Title: Viscosity measurement
Abstract: The present invention relates to a method and apparatus of determining the rheological properties of a polymer flowing in a conduit. The invention provides a method of characterising a polymer under test, comprising: Detecting acoustic emissions from said polymer flowing in a conduit to provide acoustic emission data, comparing the acoustic emissions data obtained against acoustic emission data from a polymer, or a series of polymers, of known characteristics, and thereby characterising the polymer.
Patent Number: 6,945,094 Issued on 09/20/2005 to Eggen, et al.
| Inventors:
|
Eggen; Svein (Langangen, NO);
Esbensen; Kim (Larvik, NO);
Halstensen; Mats (Hamur, NO)
|
| Assignee:
|
Borealis Technology Oy (Porvoo, FI)
|
| Appl. No.:
|
451226 |
| Filed:
|
December 21, 2001 |
| PCT Filed:
|
December 21, 2001
|
| PCT NO:
|
PCT/GB01/05771
|
| 371 Date:
|
October 8, 2003
|
| 102(e) Date:
|
October 8, 2003
|
| PCT PUB.NO.:
|
WO02/05224 |
| PCT PUB. Date:
|
July 4, 2002 |
Foreign Application Priority Data
| Current U.S. Class: |
73/54.41; 73/54.01; 73/587 |
| Intern'l Class: |
G01N 011/00; G01N 017/00; G01N 029/02 |
| Field of Search: |
73/5401,542.4,544.1,587
|
References Cited [Referenced By]
U.S. Patent Documents
| 3816773 | Jun., 1974 | Baldwin et al.
| |
| 3906780 | Sep., 1975 | Baldwin.
| |
| 4339944 | Jul., 1982 | Abts et al.
| |
| 4392374 | Jul., 1983 | Liebermann.
| |
| 4779452 | Oct., 1988 | Cohen-Tenoudji et al.
| |
| 4869233 | Sep., 1989 | Stulen et al.
| |
| 4979124 | Dec., 1990 | Sachse et al.
| |
| 5148405 | Sep., 1992 | Belchamber et al.
| |
| 5302878 | Apr., 1994 | Soucemarianadin et al.
| |
| 5317908 | Jun., 1994 | Fitzgerald et al.
| |
| 5433112 | Jul., 1995 | Piche et al.
| |
| 5459677 | Oct., 1995 | Kowalski et al.
| |
| 5568400 | Oct., 1996 | Stark et al.
| |
| 6439034 | Aug., 2002 | Farone et al.
| |
| Foreign Patent Documents |
| 1233752 | Mar., 1999 | CN.
| |
| 0 390 835 | Jun., 1989 | EP.
| |
| 0 317 322 | Jan., 1993 | EP.
| |
| 0933632 | Aug., 1999 | EP.
| |
| 1 346 095 | Jun., 1974 | GB.
| |
| 2 038 851 | Jul., 1980 | GB.
| |
| WO 97/3829/2 | Oct., 1997 | WO.
| |
| 00/33051 | Jun., 2000 | WO.
| |
Other References
Kim et al, "A Study on treeing Breakdown and Fractal Characteristics according
to Method of Acoustic Emission Detection in High Polymer", Proceedings of the 5th
International Conference on Properties and Applications of Dielectric Materials,
May 25-30, 1997, Seoul, Korea, pp. 434-438; XP010242484.
Bettinger, "Microprocessor Based System for the Detection and Characterization
of Acoustic Emissions for Materials Testing", Bradley Dept. of Electrical Engineering,
Virginia Polytechnic Institute and State University, Blacksburg, VA 24061-0111,
pp. 2364-2367, Industrial Electronics, Control, and Instrumentation, 1993. Proceedings
of the IECON '93, International Conference on Maui, Hawaii Nov. 15-19, 1993; XP010109392.
Chen et al, "Measure of Molecular Mass of Polyacrylamide with Intrinsic Viscometry",
Technology Supervision in Petroleum Industry 2001, vol. 1, pp. 22-24 and English
language translation.
|
Primary Examiner: Williams; Hezron
Assistant Examiner: Hanley; John C
Attorney, Agent or Firm: Nixon & Vanderhye P.C.
Parent Case Text
This is the US national phase of international application PCT/GB01/05771 filed
21 Dec. 2001, which designated the US.
Claims
1. A method of characterising a polymer under test, comprising:
flowing said polymer through a conduit in a controlled manner such that said
polymer experiences a predetermined flow rate,
detecting acoustic emissions from said polymer flowing in the conduit to provide
acoustic emission data relating solely to said polymer, and
comparing the acoustic emissions data obtained against acoustic emission data
from a polymer, or a series of polymers, of known characteristics, and-thereby
characterising the polymer.
2. A method as claimed in claim 1 wherein a rheological property of the polymer
under test is thereby determined.
3. A method as claimed in claim 1 wherein the identity of the polymer under test
is thereby determined.
4. A method as claimed in claim 2 wherein the rheological property under test
is the viscosity of the polymer.
5. A method as claimed in claim 1 wherein the acoustic emissions are detected
by means of an accelerometer.
6. A method as claimed in claim 1 wherein said conduit is associated with a pipe
leading directly from a polymerisation reactor.
