Structural Analysis Questions

Q: What is the reason of the warning
*** WARNING: Releases Created Element Singularity, Problems May Result Due to Fixup
and how to do I solve this problem?
REV 7.06

A: This message means that you have too many releases on a beam.

Most common reason is releasing torsion on both ends. Likewise releasing Axial, y shear, or z shear on both ends could cause the problem.

I personally think only in very rare cases should there be any releases.

Q: Why is the hydrostatic bending moment different from the structural one computed from RAOs?
REV 7.05

A: The two results deserve to be different:

For the hydrostatic bending moment, MOSES assumes that the wave changes the shape of the water surface into a sine wave, finds equilibrium with this "lumpy" sea and then computes the moment.

For the hydrodynamic results, linear hydrodynamics is assumed; i.e. the sea remains flat but there is a pressure field and the sum of the forces is no longer zero, but is Ma.

Thus, on the one hand you have a nonlinear sea (but ignoring radiation and diffraction) that is equilibrated with buoyancy and the other you have a linear sea that includes all of the effects and that is equilibrated with inertia. The differences get greater the more that the mass distribution differs from the buoyancy distribution.

Q: What is the local axis system of a vertical member in a topside which is sitting on a barge?
REV 7.05

A: The answer depends on your defaults (if you have what we shipped, MOSES will have the same system as SACS), the -CA, -REFN, and -DIR_LOC options on the element definition. This is described here Probably the best way to really see that the beam is properly positioned is to look at a GL rendered picture of the system. Finally, a BEAM PROPERTIES summary will give you the cosines of the angles between the element Y axis and the part X, Y, and Z axes.

Q: What is the difference between LCASE -TIME and CASES -TIME?
REV 7.03

A: Basically, CASES -TIME combines the results of LCASE -RAO. LCASE -TIME combines the input loads which would be used for LCASE -RAO before the stiffness analysis; i.e. you can get good stuff with a nonlinear solution with LCASE -TIME, but there are no nonlineraties with CASES -TIME.

Q: How can I choose the times for load cases after separation during a launch?
REV 7.01

A: You can do them every second or two from separation until both the top and bottom have there greatest submergence. This will make sure that you have properly checked hydrostatic collapse.

Q: How does MOSES compute the membrane stress for plate elements?
REV 7.01

A: The plate is composed of either three or four constant strain triangles. The stress in each triangle is computed from the displacements at the nodes and then averaged.

Q: What does the message *** ERROR: Singularity In Stiffness At *X.XXX" mean?
REV 7.00

A: This means there is a problem with the model you used for the structural analysis. In particular, it is reporting a singularity in the stiffness matrix at the referenced node. In general, there are two classes of singularity: local and global. The local ones are caused by stiffness mismatches between elements and the global ones are caused by not completely restraining the system; i.e. here the structural system can move without causing a change in load. This often happens when one has a simple "stick" model of a vessel. MOSES adds water springs to restrain the vessel, but since all the nodes are in a line you still get a rigid singularity in RX.

Q: What does the message "*** WARNING: xxx Sig. Figures Lost At *X.XXX" mean?
REV 7.00

A: This is an indication of problems with the model, related to relative stiffness at the referenced node. By looking at all the elements that frame into this node, it should be apparent which element is causing the problem. Significant figures lost up to 4 or 5 are typically ignored, as long the overall structural results look good. Somewhere beyond 6 significant figures lost, the stiffness matrix is declared singular.

Q: What does the message "*** WARNING: Fixing Zero On Diagonal Of Stiffness At *X.XXX" mean?
REV 7.00

A: There is a zero on the diagonal of the stiffness matrix at the specified node, and MOSES is adding a large spring at that degree of freedom.

Q: What does the message *** FATAL ERROR: Nodes Are Not Connected to Rest mean?
REV 7.00

A: It means that you are trying at least two, independent structures. In MOSES the subset of the model you are trying to solve must be connected. This error normally occurs when one is trying to transport more than one piece of cargo on a single barge. To make this problem "go away" you can either make each piece of cargo a separate piece and solve them one at a time, or connect them with weak beams. Many of these details are taken care of automatically with the Automated Installation Tools. To see which elements are not connected, take the list of nodes that comes with this error message, create a selection criterion. Use the criterion to show a picture of the structure: &PICT ISO -ENDS :LIST -COLOR SELECTED.

