Frequency Domain Questions


Q: Why do I get ERROR: RAO Computation Did Not Converge when I am trying to compute RAOs?
REV 7.06

A: This error is because one of the following:

Way to big of damping,

Way to big of stiffness, or

No mass is defined.


Q: Why is there a g*sin(theta) correction and not a g*cos(theta) term?
REV 7.05

A: Think of the incline free body diagram from your college text book. The incline is the roll or pitch in our case. In the free body diagram it is just theta.

What we are trying to find is the change from one position to the next. First, we should look at the mean (position 0). Lets assume theta0 = 0. So, the equations are:

F0 = W*sin(theta0) = W*sin(0) = 0 N0 = W*cos(theta0) = W*cos(0) = W

At position 1, the equations are:

F1 = W*sin(theta) N1 = W*cos(theta)

Let's look at the change in friction force. So, dF = F1 - F0 = W*sin(theta) - 0 = W*sin(theta)

With the assumption of a small theta, sin(theta) -> theta. Therefore, dF = W * theta

or

dF = W * roll

Now, let's look at the change in the normal force. dN = N1 - N0 = W*cos(theta) - W = W*(cos(theta) - 1)

for a small theta, cos(theta) -> 1. therefore,

dN = W*(1 - 1) = 0

So, there is a g*sin(theta) correction, but the g*cos(theta) term cancels out.

The same works for pitch.


Q: I have a piece that I want to be included for hydrodyanmic calculations but I do not want it included for hydrostatic calculations. What options do I used in the definition.
REV 7.04

A: You need to give the part just a little bit of displacement. You do this by defining it with a very small permeability. A permeability of 0.001 should suffice. You can use the -PERMEABILITY option when you define the piece.


Q: Is the viscous damping linearized stochastically or harmonically?
REV 7.02

A: By default MOSES linearizes harmonically via a "equivalent linearization". If you tell it to, MOSES will perform a spectral linearization. Remember that the viscous forces depend on the relative velocity so that you are not only linearizing the damping, but also part of the force. Click here (Section III.G) or Here(Section XII) for details.


Q: How do I get only the dynamic response excluding the mean?
REV 7.01

A: Define the envinment with the option

     -USE_MEAN NO
This is an option on the &env command. If you are working with only the connector forces this is also an option on the ST_CFORCE command.


Q: Do the damping matrices obtained using the v_matrices command include both linear radiation damping and viscous damping coefficients?
REV 7.01

A: No, what you get with MATRICES in FREQ_RESP contains both, what you get with V_MATRICES in HYDRODYNAMICS is only radiation. This menu does not know about the response.


Q: How can I compute the Cargo Forces on many points and how can I find the maxima for different periods and headings?
REV 7.01

A: (This is actually a sample. Click Here for the test files.) The quickest way to do this is with loops. This file shows how to report forces for a set of points and how to report forces for each point for a set of environments. Basically, the sample is a sample of a loop within a loop. Briefly, the points are first defined and then a macro is used to loop on the point names. In the file the points are defined within the MEDIT menu. This section is labeled as "Define Report Pts". The poins can also be defined as part of the body description in the data file. A variable, which is the list of points, is made after the points are generated. In this file a set of environments is generated with a loop. This section is labeled as "Devine env". After the environments are generated the frequency response menu is used for caluations and reporting. The rao's are computed outside the loop. The rest of the computations are done within the loops. The outer loop (using p as the index) uses the point names are first used to generate the response and force operators at the location of the point. This is done with the fr_point and fr_frcargo commands. The inner loop (using e as the index) uses the list of environment names to calculate and report the forces for each point-environment combination. Calculation of the cargo forces is done with the st_fcargo command. There is a section labeled "get max" where variables are set so that the user will have the data available to sort through. The command set_var is used to retrieve the data from the database and the command &set is used to save the maximum of the values calculated to that point. Once all the point names and the environment names have been exhausted the loops are closed. Be sure to close both loops. The final section is labeles as "report". Here the maximum of the values collected earlier are printed to the screen and the log file.


Q: Is the static angle due to wind included in the "Cargo Force Statistics"?
REV 7.01

A: No, it is not. Neither is the static component of gravity. (Notice that the heave force is around .1 to .3 instead of being greater than 1) The manual clearly states that these are "DYNAMIC FORCES". To get the total force you should add the static force to the dynamic one. Notice the cargo force results are maxima for each degree of freedom. Thus you should take the mean (static force) and add/subtract each component which yields quite a few cases.