7. A method as claimed in claim 1 wherein said conduit is associated with an extruder.
8. A method as claimed in claim 1 wherein said conduit comprises a structural
detail altering flow characteristics of the polymer flowing in the conduit.
9. A method as claimed in claim 1 wherein the acoustic emissions data is analysed
using Principal Component Analysis (PCA) techniques.
10. An apparatus for characterising a molten polymer, comprising:
a) means for controlling the flow of said polymer through a conduit such that
the polymer experiences a predetermined flow rate;
b) an acoustic sensor capable of detecting acoustic emissions from the polymer
and thereby generating a signal relating solely to acoustic emissions data from
said polymer; and
c) means for comparing the signal against acoustic emissions data from molten
polymers of known characteristics.
11. An apparatus as claimed in claim 10, wherein said apparatus is adapted whereby
to determine a value for a desired rheological property of the polymer under test.
12. An apparatus as claimed in claim 10, wherein said apparatus further comprises
means for identifying the polymer.
Description
The present invention relates to a method of determining the rheological properties
of a polymer flowing in a conduit. The method is particularly suitable for determining
the viscosity of the polymer, and other properties such as the molecular structure
or chemical composition of the polymer can also be determined. Also provided is
an apparatus for carrying out the method.
Polymers are generally manufactured using chemical synthesis reactions between
one or more basic molecules, known as monomers, which react together under favourable
conditions to form a polymer, which consists of long chains of the monomers joined together.
In general, in polymer manufacturing processes, the composition of the polymer
chain (i.e. the molecular structure of the polymer) is carefully controlled by
adding the monomer(s) to the reaction mixture at a carefully controlled rate. Where
there are two or more monomers, these are added to the reaction mixture in strictly
controlled proportions to one another e.g. in a constant ratio. It is also necessary
to maintain the reaction conditions at the correct levels in order to control the
rate at which each monomer reacts with the other monomer(s) and hence control the
resulting molecular structure. Reaction conditions include temperature, pressure,
rate of mixing, rate of shear etc. Of course, even when a single monomer is used,
such as in the manufacture of polyvinyl chloride (PVC) or polyethylene, the molecular
structure is also affected by the reaction conditions because the length of each
polymer chain can vary and the chains can be branched or unbranched to varying
degrees. This degree of "branching" of the polymer chain affects the physical properties
(e.g. density and strength) of the polymer product.
Clearly then, the molecular structure of a polymer product must be carefully
controlled during polymerisation reactions. However, measurement of the polymer
properties during reaction is extremely difficult. Properties such as the viscosity
or melt flow index (MFI) of the polymer when melted are very good indicators of
the molecular structure, and hence the physical and chemical properties of the
polymer. However it is necessary either to take samples of polymer from the reactor
in order to carry out conventional measurements of viscosity and MFI in the laboratory,
or in some cases an "on-line" rheometer may be fitted in the outlet pipe from a reactor.
Sampling techniques are time-consuming and introduce delays in obtaining
the information—therefore this is not an effective way of continuously controlling
the reaction since by the time the results are analysed and appropriate action
taken, the reaction conditions will be different and some of the polymer product
may already be adversely affected.
On-line rheometers generally work on the principle that a small amount of
molten polymer is syphoned off into a smaller "by-pass" duct, and the rheological
properties of the polymer, such as the MFI or viscosity can be measured. The rate
of flow of the polymer in the by-pass line at a given pressure (or load) is dependent
on the viscosity or MFI of the polymer, at a known shear rate. Hence on-line viscosity
or MFI can be measured. Unfortunately, though, this form of measurement is theoretically
complicated and involves the use of sophisticated and expensive equipment for example
transducers may be needed to measure pressure and flowmeters and sometimes also
gear pumps are required.
Another approach to this problem is found in GB 2 038 051, published in 1980,
which discloses the idea of an "acoustic probe" which can be immersed in polymerising
mixture in the reactor and used to monitor the rheological properties of the polymer.
The probe was intended to pick up sound-wave signals from the polymer flowing inside
the reactor, and amongst other things, it was intended to help to monitor the viscosity
of the polymer by correlating viscosity with the logarithmic decrement of sound-wave oscillations.
However, in order to pick up useful measurements, the probe needed to be
positioned in a region of polymer flow, such as near to the stirring device in
the reactor. This creates practical difficulties in that the probe is liable to
be damaged and is difficult to maintain in position in the reactor. Any device
which has to be immersed in the polymer melt itself is inherently difficult to
operate and is generally best avoided wherever possible. Furthermore, measurement
of polymer properties in the reactor has problems because the properties in the
reactor are not necessarily the same as the properties of the final polymer produced.
In Esbensen et al (1998); "Acoustic chemometrics; from Noise to information",
Chemometrics and and intelligent laboratory systems 44 (1998) 61-76, an acoustic
device is described for use with particulate materials.
Viewed from one aspect, the invention provides a method of characterising
a polymer under test, comprising:
detecting acoustic emissions from said polymer flowing in a conduit to
provide acoustic emission data,
comparing the acoustic emissions data obtained against acoustic emission
data from a polymer, or a series of polymers, of known characteristics, and
thereby characterising the polymer.
