Note: Descriptions are shown in the official language in which they were submitted.
~2~ ~87~
o96/11395 l ~ S/~
~m for m~lr;n~l~lt~qnn~lly ~ e~ic
~l~Lies of a m3ving p~ web.
This invention conc~rns the measurement of the velocity of
ultr~o~ln~, in-plane, for a moving paper web. The ultrasound
velocity in paper is known to be related to various measures
of paper ~LLenyLh and stiffness.
BACKGROUND OF THE lNv~NllON
The most important values for the papermaker to consider from
ultrasound velocity measurements on paper web are:
lO TSO Tensile Stiffness orientation, i.e. the orientation
of the elastic ~o~lLies in-plane of the paper
sheet,
TSIMD Tensile S~; ffn~c Index in the machine direction of
the paper machine,
15 TSICD Tensile Stiffness Index in the cross direction of
the paper machine.
It is possible to determine these quantities and also the
anisoL~u~ ratio TSIMD/TSICD by performing ~he ultrasound
~elccity measurements ;~n the machine direction (MD), cross
direction (CD), and directions between (MD) and (CD). The
tensile stiffness and anisoL~v~y ratio characterize the paper
quality.
The velocity of an ultrasonic pulse propagating in-plane of a
paper sheet corresponds with the sheet's elastic properties,
i.e. the TSI. TSI can be compared to Young's modulus (or "E-
modulus") for other materials. The relation~h;p can be
expressed by:
TSI = v2 * c
where TSI is measured in kNm/g, ~ is the propagation velocity
(km/sek) for the ultrasonic pulse, and c is a dimensionless
constant close to l dep~;ng on Poisson's ratio for the
paper. The velocity is easily determined by measuring the
propagation time for an ultrasonic pulse between a
~ ~ WO96111395 ~ 2 ~ 1 8 7 4 ~ J~
transmitter and a receiver.
These quantities are often measured statically on samples
taken from a paper we~. Houevel it is desirable to measure
these paper quantities on-line by an on-line meter used as a
sensor for the continuous co~ ol of a paper manufacturing
~ i~--S .
Most of the known on-line meter arrangements (U.S. Pat. No.
4,291,577, U.S. Pat. No. 4,688,423, U.S. Pat. No. 4,730,492)
employ rotating whe~ which contain transmitters and
receivers of ultrasonic waves. These wheels are rotated by a
moving paper web, which requires a direct physical contact
between the wheels and the web. The ultrasound velocity is
usually determined from the delay tIme of an ultrasonic
signal between the particular transmitter and receiver.
.
In order to obtain a reasonable mea~ ~t accuracy, the
wh~ls must be synchronized which makes the system extremely
complicated and unr~ hle~ An arrangement described in U.S.
Pat. No. 4,688,423 oveL~- ?~ this drawback by exploying disk
type transducers which can be excited continuously and,
therefore, synchronization of the wheels is not necec~ry.
However, the arrangements described in the above-mentioned
patent ~r~c; fications need a direct me~h~n;cal contact
between the ultrasonic trAnC~nc~rs and the web.
In a papermaking machine the fast moving web vibrates in the
direction normal to the web surface, creating a randomly
changing force applied to the wheels. The amplitude of
excited and received ultrasonic waves depends on the pressure
between particul~ ultrasonic transducer and the web. Due to
the randomly changing force, the amplitudes of received
5t~n~1c fluctuate, thereby making the results of measurements
less accurate.
The physical contact with the web is not ~e~ if ultr~c~n;c
waves are excited and detected optically, as described in
~ F ~ 2 0 ~ 8 7 4
~ .
U.S.Pat. No. ~, 025, 66~. Ultrasonic waves in the paper web
are generated by means of a laser. This wave is detected at a
determined distance from the excitation point by means of
another laser beam, reflected from the web. The velocity of
the ultrasonic wave is found from the measured delay time
between the excitation instant and the time of the wave
arrival.
The disadvantage of this optical system is that the amplitu-
des of the ultrasonic waves propagating in-plane of the web
are very small. A very strong acoustic noise exists in paper-
-making ma~h;ne~, which is ~c~m~anied by the vibrations of
the moving we~. In ~act this makes the optical detection of
the lowest orders symmetrical Lamb waves impossible, and only
these waves are suitable for the stiffness and tensile
strength measuraments of paper.
