The major enhancements in version 9.4 fall into the following categories:

• Improved performance of the OrcaFlex post-processing spreadsheet.
• Enhancements to modal analysis: whole system analysis, better performance.
• Enhancements to fatigue analysis: mean stress, mooring fatigue using T-N curves, SHEAR7 fatigue.
• Line contents modelling: free-flooding, slug flow, spatial and temporal variation of contents.
• Vessel data import enhancements for generic text files.
• Better handling of vessel RAO data which omits zero period response.
• Improved modelling of solid contact at edges and corners.
• Line floats wizard now supports Reynolds number dependent drag.

As usual there are a large number of improvements to the program, far too many to discuss here. Full details can be found in the OrcaFlex help file.

What follows are brief introductions to the new features that we consider to be most significant.

OrcaFlex spreadsheet

The Excel spreadsheet, supplied with OrcaFlex, that is used for pre-processing and post-processing has been completely re-written for version 9.4.

Previous versions used Excel VBA code to interface with OrcaFlex. Excel VBA has the significant limitation that it cannot be used to create multi-threaded code and so cannot take advantage of modern multi-core and multi-processor hardware. The version of the spreadsheet supplied with version 9.4 has been re-implemented as a COM add-in.

From a user's perspective this change is largely immaterial – the operation of the spreadsheet remains essentially unchanged. The principal benefit is that the code is now multi-threaded and can process multiple load cases in parallel. Naturally, this can result in significant performance gains.

There are a number of other more minor enhancements to the spreadsheet, the most notable being:

• The spreadsheet can now take advantage of Excel 2007 and later supporting very large worksheets. Older versions of Excel were limited to 65536 rows and 256 columns, but the more recent versions of Excel have effectively removed these limits.
• A new facility has been added to automate generation of text data files. This is very similar to the existing facility for generating batch script files.

Modal analysis

The modal analysis facility has also received significant attention, although the basic operation is largely the same as in previous versions.

Possibly the most important development is the implementation of a new eigen-solver, using an iterative Lanczos algorithm, which has significant performance benefits. This new solver is limited in that it cannot extract all modes of the system. However, provided that you require no more than a few hundred modes, the new solver will be used and will solve very quickly. The benefit is felt for large systems and the new solver will typically solve in a matter of seconds large systems that require minutes or even hours for the old solver.

The other major enhancement is the implementation of modal analysis for the whole system. Previous versions of the program could only perform modal analysis for a single line within the model. The new capability fits naturally into the existing modal analysis user interface. We have no doubt that a great many users of the program will benefit from this development. Indeed, we have made extensive use of it to help deal with client support queries.

Together with a number of other minor improvements, these changes to the modal analysis capability greatly improve its utility.

Fatigue analysis

OrcaFlex's fatigue analysis feature has seen regular development over recent releases, and version 9.4 is no exception.

The S-N curve data specification now allows for mean stress effects. Three different models have been implemented to describe these effects, namely Goodman, Soderberg and Gerber. Mean stress effects can be included for all types of fatigue analysis that make use of S-N curves.

Two new damage calculation options have been added: Mooring and SHEAR7.

Mooring fatigue is based on T-N curves. These are similar to S-N curves, but instead relate number of cycles to failure (N) to tension range (T).

SHEAR7 fatigue is somewhat different from the other damage calculation options. The difference is that the damage is actually calculated by SHEAR7. The OrcaFlex code merely collates the output from a series of SHEAR7 analyses. OrcaFlex reads the SHEAR7 output files, sums the damage and presents the results in either graphical or tabular form. Combined with the direct SHEAR7 interface and the OrcaFlex batch form, this allows you to automate an entire SHEAR7 fatigue analysis.

Line contents modelling

All previous versions of OrcaFlex included a very basic and simple model for line contents. Contents density could be specified, but it was assumed to be constant for the entire length of the line. OrcaFlex 9.4 introduces a host of new features to make contents modelling much more flexible.

Line contents can now be specified to be free-flooding. This results in the line being filled with seawater up to the instantaneous water level. Another new option has been added which allows axial inertia due to contents to be omitted from the analysis.

The other new contents modelling option is named slug flow. Although intended primarily for modelling of slugs, this option allows for both spatial and temporal variation of contents. We envisage the feature also being used to model static density variation, flow-lay, etc.

The slug flow option takes as input a specification of the density variation along the line. In addition a velocity is defined which determines how the contents flow. This velocity can vary over the course of the simulation. We have taken care to ensure that contents progress smoothly between line elements to minimise noise due to transient loading. The code accounts for the resulting variations of mass, weight and centrifugal and Coriolis forces.