Preferably, said method enables a rheological property of a polymer under
test to be determined, by comparing the acoustic emissions data against such data
from a polymer, or a series of polymers, of known rheological properties, and thereby
determining the rheological property of said polymer under test.
It will be appreciated that in order for the polymer to flow and a meaningful
evaluation of its properties to be deduced, it will generally be necessary to melt
the polymer so that it is no longer in solid form. Hence reference to a "polymer
flowing" as used herein should be understood in general as reference to a molten
polymer which is able to flow.
Thus, the invention is based on the discovery that it is possible to determine
the characteristics, preferably the rheological properties, of a polymer flowing
in a conduit, without the need for expensive and complex equipment and without
the need to immerse a probe or sensor in the flowing fluid. It has furthermore
surprisingly been found that the acoustic emissions from a particular polymer are
sufficiently characteristic for each different type of polymer to be identified.
Also, for any given polymer for which molecular structure may differ from batch
to batch or over time during continuous processing, this variation can be monitored.
In fact, the composition of the polymer can be determined from the acoustic emission
data of that polymer.
The composition of the polymer can of course be inferred or determined from any
values of the rheological properties obtained, e.g. from the viscosity of the polymer,
but it will be appreciated that direct comparison of emission data alone from polymers
of known identity can also be made. Thus, in order to identify a particular polymer
according to its composition, a value for the viscosity or other rheological property
of that polymer need not actually be determined from the acoustic emission data
in order to identify the polymer.
Thus viewed from another aspect, the invention provides a method of identifying
a polymer under test, comprising:
detecting acoustic emissions from said polymer flowing in a conduit,
comparing the acoustic emissions data obtained against acoustic emission
data from a polymer of known identity, and
thereby determining the identity of the polymer under test.
In this case, the identity of the polymer may be in the form of an accurate determination
of the molecular structure of the polymer, or it may be simply be an indication
of the type of polymer being produced (e.g. determining whether it is polyethylene,
polypropylene or a particular co-polymer, or even the particular composition).
Rheological properties as referred to herein include viscosity (intrinsic,
extrinsic, kinematic or dynamic viscosity), shear-strain or shear-stress, melt
flow index (MFI) or any other rheological parameter which is characteristic of
a given polymer. [The term "rheological property" as used herein however does not
include parameters such as flow rate or flow velocity, temperature, pressure, load
or pressure drop which may or may not be determined incidentally when the method
of the invention is carried out. These and many other properties of a fluid flowing
in a conduit are not "rheological properties" within the meaning of the invention
since they are not characteristic of any given fluid or polymer].
It will be appreciated by those skilled in the art that rheological properties
are generally determined for a given fluid at a pre-determined or preferably constant
value of the non-rheological properties. Thus for example the viscosity of a fluid
may vary with temperature, flow rate, pressure etc., hence a value of viscosity
should ideally be compared against another at a given temperature and under given
flow conditions etc. Since it is the molecular weight and molecular weight distribution
(MWD) which is of prime interest in controlling the properties and hence quality
of the polymer product, it is a change in any of these properties which is of interest
rather than measurement of an absolute value, in most cases.
In fact, the viscosity of a polymer also varies with other rheological properties,
e.g. shear stress. If a graph of shear stress against viscosity is plotted for
a given polymer, the shape of the curve is indicative of the molecular weight distribution
of the polymer. However, by comparing the acoustic emission data obtained in accordance
with the invention against emission data from known polymers under the same flow
conditions e.g. at a given temperature, flow rate etc., complex calculations of
the polymer properties can be avoided and the identity and/or rheological properties
of a polymer can be determined directly.
It is therefore preferred that the method of the invention be performed by detection
of acoustic emissions from the polymer at a pre-determined flow rate, pre-determined
pressure and/or a pre-determined temperature. In particular, it is advantageous
to control the flow rate of the polymer in order that the shear rate of the polymer
is known. For example, the flow rate of the polymer may be controlled over a pre-determined
range corresponding to a desirable shear rate range for the polymer under test.
In this way, it is possible to optimise the flow rate to provide a shear rate in
which the best possible distinction in measured characteristics (e.g. viscosity)
is obtained for any given polymer. The skilled person will readily understand how
to determine the optimal flow rate range by carrying out simple tests at different
measured flow rates. The optimal flow rate range for any given polymer will depend
on the characteristic of the polymer which is to be determined.
Apparatus to measure the temperature of the polymer in the conduit is well-known
in the art and may for example be a thermocouple device contained in or placed
on the conduit. Alternatively, the temperature of the conduit in which the polymer
flows can be measured either at or near to the point at which the acoustic sensor
is located, or at another convenient point e.g. at the nozzle outlet of an extruder.
All that is required is that the temperature should be pre-determined at a given
point which is indicative of (i.e. related to or dependent on) the temperature
of the polymer at the point where the acoustic emissions are being detected.
In many cases, the temperature at which a polymer melts will be significantly
above ambient temperature. Typically, temperatures of a polymer melt may exceed
100° C. and may be as high as 125 to 250° C. or higher. The sensor means
used to direct the acoustic waves emitted from the polymer must therefore in many
applications be able to withstand these high temperature.