Q: Can I find out the reactions between the bottom of a jacket and the ground during a deck mating?
REV 7.00

A: If you believe that the jacket is a rigid body then the reactions are part of the analysis. To include the stiffness of the jacket you can do a "multi body" stress analysis:

        LCASE -PROCESS T1, T2, .....
This will consider the stiffness of the jacket but not any dynamics of the elements in the deck and jacket.

Q: How can I analyze a situation where extra braces are added after cargo has been placed on a barge?
REV 7.00

A: I suppose you are asking how you can get a solution in the time domain which has the effect of the extra braces on dynamics, but not the incorrect effect on statics. This can be accomplished as follows:

       ~BRACE TUBE 36 1
       BEAM ~BRACE *B1 *C1
        LCASE -PROCESS T010  10
       CASES -COMBINE A010 T010 1. BEFORE 1. AFTER -1
       BEAM CODE -LOAD A010
Here we make three trips through the structural solver:
  • The first captures the correct static situation before the braces are added,
  • The second captures the incorrect static situation after the braces are added (the braces have static load here), and
  • The last captures the incorrect dynamics.
The CASES -COMBINE fixes the problem. Here, the "incorrect" static situation is subtracted from the dynamics and the "correct" static situation is added.

Q: I have a rectangular barge modeled as a line of beams with a uniform load distribution. Why are the structural bending moments and shears not what I get from statics?
REV 7.00

A: The difference is the way the loads are applied. The weight is divided equally among the nodes. The buoyancy forces, however, are "lumped" to the nodes at the ends of the panels. Now, suppose that you have N+1 nodes and a weight (or buoyancy) of L. We have

-L/(N+1) force applied to each node due the weight, and
for each interior nodes
L/N force due to buoyancy.
At the ends, however, you do not get to add the contribution due to a panel on each side. Here you have
L/(2N). Thus you have a net force on each node of
L [ 1/N - 1/(N+1) ]  in the interior, and
L [ 1/(2N) - 1/(N+1) ] at the ends. Using your data, this
12500 [ 1/100 - 1/51 ] = 120

Q: How does MOSES do modal analysis?
REV 6.02

A: MOSES actually does not perform modal analysis. MOSES does extract modes and their corresponding natural frequencies. To perform this task, the method of subspace iteration is used as outlined in Bathe's book, Finite Element Procedures in Engineering Analysis (ISBN 0-13-317305-4). This is a robust method which deals with the complete stiffness and mass matrices and does not require a "Guyan Reduction" for efficiency. With most other programs, the modes are the solution; in MOSES, they are the beginning. Since the mass and stiffness matrices will normally change as a simulation progresses, MOSES uses the modes as additional degrees of freedom during a simulation. In this way, the natural frequencies for the current situation are used instead of those at some nominal position.

Q: How does MOSES "inertia relieve" structural load cases?
REV 6.01

A: For all load cases except for those created using LCASE -STATIC, the loads will sum to zero. For this to happen, MOSES will compute an extra acceleration so that the total force, including this extra acceleration, sums to zero. This results in a change in force due to inertia for every little piece of the model.

Q: I have a hull hydrostatic model and a hull structural model , whereby I compare the deflections of the two models. The structural model is composed of two longitudinal beams and some transverse link beams, both made up of prismatic sections. The deflection results do not compare nicely. Why is it so ?
REV 5.10

A: Without having the data it is impossible to say exactly what is going on, but:

  • Never model the structure of a vessel as two beam (unless,of course, you are modeling a catamaran). It is impossible to get the shear and torsion correctly this way. Use either a single beam or a complete beam/plate model.
  • Is the difference a rigid motion? Longitudinal strength uses a beam pinned at the bow and stern, did you restrain your structural model the same way?
  • The answers will never be exactly the same because of the way the load is applied for a structural analysis. For more information, click here.

Q: What are the properties of the water springs that MOSES automatically generates during an upending analysis.?
REV 5.08

A: I am assuming here that you are performing a structural analysis of a jacket upend. The water springs are computed on a node by node basis, and are an approximation of the waterplane area associated with each node. Specifically, the spring constant at a node is the total waterplane area divided by the total number of nodes under water. If you do not want the water springs, add your own restraint and the water springs will not be provided by the software.

Q: I want to export the applied loads, so I issued the commands

Why do I get an error message?
REV 5.07

A: You can get this error message for quite a few reasons. In fact, what MOSES is telling you with this message is that he is missing some information. In your case, it is the load information. The EXP_ALOAD command says to export the loads for each defined load case, but you have defined none! Hence the message. You need

where what follows on the LCASE command defines the load case to be exported.