 6 dof model, this effect is considered only in the K matrix but not in M matrix.

[jguarato]

Apparently, the problem in the UCS map has already been resolved with the correction applied in PR #931.

[jguarato]

n_links still have to be implemented in bearing element matrices for the 6 dof model. When running some analyses, I got errors related to singular matrices.

I was running the following example of the 4 dof model, but which I converted to the 6 dof model:

n = 6

shaft_elem = [
    rs.ShaftElement(
        L=0.25,
        idl=0.0,
        odl=0.05,
        material=steel,
        shear_effects=True,
        rotary_inertia=True,
        gyroscopic=True,
    )
    for _ in range(n)
]

disk0 = rs.DiskElement.from_geometry(
    n=2, material=steel, width=0.07, i_d=0.05, o_d=0.28
)
disk1 = rs.DiskElement.from_geometry(
    n=4, material=steel, width=0.07, i_d=0.05, o_d=0.28
)
disks = [disk0, disk1]

stfx = 1e6
stfy = 0.8e6
bearing0 = rs.BearingElement(0, kxx=stfx, kyy=stfy, cxx=0, n_link=7)
bearing1 = rs.BearingElement(6, kxx=stfx, kyy=stfy, cxx=0)
bearing2 = rs.BearingElement(7, kxx=stfx, kyy=stfy, cxx=0)

bearings = [bearing0, bearing1, bearing2]

pm0 = rs.PointMass(n=7, m=30)
pointmass = [pm0]

rotor1 = rs.Rotor(shaft_elem, disks, bearings, pointmass)

First I tried the analyses: run_static(), and run_modal(). In both cases, the singular matrix error appeared, because the K matrix of the rotor for 6dof model is singular.