Present: Keith
Austin (Flowmaster), Christina Giannoppa (King’s College), Shusheng He
(Robert Gordon University), Thorsten Neuhaus (UMSICHT), Ivo Pothof
(Delft), Arris Tijsseling (Eindhoven University), Della Leslie, Alan
Vardy, Jim Brown and (Dundee University).
Chairman: Alan Vardy
Minutes: Della
Leslie
Meeting
commenced at 09:30 hours.
Apologies:
Anton Bergant (Litostroj E.I. d.o.o.), Bruno Brunone (Perugia
University), Mohamed Ghidaoui (Hong Kong)
Chairman’s Introduction:
Welcome
to all. Alan Vardy welcomed the group, particularly the new members to the
group.
Item 1.
Weighting function model of
unsteady friction in ROUGH-walled pipes
Jim
Brown presented some of his and Alan’s current work developing an
unsteady friction model for rough walled pipes. He gave an introduction to
the problem, including previous developments.
Jim
explained how the Weighting function approached, originally used by Zielke
for laminar flow, has been developed over the years. Vardy and Brown have
used this for turbulent flow (various Reynolds numbers) in smooth wall
pipes. This method for unsteady friction assumes a frozen viscosity
distribution, the most recent version being a two-layer model with
constant core viscosity and a linear annulus viscosity. Jim explained that
the values used previously for the core viscosity and annulus thickness
did not seem the most realistic values when examining the data available.
In current work they have change these from 0.07 to 0.065 (core viscosity)
and 0.175 to 0.2 (annulus thickness) (both non-dimensional).
Jim
emphasised that all their papers have assumed frozen turbulent and
viscosity and Alan added that there was clear experimental evidence to
support this and it was a fundamental assumption.
The
main aim of the presentation was to show how this method has been adapted
for rough wall pipes. The effect of addition drag associated with rough is
introduced through an effective wall viscosity. This is found by trial and
error. The resulting equations take the same form as for the smooth pipe
formulation. Jim showed had the equations are solved resulting in an
inverse weighting function. This function is very complex and difficult to
invert. He continues by explaining how this is overcome by approximating
this function by a form that can easily be inverted. The final weighting
function is dependent on the roughness and Reynolds number. There was much
discussion on the results. In particular was the connection with Bruno
Brunone work (estimate for coefficient including roughness). Alan said
that the results were very sensitive to the wall region. Jim said that the
theory was a “leap in the dark” because there is currently very little
(or no) experimental evidence.
Item 2.
Time-dependent turbulence in
transient pipe flows
Shusheng
He gave a presentation on some of his research on time-dependent
turbulence in transient pipe flows. He informed the group that he had just
moved to Aberdeen to take up a post at the Robert Gordon University
(previous he had been at British Energy).
First
he outlined what had been done both experimental and modelling (using
various turbulence models).
There
were two experiments described: ramp-type transient and pulsation. The
first was aimed at understanding basic features of turbulence responding
to flow transients. He described in details how the experiments were
set-up. The test area was approximated 100*diameter of pipe (2”
diameter) and pipe was constructed of plastic and glass. There were many
supports on the pipe – to reduce vibration initiated by pump (this was
isolated using flexible tubing). For ramp-type transient the flow rate was
1.0-6.5m3/hr, and Reynolds 700-4600, and various time periods
used. He showed some of the results and there was much discussion. The
results illustrated a delay in the response (i.e. frozen turbulence) and
it was shown the delay decreased with increasing Reynolds number. He
continued by describing the pulsating flow experiments, again illustrating
the frozen response of the system.
He
continued his presentation describing computational studies. He compared
several models, ranking each according to performance (near-wall response,
core response, delay stage response and general agreement). In particular
the Launder-Sharma model predicted the delayed response quite well. The
results found were discussed.
Item 3.
Unsteady skin friction
before Joukowsky:- Is nothing new?
Arris Tijsseling gave a presentation
on the history of unsteady skin friction. This dated back to 1883, with
von Kries publishing work on flow in arteries. This included work
traditionally attributed to Joukowsky (1898). Arris included a time-line
of developments in unsteady friction (see attached sheet). It was noted
that the references stated would not be common to every research group in
the field – Dr. He said that he would have listed a completely
independent line of research. It was interesting to see that throughout
the history of the field there have been many independent (but parallel)
lines of research. Arris continued by outlining some of the work completed
by von Kries (including experimental work, which would have been very
difficult, but gave excellent results). Arris also highlighted work by
Witzig (1914) on the forced wave motion of viscous incompressible fluid in
flexible tubes. Witzig had defined the quantity z=R(w/n)˝,
commonly known by many names: shear wave number, Wanersley number, Valensi
number but never as Witzig number (even though he was possibly the earlier
person to use it).