The "Cargo Force Statistics" output is generated with the ST_FCARGO command.


Q: What coordinate system is used to report the frequency domain results?
REV 7.00

A: The body system is used for almost everything is the frequency domain. The titles Surge, Sway, Heave, Roll, Pitch, and Yaw are body system terms. For forces we use Longitudinal, Transverse, and Vertical which again are body system terms. One thing that needs clarification is relative motion which is reported is the Body system of the first point.


Q: Does MOSES consider hydrodynamic interaction among multiple bodies when calculating the hydrodynamic coefficients and raos?
REV 7.00

A: The precise answer is, not in general. There are two alternatives, however. First, a body can interact with an "obstacle", or you can connect the multiple bodies, compute the modes of the assembly, and use the modes as generalized degrees of freedom.


Q: I am looking at MOSES results from the output report "CARGO G FORCE STATISTICS". Do these G forces include the g*sin(theta) effect? If so, how can the Longitudinal G force be reported as less than half of g times the maximum pitch angle?
REV 7.00

A: Of course, the G*sin(theta) effect is included in the CARGO G FORCE STATISTICS. If you examine your motion response operators, you will notice that surge and pitch are nearly 180 degrees out of phase. This means the surge component of the G force is opposing the pitch component, resulting in the lower G value. See related questions concerning phasing: Click Here for one and here for another.


Q: I have cargo placed symmetrically about the barge centerline. Why do I get very different vertical forces for the port and starboard cargo when I look at the results for a 90 degree heading?
REV 7.00

A: This is due to the heave/roll phase relationship. The vertical component of roll will change signs depending on the direction of roll. Sometimes this will be additive to heave, sometimes this will subtract from heave. This depends on the position of the cargo relative to the barge centerline, and the wave approach.


Q: Why does to force I computed on some cargo using the motions and the accelerations not agree with the MOSES force.
REV 7.00

A: The answer to your question is a word, phasing. MOSES computes the forces correctly accounting for phasing: i.e.

      fx = M INT [ { ax + g sin(pitch) }^2 s(w) dw ]
What you are doing above is applying the triangle inequality to this expression as
      fx <= M INT [ ax ^2 s(w) dw ] + M INT [ {g sin(pitch)}^2 s(w) dw ]
This is a mathematical fact, the force statistics MOSES computes will always be less than or equal what you get with a check of the motion statistics. The reason you get the same results for heave is that there is no angle involved, in this case, equality holds.


Q: If one does not specify the periods around the peak of the Roll RAO curve, will MOSES statistics of Roll motion results be inaccurate?
REV 6.02

A: In general, your statement is correct, and it is always good practice to provide a reasonable refinement of periods in the range of the anticipated peak roll response. However, this is more an issue of engineering judgment than one of running MOSES. For most common barge transports, the default encounter periods provided in MOSES will suffice. Normally, there would not be much difference in results between a run with the default periods, and a run with more periods in the region of the peak roll response. Of course, if these results are different, or your results are close to a pass/fail situation, further study would be required.


Q: I have computed pressures on the hull for encounter periods ranging from 4 seconds to 30 seconds. Can I uses these to look at a sea state with a peak period of 4 seconds?
REV 6.02

A: No, this is not good practice. You should always make sure you have computed hull pressures for several periods either side of the peak period of the spectra you expect to use. MOSES will perform an extrapolation where no data exists, but you will always get better answers using computed data.


Q: If I want to obtain the maximum roll response, do I compare the period of peak roll response with the environment peak period or the environment mean period?
REV 6.01

A: Peak period.


Q: Why is it that when I use strip theory I get a roll angle of 33 degrees and yet in diffraction I get a roll angle of 22 degrees?
REV 6.01

A: You do not have enough panels for a good diffraction solution. As a general rule of thumb, you want to have approximately 1000 panels for most monohull vessels when using 3D Diffraction. It is not surprising your 273 panel diffraction answer is different from your strip theory answer.

A final word is in order here. Strip theory is quite good for monohull vessels, particularly in roll. It is widely accepted for most deck barge transports. It is not so good for surge, and is simply not correct for multi-hulled vessels, such as semisubmersibles. The moral of this story is do not be a man with two watches, you will never know the correct time. Stick with strip theory for most common barge transports, go to 3D Diffraction when surge is important in a mooring problem, or when analyzing mufti-hulled vessels.


