Extensional Flow Properties of Polymers Using Stretching Flow Methods

Measurement Good Practice Guide No.
63
Extensional Flow Properties of Polymers using

Stretching Flow Methods
Martin Rides and Crispin Allen

Materials Centre
National Physical Laboratory
Abstract:
In many forming processes the polymer melt undergoes predominantly extensional

deformation, for example in blow moulding and film blowing. Furthermore, the extensional
flow behaviour of melts differ significantly from their shear flow behaviour. Thus extensional
flow measurement methods are necessary to appropriately characterise the melt behaviour.
Such data can be used for the development and selection of materials, and for improving
processing.
This Guide discusses various issues related to the need for, and measurement of the
extensional flow properties of polymer melts using stretching methods. It describes practical
considerations for measuring extensional flow behaviour, including measurements at very
high and low strain rates, and includes details on the analysis of the raw data and assessment
of the effect of uncertainties on the measured values. To illustrate the measurement technique,
typical data obtained using the NPL extensional rheometer are presented along with results
obtained from an international intercomparison.
This Guide is aimed at those carrying out, or intending to carry out, such testing or are
building/modifying test equipment. It complements the Standard that is being developed in
ISO (ISO/CD 20965 Plastics – Determination of the transient extensional viscosity of
polymer melts) and provides additional information to the Standard that will assist in making
and analysing such measurements, and in understanding the results..
ã Crown copyright 2002
Reproduced by permission of the Controller of HMSO
ISSN 1368-6550
December 2002
National Physical Laboratory

Teddington, Middlesex, United Kingdom, TW11 0LW
www.npl.co.uk
This Guide was produced as part of the Materials Measurement programme, a programme of
underpinning research supported by the Department of Trade and Industry and disseminated
by the National Physical Laboratory.
For further information on Materials Measurement contact the Materials Enquiry Point at the
National Physical Laboratory:
Tel: 020 8943 6701
Fax: 020 8943 7160
E-mail: [email protected]
Extensional Flow Properties of Polymers using
Stretching Flow Methods
Contents
Foreword
1. Scope..................................................................................................................................1
2. Introduction ......................................................................................................................1
3. General measurement considerations ............................................................................3

3.1 Extensional flow behaviour of polymer melts ......................................................3
3.2 Extensional flow stretching methods ....................................................................3
3.3 Industrial needs .....................................................................................................4
3.4 Advantages and disadvantages of stretching flow methods ..................................5
4. Definitions .........................................................................................................................6
4.1 General ..................................................................................................................6
4.2 Elongation ratio.....................................................................................................6
4.3 Hencky strain.........................................................................................................7
4.4 Hencky strain rates ................................................................................................7
4.5 Net tensile stress....................................................................................................7
4.6 Tensile stress growth coefficient...........................................................................7
4.7 Tensile viscosity....................................................................................................8
5. Stretching flow methods ..................................................................................................8

5.1 General ..................................................................................................................8
5.2 Description of stretching flow methods ................................................................8
5.3 Limitations in using stretching flow methods .....................................................10
5.4 Measurement considerations ...............................................................................12
5.4.1 Specimen preparation..........................................................................12
5.4.2 Clamping.............................................................................................14
5.4.3 Temperature control............................................................................16
5.4.4 Specimen support................................................................................17
5.4.5 Effect of the supporting medium on the specimen properties ............18
5.4.6 Strain and strain rate ...........................................................................19
5.4.7 Stress...................................................................................................21
5.4.8 Calibration ..........................................................................................22
5.4.9 Summary.............................................................................................23
6. Measurement of extensional flow properties...............................................................25
6.1 General ................................................................................................................25
6.2 Example extensional rheometer ..........................................................................26
6.2.1 General................................................................................................26
6.2.2 Low strain rate testing.........................................................................27
6.2.3 High strain rate testing ........................................................................28
6.2.4 Measurement of strain rate using a video camera...............................28
6.3 Theory for the analysis of stretching flow data ...................................................29
6.4 Specimen preparation..........................................................................................30
6.5 Experimental procedure ......................................................................................30
7. Experimental data for polymer melts...........................................................................33

7.1 Introduction .........................................................................................................33
7.2 Typical experimental data ...................................................................................34
7.2.1 General................................................................................................34
7.2.2 Measurement of three different polymers ...........................................35
7.2.3 Measurement of three similar high density polyethylenes..................36
7.3 Low strain rate data .............................................................................................37
7.4 High strain rate data ............................................................................................37
7.5 Results of an intercomparison of stretching techniques......................................39
8. Estimate of uncertainties in extensional testing ..........................................................39

8.1 Intermediate strain rate testing ............................................................................39
8.2 Low strain rate extensional viscosity testing.......................................................40
8.3 High strain rate extensional viscosity testing ......................................................40
9. Summary .........................................................................................................................41
10. Acknowledgements.........................................................................................................42
References ...............................................................................................................................42
Figures .....................................................................................................................................45
Appendix A1 Definitions of strain, strain rate, stress and material
properties functions in tensile (simple) extension..........................59
Appendix A2 Analysis of extensional rheometer testing........................................62
Appendix A3 Analysis of the uncertainties of measurement of transient
extensional viscosity .........................................................................66
Appendix A4 Effect of viscous dissipation in the specimen during
stretching on measurements ............................................................71
Measurement Good Practice Guide No. 63
Foreword
This Guide describes methods for determining the extensional flow behaviour of plastics
melts using tensile stretching methods. It builds on and supercedes the earlier Measurement
Good Practice Guide No.18 - Extensional flow properties of polymer melts using stretching
flow methods (May 1999). In particular, it provides guidance on testing at very high strain
rates and very low strain rates; contains an improved analysis of the uncertainties in testing;
and presents results of an intercomparison of extensional flow testing. Since the original
publication of the Measurement Good Practice Guide No. 18, a draft ISO standard has been
prepared and is currently being developed in ISO (ISO/CD 20965 Plastics – Determination of
the transient extensional viscosity of polymer melts). Persons carrying out such measurements
must also refer to the latest version of that Standard. This Guide provides additional
information to the Standard that will assist in making and analysing such measurements, and
in understanding the results.
In the drafting of this Guide, it has been assumed that the collection and analysis of data will
be carried out by suitably qualified and experienced personnel.
Users of this Guide are invited to give feedback to the authors on any issues related to its use
or content.
1. Scope
This Guide:
• summarises the key issues in measuring the extensional flow behaviour of polymer
melts
• gives practical advice and defines good practice on how to obtain extensional viscosity
values of polymer melts using stretching methods, including at high and low strain rates
• presents an assessment of the effect of uncertainties on the measured extensional

viscosity values
It illustrates the measurement of extensional flow properties with reference to results

obtained using an instrument developed by NPL, and also by reference to results obtained
in an international intercomparison of extensional flow properties.
2. Introduction
In many plastics forming processes the melt undergoes predominantly extensional

deformation, for example in blow moulding, fibre spinning, film blowing and vacuum
forming [1-3]. Extensional flow also occurs in profile extrusion and injection moulding,
for example in the nozzle of the injection moulding machine and in passing from the
runner to the gate of the moulding. The extensional flow behaviour of polymer melts can
differ significantly from their shear flow behaviour. For example, the ranking of materials
in order of increasing extensional viscosity may not be the same as their ranking in order
of increasing shear viscosity. Polymers may also exhibit additional trends in extensional
flow behaviour not normally exhibited in shear, for example strain hardening. Thus
extensional flow measurements are necessary to appropriately characterise the melt
behaviour. Such data can be used for the development and selection of materials, quality
control and for process design and optimisation.
There are two basic approaches to determining extensional flow properties of polymer
melts: either tensile testing or converging flow methods. Converging flow methods were
critically reviewed and guidance on their use is presented elsewhere [4,5]. Stretching flow
methods were also critically reviewed elsewhere [6]. These reviews also identify sources
of information on testing of a range of materials.
1
The converging flow approach for determining the extensional flow behaviour of polymer
melts is based on the measurement of the resistance to flow in a contraction region. A
converging flow is, by definition, a type of extensional flow. It is expected, therefore, that
extensional viscosity might be obtained from the measurement of pressure drop and flow
rate in a converging flow. The approach of using converging flow methods is appealing as
experimental data are easily obtained from capillary extrusion rheometry testing, normally
carried out to determine the shear viscosity of materials. Thus extensional viscosity values
can be obtained from the same capillary rheometry tests as already carried out to determine
shear viscosity at little or no extra experimental cost. However, in using this approach it is
assumed that extensional viscosity is strain independent, and other significant assumptions
are necessarily made in the analysis. Consequently, there is debate about the ability of
converging flow methods to generate quantitatively accurate data, although such methods
are considered to be suitable for qualitative characterisation [5].
In comparison, stretching flow methods can be used to generate quantitatively accurate

data on the extensional viscoelasticity of polymer melts. These methods are the subject of
this Guide.
In summary, this Guide:
• discusses various issues related to the need for, and measurement of the extensional
flow properties of polymer melts (Section 3)
• defines approved terminology and presents definitions (Section 4)
• describes important considerations for stretching flow measurements (Section 5)
• describes practical considerations for measuring extensional flow behaviour, including

testing at high and low strain rates, analysis of the raw data and assessment of the effect
of uncertainties on the measured values (Sections 6 and 8)
• presents typical data obtained using the NPL extensional rheometer and from an
international intercomparison (Section 7).
Details of the analysis of experimental data and of the uncertainty analysis are presented in
the Appendices.
2
3. General measurement considerations
3.1 Extensional flow behaviour of polymer melts
In contrast to shear flows where reference is normally made only to steady shear flow
behaviour, extensional flow behaviour is best described as being transient. The extensional
viscosity may vary as a function of both strain and strain rate, particularly at low values of
strain. This transient behaviour is due to the continued development of the molecular
orientation caused by the extensional flow field over the range of strains typically
encountered in testing and processing. In comparison, in processes that are predominantly
shearing flow the shear strain values tend to be significantly higher and the process is
dominated by the material's steady state rather than transient shear flow behaviour.
In describing the transient behaviour of materials in extension they may exhibit either an
unbounded stress growth1 behaviour [for example reference 7, 8] or an equilibrium
extensional viscosity value typically at large strains [for example references 9-11] - see
also Figure 15. An equilibrium extensional viscosity is dependent on strain rate but not on
strain or time. A maximum in the tensile stress growth coefficient value was observed [8]
although this was possibly due to large experimental errors occurring at large strains,
rather than to a true material response. This topic is discussed in more detail elsewhere [6,
12].
In conclusion, to fully understand the flow behaviour of polymer melts in extension it is

necessary to characterise their transient flow behaviour.
3.2 Extensional flow stretching methods
In carrying out extensional flow measurements there are four types of measurement that are
normally made: constant strain rate, constant stress, constant force or constant speed.
Extensional techniques can be described, following Nazem et al [13], as ‘controllable’ or
‘non-controllable’ [14, 15]. ‘Controllable’ implies that the instantaneous values of strain rate
and stress are uniform throughout the specimen and that either the strain rate or stress is held
constant with time.
The results of non-controllable methods, for example fibre spinning and converging flow
analysis [16-18] tend to be difficult to interpret [4-6, 19] as they are neither constant stress
nor constant strain rate [14]. This is emphasised by Gupta et al [16] who commented that
uniaxial stretching methods are the preferred methods for obtaining fundamental data using
constant strain rate or constant stress, but not constant force or speed methods. Cogswell [20]
1 For definitions see Section 4.

3
also concluded that constant speed or constant force methods provide no information that
cannot be provided better by the ‘direct measurement’ constant stress or constant strain
methods.
3.3 Industrial needs
The industrial needs for extensional viscoelasticity testing, specifically in terms of the strains,
strain rates and temperatures were assessed by Rides [21]. In summary, it was concluded that
there was a need for measurements typically up to Hencky strains of ≈ 2 and strain rates of
up to ≈ 5 s-1 at temperatures of up to ≈ 260 oC, although higher values in both strain and
strain rate will sometimes be required, for example higher strain rates for wire coating. A
Hencky strain of 2 corresponds to a change in length by a factor of ≈ ×8.
In carrying out extensional flow measurements there are four basic categories of
measurement that can be made: constant strain rate, constant stress, constant force or
constant speed. However, industrial processes can rarely, if at all, be described by a single
one of these categories. For example although film blowing may be carried out at constant
bubble pressure and constant nip (or wind-up) tension, modelling of the process is
complex [3] and it may not be possible to approximate it as a constant force process. Thus,
in characterising the material’s flow behaviour the measurements should preferably be of
true material properties, i.e. using constant strain rate or constant stress, rather than yielding
results that are dependent on the measurement method, e.g. constant force or constant speed
measurements. However, constant stress or constant strain rate measurements tend to be
more difficult and expensive to make, and measurements that more closely mimic the
process, for example those that are constant force or speed, may be considered to be more
appropriate.
In conclusion, there is clear need to characterise the transient extensional flow behaviour
of polymer melts if one is to obtain a full understanding of their behaviour, for example for
modelling purposes. However, certain requirements, for example for quality control and
materials comparison, may be met by qualitative ‘non-controlled’ measurement methods,
rather than quantitatively accurate ‘controlled’ methods, and this may be the preferred
route.
4
3.4 Advantages and disadvantages of stretching flow methods
A summary of the main advantages and disadvantages of stretching flow methods is

presented. This represents, in part, a comparison with converging flow methods that can be
used to generate qualitative data.
• Transient behaviour
Stretching techniques can be used to characterise the transient extensional
viscoelastic behaviour of polymer melts. In comparison, converging flow
methods typically determine apparent equilibrium behaviour.
• True properties
Stretching instruments can be developed to operate in either constant strain
rate or constant stress modes thereby yielding true material properties rather
than data that are dependent on the measurement method.
• Experimental difficulty
The concept of tensile testing is simple. However, the implementation of
the method is not so simple as experimental difficulties can be
considerable. The analysis of the raw experimental data obtained using
constant strain rate methods is however relatively straight forward. In
comparison, non-controlled methods, e.g. converging flow, may be easier to
implement and may be appropriate for quality control type applications but
are more difficult, if not impossible, to interpret in terms of true materials
properties data.
• Strain and strain rate range

The maximum strain that is achievable need not be limited by the design of
the instrument. Maximum achievable strain rates are typically up to ≈ 1 s-1
although significantly higher values have been achieved using the NPL
instrument. The strain rates obtained using tensile methods are not as high
as can be achieved using converging flow methods, but the latter do not
yield a strain dependence of properties.
• Closely mimics many processing methods

It is preferable to use techniques that mimic the processes for which the
data are being sought. In so doing the testing conditions more closely match
the conditions experienced in processing, and thus the data are likely to be
more appropriate. The stretching flow method is widely applicable in the
polymer processing sector as many processes are based on stretching the
melt.
5
In selecting the appropriate technique one obviously has to consider the above advantages
and disadvantages in the light of the industrial need for the data. Where qualitative data are
required then experimentally simpler techniques such as converging flow or fibre drawing
[spinning] may be more suitable. However, where the need is for quantitatively accurate
transient extensional stress growth data then one has to resort to ‘controlled’ methods,
either constant strain rate or constant stress stretching methods. Constant strain rate
measurements tend to be more commonly carried out than constant stress measurements.
This reflects their relative ease of execution, that they are more appropriate for
determining the transient nature of the flow, and that they are more representative of
deformations occurring in processing (see Section 5.2).
This Guide specifically addresses constant strain rate methods (‘controlled’) although many
issues addressed are relevant to other stretching methods.
4. Definitions
4.1 General
The following definitions are given by Whorlow [14] for strains and strain rates, and by the
Nomenclature Committee of the Society of Rheology for start-up flow in tensile (simple)
extension at constant (Hencky) strain rate ε [Dealy 22]. Additional detail is presented in
Appendix A1. Further descriptions are given by, for example, Gupta et al [16] and Dealy
[19, 23].
In describing and modelling plastics processing the Hencky strain is preferred as the rate of
strain of an element of fluid within a flow is then independent of its original length and is
determined only from the velocity field of that element. Strain or strain rate is taken by
default herein to imply Hencky values.
4.2 Elongation ratio
Elongation ratio, ER (dimensionless): the ratio of the current length l to the initial length
l o of the specimen:
ER = l / l o (1)
6
4.3 Hencky strain
The Hencky strain ε (dimensionless) is also referred to as the natural or true strain. It is
given by the natural logarithm of the elongation ratio: where l is the specimen length (m)
and l o is the original specimen length (m):
ε = ln( l / l o) (2)
4.4 Hencky strain rates
Hencky strain rate ε (s-1): is given by
ε = 1/ l × ∂ l /∂t. (3)
where t is time. It is independent of the original length l o .
4.5 Net tensile stress
The net tensile stress σE (Pa): in tensile (simple) extension it is defined by
σE = σ11 - σ22 = σ11 - σ33 = σzz - σrr (4)
where σii is a stress tensor in rectangular or axisymmetric coordinates.
4.6 Tensile stress growth coefficient
The tensile stress growth coefficient ηE+ (Pa.s): the ratio of the net tensile stress to strain
rate
ηE+(t, ε ) = σE/ ε (5)
where t is time, and + indicates start-up of flow. The above term is a transient term and is
used to present the results of testing herein.
7
4.7 Tensile viscosity
The tensile viscosity ηE (Pa.s) is defined by
η E (t , ε ) = lim [η E+ (t , ε )] (6)
t →∞
It is the limiting tensile stress growth coefficient value and represents an equilibrium
extensional viscosity if a steady value is achieved. However for materials that do not
exhibit a steady state behaviour the use of an ‘equilibrium extensional viscosity’ such as
this is obviously not appropriate.
5. Stretching flow methods
5.1 General
This Section discusses some of the more important considerations for stretching flow
measurements of polymer melts based principally on a review of the literature [6]. It is
presented as it provides the background information for understanding the important aspects
of extensional testing, which may differ from one material to another, and thus the basis for
good measurement practice.
The methods used for tensile testing of polymer melts are described (Section 5.2) and their
limitations and the assumptions made in their use are discussed (Section 5.3). This is then
followed by a more detailed consideration of the measurement issues (Section 5.4),
specifically: preparation of test specimens (Section 5.4.1); methods used for clamping
(Section 5.4.2); methods used to control the temperature of the test specimen (Section
5.4.3); the use and effect of the supporting medium (typically silicone oil) on the polymer
(Sections 5.4.4 and 5.4.5); the control and/or measurement of strain, strain rate and stress
(Sections 5.4.6 and 5.4.7); and the calibration of instruments (Section 5.4.8). The critical
features of such instruments are then discussed in a broader context and concluding remarks
are drawn (Section 5.4.9).
5.2 Description of stretching flow methods
The basic principle behind stretching flow measurements is to subject a specimen to a

tensile stretching deformation, and through the measurement of the force and deformation
the stresses and strains and hence strain rate can be determined. The restriction to
‘controlled’ methods implies that either the strain rate or stress is kept constant during the
test. The use of haul-off or fibre spinning device on an extruder [11, 24, 25] are typically
8
‘uncontrolled’ measurements [13] and are not considered any further here. However, they are
reviewed elsewhere [6].
Cogswell [20] commented that constant stress measurements are preferred if characterisation
of the steady state behaviour is required as steady state conditions are more readily achieved
than when using constant strain rate measurements. Conversely, the constant strain rate mode
is more suited to providing information about the transient behaviour of the material. Axtell
et al [26] similarly commented that in using the controlled stress mode of operation the strain
rate and recoverable deformation attained their equilibrium values relatively quickly.
However, Axtell et al [26] added that transient data from constant strain rate tests are often
more representative of the deformations occurring in polymer processing and are therefore
more desirable. It is clear that if constant strain rate methods are to be used, then strains
greater than 4 are required to characterise fully the material including its steady state
behaviour, if it exists.
The majority of methods reviewed [6] operate in a constant strain rate mode. In these
instruments the specimen was stretched typically between either two rotating clamps [11,
27-30], or a single rotating clamp and a fixed clamp [31-34], or translating clamps (or a
single translating clamp and a fixed clamp) [11, 24, 26, 35-39, 66], Figure 1. A translating
clamp is one that moves in the direction of the axis of the specimen thereby stretching the
specimen.
The following Types are defined for convenience to describe the various techniques,
Figure 1.
Type A: Single rotating clamp and a fixed clamp.

(The rotating clamp consists of either a single or a pair of rotating
elements, either smooth or textured. When used singly the clamp is
fitted with a specimen locking mechanism.)
Type B: Two rotating clamps.

(Each clamp consists of either a single or pair of rotating elements.
Clamps may be smooth or textured.)
Type C: Ones moving (i.e. translating, non-rotating) and one fixed clamp.
Type D: Two moving (i.e. translating, non-rotating) clamps.
In summarising the current state of measurement capability, extensional flow measurements

at strains up to 7, strain rates up to 30 s-1 and temperatures up to 350 oC were reported
although these conditions were not attained simultaneously [6]. The availability of
equipment to operate at these conditions is very limited. More typically, the upper limits of
9
testing were of strains up to ≈ 4, strain rates up to ≈ 1 s-1 and temperatures up to ≈ 200 oC but
such equipment is still of limited availability.
5.3 Limitations in using stretching flow methods
The requirement for extensional measurements is for methods that provide true material
properties data at testing conditions up to high values of strain, strain rate and temperature.
The difficulties that this requirement pose and the limitations of current methods are now
addressed.
Dealy [19] commented that limitations on the maximum strain, end-effects, non-uniformities
in temperature and deformation, and the measurement of very small forces are the key
experimental problems facing the measurement of extensional properties. Similarly, Gupta et
al [16] commented that the major difficulties in using instruments for constant strain rate
measurements are due to temperature and strain rate uniformity and also gravitational effects
(for low viscosity melts). The maximum strain issue can be easily overcome by instrument
design (i.e. by use of types A and B, Figure 1), and the end-effects can be easily assessed.
However, the other factors remain key difficulties in reliable measurement.
The application of stretching techniques to the characterisation of materials is limited by the

viscosity of those materials. The lower limit of viscosity is due to the resolution of the force
measuring device, instrument friction and inertia, and the effects on the specimen of its
inertia, fluid drag, gravity and surface tension. Cogswell [35] commented that the constant
stress method is suited to melts with viscosities greater than 105 Pa.s. Rhi-Sausi et al [40]
reported that improvements to this type of instrument were made but did not quantify the
magnitude of those improvements. Gupta et al [16] commented that constant stretch rate
methods are limited to polymers having shear viscosities at the test temperature in excess of
approximately 104 Pa.s [16]. Petrie [18] commented that in making constant stress
measurements of melts having viscosities less than 104 Pa.s the difficulties associated with
instrument friction and fluid drag become significant [19]. Also, the support of the specimen
against the effect of gravity is more critical. Gupta et al [16] commented that such techniques
may not be suitable for use with nylons and polyesters that may have shear viscosities as low
as 100 Pa.s. Furthermore they [16] commented that other non-controllable techniques, for
example spinning or converging flow analysis may be more suitable for such lower viscosity
materials. In using the instrument by Meissner et al [27] extensional viscosities as low as
3 × 103 Pa.s were reported for HDPE at 150 oC, 8 × 103 Pa.s for nylons at 250 oC and 7 × 103
Pa.s for poly(ethersulfone) at 350 oC. The instrument used by Meissner et al [27] was a Type
B instrument in which the specimens were supported on a nitrogen cushion and tank-tracks,
rather than circular clamps, were used to impose the strain. Special fittings were available for
testing low viscosity materials. Thus the lower limit for testing using existing stretching
10
techniques was as low as 3 × 103 Pa.s although more commonly the value is of the order of at
least 104 Pa.s.
The most fundamental difference between Types A and B and Types C and D is that the
maximum strains achievable using Types C and D instruments are limited by the physical
sizes of the instrument and the original specimen length [19]. For Types A and B the gauge
length is constant and the maximum strain is unlimited. Furthermore, the strain rate is
proportional to the take-up speed of the clamp(s), whereas for Type C and D instruments
clamps of exponentially increasing velocity are required to produce constant strain rates in
the specimen. This also has consequences for other aspects of the measurement capability of
the instrument, in particular the maximum strain rates that can be achieved and also the level
of temperature uniformity of the specimen.
Cogswell [20] commented that translating clamp techniques (Types C and D) are limited by
the design of the instrument, typically to maximum strains of approximately 2. However,
since then strains up to a value of 4 have been achieved using such methods [30, 41]. Higher
strain values cannot simply be obtained using Type C and D devices by employing shorter
initial specimen lengths as end-effects may become more significant. This factor is addressed
in detail in Sections 5.4.2 and 7.2. Furthermore, if the length to diameter aspect ratio of the
specimen is kept the same then the magnitude of the tensile force supported by the specimen
will be reduced for shorter specimens (proportional to the cross-sectional area of the
specimen) thereby increasing errors in the measurement. To obtain higher strain values, up to
a maximum value of 7, Type A and B instruments with rotating clamps have been
successfully used [10, 27, 30,42, 43].
The limitation in the maximum strain imposed by equipment design has important
implications in terms of characterising the equilibrium behaviour of the materials,
particularly when using controlled strain rate instruments. Petrie [18], commenting on the
work by Laun et al [10], stated that the equilibrium extensional viscosity value was reached
at strains greater than 4 or if the time of elongation exceeded the greatest relaxation time of
the material. However, Meissner et al [42] observed that the specimens did not attain steady
state or equilibrium conditions even up to strains of 7 in controlled strain measurements. A
maximum in the tensile stress was observed for LDPE at a strain of approximately 5 to 6,
beyond which the stress decreased. Münstedt et al [30] commented that for constant strain
rate experiments a broad maximum was observed in stress values indicating that there was
no steady state behaviour in elongation. However, they commented that the fall in values
after the maximum may be due to inhomogeneous deformation of the specimen at the very
large strains.
At large strains the accuracy of measurement of the force supported by the specimen will
also be a limiting factor in the determination of extensional flow properties. The limit of any
measurement will be a combination of the force resolution of the instrument combined with
11
the extensional stress growth behaviour of the material. For materials that exhibit little strain
hardening the errors will be greater at relatively lower strains than for those materials that
exhibit a large strain hardening.
5.4 Measurement considerations
The following Sections contain detailed discussions of various aspects of extensional

properties measurements, which are summarised in Section 5.4.9. It is suggested that the
reader first consults the summary and then visits each of the Sections 5.4.1 to 5.4.7 as
necessary. Section 5.4.8 discusses the important issue of calibration and checking of the
instrument. These Sections provided the background to establishing good measurement
practice (Section 6).
5.4.1 Specimen preparation
It is accepted that the preparation of specimens is of critical importance for extensional