Whilst we have not attempted to model complex multi-phase flow conditions, we do feel that the new facilities are in keeping with the level of detail typically modelled using OrcaFlex.

Vessel data import

The process of importing hydrodynamic data has been improved in a number of ways.

Generic text data may now be combined in one file and imported in a single operation. Previously, displacement RAOs, load RAOs, QTFs, and added mass and damping all had to be in separate files.

Generic text files may now specify the conventions and units of their data. If the conventions or units of the data being imported differ from those in the OrcaFlex model, then the data will be converted to match them at the time of import.

Extrapolation of RAO data towards the zero period response

Extrapolation of RAO data towards zero period response is now handled differently. Previous versions did issue warnings if any extrapolation was required. However, the extrapolation typically resulted in a non-zero response at zero period which is clearly non-physical.

Although we would always encourage users to read and take notice of the warnings, it is possible to make the program handle such input data in a more useful way. So, from version 9.4, if extrapolation is required for short period waves, the program will now always extrapolate towards a zero response for zero period.

Modelling of solid contact at edges and corners

Elastic solid contact is now handled slightly differently for objects with non-zero contact diameter, e.g. line-solid contact. The algorithm used in previous versions had some weaknesses. The essential problem was that the reaction force always acted normal to the solid face closest to the node. When a node is near an edge or a corner then this results in a discontinuity which in turn can lead to poor convergence and noisy responses.

In addition, the previous algorithm had the unfortunate property that contact could, under some circumstances, occur for nodes outside the shape. This problem most commonly afflicted nodes just beyond the end of a curved plate, e.g. a bellmouth.

Version 9.4 solves these problems by adopting a minor change to the contact algorithm. The reaction force now acts in the direction of the vector p_o-p_s where p_o is the position of the node and p_s the position of the closest point on the surface of the solid to p_o. When p_s is on one of the faces of the shape then the new and old algorithms are identical. However, when p_s is on an edge or at a corner then the reaction force is no longer normal to one of the faces.

Line floats wizard

The line floats wizard is a widely used and very useful feature. However, it has not been updated for some time and consequently did not make full use of new features added in recent versions of OrcaFlex.

The most significant change is that Reynolds number dependent drag coefficients for the base line are now supported. The program still assumes that the drag acting on the floats is independent of Reynolds number.  However, this is a much more reasonable assumption given the typical size of the floats used for offshore buoyancy.

As now seems to be traditional, here at Orcina we have spent the summer months working very hard to complete a major upgrade to OrcaFlex. The result is OrcaFlex 9.3 which was formally released early in August. We have just despatched the final upgrade CD and so all customers with up-to-date MUS contracts will have the software very soon, if not already.

I should also point out that we have released a minor upgrade (version 9.3b) which fixes some bugs that have come to light in version 9.3a (the version included on the upgrade CD). As usual a patch to upgrade from 9.3a to 9.3b is available for download.

The major enhancements in version 9.3 fall into the following categories:

• Extreme value statistics post-processing of results variables.
• Text data file (in addition to existing binary data file).
• Enhanced interface to SHEAR7.
• Static state simulation files.
• Non-linear material properties for elastomeric bend stiffeners.
• API 2RD stress check.
• Coatings and linings for homogeneous pipe.
• Roll damping for vessels.
• Modal analysis improvements.
• Performance improvements.

As usual there are a large number of improvements to the program, far too many to discuss here. Full details can be found in the OrcaFlex help file.

What follows are brief introductions to the new features that we consider to be most significant.

Extreme value statistics

Extreme Value Statistics results are available for all time history variables. Return values are estimated for a user-specified time duration, based on the time history of a selected variable.

Rayleigh, Weibull or Generalised Pareto distributions can be fitted. For the latter two, confidence limits are estimated for the return level and diagnostic graphs are presented indicating the goodness-of-fit of the selected model.

Text data file

OrcaFlex models can be saved to text data files in addition to the traditional binary .dat data file. Text data files can be edited in any standard text editor and are readable, well structured and self-documenting. The text data file offers benefits for QA and automation in particular and is intended to complement, rather than replace, the binary data file.

A simple example is shown below:

General:
  StageDuration:
    - 10.0
    - 50.0
Lines:
  - Name: Line1
    Length, TargetSegmentLength:
      - [60.0, 5.0]
      - [40.0, 2.0]
      - [120.0, 10.0]

This example first defines a 10s build-up stage followed by stage 1 with 50s duration. Then an OrcaFlex Line is created and named "Line1". Finally the section data is specified: three sections are created with varying section lengths and segment lengths. Default values are used for all data which are not specified.