Q: When I lower the position of my cargo on the vessel I find that the roll angle is now increased. Why is this happening?
REV 6.01

A: Changing KG influences not just stability but also the period of peak roll response for the system. If decreasing KG moved the period of peak roll response closer to the peak period of the wave spectrum used in the analysis than that would increase the roll response. Check the roll RAO curve or table for beam seas.


Q: How can I calculate the quadratic transfer function of a vessel in MOSES?
REV 6.01

A: As I understand it, a quadratic transfer function is used to evaluate second order effects on a floating body exposed to waves. These second order effects are really slowly varying wave drift. You can view wave drift data by using the v_mdrift command. If you are expecting to take wave drift output from MOSES, and use this as input to another program, you will likely have to perform some kind of conversion for units or format. MOSES was designed to solve all aspects of floating body problems, not to directly output quadratic transfer functions.


Q: Can I get MOSES to give me a correct wave force on a tank setting on the sea bottom?
REV 6.01

A: Actually, you need to do nothing. MOSES automatically eliminates all panels below the mud line. Just make sure that the tank extends below that bottom and everything will be fine.


Q: What are the dimensions for radiation damping coefficients in the MOSES output?
REV 5.10

A: The output says "Values Normalized by Mass with Weight = xxx". What MOSES is reporting is the weight of the body in question, and needs to be converted to mass to compute a damping coefficient to be used outside of MOSES. The translational values have units of 1/sec, the rotational values have units of ft^2/sec or meters^2/sec.


Q: We have found that the solutions of the RAOs differ depending on the number of headings which were run for the G_PRESSURE command. Shouldn't they be the same?
REV 5.10

A: MOSES uses the solution for the previous heading as an initial estimate for the current heading. Thus, there will be small differences (within the convergence tolerance) in the solution depending on the number of headings. There is nothing you can do to change this fact.


Q: I am comparing motion forces obtained by using FR_FCARGO and ST_FCARGO, and defining an environment with -SPREAD 2.11. Why is there a substantial difference between these two results?
REV 5.10

A: No, you are not doing anything wrong, this is related to your use of the -spread option, and the mathematics used to synthesize the sea. The problem is that at some point in space the waves will always reinforce, sometimes they will cancel. Just imagine two regular plane waves, with the exact same period, coming from different directions. It is easy to see that there will be maximums and minimums in the ocean, spreading evenly in space, and more importantly since they have the same period they will stay stationary in space. When we transform the waves to the time domain we still get the same waves, also stationary in space with maximums and minimums.

If the characteristic periods of your sea are large enough, your vessel will have different responses depending on where it is located in space. It may be on a maximum spot, it may be on a minimum spot. It is a problem of representing the randomness of the sea with the spreading technique, which is nothing but using exactly the same sea spectrum for different headings with a scaling factor.


Q: Can I use a hull model composed of various PGEN definitions that share surfaces for hydrodynamics?
REV 5.10

A: Absolutely not! You must have a single enclosed surface for hydrodynamics computations. You should use COMPART_SUM panels to make sure your panel areas sum to zero. If you have appendages such as skegs to include for hydrostatics, use "PGEN -DIFTYPE none" to exclude these for hydrodynamics. If you want to include the effects of a moonpool on the motions, you really should use the &surface menu to create a proper 3D Diffraction mesh.


Q: What does the message "*** WARNING: Negative On Diagonal Of Damping" mean?
REV 5.10

A: It means you have negative values in either your damping values and/or your added mass values resulting from the hydrodynamic pressure computation. We discovered users were not checking their added mass and damping output, so we added this warning message. This can be caused by a bad or strange hull definition, close proximity to the bottom, or irregular frequencies.

What we mean by "bad" hull definitions are those that do not have enough detail in areas of the vessel with a change of shape, such as the bow or stern. Often, adding more planes or panels in these areas can make the problem go away. Another good check would be to square off the ends of the vessel, and see if the warning is still reported. With 3D Diffraction, adding panels is automatic by using &PARAMETER -M_DIS. Depending on the size of the vessel, a good panel size is 3 to 4 meters, and 500 to 1000 panels is a reasonable number. Some vessel shapes may require up to 3000 panels, but much more than this takes too long to compute. The optimal aspect ratio for diffraction panels is 1 to 1, meaning a square, but rectangular panels are acceptable. The program will create triangular panels when needed to avoid warped panels. The user can control the resulting mesh by careful selection of the points used to define the planes.