viscoelasticity measurements [8]. Problems related to specimen preparation include: non-
uniformity of the specimen dimensions, degradation of the material during specimen
preparation, stress relaxation of the specimen (due to internal stresses locked in the specimen
during its preparation) prior to and during testing, and impurities including trapped air
causing stress concentrations and possible degradation. The criticality of these factors is
dependent, however, on the material and its extensional flow behaviour.
Methods for preparing specimens were principally extrusion [9-11, 26, 28, 31, 32, 34, 37,
44-48] or transfer moulding that was predominantly carried out under vacuum [38, 49-53,
66]. Injection moulding [24, 39], compression moulding [27,54] and machining from larger
samples [26, 55] were also used. The references listed here are indicative rather than
exhaustive as details presented in the references were often insufficient. A large number of
the papers reported annealing of the specimen prior to testing to remove residual stresses.
However, Vinogradov [56] reported that annealing PS specimens made by extruding from a
capillary rheometer into water had no effect on their stress growth behaviour. Ballman [55]
used birefringence to identify whether specimens were free of stress.
Hingmann et al [37], for example, commented that specimen preparation was critical to
extensional measurements: a perfectly cylindrical shape is required for homogenous uniaxial
stretching. They prepared specimens by extruding melt into an oil bath, but reported that
voids in the specimen can cause premature failure. They commented that voids can be
minimised by controlling extrusion temperature and cooling rate as voidage is related to
crystallisation. They also commented that although voidage problems could be avoided by
using injection moulded specimens they are not suitable as they contain highly orientated
12
layers. These layers will relax on re-melting and may cause excessive distortion of the
specimen.
Schlund et al [50] indicated that specimen preparation, specifically the degree of chain
entanglement, can have a significant effect on the measured properties: the moulding time
(transfer moulding under vacuum and then annealed) affected their stress growth behaviour.
Similarly, Laun et al [11] identified that for melt spinning experiments there was a
significant effect of the geometry of the contraction region, and hence of the orientation of
the extrudate on the measured extensional properties of LDPE. Thus molecular orientation
clearly has an effect on the stress growth behaviour but is not expected to affect equilibrium
extensional viscosity values (if they exist) as they should be independent of the initial
orientation.
On the subject of impurities in specimens, for example, Meissner [8] reported that dissolved
air in PS will result in the formation of bubbles in the specimens on re-melting. These would
then cause premature failure of the specimens due to their stress concentration effect. The
specimens were therefore produced under vacuum [8]. Vacuum was used in conjunction
with transfer moulding by a number of other researchers to prepare specimens.
For methods of specimen preparation that do not cause significant flow to the polymer, e.g.
compression moulding, there is the risk that, when using granules, the structure of the
granules is not “destroyed” in the process and that premature failure will occur in the
specimen at the site of a granule boundary.
The homogeneity of deformation that is necessary, particularly for obtaining data at high
strains, is critically dependent on specimen homogeneity. Meissner et al [42] commented
that further increases in the strain to failure above a value of 7 are probably limited by
inhomogeneity of the specimen. It is noted, however, that this strain is significantly outside
the range encountered in most industrial practices.
Meissner et al [42] commented that a disadvantage of Type A compared with Type C and D
instruments is that larger specimen sizes are typically required. In testing materials that are
under development this may be an important consideration. However, the difference in
specimen size may not be significant compared with the sample size required for capillary
extrusion measurements to determine shear viscosity. It is also noted that larger specimens
sizes are potentially more susceptible to impurities and non-uniformity.
13
5.4.2 Clamping
5.4.2.1 General
Clamping the specimens is obviously essential but presents problems due to the relatively
low viscosity of the material that is being clamped (the term ‘clamped’ is used here to
indicate any method by which a force can be transmitted to the specimen). Two main
approaches are used, either fixing the specimen to the translating or rotating clamp using
either an adhesive or mechanical means (termed herein as ‘fixed clamp’, e.g. Type A, C or
D), or by drawing the specimen between two counter-rotating clamps that may be textured to
improve the friction between the specimen and the rotors (Type B). The use of either
approach has advantages and disadvantages.
5.4.2.2 Translating or single rotating clamps (Types A, C and D)
In using fixed clamps (e.g. Type A, C and D) the specimen is either mechanically clamped or
stuck to the clamp using adhesive. The use of mechanical clamping may require that the
clamp is cooled to avoid necking due to stress concentrations near the clamp [35]. However,
this has the disadvantage that it may also affect the temperature uniformity of the specimen
that is important for uniformity of the deformation of the specimen, see Section 5.4.3. The
alternative and more commonly adopted approach is to use an adhesive to stick etched
specimens to the clamps [for example reference 9]. This approach avoids the need to use
clamp cooling. Dealy [23] commented that the use of adhesives seems to produce minimum
end-effects and that epoxy adhesives appear to be suitable for polyolefins [23]. Alternative
approaches include the use of clamps similar to artist's pencil grips (i.e. collet-type clamps)
as used by Vinogradov et al [56], self-locking clamps [57, 58], and also the use of a hoop-
shaped specimen mounted over two PTFE posts so that gripping of the specimen is not
required [26]. The latter method would however generate stress concentrations at the ends of
the specimen that would result in non-uniformity of deformation. For higher viscosity
materials traditional clamping methods for solids were also used [59-61].
Meissner [62] commented that Type C and D instruments have the advantage that very small
specimens, stuck to clamps using adhesive, can be used but the deformation is
inhomogeneous and the maximum strain is limited. The use of fixed clamps (Type C and D)
introduces end-errors [10] that are more significant for specimens of small length to diameter
aspect ratio. It is also noted that the non-uniformity of the cross-sectional area of the
specimen is considered to be a major limiting factor in obtaining high strains and that larger
specimens are more likely to be non-uniform, particularly if produced by extrusion.
14
The effect of end-errors when using fixed clamps, i.e. the inhomogeneous deformation due to
the constraint of the end-plates, was studied both experimentally and theoretically for
polyisobutylene solutions by Spiegelberg et al [74]. They concluded that the end-effects are
apparent for Newtonian liquids for strains up to 2, and for viscoelastic fluids up to strains of
4. Vinogradov et al [56] observed that for PS specimens of length to diameter ratio greater
than approximately 10 there was no effect of the aspect ratio on the measured stress growth
behaviour. However, there was a significant effect when shorter specimens were used. In
comparison Cogswell [20] stated that for a length to diameter ratio of at least 5 the end-
effects will be comparable to the scatter in results, thus contradicting the observed results of
Vinogradov et al [56]. A ratio for the length to diameter of the specimen of at least 10 is thus
preferred.
Everage et al [60] used a modified Rheometrics Mechanical Spectrometer (rotational

rheometer) using its off-set drive and torque measuring system to perform extensional
measurements. This approach is appealing as it employs an existing instrument with limited
modification for extensional testing. Forced air heating was used as slumping of the
specimen was not found to be a problem, thus restricting it high viscosity materials, although
a silicone oil bath was reported to be available. However, a major difficulty with this
instrument was that a maximum specimen length of approximately 20 mm only could be
used, which could result in significant end-effect errors. Dealy [19], referring to Garfield et
al [63], reported that the end-effects resulted in non-uniform strain rates, particularly at high
strains and for low viscosity materials (see also Connelly et al [61]). Connelly et al [61]
related the degree of non-uniformity of the stretching process to the Weissenberg number
characterising the flow. Petrie [18] commented that according to Pearson (personal
communication referenced in [18]) the non-uniformity of the stretching of PS was most
significant at low strain rates and high temperatures.
Also see Section 5.4.6 for discussion on the uniformity of deformation of specimens.
5.4.2.3 Counter-rotating clamps (Type B)
The use of a counter-rotating clamp systems (for example Meissner [7]) has the advantages
that constant Hencky strain rates are achieved using constant rotational speeds of the rotors,
the total elongation of the specimen does not depend on the length of the rheometer, and the
clamps may be used for polymer melts of relatively low viscosity. The use of counter-
rotating clamps also avoids any necking of the polymer or end-effects as polymer is not
allowed to build up in the vicinity of the clamps. The deformation of the specimen is
therefore homogenous along the specimen length [20, 42]. However, the use of heavy
profiling of the counter-rotating clamps may result in differences between ‘circumferential’
and take-up speed due to yielding of the specimen between the clamps (i.e. calendering). In
one case the effect was reported to be up to 10% [42]. The counter-rotating clamp
15
arrangement of Meissner [7] was further developed [27] in which two pairs of tank tracks are
used to stretch the specimen.
Laun et al [43] commented that difficulties in using two pairs of counter-rotating clamps, due
to the need to both drive and measure the tension through at least one of the pairs, can be
avoided by using a single pair of counter-rotating clamps and a fixed clamp. Thus this
arrangement separates the drive and force measuring components of the instrument. The
experimental problems associated with the use of two pairs of counter-rotating clamps is
particularly great at low strain rates (<10-4 s-1), i.e. low tensile forces.
Münstedt et al [30] made steady state measurements of extensional viscosity of LDPE and
HDPE using both Type B, C and D instruments and observed that there was good agreement
of results between the methods. This indicates that the magnitude of the effects due to the
difficulties described above in using either fixed or counter-rotating clamps can at best be
negligible.
5.4.3 Temperature control
With the exception of four papers, all the work referenced herein using ‘controllable’
methods were at temperatures equal to or less than 200 oC. The four exceptions to this were
for testing up to: 250 oC for PA and 350 oC for PES [43], 210 oC for PP [36], 220 oC for
LLDPE and HDPE/LDPE blends [53], and 220 oC for PP [31]. All of these papers reported
the use of silicone oils as mediums except for Meissner et al [27] in which a cushion of
nitrogen was used to support the specimen. With reference to the use of oils, Muller [57]
commented that the oil must have relatively low viscosity at the test temperature to avoid
fluid resistance to the movement of the clamps and viscous drag on the specimen.
Muller [57] commented that the viscosity of polymers in the molten state (particularly for
amorphous polymers near Tg) is particularly sensitive to temperature variations. This is
particularly the case in making extensional measurements as, compared with injection
moulding, processing temperatures are typically lower and testing is therefore carried out
nearer the solid-melt transition temperature range over which the temperature sensitivity may
be particularly high. Good temperature control is therefore necessary to ensure homogeneous
deformation over the length of the specimen: localised hot-spots will result in excessive
localised deformation of the specimen and eventual failure at that position with consequent
errors in the measured values for the specimen. The magnitude of the effect of temperature
non-uniformity on the measured behaviour will be dependent on the temperature dependence
of the material and also its strain hardening characteristics.
The importance of good temperature control has been acknowledged by various researchers.
For example, to achieve good temperature control Muller et al [57] used a double oil bath
arrangement. The clamp in the bath was also separately heated to control effects due to heat
16
loss from the bath. The authors claimed to obtain temperature control to within ± 0.2 oC. In
using this vertical arrangement with an unstirred inner bath there may be problems due to
convection within the inner bath causing a temperature gradient along the length of the
specimen. Sebastian et al [64] also used a double-layer heating arrangement. Axtell et al [26]
commented that the use of an oil bath has the advantage over hot-air in that better stability is
obtainable.
Meissner et al [42] commented that temperature uniformity is essential, particularly for