Concluding, Arris presented the
group with gifts of Stroopwafel (thank you very much) and kindly displayed
an advertisement selling a Stroopwafel maker (only $75) for those who
wanted an endless supply!
Item 4
Performance of instantaneous acceleration models of unsteady
skin friction in practical applications
Arris
Tijsseling kindly presented this presentation on the work of Anton Bergant
together with John Vitkovsky, Angus Simpson and Martin Lambert (University
of Adelaide). For details of the work please refer to the paper
distributed. Giving a brief overview of the presentation, it included an
outline of unsteady friction model, experimental facilities used, a number
of case studies and conclusions made.
The
experimental facility at the University of Adelaide consists of two tanks
joined by a pipe, 37.2m long and 22mm diameter. There have also been field
tests carried out at a hydropower plant in Slovenia. Results from the
experiments were compared with a quasi-steady friction model and
Brunone’s unsteady friction model. Three Reynolds numbers were tested
(1870, 3570 and 5600). The quasi-steady model did not produce enough
damping in each example and there appeared to be a phase shift. This was
discussed and it was noted that each oscillation started at the correct
position but finished at the wrong place. Brunone’s model performed much
better, producing the correct damping. With the higher Reynolds numbers
there was a greater error in the phase occurred. The next case study was
gradual and slow valve closures in the experimental apparatus (Reynolds
numbers 4300, 7800). The Brunone model produced more damping than the
quasi-steady model, but the difference decreased for the higher value).
Results from tests at the HPP compared to the two models were shown
(Comparison of heads at the valve, Reynolds number 1.06*106).
There was no difference between the two models. These results was
discussed and it was noted that both model produced the initially response
incorrectly and the formulation of the boundary condition was question. It
was suggested that a solution would be to use measured pressure as the
boundary condition (up to the time of full closure of the valve) instead
of modelling the valve.
Item 5
Proposals for experiment measurements of unsteady skin
friction: - reconciling the needs of experimenters and theoreticians
Ivo
Pothof gave the final presentation of the day. He explained the MRI
Proposal (major research infrastructure, previously named LIP), the test
rig at Delft some of the proposed tests.
For
the MRI Proposal, Delft is currently waiting for a reply on it progress.
It will provide money for travel and expenses.
The
rig at Delft is quite large. It can produce constant deceleration
conditions using the two tanks in the system. The test section is a 25m
straight section. There was a discussion on sampling rates required for
measurement of velocity profiles. A fast sampling rate would provide
velocity fluctuation, or repeating the test would provide the mean
velocity. Six proposed test were described and discussed. Ivo said that
the tests described were not fixed (i.e. can be changed). The six tests
were:
A.1:
Acceleration test. Start with no flow; operation open/close tank valve;
finish with no flow.
A.2:
Acceleration test. Start with steady flow; operation open/throttle in-line
valve; finish with steady flow.
B.1:
Deceleration test. Start with steady flow; operation throttle/open in-line
valve; finish with steady flow.
B.2:
Deceleration test. Start with steady flow; operation open/close tank
valve; finish with steady flow.
B.3:
Deceleration test. Start with steady flow; operation open tank valve,
check valve closes; finish with no flow.
C.1:
Oscillation test. Start with no flow; operation periodically open/close
tank valve; finish with oscillatory flow.
Possible
ideas for experiments and measurements were discussed. Arris asked if the
system could be brought to resonance and the damping measured. He said
this should be predominately due to friction and the results could be used
in frequency domain form of models. The distance of the test section
downstream from the bend, or other disturbances was discussed. Alan asked
about methods of obtaining data required at the wall. It was concluded
that there needs to be a new device developed to measure strains at the
wall. Several possible ideas were discussed. Dr He noted that because it
was a large-scale rig it would be possible to measure closer to the wall
(non-dimensionally).
Summing
up, Ivo said that there would be 30 days approximately available for
taking measurements, that various equipment would be available (possible
to obtain more items as necessary) and that the tests that would be
carried out could be changed.
Alan
said it would be a good idea to discuss and agree upon an exact program
and what should be measured and how to measure it.
Other business:
