Examination of Witnesses (Questions 3120
- 3129)
3120. Chairman: There is also the size
of the tunnels themselves.
(Mr Thornely-Taylor) Yes,
Sir. The CTRL tunnel is broadly 8 metres diameter and we are talking
about a 6 metre diameter, which is a good deal smaller and the
machines and the whole enterprise is smaller and the consequent
energy emitted as vibration noise is less. But it is not a predictable
effect in anything like the precision that the operation of the
railway is.
3121. Mr Taylor: So when one is looking
at the financial implications of providing alternative sleeping
accommodation, in the form sought by Camden, are you able to assist
the Committee on what that would be?
(Mr Thornely-Taylor) It
would have to be assumed, because we do not have enough information,
that everybody above the tunnel might be in a position to ask
to be put up in a hotel. At the other extreme, during boring through
the London clay it probably would not be significant at all. If
I were asked by the Department for Transport how much would this
undertaking cost, I would have to say we had better assume we
will get applications from people all along the route and even
then it would be hard to know how many.
3122. I want to turn away from that particular
issue on to points that Mr Methold made regarding the design of
the North Downs Tunnel. He has produced his sheet, exhibit LBC
35. Can you explain to the Committee what it is you understand
has occurred in relation to the design of the North Downs Tunnel
and why it is you believe that noise concerns have arisen?
(Mr Thornely-Taylor) Mr
Methold explained something about the prediction method. He said
from his point of view he thought it was plus or minus 8 dB in
the 95 per cent confidence band. Some of the figures I have seen
are a little wider—plus or minus 9, which is an 18 dB difference
between the highest value you might find you get and the lowest
value you might find you get for a central prediction. The reason
for that is that the CTRL prediction model is quite different
from the one we have used on Crossrail and on the Jubilee Line
extension. It is based on a large number of measurements which
were made from operating railways, many of them actually on the
surface and the results corrected to make the results applicable
to a tunnel, with a considerable amount of uncertainty in that
correction, and then to take a wholly statistical approach to
plot the measurements and produce a regression analysis and get
coefficients from that, so you can then generate a number for
a property at a particular distance from the tunnel, at a particular
depth for a particular train speed. If you produce a chart as
they did, exhibited in evidence to the Channel Tunnel Rail Link
Committee, where along one scale is the predicted level and on
the other scale is the actual level, that is where you find you
can be at a prediction of, say, 40 and using Mr Methold's plus
or minus 8, the outturn could be anything from 32 to 48. That
is why we have the odd situation with the North Downs Tunnel of
CTRL saying, quite rightly, they discharged their obligations
because they predicted for below 40 but there are lots of problems
where they are getting more than 40 because they did not actually
predict low enough to include all of the distribution, all of
the 8dB distribution, to get 95 per cent accuracy. So it is not
surprising, it is wholly to be expected, there would be some properties
over 40. Because of the difference between that prediction and
the Crossrail prediction methods, which start from the complete
opposite end of the range of possibilities, what we do is to set
up a fine mesh which represents mathematically all the characteristics
of the physical world, the real world, and mathematically move
a train through the model and, in time steps of a few milliseconds,
each work out what every part of the modelling will be like, what
movements take place as a result of the rolling of the wheels
of the trains, and then, in the next millisecond, that propagates
out as vibration. This is done for thousands and thousands of
time steps and we actually model in a computer exactly what goes
on in the real world, so its uncertainty is limited solely on
the fact that we do not know anything about the real world as
we cannot see everything beneath the soil. Having done that, we
then went through a validation exercise, in fact, of two kinds.
One was to model and to measure the groundborne noise from a railway
where we could know almost everything there is to know from the
rail running and the wheel running upwards, and that was the Docklands
Light Railway Lewisham extension between Cutty Sark and Greenwich.
The report of that, the specialist technical report, is on the
web. That showed that in the worst case there was an unpredicted
3 dB LAmax. The other piece of validation work that was done was
on the Jubilee line extension. I mentioned on Day Seven that it
was very hard to find anywhere that would pick up groundborne
noise but it was possible to hear passing trains in the second
basement of Christies in King Street, which is also a nice, undisturbed
place to be. I both measured passing groundborne noise from the
Jubilee line extension trains and predicted, using the same model,
for that site. That, of course, included an uncertainty that was
not in the Cutty Sark/Greenwich measurement. We did not know the
wheel and rail roughness, so that included a test of the appropriateness
of the general assumptions we made about what the wheel and rail
roughness would be. The error there was 2.5 dB(A). In fact, we
combined those two and applied a 5 dB uncertainty correction to
the predictions. If I had given you the computations as they actually
came out of the computer they would have all been 5 less and with
a 5 dB uncertainty margin. That is why 40 is in D10 as the objective
for the operation of Crossrail.
