To initiate a time domain simulation, use the command:

     TDOM, -OPTIONS

where the available options are:

     -NO_CAPSIZE, YES/NO

     -EQUI

     -NEWMARK, YES/NO, BETA, ALPHA

     -CONVERGE, NUMB, TOL

     -ROD_REFINE, NUMBER

     -RESTART, RESTART_TIME

     -RESET, RESTART_TIME

     -SAVE, SAVE_INCREMENT

     -STORE, STORE_INCREMENT

When a time domain simulation is requested, MOSES will convert the frequency domain hydrodynamic pressure data for use in the time domain. This will results in a constant added mass, a convolution kernel, and a set of forces. The total time domain force is obtained from the frequency domain force by using a Fourier series for each heading and by interpolating a total force from the heading ones and the current vessel heading. If there is no pressure data available, null data will be used for the conversion. This will result in MOSES using the input added mass and damping matrices for the hydrodynamic interaction, and there will be no wave exciting force due to the vessel. The only wave excitation will be due to wave drift and any Morison's elements.

Normally, if a body has a roll or pitch angle greater than 90 degrees, the program will stop the simulation anticipating that something occurred that was not intended. If the -NO_CAPSIZE, YES option is used, no checking for capsizing will be performed.

The -EQUI option is useful for finding equilibrium solutions for particularly difficult or complex situations. With this option, MOSES will enter the time domain with only the mean forces applied. This is conceptually the same as using the &EQUI command, except that here, the user can examine the results for each time step.

Two methods are available for integrating the equations of motions: a Predictor/Corrector method, and a Newmark method. The default is to use the Newmark Method. If, however, -NEWMARK NO is specified, then the Predictor/Corrector method will be used. This method has been around for years, and works well in many cases. Its primary shortcomings are that for "stiff" systems, very short time steps must be used, and for larger time steps, considerable numerical damping is induced. The Newmark method ameliorate these difficulties. The two parameters BETA and ALPHA are the two Newmark parameters. If they are omitted, the defaults of BETA = .25 and ALPHA = .5 will be used. The -CONVERGE option defines the convergence parameters when a Newmark method is used. Here, convergence is defined as the change in location between two iterations at the same time step. MOSES will take at most NUMB iterations until the norm of the change in location is less than TOL. The default for NUMB is 5 and TOL is 5e-2. The Predictor/Corrector method does not iterate.

In general, a time domain simulation will begin at time equal zero, with current locations and velocities. The environment to which the system will be subjected is that specified on the last &ENV command. Events will be computed every DELTA_TIME seconds until time TOBSERV is reached, where DELTA_TIME and TOBSERV are also specified on the &ENV command. For systems which have rod element connectors included, one may encounter a situation where the time step must be reasonably small to maintain numerical stability. Here, one can use the -ROD_REFINE option to save computational effort. This option will instruct MOSES to solve all rods for a time step of DELTA_TIME/NUMBER, instead of DELTA_TIME, so that DELTA_TIME can be made suitable for the system in the absence of rod elements.

At the conclusion of the time domain simulation, no results are automatically reported. Instead, they are stored in the database for further use. To obtain reports, graphs, or other types of information about the simulation, one should issue the PRCPOST command to enter the Process Post-Processing Menu.

When one issues a TDOM command, the results of any previous simulation for this process name will be lost, unless one wishes to "restart" the simulation. This is accomplished by adding either the -RESTART or -RESET option to the TDOM command. In either case, RESTART_TIME is the time at which the previous simulation will be restarted. When one restarts a process, the events up to the restart event of the previous process will be saved, and a new process will be computed for events afterwards. Normally one restarts a time domain so that a different time step can be used or to extend the observation time.

In some cases, however, one wished to change the model at some time. Here one uses -RESET. Suppose, for example, that at time 100, one wishes to activate a set of connectors. This can be accomplished with:

     &INSTATE -EVENT 100
     &CONNECTOR C100@ -ACTIVE
     TDOM -RESET 100

This first sets the initial state to that at 100, activates the connectors and then "resets" the time domain at 100. With -RESET the connector and compartment settings at the initial state will be used at the reset time. With -RESTART the settings at the restart time will be used so that no changes can be made.

During the computation of a time domain, the data is stored in the database. The user has some control over this data storage with the two options -SAVE and -STORE. The state of a configuration will be written to the database at STORE_INCREMENT increments of computed steps. If the option -STORE is omitted, the data will be written every computed time step. In other words, a default of STORE_INCREMENT equal 1 will be used. Even though the data is stored in the database, it will be unaccessible unless the database itself has been saved. Control over how often the database is saved is provided by the -SAVE option. If this option is omitted, the database will be saved fifteen times during the time domain simulation. If the option is used, the database will be saved at SAVE_INCREMENT increments during the time domain. In other words, if -SAVE 30 is used, the database will be saved once every 30 computed time steps.

The string function

     &SLAM( :PNT_SELE, E(1), E(2), DE )

is useful for determining "slam events." This function looks at event between E(1) and E(2) in increments of DE in the current process and find the events where any part of the parts selected by the part selector :PNT_SELE are submerged. If E(1), E(2), and DE are omitted, the event in the database will be used. The result returned is a set of pairs (e1, e2) where some part is submerged between e1 and e2.