Examination of Witnesses (Questions 19660
- 19679)
19660. Can you tell us, please, not in any sense
of criticism but in terms of published papers, was the paper which
I think is based on Findwave, subsequently published, as it were,
as an authorised and validated publication post-conference?
(Dr Hunt) Are we talking
about one presented at the conference in Buxton in 2004?
19661. Yes.
(Dr Hunt) At the conference
at Buxton, all the papers—an invitation was extended to
have those published in a special issue of the Journal of Sound
and Vibration, and my own paper was published there. The paper
based on Findwave was not published in the journal.
19662. That is after the conference?
(Dr Hunt) After the conference.
19663. What is the procedure for securing publication
post-conference?
(Dr Hunt) The procedure
is that you are invited to submit a paper, essentially the paper
that you presented to conference and it is sent to reviewers.
You then get comments back from the reviewers and the reviewers
will say: "Yes, your paper can be published" or, perhaps,
"No it cannot be", but usually: "Yes, it can be,
subject to the following corrections and changes". You make
those corrections and changes, the paper goes back to the reviewers
and the reviewers then accept or reject these changes, and on
the basis of that the editor of the journal makes a decision whether
or not to publish.
19664. Do you recall the paper at Buxton was
based on the Findwave model?
(Dr Hunt) Yes, the paper
at Buxton was based on the Findwave model. I am sure it must be.
Mr Thornely-Taylor will confirm my answer but I think the answer
is yes. I see diagrams in it which would lead me to believe that
it is Findwave that is used, but I cannot see the word "Findwave".
Yes, it is Findwave.
19665. To sum it up, Dr Hunt, have I understood
that the thrust of your evidence is that mathematical models in
general—Findwave is no different from that generality—to
predict accuracy of 5dB is not something you can support in terms
of relying on a model to that level of accuracy? You say that
a proper figure on the current state of learning is 10dB. Is that
the broad thrust of the point?
(Dr Hunt) That is.
Cross-examined by Mr Taylor
19666. Mr Taylor: Dr Hunt, have you ever
used the Findwave model to predict groundborne noise?
(Dr Hunt) No, I have not.
19667. You have not?
(Dr Hunt) No.
19668. Do you know how Findwave deals with the
sort of non-linearities that you were talking to the Committee
about?
(Dr Hunt) No, I do not.
19669. Who knows more about the uncertainties
associated with using Findwave, Mr Thornely-Taylor, who is an
author of the model, or you?
(Dr Hunt) Mr Thornely-Taylor
will know more about the uncertainties associated with the aspects
of the model itself, but I have been working for many years—close
on two decades—on the nature of modelling and of the data
required of models and of the non-linearities inherent in the
materials used in models. So I would say that I may well know
more about certain areas of the nature of the elements that are
inherent in the Findwave model, but I do not have intimate familiarity
with Findwave itself.
19670. You have never carried out an exercise
using Findwave in relation to groundborne noise and then comparing
that with measured results?
(Dr Hunt) I have never used
Findwave, but I have seen the published results and the published
predictions that have been used by Findwave, and the error inherent
in those is 10dB.
19671. Let us bring up your slide 13, if we
may.[29]
When you are talking about error in modelling by 10dB or 5dB,
you are talking about the difference between measurement and the
forecast at a particular octave band. Is that right?
(Dr Hunt) I would talk in third
octave bands, because I think that is what we have agreed to talk
in but, yes, I am talking about the errors inherent in the modelling.
On this slide I show that two runs of the model, these two runs
as published, are different, and they are also different from
the measured data.
19672. Were the assumptions used the same in
the two runs of that model?
(Dr Hunt) I do not know;
they are two separate, published predictions. It seems a point
to note that two published predictions were different.
19673. If a different loss factor, for example,
was used between the two model runs, would that account for differences
between what might be forecast?
(Dr Hunt) My understanding
is that a different loss factor was used.
19674. So there were different assumptions.
(Dr Hunt) Yes. The question
I have is that if you make the different assumptions in the light
of having the measurements in front of you then you are likely
to shift the curve closer to the measurements because it makes
sense to do so. In the absence of the measured data in front of
you, would it not have been simpler to accept that the blue curve
as is: "I have run Findwave now and this is my prediction",
then, a few months later, when measurements were done we might
have noticed a large difference, but given that the model and
the measurements were available at the same time I wondered whether
the adjustment to the model would have been made.
19675. I think you have explained to the Committee
that your identification of the margin of error of plus or minus
10dB comes, in particular, from your experience of groundborne
noise modelling from your Pipe-in-Pipe modelling?
(Dr Hunt) It comes, in particular,
from my reading of a large number of papers on the subject, my
attending a large number of conferences, the use of my own model
and inter-model comparisons, my understanding of the nature of
damping and all sorts of aspects of my own experience in this
field of study.
19676. The one thing it does not come from is
your experience of Findwave?
(Dr Hunt) Indeed.
19677. In relation to Pipe-in-Pipe, am I right
in thinking that the Pipe-in-Pipe model assumes the tunnel is
a perfect cylinder?
(Dr Hunt) It assumes that
the pipe is a perfect cylinder.
19678. But the Crossrail tunnel is going to
have a concrete invert inside the tunnel, is it not?
(Dr Hunt) I have not used
Pipe-in-Pipe to make any predictions for Crossrail, so I can say
very simply that the Pipe-in-Pipe model would not be a good model
to use to make predictions for Crossrail and I have not used it
for such.
19679. If it would not be a good model to use
for predicting Crossrail, why is it a good model to draw conclusions
about the reliability of the Findwave model?
(Dr Hunt) Because there
are characteristics inherent in the nature of the radiation and
vibration from an underground source which the Pipe-in-Pipe model
throws up and so, in my view, it puts into question the possibility
that the vibration levels around the foundation of the building
are uniform enough to suggest that changes in detail are not going
to make much difference. I think that the important point to make
is that the Pipe-in-Pipe model is used in conjunction with other
models to test their validity by cross-model comparison and I
would be delighted if we could find time to do a cross-model comparison
between Pipe-in-Pipe and Findwave.
29 Committee Ref: A221, Findwave Validation Graph
(WESTCC-9305A-013). Back
| 
