|
Minutes
of the Meeting on 30 April 2001
Present:
Keith Austin (Flowmaster), Bruno Brunone (Perugia
University), Mohamed Ghidaoui (Hong Kong), Christina Giannoppa
(King’s College) Chris Greenshields (King’s College), Zhao Ming
(Hong Kong) Arris Tijsseling (Eindhoven University), Jim Brown,
Della Leslie and Alan Vardy (Dundee University).
Chairman:
Keith Austin
Minutes:
Della Leslie
Meeting
commenced at 10.00 hours.
Apologies:
Sebastien Caillaud (EDF), Martin Hamilton (Independent),
Anton Heinsbroek (Delft Hydraulics), Arno Kruisbrink (Delft
Hydraulics), Ruud Lemmens (Delft Hydraulics), Simon Pugh (ESDU), Bjørnar
Svingen (Kvaerner),
Patrick Vaugrante (EDF), Lixiang Zhang (Kunming University of
Science and Technology).
Preliminary
Items
- Keith Austin was welcomed back as
chairman
- The group wished Bruno a happy
birthday.
- The minutes from the last meeting
were approved.
- A revised agenda was distributed
(due to cancellations).
Item 1.
Practical
guidelines for FSI in pipelines - progress and results
+ life in France
Della
Leslie gave a presentation covering the progress of the current
project at Dundee: Practical guidelines for FSI in pipelines. She
began by listing the three objectives the project aimed to complete.
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.
Each
of the objectives was discussed. Alan suggested that draft
“final” reports on objective 1 and on objective 2 should be
completed for the next meeting; plus an item discussing these
objectives. Della specified two examples (for objective 1) already
highlighted: nozzle (EDF) and valves too close to bend. Alan
suggested as examples: bends (free), combination of bends, moving
supports, excitation. Chris asked whether it would be possible to
rank features? Alan noted that the final report (in 18 months)
should list features agreed upon by group. Mohamed asked what is
defined by damage? Arris said that this be defined as unwanted
system behaviour.
Della
said there was too much work possible for the project and it was
essential to reduce this to manageable amount. Keith said that FSI
needs lots of data and you can reduce data needed by considering
part of system. Della agreed, saying that many guesses were unsafe
because of system sensitivity and that in analysis needed to
determine the extent of this sensitivity.
Arris
said that education is important and Keith added that there was much
ignorance, even for waterhammer. Mohamed pointed out that most
people want simple things (for guidelines). Chris asked whether the
aim at the end of project would be an addition to standard codes?
Alan emphasised that never aimed to get that far and this would be a
purpose of future projects.
Della
continued by outlining the work completed so far and the plans for
the remaining 18 months of the project (including Gantt chart). Work
completed included a review paper, which was given at the last
meeting and is to be presented at IAHR meeting at Trondheim in June.
The subsystems to be studied have been chosen, software development
is on-going, experimental work has begun at EDF and in Dundee is
planned to begin in July. Also Della has made the first of two
planned visits to EDF. The next 18 months aims to study in detail
the five sub-systems – in both time- and frequency-domain. A
larger system, which incorporates the sub-systems, will also be
examined to determine if its exhibits features from the subsystems.
Most importantly is the development of guidelines. It was emphasised
that the next group meeting would be of great importance to the
project. Since the previous meeting, there had been a change of the
chosen sub-systems. A meeting with EDF in January had prompted this.
Previously four systems had been chosen – single pipe, single
elbow, 3-D elbow pair and 4-elbow system. The new choice is five
systems (removing the 4-elbow system in favour of retaining
branches) - single pipe, single elbow, 3-D elbow pair, as previous,
plus a single T-piece and an elbow and T-piece combination. Thus,
both planned experimental systems in Dundee are included as
subsystems.
Each
of the analysis approaches, time- and frequency-domain, were
discussed. Della looked at the capabilities of software, questioning
what is each capable of, what do they miss?