3123. So to avoid the sort of difficulty that
seems to have arisen in the North Downs tunnel situation occurring
with Crossrail, what is it that you do in the design process?
(Mr Thornely-Taylor) We
allow for the uncertainty in the predictions to the extent that
the probability of it turning out worse than the prediction as
presented after the uncertainty is so small as to be not a significant
risk.
3124. Thank you. Now let us turn and deal with
35 dB(A) as a design criterion, if we may. What evidence are you
aware of that establishes that a 35 dB design criterion would
make a material improvement to the living conditions of the people
above the tunnels, compared to the design criterion of 40 dB(A)?
(Mr Thornely-Taylor) I cannot
point to any scientific or academic work, and I think Mr Methold
said the same, which would demonstrate that.
3125. In the information pack, which I think
you have got before you in the blue file, there is produced some
data relating to complaints, I think, about the Victoria line.
Is that correct? In tab B?
(Mr Thornely-Taylor) Yes.
3126. As I understand it, Mr Methold does not
believe that this data should be given weight because it represents
a complaint threshold. What do you say in response to that?
(Mr Thornely-Taylor) Mr
Methold, in one of his exhibits—I do not have the number
in front of me—was suggesting that complaints are not a
good indicator of satisfaction with things. He was suggesting,
I think, that about 5-10 per cent of people complain from the
population that is upset about something. But we find, not so
much in the tables in front of us but other London Underground
research from which Mr Methold drew his figures on the number
of people affected from 35 to 40 dB(A), that the percentage of
complaints at 40 dB(A) is .05 per cent. It was misprinted on the
front page of that document as .5 per cent but in the body of
the document it was .05 per cent, which is so much lower than
the 5-10 per cent that Mr Methold postulates as being the proportion
of people who complained among a population that are disturbed.
That actually is a strong indication that people are much less
disturbed at 40 dB(A) LAmax.
3127. Mr Methold made criticisms of the accuracy
of the numbers in the document setting out the number of dwellings
that were affected. I am rapidly searching through these documents
to find the right page. It is LBC26. I think the general thrust
of what he was saying was that the numbers on the left-hand column
need to be reduced by 10 to take into account the points that
he has made in the bullets beneath the table, namely to use LAmax
FAST, and over-prediction for rail roughness as well. Can you
just explain your views on that, please?
(Mr Thornely-Taylor) Yes.
Mr Methold was not really quite right about the LAmax F point.
It was, I think, put to him by Mr Clarkson how did they do the
measurements in 1969. In fact, right up to 1994 and beyond they
used not a sound level meter but a graphic level recorder which
instead of showing you a number on a dial or on a screen uses
a pen to write a chart on a continuous roll of paper. It is neither
fast nor slow; it is governed by the writing speed of the chart
recorder. That is a problem we used to have in those days when
those instruments were widely used, and how do we relate the writing
speed to fast or slow. The answer is you cannot precisely, and
they did what is best described as "eyeballing"; they
looked at the chart and they say explicitly in the form in which
these figures are drawn they did not take the highest excursions
of the chart, they took what looked like a good figure to represent
the maximum for the pass-by noise. So it is not fast; it is not
that different from LAmax slow, in fact. The other part of the
10 dB, I think, he was attributing to over-estimated rail roughness
as one of the causes. I have to say that I do not know if there
was an over-estimation in 1994 but there are many lines which
have got worse since 1994. So even if there might have been an
over-estimation then I would not like to say there is now. I have
been revisiting some old sites recently and on one occasion found
it was worse by 7 dB(A). I think I referred to it on Day Seven.
I do not agree at all that you should shift the noise level scale
down by 10 dB(A), or down at all, in fact. Given that 56,000 to
57,000 is a very small proportion of London's population and even
a small proportion of the population of people who live above
underground lines, it feels right. All of us know we have been
to people's houses where you hear trains—it is very common—and
it is not at all surprising that it is tens of thousands.
3128. Mr Taylor: Thank you very much.
Is there time to put another point at this stage, given the time?
Perhaps we can continue tomorrow.
3129. Chairman: That is a very good idea.
The Committee will meet tomorrow morning at 10 am.
Adjourned until tomorrow at 10.00 am
|
