As mentioned in a recent post here, we are going to publish a series of articles introducing some of the new features being developed for release in version 10.2. This article is the first of that series, discussing a new vessel option that allows users to make a choice between two methods for the interpretation of diffraction analysis data.
In version 10.0 (released October 2015) we made major changes to the way fluid loading for calculated vessels was handled. We documented the reason for making the change like so:
There is a danger of double-counting some contributions to the QTF loads, which we refer to as common second order loads. OrcaFlex has always taken measures to avoid such double-counting, but our approach in OrcaFlex 10.0 is different to that used in previous versions.
Second order fluid loads are typically calculated using QTF data obtained from diffraction analysis program results. Vessels are generally subject to nonlinear fluid loads. As with most nonlinear quantities, these can be represented as a sum of linear, quadratic, cubic and higher order terms. QTF data are provided by diffraction analysis as a means to represent the quadratic part of the vessel fluid loading. But diffraction programs include in their QTF results several second order load contributions that arise due to multiplication of first order load with first order response, or constant load with second order response. Those same interactions can arise naturally in normal time domain simulation. In fact, time domain simulation can represent the full nonlinear interaction between load and response.
So to avoid double-counting the common second order loads, OrcaFlex must apply either:
method (a): prevent the common second order loads from arising in the time domain simulation. The QTF data provided by the user can then be used without modification; or
method (b): prevent the common second order loads from arising in the QTF load, by internally modifying the user-specified QTF data before they are used to calculate the QTF load, to remove the QTF contributions for the common second order loads. The time domain simulation can then run without modification.
OrcaFlex’s handling of this issue has varied over the years, as follows:
- Prior to version 9.5 no attempt was made to address this issue. Neither of the above methods was implemented.
- Versions 9.5 to 9.8 implemented method (a) using time domain filtering of the vessel response.
- Versions 10.0 and 10.1 implemented method (b).
- Version 10.2 introduces an option to allow the user to choose between method (a) and method (b).
As an aside, this implies that the documentation quoted above, “OrcaFlex has always taken measures to avoid such double-counting”, appears to have been written assuming that history begins at version 9.5!
The list above tells the tale of a very difficult problem that we have been trying to solve for some considerable time. The fundamental issue here is that applying frequency domain diffraction data in a nonlinear large displacement time domain program is a very complicated task.
The developments introduced in 9.5 were a definite improvement on earlier versions. We had hoped that the problem was cracked at that point. However, we had feedback from a number of clients that even with the time domain filtering of method (a), OrcaFlex was over predicting roll response in certain cases. Our investigations led us to the QTF modification approach of method (b). This seemed to be theoretically compelling, and did resolve the problems with roll response.
We were then faced with the choice of either retaining both method (a) and (b), or switching exclusively to method (b). We took the decision to switch exclusively to QTF modifications in 10.0; primarily because we felt the method was better, and also to avoid increasing the complexity and scope for confusion in an already complex area of the program.
Since the release of 10.0, some unanticipated consequences of our decision have become clear. The two most significant issues are described here.
Firstly, some users found that their typical use of vessels was more difficult when using QTF modifications. OrcaFlex 9.5 to 9.8 often allowed users to successfully perform an analysis applying only 1st order (wave frequency) effects without having to explicitly include either dynamic positioning and / or moorings in the OrcaFlex model to restrain the 2nd order (low frequency) vessel responses. This was possible because when 2nd order hydrodynamic excitation was not included, there was no need to include any restraints to counter it – with the overall effect that second order response was ignored. The result was a much simpler OrcaFlex model which only analysed a 1st order (wave frequency) scenario. This analysis approach is not successful in 10.0 or 10.1, because higher order vessel loading is a natural and unavoidable consequence of applying first order fluid loads.
Secondly, the QTF modifications approach requires an accurate and consistent set of hydrodynamic data. In particular the displacement RAO data must be specified, even for calculated vessels. In principle one might imagine that obtaining consistent and complete hydrodynamic data would be a prerequisite for any project, the reality is often more muddled. Many of our users found it hard to meet this particular requirement due to consistent displacement RAO data not being readily available. This in turn leads to inaccuracies in the calculated vessel response.
Much of this knowledge has developed in response to user feedback. Our own internal testing simply cannot cover the huge range of models built and analyses run by our broad user base. As a consequence of this feedback, and our own internal investigations, we have decided to revisit the decision to offer exclusively method (b). Version 10.2 will re-introduce method (a), alongside method (b), with the choice of method made by the user. The calculation page of the vessel data form will look like this:
The Calculation Mode item is the new data that allows the choice of methods to be made.
A question that I am sure most readers who have made it this far will ask is, “which option should I select?” That is a tough question to answer, and I’m not even going to attempt to do so here. The answer will vary from model to model. Some analyses will be best performed with one option, and some with the other option.
That’s all for now. We will post again soon with news of other upcoming developments to OrcaFlex.