Q: Is it necessary to generate RAO after the 'g_pressure' command if I am doing a mooring in time domain?

A: No, the RAOs have no influence on the time domain.

Q: What does the message "*** ERROR: Launch Leg Has Offset" mean?

A: When performing a launch analysis, the elements that make up the launch legs must be continuous. This message indicates they are not. The usual cause of this is bad offset information for an element in the launch legs.

Q: How can I get zero peaks from the statistics of a time domain sample?

A: Finding peaks is not quite as simple as it may seem since we must worry about noise. Let

    max  = max ( val(1) - mean , .....)
    fact = 1./max
    a    = fact * ( val(i+1) - val(i) )
    b    = fact * ( val(i)   - val(i-1) )

    abs(a) >  tol  and
    abs(b) >  tol  and
    a*b    < 0

One should not be too distressed by having zero peaks since the peaks are only used in computing the statistics of them and in predicting the maximum response.

Q: How can I plot jacket buoyancy vs time?

A: Use the Disposition Menu to do this. From the Main Menu, type PRCPOST, then TRAJECTORY. Then, issue the VLIST command to display the available variables. Jacket displacement will typically be labeled something like "Displ:JACKET", but this depends on the body names in your particular problem. Now you can issue the PLOT command with the appropriate variable numbers. For instance, "PLOT 1 24 -NO" will work, if variable number 24 corresponds to your jacket displacement.

Q: I am performing a time domain vessel motions analysis, and I am having problems like "Simulation Terminated Due To Instability" and coprocessor faults. What am I doing wrong?

A: You need some soft mooring lines to hold the barge in proximity of the waves for the time domain simulation. You need 4 lines, one at each corner, at 45 degrees off the longitudinal, similar to how a model test tank would hold the barge in the middle of the tank during a model test. Mooring lines need to be soft enough so as not to effect the motions, but stiff enough to keep the barge from drifting.

Q: If the vessel heading changes during a time domain analysis, how does MOSES compute wave forces on the vessel for the changed directions?

A: MOSES will interpolate between the headings used for computing hydrodynamic pressures on the hull. Thus, it is important to use a sufficient number of headings on the G_PRESS command.

Q: How can I perform a time domain simulation in a new process which starts at the last event in an existing process?

A: Use

     &DESCRIBE PROCESS NEW -EVENTS

Q: What do you mean by "length of leg on deck" In the jacket launch analysis output?

A: Length of leg on deck is the distance from the rocker pin to the trailing edge of the jacket. This is always measured along the of the jacket leg, regardless of jacket orientation. To determine when a jacket support point passes the rocker pin, you need to equate the length of leg on deck to the distance from the trailing edge to the support point. The string function &POINT(D_NODE *N1 *N2) is useful for this. If you were to use the automated launch macros, the load cases would be generated automatically for you, as each support point passes the pin, and as each midspan point passes the pin. If you allow the simulation to go beyond separation, the macro also creates load cases at maximum post separation velocity, to capture drag dominated members.

Q: How can I position a jacket so that the center of gravity (which is off center) is at the center line of the barge?

A: You need to specify the proper Y coordinates on the LLEG command for the beginning of the skidway, and the rocker pin using the -TPIN option. For example:

         LLEG *j1 *j2 barge 0  25+1 25 *B@ -tpin 5 300  25+1 20
         LLEG *j3 *j4 barge 0 -25+1 25 *B@ -tpin 5 300 -25+1 20

will shift the skidways one unit to starboard.

Q: Why does the results I get from my launch using &INSTATE -EVENT T followed by &STATUS XXX differ from what is reported in PRCPOST?

A: What is reported with &STATUS is for a static situation, so it is not really surprising that these results differ from those during the dynamics! This statement is valid regardless of the type of time domain simulation, but particularly for a launch. During a launch, various connectors interact and for MOSES to know the complete story, one must look at the history. For example, suppose that you selected an event after tipping. Here the pitch of the barge is different from that of the jacket. The launch way connectors will try to make the two pitches equal and what you get may have no relation to what is actually going on during the simulation.

In conclusion, &INSTATE -EVENT gets the state information at a particular event and set it as the current condition. If there is no dynamics going on, then this is a reasonable approximation to the dynamic state. If not, however, anything you get with &STATUS will be rubbish.

Q: The barge I'm considering for my jacket launch analysis has only the primary rocker beam. The total length of this beam is 15.25 m (ie 6.1m from from pin aft and 9.15m from pin forward). If I want to use a conservative load distribution of 10% 80% 10% for my launch stress analysis, what value of my "tblen" in the "-LLFOR" option should be. Is it 6.1m or 15.25/2 ?

A: To be conservative, you should use the shorter length of 6.1 meters.

Q: If I take your advice in the question above and I use 6.1m and my QMID and QBEG are 88.9 and 11.1, and the TBLEN is not 1/2 of my rocker beam (15.25/2), how will MOSES distribute the loads at the beginning of the rocker beam?

A: The loads will be distributed according to the documentation in the User's Reference Manual. Specifically, the load is distributed according

using a trapezoidal distribution. 88.9% of the load will be at the pin, with 11.1/2 at the forward end of the rocker arm, and 11.1/2 at the aft end of the rocker arm. The total length of the rocker arm will be assumed to be 6.1*2 meters, or 12.2 meters.

Q: What causes the roll and yaw angles to change radically when a body nears a pitch angle of -90 degrees?