A method and device for continuously determ; ni n~ the modulus
of elasticity o~ adv~n~;~ flexible material, such as paper
web, in a contactless ~ashion is disclosed in WO91/17435. An
ultrasonic wave train is transmitted through the air towards
the web. Fig. 6 shows an embodiment in which the ultrasonic
waves scattered through the air by the material are sensed
both at a distance d and at a distance d at the same side of
the web, no reference ultrasonic wave receiving means being
provided for receiving a reference ultrasonic wave from the
transmission point. The measured distance is between the two
pick-up ultrasonic wave receiving means, thus not between the
position where the ultrasonic wave is generated and the
position of the pick-up ultrasonic wave receiving means.
Other prior on-line paper measuring ~y~lls are disclosed in
the U.S.S.R. Pat. No. 489018 and U S.S.R. Pat. No. 489036,
and described in the publication by Kazys (the same inventor
as for the present invention), Proceedings of 20th interna-
tional conference of ultrasound, Prague, 1976, p. 192-194.
The ultrasound velocity in a moving paper web was determined
by exciting broad band noise-like ultrasonic wave by means of
a dry friction, receiving the ultrasonic wave reradiated by
A~ND~ ~H~T
. 3~ 2 2 0 1 8 7 4
the web by two non-contacting u!tr~sonic re-eivers and _al-
culating cross-correlation function between these tw~ recei-
ved signals. The first receiver was placed opposite to the
ultrasonic transmitter and the second a determined distance
from the transmitter along the web.
In order to improve signal/noise ratio, a rotating cylinder
was placed underneath the we~ close to the second ultrasonic
receiver. The delay time was determined from the delay of the
peak value of the cross-correlation function. The advantage
of this measuring system compared to the ones described above
was that it had no moving or rotating parts.
The disadvantage of the system described in the above-
A~A~?qr.-,,,~,L~---
~ WOg6/11395 4 2 2 ~ 9 8 7 4CI/SE;95/011~14
~ mentioned ~SSR-patents is that the signal/noise ratio is not
sufficiently high enough to permit reliable continous on-line
measurements in a mill environment. Another disadvantage is
that excitation and reception of the ultrasonic waves are
performed ~rom the opposite sides of the paper web. Another
problem which is ~n~o-lntered in performing measurements in
other directions than the machine direction ~MD~ is that an
even worse signal/noise ratio is then obtA; n~ due to the
higher losses of ultrasonic waves in an anisotropic material.
The main object of the invention is to provide i~vv~d noise
robustness for the system in a paper mill environment.
Another object of the invention is on-line measuring system
with single side access to the paper web, performing
measurements at different directions in-plane of a moving
web.
Still another object of the invention is to provide an
im~oved signal processing method for reliable determination
of the ultr~o~n~ velocity in the paper web.
The main object is achieved with a system having the
characterizing features disclosed in the main claim. Further
~eatures and further developments of the invention are
disclosed in the subclaims.
SnMMA~Y OF THE lN~ NllON
The present invention solves the problems associated with the
prior art and other problems by providing a system for
continuous measurements of the velocity of ultrasonic waves
in a moving paper web. The foregoing is accomp~i ch~ by
exciting a ultrasonic wave, such as broad-band noise type
Lamb wave, in the web, receivin~ contactlessly the ultrasonic
wave reradiated by the web at least at three different
points, and determining the delay time between the received
signal, received directly by a first reference receiver
microphone placed in the vicinity of the excitation point,
and the added other two signals, received by pick-up receiver
- 2~ ~874
WO96/11395 ~ J~h~ 1144
microphones separated by a half-wavelength in air of the
transmitted ultrasonic wave at the centre frequency of the
frequency band used for measurements.
The distances between the excitation point and the two pick-
up receivers is known. The delay time of the ultrasonic wave
is prefera~ly determined as a zero cross of the ~;lh~rt
transform of the cross-correlation function of the received
signals, corresponding to the maximum value of the cross-
correlation function. The source of ultrasonic waves and allthe ultrasonic receivers are preferably located on only one
side of the web.
In order to ma~e the system noise ro~ust, i.e. provide a low
signal/noise ratio, the receiving of the reradiated
ultrasonic waves is performed above a rotating cylinder in a
paper -k i ng machine at the particular position in respect to
the line, where the moving web touches the cyl;n~r for the
first time. ~he broad ~and noise-like Lamb wave in the web is
generated by means o~ dry friction contact between the moving
web and a friction head. Therefore, the system has no moving
parts and all signals are received by nor. ~o~lLacting means.