Enhanced interface to SHEAR7

Previous versions of OrcaFlex interfaced with SHEAR7 by exporting SHEAR7 input files (.dat structure file and .mds mode shapes file). This approach left the onus on the user to then run SHEAR7.

The enhanced interface calls SHEAR7 directly from OrcaFlex. One immediate benefit of the direct SHEAR7 interface is much simplified automation and file management.

The other benefit is that coupled VIV analyses can now be performed. The direct SHEAR7 interface achieves this by parsing the SHEAR7 output file to obtain the drag enhancement factors (C_f) predicted by SHEAR7. OrcaFlex then uses the enhanced drag coefficients and repeats the static analysis until a coupled solution is obtained.

Static state simulation files

One of the quirkier aspects of previous versions of OrcaFlex is that static state results could not be saved to file. The standard work-around to this limitation is to run a very short dynamic simulation with no dynamic excitation.

I'm pleased to say that this work-around is no longer needed. Static state results can now be saved to simulation files in much the same way as dynamic results can be – you simply save the model when it is in static state.

The batch form has been extended to allow specification of the analysis to be performed. If you specify dynamic analysis then the program processes the data file in the traditional way by running a dynamic simulation and then saving a dynamic simulation file. However, if you specify static analysis then the program performs the static calculation and then saves a static state simulation file.

This natural enhancement to the program is clearly useful when static analysis takes a significant amount of time. For example, the ability to save static state simulation files is particularly beneficial when using the direct SHEAR7 interface.

Non-linear material properties

The previous major upgrade to OrcaFlex (version 9.2) introduced built-in modelling support for bend stiffeners. A limitation of the built-in bend stiffener was that the material properties had to be linear. This limitation has now been removed and the material properties can be specified by a non-linear relationship between stress and strain.

This improvement has been made principally to allow for better and simpler modelling of elastomeric bend stiffeners but could also be valuable when modelling plastic deformation of steel pipes during installation.

The relationship between stress and strain can be specified either by supplying a table of values or through the Ramberg-Osgood formula. The former option is typically used for elastomeric bend stiffeners whilst the latter is applicable to steel pipes.

API 2RD stress check

A relatively simple (but hopefully valuable) addition to the program has been the inclusion of the stress check from the API 2RD code. This is a very commonly used strength criterion for the design of steel risers.

Implementation of the API 2RD stress check is a relatively simple exercise in post-processing of standard OrcaFlex output. However, the code as published contains a number of well-known errors. A recently published errata corrects some, but not all, of these errors.

OrcaFlex implements corrected forms of the code and so removes any doubts concerning the unfortunate errors in the published code documents. In addition, having the stress check built-in to the program significantly increases productivity.

In future releases of OrcaFlex we intend to implement additional code checks from other commonly used industry codes.

Coatings and linings for homogeneous pipe

The homogeneous pipe category of Line Type was introduced in OrcaFlex 9.2. This enables you to specify a homogeneous pipe line type by giving the basic structural properties such as material density, Young’s modulus, diameters etc., and OrcaFlex will calculate for you the derived structural values (e.g. mass per unit length, axial & bend stiffnesses, etc.).

This feature has been extended to cater for coatings and linings. Coatings and linings are typically used with steel pipes to model the additional mass and displacement of concrete coatings, plastic linings etc. They contribute mass, weight and displacement and also modify the pipe's inner and outer diameters. However, they contribute no additional structural strength and are assumed not to be load bearing. Stress results are calculated based on stress diameters equal to the underlying pipe diameters.

Roll damping for vessels

Calculated vessels can now model the effects of roll damping. These effects are included in a simple model offering both linear and quadratic damping terms.

Modal analysis improvements

A number of improvements have been made to the modal analysis facility included in OrcaFlex:

• Modal analysis can be performed for lines which include torsion.
• Mode shape views can be animated which greatly helps visualisation of more complex mode shapes.
• Modelling of seabed friction in modal analysis calculations is now more robust. Previous versions of the program were somewhat prone to modal analysis failures related to seabed interaction and this change has been implemented to avoid such failures.

Performance improvements

Previous releases of the program have suffered from performance problems when modelling certain drag-dominated systems. Most commonly this affected towed-array systems with high tow speeds. Such systems could suffer from poor convergence in both statics and implicit dynamics. Dynamic simulations could require very short time steps and often were faster performed with the explicit solver. These performance problems have been comprehensively addressed by this release of the program.

