Minutes of the Meeting on 11 April 2000

Present:     Bruno Brunone (Perugia University), Sebastian Cailaud (EDF), Chris Greenshields (King’s College), Martin Hamilton (EUTech), Anton Heinsbroek (Delft Hydraulics), Arno Kruisbrink (Delft Hydraulics), Ruud Lemmens (Delft Hydraulics), Bjørnar Svingen (SINTEF), Geoff Taylor (EUTech) Arris Tijsseling (Eindhoven University), Della Leslie and Alan Vardy (Dundee University). Pascal Guihot (EDF), Christ de Jong (TNO), Dave Wiggert (USA)

Chairman:                   Alan Vardy

Minutes:                   Della Leslie

Meeting commenced at 11.00 hours.

Apologies:     Keith Austin (Flowmaster), Simon Pugh (ESDU), Patrick Vaugrante (EDF), Lixiang Zhang and Yang Ke (Yunnan Polytechnic University).

Revision to Agenda: Bjørnar Svingen will present in place of Lixiang Zhang, who unfortunately was unable to attend. The meeting was scheduled to start an hour later.

Item 1.                   Existing guidelines – discussion session

Ruud Lemmens opened the discussion by given an informative introduction. He commented upon research on the EU project (mentioned previously in past meeting) in which standards and rules were examined (Europeans, Australian and USA). Transients, waterhammer and leak detection were all investigated. WP1 examined what’s available, showing very little detail in most European countries. In German codes detailed include transients in analysis. FSI is mentioned in Dutch code but this is not detailed. German GW-> Not standard, but higher than guidelines. Applied in German water companies, how to handle transients in water. Ruud noted that FSI is relevant in non-buried pipes. He also noted that there is no harmonisation between subjects. The project is due to finish at the end of 2000. Alan questions whether the aim was standards or guidelines? Ruud answered ground rules (i.e. basics). Alan asked about access to the results from WP1 and Ruud said that it might be possible. Dave said that in American it is usual to do things after the events and that might just use Joukowsky in assessing systems. He added that they are stricter with regard to Nuclear reactors. Dave said that in the nuclear industry structural failure is the biggest problem and that seismic loads are much studied (seismic codes, structurally induced FSI). Chris mentioned tables available in UK for Fatigue lifetime.

Alan said that this was a difficult subject with not everyone wanting to do entire analysis. Is it worth analysing? Anton said that for Flustrin they try to sift out systems before analysis. Alan asked how do you approach this? Anton mentioned flexibility and deformation modes. Also the need for Poisson coupling and/or junction coupling, saying that need to decided if there is a fluid/structure action OR fluid-structure interaction. Dave said that one of the biggest difficulty is in getting structural people to interact with fluids (& vice versa). Ruud suggested that process people have more idea than structural. Geoff highlighted the costs aspect, saying there is a large difference between industry and research. Ruud said that often problems are accepted.

Item 2.                   Selection of systems important for study in Dundee project

Della Leslie introduced this section. First she outlined the objectives for the new project in Dundee, before posing a number of questions as to the possibilities for the research. The objectives are:

To identify the features of pipe systems that cause greatest susceptibility to risk of damage through fluid-structure interaction;

To identify the minimum acceptable capabilities of methods of analysis suitable for assessing fluid-structure interaction;

To express these outcomes in a manner that will reduce significantly the uncertainties faced by designers of pipe systems.

To begin the project there are a number of questions that must be answered. What systems do we need to look at? This includes, configuration, supports, other features and cause (type) of excitation. What to analyse and what are we looking for? For the project it was proposed that up to six sub-systems would be investigated. There are a large number of possibilities with suggestions including: Single pipe, single elbow, T-piece, 4-elbow system or more complex. (In the last meeting, elbows and T-pieces were highlighted as problem areas by Simon Pugh.) Even with a single pipe there are a large number of quantities, properties that can be varied, including: Excitation (impact, oscillation); Supports (rigid, spring, location); Pumps, Valves; Material properties. What are the most important features? We need to limit the amount of work to be completed. An enthusiastic discussion followed.

Alan first asked: Is there a problem, and noted that engineers and NOT experts decide this. Pascal said that in the design of new nuclear plants, the velocity of the pipe must be less than 12mm/s (given by an ASME code). Dave asked about uncertainty levels in analysis. Anton noted that velocity and not amplitude was important, and Geoff suggested the gradient of pressure rise was important, in particular rapid increases. Dave gave an example of a plant in South Carolina; convert stainless steel to plastic: 25mm diameter, 100m long, high-pressure pumps. Shutting the pump was effectively like valve slam causing a displacement of ½ foot.

Event (somewhere) causes failure (weakest part) was suggested as an approach. Alan asked if the weakest link can be identified? Dave suggested a cause/effect approach to look for modes of possible failure. Geoff said that you can stiffen things too much, noting that steel pipe are forgiving but give more support failures. Ruud asked about the strength of supports.

Dave asked how far can we simplify. Alan noted that we start with an 100% uncertainty as to whether there will be a problem with the system and that if we could reduce this to 50% it would be good. Christ suggested locating the excitation source and try to prevent it, or seeing what the excitation is and asking could it cause a problem. Anton gave an example of a pipe bridge designed for thermal expansion including hinges at elbows to try to remove bending.

Alan gave an example of a single pipe: What vibrations are possible? He suggested that we don’t need to look at whole system. Christ asked what is the problem, saying it is different for different industries. Chris said that the type of material affected the type of failures, i.e. we have a configuration problem and a materials problem.