BRIEF DESCRIPTION OF '1'~1~ DRAWINGS
For a more complete unders~n~;ng of the present invention
and for ~urther objects and advantages thereof, reference is
now made to the following description taken in conjunction
with the accompanying drawings, in which:
FIG. l is a schematic side view of a measuring system
according to the prior art,
FIG. 2 is a schematic side view of a first emho~ nt of a
~ ~ing system according to the invention,
35 FIGs 3A to 3D are diagrams of signals provided in different
operation steps in searching for the delay time of
the ultrasonic wave transmitted through the paper
web,
FI&. 4 is a flow chart of the ~Loce sing operation for
22Q ~74
WO96/11395 - 6 P~ s5lcll44
providing the delay time of the ultrasonic wave in-
plane of the paper web,
FIG. 5A is a schematic side ~iew of a C~con~ embodiment of
a measuring system according to the invention,
FIGs 5B and 5C illustrate schematic view from above of two
embodiments of the system in FI~. 5A having the
possibility of measurtng the ultr~olln~ velocity in
different directions,
0 FIG. SD illustrates a graph to provide an extrapolated
value of the ultrasound velocity in the cross
directiont and
FIGs 6A, 6B are perspective views of an emho~;ment of a
transmitter/microphone element.
i5
With reference to FI&. l, a prior art on-line paper measuring
system disclosed in the U.S.S.R. Pat. No. 489018 includes a
friction head l provided on one side of a moving papeT web 2
and generating a noise-like ultrasonic signal VW as a result
of dry friction between the head l and the web 2. A random
signal with a normal law of distribution up to 70 to so k~z
is excited. ~he part o~ this signal VW propagating in the
paper web 2 as the zero order symmetrical Lamb wave sO is the
interesting one to examine. The excited wave is reradiated
partially into the ~u~o~ li nq air and, is picked up by a
contactless reference microp~on~ Mic l provided opposite the
head l on the other side of the web 2, and by a contactless
pick-up microphone Mic2 provided on the same side of the web
as the reference microphone Micl but a determined distance
away from, i.e. downstream from, the head l along the web in
its moving direction, below called "the machine direction".
In order to have an enhanced reradiation of the propagated
wave from the web to the air the web 2 is ~o~Led by a
rotating cylin~T 3 opposite the pick-up microphone Mic2. The
signals from the microphn~c Micl and Mic2 are fed to a
processing unit 4', which correlates the two signals in order
to derive the propagation time through the web, so that the
velocity of the ultrasonic wave in the paper web can be
computed and the result presented on a display 5'.
` 22~ ~1874
WO96/11395 ~ PCT/SE951oll~
In accordance with the invention measures are taken to
-enhance the signal'noise ratio of the correlated signals,
particularly in a :! . . isy en~ironment. Therefore, in accordance
with a first embod_ ent of the invention, shown in FIG 2A, a
double ~-h~nnel measuring receiving microphone device is
provided at the vicinity of teh rotating cylinder 3 to
receive the wave propagated along the web, since the lowest
signal/noise ratio is obt~; n~ at the input of the microphone
Mic2 in the prior art system shown in FIG 1. It is, however,
to be noted that more than two pick-up microrho~s can be
provided according to the invention.
In accordance with the invention at least two pick-~p
ultrasonic microphones Mic2a and Mic2b, being the pick-up
elements of the pick-up receivers, are placed a distance lm
from each other the distance being rhOS~ to be a half-
wavelength of the ultrasonic wa~e in air at the centre
freguency of the bandwidth of the ultrasonic wave transmitted
through the paper web. The microphone Mic2a is located
opposite the contact line C1 between the rotating cyl in~ 3
and the web 2 from which the best radiation into the air of
the wave propagated in the web is provided. The microphone
Mic2b is located on the side of the microphone Mic2a turned
away from the friction head 1. The lateral dimensions of the
pick-up microphones Mic2a and Mic2b,~and also of the
reference microphone Micl, are at least 10 times less than a
wavelength of the ultrasonic wave in the paper we~, and all
the microrhon~c are placed at a distance from the web less
than a wavelength of the ultrasonic wave sO reradiated by the
web into air. Noise is also radiated into the air from the
contact line C2 where the web first meets the cyl ;n~r 3,
This noise should preferably be suppressed as much as
possible. Therefore, a noise ~u~e~sir~ shield 6, for
instance made of rubber, is provided a~nd the microphones
Mic2a and Mic2b shielding them from th~ noise from the
contact line C2 and also from ambient noise. Thus, its outer
edge nearest to the contact line C2 is located downstream
this line. The microrh~n~c Mic2a and Mic2b are placed close
to the internal edge of the shield 6.