In addition, we have made performance improvements for models which include line clashing when using the implicit solver. Such systems can often now be modelled with longer time-steps than was possible in previous versions.

However, you still should use a short enough time-step to achieve your desired level of accuracy. Line clashing commonly leads to significantly non-linear responses. Hence, shorter time steps are required for accurate simulations than is the case for systems with less non-linearity.

While we are on the subject of the post-processing spreadsheet it seems appropriate to announce some of the changes we are making for 9.3. As you are probably aware the post-processing spreadsheet is implemented using Excel VBA macros. These link, in turn, to the OrcaFlex code through our OrcFxAPI DLL.

VBA macros have many strengths but we have become somewhat constrained by their limitations. For 9.3 we will be moving all the code into the OrcFxAPI DLL. One immediate benefit of this is that our spreadsheet template will no longer contain any macros and so will not be subject to Excel's macro security dialogs.

For 9.3 we are planning to keep the user-interface to the spreadsheet identical to previous versions. We are, however, planning one very significant change to the implementation. All previous versions processed instructions serially and made no use of the multi-core processors that are now commonplace. It is not possible to perform calculations in parallel with VBA macros. Moving the processing code into the OrcFxAPI DLL will enable us to process load case results extractions in parallel. This should significantly reduce the time taken to post-process results.

This development is currently partially complete and we have not yet implemented the multi-threading of the results extraction. I look forward to being able to write about the performance of the multi-threading aspect of this work when it is complete. Hopefully I'll be able to announce big improvements in extraction times.

The model is for modelling the seabed resistance for catenary pipeline contact. It takes as input the line diameter and the seabed soil geotechnical properties (undrained shear strength at the mudline, shear strength gradient with depth, soil saturated density). It then calculates the seabed reaction force on each line element as a function of the penetration, allowing for non-linear penetration resistance, hysteretic effects due to uplift and repenetration, and non-linear suction forces.

Some of the model characteristics are illustrated in the following diagram. For comparison, the standard OrcaFlex linear seabed model would give just a straight line seabed reaction curve in the diagram.

This release includes a number of bug fixes as described below:

• Certain models (typically more complex models) which use the stiffener modelling object were failing to simulate when using the implicit solver.
• Results extraction from the OrcaFlex Excel spreadsheet was sometimes hanging and eventually reporting an out of memory error in Excel.
• Batch processing had a number of separate bugs. Firstly, the auto save function was not working. In addition the option to save partially completed simulation files when cancelling a batch run was not functioning. Finally, multi-threaded batch simulations could sometimes fail, erroneously reporting "out of memory" or "could not create thread" error messages.
• Certain data items are editable even after a simulation has been run. Most notably this includes stress loading factors for Line Types. Such data items can be changed within OrcaFlex and so likewise should be editable from a batch script. However, in 9.2a the act of changing any such data item from a batch script resulted in the simulation being reset.
• There were two separate bugs relating to range graphs with vertical arc length axis (a new feature introduced in 9.2a). Such graphs were reporting incorrect values in the hint window. Additionally, instantaneous value range graphs with vertical arc length axis were not automatically ranging themselves correctly.
• Import of AQWA QTF data was failing if the data has directions which are split across more than one table.
• Cancelling certain operations (e.g. opening a graph) could leave the program in a state in which simulations could not be run.

None of these bugs lead to erroneous simulation results and so this upgrade is not a critical one.  However, a number of the bugs have significant impact on commonly used program functionality and so we would recommend that all 9.2 users upgrade to 9.2b.

A patch to upgrade from 9.2a to 9.2b can be downloaded from our website. If you would prefer to receive version 9.2b on CD please contact us.

There are a great many enhancements to the software which we described in the previous blog post. The only significant change in the software from that post is that the non-linear soil model is not included in 9.2a. We have not had sufficient time to perform the testing and validation of this development to a level sufficient for release. We are continuing to work on this and hope to include the non-linear soil model in a minor 9.2 update sometime later this year.

• Performance
• Bend Stiffener and Stress Joint modelling
• Multi-threaded batch processing
• Fatigue post-processing enhancements
• Non-linear seabed
• Wake deficit / interference modelling
• Wave spreading
• 6 DOF Vessel QTFs
• Python interface to OrcFxAPI

Performance

We have continued the trend of significant performance improvements to OrcaFlex which started with the introduction of the implicit solver in version 9.0. For 9.2 we have improved the implicit solver in the following ways:

• Increased the default dynamics convergence tolerance to 25e-6. This results in faster simulations and our extensive trials indicate no discernible loss of accuracy. Note that this convergence tolerance is safe to use with version 9.1 so you can take advantage of the speed-up today.
• We have overhauled the linear matrix solvers used by the implicit solver. This has two benefits. First, many simulations will now run even faster because the new solvers are more efficient. Second, complex cases with many connections are now significantly more robust.
• Improved the convergence still further for many cases.

