In-line monitoring with NDT flaw detection methods is not applied at this time. Only one big company in Germany uses in-line visual testing - a strong light source is placed inside the tube and shadows are recorded by CCD cameras mounted outside. Unfortunately, there are doubts as to whether this addresses the need for quantitative test results. In practice, some "NDT methods" are quite often applied for more or less marketing reasons. No standardization is currently available, nor is there a lobby for the establishment of such; in fact, it seems that any argument for not advancing the cause of standardization is welcome.
Flaw detection is normally being effected with methods that can cause damage to the product and is carried out every few hours, according to German homegeneity test standard R 14.3.1 released many years ago from the association "Gütegemeinschaft Kunststoffrohr e.V.", which claims that pipes exceeding the flaw size of 0.02 mm² are rejected! This statistical evaluation provides certain information on the quality, but 100 % detection is not feasible. Furthermore, recognition of flaws is somewhat coincidental, so correction of process parameters can be effected only alter a considerable delay in time. This results in the following risks for the pipe manufacturer:
For these reasons 'some' companies in the extrusion industry had the wish to detect flaws continuously and over the whole circumference of the pipe, to display them on a screen in order to be able to control the process without delay and to either mark flaws or to eliminate them via a signal transmitted to the cutting device. This prevents delivery of imperfect pipes in any case. A continuously working flaw detection system provides the following features:
Let's not forget an accident that occurred in Germany's plastic pipe market, and this just involved water pressure pipes and thankfully, resulted in no human injuries. Today, that company states, "If we were beginning gas pipe production, no doubt we would install a system if one were available and sleep well at night". They remember very well how many years it took for their customers to regain confidence in their production quality.
While in the metal industry ultrasonic flaw detection has been widely used for the past 40 years, high investment costs prevented the use of this technology in the plastic pipe extrusion process. This argument is no longer valid because systems with considerably reduced investment costs have been developed. We know that 6 years ago the first economic rotation head was delivered to a gas pipe plant in the UK. This was a prototype, however, and still needed to be improved, but no further action was taken. Only a year and a half ago, two German suppliers specializing in UT for plastic pipes submitted their first system to selected manufacturers, including the largest manufacturer of PE pipes worldwide, for field testing. One of the two systems showed very good results [4]. The technology was rejected for purely political reasons. During the last year, economical solutions have been developed in the steel pipe market that could also be applied to this market [6].
The next paragraphs describe this application in detail and give an outlook on what kind of system designs could be available.
What kind of flaws are relevant to plastic pipe production?
Methods for flaw detection require a precise definition of what the term
"flaw" means.
Unfortunately, production sites very often speak different languages
on this matter.
Wall thickness measurement itself could be taken as a kind of flaw
detection, as when tolerances
are exceeded or fall below the minimum; this is a kind of control, too.
However, this is not what is meant by the "classical" use of the term
"flaw detection," and wall thickness measurement is
already a standard for in-line gas pipe testing (it saves raw material
costs!). We will not concentrate on this here, but refer you to the
article, "Ultrasonic for Plastic Pipe Extrusions
(by R. Diederichs)"[1] that describes the state of the art and outlines
some basic problems
of ultrasonics applied to plastics.
The statistical/spiral wall thickness in-line measurement should be a
precondition that is applied with a separate system. This function
could also
be integrated into the flaw detection system as it is established.
Wall thickness changes occur suddenly, in a small section, and
their detection can easily be grouped under "flaw detection."
These flaw types are principally depicted in the figure above with no scale. Volume flaws are inclusions of air bubbles or foreign substances that occur during the cooling phase of the pipe or during the extrusion process. Also the raw material itself can have volume flaws, especially if the company recycles its own production wastes, as it is allowed to do. Recycled materials are not allowed for the production of gas pipes. Volume flaws must be detectable near the surface or in front of the back wall of the pipe. Another flaw type is the inclusion of foreign bodies in the finished product. Foreign bodies can also be material deposits in the pipe head which are irregularly included into the melt flow without being melted, leading to flaws in the pipe. Other reasons for surface flaws may be wrong or worn-out calibration units.
What size flaws should be counted?
Nobody can really answer this.
Some would like to meet the above-mentioned standard from the
"Gütegemeinschaft" with its 0.02 mm²! This would result in a
circle diameter of 0.16 mm. There is doubt that this would be
practical; possibly no pipe could be produced anymore! Why was this
specification never tested? It could stop a possible design
in its first stage.