.
1 8 ~ 4
WO96/11395 8 PCT/SE95~01144
The signal part of interest of the ultrasonic wave VW
transmitted through the web to be indicated is-the sO wave
signal, which corresponds to the symmetric zero order Lamb
waves propagated in the web 2, i.e. the fastest propagating
5 wave.
Thus the principle of the operation is based on a difference
of ultr~oun~ velocities in air (va=343 m/sek~ and paper
(v50-l.5 to 4 km/sek). The signals at the outputs of the
microphones Mic2a and MIC2~ are given by:
U2a(t) Ys (t)+ya(t)+nmd(t)~noa(t~
2b s Ys~ ts)+ka*ya(t~~ta)~kn*nmd(t+~t )+n (t)
where u2a(t) and u2b(t~ are the complete wave signals at the
~uL~uL of the microphones Mic2a and Mic2b, respectively,
ys(t) is the sO wave signal at the ou~uL of the microphone
~5 Mic2a, Ya is the airborne wave generated by the friction
head, nmd is the noise propagating along the -~h;n~
direction at the o~uL-of the microphone, nOa{t) and nOb(t)
are electronic noise and ambient noise propagating along
directions others than the.machine direction, kS, ka~ and kn
are the coefficients reflecting the asymmetry of the
microphones Nic2a and Mic2b for the a~r~liate waves,
AtS=lm/vso is the delay time of the sO wave between the
microphones Mic2a and Mic2b, and ~ta=lm/va is the delay time
of airborne waves between the microphones Mic2a and Mic2b
propagating along the machine direction.
Due to extensive differences in the ultrasonic velocities in
the web and in air, ~t <~t . Furthermore, ~t~<t ~ to l/fo~
where fO is the centerSfre~ ency of the signal spectrum.
Therefore, the ~e~Lral components with freguencies equal or
close to the freguency fO are approximately:
Ys(t)~Ys(t-~ts)
Ya(t)~ Ya(t ~ta)
nmd~t)~~nmd~t+~ta)
Then, addition of the signals from the two microphones Mic2a
and MiC2b gives the following result:
U2(t)=u2a(t)+u2b(t)=(l+k5)*Ys(t)+(l-ka)*Ya(t)+(l-
kn)*nmd(t)+n0a(t)+nob(t)
2~ ~8~ -
o96/11395 9 P~ 5
The coe~ficien~s k5, ka~ kn are close to l, which gives
approximately:
U2(t)=2*YS(t)+~aYa(~)+~nnmd(t)+tnoa(t)+nob(t)]
where ~a and ~n are much lower than l, which indicates that
the amplitude of the sO wave signal is amplified twice and
the amplitude o~ the wave propagating in air along the
machine direction from the friction head is substatially
re~llc~, like the noise propagating in the machine direction.
The electronic noises or the noises arriving from directions
different from the machine direction are not ~ essed and
are added as partially correlated or uncorrelated random
processes.
It is to be noted that the distance lm between the pick-up
microphones could be chos~n in another way, but then the
equations above and the combination o~ them will be changed.
The main feature of the choise of distance is that the term
y~(t) is essentially enhanced and the term Ya (t) essentially
reduced at the combination.
Referring now to an embodiment having the pick-up microphones
half-wavelength of the airborne ultrasonic wave apart, in
order to estimate the velocity of the sO wave a cross-
correlation should be made on the signals from the reference
microphone Micl and the added signals~from the two pick-up
microphones Nic2a and Mic2b. The signals are first amplified
in respective amplifiers 7, 8, 9. The signals from the
amplifiers 8 and 9 are added in an adder lO. The signals from
the amplifier 7 and the adder lO are fed to a processor 13
through h~n~rAcs filters ll and 12, respectivaly. The
proc~cor 13 is provided with a ~ OYL~ ~ for performing an
automatic time delay mea~u~. ~t in order to obtain the
~elocity of the wave in the actual paper web.
The delay time is determined from the cross _~ Lelation
function. For this purpose two methods are combined, namely,
cross-correlation function envelope pea~ detection for a
coarse eYaluation and ze.o ~ossing detection of the cross-
correlation function ~il h~rt transform for the accurate
~ 2 ~ ~ 8 7 ~
wo s6rll3ss :LO P~ sroll44
measurements. Time diagrams illustrating this tP~hn; que are
given in FIGs 3A to 3D. This t~ch; ~ue is efficient in the
case of relatively narrow-~and signals, i.e., when a cross-
correlation function has an oscillating character.