In summary, 9.2 is even faster than 9.1 and diverges less frequently.

Bend Stiffener and Stress Joint modelling

Bend stiffeners (BSR) and tapered stress joints (TSJ) are modelled in OrcaFlex by discretising the BSR/TSJ profile as part of the overall line discretisation. Currently this involves quite a lot of pre-processing of data since you need a separate OrcaFlex Line Type object for each segment of the discretised profile. To make the modelling easier we are introducing the following enhancements in 9.2:

• There is a new category of Line Type called Homogeneous Pipe. This enables you to specify a homogeneous pipe line type by giving the basic structural properties such as material density, Young's modulus, diameters etc., and OrcaFlex will calculate for you the derived structural values (e.g. mass per unit length, axial & bend stiffnesses, etc.). This is now the preferred way to define steel riser properties.
• The outer diameter of a Homogeneous Pipe Line Type can vary. You specify the variation as a piecewise-linear table on the Variable Data form. When OrcaFlex builds a line based on this Line Type it discretises the variation of mass, buoyancy, stiffness etc. automatically for you based on your specified segmentation.
• Bend stiffeners are modelled as separate OrcaFlex Lines, with a user-specified diameter profile, that can then be attached to the product (riser, umbilical, etc.) that they are protecting. This allows the stiffener line to apply bend moments etc. to the product line, but you can obtain results for the product and stiffener lines separately.

For example a stress joint may be modelled in 9.2 by first specifying a diameter profile on the Variable Data form:

This simple profile results in a linear variation of diameter over the length of the joint but arbitrary piecewise linear variations can be modelled.

The profile can then be referenced on the Line Type data form, together with material properties:

Note that we have not shown hydrodynamic data, drawing data etc. but this data is still present.

Multi-threaded batch processing

The advent of commonly available multi-core CPUs means that in order to take advantage of their capabilities, programs need to perform multiple threads of execution concurrently. The OrcaFlex batch form has traditionally processed each job in sequence. In version 9.2, jobs are processed concurrently. For example on a dual core machine the OrcaFlex batch processor can now run two simulations at the same time.

In the past you could achieve this effect by running multiple separate instances of OrcaFlex, i.e. running multiple separate copies of OrcaFlex on the same machine, but the division of tasks for the separate instances was a manual and tedious procedure. Of course on a machine with even more processor cores the tedium increases linearly! The new batch processor in 9.2 removes the tedium. This is what it looks like running on our brand new 8 core machine:

Fatigue post-processing enhancements

The fatigue post-processor in OrcaFlex has been given a thorough review for the release of 9.2.

Performance

In light of the numerous recent improvements to OrcaFlex runtimes we have realised that the fatigue post-processor has become a bottleneck. In 9.1 it can take longer to post-process the fatigue analysis than it does to run the simulations.

We have made numerous optimisations to the post-processor and it is now significantly faster. It's impossible to give a single speed-up figure, but for a typical SCR fatigue analysis, which we have used to benchmark the optimisations, 9.2 is around 20 times faster than 9.1.

The new version also uses multiple processor cores if they are available and this can result in even greater speed-up. Our 8 core machine is around 100 times faster when post-processing fatigue with 9.2 than 9.1!

The improvements give a step change in the effectiveness of the OrcaFlex post-processor. Combined with the simulation speed improvements over recent years this now opens the way for routine use of irregular wave rainflow fatigue assessments. We will be talking in more detail on this subject in the upcoming OrcaFlex User Group Meetings.

Stress factor fatigue

The fatigue post-processor was designed for use with homogeneous pipe (e.g. steel pipes), and stresses are recovered easily for homogeneous pipes. But for umbilicals and flexibles, where the construction is more complex, the homogeneous pipe stress calculation built into OrcaFlex is not appropriate.

For many years now we have made use of the Line Type Stress Loading Factor data to trick the homogeneous stress recovery into producing appropriate stresses for an umbilical or a flexible. Whilst this works well it is quite complex to explain and very error-prone. Accordingly we have now encoded this method directly into the fatigue post-processor and removed the need for any manipulation of Stress Loading Factors.