A typical acoustic sensor means for use in accordance with the invention would
be an accelerometer. Accelerometers are known acoustic sensor devices and are widely
available, for example of the type manufactured by Brüel and Kjær in
Denmark. Where high temperatures need to be withstood by the sensor means, this
should be borne in mind when selecting a suitable device. Accelerometers for example
can be manufactured to withstand temperatures up to and above 250° C. and
the technology to do this is well known to manufacturers of accelerometers.
The conduit in which the polymer flows may take any form. Preferably however
the conduit is a pipe e.g. a cylindrical pipe which may be made of any suitable
material. Steel is typically used in polymer production processes but other corrosion-resistant
materials may be used. The material of the conduit should however be suitable to
allow acoustic waves to be well conducted in order to be detected outside the conduit.
Hence acoustically conductive materials, especially metals such as steel are preferred.
The acoustic sensor means must be placed in acoustical contact with the conduit.
In order to enhance the acoustic emissions from the polymer as it flows, it is
preferable to cause a disturbance in the flow of the polymer in the conduit. For
example, the pipe may be modified in some way to alter the flow characteristics,
especially to cause a sudden change in the flow. Thus, a structural detail may
be provided in the conduit in order that the conditions of flow change, at or near
the portion of the conduit in which acoustic emissions are detected. It has been
found that the presence of a constriction in a pipe is particularly suitable. The
diameter of the constriction is not crucial but it must be sufficiently small relative
to the diameter of the conduit to allow the necessary degree of turbulence to occur.
An orifice plate of the type routinely used for flow measurement is an ideal way
of providing a constriction in a pipe. Other forms of structural detail which may
be used to create turbulence include, but are not limited to, a bend (e.g. 45°
or 90°) in the conduit, the presence of a valve or other choke mechanism.
A sudden increase in pipe diameter may also be suitable.
Where the polymer exits the reactor in molten form (e.g. low density polyethylene)
the conduit may be an exit pipe directly from the reactor, or it may be a by-pass
pipe from one of the main polymer pipelines. Where the polymer is initially in
solid form (e.g. granules or powder) a melting step is needed. The acoustic rheometer
in accordance with the present invention may be used in a similar manner to existing
or known rheometers i.e. it is suitable for use in any form of conduit and therefore
it may simply replace an existing rheometer. For example, existing and known rheometers
such as online rheometers are often situated in a by-pass line from an extruder
or they may be placed on an extruder directly. For example, the conduit in accordance
with the invention may be associated with a single or plural screw extruder.
As mentioned above, flow conditions are also preferably kept at a pre-determined
level in order to allow effective comparison of acoustic emission data with data
from known polymers. Hence, preferably the flow rate of the polymer in the conduit
is measured and/or monitored at or near the point at which the acoustic sensor
is positioned. Flow rates can conveniently be measured by any method known in the
art i.e. by any flowmeter, but it may in some instances be convenient also to measure
flow rates by acoustic means, e.g. by detecting the Doppler shift etc. However,
in order to measure the flow rate in accordance with such apparatus, it will be
noted that a sound-wave (ultrasound >25 kHz) source other than the polymer
flow itself must be present, as this technique depends on detection of ultrasound
waves which are reflected off the flowing fluid.
This differs from the detection method of the present invention which relies
on passively emitted acoustic waves from the polymer itself. However, there is
no reason why any necessary flow rate measurements cannot be taken using a separate
ultrasound sensor means in sender-receiver mode, and utilising this ultrasound
sensor means to pick up the reflected ultrasound for flow rate measurement.
Where the polymer is passing through a pipe or extruder, the pressure in the
extruder or pipe is also preferably measured and/or maintained at a pre-determined level.
As explained above, the invention relies on the principle that movement of the
polymer, for example through a constriction in the conduit, causes the polymer-conduit
assembly to produce vibrational acoustic emissions, which can then be detected.
One preferred way in which the detection takes place is to generate an acoustic
spectrum which typically may take the form of a graphical representation of the
emitted acoustic waves. An example of an acoustic spectrum is shown in FIG.
5.
However, in its simplest form, an acoustic spectrum generated in accordance with
the invention could take the form of a plot of amplitude on the Y axis against
frequency on the X axis called a "power spectrum".
The acoustic spectrum for any given polymer acts as a multivariant "fingerprint"
for that polymer, since it is different from the spectra of other polymers (and
other fluids generally) flowing at the same point in the conduit under the same
flow conditions. Hence, in accordance with the invention, a polymer can be identified
by comparing its acoustic spectrum against acoustic spectra of known polymers until
a match is found. Where the rheological properties of that polymer are also known,
the rheological properties of the polymer under test can also be determined from
a comparison of the acoustic spectra.
If the acoustic spectra are recorded e.g. in electronic form or in any other
form
of searchable database, rapid comparison of data can be carried out e.g. by computer
analysis, and swift matches for the identity of a polymer and/or the rheological
properties of a polymer can be found. The speed of response which can be achieved
using computer processing techniques means that data obtained from the detection
of acoustic emissions can be analysed in a database, and values for rheological
properties or the identity of a polymer can be determined in a matter of seconds,
or even milliseconds. Hence the method of the invention is particularly advantageous
for on-line monitoring of properties of polymers and this can be used to facilitate
process control.