determining the failure behaviour, as it influences the homogeneity of the deformation of the
specimen. They considered it to be a critical controlling parameter in testing that limits the
maximum strains that can be achieved. In earlier work a glass cover was used to improve
temperature uniformity within the oil bath. In subsequent developments a copper trough and
cover were used to further improve the temperature uniformity. The motor was also placed in
the silicone oil to further improve on the temperature homogeneity that was reported to be
better than ± 0.1 oC at 150 oC [42]. With these improvements strains of up to 7 were
obtained. Meissner et al [42] commented that further increases in the strain to failure (> 7)
were probably limited due to inhomogeneity in the specimen dimensions.
As with all rheological testing, adiabatic heating - that is heat generated due to working the
specimen - can be significant. This is particularly the case with capillary extrusion rheometry
[65]. Thus in melt spinning the effect of viscous dissipation in the extruder on temperature
control may potentially be significant. However, as discussed earlier this technique can only
be assumed to provide qualitative rather than quantitative values. Whorlow [14] commented
that for stretching techniques uniaxial extensional measurements are limited to strain rates of
approximately 10 s-1 due to adiabatic heating. Laun et al [43] estimated that for a LDPE the
adiabatic temperature increase at a strain rate of 3 s-1 and for strains less than 4 would be
smaller than 1 oC. They concluded that this would have a negligible effect on measurements.
However this assumes that the deformation and thus the adiabatic heating is homogeneous.
Where this is not the case the excessive localised deformation will generate a greater heating
effect thus resulting in even greater localised deformation and likely specimen failure.
5.4.4 Specimen support
Apart from providing a controlled temperature environment the test medium must also
provide sufficient support for the specimens otherwise the specimens will sag under their
own weight. Silicone oil was used almost exclusively with the exception of when testing was
carried out in air [e.g. 24, 60, 61, 65] or using a nitrogen cushion [27]. Specimens were either
mounted vertically in the oil [e.g. 9, 49] or horizontally either submerged in or on the surface
of the oil [e.g. 7, 10]. Utracki et al [38] commented that for low viscosity melts it is
important to minimise buoyancy or gravitational effects by matching the densities of the
polymer and the medium. They commented that for materials of viscosity higher than
17
107 Pa.s this is not a problem. However, for thermoplastics at typical processing temperatures
it cannot be ignored and the use of an oil bath or other mechanism to support the specimen is
required [19] (Cogswell commented when the viscosity < 106 Pa.s [20]). When specimens
are mounted vertically or horizontally in the oil it is necessary to match the density of the oil
with that of the specimen (both at the test temperature) so that deformation of the specimen
prior to and during the test due to gravitational or buoyancy forces is minimal. When the
specimen is floated on the surface of the silicone oil it is not necessary to match exactly the
densities but the oil must be at least as dense as the specimen. However, floating the
specimen on the oil may produce additional problems with temperature control.
5.4.5 Effect of the supporting medium on the specimen properties
A number of researchers using silicone oil as a test medium have investigated the effect of
the oil on the properties of the polymer under investigation.
Hingmann [37] reported on the absorption of oil by the specimens, determined by weight
measurements, during extensional testing. For PP it was found to increase by approximately
2% with the absorption being in a surface layer of ≈ 300 µm thickness. However, they
reported that there was no effect on measured properties. They also identified that 6% oil
content had no effect on flow properties [37].
Lanfray et al [36] commented that for PP and PS the specimens showed no evidence of
swelling or change during the time of the experiments, as observed through low strain rate
(Trouton ratio) measurements and also dynamic shear measurements.
In testing of glass fibre filled PP, Kamal et al [66] subjected the specimens to severe regimes
of 24 hours at 200 oC in oil and measured a weight loss of less than 2% in Dow-1265 but a
2% increase in 15 minutes in Dow 210H. The latter exhibited swelling which was then
followed by a weight loss indicating degradation of the polymer. The different silicone oils
were used to match the different sample densities. However the authors commented, on the
basis of the ratio of extensional viscosity and zero shear viscosity results and also on the
superposition of stress growth results, that there was no appreciable effect of the oil on
extensional viscosities.
Münstedt [9] reported that for PS after 2 hours exposure at 160 oC no silicone oil penetration
was identified using IR spectroscopy. Furthermore there was no indication of oil penetration
from zero shear viscosity measurements that are reportedly sensitive to penetration of low
molecular components into the melt and thermal degradation. Similarly for LLDPE and
LDPE, Schlund et al [50] concluded that silicone oil had no effect on results.
Thus although silicone oil was reported to be absorbed by PP, there was no reported effect on
properties. However, there appears to be no comment on the failure behaviour of the
18
polymers. Schlund et al [50], referring to work of Kurbanaliev [67] and Utracki [68],
commented that the medium used in testing could have a significant effect on the ultimate
properties of the melt in extension, specifically the maximum stress and strain at break for
1,2-polybutadiene and HDPE, LDPE and LLDPE. However the various mediums used by
Kurbanaliev [67] were alcohols and the findings may not apply to the use of silicone oils. In
the work presented by Schlund et al [50] it was shown that there was no effect within
30 minutes at 150 oC for LLDPE. This does highlight, however, that it is important to assess
the effect of the medium on properties at failure in addition to the low strain behaviour.
Utracki et al [53] commented that testing at low strain rates ( ε < 0.01) may not be limited by
the true melt strength of the specimen but by the test duration and thermal stability of the
sample and degradation effects of the silicone oil (the duration of thermal stability was
identified by dynamic viscosity measurements). At such low rates testing times increase
dramatically, for example to test at a strain rate of 0.01 s-1 to a strain of 6 will take
10 minutes. Apart from degradation effects Laun [43] observed differences in behaviour of
normal and stabilised LDPE indicating that cross-linking of the normal LDPE was occurring
at 150 oC.
Meissner [62] commented that future developments of the uniaxial and multiaxial
extensional rheometers should include eliminating the use of a supporting liquid and
increasing the maximum test temperature. The approach by Meissner et al [27] in using a
nitrogen cushion has the benefit that higher test temperatures, not limited by the thermal
stability of the test medium (usually silicone oil), can be obtained and also that materials
affected by silicone oils can be tested. However the potential for degradation effects and
further cross-linking must still be assessed. A considerable benefit in using an oil bath is that
heating times are small due to efficient heat transfer from the oil to the specimen.
5.4.6 Strain and strain rate
The predominant problem with Type C and D instruments is that the strain is limited by the
length of the instrument, and to obtain a constant strain rate the clamp separation speed needs
to be increased exponentially [57]. Rhi-Sausi et al [40] commented that strain rates above 0.1
s-1 are difficult to achieve using Type C and D instruments with screw-thread drive as high
clamp speeds are required. However, values as high as 1 s-1 were obtained by Schlund et al
[50] using a rotary motor as a winding device for a cable connected to the clamp, and 2 s-1 by
Lanfray et al [36] and also by Lacaze et al [69] using twin screw-driven clamps. Typically
the largest strains and strain rates obtained using Type C and D instruments were
approximately 4 and 2 s-1 respectively. In comparison, using Type A and B instruments
strains up to 7 were achieved by both Meissner et al [42] and Raible et al [29], and strain
rates of up to 30 s-1 were achieved by Münstedt et al [30], 10 s-1 by Laun [43] and 1.5 s-1 by
19
Ide et al [45]. Measurements carried out by Münstedt et al [30] at 30 s-1 were made in
approximately 0.1 seconds.
The use of a cam loading system to generate controlled stress tests limits the maximum
strains that can be achieved (the cam being profiled to reduce the force linearly with the
reduction in cross-sectional area of the specimen as it stretches). Larger strains require larger
cams that in turn results in a large inertia component [10] and increased frictional problems
thus increasing errors in the determination of both the stress and strain rate values. These
factors become more significant at higher strains as the tensile force supported by the
specimen decreases. These problems can be partially overcome by using two cams in series
[10]. This system was shown to produce stress values deviating by no more than 5% over the
range of deformations up to a strain of approximately 3.5 [10]. Various cam designs are
described by Whorlow [14]. An additional practical problem associated with the use of cams
is that they are designed for specimens of a specific length. Additional sources of error were
investigated by various researchers. In constant strain rate testing, inhomogeneity of
deformation due to the various reasons described previously may result in the actual strain
rates and therefore strains differing from the set values. Correction to the strain rates of the
order of 2% to 3% on the basis of optical measurements were made by Tanaka et al [44].
Gupta et al [16] commented that the specimen needs to be accelerated from rest to a finite
velocity instantaneously at the start of the test if the deformation is to be truly constant strain
rate. However, due to the inertia of the instrument [16, 19] and the response time of the
control system this will not be the case. Consequently, for controlled strain rate
measurements at short times and at high rates the strain rate is unlikely to be constant.
Ide et al [45] reported that they had made density-temperature corrections to the cross-
sectional area of the specimens to obtain the true specimen diameters at the test temperature.
It is important to know the initial diameter of the specimen at the test temperature. The
magnitude of the correction made by Ide et al [45] ranged from 16% for HDPE and PP, 12%
for LDPE and 5% for PS and PMMA.
Laun [70] briefly discussed the effect of instrument compliance on recoverable strain
measurements indicating that the compliance will result in the strain recoil being
overestimated, and the effect of machine inertia/control will result in underestimation of the
recoverable strain in stress relaxation. Instrument compliance may also affect the
measurement of strain and therefore strain rate, in particular when testing at high forces.
Consideration should be given to ensuring that the compliance of the instrument (primarily
the displacement of the force measuring device, on application of a force, affecting the
effective specimen length) is negligible in comparison with the length of the specimen. This
can effectively be checked by monitoring of the accuracy of deformation during the test by
optical means (see below and Section 6.2.4).
The uniformity of deformation can be assessed, as carried out by Meissner [8], by cutting the
stretched specimen into several sections at the end of the test. The homogeneity of the
20
deformation can then be determined by measuring the weight of each section. Optical
methods have also been employed to check on the uniformity of deformation [e.g. 32, 44, 61,
71].
See also Section 5.4.2 on clamping and its effect on uniformity of deformation.
5.4.7 Stress
The tensile force in a specimen undergoing extension will be a function of the specimen's
initial cross sectional area, strain, strain rate, the material and its extensional stress growth
characteristics (including temperature dependence). The consequence of this is that a wide
range of forces will be encountered in measurements. At high strains, e.g. ε = 6
corresponding to an elongational ratio of ≈ 400, the force becomes very small due to the
reduction in cross sectional area of the specimen [10]. The accuracy of the measurement of
extensional properties is then limited by the resolution and stability of the force measuring
device (referred to below as ‘transducer’) and also by the friction of the instrument and other
factors discussed below. The decrease in force makes force transducer selection difficult but
critically important to the accurate measurement of properties, particularly at high strains [19,
72]. Force transducers need to be of the appropriate range and also have good resolution and
stability.
Instruments reviewed elsewhere [6] had maximum force ranges up to 10 N [9] although
measurements of the order of 0.03 N peak force were identified for LDPE at ε = 0.005 s-1
and 150 oC [42], and 0.002 N for LDPE at ε = 0.01 s-1 and 160 oC [45]. Quoted values of
the resolution of stretching instruments used for testing polymer melts were of the order of
10-3 N [9, 27] and 10-4 N [43]. Tensile forces were measured either using leaf springs
combined with displacement transducers [e.g. 8, 27, 43] or strain gauges [e.g. 64], load cells
[e.g. 9, 45], or the force was applied using cam loading systems [e.g. 10, 35, 73]. Sebastian et
al [64] commented that immersion of the strain gauges in the oil bath resulted in significantly
improved voltage stability in comparison with mounting in ambient.
The effect of errors on measurements due to instrument friction, interfacial tension, viscous
or frictional drag and inertia of the specimen were investigated by various researchers.
Friction of bearings can limit the lower range of instruments to materials with viscosities
greater than approximately 104 Pa.s [73] or 105 Pa.s [19, 20]. Ishizuka et al [31] commented
that frictional forces can become significant at high strains (estimated at ≈ 10% for ε = 4,
ε = 0.6 s-1) as the tensile force on the specimen decreases considerably with increased strain.
Laun et al [43] showed, using a combined experimental and theoretical approach, that
interfacial tension effects on tensile stress measurements of LDPE at 150 oC (using silicone
oil) were negligible at strain rates above 10-4 s-1, but were dominant for strain rates less than
21
10-5 s-1. A method by which a correction for interfacial effects can be made was presented.
Interfacial tension will be more significant when using translating clamp (Type C and D)
instruments for which relative speeds between the specimen and the oil will be greater.
Ide et al [45] estimated that the effect of fluid drag on the tensile force on the specimen was
of the order of 2 - 3% for a low viscosity polymer and a silicone oil of 0.01 Pa.s. This effect
was therefore ignored in their work. Ishizuka et al [31] indicated how the effect of drag on
measurements can be taken into account.
Rhi-Sausi [40] presented evidence to indicate that inertia arising from acceleration of the
specimen was not expected to be significant in their experiments that used a moving clamp
Type C instrument. Ishizuka et al [31] concluded that the effect of inertia for their Type A
instrument was of the order of 0.02% and was therefore negligible.
5.4.8 Calibration
Each component of the measurement, e.g. force, displacement and temperature, must be
calibrated separately. The major problem in checking the calibration of an extensional
rheometer is that there are no calibrants or reference materials that could be used, except
possible limited sources established through intercomparisons [e.g. 77].
Particular problems associated with the use of such instruments will be the measurement of
the specimen's dimensions at the test temperature at the start of the test, checking the
uniformity of deformation, and the accuracy at low levels of force measurement encountered
at low and high strain values. Dealy [19] discussed some of the problems associated with
calibration of extensional rheometers. As additional means of checking the instrument, he
commented that the minimum requirement should be to check that the extensional viscosity
at very low extension rates was three times that of the zero shear viscosity (limiting shear
viscosity at zero shear rate). This is known as the Trouton ratio. However, this does not
check the instrument in the more industrially relevant high strain rate range. Dealy [19]
commented that a better verification is to compare results obtained using different
instruments, although this in itself is problematic [6, 75, also see Section 7.5].
Given the difficulty of such measurements it is essential to calibrate the individual

components of the instrument and to perform, where possible, additional checks such as
using a reference material or comparison with other instruments.
22
5.4.9 Summary
A comparison of the methods presented in the literature would indicate that the use of at least
one rotating clamp is a desirable feature of an extensional rheometer. This avoids the
restriction placed on measurements of the strain being fixed by the instrument's size rather
than by the material's response - i.e. its failure. This artificial limitation on strain would most
likely also limit the maximum strain rate that could be achieved using the instrument.
On the basis of a comparison of measurements using two different instruments the

magnitude of the effect of end-errors introduced due to using a non-rotating clamp compared
with rotating clamps was found to be negligible [30]. However, the effect of end-errors on
the uniformity of the deformation was also observed to be potentially very significant [60].
Measurements made using different length to diameter ratios for the test specimens indicated
that the length of the specimen should be at least 10 times its diameter [56].
Silicone oil was the most commonly used test medium. It provides both buoyancy for the
specimen thus reducing any sag due to gravity and also acts as a heat transfer medium. The
effect of silicone oil on the specimens was reported in several papers and although
absorption was identified there was no observed effect on the measured properties. However
the effect of the oil on the properties at failure was not reported, except for polyisobutylene
in alcohols for which there was a significant effect [67]. The effect of silicone oil absorption
on ultimate tensile properties of polymer melts requires further investigation.
The uniformity of temperature is an important factor in obtaining results at high strains.

Considerable effort has been made to minimise temperature variations, for example the use
of a copper trough [42] or double-skinned silicone oil baths [56, 58]. The use of a vertical
test set-up and a silicone oil bath may result in the development of an undesirable
temperature gradient due to density variations of the oil. Improved temperature homogeneity
would be obtained for example by using a shorter oil bath. However in reducing the
specimen length to improve on temperature uniformity the effect of any end-errors on the
homogeneity of deformation would become more significant. In order to maintain an initial
length to diameter ratio for the specimens of at least 10, specimens of small initial diameter
could be used. Shorter specimens that are homogeneous may be easier to produce but they
will also generate smaller forces (proportional to cross-sectional area) that may result in
increased measurement errors.
The selection or design of the force measuring device is clearly of considerable importance,
particularly for measurements at high strains at which the cross-sectional area of the
specimen will be considerably smaller than it was at the start of the test. Good resolution and
high accuracy at low values are important for accurate measurements at high strains.
23
To check the uniformity of deformation of the specimen and thus the quality of the test two
methods have been used. The weight of several samples cut from the deformed specimen
have been compared or the specimen's dimensions have been measured optically during the
test. The latter obviously requires the use of a transparent window, that in itself might affect
the temperature homogeneity of the specimen. However, the difficulties associated with
measuring the diameter of a filament of molten polymer in an oil bath, both of which may
well be transparent, may also result in unacceptable measurement errors.
The magnitude of the effect of various errors, for example due to frictional drag of the
supporting medium were discussed (Section 5.4). Methods for correcting for the effect of
these errors were referenced.
In summarising the current state of extensional measurement capability, flow measurements

at strains up to 7, strain rates up to 30 s-1 and temperatures up to 350 oC have been reported
although these conditions were not attained simultaneously. The availability of equipment to
operate at these conditions is very limited. More typically, the upper limits of testing were of
strains up to ≈ 4, strain rates up to ≈ 1 s-1 and temperatures up to ≈ 200 oC but such
equipment is still of limited availability.
In conclusion, the more important aspects of instruments for measuring extensional

viscoelastic properties of polymer melts are summarised as follows:
• Testing should be carried out using either constant strain rate or constant stress
deformations in order to generate quantitative data that are independent of the test
method (Section 3).
• The use of at least one rotating clamp overcomes the restriction that the maximum
strain is limited by the length of the test equipment (Section 5.4.2).
• When using non-rotating clamps the effect of end-errors on the uniformity of

deformation can be significant. The specimen length to diameter ratio should be at
least 10 (Section 5.4.2).
• The attachment of etched specimens to clamps using adhesives (typically epoxy) has
been found to be generally suitable (Section 5.4.2).
• Silicone oils have been commonly used and have been found to be suitable for use to
support specimens and to act as heat transfer media, with no effect on extensional
flow properties reported. However, there was no evidence in the literature of the
effect of silicone oil on the specimen’s properties at failure. A forced gas heating
systems has also been found to be suitable, but the use of a silicone oil bath may
permit more rapid heating of the specimen (Sections 5.4.4 and 5.4.5).
24
• The homogeneity of specimens, in terms of their dimensions and also purity, affects
the uniformity of their deformation. It is essential that this factor is controlled in order
to obtain reliable data (Section 5.4.1).
• Temperature uniformity is essential for uniform deformation of the test specimen. It

is essential that this factor is controlled in order to obtain reliable data (Section
5.4.3).
• The accurate measurement of force over a wide range is critical to the reliable
determination of extensional properties over a wide range strains, in particular at very
low and very high strains (Section 5.4.7).
• It is desirable to check the uniformity of deformation of specimens to assess the