Results presentation

The final major fatigue development is to the results presentation. The existing tabular output has been augmented by some new summary tables which draw out the variation of damage with arclength. In addition we will very shortly be adding charting facilities to make visualisation of the fatigue results more immediate.

Non-linear seabed soil model

In previous versions of OrcaFlex the seabed is modelled as a simple linear spring+damper. A real soil seabed behaves in a much more complex way, so we have added a new alternative seabed model which models seabed-pipe interaction using a fully non-linear hysteretic seabed soil model.

The new model has been developed by Prof. Mark Randolph of the University of Western Australia, an acknowledged expert on seabed behaviour. It takes as data the seabed soil properties, such as the undrained shear strength at the mudline, the shear strength gradient with depth and the saturated soil density, and the model then calculates the seabed reaction force. The model includes the non-linear and hysteretic behaviour of the seabed, and effects like suction resisting uplift.

The above graph illustrates the model for a case where a catenary line touches down on the seabed, giving initial penetration that increases as the pipe sinks into the seabed (blue curve). And the graph shows how the model's seabed reaction responds if the pipe then starts to lift again, either lifting right off the seabed (the green curve) or lifting and then repenetrating again (the purple curve). Further cycles of uplift and repenetration give suitably calculated hysteresis loops of seabed reaction.

Wake deficit / interference modelling

The new version will include the option to model wake deficit and interference effects. Wake interference effects occur when a downstream line is in the wake of an upstream line. The downstream line experiences reduced flow velocity in the wake, and so is subject to reduced drag loads (wake deficit). It can also experience a transverse lift force that tends to pull it in towards the centre of the wake of the upstream line. These effects can lead to dangers such as riser clashing, so it is very useful in some analyses to be able to include modelling of wake interference effects.

In the new version you have the option of including these wake interference effects in your model, using one of the following 3 wake interference models:

• Huse model. This is a well known drag reduction model, but it does not include any transverse lift force effect.
• Blevins model. This is a more recent model developed by RD Blevins, which models both the drag reduction and also the transverse lift effect.
• User-specified model. This allows you to specify the wake interference effects yourself, by specifying a table of drag coefficient factors and transverse lift coefficient factors, as a function of the position of the downstream line element relative to the upstream line. This option is useful if you have experimental data available that you believe is suitable for the cases that you want to model.

These wake interference models define the wake effects in the simple situation of two rigid parallel cylinders, one downstream of the other, subject to flow normal to their axes. The models are intrinsically 2D and the most challenging aspect of the development was coming up with an effective generalisation to arbitrary 3D configurations that are common in OrcaFlex. These generalisations, and the OrcaFlex wake interference modelling, are discussed in a paper (Innovative Riser Wake Interference Assessment) to which we have contributed, and which will be presented at the ISOPE conference in July.

Wave spreading

OrcaFlex irregular wave trains are currently unidirectional. In 9.2 we have added support for directional spread spectra. This augments the existing frequency spectrum.

The directional spectrum that we have chosen is a cos²ⁿ spread over 180°. The new user data items are the number of wave directions and the exponent 2n.

We have adopted an equal energy approach to discretising the directional spectrum. This is similar to our handling of the frequency spectrum. Instead of using equally spaced wave directions OrcaFlex chooses wave directions that are concentrated towards the mean wave direction, with a distribution specified by the direction spectrum. For a given total number of wave components this results in a much more efficient discretisation than using equal spacing in direction.

6 DOF Vessel QTFs

For many years now the 2^nd order wave drift facility in OrcaFlex, useful for coupled analyses, has catered for drift loads only in the 3 horizontal degrees of freedom: surge, sway and yaw.

For a barge or tanker shaped vessel then this is adequate but for a SPAR, for example, with a much smaller water plane area, wave drift becomes significant in the other degrees of freedom. In 9.2 we cater for this by adding to the existing data some extra columns for heave, roll and pitch wave drift coefficients.

Python interface to OrcFxAPI

The programming interface to OrcaFlex, OrcFxAPI, allows full access to OrcaFlex's capabilities from other programs. The native language for this interface is C and we also provide full support for accessing the interface from Delphi and Visual Basic.

Accessing OrcFxAPI from these languages is very powerful but also quite a complex process. Quite a significant amount of relatively low-level programming is required to perform memory management, error checking etc., and this makes it much harder to automate OrcaFlex than we would like. We have now addressed this issue by developing a powerful new interface with the Python programming language.

Python is a very high level language that is very widely used and is freely available. Automating OrcaFlex is greatly simplified by using the new Python interface. We recommend that anybody wishing to automate OrcaFlex gives very serious consideration to using Python to do so. The Python interface supports version 9.1 and we can supply a pre-release.