Thus in a preferred aspect, the invention provides a method for the determination
or on-line measurement of the rheological properties of a polymer, comprising:
detection of acoustic emissions from said polymer flowing in a conduit, and
comparison of the acoustic spectrum generated against the acoustic spectra
of polymers of known rheological properties, whereby to determine the rheological
properties of the polymer under test.
The range of acoustic emissions detected may be anywhere in the acoustic frequency
range of 0 to about 25 kHz.
As explained above, the acoustic emissions detected can provide a set of data
which can provide a "fingerprint" of the polymer concerned.
In a simple case, acoustic emission spectra can provide a set of numbers which
is characteristic of the particular polymer produced. This set may be compared
with a corresponding set which is known to relate to acceptable products (e.g.
from previously produced product). By determining whether the numbers are sufficiently
similar (e.g. within previously specified tolerances) it may be determined whether
the fluid is itself acceptable. It will be appreciated that these numbers relate
indirectly, but unambiguously to molecular weight and molecular weight distribution,
although absolute values need never be found for these parameters. Nevertheless,
it may in practice also be useful to do so.
The previously acquired sets of acoustic emission data may have been obtained
by making similar measurements of known polymers having desired characteristics.
For example, sets of data may be obtained for each polymer which it is desired
to produce corresponding to the ideal conditions for producing that polymer.
Close similarity between the measured data and one of these previously acquired
sets of target data may then be used to identify the polymer concerned and/or to
determine whether a desired polymer is being produced with the correct characteristics.
It will be appreciated that this comparison could be performed in numerous ways
and in the simplest case useful information could be obtained even from visual
comparisons of plots of the various data sets. However, these comparisons are preferably
automated. In practice this means that the comparisons are carried out by a computer.
Numerous known computational techniques may be used to perform the analysis,
but it is has been found that multivariate calibration is particularly effective
and accurate (see Martens and Naess 1989 "Multivariate Calibration" published by
John Wiley, Esbensen (1998)). Thus, a latent variable corresponding to an optimal
linear combination of the measured frequency data may be introduced. The data are
then redefined in relation to this latent variable.
In a particularly preferred form of the invention, Principal Component Analysis
(PCA) of the acoustic emissions data is used for classification of new samples
in relation to old samples of known properties. The raw data may be subjected to
preprocessing such as e.g. transformation, centering, smoothing or scaling. Subsequently,
from a set of samples ("calibration set") of known properties a data subspace is
empirically identified into which the test sample data points may be projected.
This subspace is described by a set of "latent variables", spanning individual
axes in the subspace and is denoted the "model" of the given class of samples.
The number of latent variables are then empirically found as those needed to give
representative information related to flow properties of the fluid in question
based on casual knowledge by the operator. It will be noted that it is not necessary
to run any transformation to align with rheological parameters.
If a visual evaluation is desired, a plot of the data may be produced where the
axes are given by the latent variables, and where new samples are compared to the
set of known samples, and to limiting values based on the same samples. For a mathematical
evaluation (classification) upper and lower limiting values may be defined for
the value of the latent variables, and for residuals of the raw data after projecting
into the subspace an upper limiting value is defined. Then new samples may then
be classified according to these limiting values. This approach has been termed
the SIMCA approach, as referred to in Esbensen 1998 and numerous other references herein.
Typically, when using PCA, the latent variables are defined by the eigenvectors
of the (n×k) matrix e.g. where n is the number of samples in the calibration
set and k is the number of values measured for a given variable. Each sample in
the calibration set, and future test process samples, may then be described by
their score values along the individual latent variables thus defined.
By calculating the correlation of the latent variable with polymer property parameters
like MWD, MFR (melt flow rate), etc. one will obtain knowledge of along which direction
these parameters have their largest variability in the latent variable data space.
This information may be compared to the position of the individual samples in the
same data space, to evaluate their score in relation to the different parameters.
By calculating the correlation of the latent variable with processing parameters
like reactor temperature, reactor feed compositions etc., one will obtain knowledge
of along which direction these parameters have their largest variability in the
latent variable data space. This information may be compared to the position of
the individual samples in the same data space, to evaluate their score in relation
to the different parameters, and it may be used to estimate how process parameters
should be changed to change the positioning of the product in the latent variable
space to have the selected flow properties represented by the acoustic emission
data values.
It is particularly preferred for the method to be implemented using a computer
arranged to display a score plot representing the data at least substantially in
real time. In this way, as new data is acquired and new plots are added to the
score plot, changes in the fluid (polymer) characteristics may be followed. It
is helpful for an indication to be provided on the display of where the boundaries
between acceptable and unacceptable points lie, for example based on statistical
quantities. The indications may be a boundary line in the form of an ellipse. Points
falling outside the boundary correspond to unacceptable product.
As discussed above, the score may be evaluated in relation to different parameters
and so it is possible to correlate the position of a point outside the boundary
with the corresponding deficiency in its properties. This information may then
be used to enable appropriate corrective action to be taken by a plant technician.
For example, the previously acquired data sets could include data corresponding
to known incorrect settings for the desired product from which previously determined
corrective action may be taken. Such previous data sets could have been deliberately
produced or they could be learned automatically from analysis of previous operations
of the plant. Alternatively the plant may be adjusted in an iterative manner based
upon the nature of the deviation of the measured data sets from the desired data set.
In particularly preferred forms of the invention, means is provided to automatically
adjust the operating conditions of the plant in order to ameliorate the deficiency.
Of course, there need not be a display for this to be effective—the "ellipse"
may simply be a defined volume of data space.
Another advantage of this form of the invention is that even if a product
is determined to be acceptable, it is possible to monitor variations in where points
are plotted (or located in data space) in order to determine trends which may be
used to anticipate future deficiencies and to take corrective action before they
occur. Preferably this is also implemented automatically.
In this context PCA represents one way of identifying the latent variables. However,
it will be appreciated that any other mathematical method involving linear or non-linear
transformation of the relevant process data into a set of latent variables may
be used. Examples of other methods are Partial Least Squares Regression (PLSR),
Neural Networks (NN) and curve fitting of the pressure data or preprocessed pressure
data to a curve of selected exponential degree.
A particularly preferred aspect of the invention is to use the acoustic emission
data for quantification of selected polymer properties, e.g. MFR or MWD. Again
the raw data may be subject to preprocessing such as e.g. transformation, centering,
smoothing or scaling. From a set of samples ("calibration set") of known properties
it is then possible empirically to identify a mathematical relation (the "model")
to quantify the selected properties based on the preprocessed pressure. This model
may be any linear or non-linear relation defined by methods like Principal Component
Regression (PCR), Partial Least Squares Regression (PLSR), Neural Networks (NN), etc.
When using PCR and PLS, latent variables may be identified in a modified form
closely related with PCA (above), and then a linear regression model is developed
between the polymer property and this type of latent variable. In the same way
as when doing classification above, the score values in the latent variable space
may then be used for visual and mathematical evaluation. Correlation between the
latent variables and process parameters may be used to identify how the process
parameters should be changed to adjust the selected property of the polymer being produced.
It will be appreciated from the foregoing that the present invention is useful
in the field of polymer production and so acoustic emission detection means is
preferably situated on-line and may be associated with an extruder used in such
a context. Polymer may be fed from the extruder directly into a suitably modified
conduit for acoustic emissions to be detected, e.g. by means of a bypass. Because
of the speed of operation and the improved accuracy of the method of the invention,
if the properties of the polymer are as desired, this will be known much more speedily
than in the prior art system. Furthermore, it is also possible to determine more
quickly if the measured characteristics are not as required and then to adjust
the operating conditions of the reactor accordingly in order to obtain the desired
characteristics. Consequently, wasted production may be greatly reduced.
It is possible to apply the method of the present invention either only when
the
reactor is first set up for a given production run, or at occasional intervals
as required by quality control. However, since the method may operate automatically
it is particularly preferred that regular and comparatively frequent measurements
be made, say around every 10 minutes.
Polymer producing plants are normally operated continuously and if it is
desired to change from production of one polymer to another this is done without
closing down the plant. Instead, the reactor operating conditions are adjusted
in order to change the polymer thereby produced and fed to the extruder. Thus,
preferably the method of the invention is used to obtain data which is used to
monitor the transition between products. Since in the preferred forms of the invention
the data acquisition and subsequent comparison steps are carried out by computer,
this may be done rapidly. Consequently, the transition may be effected more smoothly
and quickly than in the prior art and moreover the operator can determine more
quickly when the desired product starts to be produced. It will be appreciated
that this significantly reduces the amount of wastage associated with operation
of the reactor therefore a significant advantage in terms of saving time and materials
and thereby costs.
The invention also provides an apparatus, also referred to herein as an acoustic
rheometer, for carrying out the method of the invention, and the use of the acoustic
rheometer to control a polymerisation reaction. Thus viewed from a further aspect
the invention provides an apparatus for characterising a polymer, comprising:
a) an acoustic sensor capable of detecting acoustic emissions from the polymer
and thereby generating a signal;
b) means for comparing the signal against acoustic emissions data from polymers
of known characteristics. This data may for example be stored in a computer memory
either provided within the apparatus or remotely.
The invention also provides the use of an acoustic rheometer comprising
a) an acoustic sensor capable of detecting acoustic emissions from a polymer;
for controlling a polymerisation reaction producing said polymer. Preferably,
in this aspect, the acoustic rheometer further comprises means for comparing the
signal against acoustic emissions data from polymers of known characteristics,
as defined above.
Preferably, the apparatus is adapted for determining the rheological
properties of a polymer, comprising:
a) an acoustic sensor capable of detecting acoustic emissions from the polymer
and thereby generating a signal;
b) means for comparing the signal against acoustic emissions data from polymers
of known rheological properties whereby to determine a value for the desired rheological
property of the polymer under test.
The apparatus may further comprise means for identifying the polymer. Preferably,
the apparatus comprises an acoustic sensor which is capable of detecting vibrational
acoustic emissions in the interval 0-25 kHz.
The acoustic sensor may be as described above. The means for comparing the signal
(referred to hereinafter as "comparison means b)") may if necessary or desired
comprise means for amplifying or processing the signal from the acoustic sensor.
For example the comparison means b) may be a computer which in turn may be connected
e.g. to a signal amplifier or preprocessor. The computer will preferably be loaded
with suitable software. Conveniently, the comparison means b) may be provided by
a package such as the Multi-Channel Spectrum Analyser available from Applied Chemometrics
Research Group (ACRG), Tel-Tek, Porsgrunn, Norway.
The apparatus of the invention is set up such that the acoustic sensor means
is positioned in acoustic contact with (preferably touching) the conduit through
which a polymer can flow. The conduit is preferably a straight segment of a pipe
and preferably this has a structural detail as hereinbefore described. The acoustic
sensor means is therefore positioned whereby to detect acoustic emissions from
the flowing polymer as it passes through the structural detail in the pipe.
It has been found in particular that the acoustic sensor means can be placed
in
a variety of positions in relation to the conduit in order to successfully determined
rheological properties of a polymer. For example, it could be placed before or
after the structural detail e.g. within about 5-20 cm or 5-10 cm of the structural
detail (relative to the direction of flow) but preferably it should be placed before
the structural detail. Alternatively it could be positioned at the position of
the structural detail itself, which is particularly preferred.
The invention also extends to a polymer production plant incorporating the method
or apparatus of the invention as set forth above and also to polymer products thereby produced.
Certain embodiments of the invention will now be described, by way of example
only and with reference to the accompanying drawings in which:
FIG. 1 is an acoustic rheometer according to the invention;
FIG. 2 is a schematic flow diagram showing the data path for analysis of the
acoustic emissions data;
FIG. 3 is a diagram of one possible configuration of an acoustic rheometer according
to the invention;
FIG. 4 is another diagram showing a different possible configuration of an acoustic
rheometer according to the invention;
FIG. 5 shows an acoustic spectrum in the frequency range 0-25 kHz, for each
of the four different polymers as described in Example 1.
FIG. 6 shows a score plot acoustic spectrum for PCA (principal component analysis)
of these four polymers. The percentage score of Component 2 (22.3%) is plotted
against the percentage score of Component 1 (35.7%).
FIG. 7 shows the PLS model (partial least squares), with the viscosity of the
modelled values plotted against the measured viscosity for each of the polymers.
FIG. 8 shows the variation in viscosity of the four different polymers as measured
by a rheometric dynamic analyser in a frequency sweep mode (190° C. melt temperature).
On the Y axis the crossplot viscosity is given at 300 rad/sec and on the X axis
it is given at 0.05 rad/sec.
FIG. 9 shows a score plot (t1t2) of five replicates of each of
the polymers designated A, B, C and D.
FIG. 1 shows an ultrasound rheometer for operation in accordance with the invention.
Typically, the polymer melt leaving the polymerisation reactor will be processed
through an apparatus
1 which consists of a conduit
2 with a constriction
3 allowing the polymer to pass through. The acoustic sensor means 4 e.g.
an accelerometer may be placed in any one or more of positions A (before the constriction),
B (at the constriction) or C (after the constriction). The accelerometer
4
detects acoustic emissions from the polymer flowing through the apparatus
1
and generates an acoustic spectrum which is characteristic of the polymer. The
signal is amplified by an amplifier/preprocessor
5 and data analysis is
carried out by a computer
6 or other suitable means. Data analysis can for
example be carried out by mulitvariate analysis techniques such as principle component
analysis (PCA) or partial least squares (PLS). Information on the viscosity, molecular
structure, MFI and other polymer properties can then be calculated by comparison
with information from known polymers.
FIG. 2 is a schematic flow diagram showing the data path from the acoustic emissions
(vibrations) generated by the polymer and how that data is analysed numerically.
Box
7 represents polymer flow through the constriction in the conduit, from
which the acoustic signal is detected by the sensor accelerometer
8. Box
9 represents signal processing by adaptation of the signal through a lowpass
filter and analog-digital conversion to allow analysis of the signal. Multivariate
analysis of the signal data is then carried out, as represented in box
10.
FIG. 3 shows one possible configuration for the acoustic rheometer mounted on
a by-pass from an extruder. The extruder barrel
11 is shown with the by-pass
line
12 leading from it and round the by-pass "loop" back into the extruder
11. Polymer is pumped round the by-pass pipe
12 by means of a gear
pump
13, piston, or any other suitable device for generating flow, and through
a constriction
14 in the by-pass pipe. The acoustic sensor means
15
is placed outside the pipe in acoustical contact therewith, and leads
16
transmit the signal to an amplifier
17 and then to a personal computer
18
which is capable of analysing the data by means of multivariate analysis (MVA).
FIG. 4 shows another possible configuration for an acoustic rheometer. The polymer
process flow from the polymerisation reactor is in powder form and is transported
through pipe
19 from the reactor. A portion of the polymer is drawn off
from the main flow pipe at a sampling point/system
20 and passed through
a single screw extruder
21 or any other suitable device where it is heated
and melted to allow it to flow. An acoustic sensor means
22 is placed in
acoustical contact with the single screw extruder pipe at a point before where
the polymer flows through a constriction
23. The signal detected by the
acoustic sensor means
22 is transmitted to a data analysis unit such as
a computer (not shown).
EXAMPLE 1
Comparison of 4 Different HDPE Resins using Acoustic Rheometer
The aim with this study was to compare the acoustic spectrum recorded as described
in the patent with viscosity data obtained using a conventional rheometer (plate—plate
dynamic rheometer; Rheometrics dynamic spectrometer, RDA-II) For this purpose 4
commercial HDPE (high density polyethylene) polymers manufactured by Borealis were chosen:
- HE8168, HE8343, LE7520, LE0400
Viscosity data obtained by the dynamic rheometer (at 190° C.) is shown
in table 1 below:
| TABLE 1 |
| Viscosity vs. shear rate at 190° C. |
| |
viscosity (Pa · s) at different shear rates |
|
| |
polymer |
|
|
|
| shear rate |
LE7520 |
LE400 |
HE8343 |
HE8168 |
| 25 |
511 |
1986 |
3793 |
860 |
| 38 |
419 |
1516 |
2973 |
857 |
| 50 |
368 |
1269 |
2534 |
855 |
| 74 |
306 |
985 |
2017 |
840 |
The experimental setup for the rheometer is shown in FIGS. 1 and 2;
The rheometer is basically a heated pipe in which a die is inserted in order
to create a constriction in the pipe. At the flow inlet of the die is placed an
accelerometer in order to record sound generated by the flowing polymer. Polymer
is being fed by a 30 mm extruder (manufactured by company Collin GmbH)
The procedure of collecting and numerically treating the data is shown schematically
in FIG. 2. (for further explanation refer to Esbensen et al (1998); "Acoustic
chemometrics; from Noise to information", Chemometrics and and intelligent laboratory
systems 44(1998) 61-76.
Each polymer was extruded at 4 rpm's (30, 45, 60, 90). With the die chosen for
the experiments (7 mm diameter) these rates equal shear rates as shown in table
1. In table 1 (above) viscosities for the 4 polymers at the given shear rates are
given based on laboratory measurements. During the experiment with the acoustic
rheometer, the following data were recorded:
- polymer temperature (end of extruder)
- polymer pressure
- Polymer temperature at the measurement point
- acoustic spectrum (FIG. 5)
MVA Analysis
A plot of the different acoustic spectra is shown in FIG. 5 A PCA (principal
component
analysis) analysis is shown in FIG. 6: the scores of the first two latent
variables show that the spectra are able to distinguish between the different polymers
in a systematic manner.
By combining table 1 and the recorded spectra we can use PLS regression technique
to study how the acoustic spectra explain the variation on viscosities at the actual
shear rates.
The PLS model (FIG. 7) explains the measured viscosities by 94% in 2 comp\96%
in 3 com. Cross validation reduces the explained variance to around 60%.
From this it can be concluded that the recorded spectra at a given flowrate
relate to one point on the dynamic spectra curve. As done in this experiment, running
at four different flow rates (or using 4 different dies) one can put the spectra
together to characterize the flowcurve of the polymer.
EXAMPLE 2
Study Lot Variation Within a Single Product by Use of Acoustic Rheometer
It is well known that any commercial polymerisation process will be subject to
certain variations in the molecular structure of the polymer produced. The amount
of variation is usually low and in some cases difficult to quantify. Online methods
are used to measure this variation in properties. The accuracy of the online rheometer
will determine how well small variations can be detected and thus in the long run
avoided. To test the acoustic rheometer of the invention, 4 different lots of a
polymer grades with known difference in molecular structure were tested using the
same setup as in example 1.
FIG. 8 shows the variation in viscosity of the 4 lots as measured by means of
a rheometrics dynamic analyser in a frequency sweep mode (190° C. melt temperature).
Each sample was extruded at 4 rates (30,50,70.100 rpm on the 30 mm extruder).
Spectra in the range 0-25 KHz were recorded on a Bruel & Kjaer acclerometer (nr 4384)
| TABLE 2 |
| Part-list (high temperature equipment) |
| |
| 1 Accelerometer 250° C. |
Brüel & Kjær, |
Number: 4384 |
| |
Denmark |
| 2 Coaxcables, 2 mm, |
Brüel & Kjær, |
Number: AO 0038 |
| 250° C. |
Denmark |
| 1 Charge/Deltatron |
Brüel & Kjær, |
Number: 2646 |
| ampl. |
Denmark |
| 1 UNF to BNC adapter |
Brüel & Kjær, |
Number: JP 0145 |
| |
Denmark |
| 25 Cement studs |
Brüel & Kjær, |
Number: UA 0866 |
| |
Denmark |
| 25 Extension |
Brüel & Kjær, |
Number: UA 0186 |
| connectors |
Denmark |
| TABLE 3 |
| Recording Parameters: |
| |
| |
Sampling frequency: |
50 kHz |
| |
Frequency range: |
0-25 kHz |
| |
Number of variables: |
512 fewq. + 1 Temp. = 513 |
| |
|
variables |
| |
window size: |
1024 data-points |
| |
Transformation window |
Blackman Harris |
| |
type: |
| |
Number of replicates: |
5 |
| |
Recording length each |
0.02 sec. |
| |
replicate: |
| |
Averages each replicate |
100 |
| |
spectrum unit: |
dBV rms |
| |
Data were pretreated as shown in FIG. 2 example 1. FIG. 9 shows a score plot
of results: for each resin five replicates were run.
The data show the method to be able to separate between individual lots from
a commercial polymerisation.
*