Therefore, as shown in FIG 3A, a cross CVL ~elation function
Rxy(r) between transmitted and received s~ wave signal at the
~L~Ls of the receivers Micl, 7, 11, and Mic2a, Mic2b, 8, 9,
10, 12 is provi~ed
1~ Rxy(~ /T)l~x(t)+nl(t)]*~2Ya~t+~)+~Ya(t+r)+nmd(t+~)]+
n2(ttr)]dt~
where T is the signal duration used for calculation, x(t) and
y(t+~) are the signals from the input channel Nicl, 7, 11,
and the ~u~ r~Ann~l Mic2a, Mic2b, 8, 9, 10, 12,
respectively, and nl(t) is the noise received by the
microphone Micl and n2(t+~) is the added noise received by
the mi~.o~olles Mic2a and Mic2b.
A zero-cross of the ~;1h~t transform of the cross-
correlation correspon~;ng to the maximum value of the cross-
correlated function is made.
Then, the envelope, as shown in FIG 3B, of a cross-
correlation function Rxy(~) is obtained by means of the
~ilh~rt transform:
AXy(~ tRxy (~) + R 2 (~)]
-(see FIG 3C), where ~
Rxy (r) =HtRxy(~)3=l RXy(t)/t~*(~~t)] dt
is the ~il ~rt transform of a cross-correlation function
~ (~) and shown in FIG 3D. FIG 3C shows the detection of
the envelope peak shown in FIG 3B.
In the presence of signals propagating through multiple
paths, the cross-correlation function has a few peaks,
35 ~Or ~ JOr~ling to different delays. Then the envelope function
can be prese~ted as
Axy(~ Ai(~-rdi)
where ~dl~ ~d2 - are the de3ays in the corr~pon~;ng
paths. Therefore, in a general casen, not just one but a few
Z ~ 8 7 4
WO 96/11395 11 1 ~1ID~SI~1144
peaks will be detected. The proper peak is found t~k; ~ into
a~c~.L a prior-knowledge about an expected time of the
arrival and usually is the peak closest to the zero instant.
The obt~; n~ rough estimate of the delay time ~di iS used to
produce a window H(t) in a time ~t -; n the width of which
is slightly less than half a period of oscillation of the
band-limited cross-correlation function
~<to/2
The window is located symmetrically in respect to the
determined de ay time ~d~
l~ for ~d~ /2) < t S di
H(t-rdi)=
0, otherwise.
The accurate delay time estimation is ob~ine~ from the
windowed ~;lh~t transform RW(t) of the initial cross-
correlation function:
RW(t)=H(t-~di~ ~ (t)
The peak value of the envelope function Axy(~) c~LLe~onds to
the peak value of the cross-correlation function Rxy(~) only
in the case of non-~i~p~sive propagation. As it was noticed
above, the symmetrical s0 wave used for the measurements
propagates without a noticeable ~;cr~rsion.
On the other hand, the ~l~e~Lainty in detecting the rough
delay time cho~ be less than to/2. For 35 kHz center
frequency, rough delay time uncertainties of as much as
to/2=14 ~s can be allowed. Usually this requirement is easy
fullfilled and no am~iguity O~uL~.
The peak values of the cross C~L elation function Rxy(~)
correspond to the zero values of the ~;l~rt transform
~ (r). Hence, the time of signal arrival now can be found
using simple zeLo _Lossing te~hn;que (FIG. 3D~:
~ (t)t ~dl=H(t ~d1)* ~ (~ dl='
It is worthwhile to remember, that by shifting the window
function H(t) to the locations of other envelope peaks
the accurate delay times of s;gn~le propagating tLl~uyh
different paths may be automatically determined.
2~ ~87~
WO9611139S 12 ~ J~ 44
A flowchart of a p~Gyr ~m in the pro~so~ 13 for
automatically deriving the time delay is shown in FIG. 4 and
includes shifting of the window ~, shown in FIG 3D, in
several steps in order to find the searched time delay ~d for
the paper web 2.
The algorithm consists of three main stages: cross-
correlation envelope function fitting by 2nd order
polynomial; f; n~; ng the peaks; and f i n~; ng their
A~i fication according to a sharpness.
The algorithm starts from the window generation in the time
domain. The width o~ the window is given in terms of sampling
points and defines the number of points used in the analysis.