Miscellaneous developments

As usual there are a host of other smaller developments to the program that improve it (we hope!)

Release schedule

We are currently planning to release 9.2 towards the end of July. We hope that you find it as big an improvement as we believe it to be.

Version 9.1c fixes two bugs as described below:

Vessels

Versions of 9.1a and 9.1b had a bug which affected vessel motions and runtime performance when all of the following conditions were met:

• The implicit integration scheme was in use.
• A vessel had Primary Motion set to "Prescribed" or "Time History".
• That same vessel had Superimposed Motion of "None".

When all these conditions were met then vessel motions contained errors. The errors build up gradually during the simulation which means that longer simulations are more seriously affected. In addition simulations were significantly slower than they should have been.

In version 9.1c the problem has been corrected.

Lines

Versions 9.1a and 9.1b had a bug related to the new hint window results feature introduced in version 9.1a. If a replay is paused and you hover the mouse pointer over a line then a hint window is displayed. This hint window now includes results. The code to derive these results contained a bug which corrupts Summary and Full Results for the line, and also corrupts the 3D View of the latest simulation position.

Note that the bug does not affect other results such as Time History, Statistics, Range Graph etc. It also does not affect replays. The bug is corrected in version 9.1c.

Obtaining the upgrade

A patch to upgrade from any earlier version of 9.1 to 9.1c is available for download from our website. When downloading the patch you will be asked for a username and password which you can find in the OrcaFlex 9.1 About box (see About on the Help menu).

If you would prefer to receive version 9.1c on CD please contact us.

In August 2007 we released OrcaFlex 9.1, which introduced many further improvements to the implicit solver — version 9.1 is now significantly faster than 9.0. The purpose of this blog post is to discuss some of the improvements in version 9.1, present a runtime comparison with Flexcom, and to discuss some future developments we plan to make OrcaFlex even faster still.

Integration schemes and dissipation

In version 9.0 we used the Newmark integration scheme with γ = ½ and β = ¼. This choice of parameters is known as the constant average acceleration (CAA) scheme.

Compared to the OrcaFlex's explicit integration scheme this resulted in huge improvements in run times. However, we weren't completely happy with the performance.

The principal cause for concern was the fact that some models would only run to completion with relatively small time steps. We implemented a variable time step algorithm which allowed OrcaFlex to use longer time steps if possible. For the time frame of the 9.0 release we were simply unable to investigate and implement a better solution.

After releasing version 9.0 we then had time to study the problem in more depth. The key is a feature present in most integration schemes (but not Newmark CAA) called numerical dissipation. This dissipation is designed to damp out spurious responses in non-physical high-frequency modes of the numerical model. Such spurious high-frequency responses are inevitable in structural finite element programs.

There are versions of the Newmark scheme which include numerical dissipation, but these are known to over-damp responses at lower frequencies. In other words they can affect the signal as well as removing the noise. Other commonly used integration schemes which include dissipation are:

• The Wilson-θ method. This has very good dissipation of the spurious high-frequency response but unfortunately over-damps the lower frequency modes.
• The α method of Hilber, Hughes and Taylor (HHT).
• The Generalised-α method of Chung and Hulbert.

For OrcaFlex 9.1 we chose to use the Generalised-α scheme. It has some rather obscure technical advantages over the HHT method, but in reality there is little to choose between Generalised-α and HHT in terms of run time and accuracy.

The Generalised-α scheme results in much better and more stable convergence of the implicit iteration. The net result of this is much faster simulations even than OrcaFlex 9.0.

The OrcaFlex "High frequency dissipation" parameter

OrcaFlex includes a data item called "High frequency dissipation". This is the parameter denoted ρ_∞ in Chung and Hulbert's paper. It controls how much numerical high frequency dissipation is provided by the Generalised-α integration scheme.

The high frequency dissipation must take a value between 0 and 1. Perhaps counter-intuitively, larger values correspond to lower levels of dissipation. A value of 1 gives no dissipation and a value of 0 gives asymptotic annihilation, whereby high frequency response is annihilated after one time step.

The default value is 0.4 which has been chosen to give fast simulation run times without compromising accuracy. We have yet to come across an OrcaFlex model for which it is better to use some other value. So it is likely that in a future release of the program we will remove this data item and use an in-built value of 0.4.

Variable and constant time step schemes

The other problem with the Newmark implicit solver of OrcaFlex 9.0 is due to its use of a variable time step algorithm.