quality of each test (Section 5.4.6).
The relative importance of these factors and others discussed elsewhere in Section 5 is
likely to differ from one material to another so general guidance is not possible. However,
the uniformity of the specimen and test temperature are recognised as being critical to
good measurements, but this should not undermine the importance of the other test
parameters (e.g. force and strain rate) and their effect on the quality of the results.
Quantification of some of these factors, i.e. the permitted tolerances, is provided in the
latest version of the standard ISO/CD 20965 Plastics – Determination of the transient
extensional viscosity of polymer melts, to which the reader is referred.
6. Measurement of extensional flow properties
6.1 General
Section 6 is aimed at illustrating various aspects of the measurement of the extensional

properties of polymers using stretching methods, with reference to the NPL instrument and
results obtained with it. It describes:
• the instrument (Section 6.2) and outlines the procedures adopted for using it to measure
the extensional viscoelasticity of molten polymers (Sections 6.4 and 6.5),
• the analysis for interpreting the experimental data (Section 6.3), and
An assessment of the effect of uncertainties experimental data on the derived extensional

viscosity values is presented in Section 8.
25
Many of the issues and analyses for this type of instrument are valid for the other
controlled stain rate and controlled stress rheometers (Figure 1).
Use of this Guide must be in conjunction with the latest version of the Standard ISO/CD
20965 Plastics – Determination of the transient extensional viscosity of polymer melts.
6.2 Example extensional rheometer
6.2.1 General
An instrument for measuring the extensional viscoelasticity of polymer melts over a wide
range of strain rates, developed by NPL, is shown in Figure 2. The operating principle of the
instrument is to stretch a molten specimen between a rotating clamp and a fixed clamp (Type
A instrument, Figure 1).
In this instrument the specimen deformation is achieved by the use of a rotating clamp that,
by winding up the specimen, controlled the rate at which the specimen was stretched. The
clamp’s rotational position, and hence speed is controlled using a servo motor. In using a
rotating clamp, measurements at constant strain rate can be achieved simply by using a
constant rotation speed. The measurements are therefore ‘controlled’ [13]. Furthermore, the
use of a rotating clamp results in an instrument that is not limited in its maximum strain
capability. For an instrument that uses translating clamps the maximum strain that can be
achieved is limited by the dimensions of the instrument. The rotating drum is raised during
the test to prevent overlap of the specimen on winding more than once around the drum.
The drum was roughened to prevent slip of the specimen occurring at the surface.
The fixed clamp is used to measure the tensile force developed in the specimen during
stretching. This clamp is mounted on a leaf spring, the displacement of which is measured
using a non-contacting optical transducer. The force is then calculated from the force -
displacement characteristics of the leaf spring, determined through calibration using known
weights. To improve the stability of the force measuring device compressed air was bled
over it to prevent contamination of the optical system through condensation of silicone oil
vapour. This was found to be effective and very desirable when testing at very low strain
rates, and thus long test durations, and resulted in reduced drift in the sensor. To obtain
different force sensor ranges and thus sensitivities, which is desirable for testing materials
of different properties or at extremes of strain rate, leaf springs of different thickness were
used: e.g. a thicker leaf spring for high strain rate measurements. Thus the performance of
the force sensor was tuned to the measurement requirements.
The specimens used were typically either approximately 20 mm or 80 mm in length and

3 mm in diameter. Shorter specimens of smaller diameter could be used, provided the
length to diameter aspect ratio of at least 10 is maintained, but this obviously reduces the
26
magnitude of the tensile forces measured during testing. Two drum diameters of 40 mm
and 80 mm were also available. By using the larger diameter winding drum and short
specimen length, high strain rates up to a nominal value of ≈ 100 s-1 have been achieved
using this instrument. Conversely, by using the smaller diameter winding drum and longer
specimen length, low strain rates down to ≈ 0.001 s-1 have been achieved.
Testing has been carried out at up to 230 °C. The instrument has a maximum operating
temperature of ≈ 260 °C, limited by the maximum operating temperature of the silicone oil
into which the specimen is immersed.
6.2.2 Low strain rate testing
Particular difficulties that had to be addressed in developing the equipment to operate at

low strain rates related to the control of specimen temperature and the measurement of low
tensile forces. The heating system had to be developed to enable the temperature of the
specimen to be controlled for the duration of the test. This was achieved by using a second
oil bath to feed a heater coil in the test bath thereby providing heat input to counter the loss
of heat from that bath to the surrounding environment. This solution was adopted as it
provided the heat input to the test bath without causing excessive vibrational noise or
excessive stirring of the silicone oil in the bath, which would otherwise swamp the
measured force signal.
In testing at low strain rates, the tensile force generated in the specimen due to its
deformation is very low. A thinner and thus less stiff leaf spring was used for testing at
low rates to improve the accuracy of measurements. Also, improvement to the procedure
for using the non-contacting force measuring device enabled the drift in the sensor due to
contamination to be minimised. This resulted in greater confidence in defining the base
line for the sensor which, because of the lower forces and longer durations of testing,
became much more significant. Testing to failure, and thus testing to very long times, was
important as it was necessary to determine a base line which was subsequently used to
calculate the transient extensional viscosity values. As an indication of the maximum test
duration, for materials that failed at a strain of three and were tested at 0.001 s-1 the test
duration was 3000 s, or 50 minutes. Testing at low rates also presented problems for
logging where relatively high logging rates were required at the start of the test, to capture
the transients at the start, yet needed to continue for some considerable time thereafter.
27
6.2.3 High strain rate testing
High strain rate testing puts particularly strenuous demands on extensional measurement
instrumentation in terms of the high speeds required and very short test durations. To
obtain true tensile stress growth coefficient data one should have constant rate deformation
of the specimen from the start of the test. For very high rate testing in particular this is not
possible due to the inertia of the instrument and the very short test durations.
The performance of the motor over the first revolution was predicted for the NPL
rheometer using the specification data for the drive motor and was also measured in
calibration tests to determine its true response, Figure 3. The specification for the
acceleration of the servo motor was 250 rev/s2. Higher accelerations were theoretically
possible using the motor, but were not achievable whilst keeping the motor under
programmed control. The results demonstrated that the motor’s performance was slightly
worse than that predicted, presumably due to the added inertia elements of the specimen,
drum and drum support. For tests at strain rates up to approximately 50 s-1 the error in time
(and thus average rotation speed) taken to complete one revolution from a standing start
was less than 5%. For the second rotation from start-up the error between the set and
measured speeds was negligible.
In the more extreme case of testing at 100 s-1, for tests up to a strain of 4 the predicted
error in the time taken was approximately 30%. For shorter tests up to a strain of 3 this
value increased to approximately 35%. These values fall below 10% for tests at a strain
rate of approximately 50 s-1. The strain at which the velocity reaches the set value for a test
at 100 s-1 is estimated to be approximately 1.6, and 0.4 at 50 s-1. Thus, as specimens
normally failed at greater strain values than these, the strain rates at failure were equivalent
to those set. Also, as a first approximation, only correction for the strain/time error need be
applied for tests above approximately 50 s-1 as the magnitude of the errors are considered
not to be too significant below that strain rate. For strain rates greater than 50 s-1 the
magnitude of the correction can be estimated using linear interpolation of the above
information.
For testing at high strain rates the data logging facilities captured data at high sampling
rates up to 2000 per second. At 100 s-1 it would take 0.04 s to reach a strain of 4, thus
giving 80 data points to define the shape of the test curve.
6.2.4 Measurement of strain rate using a video camera
It is desirable to check that the specimens deform as anticipated, given the set speeds of
the instrument. For this purpose a video camera was used to record and hence measure the
diameter of the specimen during testing. This information permitted the strains in the
specimen during testing to be determined, and hence also the strain rates. This has
28
demonstrated that the instrument performs well, with measured strain rates being
approximately 95% of the set strain rates over a range up to 25 s-1, Figure 4. Evaluation at
higher strain rates was not possible as a shorter sample length of 20 mm length was used
and it was not physically possible to get the camera into a suitable position to record the
test.
All strain rates quoted below are nominal strain rates, without correction for acceleration
and other effects (see Section 6.2.2). The implications of these errors for determination of
extensional viscosity values are considered later.
6.3 Theory for the analysis of stretching flow data
The theory for the interpretation of the transient experimental data generated by the
various types of constant strain rate instrument (Figure 1) is given in full in Appendix A2.
In summary, for an instrument (Type A) with one fixed clamp and a rotating clamp of
radius r rotating at ω rad/s with a specimen of length l o between the fixed and rotating
clamps then the strain rate is constant for a fixed rotation speed.
The strain rate ε is given by
ωr
ε& = (7)
lo
The strain ε is simply determined as the integral of strain rate with respect to time t and, as
strain rate is constant, is given by
ω rt
ε= (8)
lo
The cross-sectional area of the specimen is determined from the original cross-sectional area
Ao of the specimen and its strain. Assuming conservation of volume on stretching, the
transient tensile stress growth coefficient ηE+ is given by
æ ω rt ö
çç ÷÷
è lo ø
F lo e
ηE + = (9)
Ao ω r
where F is the force supported by the specimen. Equivalent expressions for other geometry
types (see Figure 1) are presented in Appendix 2.
29
6.4 Specimen preparation
In carrying out specimen preparation, the operator is referred to the latest version of the
standard ISO/CD 20965 Plastics – Determination of the transient extensional viscosity of
polymer melts.
Specimens were prepared by extrusion using a capillary extrusion rheometer, as the

preferred method, taking care to avoid entrapped air and extruding relatively slowly.
Specimen lengths were selected from the extrudate to minimise variations in the diameter
of the specimens and to avoid any defects that were visible. Typically the prepared
specimens were a maximum of 100 mm in length and approximately 3 mm in diameter.
Alternative methods of preparing specimens and the potential problems with these
methods are discussed in Section 5.4.1.
The specimens were attached to small end-clamps before they were mounted onto the
extensional rheometer. This approach enabled many specimens to be prepared at a time. The
specimens were fixed to the end-clamps using an epoxy adhesive as follows. The ends of the
specimen were treated by passing them through a butane flame. Care was taken to prevent
any other part of the specimen, except that that was to be stuck, from being exposed to the
flame. The ends were then dipped into the adhesive and attached to the end clamps. The
specimen and end-clamps were then placed into an oven and the adhesive cured using a
suitable time-temperature cycle. For the materials tested so far an epoxy adhesive (Ciba
AV119) has proved to be suitable, with a cure cycle of 100 oC for 1 hour. The specimens
were then allowed to cool before handling.
Normally the above procedure was found to be adequate. However, improved adhesion could
potentially be obtained by dipping the ends, after flaming, into concentrated sulphuric acid
for 30 seconds. However, care should be taken to prevent any other part of the specimen,
except that to be stuck, from being exposed to the acid.
6.5 Experimental procedure
In summary, the following experimental procedure was developed for constant strain rate
instruments. Many of the points are generic to testing using the various “types” of
rheometer (see Figure 1), although some are stages are specific to the instrument
developed by NPL. In carrying out experiments, the operator is referred to the latest
version of the standard ISO/CD 20965 Plastics – Determination of the transient
extensional viscosity of polymer melts, for specific details on testing such as the permitted
tolerances on specimen dimensions, force measurement, strain and strain rate
determination and variations in temperature along the specimen. Such values for the
current version of the Standard are presented below. Use of this Guide must be in
conjunction with the draft Standard.
30
Measure the diameter of the test specimen at at least five positions along its length and
repeat after rotating the specimen by 90o. An average value for the diameter was then
calculated from these ten measurements. For short specimens of 20 mm length, five
measurements along the length may not be practically possible using a hand-held
micrometer, in which case measurements should preferably be made at at least three
positions.
Clamp the specimen in place on the rheometer between the drum and the leaf spring using
the end-clamps stuck to the specimen.
Measure the length of the specimen between the clamps before immersing the equipment
into the oil bath.
A specimen length to diameter aspect ratios greater than 10 should be used. Specimens
typically had a length to diameter ratio of ≈ 30. However, a longer specimen will result in
a reduction in the maximum strain rates that can be achieved. The magnitude of the end-
errors can be assessed by using specimens of different length or diameter to produce
different length to diameter aspect ratios. The effect on measured values can then be
assessed.
After loading the specimen into the instrument, immerse the specimen into the silicone oil
bath that had previously been allowed to reach equilibrium at the test temperature. The use of
the oil bath provides temperature control and also support for the specimen during testing.
The oil should be selected to provide the necessary temperature range for testing but also
should be of a density that is similar to that of the materials to be testing thereby optimising
the specimen support.
Measure the temperature in at least two positions along the length of the specimen to
monitor temperature uniformity. The temperature should be measured near the specimen
without contacting the specimen.
Allow the specimen to reach thermal equilibrium. Four minutes immersion time for testing
at 150 oC was found to be suitable for specimens of 3 mm in diameter. The adequacy of
the time allowed for the specimen to reach thermal equilibrium and the effect of the
silicone oil on the specimen can be checked by varying the time for which the specimen is
immersed in the oil bath before testing. The effect on measured values, obtained using
different immersion times, can then be assessed.
In addition to the adequacy of time to reach thermal equilibrium, materials may be

susceptible to degradation or further cross-linking and this may affect results. Such
behaviour can be readily identified using an oscillatory rheometer, for example.
31
Peak forces measured in testing were up to ≈ 2 N for specimens approximately 3 mm in

diameter. Peak forces for any configuration can be estimated as described in Appendix A3
(Equation A3.24) using data of the type presented, for example, in Figure 10. This will
enable the operator to optimise the experimental set-up (where possible) to minimise
errors, particularly in the measurement of force.
A correction for thermal expansion of the specimen diameter during heating may be made.
Measurements of the density of a HDPE (HCA000) indicated a 20% decrease in density on
heating from 25 °C to 150 °C and thus a corresponding increase in cross-sectional area,
assuming the length is unchanged (due to internal stresses).
The diameter of the specimen may also be measured during the test by use of optical or
cutting methods to derive true strains and strain rates and to assess the uniformity and
accuracy of deformation. Obviously the cutting method results in the test being terminated
once the cuts have been made and prevents data to failure from being obtained, and is thus
not preferred. Such a check will also check for the effect of the compliance of the
instrument on measured values.
Repeat testing is desirable, particularly on materials for which the operator has no previous
experience, to establish the repeatability of the measurements and, in particular, that of the
point of specimen failure. Repeat measurements do not, however, give an indication of the
reproducibility and accuracy of the results.
When testing at very high strain rates correction or allowance for the inertia of the
instrument may be necessary.
In accordance with the latest version of the Standard, ISO/DIS 20965 Plastics –
Determination of the transient extensional viscosity of polymer melts:
• the diameter or width and thickness of the specimen, as appropriate, shall each be
determined to and be uniform to within ± 2% of their average value,
• the length of the specimen between the clamps shall be determined to within 1% of
its absolute value.
• the resolution of the force measuring device should preferably be no greater than
0,1%.
32
• the accuracy of force measurement shall be within ± 2% of the full scale value,
although it is desirable and preferable, in particular for accurate measurements at
low forces, that the accuracy of the force measurement device should be within
± 2% of absolute. However, this may be difficult to achieve in the lower part of the
force transducer’s range.
• the accuracy of strain determination or measurement shall be within ± 3% of the

absolute value.
• the accuracy of rate of strain determination or measurement shall be within ± 3% of

the absolute value.
• the spatial temperature uniformity shall be within ± 0,75 °C.
• the temporal temperature variation shall be within ± 1,0 °C of the set temperature.
• the temperature measuring device shall have a resolution of 0,1 °C and shall be
calibrated using a device accurate to within ± 0,1 °C.
The user of this Guide should check, using the latest version of the Standard, whether
these specifications have since been revised.
7. Experimental data for polymer melts
7.1 Introduction
To illustrate the various measurement issues, results obtained using the NPL constant
strain rate instrument are presented and cover:
• typical experimental data, including the characterisation of three different polyethylenes

(HDPE, LDPE and LLDPE) and three similar high density polyethylenes (Section 7.2)
• high strain rate experimental data (Section 7.3)
• low strain rate experimental data (Section 7.4)
Furthermore, a summary of results of an international intercomparison of stretching

techniques is also presented (Section 7.5).
33
7.2 Typical experimental data
7.2.1 General
Tests were carried out on a range of polyethylenes at 150 °C using the extensional rheometer.
Typically the specimen length was a maximum of ≈ 100 mm and the initial diameter was
≈ 3 mm. The experimental data of force versus strain are illustrated in Figure 5. As a
constant strain rate was used in testing, the strain is directly proportional to time. In this case
a strain of 1 corresponds to a time of ≈ 1 s. The trace shows that at the strain rate of 1.02 s-1
the force rose rapidly and peaked at ≈ 0.85 N in ≈ 0.6 seconds. The corresponding tensile
stress growth coefficient curve, Figure 6, shows a continual increase in value up to a strain of
≈ 3 at which point the specimen failed. A strain of 3 corresponds to a reduction in the cross-
sectional area of the specimen by a factor of ≈ ×20 and occurred within a duration of ≈ 3
seconds from the start of the test. The tensile stress growth coefficient data, Figure 6, take
into account the exponential decrease in the cross-sectional area of the specimen (see
Equations A2.4 and A2.5). The failure of the specimen is more easily observed in the tensile
stress growth coefficient data than in the force data.
These curves highlight various experimental difficulties. For example, the maximum load to
which the specimen was subjected corresponded to ≈ 0.85 N (equivalent to the force exerted
by ≈ 85 g under gravity), Figure 5. At the point of failure the load had dropped to ≈ 0.1 N.
Thus the forces measured during stretching of the specimen are relatively small, in
particularly at large strains due to the considerable reduction in the cross-sectional area of the
specimen. As the strain increases the percentage error in the measurement of force will
increase significantly with consequent effects on the errors in the derived values of the
extensional properties.
The use of different specimen length to diameter ratios in the range ≈ 7 - 30 (different lengths
but of the same diameter of ≈ 3.2 mm) indicated there was little effect at strains above 0.5,
Figure 7. However, at low strains of the order of 0.1 there was a possible correlation of
aspect ratio with measured values. The use of shorter specimens resulted in higher tensile
stress growth coefficient values, although the total variation was within approximately ±
10%. This trend is as expected as the effect of end-errors would increase for shorter
specimens.
The repeatability of measurements, for example Figure 8, is shown to be good with variation
less than approximately ± 5% for strains above 0.1.
34
Results of extensive testing of a HDPE (HGH0002) over three decades of strain rate from
0.01 s-1 to 10 s-1 are presented in Figures 9 and 10. When plotted as a function of time the
data obtained at different strain rates overlap at low times forming an apparent master curve,
Figure 10. The effect of increasing the strain rate is to shift to earlier time the point at which
the curve departs from the master curve. When plotted as a function of strain the effect of
strain rate is seen as a shift of data predominantly in the tensile stress growth coefficient axis,
with lower strain rates corresponding to higher tensile stress growth coefficient values,
Figure 9. For the strain rate of 10 s-1 it is observed that failure occurred within ≈ 0.25 seconds
from commencement of the test indicating the need for high sampling rates to obtain
adequate data, Figure 10.
Viscous heating of the specimen due to the work done in stretching, often a source of
significant experimental error in shear flow testing, has been (over)estimated on the basis
of data presented in Figure 5 and shown to be small, Appendix A4. The effect of viscous
heating on the data presented here is therefore considered to be negligible.
7.2.2 Measurement of three different polymers
Three grades of polyethylene, a linear low density polyethylene (LLDPE), a low density
polyethylene (LDPE) and a high density polyethylene (HDPE) were investigated using this
instrument. Tensile stress growth coefficient data for the three polyethylenes are presented in
Figures 11 to 14 and also combined in Figure 15 for a strain rate of ≈ 0.7 s-1 only.
These materials exhibited significant differences in their extensional strain hardening

behaviour. The LDPE exhibited greatest strain hardening with tensile stress growth
coefficient values increasing significantly with increasing strain. In comparison, the LLDPE
exhibited no further strain hardening beyond a strain of ≈ 0.5. The HDPE exhibited an
intermediate level of strain hardening. For these materials an equilibrium extensional
viscosity (or tensile viscosity) could be determined for the LLDPE but not for the HDPE or
LDPE.
It is considered that the high degree of scatter for the LLDPE data, Figure 14, is due to the
low rate, low viscosity test approaching the lower limit of the force measuring sensor used.
Both the low rate and low viscosity contribute to the low magnitude of the measured force
values.
The peak tensile stress growth coefficient values, obtained prior to failure of the
specimens, for these three polyethylenes and also for an additional HDPE (HGH000) have
been plotted as a function of strain rate and clearly indicate the rate dependence of their
behaviour, Figure 16. The behaviour of the LLDPE is relatively independent of the rate
2
NPL references to the materials, e.g. HGH000, are used to identify different grades of materials.
35
while the HDPEs and the LDPE have similar rate dependencies. The difference in tensile
stress growth coefficient values was over two decades demonstrating a significant
difference in the behaviour of these materials.
It is noted that the shear viscosities and extensional viscosities of the materials HDPE,
LDPE and LLDPE exhibit different trends. In order of decreasing shear viscosity they rank
LLDPE, LDPE and then HDPE [5], whereas in order of decreasing extensional viscosity
they rank LDPE, HDPE and then LLDPE, Figure 15. The latter trend is exhibited by both
the converging flow [5] and tensile stretching methods. This highlights the importance of
extensional flow measurements in fully characterising the flow behaviour of materials,
especially where those materials are to be used in processes in which the deformation is
predominantly extensional. The reliance on shear flow data alone can be misleading,
particularly when using that information to predict how a material will perform in a
predominantly extensional flow process.
7.2.3 Measurement of three similar high density polyethylenes
Obviously, the results presented above clearly demonstrate the importance of extensional
flow measurements. However, it does not demonstrate the capability of the method for
discriminating between similar materials. After all, the designer or processor usually has to
make a choice between similar materials rather than different material types. Results for
three similar HDPEs, each having PE100 rating and referred to as HGZ000, HHA000 and
HHB000, showed that at low shear rates there was no significant difference in their shear
viscosities. At shear rates greater than 100 s-1 melt distortion appears to have occurred,
Figure 17. These shear flow data thus provide a limited basis on which to compare the
materials. Analysis of entrance pressure drop data at the lowest rates indicated that
HHB000 had the highest values, by ≈ 50%, and HGZ000 and HHA000 had similar values
but HGZ000’s were perhaps marginally higher.
The extensional rheometer developed at NPL was used to test these materials. Testing was
carried out at three strain rates 0.1 s-1, 1 s-1 and 7.9 s-1 at 200 °C, Figure 18. At each of
these three rates the trend in extensional viscosity values was, in order of decreasing value,
HHB000, HGZ000 and HHA000. The difference between HHB000 and HGZ000 was
≈ 30%, and the difference between HGZ000 and HHA000 was ≈ 10%.
One of the reasons for wanting to characterise these materials was to understand how they
compared, and to relate their measured behaviour to their performance both in extrusion
and injection moulding. The differences apparent in the extensional flow behaviour
suggest that the material HHB000, having the highest extensional viscosity, may be more
suitable for thick-walled pipe extrusion being less prone to wall thickness variations in the
pipe.
36
Because of pipe specifications, it is necessary to use the same quality of material for the
pipe fittings as is used for the pipe. Thus the suitability of these materials for injection
moulding is also important. The results also indicate that HHB000, having the highest
extensional viscosity, may be the more problematic of the materials for injection
moulding, requiring higher pressures to force material through the gate region for example.
7.3 Low strain rate data
Low strain rate testing of various polyethylenes (HDPE, LDPE and LLDPE) has clearly
illustrated the measurement difficulties that are prevalent in such testing. Data obtained at
low strain rates are presented, for example, in Figure 19 [75]. Data at strain rates as low as
0.001 s-1 on HDPEs have since been obtained [75]. For the LDPE (HGE000) considerable
strain hardening was exhibited, Figure 19. Transient extensional stress growth coefficient
values exhibited a decade increase in magnitude on departing from the master curve3.
However, at the lowest strain rate used (0.002 s-1) no apparent strain hardening was
observed. The reason for the absence of strain hardening is unclear. It may have been due
to premature failure of the specimen. The lack of strain hardening means that the specimen
is relatively unstable and should any necking occur it is likely to fail prematurely. However
the repeat test showed almost identical behaviour at long times indicating the high level of
repeatability of the behaviour.
The peak force measured in the 0.002 s-1 strain rate test was less than 0.005 N (equivalent
to 0.5 gf), thus indicating the small forces that need to be measured. The resolution of the
force measuring device was 0.000024 N. Scatter in data due to vibrational noise, in part
from the drive system, was more apparent when testing at the lowest strain rates but could
be reduced further by modifying the filtering of the force sensor signal. These values thus
demonstrate the severe demands placed on such an instrument for use when testing at low
strain rates.
7.4 High strain rate data
Results for various polyethylenes (HDPE, LDPE and LLDPE) clearly demonstrate
measurement issues related to high strain rate testing. For the LDPE (HGE000) strain rates
up to 100 s-1 were achieved with test times less than 0.1 s, Figure 20. It is noted that strains
are the natural logarithm of the extension ratio and thus a strain rate of 100 s-1 is
significantly greater than that of 10 s-1 in terms of the rate of deformation occurring: a
strain rate of 1 s-1 corresponds to stretching the sample to 2.7 times its length per second, a
3The master curve refers to the superimposition of data obtained at different strain rates when
plotted as transient stress growth coefficient versus time. Additional strain hardening refers to the
departure of data from this master curve at high strain values.
37
rate of 10 s-1 corresponds to 22000 times its length per second, and 100 s-1 to 2.7 x1043
times its length per second. However, in practice the times for which these latter
deformation rates occur is significantly shorter than one second and thus the total strains
are also significantly shorter.
The repeatability of results is shown to be quite good, Figure 20, and even at the highest
strain rate of 100 s-1 [76]. The feature of the extensional behaviour that is hardest to achieve
repeatably is the point of failure of the specimen. It is reasonable to assume that where
repeat testing is carried out the highest strain to failure is the most reliable: any premature
failures being due to inhomogeneities in the specimens.
It was noted that the oscillatory waveform evident in some of the results was particularly
pronounced at a test rate of 40 s-1, Figures 20 and 21 [76]. This was considered to be due to
the fact that the resonant frequency of the force measuring system, which was approximately
40 Hz, corresponded with the effective rate of loading at this strain rate. The resonance
appeared to be less pronounced at 100 s-1. Data obtained in the presence of resonance was in
agreement with data obtained using a longer specimen for which resonance was not such a
significant problem [76]. The main cause for concern with the resonance is its effect on the
accuracy of the values obtained at short time scales that are comparable with the period of
oscillation of the resonance. Resonance of the force sensor is considered difficult to
overcome due to the need for a sensor of low force capability, and thus low stiffness.
However, the use of specimens of different length may provide a means to minimise the
effect.
From the repeat measurements, Figure 20, the use of the short length specimens of ≈ 20 mm
was shown not to be problematic at strains in excess of ≈ 0.5 at 25 s-1 (i.e. > 0.02 s), Figure
20 [the use of short specimens is essential for obtaining high strain rate data]. Below this
values of strain the results differ more markedly due to the very short times and low forces
that they correspond to, along with the resonance of the force sensor. Although not
conclusive, but indicated by HGE000, Figure 20 and [76], an examination of the viscosity
versus time curves showed that at times less than 0.01 s the lowest value extensional
viscosity data corresponded to that obtained at the highest rates. This is consistent with the
effects of the initial acceleration of the drum in that the forces generated will be lower as the
strains are lower than expected. As a consequence the derived extensional viscosity values
will also be lower. On this basis, when examining viscosity versus time plots for high rate
data it is more appropriate to assume that, at short times, the data obtained at the lower strain
rates are more reliable.4
4 Tensile stress growth coefficient versus time plots are found useful as data obtained at different
strain rates tend to form a master curve, with departure from that master curve occurring at high
strains due to additional strain hardening.
38
7.5 Results of an intercomparison of stretching techniques
The results of an intercomparison of stretching methods comprising 6 laboratories

indicated that the variation in tensile stress growth coefficient values was estimated to be
up to approximately ± 60%, Figures 22 and 23 [77]. Measurements were performed mainly
over a strain rate range of 0.01 s-1 to 1 s-1, and measured tensile stress growth coefficient
values were predominantly in the range 5 x 104 Pa.s to 3 x 106 Pa.s. The observed
variation in peak tensile stress growth coefficient values was up to approximately ± 100%
indicating the greater difficulty in determining these failure values. A high density
polyethylene that was stable at 190 °C for in excess of two hours was used as the
intercomparison material. Samples of the material were provided to each of the
participants who produced their own specimens. Measurements were made at 150 °C and
190 °C [77].
NOTE: The quoted variation in tensile stress growth coefficient values excludes data at
short times/low strains for which the variation is greater due to instrument inertia effects
and that the resolution of the measuring equipment may be insufficient for reliable
measurement of the low forces generated. Measurement errors will be strongly dependent
on the magnitude of the force to be measured which, in turn, will be strongly dependent on
the strain. The values quoted are indicative for the various instruments that were involved
in the intercomparison.
8. Estimate of uncertainties in extensional testing
8.1 Intermediate strain rate testing
Due to the considerable stretching that the specimen normally undergoes during testing,
particularly on reaching high strain values, there is a very significant decrease in the cross-
sectional area of the specimen, accompanied by a similarly large decrease in the measured
force. As a consequence there is a significant increase in the uncertainty in the
measurements. To understand and quantify the level of confidence that one can put in the
accuracy of the results obtained, it is important that the uncertainties in the measurements,
in particular that due to the accuracy and resolution of the force measurement, are known.
An analysis of the uncertainties of extensional measurements is presented in Appendix A3

[78]. The equations can be used to calculate the uncertainties in the derived tensile stress
growth coefficient values, given the uncertainties in each of the components of the
measurement. The analysis can thus be used to identify the critical parameters that
generate the uncertainties in the measurements and can thus be used to improve on the
measurement method.
39
The example presented, Appendix A3 and Figure 24, is based on a HDPE (HGH000) at
150 °C and a strain rate of ≈ 1 s-1 (Figure 10). It illustrates the significant increase in
measurement uncertainties at high strains due, predominantly, to the increase in the
uncertainty in the measurement of force resulting from the significant reduction in the
cross-sectional area of the specimen. The estimated uncertainties increased dramatically at
high strain values, Figure 24. Expanded uncertainties representing 95% confidence limits
are presented and were estimated to be of the order of ± 20% at a strain of 3.3, ± 50% at a
strain of 4.8 and ± 100% at a strain of 5.7.
8.2 Low strain rate extensional viscosity testing
At low strain rates the uncertainties in the low forces generated at low strain values are
potentially significant. The analysis was used to estimate the uncertainties in testing for the
HDPE (HFU000) at 0.001 s-1, Figure 25. The results show measured tensile stress growth
coefficient data along with calculated values, assuming a power-law fit, and 95%
confidence limits to the calculated data. The relative combined uncertainty is also
presented. To provide an improved analysis the strain range had been split into two
regions, that above and that below a strain value of approximately 0.4. This enabled
different power-law parameters to be used in each region, thereby more accurately fitting
both the master curve and additional strain hardening parts of the experimental data.
The principal difficulty in carrying out the uncertainty analysis for these low rate tests was
in establishing the value for the uncertainty in the force measurement at the very low
values of force involved. A value for the error in measurement of the force was determined
from the calibration curve for the force sensor. These uncertainty calculations were very
sensitive to the magnitude of that chosen value, and consequently the uncertainties in
measurement of transient stress growth coefficient values may differ quite markedly from
those plotted. However, the results for the uncertainty calculations appear to be reasonable
(if not excessively large for the lower 95% limit at low strain values) and reflect the
relatively high confidence in results of transient stress growth coefficient data at low
strains based on the fact that the low rate data lie on or near the master curve generated by
data obtained at higher strain rates, e.g. Figure 19 [75].
8.3 High strain rate extensional viscosity testing
As demonstrated earlier, the uncertainties in extensional viscosity determination grow

rapidly at large strains primarily due to the rapid decrease in cross-sectional area of the
specimen with a resultant rapid decrease in the measured force. At high strain rates
acceleration issues also become significant, as discussed in Section 6.2.3. Due to the
acceleration behaviour of the instrument the strain achieved in a given time is less than
that that would be achieved if an instantaneous acceleration were assumed. Furthermore,
40
the difference in the measured to set strain rates, as observed using the video camera, has
an effect on the accuracy of the derived results. The analysis of the uncertainties in
extensional testing was used to estimate the uncertainties in testing for the HDPE
(HFU000) at 25.5 s-1 and 100 s-1, taking into account these additional errors, Figures 26
and 27. The errors in the strain rate due to the acceleration of the drum and the visual
observations has been accounted for by assigning a 15% or 40% error to the angular speed
of the drum for set strain rates of 25.5 s-1 and 100 s-1 respectively. The contribution due to
the visual observations was set at 5%, the remainder being due to the acceleration of the
drum. The plots, Figures 26 and 27, show the measured tensile stress growth coefficient
data along with calculated values, assuming a power law fit, and 95% confidence limits to
the calculated data. The results show a rapid increase in the uncertainties as strains
increases beyond a value of approximately 1, with the effect becoming more significant at
higher strain rates. The 95% confidence limits to the data are not true confidence limits
because the errors due to the acceleration and strain rates are systematic. The effect of both
the acceleration phase and the error in strain rates observed using the video camera would
be to result in lower transient stress growth coefficient values being determined than
actual. Thus the upper 95% limit gives an indication of the maximum shift in data that
would be necessary to correct for these effects, rather than being due to random
uncertainties in the various parameters.
9. Summary
Extensional flow characterisation yields information about the materials not revealed by
shear flow measurements. A comparison of extensional viscoelasticity data with shear
viscosity data [5] demonstrates that materials characterisation solely on the basis of shear
flow measurements is inadequate, particularly when the data are required for predicting the
performance of materials in processes that are predominantly extensional flows.
Guidance and recommendations for the measurement of the extensional flow properties of
polymer melts, including at very high and very low strain rates, have been presented.
Analyses to determine tensile stress growth coefficient values and an analysis of the
uncertainties in data have also been presented. The method developed by NPL has been
used to illustrate the measurement of the transient extensional flow behaviour of polymer
melts, using a range of polyethylenes, and has highlighted the measurement issues that
need to be addressed in order to obtain reliable data. Results of an international
intercomparison are also presented.
41
10. Acknowledgements
The work reported in this paper was carried out as part of a programme of underpinning
research supported by the Department of Trade and Industry on measurements related to
the processability of materials.
References
1 Han, C.D., Rheology in Polymer Processing, Academic Press, London 1976.

2 Tanner, R.I., Engineering Rheology, Clarendon Press, Oxford, 1985.
3 Pearson, J.R.A., Mechanics of Polymer Processing, Elsevier Applied Science
Publishers Ltd, London, 1985.
4 Rides, M. and Chakravorty, S., Review of converging flow methods for
determining the extensional flow behaviour of polymer melts, NPL Report
CMMT(A)80, August 1997.
5 M. Rides, M. and C.R.G. Allen, Measurement of the extensional flow properties of
polymer melts using converging flow methods, NPL Measurement Good Practice
Guide No.16, January 1999.
6 Rides, M., Allen, C.R.G., and Chakravorty, S., Review of extensional
viscoelasticity measurement techniques for polymer melts, NPL Report
CMMT(A)44, October 1996.
7 Meissner, J., Trans. Soc. Rheol., 16 (1972) pp.405-420.
8 Meissner, J., Chem Eng. Commun., 33 (1985) pp.159-180.
9 Münstedt, H., J. Rheology, 23 (1979) pp.421-436.
10 Laun, H.M., and Münstedt, H., Rheol. Acta, 15 (1976) pp.517-524.
11 Laun, H.M., and. Schuch, H., J. Rheology, 33 (1989) pp.119-175.
12 C.S. Petrie, Rheol. Acta, 34 (1995) pp.12-26.
13 F. Nazem and C.T. Hill, Trans. Soc. Rheol., 15 (1974) pp.87-.
14 Rheological Techniques, R.W. Whorlow, Ellis Horwood, London, 1992.
15 Rheometry, K. Walters, Chapman and Hall, London, 1975.
16 R.K. Gupta and T. Shridar, Elongational Rheometers, in Rheological Measurement,
Ed. A.A. Collyer and D.W. Collyer, Elsevier Applied Science, London, 1988.
17 D.M. Binding, Contraction Flows and New Theories for Estimating Extensional
Viscosity, in Techniques in Rheological Measurements, Ed. A.A. Collyer, Chapman
and Hall, London, 1993.
18 Elongational Flows, C.J.S. Petrie, Pitman, London, 1979.
19 J.M. Dealy, J. Non-Newtonian Fluid Mech., 4 (1978) pp.9-21.
20 F.N. Cogswell, Trans. Soc. Rheol., 16 (1972) pp.383-403.
21 M. Rides, Industrial need for extensional viscoelasticity measurements, NPL Report
CMMT(A)38, September 1996.
42
22 Dealy, J.M., J.Rheology 39(1) January/February 1995, pp.253-265.

23 J.M. Dealy, Polym. Eng. Sci., 11 (1971) pp.433-445.
24 J. Revenu, J. Guillet and C. Carrot, J. Rheology, 37 (1993) pp.1041-1056.
25 L.E. Bailey, D.G. Cook, J. Pronovost and A. Rudin, Polym. Eng. Sci., 34 (1994)
pp.1485-1491.
26 F.H. Axtell and B. Haworth, Polymer Testing, 9 (1990) pp.53-70.
27 J. Meissner and J. Hostettler, Rheol. Acta, 31 (1994) pp.1-21.
28 M. Takahashi, T. Isaki, T. Takigawa and T. Masuda, J. Rheology, 37 (1993) pp.827-
846.
29 T. Raible, S.E. Stephenson, J. Meissner and M.H. Wagner, J. Non-Newtonian Fluid
Mech., 11 (1982) pp.239-256.
30 H. Münstedt and H.M. Laun, Rheol. Acta, 20 (1981) pp.211-221.
31 O. Ishizuka and K. Koyama, Polymer, 21 (1980) pp.164-170.
32 T. Takahashi, H. Toda, K. Minagawa, J-I. Takimoto, K. Iwakura and K. Koyama, J.
Applied Polym. Sci., 56 (1995) pp.411-417.
33 M. Kobayashi, T. Takahashi, J. Takimoto and K. Koyama, Polymer 36 (1995)
pp.3927-3933.
34 W. Minoshima and J.L White, J. Non-Newt. Fluid Mech., 19 (1986) pp.251-274.
35 F.N. Cogswell, Plastics and Polymers, April 1968, pp.109-111.
36 Y. Lanfray and G. Marin, Rheol. Acta, 29 (1990) pp.390-399.
37 R. Hingmann and B.L. Marczinke, J. Rheology, 38, (1994) pp. 573-587.
38 L.A. Utracki and P. Sammut, Polym. Eng. Sci., 30 (1990) pp.1019-1026.
39 R. Fulchiron, V. Vernay and G. Marin, J. Non-Newt. Fluid Mech., 48 (1993) pp.49-
61.
40 J. Rhi-Sausi and J.Dealy, Polym. Eng. Sci., 16 (1976) pp.799-802.
41 P.K. Agrawal, W.K. Lee, J.M. Lorntson, C.I. Richardson, K.F. Wissbrun and A.B.
Metzner, Trans. Soc. Rheol., 21 (1977) pp.355-379.
42 J. Meissner, T. Raible and S.E Stephenson, J. Rheology, 25 (1981) pp.1-28.
43 H.M. Laun and H. Mnstedt, Rheol. Acta, 17 (1978) pp.415-425.
44 K. Tanaka, K. Koyama and S. Kurita, J. Rheology 37(4) (1993) pp.609-620.
45 Y. Ide and J.L. White, J. Applied Polym. Sci., 22 (1978) pp.1061-1079.
46 L. Li, T. Masuda and M. Takahashi, J. Rheology, 34 (1990) pp.103-116.
47 Y. Suetguga and J.L.White, J. Applied Polym. Sci., 28 (1983) pp.1481-1501.
48 P. Riha, J. Watanabe, K. Minagawa and K. Koyama, Proc. Fourth European
Rheology Conference, Sevilla, Spain, 4-9 September 1994, pp.374-376.
49 L.A. Utracki and A.M. Catani, Polym. Eng. Sci., 25 (1985) pp.690-697.
50 B. Schlund and L.A. Utracki, Polym. Eng. Sci., 27 (1987) pp.380-386.
51 D.M. Kalyon and D.S. Czerwonka, Plast. Rubber Proc. and Applications, 14 (1990)
pp.29-33.
52 B. Schlund and L.A. Utracki, Polym. Eng. Sci., 27 (1987) pp.1523-1529.
53 L.A. Utracki, M.R. Kamal and N.M. Al-Bastaki, Conference Proc., ANTEC 1984,
New Orleans, pp.417-420.
43
54 F.N. Cogswell and D.R. Moire, Polym. Eng. Sci., 14 (1974) pp.573-576.
55 R.L. Ballman, Rheol. Acta, 4 (1965) pp.137-140.
56 G.V. Vinogradov, V.D. Fikhman, B.V. Radushkevich and A.Ya. Malkin, J. Polym.
Sci., A-2, 8 (1970) pp.657-678.
57 R. Muller and D. Froelich, Polymer, 26 (1985) pp.1477-1482.
58 D. Froelich, B. Muller and Y.H. Zang, ACS, Rubber Div., 128th Meeting - Fall
Cleveland, Ohio, October 1985, paper 66, pp.23.
59 E. Kamei and S. Onogi, Appl. Polym. Symp., 27 (1975) pp.19-46.
60 A.E. Everage and R.L. Ballman, J. Applied Polym. Sci., 20 (1976) pp.1137-1141.
61 R.W. Connelly, L.G. Garfield, G.H. Pearson, J. Rheology 23(5) (1979) pp.651-662.
62 J. Meissner, Polym. Eng. Sci., 27 No. 8 (1987) pp.537-546.
63 L.G. Garfield, G.H. Pearson and R.W. Connelly, Annual meeting, Society of
Rheology, New York (1977) Abstract E15.
64 D.H. Sebastian and J.R. Dearborn, Polym. Eng. Sci., 23 (1983) pp.572-575.
65 M.R. Kamal and H Nyum, Polym. Eng. Sci., 20 (1980) pp.109-119.
66 M.R. Kamal, A.T. Mutel and L.A. Utracki, Polym. Composites, 5 (1984) pp.289-
298.
67 M.K. Kurbanaliev, G.V. Vinogradov, V.E. Dreval, A.Ya. Malkin, Polymer, 23
(1982) pp.100-104.
68 L.A. Utracki and J. Lara, Paper at the Table ronde internation. Ecoulement
élongationnels, La Bresse, France, Jan 23-28, 1983 (Ref. 17 of Schlund et al (54)).
69 J.M. Lacaze, G. Marin and Ph. Monge, Rheol. Acta, 27 (1988) pp.540-545.
70 H.M. Laun, J. Rheology, 30 No.3 (1986) pp.459-501.
71 T. Takahashi, J. Watanabe, K. Minagawa and K. Koyama, Polymer 35 (1994)
pp.5722-5727.
72 Rheometers for Molten Plastics, J.M. Dealy, Van Nostrand Reinhold Company,
London, 1982.
73 H. Münstedt, Rheol. Acta, 14 (1975) pp.1077-1088.
74 S.H. Spielgelberg, D.C. Ables and G.H. McKinley, J. Non-Newtonian Fluid
Mechanics, 64 (1996) pp.229-267.
75 Low rate extensional rheometry of polymer melts, M Rides and CRG Allen,
MATC(A)114, July 2002.
76 High rate extensional rheometry of polymer melts, M Rides and CRG Allen, NPL
Report MATC(A)71, December 2001.
77 Rides, M., Allen, C.R.G., and Chakravorty, S., Intercomparison of extensional flow
characterization techniques for polymer melts: tensile stretching and converging flow
methods, NPL Report CMMT(A)171, April 1999.
78 Rides, M., Determination of the uncertainties in transient extensional viscosity
testing, NPL Report CMMT(A)303, November 2000.
79 J. Meissner, S.E. Stephenson, A. Demarmels and P. Portman, J. Non-Newtonian
Fluid Mechanics, 11 (1982) pp.221-237.
44
Figures
ω ω
F
F
ω
ω ω
F
ω
Lo
lo
Type B
Type A
V V
F V
F
lo Lo
Type C
Type D
- Rotating clamp or angular speed ω
- Fixed or translating clamp of speed V
F - Force
Figure 1: Schematic of test instrument Types A to D.
45
Motor Camera
Displacement
Specimen transducer
clamps
Leaf spring
Rotor
Oil bath
Specimen
Figure 2: The NPL extensional rheometer showing the sample clamped between the
rotating drum and force measuring device, suspended above the silicone oil bath.
25
Measured error
Error in speed for first revolution, %
Calculated error
20
15
10
0
0.001 0.01 0.1 1 10 100 1000
-1
Strain rate, s
Figure 3: Error in rotation speed due to drum acceleration (using specimen length of
≈ 20 mm and drum radius of 40 mm).
46
30
25
Measured strain rate, 1/s
20 y = 0.950x - 0.241
R2 = 0.998
15
10
0
0 5 10 15 20 25 30
Set strain rate, 1/s
Figure 4: Comparison of set with optically measured values of strain rate for the
extensional rheometer, the measured values being determined using a video
camera.
0.9
0.8
0.7
0.6
0.5
Force, N
0.4
0.3
0.2
0.1
0.0
-1.0 0.0 1.0 2.0 3.0 4.0
-0.1
Strain
Figure 5: Extensional testing of a HDPE at a strain rate of 1.02 s-1 and 150 oC showing the
raw force versus strain data necessary for determining extensional parameters
(HGH_EXT_074).
47
6.E+05
Tensile stress growth coefficient, Pa.s
5.E+05
4.E+05
3.E+05
2.E+05
1.E+05
0.E+00
0.0 0.5 1.0 1.5 2.0 2.5 3.0
Strain
Figure 6: Transient extensional viscosity (or tensile stress growth coefficient) versus strain
plot for the same test as presented in Figure 7 (HGH_EXT_074).
1000000
100000
EXT094, 22 mm
EXT093, 24 mm
EXT088, 47 mm
EXT089, 68 mm
EXT074, 98 mm
EXT073, 101 mm
EXT075, 101 mm
10000
0.01 0.1 1 10
Time, s
Figure 7: Effect of specimen length in the range 22 mm to 101 mm on the tensile stress
growth behaviour of a HDPE (HGH000) at 150 °C and a strain rate of 1 s-1.
48
1000000
100000
EXT073, 101 mm
EXT074, 98 mm
EXT075, 101 mm
10000
0.01 0.1 1 10
Time, s
Figure 8: Effect of immersion time in silicone oil on extensional flow behaviour of a HDPE
(HGH000) at 150 °C and a strain rate of 1 s-1 (EXT073 & EXT075: 4 minutes,
EXT074: 8 minutes).
10000000
EXT077, 0.01
EXT076, 0.1
EXT073, 1
EXT074, 1
EXT075, 1
EXT088, 1
EXT089, 1
EXT093, 1
1000000 EXT094, 1
EXT079, 3
EXT080, 6.28
EXT092, 10
100000
10000
0.001 0.01 0.1 1 10
Strain
Figure 9: Effect of strain rate from 0.01 s-1 to 10 s-1 on tensile stress growth behaviour of a
HDPE (HGH000) at 150 °C.
49
10000000
1000000
EXT077, 0.01
EXT076, 0.1
EXT073, 1
EXT074, 1
EXT075, 1
EXT088, 1
100000 EXT089, 1
EXT093, 1
EXT094, 1
EXT079, 3
EXT080, 6.28
EXT092, 10
10000
0.01 0.1 1 10 100 1000
Time, s
Figure 10: Effect of strain rate from 0.01 s-1 to 10 s-1 on tensile stress growth behaviour of a
HDPE (HGH000) at 150 °C.
1E+08
HFU000 ext 043, 0.0138 1/s