About 0.5 - 1.0 mm should be considered a practical size; this
should also be sufficient to minimize risk of possible mechanical
damage to the pipe.
What test speed is necessary, relative to pipe production speed?
The production speed for plastic pipes is the precondition that must be
matched, and this speed varies with the size of the pipe. This also
varies between production sites, and the different capacities of the
machines in use. Modern machines can produce about 1000 kg/h but the
length of the
cooling bath also determines the possible production speed.
If we consider the worst case, for the fastest necessary speed, we choose the smallest possible sound field of about 5mm - for this a rotation speed of 3000 rpm would be necessary for just 1 channel to operate. The mechanical engineering site would prefer an economically rotary head design of only 600 rpm maximum, and a pipe diameter of up to 250mm; for bigger pipes the rotation speed can and must decrease. We could consider using a fixed transducer array for the smallest size, but that would not be economical for a size of say, 110 mm, so we leave this for other pipe markets, e.g., medical, automotive, to figure out.
Advances in transducer design offer a good chance for finding a design with a sound field of about 10 - 15mm, so the simplest arrangement could operate with two transducers or even one. At this stage we presume that a rotary speed of 600 rpm and 2 channels would be sufficient for the task.
All of these issues are actually the basic disadvantages of UT in terms of capability of determining the real size of a flaw, or specifying an exact rejection threshold. This raises the question of why equipment manufacturers compete to increase the number of digits of dB after the decimal point. Returning to our task; obviously, for the most accurate test results, we should consider at least one more test axis with another transducer applied in angle This transducer will improve the reliability of the test, regarding: oblique flaw, surface flaw, close under surface flaw, close to backwall flaw.
| The straight transducer still provides some advantages due to its displayed interface and backwall echo signals. A problem in UT is that natural flaws sometimes have no reflection echo. To compensate for this, a separate evaluation called "anticoincidence" is available; the reject level is adjusted at the back wall echo. (see illustration) . This also facilitates coupling control. Furthermore the straight transducer will allow us to check some surface displacements or to operate a TGC/DAC curve. An integrated wall thickness test could be applied, too. |  |
The angle applied transducer needs a calculation of incidence angle value [5], and a further determination if applied longitudinally or transversally to the pipe axis. Because certain orientation of inclusion is not yet predictable for plastic pipes, the suggested 45 degrees is a compromise. The transducer axis applied through water is calculated to be 22 degrees according to our equation (see fig.2b). In pipe extrusion, continuous longitudinal surface grooves are often generated from the die; such grooves are permitted within limits. If an angle transducer tested transversally we would receive a large echo from this groove. For this reason a longitudinal test axis is essential. Otherwise we wind up in the well-known discussion of conforming ultrasonics to standards based on other measurements.
In order to minimize the rotation speed or number of channels we will
select a large transducer. The result on test flaws, e.g., flat
bottom holes shows the
maximum size that should be chosen [1]. Increasing the gain to the maximum
of which the equipment is capable is not recommended. Especially for
in-line testing, we will need a high rejection reliability and
this under strong production electrical interferences. Also 100 meter
coils
are produced and should not be cut because of a failed rejection
caused by
interferences.
With a large crystal size near to surface resolution
capability will also be reduced.
On consideration of all the restrictions mentioned, a good mix will
consist of a
5 MHz transducer containing a good piezoceramic, balanced damping and a
rectangular crystal of about 20x8 mm in size. The 5 MHz frequency is
suitable up to a thickness of approx. 24 mm;, above that a 2 MHz
or 1 MHz transducer should be considered (sound attenuation in plastic
is high! ) Graph 
Ref. [1] When using large crystals, a rectangular shape is better for
testing on pipes. This transducer will show an effective 16 mm sound
field (-6dB) in a longitudinal pipe axis. The mentioned article [1] is recommended reading. It describes the
state of the art and some
basic problems of ultrasonics as applied to plastics.
The place to be tested should be as close as possible to the end of
the cooling line [3]. This is necessary because of temperature influences. Graph 
Ref. [2]
Sound velocity calibration must still be watched carefully, to ensure
that the
transducer's flaw gates are operating at the predetermined positions.
| The recommended test procedure For pipe dimensions: Wall thickness range: 1.9 - 24.0 mm Diameter range : 20 - 250 mm | |
|
1. Flaw detection channel with vertical sound transmission, shown in
Fig. 2a
* Flaw size 0.5mm, as substitute flat bottom hole 0.5 mm
2. Flaw detection channel with angular sound transmission,
* Flaws on inner surface
|
Fig.2 a,b Transducer arangement
|
The system consists of the following units:
Some important issues are:
Electric interference (EMV) must be prevented by an ultrasonic transmitter/preamplifier mounted directly on the mechanism, and by further transmission to the evaluation cabinet by double-screened coaxial cables.
Reliable measuring mechanics Compared with the rotary heads used in the metal industry, the application in the extrusion process requires some considerable mechanical modifications. The most important feature is the reliable coupling of the ultrasonic signal to the plastic pipe via a coupling medium as the penetration of thermoplastic materials requires more energy than that needed for steel. The system has to be reasonably priced and designed for a long service life.
The rotary head should work with min. 600 r.p.m. and with one or more transducers, depending on the task. In immersion technique, the transducers operate inside the rotating water chamber. They should assemble easily in a holding device and work in a fixed scanning pattern, depending on the pipe diameter. The transducer signals are transmitted via full silver slip rings, each with a separate ring for grounding.
The device must guide the pipe centrically through the rotary head along the inlet/outlet guide and sealing rings. A special guidance facility is necessary for a diameter over 250 mm, however. The rotary head must consist of takeover controls for functions such as operation, safety circuit and angle recording. Reliable flaw detection requires perfect water quality. The head must receive a water supply that is free of air bubbles and floating particles; a water circulation system with a 25 micromm filter is necessary.
Adjustable sensitivity . In order to adjust ultrasonic systems for flaw detection to the requirements of various products, the supervisor must be able to adjust the most important parameters such as amplification, attenuation, amplitude, and flaw size levels. It is equally important that the system be equipped with an echo display in order to facilitate adjustment of these parameters. For the machine operator these settings must be stored parameters which means they are reproducible under the pipe denomination and file name.
Flaw results - Alarm/Visualization . Flaws should be visualized in various ways. It is essential to select a certain pipe section where measuring values are continuously actualized. When a flaw is recognized, an alarm should be triggered; the alarm output can be either visual or displayed as an impulse from the cutting unit involved, a marking consistent with the path delay. The output must include the flaw type as well as the flaw position at the pipe circumference. Furthermore, in order to evaluate flaws, zooms and 2D-displays are required.
A complete real time evaluation should consist of: Graphic output of the flaws in lengthwise direction with selectable pipe section, continuous renewal and meter counter/coil number. According to the task, the system should be presettable for logical flaw types, e. g., surface flaws, volume flaws, and detection channel number . The flaw size level is displayed in a 2 color graphic with digital information on flaws such as "serial number", "flaw number", and "pipe length". "Flaw size" (amplitude) and "flaw depth" is displayed on the monitor, transferred to a data file and can be hardcopied by a printer. Results are displayed in real time for post-processing. At the end of the test, the graphic should show a two-dimensional (2D) image that can be scrolled over the production length, and printed out in list format, either on-screen or on a printer. The detection result (file) should be available on a floppy disk or via network for OFF-LINE evaluation.
The described concept could be a valuable start at the production site. It should be considered that investigations are still necessary to establish this method and develop future standards. The picture shows a production line with the equipment and a flow chart. It is essential that the laboratory support the initial phase and follow through with the certain establishment of a organization for this new NDT. This is necessary work that should be carried out. The main elements are:
|
Verification - Comparison - Evaluation - Statistics
During the initial phase intensive production support takes place. Personnel are trained and new techniques are learned. The picture (a,b,c) describes steps for verification of detected flaws. | |
|
A. US manual testing
B. Semi automated testing
C. Microtom |  |
To establish UT for such an important task is not comparable to a normal order for a new machine. Ultrasonic flaw detection begins in a new field with a learning process that will let us discover things about our pipes we didn't know before. The illustration also shows us some accessories that will be necessary for carrying out this procedure and new for plastic pipe manufacturers. Plastics manufacturers are not yet used to NDT and further personnel training will be needed. We must have the confidence to do this.
Today's expanding technology should not only result in increased production and profit, but also a heightened awareness of its relevance to safety considerations. If there are any further questions the author will be glad to help.
Author