Open sdeinert opened 2 years ago
This is a fairly legthy issue, so I will need to take some time to get through to all the details. What I can already say, is that im pretty decidedly against removing the "obsolete" options for bulkhead reinforcements, since those provide as imple as possible parametrizations of established bulkhead designs, and I know of several tools, that use them. So the way I see it, we should rather consider the eta-ksi way you propose as an additional option for more complex designs.
Another point, since eta-ksi-coordinates are a known concept from the wing, where it is in the interval of [0,1], it might be confusing to have it in the interval [-0.5, 0.5] here (incidentally, I think your coordinate system origin in the sketches might be wrong). Anyway, for the walls, which are not subject to configuration symmetry like bulkheads the (0, 0) coordinate would be pretty meaningless.
Sorry, I didn't mean to step on any toes with "obsolete". I just wanted to mention that there is already something similar in CPACS which this new definition could also cover.
The main drawback on the existing reinforcement definition is that it is placed under structuralElements
so this means the reinforcements are placed without having any idea how the actual geometry of the wall or pressure bulkhead will look like, as it could be referenced by several ones. In practice this has created issues very regularly.
Usually the reinforcements are the last in line to be placed after all the intersections with other main structural components like walls or pressure bulkheads are defined. Therefore, the reinforcements need to be placed individually around the already existing structure intersections which is quite difficult to achieve with only equal distanced lattices:
In the picture above one can see that the equally spaced horizontal and vertical reinforcements (yellow) of the pressure bulkhead interfere in some areas with the walls and floors intersecting this pressure bulkhead (red lines). Also, vertical and horizontal reinforcements may intersect each other very close to the border of the bulkhead leading to tiny patches in between. These effects yield to not very reasonable structural designs when looking at it in detail. One may be able to solve this playing around with the number of reinforcements in both directions. This solution, however, would only be useful for that single bulkhead and the process needs to be repeated for all other bulkheads as well so that in the end each pressure bulkhead would reference its individual structuralElement definition. Most likely the number of reinforcements would be different in adjacent pressure bulkheads making it very complicated to define reinforcements in walls/floors connecting those bulkheads. For this reason we developed the more flexible solution described above which can also cover such equal spaced lattices.
Concerning your argument about the range of eta/xsi values, I am aware of that. The idea behind -0.5 to 0.5 was to simplify creating symmetric definitions by flipping the sign. Walls may also benefit from that, if used as floors. Here, xsi=0 would yield the fuselage center line. If 0 to 1 is a fixed requirement, this can be changed rather easily. Alternatively, eta/xsi could also be renamed in this definition, as I think they are not used anywhere in the fuselage yet. So we could keep eta/xsi exclusively for the wing and use different names for the fuselage.
You are right about the pressure bulkhead sketch, it needs updating. But I think it still illustrates more or less what is intended.
Not to worry, my toes are just fine. :) To be honest, my main concern is retaining the old definition. I completely see the use of the new definition, and welcome it as another option, but simple is better than complex, so why not keep the simple established solution and only use the complex way if we need to? Certainly, having more than one way to do things is not unprecedented in CPACS (looking at you, positionings
).
As for the parametrization of the underlying surface, I surely don't insist on the [0, 1] interval, but I agree, that a changed naming would go a long way to avoid confusion. Another thing: would it maybe be better to change the interval to [-1,1]. If I remember correctly from FEM class, it's usually either that or [0, 1] for local parameters, whereas bounds at 0.5 seem uncommon.
Removing or leaving the existing definition is equally fine for me. I just want to stress the importance of such a more detailed definition.
Regarding the renaming of eta/xsi in this definition I'd be thankful for any suggestions. Any favorite Greek letters that are not taken elsewhere yet?
We can surely change the interval from -1 to 1.
