VIVANA Release Notes

Fatigue in frequency domain using Rayleigh distribution can now be calculated using bilinear SN-curves.

The system may now include SIMO bodies. The mass and stiffness of the SIMO body elements will be included, but there will be no VIV loads on them.

Bar elements may be included in the system. VIV loads will only be applied if the system does not contain any beam elements.

Note that these new modelling features are under development.

Corrected an error in the calculation of element forces using stiffness for systems with branches. The calculation was incorrect for elements whose end nodes were not sequentially numbered. This will occur only in systems with branches.

The standard deviation of the response used to calculate the drag amplification factor was calculated assuming that all response frequencies act simultaneously. This is corrected.

Longstanding error.

Both the response frequency number and the response frequency are now included in the MatrixPlot legends for matrices with results for multiple frequencies. This is done to ensure a unique identification of each response frequency even if some response frequencies are similar enough to be printed identically.

The response frequency number is now included as the first column of the convergence summary table on the _vivana.mpf file.

The spelling of summary is corrected on the _vivana.mpf file.

A group of segments and elements can now be selected using ALL for fatigue calculations. This means that - All segments in a line can be selected - All elements in a selected line and segment can be selected

Allow cross-sections to be specified for fatigue calculated from Rayleigh distributed stress cycles so that stress concentration factors (SCFs) may be specified.

An option to re-try failed response iterations has been added. If selected, the other of the two response iteration methods will be tried if the response iterations do not converge with the specified method.

If neither method converges, the results from the best iteration in the second method tried will be used.

An optional transition to constant added mass for higher modes was introduced in VIVANA 4.20.0. The transition is specified by the first and last active mode number for the transition. Specifying a transition from mode 0 to mode 0 will now give the same reduction in calculations as specifying mode 0 to mode 1.

No change in results, but improvement in computational time.

Time series of curvature are now also available for cases with nonlinear bending stiffness. Calculation of forces, stress and fatigue damage will be skipped.

Previously, this gave an error exit.

A warning is printed to the res-file if nonlinear bending stiffness is found.

The option for a smooth transition from frequency-dependent added mass for lower modes to constant, still-water added mass for higher modes has been implemented.

The transition zone is given by the input parameters ITRANS_LOW, the last mode number for which the full frequency-dependent added mass curves and ITRANS_UP, the first mode number for which the constant still-water added mass will be used. The mode number refers to the modes in the active VIV direction. In the transition region, a linear interpolation is used between the frequency dependent added mass and the constant still-water added mass.

At present, the default values cause the added mass to be taken from the selected curves for all modes.

Error if both CF and IL loads and forces are calculated from curvature and IOPPR = 0 (print results for only the most critical point point in the cross-section).

Long-standing error

The initial response estimate is improved by including nodal rotations in addition to translations.

This gives improved convergence for cases with curvature-dependent damping levels. Minor changes for other cases with Newton-Raphson iterations. No change for cases with fixed point iterations.

The data group STRUctural DAMPing SPECification allows additional material and slip damping to be specified for some or all segments in the system. This structural damping is read from separate files and is applied in addition to the relative structural damping level given by RELDAM. The structural damping is given as a function of the response curvature and is therefore updated during the response iterations.

At present, the calculated additional damping for an element is added using the total element stiffness matrix. Axial and torsional degrees of freedom will therefore also be damped.

Newton-Raphson response iterations must be used; i.e. IFRIT = 2.

Stiffness-proportional cross-sectional damping specified in INPMOD will now be included in VIVANA. Like the seafloor damping, the cross-sectional damping will come in addition to the relative damping specified in VIVANA. The damping will never be decreased under the level specified in VIVANA.

Currently, the stiffness-proportional damping coefficients must be identical for axial, torsional and bending displacements. Additional cross-sectional damping may not be applied to elements that have boundary conditions in skew or vessel coordinate systems.

The algorithm for generating pseudo-random numbers may be selected by the user. The choice of random number generator will apply to: - generation of stress time series for fatigue calculations using the Rainflow method. - output of curvature time series consistent with the stress time series generated for the Rainflow method.

The default in-line added mass curves were used even if user-defined added mass curves were given.

If user-defined in-line added mass was specified, the cross-flow added mass was used if it also was specified by user-defined curves. If default cross-flow added mass curves were specified, incorrect accessing could give program termination or possibly incorrect results.

For IRSTYP=3; combined cross-flow and in-line VIV; the in-line added mass is used in the initial eigenvalue analysis for the in-line response frequencies. These frequencies are used as the starting values if in-line frequency iterations are performed. If not, the calculated initial mode shapes are used to establish initial response estimates for the response iterations.

An uninitialized error flag could cause sporadic, false error exits after writing to or closing the _vivana.mpf fie.

The input parameter TSIOPPR for print of tension and curvature time series has been reactivated for concurrent response frequencies; IOPTSH = 0.

The stress and fatigue damage can be estimated using element dependent stress concentration factors. Note that the stress is calculated using default stress concentration factors for the other elements.

The in-line frequency iteration for combined cross-flow and in-line VIV can now optionally be avoided. This is done by defining the constant added mass by a single value.

If IRSTYP=3 (Combined cross-flow and in-line VIV) and constant added mass curve with NAMPT=1 is used for all sections, no IL frequency iteration is performed. The IL frequency is set to twice the CF frequency and the in-line added mass is not adjusted.

The relative duration; i.e. the probability; of the current condition; may now be specified in the VIVRESPONSE FATIGUE DAMAGE data group. If a non-zero relative duration is given, the fatigue damage contributions will be found as the calculated fatigue damage scaled by the relative duration. The fatigue contributions will be written to the _vivana.mpf file in addition to the calculated fatigue damage.

Corrected a long-standing error in stiffness contributions from flex-joints. The error could cause incorrect response for systems with flex-joints.

Also corrected in VIVANA 4.10.5.

Some input parameters were not set to their default value if input was not given by the user. This has been corrected.

The accumulated line length used in result presentation is now calculated from the actual stress-free segment lengths given in the line type definition in INPMOD; i.e. parameter SLGTH0 if it is given, otherwise SLGTH.

The line length is used for results on the _vivana.mpf file and for fatigue damage on the _vivana.res file. Previously, the line length in VIVANA was calculated from the final static nodal coordinates.

Significant changes are expected for cases in which all elements do not follow each other sequentially.

The change in line length will be small for cases without - segment length change; initially pre-stressed segments in STAMOD or segment length variation in DYNMOD - length changes due to temperature or pressure changes - element length modification for elements on / leaving a winch - large static elongation / compression (VIVANA and OUTMOD)

A longstanding error in the calculation of mode shapes in the VIV active plane has been corrected. The error occurred for systems where the elements were not all in a single row one after the other. The error could cause incorrect ranking of response frequencies or program termination.

No consequence for simple systems where all elements are in a row one after the other.

Error since VIVANA 4.0. Also corrected in VIVANA 4.10.2.

Corrected storage of input SN curves defined with five straight lines. No consequence is all SN curves were defined with fewer than five straight lines.

Also corrected in VIVANA 4.10.2.

Fixed the format of a figure in the pdf version of the VIVANA Theory Manual that caused errors when printing or paging through the chapter "Calculation of Response Frequencies".

Also corrected in VIVANA 4.10.1.

User-defined excitation, added mass and excitation zone coefficients may now be given for pure in-line loading, IRSTYP = 2.

User-defined excitation curves in table format may now be given for combined cross-flow and in-line loading, IRSTYP = 3.

The initial response estimate is written to the _vivana.mpf file for all response frequencies.

Final results are written even if the response is below the limit.

Avoid duplication of matrices when response iterations have not converged and the last iteration is identified as the best.

For combined cross-flow and in-line VIV loads (IRSTYP = 3) and IPRINT = 5, write both CF and IL response for all iterations.

For IOPTSH = 1; consecutive response frequencies or time-sharing; the relative duration of each response frequency is now reported on the _vivana.mpf file.

A warning is printed on the _vivana.res file if any of the response calculations did not converge. Detailed information about the response convergence is then also written on the _vivana.res file.

Detailed information about the response convergence is written on the _vivana.mpf file.

The matrix All response iterations converged? with three values is also written on the _vivana.mpf file. The first two numbers are 1 if all response calculations have converged and 0 if any did not converge. The third is 1 if all the response calculations with significant response converged, 0 if any of these did not. The key for the two Y values give the number of response calculations that converged.

Seafloor damping may be specified for analyses using the 3D seafloor formulation; i.e. IBOR3D = 1 or IBTANG = -1. A longstanding error has caused this damping to be incorrectly applied in VIVANA since 3.6. No error if the damping was less than the damping specified in VIVANA. The error has been corrected.

Corrected in VIVANA 4.8.7

The record length of the binary files was set to four times the correct value. The .ffi, .sam, .raf, and .bin files were therefore four time their necessary size. The .bin files from STAMOD were not comparable with their documentation and pre-existing tools for reading them.

Error since 4.8.0. Corrected in VIVANA 4.8.3

The new Linux release of SIMO, RIFLEX and VIVANA is 64-bit and solves several issues. Unfortunately, this means that 32-bit Linux operating systems are no longer supported.

The package has been tested on the following Linux distributions:

Hydrodynamic properties are specified in the data group SECTION PROPERTY SPECIFICATION. VIVANA will now check that coefficients are specified for all segments in the model. This was not previously checked and could give errors.

The radius of gyration for CRS0 cross sections has been corrected in RIFLEX. This concerns both CRS0 cross sections explicitly specified in the model as well as CRS0 cross sections generated for a stress joint specification. Please see the RIFLEX 4.4.1 Release Notes for details.

For most cases, the correction is not expected to cause significant changes in the response. For some cases, however, the correction may lead to changes in convergence and estimated response. Note that changes in estimated stress are exponentially increased by the slope of the SN-curve to changes in estimated fatigue damage. With a slope of 4.0 in the SN curve, a 2 % increase in stress will give an 8 % increase in estimated fatigue damage.

Corrected in VIVANA 4.4.1.

If more than 20 response frequencies contribute to the response, the plots of response amplitude, curvature, axial force, bending moment and stress are split into several matrices on the _vivana.mpf file. These matrices will now have unique names. Truncation of the legends is now avoided, so all frequencies will be plotted.

These corrections aid examining and extracting intermediate results. No change to final results; e.g. fatigue damage.

Non–unique matrix names may still occur if the response iterations do not converge and the last iteration is identified as giving the best response estimate.

Corrected in VIVANA 4.4.2.

Formatting of units in .mpf files are now consistent with RIFLEX.

Blank space is added around some numbers printed on the .res files. This has been done to increase readability and to facilitate comparison of numerical results.

Minor improvements in error handling and layout.

Windows versions 4.4 and higher are 64 bit executables and therefore require different Fortran and Java DLLs than earlier version. The necessary DLLs are included in the download package. The preformance is improved.

INPMOD and STAMOD 4.0 input files may be used unchanged.

Input for new functionality is described in the RIFLEX User Manual.

The input file version CHVERS must now be given in the data group VIVANA CONTROL INFORMATION at the beginning of the input file:

At present CHVERS must be 4.0.0 or higher.

The input parameters SCAINI, REXPAL and REXPAH have been added to the data group RESPONSE ANALYSIS PARAMETERS:

SCAINI specifies the scaling of the initial response estimate. The default value is 0.5, giving the same initial response estimate as in VIVANA 4.0.

REXPAL and REXPAH are used to limit the number of response frequencies. The default values are 0.0, i.e. no cut-off applied.

A new option for the fatigue analysis IOPDAT has been added to the data group VIVRESPONSE FATIGUE DAMAGE. For IOPDAT = 3 the fatigue damage is calculated directly from the harmonic stress amplitudes. Valid for single-frequency response only.

The new input parameters REXPAL and REXPAH allow the user to limit the number of response frequencies. Response frequencies with excitation parameter below the user-specified cut-off values are omitted. The cut-off values are given as fractions of the highest excitation parameter value, which is for the dominating frequency. Separate cut-off values are given for response frequencies below and above the dominating frequency.

At present, the default values are zero so no response frequencies will be omitted. Specifying REXPAL = 0.2 and REXPAH = 1.0 will omit all response frequencies above the dominating frequency and those below that have excitation parameters of less than 20 % of the peak value. This agrees well with experiments with linearly sheared current and a constant diameter.

The user may give a scaling factor for the initial response estimate or, alternatively, the scaling may be based on the excitation curves in the excitation zone. The input is added to aid in analyses with convergence problems. For some cases, the scaling based on the excitation curves gives improved convergence. The default is to scale so that the maximum amplitude is 0.5 average diameters, corresponding to the built-in values in previous versions.

Add option IOPFAT = 3 for direct calculation of fatigue assuming constant stress amplitudes. This option may be used for single-frequency response only. Currently available for IRSTYP = 1 or 2; i.e. pure CF loading; or pure IL loading; and a single response frequency.

The case of no VIV response is no longer reported as an error. Note that the _vivana.mpf file will in this case be empty,

Please refer to the RIFLEX 4.2 Release Notes.

The included Java folder and the HLALIB.jar file have been updated in the Windows installation .zip file.