The window is shifted step by step in sl~hC~quent algorithm
loops. The size of this step defines the separation between
two neighbouring peaks and can be chosen in such a way that
minor peaks caused ~y a random noise or spurious waves would
~e ignored.
The cross eoL ~ elation envelope ~unction ~itting is needed for
fi~inq the peak and is performed by the least-square method
using the 2nd order polynomial. Such a polynomial can have a
positive or negative -u~v~Lule ~p~n~;ng on what kind of
2~ local extremity - a peak or a minimum has been found.
Strictly ~re~k; n~, the 2nd order polynomial fitting always
finds a local minimum or maximum independently of how they
were created - by delayed signals or by random noise
fluctuations. The influence of local fluctuations can be
re~llc~ by increasing the width of the window. Then the peaks
c~l~c~ by delayed waves are usually sharper than the other,
spurious, peaks.
Therefore, the peak f;n~;ng proce~n~e consists of the first
order derivative calculation, which enables the determination
of the locations of all exL ~mities and the 2nd order
derivative calculation, which allows sorting them into
maximums and minimums and, ronc~quently, selection of the
~20 ~874
W096/11395 13 PCT/SE95/011~
proper pea~ (or peaks) according to its (or their) sharpness.
The sharpness ~ is given by the magnitude of the 2nd
derivative of the peak.
The delay time estimate rdi obtA; n~ from this peak is used
to generate the window H(t) mentioned a~ove.
The ~;lh~rt transform of the cross-correlation function
RXy(t~ is multiplied by the windowing function H(t~. All
these functions are discrete in the time ~ -; n . The,spacing
between two ad~acent points is equal to the sampling period
~tS. In order to o~tain measur~ent errors less than the
signal sampling interval ~tS, the se j -nt of the Hilbert
transform is fitted using the least-square method by the 5th
order polynom-ial. Then the Equation has five roots, but only
the root inside the created window is selected. This root is
a fine time delay tdi estimation. The wave velocity
vo=lo/tdi, and the tensile stiffness TSI=cl~vo2, where cl is
a ~; ^n~ionless constant close to 1 der~n~;ng on Poisson's
ratio for the paper. The flowchart in ~IG 4 is believed to be
self-explanatory and is therefore not described in further
detail.
It is n~C~sc~y to point out that if the peak of the cross-
correlation function c~nc~ by the sO-~amb wave is the
biggest, then the envelope function fitting can be omitted
and the rough estimate of the peak delay obt~; n~ directly
from the measured cross-correlation or envelope function. The
other steps in the algorithm remain the same.
O
From a commercial point of view, a ~-snring system in which
all units are located at the same side of a paper web has
many advantage; However, in order to implement the single
side access approach it is n~c~c~ry to ove~e a 7 ot of
problems.
l. According to prior art (FIG l), the reference microphone
could not be put at the same distance from a signal
source as the ~on~ ~h~nn~l microphone from a paper
22~ ~874
WO96/11395 14 ~ s~0ll44
web, because both the reference microphone and the
signal source had to be located on the same side of the
web. For the same reason the reference microphone
surface could usually,not be perpendicular to a
propagation direction of ~he signal in air, and that
caused a significant reduction in a normalized cross-
correlation (covariance) function value or a distortion
of its shape.
2. The location of the signal source unit and ~oth the
reference microphone and the receiving microphone (see
prior art in FIG l1 for the waves propagated along the
web on the same side creates a direct wave propagating
in air that is much ~ ~ Gl.~er than in the case of a two-
side ~rC~cs, due to no shielding of airborne waves,
h~C~l-ce then the paper web is not shi~l~; ng the airborne
ultrasonic waves. It r~-~c~s a degree of correlation
between the transmitted and recieved signals too.
3. The ~riction head causes an abrasion of the paper and
scrapes off fibres which produces dust. If it is placed
on the same side of the web as the microphones this dust
will ~e tra~ Led to the microphones, which will
reduce not;ceAhly their sensitivity and distort their
frequency ~e~ol.ce, if the same kind of friction heads
are used as in prior art.
Therefore, a new kind of friction head 20 adapted to a
reference microphone 2l,is provided according ~o a further
development of the invention illustrated schematically in the
second embodiment of the invention shown in FIGs 5A, 5B, and
5C.
The main feature of the combination of the friction head and
the reference microphone is that friction and microphone
elements are provided symmetrically to each other. This means
, that there could be one friction element and an even number
of microphone elements provided symmetrically in relation to
the friction ,element such that the microphonos in each pair
2 2 ~ 11 8 7 4
WO 96111395 lS P~ ,J'~,1144
have the same distance to the friction element, or there
could be one microphone element and an even number of
friction elements placed around the microphone element. The
~riction elements have preferably a nearly pointlike contact
with the paper web.
