Hi,
I have recently analyzed a complete jacket model using the Full Plastic Collapse/Pushover analysis. My main goal was to obtain the RSR value for the structure in eight wave directions, therefore I created 8 different "clpinp." files for these directions and run them one by one. However, there are some questions that I'd like to ask regarding this subject:
1- What type of loading should be exactly used as the gravitational loading? As I mentioned earlier, I used a real jacket model with real design calculations. I used a combination of dead loads (i.e. deck dead loads) and live loads using the factors that were used in the Inplace analysis.
2- In most of the directions that I run the program, the structure collapsed due to exceeding maximum displacement or rotation warning. As I understood from the SACS manual, the rotation limit is set to 2 radians and cannot be changed but the displacement criterion can be adjusted using the collapse deflection input. However, changing this input did not make any difference and the structure still collapsed with almost the same ultimate base shear. Should I conclude that it is collapsed because of exceeding the maximum rotation (i.e. 2 radians)? As a matter of fact, it collapses without a singe member or pile enters the plastic region and the plasticity ratio is extremely small. Is there an issue with the options that I'm using?
API RP 2A 21st Edition Section 17 describes reserve strength ratio calculations as it pertains to existing structures. The reserve strength ratio is calculated relative to the lateral load of a 100 year storm so I would argue that the gravity loads should correspond to the lateral load case (i.e. the 100 year gravity load combination). However, this is will be dependent upon the codes and standards used for the project, so I would discuss this with your client if you are unsure.
It's possible that the structure has reached its lateral capacity. Imagine the the lateral stiffness of the structure approaching zero so that the force-deflection curve is flat near your ultimate base shear load. I'd recommend reviewing the collapse results graphically in Precede where you can visualize the load sequence and structure plasticity, generate force-deflection curves and other curves to help you determine the failure mode of the analysis. It is possible that there are convergence issues in the analysis as it approaches failure which can contribute to a false positive on the collapse. You need to review the listing report to verify that as well.
You may also want to consider using Collapse Advanced which includes automatic load sub-incrementation and an automatic arc-length method implementation which can improve accuracy of the analysis around areas of large plastification and buckling. Check out this video for more info: 2018-07 Bentley Offshore SIG – Non-Linear Analysis with Collapse II Video. It's a bit out of date, but should give you a pretty good idea about it.
Regards,
Geoff
Answer Verified By: Behrooz Tadayon
I really appreciate your response sir. It helped me a lot. As I was checking my model, I found out that in one of the joints connecting the J-tubes to the jacket structure, a high amount of rotation or displacement occurred and the analysis was terminated due to this fact (although no significant plasticity occurred in jacket legs or bracings).
I removed these members from the model and re-run the program. However, my question is that what exactly happens when I specify that a certain group of members should be treated as elastic members (using the GRPELA command line)? Doesn't it imply that the large deflections that happen in these members should not contribute to the collapse of the structure? I'm asking this question since I had already added these members as elastic members but it seemed that their deflection was considered in the analysis after all. Or, in other cases, some joints in the deck of the platform caused one of the stiffness matrix components to become negative and the analysis was terminated. In summary, should I consider these happenings as the real collapse level of the structure (after checking the force-deflection curves and finding a flat curve, though for example one of the joints in the deck has been considered as the one which made the stiffness negative)? Can I conclude the the real capacity of this specific structure has been reached?
Ah, I see what you are doing. The GRPELA prevents the element from yielding, but does not remove it from the collapse criteria. Really, the collapse criteria uses joint displacements which are independent from the element definitions. When we define elements as elastic, Collapse will treat the elements as if the have a linear stiffness instead of a bi-linear stiffness (i.e. post yield behavior). Essentially the element will never yield or fail.
For non-critical elements like risers which can cause a false positive for collapse due to premature failure, I'd recommend utilizing Seastate dummy structures or appurtenance structures. Both of these non-structural element definitions will be deleted by Seastate when computing the hydrodynamic load on the elements, and transfer the loading to the designated structural elements. This means that the riser loading will be applied to the structure, but the riser will not be present for the non-linear analysis. See Section 2.7 and Sample Problem 2 of the Seastate manual for more on non-structural element definitions.
Regards,Geoff
I'm really grateful for your response sir. Now I truly understand what happens when I use the GRPELA line, which seems to be misunderstood by me in the beginning. Your valuable guidance is really helpful in this regard sir.
There is just another point that I have some doubt about and I'd like to mention it here as well. In some models and analyses (specially the ones used in research papers), I noticed that usually the jacket substructure is only modelled for the pushover analysis and the aim there seems to be investigating only the jacket itself. As for the other parts (the superstructure, appurtenances, ...), only the loading is applied and the elements themselves are not included in the model (not even as dummy structures. They are completely removed from the model). However, I don't aim to do that and I'd like to investigate the whole platform's capacity (including all members and piles) and as you said, several non-structural members have been assigned as dummy structures. Having all these properties, when I run my model, in most environmental directions, the collapse occurs due to a non-positive component in the stiffness matrix in the joints that exist in the deck of the platform. I doubt it whether this outcome can be considered as the ultimate strength of my platform (as I said, the whole platform, not just the jacket) or not. It is worthwhile to mention that now that the non-structural elements are not present in my non-linear analysis, some jacket bracings enter the plastic region as well (even with complete plasticity occurring in some of them) but the analysis carries on until what I mentioned about the stiffness matrix happens in the deck and the "structure collapsed" message is shown.
I'd like to apologize in advance for taking your time sir, but as I mentioned earlier, your comments are greatly helpful and beneficial and it would be an honor for me to know your opinion in this regard as well.
Behrooz Tadayon said: In some models and analyses (specially the ones used in research papers), I noticed that usually the jacket substructure is only modelled for the pushover analysis and the aim there seems to be investigating only the jacket itself. As for the other parts (the superstructure, appurtenances, ...), only the loading is applied and the elements themselves are not included in the model (not even as dummy structures. They are completely removed from the model).
As I stated in the previous post, Seastate removes the non-structural elements (dummy and appurtenance) and transfers the loading to the remaining structural elements. They are not present in the non-linear analysis, which is what you are describing here.
Behrooz Tadayon said:Having all these properties, when I run my model, in most environmental directions, the collapse occurs due to a non-positive component in the stiffness matrix in the joints that exist in the deck of the platform.
A non-positive stiffness matrix indicates that a mechanism has formed in the structure due to plasticity and/or buckling. If you are interested in the behavior of the jacket, I would recommend making the deck structure elastic so that you don't get a failure due to the deck structure. Since you are performing a pushover analysis be sure to only scale the 100 year storm loading and use a gravity load factor of 1.0. You shouldn't see any failures in the deck if you take this approach.
Another approach you could take is to model the deck structure as a superelement which will resolve the linear stiffness of the deck and its load vector down to the interface joints at the jacket. Because the deck is modeled as a superelement the joints and elements will not be considered in the non-linear Collapse analysis so they won't cause any false positive failures. This would be more accurate than simply removing the deck from the analysis as the interaction between the jacket legs would be retained. See the Superelement manual and in particular Sample 1 for more details on superelement creation.