Dave gave an example of a case study used in the waterhammer course he teaches. He drew a picture of double elbow, saying that they isolate the structure, but do a complete waterhammer analysis on the system. Geoff asked about coincidence between fluid and structure modes.

Geoff said that we need to be aware of loads. Alan asked how strong should support be? Geoff said it is difficult to identify rapid events, therefore limit pressure gradient in order to limit loads. He asked if you can isolate an element of a system? Martin asked will it really happen? No examples?! Anton suggested validation experiments in order to convince problem owners

Item 3.                   FSI code Presto and some applications

Christ de Jong and others developed Presto 6 years ago at TNO. The research carried out is for both government and industry (includes funding by defence for researching into noise in ships). Investigation is carried out into fluid pulsations and mechanical vibrations, including: Noise transmission, fatigue, incorrect reading of flow meters and machinery condition monitoring. They have looked at the effect of pulsation on flow meters and also the effect of vibration, how to mount and where to mount meters. There can be large deviation of "10% on meter readings. Numerical simulation is used for the prediction of noise and vibration levels, evaluation of measurement methods and evaluation of noise control methods.

Presto is a frequency-domain code using the transfer matrix method.

Christ described a validation study of a singe elbow: 16cm diameter, 4.5mm thickness with an annular mass at top of 176kg (changing the mass, scales the modes) and an end mass of 572kg. For high frequencies this end mass is considered as rigid but for low frequencies moves. Flanges and bolts (with o-rings) join each part of the structure (cause additional complexity). Results compare numerical and experimental: results are good for low frequencies. For high frequencies Christ said the method of statistical energy could be used. Christ showed shows picture of modes (forced response). He said that the program is not used for analysis in real systems and gives results for first 8-10 modes. Christ showed a diagram of an experimental test loop (TU Darmstadt). The loop vibrates out of planes (pump vibrates in 3-D). Graphs give pressure at pump flange and pipe velocity at top. Various simulations are performed, including with pump only and with both pump and external excitation.

Work is underway with EDF looking at elbows (single and double elbows): Estimate fatigue due to permanent vibrations and compare numerical and experimental results. Also work on uncertainties with full coupling.

Item 4.                   Some experimental results of pressure peak attenuation

Bruno Brunone gave a short presentation on some recent research carried out at Perugia University. Bruno described two approaches for the analysis of pressure peak attenuation: A global approach and a local approach. In the global approach, analysis of attenuation (& rounding) of pressure peaks is with 1-D models and main outcomes are concerned with unsteady friction and the proper use of the MOC. In the local approach a 2-D numerical model is used and the measurement of unsteady velocity profile gives the shear stress at wall. He showed a diagram of the experimental set-up (252m long, 93mm diameter). Aims of the research are to evaluate influence of the duration of valve operation, the steady-state value of the Reynolds number (initial for closing, final for opening), the amount of steady-state energy dissipation and, for a given pipe roughness, the behaviour of instantaneous velocity profiles (with comparison to the corresponding steady-state profiles, with same discharge).

Bruno showed some graphs comparing experimental and numerical results and also illustrated how the choice of numerical method affects the results (symmetric, anti-symmetric, with or without interpolation). In one example (Re = 7668.05, K = 0.14) there was a problem with negative peaks – numerical gave a lower value than experimental with positive peaks showing good agreement. Alan suggested that if lowered experimental then both positive and negative peaks would be slightly out, asking whether the means were wrong?

Bruno gave some results showing relation between Re and K and compares to work by Alan Vardy & Jim Brown on smooth pipes. He showed there was a gap in the data.

Alan mentioned work by Ghidaoui looking at stability. In Dundee model there is no memory so the conditions stay the same. Ghidaoui showed there is change. Anton asked about FSI occurring. Bruno said there are still some unsolved problems in steady state.

Item 5.                   Some experimental results

Bjørnar Svingen gave a brief overview of current research at SINTEF. His presentation was entitled: Ongoing fundamental research in transient turbulent boundary layer, including transient friction in a large-scale plexi-glass tunnel. The research examines fundamental transient losses in tunnels and pipes.

The tunnel model is large (230x230mm, 21m long) with 2 propeller pumps (each end) and smooth walls. LDV is used for measurements. Steady and steady-oscillatory flows are examined. Absolute pressure and differential pressure are measured. Bjørnar illustrated some of the results obtained using a computer to show real-time changes in measurements. Measurements were taken for: frequency 0, 0.05, 0.1; Mean flow 0, 13, 23, 32 lit/s and oscillatory flow 22-25 lit/s. Results are given for very close to wall (0.005mm).

Measurements using LDV are given as: U = U(mean) + u (turbulent velocity); U = U(mean) + RMS. Both x and y directions were shown. The results were very interesting; in particular it showed the turbulent increasing and decreasing but not going to zero. It also show disturbance near the mean flow (examples given with zero mean flow & with steady-state mean flow, both with oscillation on top). Increasing the frequency of oscillations gave a higher turbulence. It was noted by many of the group that these results could be of great use and expressed interest in obtaining them for further use.

Item 6.                   Delft – a short tour of facilities

Arno Kruisbrink kindly showed the group around experimental facilities at Delft. It clearly illustrated the large scale of some of the testing undertaken.

Item 7.                          Place and date of next meeting

Dundee, Monday 23 October 2000.

Item 8.                          Chairman’s Closure

Closure of meeting at 17:00 hours (approximately).