In
the time-domain, software is being developed using Fortran. It uses
a MOC approach with reach-back and adjusted material densities to
avoid interpolation. It includes Poisson and Junction Coupling but
no damping. The program is structured so that features are isolated
in their own procedures and can be tested individually. Each of the
coupling mechanisms can be switched off to enable direct comparison
between levels of analysis. Examples of results were shown for
single pipe and single-elbow subsystems.
In
the frequency-domain there are two pieces of software to be used for
analysis: software using Mathcad has been developed for simple
systems and Circus (EDF). Della explained the advantages and
disadvantages of using own/commercial software. Primarily the most
important differences are in the speed of calculation (Mathcad
calculations are very slow compared to Circus) but in commercial
software must be careful to fully understand what each feature
actually does. The aim is to compare own software with Circus to
ensure a complete understanding. The testing of software was
discussed. Della showed an example that compared a single elbow
system to a single pipe. By taking limit the second length in the
single elbow system to zero, the system should behave like a single
pipe – frequency responses showed that this was true. Black-box
type testing of modules was discussed. Della said that she wanted to
be certain that program is without errors before completing complete
parameters studies.
Della
summarised the on-going work in the project and a general discussion
developed. Della showed the group an EPRI report on a project from
1992 (provided by EDF) that presented guidelines for waterhammer
(not all volumes of report were available). Keith said that this
appeared very similar to the EU project (WG1-6). Within this report,
eight benchmark problems (very complex!) were described that aimed
to test commercial software. Della also described a discussion at
EDF with an on-site engineer. He emphasised that one of the main
problems with on-site work is to determine where to put gauges
(quickly) in order to find all characteristics needed. Must be
carefully not to position on nodes, but often there is too much
information provided - global information can mask local
information. Della completed the presentation with a summary of her
visit to EDF. She outlined to aims for the visit, including: to
understand and use Circus; evaluate other software available; learn
about experimental facilities. A discussion of software available
developed, in particular: Fortran90 vs. Fortran77 and Maple vs.
Mathcad.
Item 2.
Modelling Blood flow in large arteries and other things
Chris
Greenshields outlined plans for new/current project with Christina
Giannoppa. He first described the problem of modelling blood flow in
large arteries and proposed a solution method:
Poisson
Coupling is the dominant phenomena in flexible tubes (junction
coupling is usually dominant for rigid pipes). The wave speed in
artery is about 5-10m/s with a disturbance speed of 1m/s (In rigid
pipes 500m/s wave speed plus disturbance of 0.5m/s). Chris described
the need for FSI in blood flow modelling. Wall shear influences
development of stenoses (mechanical action and physiological
processes) and shear depends on velocity gradient at wall. Chris
said that current research doesn’t model problem
as wave propagation phenomena and that no
deformation is used (only 10% affect therefore assumed not
important).
Summary
of the problem: Blood and artery interact. For the blood we need a
fluid material model (factors to be considered/included -
Newtonian/non-Newtonian, non-homogeneous; turbulence; Near wall
behaviour) and for the artery a solid material model (factors to be
considered/included – viscoelastic; large deformation; geometric
model)
Current
modelling of FSI involves commercial fluid dynamics software (FV)
and commercial stress analysis (FE) coupled using an in-house
interface. Chris suggested that a better alternative should treat
fluid and structure in single package (Chris chooses FV package).
There was a discussion of the modelling techniques and what features
are included.
Chris
proposed a revised view of blood flow in large arteries – to view
the blood + wall as single entity, using a parameter such that
1=fluid, 0=solid. This would enable the same set of equations
independent of material, i.e. a general model plus type specific
(solid: large deformation, fluid: turbulence, near wall behaviour).
It also allows for “in-between” materials. He continued by
describing the advantages of using a velocity formulation for
non-linear solids and a pressure formulation, before finally
highlighting the advantages of single fluid/structure formulation.