.
However, friction elements will cause dust in the environment
and measures must be taken to minimize the influence of dust
on the microphone(s). Thus, the prefered embo~i nt is to
o have a microphone between two pointlike friction elements
placed along a line perp~n~ r to the machine direction,
i.e. the moving direction of the web. I~ more than two
friction elements are provided they must all be provided at
the side of a line through an ultrasonic sound receiving
element of the reference means directed in the machine
direction of the moving paper web in order to prevent dust
from coming directly on the microrhon~.
An emho~;ment of the unit of the friction head and reference
microphone is shown in FIGs 6A and 6B. ~IG 6A shows the
actual appearance of the units when a noise shield is
provided, and FIG 6B shows the unit without the noise shield.
The new friction head comprises two friction parts 22 placed
along a line perpendicular to the machine direction of the
paper web. The friction parts are preferably made from a hard
alloy material, for instance wolfram carbide. The friction
parts 22 are held by a holder 23.
The reference microphone 21 is located between the two
f~iction parts 22 of the friction head 20. The distance D ~
between the friction parts 22 is much less than the
wavelength of the ultrasonic wave in the web. Also, the
dimensions of the contact area between the friction head 20,
and the paper web are less than the wavelength of the
ultrasonic wave in the web. Thus, this kind of ultrasonic
sound source acts substantially like a two-point-source. The
distance between the contact areas and the pla-ne of the
reference microphone is comparable with the wavelength in air
(for fO=40 kHz, ~a/2=4~3 mm).
,
8 7 ~ -
- ~ W096nl3g5 16 P~~ I44
- In order to provide a good correlation between the signals
~Luled by the reference microphone 21 and the microphones
Mic2A and Mic2B located at the rotating cyl ;n~r 3 (FIG 5A)
it is n~cPcc~ry that the reference microphone 21 is placed as
s accurately as possible at the same distance from the contact
areas between the friction parts 22 and the web. In order to
reduce the waves radiated other than by the contact area by
the friction parts and transmitted through the air to the
reference microphone, a noise shield 24 lFIG. 6A) is placed
~round the ~riction parts 22 and held by their holder 23 and
is provided with an opening adap*ed to hold the reference
microphone 21 in place.
Each of the two parts of the friction head 22 in FIGs 6A and
6B are formed as hem;crheres. The friction parts of the head
22 are made of hard alloy and comprise tips, covered by a
material absorbing ultrasonic waves, for example, a soft
rubber contacting the web.
Referring ~ack to FIG 5A, in order to make an extra shield
for the microrho~c Mic2A and Mic2B, ~oth regarding the
air~orne noise from the friction head and against the dust
from it, a number of shields 35 are provided above the paper
web between the microphone 21 and the rotating cyl; n~Pr 3.
Also, as in the embodiment shown in FI& 2, a noise reducing
shield 36, for instance made of rubber, is placed around the
microrhonPs Mic2A and Mic2B in order to reduce the noise from
the noisy ~ v~ l; ngs . The shield 36, having the same
function as the shield 6 in the ~ hoAiment shown in FIG 2,
has preferably the shape of its lower side adapted to the
shape of the paper web when it is transferred over the
rotating cyl ;n~e~ 3, as seen from FIG 5A (as well as from FIG
2).
The method above has been described for measurement of the
time delay in the machine direction and this will give the
tensile stiffness index TSIMD in the machine direction of the
paper machine. The friction head 20, the microph~n~c 21,
Mic2A and Mic2B are then located in line with the mar-h; n~
2 2 0 ~ 8 7 4
WO9611139S 17 r~-~
direction. However, as mentioned in the i~ uctory part of
the specification, the tensi-le stiffness ;n~Y TSICD in the
cross direction of the paper machine, and in directions
between TSIMD and TSICD, are also needed in order to
calculate the anisG~y ratio and the tensile stiffness
orientation. An : ~o~i nt for providing also these
quantities will now be described with reference to FIGs 5B
and 5C, even though the same feature naturally can also be
provided for the : ho~; ~nt shown in FIG 2.