Strange hull shapes can include floating jetty type structures that look like an upside down "T" in cross section. Here you have a horizontal vessel surface in close proximity to the water surface, and this can cause negative added mass. In this case, the negative values and resulting motions are probably real, but thorough checking is advised. If you get the warning with both Strip Theory and 3D Diffraction, this increases the chances the answers are real.

Negative damping can also occur when the keel of the vessel is close to the seafloor. In this case, the water under the keel gets "pushed" against the bottom, and this "hydraulic spring" effect is captured in MOSES as negative damping.

Occasionally, negative damping is caused by irregular frequencies, a numerical phenomenon where the computational algorithm breaks down at particular frequencies. In essence, these are the natural frequencies at which water would slosh inside the vessel. Numerics being what it is, the problem being solved becomes "ill conditioned" in the neighborhood of these frequencies.

The onset of irregular frequencies depends on draft, length, and beam, but for a 400 X 100 foot barge it is around 3.5 seconds. For H851, is it about 6.5 seconds. More panels makes the ill conditioning better. Therefore, before you can declare you are subjected to irregular frequencies, you need to fine tune and check your model carefully. If the negative damping occurs at only one or two periods where you know there is little or no spectral energy, these periods can be safely deleted from the computed periods list. Another option is to remove the offending periods from the g_press command, but use them on the rao command. MOSES will interpolate or extrapolate results for the removed periods.

Many research papers have been written concerning this topic. One of the more useful ones is "On the Significance of Negative Added Mass", by T. Vinje, presented at the Eighth International Conference on Offshore Mechanics and Arctic Engineering, The Hague, March 19-23, 1989.


Q: How are the phases defined for RAOs?
REV 5.06

A: See the MOSES Sign Convention document.


Q: Why are the heave and rotational "unit-forces" computed from ST_POINT not the same as when computed from ST_FCARGO?
REV 5.06

A: What you are seeing here is numerical differences. In MOSES, we "work hard" to compute the moments of a spectrum. In particular, we assume that the response operator varies linearly between the points at which they are computed, and then carefully integrate this with the wave spectrum values. This is all fine and good except for your problem. You see that the acceleration is the second moment of the motion so it is computed according to the above procedure. The heave force, however, is the zeroth moment of its spectrum. Thus, according to the above rule, it is computed assuming a linear variation. Any differences should be minor. If they are not, you need to compute the results at more periods.


Q: I have a body, composed entirely of tubular members. When I compute frequency domain results for this body, I get all zeros for the added mass, damping, and exciting forces. How can this be?
REV 5.05

A: There are three reasons that you can receive a report of all zeros:

  • The body is not in the water.
  • You are looking at the wrong report. There are two reports which give the information you described. One of these is is obtained in the HYDRODYNAMICS Menu, the other in the FREQUENCY RESPONSE Menu. The one from HYDRODYNAMICS reports results only for panels. You have no panels, hence these should be zero. The report you get in the FREQUENCY RESPONSE menu will contain results for both panels and Morrisons elements.
  • You asked for the results to be reported before they were computed. Hydrodynamic forces due to Morrisons equation are computed when they are needed. Hence, there are no hydrodynamic forces on your body until after the first RAO or SRESPONSE command is issued.


Q: If I use the command FR_POINT &BODY(CG DECK) will I get the RAOs of the motion of the DECK cg in its current position?
REV 5.05

A: Maybe. The FR_POINT does yield the RAOs of a point about its mean position. These RAOs are represented in the body coordinate system. If, however, you have more than one body the syntax you described may give the results for the "wrong" body. The documentation of FR_POINT (For details, click here) warns you that if you use the syntax above, MOSES will use the "current body". This may not be the body you want unless you explicitly use the command &DESCRIBE BODY DECK immediately before FR_POINT. It is easier to use FR_POINT *DECKCG where the point *DECKCG has been defined to have the location of the CG. Also, I am not quite sure what you mean by "in the correct position". The way moses views life, points are "particles" attached to bodies, not locations. Thus, the RAOs you get are the raos of a fixed particle of a body - regardless of the location of the particle. I am afraid that you are confusing the representation of the RAOs with the location of the point. As stated above, the RAOs are represented in the body system, i.e. a heave is perpendicular to the "deck" not the waterplane.


Q: How does MOSES treat nonlinear connectors in the frequency domain?
REV 5.05

A: MOSES uses the tangent stiffness matrix at the mean position in the frequency domain. This statement applies equally well to all stiffnesses, hydrostatic, weight, etc. What is applied in the frequency domain is the derivative of the force evaluated at the current configuration when the RAO or SRESPONSE command was issued. Of course if a tension only (compression only) connector has zero load at this position, it will not contribute to the stiffness.


Q: What coordinate system is used when reporting the results of the FR_POINT command?
REV 5.05

A: The body system of the body containing the point. Normally, one should specify a point name on the FR_POINT command to define the point of interest. Thus, MOSES will get the coordinates in the system it wants, and it will know the body being considered. The results are reported in the body system. If you are interested in global motions, you can use the PMOTION command.


Q: Are the heave motions from the frequency domain coupled with rotations?
REV 5.05

A: Yes.


Q: How does MOSES predict the "maximum" response of quantities in an irregular sea?
REV 5.03

A: As stated in the MOSES Reference Manual, &DEFAULT -PROBABILITY STAT PDATA controls the statistics which will be defined when computing the statistics of quantities in an irregular sea. The default is for STAT to refer to the MAXIMUM, and PDATA is 3.72 which provides the 1/1000 highest response, based on a Rayleigh distribution. This is the statistical multiplier the root mean square will be multiplied by to obtain a maximum. When DURATION is used for STAT, the time specified is used to determine the statistical multiplier. See equation 11.22 of the Time Series, Spectra and Extreme section in How MOSES Deals With: (PDF).


Q: Where is the assumed center of rotation for vessel motions?
REV 5.03

A: Our software does not assume a center of rotation for vessel motions, since there really is no such thing. This question normally arises from the old pendulum motion techniques, assuming 20 degrees roll in 10 seconds for instance. Sometimes this question also stems from supplying accelerations for use in a structural software package. Some of these packages require not only a center of rotation, but also angles, to calculate the g*sin(theta) effect. When using the Statistics of Forces report from our software, this effect is already included.


Q: Surely your answer to the above question is not correct! The moment of inertia depends on the axis system and the roll depends on the moment of inertia. How can the roll not change with a change in axes?
REV 5.03

A: The mistake in your argument is that you are thinking in only one degree of freedom! If I may for a moment change your argument slightly, if I fix a point on the barge, and watch the roll, it will change when the fixed point changes. This is a true statement. The fallacy lies in the fact that normally a barge does not have a point which is fixed! When the axis system is changed, more must happen than a change in the roll moment of inertia - the entire inertia matrix must change. Changing the inertia matrix changes the coupling between the degrees of freedom (in this case roll and sway) and it also changes the applied roll moment. If this change is done correctly, the roll RAO will not change, but the sway one will! The change in sway is not a real change in the sway at all, it is simply a manifestation of the fact that by changing the origin of the axis system we are now computing sway at a different physical point. If you take the RAOs computed with the new axis system and use them to compute RAOs at the old origin, you will find that the results are the same.


Q: Why do my connectors have no load in the frequency domain, yet indicate load in static equilibrium?
REV 5.03

A: The RAO command was executed prior to the &EQUI command.


Q: Why do I get a substantial roll response in head seas?
REV 5.03

A: This can happen for a variety of reasons, including an asymmetric hull description. The most likely cause, however, is gyroscopic coupling. We solve the problem correctly using a full NxN mass and stiffness matrices. In most cases, the distribution of mass produces an inertia matrix which has roll/pitch coupling. Thus, the pitch excites the roll through mass coupling.


Q: How can I find the maxima of a value defined with RAOs?
REV 5.03

A: There are several way of estimating the maximum. Normally, one does this statistically by assuming that the peaks are Raleigh distributed. If one makes this assumption, then he only needs to define what he means by maximum. In MOSES, you define what you mean. This is done by giving a duration to the seastate, or defining a multiplier. If you give a duration, MOSES will use the moments of the response spectrum and the duration to produce the probable maximum value. Alternatively, you can define a multiplier for the RMS of the spectrum. The default is to use a multiple of 3.72 which defines the maximum to be the average of the 1/1000th highest values. Other values for this multiplier are:

        2.00 - Significant value
        2.54 - average of 1/10th highest values
        3.03 - average of 1/100th highest values