The
aims of the project are to: validate (get working) the model
presented (using FOAM); implement a viscoelastic model; implement a
formulation to handle large deformation in the solid; validate
against available experimental data and analytical models; and
finally to perform simulations.
Item 3.
Experimental plans
This
item extended item 1 to cover in more detail the experimental plans
in connection with the current project at Dundee. Experimental work
is planned at both Dundee and at EDF:
At
Dundee we intend to use two 3D configurations (2 elbow pair and
elbow and T-piece system). For this Arris has agreed to assist and
will visit Dundee for two weeks in July. Initially we will repeat
older/existing configuration in order to test the apparatus before
beginning the new configurations. The possible problems associated
with the proposed systems were discussed – in particular, because
the system in 3D, maintaining tension in supporting wires. We aim to
keep set-up simple so that can it can be accurately modelled. The
reason for choosing the systems were also discussed – new
configurations not completed before and they should include all
types of motion.
Della
presented some of the worked being carried out and planned at EDF.
Currently they are examining a single elbow set-up. Initial tests
have been completed with the system air filled. This has provided
good results – Della presented some of the results found. The
water filled system appears to have problems (due to air in liquid)
and testing of this is on-going. They also plan to test an in-plane
2-elbow system.
Item 4.
An integrated Air conditioning circuit and cooling circuit
simulation model
Keith
Austin gave an interesting presentation describing a simulation
model for an integrated Air conditioning circuit and cooling circuit
developed within Flowmaster. First he gave an-overview of the models
used for major components of the AC system (compressor, condenser,
evaporator and expansion). The model is a 1D simulation, so that it
is reliable, stable, accurate and fast, but it does need good
performance data. Each component was validated separately. The model
shows how the AC circuit and engine cooling affect each other. The
combined analysis leads to better understanding of different cooling
strategies and the resultant economic effect (main incentive for
such models). A discussion followed, including the merits of 1-D vs.
3-D models and of experimental vs. CFD. Alan emphasised that
inherently 1-D behaviour is better modelled by 1-D model and NOT
3-D, saying that we often allow 3D modelist to claim superiority
when sometimes there are not.
Item 5.
Frequency-domain Analysis
Arris
Tijsseling gave the last presentation of the day covering the basics
of frequency-domain analysis. He began with the fundamental
connection between the time-domain and frequency-domain. First,
examine a single sinusoidal wave in the time domain – this
equivalent to single point in frequency-domain (amplitude and phase,
at given frequency, i.e. 3 numbers represents signal); Two points in
the frequency-domain map to a more complex wave; Many points in
frequency-domain map to a very complex wave and finally a continuous
line (i.e. continuous frequency spectra) in frequency-domain maps to
non-periodic signal.
Next,
he covered the definition of a Laplace transform – in which s is
the Laplace parameter, s is generally contains real and imaginary
component. Using s = iw,
the Laplace transform maps time-domain to frequency-domain and is
used to transform a PDE to an ODE.
In
the frequency-domain the TMM (transfer matrix method) relates
upstream to downstream values of parameters by a transfer matrix
(which includes mechanical impedence, dynamic stiffness apparent
mass and boundary conditions). Arris shows an example of the method
for a single elbow system. Arris notes Abel’s formula, which
relates the limits of frequency to the time domain: limit f =
0 is equal to final steady state; limit f infinite gives
initial state.
Finally,
Arris discusses the frequency domain in relation to the Dundee
experiments: The pressure (using transducer) is measure for 1.5s and
using a discrete Fourier Transform gives spectrum. Similar measuring
the bending data (using strain gauges) can be transform to provide a
spectrum. The results provided good comparison with theoretical
results. Arris concluded questioning the life expectancy of the
Dundee pipes? Perhaps more than 20 years?
Any
other business: Date
of next meeting: October 22nd 2001
Item 6.
Chairman’s Closure
Closure
of meeting: at 17:40:07 hours.
|