As is apparent from FIG 5B, sev~ al sets of microphones
Mic3A, Mic3B; Nic4A, Mic4B etc are shown located parallel to
each other and oblique to the microphone 21 in relation to
the machine direction (the respective angular directions ~N-
lt ~N etc)~ such that each microphone Mic3A~ Mic4A is
situated tangentially in the same location above the rotating
cyl; n~ 3 as the microphone Mic2A. The delay time of the
symmetrical Lamb wave propagating in that o~lique direction,
~N-l ' N etc, is measured in the same way as described above
for the ultrasonic Iamb wave propagation in the machine
direction t~kin~ ac~ of the somewhat longer propagation
path for each set.
Instead of providing an array of receiving pick-up microphone
sets only one set need be provided, said set being movable
along the cylin_-~ above the web so as to be placed in
different oblig~: pos~tions, i.e. sc~nn;~g along the line Cl.
In this instance it is important to place the set of
microph~n~c in accurately precise positions above the paper
web (same distance to the web and along line Cl) in order to
have the same measuring conditio~s for each measured oblique
settina fnot shown in a separate figure, however the pick-up
microp~ ~ set will be placed in the same way as shown in F-G
5B).
Another emho~; ment, shown in FIG 5C has only one pick-up
microphone set Mic2A', Mic2B' and moves, as a unit, friction
head 20 and reference microphone 21 across the web, for
instance along a straight line Fl parallel to the line Cl, as
,
- 2~ ~8~
096/ll3g5 18 ~ 44
` ' shown, and to derive the delay time for the sO wave for a
oCo~ amount of settings of the unit 20,21 having different
angular positions in relation to the pick-up microphone set.
Tt is also possible to move the friction-head/microphone set
20,21 along a curved line F2 (~h~), or to provide the
velocity measurement along the maçhi n~ direction separately
and the measurements in the oblique directions along a line
F3 (dot/~h~A) perpendicular to the line Cl.
It should be noted that even for the em~o*iments having
C~nn; ng elements along a line and one element constantly in
the same position, each measuring result is provided having
both kinds of elements in the same position in relation to
each other during the time it takes to get the measuring
5 result.
Many different kinds of numerical methods may be used to
provide a quite precise estimation a~out the s0 wave rate in
the cross direction of the paper web. One method is to fit
the measured sO wave rates for the different o~lique
positions in some kind of periodic ~unction, e.g. the
function for an ~ ce or some kind of Fourier serie.
Example in which a trigonometric first order Fourier series
is used: ,
We assume that the ultrasonic velocity of the sO wave has
been measured in three different directiohs and these three
different values are used ~or determining constants ao, al
and bl. The constants are then inserted in the following
3~ formula:
f(~)=aO+al*cos2~+bl*sin2~ (1)
The estimated velocity is also ~p~n~nt on formula 2:
f(x)=kl*x+k2 (where x=f~)max/f~)min) ~2)
The constants kl and k2 are known. A combination of the
functions 1 and 2 will give the following function which
determines the s0 wave velocity in the cross direction
(~=90)
v(CD)=f(x)*(aO-al) (3)
By changing the constants kl and k2 it is possi~le to get the
- 2~0~87~
~ ,
19
velocity in any direction from the formula 4:
v(a,max/min)=(kl(a)*x+k2(a))*(aOIal*cos2a+bl*sin2a) (4)
Another advantageable way to derive the velocity of the sO wave
in the cross direction from the results from the different
settings of the friction-head/reference-microphone and the
pick-up microphones in relation to each other is to set the
measuring results o~ the sO wave rates in a coordinate system,
having the rate in the m~h; n~ direction along the X-axis and
0 the rate in the cross direction of the web along the Y-axis, in
relation to the respecive angular deviation a~I, aN etc of each
set to ~he m~,h; n~ direction in the way shown in FIG 5D. A
curve is drawn through the different measuring results and
extrapolated to cut the Y-axis in order to provide the velocity
of the sO wave in the web in the cross direction. A small
extrapolation error is unavoidable but is m;n;m;~ed by having a
lot of settings of the friction-head/reference-microphone in
relation to the pick-up microphones, the more the better.
The same extrapolation t~hnique as shown in FIG 5C can be used
also for the embodiments shown in FIG 5D.
While the invention has been described with reference to
specific embodiments, it will be understood by those skilled in
`5 the art that various changes may be made and equivalents may be
substituted for elements thereof without departing from the
scope of the invention as apparent from the Claims. In
addition, modifications may be made without departing from the
essential t~ch;ngs of the invention. For instance, more than
two pick-up microphones could be provided at the rotating
cylinder.
h~ n,~.
ANt~N~ED Sl lEET