A: What you are seeing is correct and there is no fix for it! This is a problem which has plagued all for centuries: any set of Euler angles has a "singularity". For our set the singularity is at a pitch of 90 degrees.

At a pitch of 90 degrees, there is no unique way to compute roll and yaw. As a result, when you pass through this angle, "strange things appear to happen", the roll and yaw angles may change by 180 degrees. Nothing, however, is really happening - as a plot of the trajectory will show. This is simply numbers changing to protect their integrity.

The only advise I can give here is to change your coordinate system so that you do not go through a pitch of 90; i.e. change the coordinate system so that during the process the structure rolls instead of pitches (use a &DESCRIBE PART XXX -MOVE).

This works so long as the pitch does not change by over 90 degrees during the simulation.

Q: Is it possible to obtain the minimum GM of the jacket during a launch simulation?

A: GM is a measure of stability and stability only makes sense with respect to a "singular point" of the equations of motion (an equilibrium position or a constant velocity state). During launch, there is acceleration and hence, GM would not be useful here.

I think that the reason you are interested in GM is to see what happens if the jacket stops during the launch. If this is the case, you can: move the jacket to a position, find equilibrium, and then see how stable it is.

Q: When I issue, the command POINTS from the PRCPOST Menu, I get things labeled "X:*NODE", ..., M-X:*NODE". I thought that the M- columns would be the angles of rotation for the body, but they appear to be translations. How can I get the angular motion of the body?

A: A quick answer is that you use TRAJECTORY. This gives you the location (including Euler angles), velocity and acceleration of either the origin or the CG of the body and some other general data.

The columns that you thought were angular motions are really the motion (displacement vector) of the point from its reference position. For more details, Click Here.

Q: I need a really small time step, but below, you said that with MOSES one can have too small a step. How can I get around the problem?

A: As was said, the problem with a small time step is that MOSES solves a system of integro-differential equations. The velocity history is what produces the radiation damping. There is a limit to the amount of history that can be stored, and with a step too small, you may be ignoring part of the history that is important. The are no good rules as to what is too small, but when you start getting "strange things" and your step is somewhat below .1 second, history truncation may be your problem.

The work around is to not use the velocity history to create the radiation damping! Simply issue the following:

       HYDRODYNAMICS
          A_PRESSURE yy zz -PERI_USE xx -FACT_CONV 0
       END

Q: How can I find out the velocity of a point during a time domain simulation?

A: You need to use the POINTS command in the Process Post-Processing Menu.

Q: I am going to do a time domain simulation of a vessel in an irregular sea. What are your recommendations as to periods used and time step?

A: In the absence of any more concrete information, I would say that the defaults should suffice. There are three sets of periods that are involved here: those specified by -PERIOD on the G_PRESS command, those specified by -PERIOD on the &ENV command, and those specified by -P_DRIFT on the &ENV command. The defaults for those specified by -PERIOD are those specified on an &DEFAULT command (in the last moses.cus file). Unless you changed them, the defaults are:

     -PERIOD     25.  20.  19.  18.  17. 16.  15.  14.5    \
                 14. 13.5  13.  12.5 12. 11.5 11.  10.5    \
                 10.  9.5   9.   8.5  8.  7.5  7.   6.5    \
                  6.  5.5   5.   4.5  4.  3.               \
     -P_DRIFT   60 70 80 90 100 120 140 160 180 200        \

The periods specified via -P_DRIFT are a more difficult question. These define the periods at which second order wave forces will be generated. The important thing here is to have some periods in the neighborhood of the "natural frequency" of the mooring system. You may need to play with these using SRESPONSE to see where to place them in the time domain.

While you need to have a time step suitably small so that the wave forces can be accurately integrated, the stiffness of your system is normally the governing factor. For many problems, a time step of .75 - 1. works quite nicely. This is for a free vessel, or one with a "soft" mooring. As the stiffness of the connection increases, the time step must decrease! There is no general rule, but normally .05 is small enough. One thing to watch out for here is that MOSES is solving integro-differential equations and keeps a limited history. As you make the time step smaller, portions of the history are neglected. Thus, with MOSES, you can have too small a time step, i.e. too small a time step may have more numerical error than a larger one.

Q: What does the message "*** WARNING: CONVOLUTION KERNEL HAS BAD PROPERTIES" mean?

A: In general, MOSES uses an inverse Fourier transform to account for the frequency dependence of the added mass and damping in the time domain. The result of this procedure is that the equations of motion are no longer differential equations, but integro-differential ones. The numerics of this inverse transform are quite sensitive to the shape of the added mass and damping. We do our best to compute the inverse, but we must deal with the data at the discrete points you used in computing the hydrodynamic pressures. Since the form of the kernel can have important consequences on the motion of the system, we check our computation. In particular, we compute the Fourier transform of the kernel and compare it with the original damping matrix. If the comparison is not good, we give this warning. Normally, the warning can be safely ignored. If, however, you get "bad results" (instabilities, large motions, etc.) you better heed the warning. The "fix" here is to not use the convolution, but in its stead, use an added mass and damping matrix at a single frequency. Normally, one chooses the frequency of the mean period of the sea being applied. Also, using a smaller time step has been known to remove the warning message.

Q: can I change the time step in the middle of a time domain solution?

A: No. The convolution used for the hydrodynamic coefficients does not allow for this. However, any time domain simulation can be stopped and then restarted with a different time step.

Q: Is there a way to obtain the wave velocity for a wave slamming problem?

A: No, not at this time. On our list of things to do is to develop a reasonable method for dealing with wave slam.