Variable time step schemes can introduce high frequency noise into a system which in turn can lead to inaccurate results, for example noisy time histories, non-physical spikes in results etc. This is a feature of all variable time step algorithms and not something particular to OrcaFlex. For the majority of systems no problems arise when using a variable time step. However, if you are using variable time steps then we do recommend that you check the quality of your results.

For OrcaFlex 9.0 we had no choice but to use a variable time step scheme, because we had no numerical dissipation built in to the integration scheme. Changing to use the Generalised-α scheme allowed us to offer a constant time step option in OrcaFlex. This is now the default setting and you should use a constant time step whenever possible.

Run Times

After all that theory, what difference do these changes actually mean in terms of run times? Well, that's an impossible question to answer comprehensively because the answer varies for different models. However I can say that the majority of cases are many times faster in 9.1 than in 9.0, and we have not seen any case that is slower in 9.1 than in 9.0.

We have performed a comprehensive timing comparison for the deepwater SCR case as described in our validation document 99/101. This is a model of a 12" SCR in 1800m water depth. Results of various analyses are compared against Flexcom, and extremely close agreement is achieved for all cases considered.

For the timing comparison we considered three environmental variants on the basic model as follows:

These were run in OrcaFlex 9.1 and Flexcom 7.3. The models had identical spatial discretisation and a variety of time steps were used. Here we present results for time steps 0.1s and 0.25s.

For this model OrcaFlex is between 3 and 6 times faster than Flexcom, depending on the time step and case considered. Although we have only varied the time step and the wave, we would expect run times to be sensitive to other model parameters. So it is impossible to make any general statement comparing the run times of the two programs. However, it is clear from this that the simulation run times for OrcaFlex 9.1 are extremely competitive.

These comparisons were performed by an independent consultant with access to both OrcaFlex and Flexcom.

Future improvements

We have recently made various improvements to the linear solvers that the implicit solver uses for each iteration of a time step. These changes, to be released as part of OrcaFlex 9.2, yield between 10% and 20% improvements on the run times of 9.1

Another area where we can see room for improvement is in the multi-threading performance of OrcaFlex. By multi-threading we mean the capability of processing a single job in parallel on each processor core of a computer.

OrcaFlex has been able to do this since version 8.7. However the facility is limited by the fact that to achieve best speed-ups you need at least as many OrcaFlex Lines in your model as processor cores. So as computers with more and more cores become commonplace, there will be more cases where the hardware is not being used as effectively as possible. In addition the speed-ups are not as significant when using implicit as when using explicit.

This is an area which we need to improve as multi-core machines are now becoming widespread. I hope that we can make significant improvements to our multi-threading capability over the next 18 months.

Of course, it's quite common to have a lot more load cases than processor cores. In this scenario you can run multiple copies of OrcaFlex and process the load cases in parallel that way. This becomes difficult to manage if you have a large number of processor cores available and so one option is to use Distributed OrcaFlex (DOF).

As a simpler alternative to DOF, we are currently changing the batch facility in OrcaFlex to process multiple simulations in parallel, one per processor core. We expect to release this in version 9.2. This form of parallel processing (one simulation per core) allows pretty much optimal speed-up, i.e. n processor cores gives an n times speed-up.

In a similar vein we have recently revamped the fatigue calculation also to make use of multiple cores. This also yields close to optimal speed-up.

Finally, we are also planning to implement a frequency domain option and possibly a linear time domain option (OrcaFlex is currently fully non-linear). These obviously improve run time at the expense of accuracy. At the moment we anticipate that these options would be available as part of the standard OrcaFlex package — in other words we don't expect to charge extra for frequency domain or linear time domain capabilities.

As always we would encourage and welcome any feedback you have on this topic. So please do e-mail us or post comments on the blog.

To make room for these changes we had to rearrange a number of other shortcuts. In particular the shortcuts to rotate the 3D View were changed to CTRL+ALT+ ← → ↑ ↓. Shortly after releasing version 9.1 we began to hear reports from some users that their laptop screen would suddenly be displayed upside down, or on its side, after using one of these shortcuts.

After a little detective work we have traced the issue to a piece of Intel software that is stealing these keyboard shortcuts before they reach OrcaFlex. The solution is as follows:

1. Open the Control Panel and make sure it uses Classic View.
2. Double click the Intel GMA Driver for mobile icon.
3. Select the Hot Keys page.
4. Deselect the "Enable Hot Keys" option.

This will fix the problem for Intel graphics cards, but there may be similar problems for cards from other manufacturers, or even for different Intel cards. I'm guessing that similar solutions will be available and that the steps described above should help you solve such problems.