Chairman: Keith Austin

Minutes: Della Leslie

Meeting commenced at 09.00 hours.

Apologies: Anton Bergant (Litostroj E.I. d.o.o.), Bruno Brunone (Perugia University), Chris Greenshields (Nabla), Anton Heinsbroek (Delft), Christ de Jong (TNO), Arno Kruisbrink (Delft), Ruud Lemmens (Delft), Ivo Pothof (Delft), Patrick Vaugrante (EDF), Lixiang Zhang (Kunming University of Science and Technology).

• The minutes from the last meeting were approved.

Item 1. Progress at Dundee

Della gave an overview of the project and recent progress. She began by outlining the 5 systems that are being investigated and the methods used. In previous meetings the focus had been on the Fortran program that is being used to solve each system in the time-domain. However, it has become quite clear that this is only a small part of the process. Firstly, for system the cases to be examined have to be decided, then for each of this what variations to be performed. This is accomplished through batch files which set-up the material and geometric properties, boundary conditions and supports options. Next the Fortran program is used. But after this we are left with a very large number of data files and the job now is to arrange them into manageable groups and a manner that can be visually assessed and useful to readers.

The three main objectives of the were highlighted and the work completed in connection with each. For objective 1, features have been identified and the simulations currently being performed should provide evidence for these features. For objective 2, Della said that a select number of examples were to be chosen and different analysis methods used and compared. Arris suggested that there should be graphs comparing each method.

For objective 3, Alan has prepared a draft version of guidelines which is available on the web site. It was emphasised that this was a first draft which is clearly evident with later sections consisting of titles indicating what will be added as more results are completed.

Della continued, telling the group that the web-site had been updated, including the first set of results from the first system. Della outlined how the results will be (and have been) presented. Each figure contains 6 graphs, each one representing a different set of hydraulic boundary conditions. There was extensive discussion on the presentation of results, some preferred fixing the fluid conditions and varying the structural conditions, and vice versa. Della said she would try and satisfy each groups preference.

For the progress of experimental work, Della said that Arris would be talking about it later in the day.

Della concluded by adding that she will be leaving for Ireland of the end of the year, so that this would be the last meeting she would be organising. However, she said that in the same manner as when Arris left for Eindhoven and has subsequently invited back, the same would happen for her. But as an extra incentive she reminded the group that Arris kindly often brings gifts of stroopwafel for participants, a traditional biscuit from the Netherlands. Ireland has some traditional food stuffs, namely Guinness (or perhaps potatoes).

Sebastien Caillaud gave a presentation on the vibroacoustical analysis of pipe shell deformations. The problem in question was a pure tone is generated in open gate valves on main steam lines ('singing' steam generator) of frequency 460Hz (440Hz= 'A'). Simon asked if the aim was to remove the noise. Sebastien said it was to suppress it.

The initial objective is to understand the way of energy transfer from fluid to main pipe. Sebastien showed a diagram of the system. It included 7 safety valves with long outlets, but these were not part of the problem.

Two methods have been used, finite element analysis and a method based on dispersion diagrams. The results showed that plane wave modes did not provide any insight, but the transverse acoustical waves did provide a response at the given frequency. Sebastien concluded that flexural ovalisation of the pipe coupled with transverse acoustic modes were responsible for the unwanted noise.

Next, he describe a second problem: an unlucky high frequency excitation in a pipe with an orifice (the orifice is there to create a pressure drop in system). Dispersion diagrams could not explain the excitation.

Pierre Moussou gave a presentation that investigated an alternative approach to examining FSI – a kinematic variational approach. The objectives of the research were:

Define indications of strength of coupling

Relate acoustic and structural modes to coupled

Starting with Hamilton’s equations, Pierre presented the theory in which the coupled equations are transformed from temporal to spectral domain and gave a variational formulation for FSI. This gives a form of the equations in which there is a separate coupling term. Effectively the fluid and structural terms are separated in the form of an ‘amended structure’ + ‘resonant fluid’. Pierre said he could present many pages of equations if the group wished. Arris suggested that he should prepare a very simple example to show how the theory works.

Prior to item 4, Della gave a short display of the web site

This item was a general discussion on the future of FSI and the guidelines. Many contributions and suggestions were made from the group members. Below is some of the discussion that proceeded: (please excuse the disjointed form of these notes)

Keith began presenting a summary of what engineers like/want:

A process A Checklist

A reason to change some validations

For example FSI vs. waterhammer; need a reason why to change and validation.

Simon asked what happens in real life, i.e. who wants the information. Pierre replied that mostly process engineers will try to fix problems themselves on site, and only then if they fails will the problem be directed to research. In EDF problems are dealt with case by case and not all are FSI related.

Martin said there was two parts:

design stage (by designers)

upgrade stage (improvement or change of use)

broken/troubleshoot (guys on site)

Simon said that FSI guidelines were a good aim, but guideline on surges were also needed, i.e. we need to start further back. Alan said that the introduction in guidelines does some of this, but added that the web was aimed at non expert with a little knowledge - i.e. have heard of FSI. Simon said that within the guidelines/web-site there should be a link to consultants/experts on FSI.

Arris lists hierarchy of methods (based on time scales)

   quasi-steady           no inertia

   rigid column            no elasticity

   waterhammer         no pipe vibration

  1D-FSI                    (Us) low frequency, vibration

   2D-FSI                  (EDF) medium frequency, noise

   3d-FSI                   high frequency, ultra-sonics

Simon said that this list is numerical treatment, and Keith added that a process engineer doesn't know which one would apply. Simon suggested thinking in terms of systems and not the method.

Alan reminded the group about the Surge Net group: a thematic network and suggested anyone interested should join. The next Surge Net meeting would be in Delft next April 10^th/11^th. It was proposed the next FSI group meeting should coincide with this date and take place also in Delft. This group is broader than just FSI. Alan also told the group that the next Pressure Surges conference would be in November 2003 in Lisbon.

Christina gave a presentation on the progress made in her PhD project – FSI in flexible tubes. She reported that she was at the end of her 2nd year. She reminded the group that for her research, she was interested in local deformation, e.g. blood flow. Traditionally an uncoupled FSI approach in used, where the fluid and solid and solved individually with data passed between them. This method is computationally expensive and potentially unstable. Within her work a single formulation method is employed. This requires the solid to be written in terms of pressure and velocity. This method needs to be validated with respect to the solid, and the example being used for this is a 2D beam subject to an end shear.

Using a standard analysis with this example Christina illustrated numerical dispersion. Using a velocity formulated stress analysis there was no dispersion, but a beat was present (as reported at the last meeting). She continued describing the method with pressure included. Options for formulating the boundary conditions were: using the continuity equation, which she had reported previous that it gave dissipation, or the momentum equation, with a fixed gradient or fixed value. Arris asked why not use constant pressure at boundary, the reply was that pressure was unknown at the boundary. Results using the momentum equation were much better, showing dissipation only for a very high numbers of time steps. Concluding, Christina said that more investigation was needed, but it had been shown that the PISO algorithm (an algorithm commonly used for fluid) could be applied in structures, though care was needed in using boundary conditions. Also she said that it appeared that the use of velocity in solids gave 'better' results than standards stress analysis.

First of all, Arris kindly provided the group with gifts of stroopwafel. He continued by giving a summary of his history with the experimental apparatus at Dundee. This dated back to September 1989, with measurements on a single pipe subject to cavitation. This followed by more experiments on an L-shaped pipe in May 1990, again with cavitation. Arris said that he had requested to continue this series by proposing to repeat the cavitation experiments on a T-shaped system in 1991, but that this was denied. There were two reasons for this, one from his PhD supervisor, who believed he had done enough experiments to complete his thesis, and secondly, from Professor Vardy, due to the damage of three pressure transducers during the previous tests.

Last year, July 2001, experiments resumed at Dundee and in a previous meeting Arris had reported an ambitious program. Only a small part of the program had been completed due to problems and Della had continued with the planned experiments earlier this year. Cavitation at this point was still excluded. However, recently due to the purchase of new, spare transducers at Dundee Arris was given permission to carry out the cavitation tests previously denied.

Arris outlined the test that were performed. In particular he illustrated the repeatability of the experiment by comparing two sets of results for the T-system subject to an axial impact, resulting in cavitation. The results were very close and only then did Arris disclose that he was comparing results for those taken this year and older ones from 1989.

Arris illustrated some of the newer results, in particular the symmetry experienced in the T-section and also the phenomena that cavitation appeared to introduce out of plane motion, when only in-plane motion was expected.

Next meeting: April 10^th or 11^th 2003, Delft, The Netherlands

Item 7. Chairman’s Closure

Closure of meeting: at 17:30 approx.