In reality large planar structural components such as fuselage walls or pressure bulkheads need to be reinforced to increase stiffness and prevent stability issues such as buckling. Such reinforcements are quite similar to wing stringers. Depending on the specific design, these reinforcements can have quite complex geometries as they need to consider other intersecting structures and cutouts. Therefore, a logic similar to wing spars or explicit wing stringers is proposed using a list of positions and segments connecting these positions.
CPACS extension proposal:
To allow modelling this in CPACS, the following extensions to the types
pressureBulkheadAssemblyPositionType
andwallSegmentType
are proposed:pressureBulkheadAssemblyPositionType
the following optional element needs to be added:The
pressureBulkheadReinforcementsType
looks like this:The
wallSegmentType
is extended with a similar optional element:with
The
pressureBulkheadReinforcementsType
andwallReinforcementsType
each contain a list of positions and a list of segments similar to the wing spar logic:Reinforcement positions are defined using relative coordinates (
eta
/xsi
) inside the shape of the parent component (wall/pressure bulkhead). For walls an additionalsubSegmentIndex
is required as they can consist of several segments. ThesubSegmentIndex
selects the wall segment relative to which theeta
/xsi
coordinates shall be evaluated. An alternative placement option toeta
/xsi
is the direct reference of another structural component intersecting the parent component (i.e. other walls, pressure bulkheads, reinforcements of other walls/pressure bulkheads, frames, stringers). In case this intersection with the parent component yields a wire, therelPos
parameter is applied to select a point along the intersection wire. In case of a point as result to this intersectionrelPos
is not required.The
fuselageStructureReinforcementSegmentType
is defined as follows:structuralElementUID
: References a profileBasedStructuralElement to assign physical properties to the segmentoffsetSide
: Defines on which side of the parent geometry face the reinforcement shall be located using an offsetsinglePosition
: Referencing only one position + an angle and a flag to which side the reinforcement shall extendpositionUIDs
: A list of position references connecting at least two positions with straight linesAnd some additional types used by the definitions above:
Geometric construction
With this definition the geometry for pressure bulkhead reinforcements can be realized like this:
The eta/xsi space is defined by the bounding box (grey dashed box) of the pressure bulkhead so that also positions outside the actual geometry could be defined. The geometry construction must be robust enough to support this.
Two options are shown above, a straight reinforcement segment with a single position definition plus an angle, and a multi-position reinforcement segment based on four positions with the last one placed outside the actual loft.
Eta
runs from -0.5 (left border) to +0.5 (right border), whilexsi
ranges between -0.5 (lower border) and 0.5 (upper border).A similar geometry construction logic is applied for wall reinforcements:![wall_reinforcements](https://user-images.githubusercontent.com/15669066/184325071-d7ab281d-ddd1-42b7-a268-8541ceff961d.png)
Eta
/xsi
locations are always relative to the bounding boxes of the respective wall (sub-)segment that contains them. The example in the figure above shows a reinforcement segment with three positions, where the last position lies in wall sub-segment 2. For each wall sub-segmenteta
ranges between 0 (start position border of wall subsegment) to 1 (end position border of wall subsegment) andxsi
from 0 ("left" border of wall subsegment) to 1 ("right" border of wall subsegment) depending on the orientation with which the wall was defined. Also in this case the geometry construction must be robust enough to support positions lying outside the shape of the wall segment but inside its bounding box.As reinforcement segments may reference common position definitions, fixed intersections between reinforcements can be defined explicitly.
An example for a pressure bulkhead reinforcement definition created with a TiGL prototype can be seen here:![pressureBulkheadReinf](https://user-images.githubusercontent.com/15669066/184327893-639e5e0e-0d37-48c4-8f6c-6ad54cf34993.png)
Obsolete CPACS definitions
With the new definitions detailed above the following CPACS definitions become obsolete:
structuralWallElementType
: RemovelongitudinalReinforcement
andlateralReinforcement
definitionspressureBulkheadType
: Remove choice for horizontal/vertical or radial reinforcements: