Вы находитесь на странице: 1из 12

Sensors and Actuators A 182 (2012) 1627

Contents lists available at SciVerse ScienceDirect

Sensors and Actuators A: Physical


journal homepage: www.elsevier.com/locate/sna

A review and evaluation of melt temperature sensors for polymer extrusion


Chamil Abeykoon a, , Peter J. Martin a , Adrian L. Kelly b , Elaine C. Brown b
a
School of Mechanical and Aerospace Engineering, Queens University Belfast, Belfast, BT9 5AH, UK
b
IRC in Polymer Science and Technology, School of Engineering, Design and Technology, University of Bradford, Bradford, BD7 1DP, UK

a r t i c l e i n f o a b s t r a c t

Article history: Melt temperature is one of the key variables in polymer extrusion which determines process thermal
Received 24 December 2011 stability and hence melt quality. Therefore, melt temperature is commonly measured in polymer pro-
Received in revised form 19 April 2012 cessing and point/bulk melt temperature measurement methods are widely used in the present industry.
Accepted 20 April 2012
Some thermal prole measurement methods have also been attempted in research. This study presents a
Available online 27 April 2012
review of melt temperature measurements in polymer extrusion in research and industry, and describes
the results of an experimental evaluation carried out to explore the performance of ve melt tempera-
Keywords:
ture measurement techniques. In addition, an investigation was carried out on the fully developed melt
Polymer extrusion
Process monitoring
temperature prole of a cylindrical rod die. Moreover, the existing challenges and possible future appli-
Melt temperature sensor cations for extrusion thermal monitoring are discussed. The results conrmed that melt temperature
Melt temperature prole varied signicantly at different radial locations within the die. Point/bulk melt temperature measure-
Frequency analysis ments were found to provide relatively limited information on process thermal quality close to the die
wall. Hence, information provided by these sensors is less representative of actual thermal conditions as
they are unable to capture thermal information from the whole melt ow cross-section. Therefore, the
importance of the development of industrially compatible thermal prole measurement techniques is
emphasised.
2012 Elsevier B.V. All rights reserved.

1. Introduction 1.1. Polymer extrusion

Polymers are among the most important materials available to Different types of polymer processing extruders are available in
manufacturing industry. Many conventional raw materials such as industry: single/multi screw extruders, disk/drum extruders, etc.
steel, glass and wood are being replaced by various types of polymer Of these extruders, single screw continuous extruders are the most
materials or polymer composites which perform the same function commonly used in the plastics industry [1]. The basic components
while offering a number of other advantages. Some of the attractive of a single screw extruder are shown in Fig. 1. The screw is the
properties of polymeric materials are ease of forming into complex key component of extrusion machines and can be divided into
shapes; high strength to weight ratio; high temperature resistance; three main functional/geometrical zones (i.e. feed or solids con-
high chemical resistance; high clarity; re-processability; low cost; veying, compression or melting, and metering or melt conveying)
etc. Extrusion is the most fundamental method in polymeric mate-
rial processing and it is involved in the production of commodities
in diverse industrial sectors such as packaging; household; auto-
motive; aerospace; marine; construction; electrical and electronic;
and medical applications. However, effective thermal monitoring
and control still remains an issue despite the signicant develop-
ments to the process that have been introduced over the last few
decades.

Corresponding author. Tel.: +44 2890974236.


E-mail address: yabeykoon01@qub.ac.uk (C. Abeykoon). Fig. 1. The basic components of a single screw extruder.

0924-4247/$ see front matter 2012 Elsevier B.V. All rights reserved.
http://dx.doi.org/10.1016/j.sna.2012.04.026
C. Abeykoon et al. / Sensors and Actuators A 182 (2012) 1627 17

in the case of simple, single ighted screw geometries based on its [14] developed a non-contact IR pyrometer for temperature moni-
primary operations. The feedstock material fed into the machine toring in a microbre-extrusion process and achieved satisfactory
through the hopper is conveyed along the screw while absorbing results. Steward [6] used a hand held IR pyrometer to measure
heat provided by the barrel heaters and through process mechan- extrusion melt temperature and argued that the pyrometer elim-
ical work. Eventually, a molten ow of material is forced into the inates the shear heating effect and gives a more reasonable melt
die which forms the material into the desired shape. More details temperature value, but does not capture temperature variations.
of the mechanisms of polymer extrusion can be found in literature Christie [15] used a non-contact IR thermal imager to measure
[2]. the extrudate temperature in an extrusion coating process. A spe-
Process monitoring is crucial to maintain product quality in cial device was developed to scan and measure temperature eld
polymer extrusion. Among the process variables, melt temperature across the melt curtain. No conclusions were drawn relating to pro-
is one of the most important variables which have to be main- cess melting stability using this technique. Another study [16] used
tained very carefully. It is essential to ensure that the polymer is not a thermal imaging system for micromoulding process monitoring
degraded or overheated during processing [3]. The nal melt output and improvements to the device are underway to measure polymer
from the machine should be homogeneous in colour, composition melt surface temperatures.
and temperature to produce a good quality product. Therefore, melt Noriega et al. [17,18] proposed a non-invasive measuring tech-
temperature is one of the most commonly measured parameters in nique to capture experimental data and images inside the extruder
polymer extrusion and several measurement methods are available through a small quartz window and a rigid boroscope with a
in the present industry/research. Of these, some methods can only fast response time. A 45 mm diameter single screw extruder was
provide point/bulk melt temperature measurements while other used for the experiments and the authors claimed that the cap-
methods can offer thermal prole measurements. tured images allowed study of the melting, material conveying
The aim of the rst part of this study was to carry out a review and behaviour of the solid bed and melt pool including solid bed
of previously attempted work on extrusion melt temperature mea- breakup information. They observed that the formation of the melt
surements. In the second part, an experimental evaluation of ve pool was delayed for a few turns as melting progresses due to the
melt temperature sensors was carried out to explore their perfor- fact that before a melt pool forms, the melt lls the gaps between
mance in presenting actual process thermal stability. A frequency the granules. Modications to the widely accepted Tadmor melt-
analysis of each sensor signal was carried out to identify the crit- ing model [19] were proposed based on this observed delay and
ical frequencies of thermal uctuations. Also the fully developed good agreement was found between the modied model and exper-
melt temperature proles were investigated in an adaptor prior to imental results. The authors claimed that this was a good method
entrance into a rod die at different screw rotation speeds. Finally, to monitor process thermal behaviour without disturbing the melt
the existing problems and challenges in polymer extrusion ther- ow.
mal monitoring are discussed. A single screw extruder was used in Other work has been reported on ultrasonic melt tempera-
the experiments as it is the most commonly used type in industrial ture measurements in extrusion [2022]. Wen et al. [23] used an
polymer extrusion. ultrasonic sensor to measure the surface temperature of extruded
polymer and found that the sensor had a faster response time than
conventional wall mounted thermocouples. Muller et al. [24] pro-
2. Review of previous attempts to monitor melt posed a new ultrasonic probe for in-line monitoring of polymer
temperature melts and measured ultrasonic attenuation spectra experimen-
tally with different types of polymer processes. Another study
2.1. Point/bulk melt temperature measurements [25,26] used ultrasonic techniques to measure real-time melt tem-
perature in extrusion and claimed that the ultrasonic technique
An early study made by Maddock [4] used wall mounted ther- provided a sensitive indication of the melting process and material
mocouples to monitor extrusion melt temperature under different variability. Kelly et al. [11,13] made ultrasonic melt temperature
processing conditions and the product quality was dened in terms measurements and compared them to an IR temperature sensor
of melt temperature and pressure uniformity. Later, a number of measurement and an average thermocouple mesh temperature
studies were carried out to observe the extrusion melt temper- where good agreement was achieved.
ature by using wall mounted thermocouples [3,5,6]. In most of Bur et al. [27] used a uorescence technique to measure the bulk
these studies, melt temperature measurements were taken from melt temperature of a 19 mm diameter single screw extruder. A u-
the thermocouples xed at the end of the extruder barrel or at the orescence sensor, an IR sensor and a wall mounted thermocouple
die. were placed on the extruder barrel wall in the same plane near the
Infrared (IR) sensors are a less commonly used melt temperature screw tip. Temperatures from the three sensors were recorded by
measurement method in the present industry. A few IR temper- applying screw speed step changes and it was claimed that mea-
ature measurement devices are commercially available such as surements made by the uorescence and IR sensors were in good
wall mounted IR probes (which make measurements in-contact agreement with a difference of less than 3 C over the measure-
with the polymer melt) and non-contact temperature measuring ment period. However, the thermocouple showed a large lag from
devices such as hand held IR pyrometers and IR thermal cameras. the measured IR and uorescence measurements.
The principles of operation of all of these sensors are based on Recently, an attempt was made to use a nuclear magnetic reso-
the emitted IR energy from the polymer melt but nal temper- nance (NMR) technique to monitor the extrusion melt temperature
ature is derived using different concepts and the output may be [28]. The authors argued that this technique can be used to deter-
displayed in different forms (e.g. a value, a thermal image/map). mine the melt ow composition and homogeneity in addition to
Several attempts have been made to monitor extrusion bulk melt the melt temperature. However, signicant research and develop-
temperature from IR wall mounted probes [3,713]. It has been ment will be required before applying this technique in industrial
found that IR sensors can detect rapid variations (response time process measurements.
is around 10 ms) in melt temperature better than the typical wall Work by the present authors [29,30] reported an attempt to
mounted thermocouples. However, IR signal penetration depth is predict process thermal stability inferentially. The correlations
affected by material properties such as melt emissivity and these between screw load torque, melt pressure and melt temperature
affect the sensor performance [8,12,13]. Bendada and Lamontagne uctuations were examined by analysing experimentally measured
18 C. Abeykoon et al. / Sensors and Actuators A 182 (2012) 1627

signals. However, no strong correlations between these signals and viscous dissipation were observed on the temperature prole.
could be observed. It was found that as the screw load torque sig- The authors mentioned that high screw speeds were not achieved
nal is dominated by the solids conveying torque it is not sensitive due to system limitations and hence further developments to the
enough to identify unstable melting issues. Pressure uctuations measurement system were required before use in industrial appli-
had slight correlations with melt temperature uctuations partic- cations.
ularly at low screw speeds. However, none of these signals showed Rasid and Wood [38] used a thermocouple mesh in-between
sufciently good performance for them to be used as a powerful the screw tip and the die to measure the melt ow temperature
tool of monitoring the process thermal stability inferentially. prole with a low density polyethylene. Tests were carried out to
Obviously, the majority of commercially available melt tem- investigate the effects of individual process settings on the melt
perature sensors for polymer extrusion are based on point/bulk temperature and found that of the barrel temperature settings, the
measurements and hence are the most common in the present metering zone temperature had the highest inuence on the melt
industry. Of these techniques, wall mounted thermocouples (i.e. temperature whereas the melting zone and die temperatures had
based on the thermoelectric effect) are the most commonly used. negligible effects. However, they did not observe a symmetrical
However, typical wall mounted thermocouples are highly affected temperature prole around the centre of the melt ow, which was
by the barrel metal wall temperature and also they are slow in later attributed to an alignment problem of the apparatus.
response time (typically around 1 s or greater) [2,3,31]. Also, IR Kelly et al. [3942] used a thermocouple mesh to monitor
and ultrasonic thermal monitoring techniques are commercially the melt temperature prole and presented a number of com-
available and used by some producers in the polymer industry. prehensive studies on factors inuencing the melt temperature
Moreover, resistance temperature detectors (RTD) are commer- homogeneity other than the process settings. The mesh was placed
cially available and mostly these are made out of platinum and across the melt ow at the screw tip or in-between the adapter and
known as platinum resistance thermometers (PRT). Normally, a die. They observed that a maximum temperature exists at the cen-
PRT is used where very precise measurements are required and/or tre of the melt ow and the melt temperature prole varied across
where high temperatures are involved [2]. However, a PRT also pro- the melt depending on the proximity to the screw root, barrel wall
vides a point measurement only. In fact, there are no currently or melt pool. The extruder screw geometry and polymer properties
available established techniques to measure thermal proles in were found to be the most signicant parameters.
industry. Previous work by the authors [29,43,44] examined the effects of
screw speed and barrel temperatures on the die melt temperature
2.2. Temperature prole measurements prole. A thermocouple mesh was used to measure the temper-
ature prole of the melt ow prior to die entry (a 174 mm long
Work presented by Schlaffer and Janeschitz-Kriegl [32] used adapter was placed in-between screw tip and the die). It was found
thermocouples mounted on a radially movable bar to measure that temperature variations increased as the screw speed increased
the melt temperature prole just in front of the screw tip of a and that the highest uctuations existed a few millimetres away
30 mm diameter single screw extruder. The lowest temperature from the die wall, particularly at high screw speeds. Barrel tem-
was observed close to the barrel wall and the temperature prole perature changes were also shown to inuence the prole, but to a
appeared to be symmetrical around the barrel centreline. The high- lesser effect.
est temperature was observed around 7.5 mm distance from the In another study by Kelly et al. [45], a traversing thermocouple
barrel centreline. The temperature of the middle of the melt ow was used to measure the melt temperature prole, and this gave
was slightly lower than the highest temperature observed and this similar measurements to the thermocouple mesh at the centre of
difference increased as the screw speed increased. the melt ow, but measured lower temperatures close to the die
An up-stream temperature sensor [33,34] was used to mea- wall. It was felt that these lower observed temperatures were due to
sure the melt temperature at a number of different radial positions conduction errors and insufcient time to reach equilibrium at each
across the melt ow. A detailed description of the probe design was position. Sabota et al. [10] used a Dynisco auto traversing probe to
provided while the probe was used to measure the radial temper- monitor the radial melt temperature prole of a 50 mm diameter
ature prole with a few different materials. The shapes of the melt single screw extruder. The lowest and highest melt temperatures
temperature proles varied with the material. The authors stated were observed at the melt close to the wall and middle of the ow
that further development of the probe was required before its use respectively. Moreover, melt temperature proles were shifted up
in industrial applications. from the level of temperature as screw speed increased. In other
Bruker et al. [35] measured the radial melt temperature distri- work reported by Shen et al. [3], the die melt temperature prole of
bution of a 63.5 mm diameter single screw extruder at the end of a 50.8 mm diameter single screw extruder was observed by using an
the screw by using a ring/bar mechanism. Ten thermocouples were auto-traversing probe. They found that the temperature prole was
mounted on the system in the radial direction opposite to the melt highly affected by processing conditions and material properties
ow. The lowest and almost identical melt ow temperatures were such as screw speed or output, barrel set temperatures, die wall set
observed in the middle of the melt ow and in the melt close to the temperatures, material viscosity, etc. Moreover, they found that the
barrel wall. The highest melt temperature was observed around the radial temperature prole was more signicantly inuenced by the
middle of the melt ow in-between the barrel centre and wall. set die wall temperature at low screw speeds than the high screw
Miaw et al. [36] used a device consisting of a bar with thermo- speeds.
couple probes facing upstream to measure the melt temperature Bur et al. [27] used a non-invasive uorescent measurement
prole of a 63.5 mm diameter single screw extruder just after the technique to measure the vertical temperature prole across the
screw tip. A roughly symmetrical temperature prole was observed melt ow inside the extruder barrel (i.e. the screw root to barrel
around the melt ow centreline. The highest and the lowest melt wall) and discussed the effects of screw speed, screw geometry
temperatures were observed in the centre of the melt ow and and material specic behaviour on the melt temperature prole.
the melt close to the barrel wall respectively. The atness of the They observed quite a at vertical temperature prole at low screw
temperature proles decreased as the screw speed increased. speeds with a conventional screw. The conventional screw temper-
A cam-driven thermocouple system was used to measure the ature prole became curvilinear as the screw speed increased since
radial temperature distribution along the centreline of a single the melt temperature at the middle of the melt ow was lower
screw extruder screw channel [37]. The effects of barrel cooling than the melt temperatures close to the screw root and the barrel
C. Abeykoon et al. / Sensors and Actuators A 182 (2012) 1627 19

wall. There were distinct differences in temperature prole shapes


between a mixing screw and the conventional screw. The lowest
temperature was observed in the melt near the barrel wall with
the mixing screw rather than the middle of the melt ow.
Schoppner et al. [31] used an IR thermometer with a 10 ms
response time to measure the melt temperature prole across the
screw channel at different locations of the screw. The solids convey-
ing zone melt temperature prole showed a maximum peak close
to the passive ight (i.e. in the melt pool area) and the minimum
peak close to the active ight (i.e. in the solid bed area) while the
metering zone channel prole showed the opposite order of peaks Fig. 2. Details of the rapid compression screw.
at low screw speeds (10 and 15 rpm). However, no signicant dif-
ferences in the patterns of the temperature traces at high screw
speeds were observed other than the magnitude differences.
A non-invasive electrical capacitance tomography (ECT) tech-
nique was proposed to measure the melt temperature distribution
in polymer extrusion [46]. An ECT sensor with eight electrodes was
developed and studies are underway to carry out experimental tri-
als on a laboratory scale extruder. The authors believe that ECT will
permit useful temperature measurements of a full two dimensional
melt plane in real-time with a fast response time.
From the discussion above, it is evident that numerous attempts
have been made at extrusion thermal monitoring over the last
few decades. However, it seems that none of the well established
industrial thermal monitoring methods can provide a thermal pro- Fig. 3. Dimensions of the equipment.
le measurement with detailed thermal information across the
melt ow. Some of the temperature measurement techniques in
research are capable of thermal prole measurements but these adapter and die. All of these temperature zones are equipped with
techniques are not yet suitable for use in a production environ- Davis Standard Dual-Therm controllers. The extruder drive is
ment due to constraints such as their complexity, limited durability, a horizontal type separately excited direct current (SEDC) motor
access requirements, disruptive effects on melt ow and out- which has ratings: 460 Vdc, 50.0 hp (30.5 kW), at speed 1600 rpm.
put, etc. Therefore, further improvements to extrusion thermal The motor and screw are connected through a xed gearbox with
monitoring to capture thermal variations across the melt ow a ratio of 13.6:1, and according to the manufacturers information
are invaluable. The development of such advanced process ther- the gearbox efciency is relatively constant at all speeds (96%).
mal prole monitoring techniques may enable improved process The motor speed was controlled by a speed controller (MENTOR II)
thermal monitoring and control and help to overcome prevalent based on speed feedback obtained through a direct current (d.c.)
problems in extrusion industry which may result in high defec- tachometer generator.
tive product rates, long down times, and waste of energy, labour Melt pressure was recorded by a Dynisco TPT463E pressure
and materials. In the next section, an experimental evaluation transducer close to the screw tip. The point/bulk temperature of the
is carried out of ve different melt temperature measurement die melt was measured by four different sensors: a 1 mm diameter
methods discussed in the literature review. Three different types non-insulated wall mounted thermocouple (Non-ins TC), a 3 mm
of wall mounted thermocouples and an IR temperature sensor diameter semi-insulated (a small air gap available surrounding
were selected for experiments as these are used in industrial the sheathed thermocouple junction) wall mounted thermocouple
applications. Also, a well established thermal prole measure- (Semi-ins TC), a 0.5 mm diameter insulated wall mounted thermo-
ment technique in research, a thermocouple mesh, was selected couple (Ins TC) and a Methotrexate (MTX) IR sensor (Dynisco MTX
for temperature prole measurements. Using these ve techniques 922-6/24). More details of the arrangement of the sensors on the
allowed a detailed comparison to be made between their perfor- machine are shown in Fig. 3. Moreover, melt temperatures at the
mance. different radial locations of the die were measured using a thermo-
couple mesh (TCM) with seven junctions (i.e. with 1 negative and
7 positive thermocouple wires) placed in-between the adapter and
3. Experimental evaluation on ve different melt die as shown in Fig. 4.
temperature measurement methods

3.1. Equipment

All measurements were carried out on a 63.5 mm diameter (D)


single screw extruder (Davis Standard BC-60) at the IRC laboratories
of the University of Bradford. A tapered rapid compression screw
with 3:1 compression ratio (feed or solids conveying-length: 12 D
and channel height: 10.53 mm, compression or melting-length:
2 D, metering-length: 10 D and channel height: 3.50 mm) as
shown in Fig. 2 was used to process the materials. The extruder
was tted with a 38 mm diameter adapter by using a clamp ring
prior to a short 12 mm diameter capillary die as shown in Fig. 3.
The extruder barrel has four separate temperature zones and
another three separate temperature zones at the clamp ring, Fig. 4. Extruder die, adapter, and thermocouple mesh.
20 C. Abeykoon et al. / Sensors and Actuators A 182 (2012) 1627

4. Results and discussion

In general, melt pressure and melt temperature are the most


important parameters in polymer extrusion [2], and indicate pro-
cess functional quality. Therefore, melt pressure was also measured
during the experiments. Melt pressure development during the
experiment with screw speed changes are shown in Fig. 6(a). Melt
pressure signal indicates that the process operation was under nor-
mal conditions. Melt temperature measurements made by each
method are shown in Fig. 6(b) and this provides a good comparison
of melt temperature measurements made by the different meth-
ods. The measured melt temperature signals corresponding to the
90 rpm screw speed are shown in Fig. 6(c) for more clarity. The
measurement made by the semi-insulated thermocouple is similar
to and overlaps the insulated wall mounted thermocouple signal.
Therefore, the semi-insulated thermocouple signal is not shown in
Fig. 6(b).
Wall mounted thermocouples are very poor at detecting uctua-
tions in melt temperature which is clearly noticeable from Fig. 6(b).
This is due to the slow response time of these sensors as discussed
Fig. 5. The thermocouple mesh arrangement.
above. Also, these readings are highly dominated by the barrel/die
wall metal temperature which may be another reason for having
Work by Brown and Kelly [40,42] has previously presented relatively constant readings during screw speed changes. However,
this technique in detail and has conrmed that the temperature of these wall mounted thermocouples, the non-insulated thermo-
measurements across the melt ow are symmetrical across the couple shows a slightly higher temperature reading throughout the
thermocouple mesh centreline when averaged over signicantly period than the insulated and semi-insulated thermocouples.
long periods of time. Therefore, the thermocouple mesh junctions The experimental observations showed that the IR sensor
were placed asymmetrically across the melt ow along the mesh captures the same pattern of temperature variations as the ther-
diameter (distance from the die centreline to each junction: 0 mm mocouple mesh. Also, it shows slightly higher melt temperature
(TCM-R0), 3.5 mm (TCM-R3.5), 4.5 mm (TCM-R4.5), 9.2 mm (TCM- readings than wall mounted thermocouples at 10, 30, and 50 rpm
R9.2), 10.5 mm (TCM-R10.5), 15.8 mm (TCM-R15.8) and 16.5 mm screw speeds. As shown in the measured IR sensor signal, melt
(TCM-R16.5)) as shown in Fig. 5. Then, the melt temperatures mea- temperature increases until 30 rpm by following the screw speed
sured at different points across the die melt ow can be mirrored increments and afterwards melt temperature reduces as screw
over the die centreline to obtain the complete die melt temperature speed increases. The IR sensor captured up to a 30 C magnitude
prole. A data acquisition programme developed in LabVIEW was of temperature variations over the measurement period.
used to communicate between the experimental instruments and The thermocouple mesh reads the melt temperatures of 7 dif-
a PC. All signals were acquired at 10 Hz using a 16-bit DAQ card, ferent radial locations of the melt ow. According to all radial
National Instruments (NI) PCI-6035E, through a NI TC-2095 ther- measurements, the thermocouple mesh shows melt temperature
mocouple connector box and a NI low-noise SCXI-1000 connector variations in the range of 075 C throughout the measurement
box. period. In particular, the temperature of the melt at the middle
of the melt ow (i.e. TCM-R0, TCM-R3.5 and TCM-R4.5) shows a
3.2. Materials and experimental conditions gradual increase as screw speed increases while increasing the
magnitude of variations. Melt temperatures at the TCM-R9.2 and
Experimental trials were carried out on a virgin low density TCM-R10.5 radial positions start to decrease with the step increase
polyethylene (LDPE), Dow LD150R (density: 0.921 g/cm3 and MFI: of screw speed from 10 to 30 rpm and also this happens with the
0.25 g/10 min). The melt ow index (MFI) value is presented accord- step increase of screw speed from 30 to 50 rpm at the TCM-R15.8
ing to the ISO 1133 standard (190 C, 2.16 kg). LDPE was selected for and TCM-R16.5 radial positions. Therefore, it is evident that the
the experiments as it is known to be a material with less process- thermal behaviour of the melt is dependent upon the die radial
ing problems and hence may not damage the thermocouple mesh, position.
particularly at high screw speeds. The screw speed was adjusted In comparison, it seems that wall mounted thermocouples are
from 10 rpm to 90 rpm in steps of 20 rpm. The extruder tempera- unable to provide details of the actual melt thermal behaviour other
ture settings were xed as described in Table 1. These temperature than a constant value of melt temperature which may be signi-
settings were selected from the middle of the operating window cantly inuenced by the metal barrel wall temperature. Although,
of the extruder for this particular material. This therefore allowed the diameters of the wall mounted thermocouple probes are dif-
operating the process under normal conditions as process speed ferent, no noticeable differences between their readings can be
changed from 1090 rpm. Process data was recorded continuously observed. This may be due to the fact that all of these sensors
whilst the extruder was allowed to stabilise at each screw speed. were ush mounted to the barrel wall and hence there are no shear
heating effects at this position. The IR sensor provides the process
thermal information related to a material specic volume of poly-
Table 1
Extruder barrel temperature settings. mer but it is incapable of providing position specic thermal details.
Also, it reads lower magnitudes of melt temperature uctuations
Temperature settings ( C)
than the thermocouple mesh although it captures the same pat-
Barrel zones Clamp ring Adapter Die tern of melt temperature variations as the thermocouple mesh to
1 2 3 4
some extent. Moreover, the IR measurements may be affected by
the polymer type and the sensor calibration accuracy [8]. Among
140 170 185 200 200 200 200
the 5 methods, the thermocouple mesh provides more detailed
C. Abeykoon et al. / Sensors and Actuators A 182 (2012) 1627 21

30
90 rpm

Melt pressure (bar)


25
70 rpm
20
50 rpm
15 30 rpm
10 10 rpm
5
0 rpm (a)
0
230
TCM-R4.5, R3.5, R0
TCM-R10.5
220

210
TCM-R9.2
Melt temperature (oC)

200

190
TCM-R16.5 Insulated-TC TCM-R15.8
TCM - R0
180 TCM - R3.5
MTX-IR Non-insulated-TC
TCM - R4.5
170 TCM - R9.2
TCM - R10.5
TCM - R15.8
160
TCM - R16.5
MTX - IR
150 Non-In-wall-TCTC
Non-insulated-TC
In-wall-T-TC
Insulated C (b)
140
0 320 1345 1760 2140 2400 2600
230
TCM-R4.5, R3.5, R0 TCM-R9.2
220

210
Melt temperature (oC)

200

190

180

170
Insulated-TC
MTX-IR
160
TCM-R16.5 Non-insulated-TC
150
TCM-R10.5 TCM-R15.8
(c)
140
2400 2500 2600
Time (s)

Fig. 6. (a) Measured melt pressure signal during the experiment, (b) melt temperature measurements provided by each method, (c). melt temperature measurements at
90 rpm.

information regarding melt thermal behaviour particularly on posi- However, mesh wires may be damaged under poor melting con-
tion specic thermal uctuations. However, the melt ow may be ditions (i.e. as the melt ow becomes highly viscous as un-melted
slightly disturbed by the thermocouple mesh as it is positioned materials come out with the extrudate) and through melt solidi-
across the melt ow. Furthermore, thermocouple mesh measure- cation on cooling. Therefore, the thermocouple mesh technique is
ments are slightly inuenced by shear heating depending on the not yet suitable for extensive industrial applications.
size of the mesh wires and the mesh junctions [42]. However, the
design of the thermocouple mesh is carried out to avoid these prob- 4.1. Frequency analysis
lems by using thermocouple wires and junctions which are small
in size. In fact, the thermocouple mesh provides highly accurate It is useful to examine the critical frequencies of melt temper-
and detailed measurements under normal operating conditions of ature uctuations given by each method for further evaluation
the process and hence is highly useful in a research environment. of sensor performance and understanding of the actual process
22 C. Abeykoon et al. / Sensors and Actuators A 182 (2012) 1627

Fig. 7. PSD spectra of melt temperature of different sensors at a screw rotation speed of 90 rpm.

thermal dynamics. Therefore, a frequency analysis of the nor- components above 1.5 Hz. Therefore, only frequencies up to 1.5 Hz
malised melt temperature signals measured by each method at are displayed in the PSD spectra for more clarity. The PSD spectra at
90 rpm was carried out to produce a power spectral density (PSD) 90 rpm for all the sensor signals are shown in Fig. 7. Moreover, the
spectrum. This speed was selected as the highest temperature uc- minimum, mean and maximum temperatures measured by each
tuations were observed here. The Welch method [47] was applied sensor at 90 rpm (i.e. of the same 2000 data points used for the PSD
in a Hamming window equal to the data length. This method tends analysis) are shown in Table 2 together with the standard deviation
to decrease the variance due to spectral leakage of the power spec- (SD) at each condition.
tral density estimate relative to computing the power spectrum As shown by the PSD spectra in Fig. 7, it is evident that melt
over the entire data. In general, the magnitude of a PSD spectrum temperature uctuations are very slow and non-periodic in nature.
(i.e. Y-axis) shows power inclusion of the signal at each frequency Most of the signicant uctuations have frequencies below 0.5 Hz.
indicated along the X-axis. Only 2000 data points of each sig- All wall mounted thermocouples show the dominant frequency
nal (i.e. data of 200 s shown in Fig. 6(c)) at 90 rpm were used component at 0.005 Hz and hence a time period of 200 s. This means
in the frequency studies. While frequencies of up to 5 Hz can be it is purely due to a drift in the mean data over the measurement
observed from the data, no signal showed any signicant frequency period and therefore is fairly irrelevant to the process. Additionally,
C. Abeykoon et al. / Sensors and Actuators A 182 (2012) 1627 23

Table 2 0.395 Hz, and 0.435 Hz. However, the power included in the fre-
Minimum, mean and maximum melt temperatures measured by each sensor at the
quency components is not equal as evidenced by different PSD
90 rpm together with the corresponding standard deviation.
spectrum magnitudes. Moreover, the standard deviation of the
Sensor Measured melt temperature ( C) TCM-R16.5 signal is around 3.25 times higher than the standard
Minimum Mean Maximum SD deviation of the IR sensor. As mentioned earlier, the magnitude
of a PSD spectrum shows power inclusion of the signal at each fre-
TCM-R0 179.4 214.0 222.4 4.85
TCM-R3.5 182.2 213.4 221.7 5.03 quency. By having a lower magnitude in the PSD spectrum together
TCM-R4.5 194.8 211.8 221.2 4.68 with a lower standard deviation than the TCM-R16.5 signal, the
TCM-R9.2 157.0 199.9 217.1 8.84 IR sensor shows its poor performance in capturing thermal uc-
TCM-R10.5 143.9 194.1 212.9 13.11 tuations over the thermocouple mesh. This may link with the IR
TCM-R15.8 157.6 183.7 203.0 8.87
TCM-R16.5 172.6 186.9 197.5 5.92
sensors conical measurement volume as readings are mostly con-
MTX-IR 186.2 190.0 193.9 1.82 centrated close to the sensor tip.
Non-ins TC 198.2 196.8 197.9 0.49 The PSD spectra of thermocouple mesh signals at different radial
Ins TC 191.8 193.7 195.1 0.70 positions show different patterns. As shown in Fig. 6(b), the TCM-
Semi-ins TC 192.2 193.2 194.7 0.60
R9.2, TCM-R10.5, TCM-R15.8 and TCM-R16.5 signals contain higher
magnitudes of variations than the TCM-R0, TCM-R3.5 and TCM-
R4.5 signals. These higher thermal variations are reected by the
magnitudes of the PSD spectra as well (see Fig. 7) by having lower
magnitudes of the PSD spectra for the TCM-R0 and TCM-R4.5 sig-
nals than the PSD spectra magnitudes of mesh wires close to the
barrel wall. According to the thermal information given by the ther-
mocouple mesh data (i.e. the measured signals shown in Fig. 6(b),
PSD analysis, and standard deviations), the highest melt tempera-
ture uctuations in the melt ow can be observed at the TCM-R10.5
(i.e. 10.5 mm away from the die wall) rather than the middle of the
melt ow or melt close to the barrel wall.
In summary, this analysis conrmed that wall mounted thermo-
couples give very poor performance in capturing melt temperature
uctuations as shown by very low magnitudes in the PSD spectra.
The IR sensor can capture thermal uctuations up to a considerable
Fig. 8. A comparison of the frequency components with a lower power of the MTX-IR level and its PSD spectrum is quite similar to the TCM-R16.5 signal
and TCM-R16.5 PSD spectra.
PSD spectrum. This means that the IR sensor readings only show
the melt ow information close to the barrel wall. Also, it seems
the PSD spectra of the insulated and semi-insulated thermocou- that the thermocouple mesh measurements are highly useful in
ples consist of another signicant frequency component at 0.035 Hz capturing detailed thermal information across the melt ow.
and the magnitude of this frequency component is higher for the
insulated thermocouple than the semi-insulated thermocouple.
However, this component is not noticeable in the PSD spectrum of 4.2. Covariance analysis
the non-insulated thermocouple. Perhaps, it cannot capture some
of the uctuations in melt temperature as it might be highly inu- Correlation coefcients were calculated between each of the
enced by the barrel metal wall temperature. Overall, all of the wall individual sensor measurements at 90 rpm and are given in Table 3.
mounted thermocouples show very low standard deviations (i.e. The coefcient matrix is symmetric and hence only a half of the
below 0.7 C) of their measurements as shown in Table 2 while hav- matrix is shown in Table 3. Moreover, the correlation coefcients
ing very low power in their PSD spectra as shown in Fig. 7. These of the Non-ins TC, Semi-ins TC, Ins TC and MTX-IR with each ther-
indicate their poor performance in capturing thermal uctuations mocouple mesh junction are shown in Fig. 9. Normally, correlation
in the melt ow. coefcient represents the normalised measure of the strengths and
A comparison of the frequency components of the MTX-IR and directions of the linear relationship between two variables and this
TCM-R16.5 signals PSD spectra (i.e. except the components with ranges from 1 to 1, where positive values indicate that variables
higher power which are shown in Fig. 7(b) and (h)) is shown in are positively correlated (i.e. variables vary in the same direction)
Fig. 8. As shown in Figs. 7(b), (h) and 8, the melt temperature sig- and the strength of the positive correlation increases from 0 to 1.
nals of the IR sensor and TCM-R16.5 hold critical frequencies in Negative values specify that variables are negatively correlated (i.e.
the same locations: 0.01 Hz, 0.055 Hz, 0.155 Hz, 0.32 Hz, 0.375 Hz, variables vary in the opposite direction) while the strength of the

Table 3
Correlation coefcients between each of the individual sensor measurements at 90 rpm.

TCM-R0 TCM-R3.5 TCM-R4.5 TCM-R9.2 TCM-R10.5 TCM-R15.8 TCM-R16.5 MTX-IR Non-ins TC Ins TC Semi-ins TC

TCM-R0 1.000
TCM-R3.5 0.850 1.000
TCM-R4.5 0.426 0.048 1.000
TCM-R9.2 0.170 0.361 0.226 1.000
TCM-R10.5 0.011 0.216 0.444 0.326 1.000
TCM-R15.8 0.019 0.084 0.071 0.104 0.189 1.000
TCM-R16.5 0.144 0.101 0.135 0.004 0.098 0.022 1.000
MTX-IR 0.071 0.038 0.260 0.201 0.229 0.100 0.739 1.000
Non-ins TC 0.099 0.082 0.008 0.002 0.004 0.337 0.375 0.425 1.000
Ins TC 0.086 0.050 0.080 0.027 0.014 0.338 0.522 0.532 0.766 1.000
Semi-ins TC 0.108 0.054 0.098 0.048 0.028 0.237 0.370 0.411 0.685 0.822 1.000
24 C. Abeykoon et al. / Sensors and Actuators A 182 (2012) 1627

1 16.5 mm, 15.8 mm, 10.5 mm, 9.2 mm, 4.5 mm, 3.5 mm,
MTX-IR Ins TC Non-ins TC Semi-ins TC
0.8 0 mm radial positions) of the die over the time with applied screw
0.6 speed step changes (i.e. 050 rpm) is given in Fig. 10. The time peri-
Correlation coefficient

ods relevant to the applied screw speed step changes are shown in
0.4
Fig. 6. As shown in Fig. 10, an almost at temperature prole is
0.2
shown when the screw speed is zero (i.e. no material in the die)
0 but at a lower temperature than what was set at the die wall.
-0.2 The melt temperature suddenly increases just after the increase
-0.4 of screw speed with the addition of process mechanical heat (i.e.
-0.6 heat from frictional and viscous mechanisms) and keeps increas-
ing as the screw speed further increases, while the atness of the
-0.8
temperature prole decreases.
-1
0 3.5 4.5 9.2 10.5 15.8 16.5 19 To further understand the effects of screw speed on the die melt
Thermocouple mesh radial position (mm) temperature prole, the fully developed melt temperature of each
screw speed was plotted by rotating the single measured axis (i.e.
Fig. 9. Correlation coefcients between each thermocouple mesh junction mea- average values of the measured melt temperature at each radial
surement and the other temperature sensors. position) around its centre point and the proles are shown in
Fig. 11. The die wall set temperature was used as the melt temper-
negative correlation between signals increases from 0 to 1. Val- ature at the 19 mm radial positions. Then, 15 radial positions (i.e.
ues close to or equal to 0 suggest that there is no linear relationship distance to each radial position from die centreline of: 19 mm,
between the variables. 16.5 mm, 15.8 mm, 10.5 mm, 9.2 mm, 4.5 mm, 3.5 mm,
As shown in Fig. 9, all sensor signals show their strongest corre- 0 mm) make a complete temperature prole for this study as shown
lation with the TCM-R16.5 measurements where the IR sensor has in Fig. 5. At low screw speeds, the highest temperature exists in the
the highest correlation with the TCM-R16.5 of these sensors. Such middle of the melt ow and temperature decreases towards the die
a close correlation between the IR sensor and the TCM-R16.5 sig- wall, while the melt close to the die wall has the lowest tempera-
nals was also conrmed by PSD analysis. As previously mentioned, ture. The difference between the highest and lowest temperatures
this is linked to the effective measurement volume and effective of a particular die melt ow cross-section is increased with the
penetration depth of the IR temperature measurement technique. increase of screw speed and these differences are around 11 C,
Moreover, none of the other sensors shows good correlations with 13 C, 17 C, 19 C and 33 C at 10, 30, 50, 70 and 90 rpm respec-
the TCM-R0, TCM-R3.5, TCM-R4.5, TCM-R9.5 and TCM-R10.5 mea- tively. Moreover, changes to the shape of the melt temperature
surements. In general, correlation coefcient analysis also conrms prole can be observed as screw speed increases. A lower temper-
that the most commonly used point/bulk temperature measure- ature region is created in the melt a few millimetres away from
ment techniques in polymer extrusion are poor in performance and the adapter wall by generating a shoulder region and this can be
provide only information close to the barrel/die wall to which they clearly observed at 70 and 90 rpm. This means that the tempera-
are xed. tures of some positions of the melt ow are lower than even the
adapter wall set temperature. This may happen due to poor or late
4.3. Die melt temperature prole melting (i.e. maybe due to poor heat generation and transfer) and/or
poor mixing of material at high screw speeds as the reduction of
Apart from the evaluation of the sensors performance, an inves- material residence time which may signicantly affect the prod-
tigation of the melt temperature prole at the entrance to the uct quality. Perhaps, the rate of transfer of heat supplied by barrel
cylindrical die was carried out. Die melt temperature proles were heaters (i.e. conduction heat) to the material may be reduced as
determined from the measured thermocouple mesh data at differ- the material residence time decreases with the increase of screw
ent radial locations. The dimensions of the experimental setup and speed. It can be seen that the temperature of the shoulder regions
the placement of the thermocouple mesh across the melt ow are further reduces as screw speed increases. Also, a large amount of
shown in Fig. 3. An illustration of the symmetrical 3-D temperature process mechanical heat is generated near the barrel and screw sur-
prole (which is relevant to the measured melt temperatures of the faces through friction. Therefore, the materials at the middle of the

Fig. 10. Melt temperature prole from the extruder start-up conditions to the fully developed ow at 10, 30 and 50 rpm.
C. Abeykoon et al. / Sensors and Actuators A 182 (2012) 1627 25

Fig. 11. Radial melt temperature proles (i.e. plotted by rotating the average melt temperature of the thermocouple mesh around the central ow axis) in fully developed
ow at 10, 30, 50, 70 and 90 rpm.

melt ow between the barrel and screw surfaces may obtain less cross-section. They can only capture some thermal details of the
heat for melting which causes poor melting and hence a lower tem- melt close to the barrel wall. Therefore, the process thermal stabil-
perature shoulder region in this area particularly at higher screw ity cannot be guaranteed if these point/bulk measurement methods
speeds. Brown et al. [48] argued that the formation of the lower show a consistent melt temperature during the process and this
temperature shoulder region can be attributed to polymer owing is absolutely clear from the thermal measurements presented in
from the melt pool which develops along the screw length. More- Fig. 7. This fact was further conrmed by observation of the melt
over, geometrical effects of screws have a great inuence on these temperature proles at different screw speeds. In fact, there can
prole shapes while material properties also have an impact up to be thermal uctuations within the melt ow which cannot be
some extent in addition to the processing conditions [22,27,29,40]. diagnosed by these point/bulk methods. However, it is difcult
In fact, these proles show how signicant position specic thermal to make more detailed thermal prole measurements without
uctuations in the melt ow cross-section are in polymer extrusion. disturbing the melt ow with the existing thermal prole measure-
Obviously, the existing point/bulk melt temperature measurement ment methods and hence non-invasive measurement techniques
techniques cannot indicate most of these signicant thermal uctu- may be invaluable to overcome this problem. Although, a number
ations in the melt ow. Therefore, details provided by these sensors of non-invasive techniques are available within the industry they
are unrepresentative of the actual process thermal behaviour. make point/bulk measurements only and do not offer detailed and
accurate process thermal information. Therefore, the development
4.4. Importance of thermal prole measurement of powerful soft sensor approaches for thermal prole prediction
may be one of the future possibilities for lling the existing gaps
The PSD spectra of thermocouple mesh junctions at different in extrusion thermal monitoring. For example, techniques such as
radial locations show signicant differences and this illustrates the the thermocouple mesh can be used in a research laboratory to
differing thermal behaviour at different radial locations of the melt make actual thermal prole measurements of the melt ow and
ow. Melt temperature uctuations in the middle of the melt ow then these measurements can be used to develop process mod-
are not as severe as the uctuations which occur in the melt close els to predict a die melt temperature prole incorporating other
to the barrel wall. Also, these observations prove that typical wall readily measurable process parameters (i.e. a soft sensor approach
mounted thermocouples and IR temperature sensors cannot pro- for melt temperature prole prediction). Furthermore, generalisa-
vide a good indication of the thermal behaviour across the melt ow tion of these models may enable their use with different materials,
26 C. Abeykoon et al. / Sensors and Actuators A 182 (2012) 1627

machines and processing conditions. In the past, numerous showed that the thermal uctuations varied signicantly across
research efforts have focused on the development of rst principle the melt ow cross-section and these variations increased with the
models to represent the polymer extrusion process mechanisms screw speed. Therefore, point/bulk measurement techniques are
and dynamics [19,49,50]. Such models involve complex, coupled not capable of detecting most of these important thermal varia-
partial differential equations and require computational solutions tions although they are commonly used in practice. Based on the
by nite difference techniques. Therefore, these approaches were results of this study, it seems that the use of thermal prole mea-
not practical at that time as these equations/models could not be surements are highly appropriate in determining the actual process
solved without powerful computers. However, solving these kinds thermal stability. Although, the thermocouple mesh technique is a
of complex models may be possible today with the development of highly useful technique of capturing detailed thermal information,
super computers with very high speed data processing capacities. it is not yet robust enough for extensive industrial applications due
Conversely, some of the presently available modelling techniques to its greater fragility. Also the placement of physical sensors across
may be used to develop simple and highly accurate process models the melt ow for thermal prole measurement are less attractive in
which can be used in real-time applications without demanding industrial environments due to their disruptive effects on the melt
high computational facilities. Hence, the development of accurate ow. Presently, an industrially usable thermal prole technique is
process models may permit soft sensor approaches which can pre- not available and all the industrial processes use point/bulk tech-
cisely represent the process dynamics. These will help not only to niques for process thermal monitoring and control. In fact, detailed
strengthen the process monitoring abilities but also the process and accurate process monitoring techniques are fundamental to
control strategies. advanced process control techniques. Therefore, process thermal
Conversely, the ability to accurately measure melt tempera- control strategies based on conventional point/bulk sensor feed-
ture proles as opposed to single point or bulk measurements back may not provide a satisfactory control performance and may
would be a useful tool in solving a range of processing prob- be responsible for the existing processing problems such as high
lems currently found within the industry. In polymer extrusion, defective product rates and long down times resulting in signicant
melt output from machines should be uniform (i.e. uniformity of material and energy waste. In the future, the extrusion industry will
melt temperature over time and radial homogeneity across the look for more detailed (e.g. thermal prole measurements), precise,
melt ow) throughout the process to achieve good quality prod- non-invasive, simple, economical and durable thermal monitoring
ucts. Temporal thermal variations within the melt output may techniques. Therefore, the development of powerful soft sensor
cause variations of melt pressure and hence output rate, and result approaches should be useful as one of the solutions to overcome
in non-uniformities of optical/mechanical/chemical properties of the existing process thermal monitoring problems. Development
extruded parts, or extrudate with un-melted particles, etc. [51,52]. of such techniques will strengthen process control performance as
Excessive variation in melt temperature may also lead to varia- well to solve major processing problems.
tion in mechanical properties and problems during cooling such as
localised warpage and frozen-in stresses [52]. As previously dis- References
cussed, most of the point/bulk measurements are restricted to the
barrel/die wall temperature or melt close to the barrel/die wall. [1] M.A. Spalding, K.S. Hyun, Troubleshooting mixing problems in single-screw
However, barrel/die wall temperature is not indicative of the poly- extruders, SPE ANTEC Technical Papers (2003) 229233.
[2] C. Rauwendaal, Polymer Extrusion, 4th ed., Hanser, 2001.
mer melt temperature [32]. Therefore, the measurement of melt [3] X. Shen, R. Malloy, J. Pacini, An experimental evaluation of melt tempera-
temperature prole helps to determine the actual thermal homo- ture sensors for thermoplastic extrusion, SPE ANTEC Technical Papers (1992)
geneity of the melt ow cross-section and hence to control process 918926.
[4] B.H. Maddock, Measurement and analysis of extruder stability, SPE Journal 20
parameters for achieving the optimum melt ow homogeneity. (1964) 12771283.
Conversely, proper understanding of the actual thermal behaviour [5] E.L. Steward, Recording pressure and melt temperature in extrusion, SPE ANTEC
of the melt ow cross-section can greatly assist process optimisa- Technical Papers (1998) 320324.
[6] E.L. Steward, Control of melt temperature on single screw extruders, SPE ANTEC
tion, for example through improved product design, selection of
Technical Papers (1999), Paper No: 813.
set processing conditions and die/screw design while permitting [7] C. Warren, IR sensors: a good approach for industrial and process applications,
the selection/design/optimisation of these processing factors to be Chiltons Instruments and Control 58 (6) (1985) 3638.
[8] C. Maier, Infrared temperature measurement of polymers, Polymer Engineering
driven into a new direction.
and Science 36 (11) (1996) 15021512.
[9] W. Obendrauf, G.R. Lagecker, W. Friesenbichler, Temperature measuring in
plastics processing with infrared radiation thermometers, International Poly-
5. Conclusions mer Processing 13 (1) (1998) 7177.
[10] K.D. Sabota, D.R. Lawson, J.S. Huizinga, Advanced temperature measurements
in polymer extrusion, SPE ANTEC Technical Papers (1995) 28322842.
This work aimed to review of previous attempts made in [11] A.L. Kelly, E.C. Brown, M. Woodhead, P.D. Coates, Extrusion temperature eld
extrusion melt temperature measurements and to carry out an measurement: a soft sensor approach, SPE ANTEC Technical Papers (2001),
Paper No: 240.
experimental evaluation of the selected melt temperature sen- [12] A.L. Kelly, E.C. Brown, P.D. Coates, Infrared melt temperature measurement:
sors to understand the actual thermal behaviour of the melt ow effect of ller content on penetration depth, SPE ANTEC Technical Papers (2003)
while evaluating sensor performance. A number of melt tem- 33063310.
[13] A.L. Kelly, E.C. Brown, M. Woodhead, P.D. Coates, Temperature measurement
perature measurement techniques have been used in polymer methods for polymer processing, in: PPS 18th Annual Meeting, Guimares,
extrusion with both point/bulk and prole measurements. Exper- Portugal, 2002, Paper No: 122.
imental evaluation showed that the most commonly used wall [14] A. Bendada, M. Lamontagne, A new infrared pyrometer for polymer tempera-
ture measurement during extrusion molding, Infrared Physics and Technology
mounted thermocouples are poor in capturing process thermal
46 (12) (2004) 1115.
variations. However, such measurements are useful to obtain rough [15] A.W. Christie, Using thermal process imaging for extruder analysis and
estimates of the melt temperature. The IR sensor showed better improvement, SPE ANTEC Technical Papers (2005) 335339.
[16] R. Spares, B.R. Whiteside, P.D. Coates, High speed thermal imaging of micro-
performance in detecting thermal variations than the thermocou-
moulding, SPE ANTEC Technical Papers (2009) 27502753.
ples. Frequency and covariance analysis indicated that the thermal [17] M.D.P. Noriega, T.A. Osswald, N. Ferrier, In line measurement of the polymer
information provided by typical wall mounted thermocouples and melting behavior in single screw extruders, Journal of Polymer Engineering 24
an IR sensor are limited to a small volume of the melt ow (i.e. (6) (2004) 557578.
[18] M.D.P. Noriega, T.A. Osswald, N. Alberto, Comparison of the melting behaviour
a point or bulk measurement) where they were xed in the bar- of HDPE and PP in single screw extruders, SPE ANTEC Technical Papers (2011)
rel or die wall. Measurements made by the thermocouple mesh 11661173.
C. Abeykoon et al. / Sensors and Actuators A 182 (2012) 1627 27

[19] Z. Tadmor, Fundamental of plasticating extrusionc I. A theoretical model for [46] Y. Yang, W. Yang, H. Zoang, Temperature distribution measurement and control
melting, Polymer Engineering and Science 6 (3) (1966) 185190. of extrusion process by tomography, in: IEEE IST 2008, 1, Crete, Greece, 2008,
[20] T.-F. Chen, K.T. Nguyen, S.-S.L. Wen, C.-K. Jen, Temperature measurement of pp. 170174.
polymer extrusion by ultrasonic techniques, Measurement Science and Tech- [47] P.D. Welch, The use of fast fourier transform for the estimation of power spec-
nology 10 (3) (1999) 139145. tra: a method based on time averaging over short, modied periodograms, IEEE
[21] E.C. Brown, P. Olley, P.D. Coates, In line melt temperature measurement during Transactions on Audio and Electroacoustics AU-15 (1967) 7073.
real time ultrasound monitoring of single screw extrusion, Plastics, Rubber and [48] E.C. Brown, A.L. Kelly, P.D. Coates, Temperature measurement in single screw
Composites 29 (2000) 13. extrusion: late melting and thermal instabilities, SPE ANTEC Technical Papers
[22] E.C. Brown, A.L. Kelly, P.D. Coates, Melt temperature homogeneity in single (2003) 7781.
screw extrusion: effect of material type and screw geometry, SPE ANTEC Tech- [49] Z. Tadmor, S.D. Lipshitz, R. Lavie, Dynamic model of a plasticating extruder,
nical Papers (2004) 183187. Polymer Engineering and Science 14 (2) (1974) 112119.
[23] S.S.L. Wen, T.F. Chen, D.R. Franca, K.T. Nguyen, C.K. Jen, I. Ihara, A. Derdouri, A.G. [50] K. Amellal, P.G. Laeur, Computer simulation of conventional and barrier screw
Rejon, Resent progress in on-line ultrasonic process monitoring, Proceedings extruders, Plastics Rubber and Composites Processing and Applications 19 (4)
of the Society of Photo-Optical Instrumentation Engineers 3399 (19) (1998) (1993) 227239.
122131. [51] P.H. Squires, F.W. Wolf, Designing extruders for quality performance, SPE Jour-
[24] J. Muller, S. Kummer, D. Fischer, New ultrasonic probes for in-line monitoring nal 27 (4) (1971) 6873.
of polymer melts, Measurement Science and Technology 20 (9) (2009) 14. [52] S.H. Tabatabaei, P.J. Carreau, A. Ajji, Effect of processing on the crystalline ori-
[25] E.C. Brown, T.L.D. Collins, A.J. Dawson, P. Olley, P.D. Coates, Ultrasound: a vir- entation, morphology, and mechanical properties of polypropylene cast lms
tual instrument approach for monitoring of polymer melt variables, Journal of and microporous membrane formation, Polymer 50 (17) (2009) 42284240.
Reinforced Plastics and Composites 18 (4) (1999) 331338.
[26] K. Trivedi, D. Cuff, C.L. Thomas, In-process temperature measurement in extru-
sion using continuous wave ultrasound, SPE ANTEC Technical Papers 1 (2004) Biographies
12711273.
[27] A.J. Bur, S.C. Roth, M.A. Spalding, D.W. Baugh, K.A. Koppi, W.C. Buzanowski,
Temperature gradients in the channels of a single-screw extruder, Polymer
Chamil Abeykoon received his BSc degree in mechanical
Engineering and Science 44 (11) (2004) 21482157.
engineering from the University of Peradeniya (Sri Lanka)
[28] A. Gottwald, U. Scheler, Extrusion monitoring of polymer melts using a high
in 2007 and PhD degree in mechanical engineering (pro-
temperature surface-NMR probe, Macromolecular Materials and Engineering
cess control) from the Queens University Belfast (UK) in
290 (5) (2005) 438442.
2011. His research interests include process monitoring,
[29] C. Abeykoon, M. McAfee, K. Li, A.L. Kelly, E.C. Brown, Monitoring the effect of
modelling and control; soft sensors and soft sensing; and
operating conditions on melt temperature homogeneity in single-screw extru-
heat transfer.
sion, SPE ANTEC Technical Papers (2010) 17991806.
[30] C. Abeykoon, M. McAfee, K. Li, P.J. Martin, A.L. Kelly, The inferential monitor-
ing of screw load torque to predict process uctuations in polymer extrusion,
Journal of Materials Processing Technology 211 (12) (2011) 19071918.
[31] V. Schoppner, K. Enns, K. Anger, E. Giese, Dynamic temperature measurement
in polymer processing, SPE ANTEC Technical Papers 1 (2008) 228232.
[32] W. Schlaffer, H. Janeschitz-Kriegl, Measurements of radial temperature proles Peter J. Martin is a Reader in mechanical engineering at
in a single-screw extruder, Plastics and Polymers 39 (141) (1971) 193199. Queens University Belfast in Northern Ireland, and a Fel-
[33] J.V. Leeuwen, Stock temperature measurement in plastifying equipment, Poly- low of the Institution of Mechanical Engineers (IMechE).
mer Engineering and Science 7 (2) (1967) 98109. He has more than 20 years experience of research into
[34] G. Menges, M. Meissner, Improvement in extruder melt temperature control, polymer processes with particular emphasis on extru-
Journal of Macromolecular Science: Part AChemistry 6 (3) (1972) 641656. sion, thermoforming and blow moulding. His current
[35] I. Bruker, C. Miaw, A. Hasson, G. Balch, Numerical analysis of the temperature research interests include viscoelastic/plastic modelling
prole in the melt conveying section of a single screw extruder: compari- of polymers, nite element simulation of processes, large
son with experimental data, Polymer Engineering and Science 27 (7) (1987) strain biaxial testing, in process measurement and control,
504509. micro-forming and thermoplastic composites.
[36] C. Miaw, A. Hasson, G. Balch, Melt temperature measurement in a single screw
extruder, SPE ANTEC Technical Papers (1985) 7986.
[37] M. Esseghir, V. Sernas, On the measurements of the radial temperature distri-
bution in an extruder channel, Advances in Polymer Technology 13 (2) (1994) Dr Adrian L. Kelly is a senior lecturer in polymer engi-
133140. neering at the University of Bradford, working as part
[38] R. Rasid, A.K. Wood, Effect of process variables on melt temperature proles of the IRC Polymer Engineering research group. Adrians
in extrusion process using single screw plastics extruder, Plastics, Rubber and research interests include inprocess monitoring of extru-
Composites 32 (5) (2003) 193198. sion and injection moulding processes, pharmaceutical
[39] A.L. Kelly, E.C. Brown, P.D. Coates, Melt temperature eld measurement in sin- polymer engineering, nanocomposites and polymer recy-
gle screw extrusion: inuence of melt pressure and die geometry, SPE ANTEC cling.
Technical Papers 1 (2005) 291295.
[40] A.L. Kelly, E.C. Brown, P.D. Coates, The effect of screw geometry on melt tem-
perature prole in single screw extrusion, Polymer Engineering and Science 46
(12) (2006) 17061714.
[41] A.L. Kelly, E.C. Brown, K. Howell, P.D. Coates, Melt temperature eld mea-
surement in extrusion using thermocouple meshes, Plastics, Rubber and
Composites 37 (24) (2008) 151157. Dr Elaine C. Brown is a Chartered Mechanical Engineer,
[42] E.C. Brown, A.L. Kelly, P.D. Coates, Melt temperature eld measurement in working as a University Lecturer and in the IRC Polymer
single screw extrusion using thermocouple meshes, Review of Scientic Instru- Engineering Research Group at the University of Bradford.
ments 75 (11) (2004) 47424748. Elaines research on inprocess measurements has focused
[43] C. Abeykoon, K. Li, M. McAfee, P.J. Martin, J. Deng, A.L. Kelly, Modelling the on applying ultrasound to measure melt and/or process
effects of operating conditions on die melt temperature homogeneity in single properties in-line during extrusion, injection moulding
screw extrusion, UKACC International Conference on Control (2010) 4247. and micromoulding, with recent applications to pharma-
[44] C. Abeykoon, K. Li, M. McAfee, P.J. Martin, Q. Niu, A.L. Kelly, J. Deng, A new model ceutical and nanocomposite extrusion, and on new and
based approach for the prediction and optimisation of thermal homogeneity innovative methods of measuring melt temperature in
in single screw extrusion, Control Engineering Practice 19 (8) (2011) 862874. extrusion and injection moulding.
[45] A.L. Kelly, E.C. Brown, M. Woodhead, P.D. Coates, Melt temperature eld mea-
surement and modelling in extrusion, SPE ANTEC Technical Papers (2002)
33673371.

Вам также может понравиться