1E+07 HFU000 ext040, 0.685 1/s

HFU000 ext 046, 2.69 1/s
1E+06
1E+05
1E+04
1E+03
1E+02
0.001 0.01 0.1 1 10
Strain
Figure 11: Extensional flow measurements on a HDPE (HFU000) at 150 oC plotted as a

function of strain, indicating the effect of strain rate on the flow behaviour.
50
1E+07
HFU000 ext 046, 2.69 1/s
HFU000 ext040, 0.685 1/s
HFU000 ext 043, 0.0138 1/s
1E+06
1E+05
1E+04
0.01 0.1 1 10 100 1000
Time, s
Figure 12: Extensional flow measurements on a HDPE (HFU000) at 150 °C indicating the
overlap at short times of data obtained at different strain rates.
1E+08
HGE000 ext 044, 0.0136 1/s
1E+07 HGE000 ext041, 0.707 1/s
HGE000 ext 047, 2.86 1/s
1E+06
1E+05
1E+04
1E+03
0.001 0.01 0.1 1 10
Strain
Figure 13: Transient extensional viscosities for LDPE (HGE000) at 150 °C at three strain
rates.
51
1E+08
HGF000 ext045, 0.0139 1/s

HGF000 ext042, 0.728 1/s

1E+07
HGF000 ext 048, 2.91 1/s
1E+06
1E+05
1E+04
1E+03
0.001 0.01 0.1 1 10
Strain
Figure 14: Transient extensional viscosities for LLDPE (HGF000) at 150 °C at three strain
rates.
1E+06
HGE000 ext041
HFU000 ext040
8E+05 LDPE
HGF000 ext042
6E+05
4E+05
HDPE
2E+05
LLDPE
0E+00
0 0.5 1 1.5 2 2.5 3
Strain
Figure 15: Extensional flow measurements on three polyethylenes at 150 oC exhibiting

significant differences in strain-hardening behaviour at a strain rate of ≈ 0.7 s-1.
52
1.E+08
Maximum tensile stress growth coefficient,
HGE000
LDPE
1.E+07 HGH000
HDPE
HFU000
HDPE HGF000
Pa.s
1.E+06
1.E+05 LLDPE
1.E+04
0.001 0.01 0.1 1 10 100
-1
Strain rate, s
Figure 16: Comparison of maximum tensile stress growth coefficient values obtained by a
stretching method for four polyethylenes LDPE (HGE000), LLDPE HGF000),
HDPE (HFU000) and HDPE (HGH000) at 150 °C.
10000
HHA000
HGZ000
HHB000
Shear viscosity, Pa.s
1000
100
10 100 1000 10000
-1
Shear rate, s
Figure 17: Comparison of shear viscosities of three polyethylene materials at 200 °C,
suggesting melt fracture at shear rates above 100 s-1.
53
1.E+07
Black symbols: HHB000
Blue symbols: HGZ000

Red symbols: HHA000
0.1 s-1
1.E+06
7.9 s-1
1 s-1
1.E+05
1.E+04
0.01 0.1 1 10
Strain
Figure 18: Comparison of the tensile stress growth coefficient behaviour of three high
density polyethylenes at three different strain rates at 200 °C.
10000000
Tensile stress growth coefficient, Pa.s .
Test ref, strain rate (1/s)

222, short, 1
206, 0.2
237, 0.2
1000000 238, 0.2
239, 0.02
242, 0.02
240, 0.002
241, 0.002
100000
HGE000, 150 °C
10000
0.001 0.01 0.1 1 10 100 1000
Time, s
Figure 19: Extensional viscosity data as a function of time for a LDPE (HGE000) at 150 °C.
54
10000000
HGE000
1000000
100000
10000
Test ref.
206 long 0.2

1000 205 long 1
222 short 1
204 long 3
203 long 10
100 201 short 10
202 long 25
200 short 25
199 short 40
10
198 short 100
1
0.001 0.01 0.1 1 10 100
Time, s.
Figure 20: Extensional viscosity data as a function of time for a LDPE (HGE000) at
150 °C. The legend presents: test reference, specimen length (long or short) and strain rate.
1.6
ext181, HFU000, 150
1.4
1.2
Force, N, absolute
0.8
0.6
0.4
0.2
0
0 0.5 1 1.5 2 2.5 3 3.5 4
Strain, absolute
Figure 21: Resonance of the force measuring system as observed when measuring the
extensional viscosity of a HDPE (HFU000) at 150 °C. Strain rate = 40 s-1, thus a
strain of 1 corresponds to a time interval of 0.025 seconds.
55
1.E+07
Extensional stress growth coefficient, Pa.s
1.E+06
1.E+05
1.E+04
1.E+03
0.001 0.01 0.1 1 10 100 1000
Time, s
Figure 22: Intercomparison of stretching flow methods for a HDPE (HGH000) at 150 ºC:
laboratory 3 and short time laboratory 8 data removed.
1.E+07
Extensional stress growth coefficient, Pa.s
1.E+06
1.E+05
L6 190 0.962 L6 190 0.095

L6 190 0.009 L10 190 0.555
L10 190 0.103 L10 190 0.01
1.E+04 L9 190 081- 1 L9 190 082- 3

L9 190 083- 6.28 L9 190 084- 0.1
L9 190 086- 0.1 L9 190 085- 0.01
1.E+03
0.01 0.1 1 10 100 1000
Time, s
Figure 23: Intercomparison of stretching flow methods for a HDPE (HGH000) at 190 ºC:
laboratory 3 data removed.
56
10000000 0.5
Upper 95% limit
Transient extensional viscosity, Pa.s
Lower 95% limit
Relative combined uncertainty in

0.4
transient extensional viscosity

Transient extensional viscosity
Relative combined uncertainty
1000000
0.3
0.2
100000
0.1
10000 0.0
0.01 0.1 1 10
Strain
Figure 24: Analysis of the uncertainties in transient extensional viscosity values yielding
95% upper and lower confidence bands, based on a HDPE at 150 oC and 1 s-1.
The relative combined uncertainty values presented (right-hand y-axis) are for one
standard deviation.
10000000 0.5
Relative combined uncertainty in tensile .

0.4
1000000
stress growth coefficient
0.3
100000
0.2
10000
Upper 95% limit 0.1
Lower 95% limit
1000 0.0
0.01 0.10 1.00 10.00
Strain
Figure 25: Assessment of the uncertainties in extensional viscosity when testing at low strain
rates – HFU000 at 0.001 s-1 (test reference ext228).
57
1000000 0.5
Measured viscosity, Pa.s, ext185, 25.5 1/s
Upper 95% limit
Relative combined uncertainty in tensile

Lower 95% limit
100000 Transient extensional viscosity 0.4


10000 0.3
1000 0.2
100 0.1
10 0.0
0.1 1 10
Strain
Figure 26: Assessment of the uncertainties in extensional viscosity when testing at high rates
– HFU000 at 25.5 s-1 (test ref ext185).
1000000 0.5
Measured viscosity, Pa.s, ext197 100 1/s
Upper 95% limit
Lower 95% limit Relative combined uncertainty in tensile
100000 Relative combined uncertainty 0.4

10000 0.3
1000 0.2
100 0.1
10 0.0
0.1 1 10
Strain
Figure 27: Assessment of the uncertainties in extensional viscosity when testing at high strain
rates – HFU000 at 100 s-1 (test ref ext197).
58
Appendix A1 Definitions of strain, strain rate, stress and

material properties functions in tensile (simple) extension
The following definitions are given by Whorlow [14] for strains and strain rates. Further
descriptions are also given by, for example, Gupta et al [16] and Dealy [19, 23].
A1.1 Elongation ratio
The Elongation ratio (ER) is the ratio of the current length l to the initial length l o of the
specimen:
ER = l / l o. (A1.1)
It is dimensionless.
A1.2 Cauchy εC and Hencky ε strains
The Cauchy strain εC is given by the ratio of the change in length δ l to the initial length l o of
the specimen:
εC = δ l / l o. (A1.2)
The Hencky strain ε (also referred to as the natural or true strain) is given by the natural
logarithm of the elongation ratio:
ε = ln( l / l o) (A1.3)
Thus the elongation ratio is related to the Cauchy strain by
ER = 1 + εC (A1.4)
i.e. l / l o = 1 + δ l / l o. (A1.5)
To illustrate the difference between these measures of strain a specimen that was stretched to 10
times its original length has an elongation ratio of 10, a Cauchy strain of 9 and a Hencky strain
of 2.3. These strains are dimensionless.
A1.3 Cauchy ε c and Hencky ε strain rates
The Cauchy strain rate is given by
ε c = 1/ l o × ∂ l /∂t (A1.6)
and the Hencky strain rate by
ε = 1/ l × ∂ l /∂t (A1.7)
59
In describing and modelling plastics processing the Hencky strain is preferred as the rate of
strain of an element of fluid within the flow is independent of its original length and is
determined only from the velocity field of that element. It is thus a more suitable characteristic
of the flow.
These strain rates have units of reciprocal seconds (s-1).
A1.4 Extensional flows: velocity fields
Terminology and definitions for materials functions describing the response of viscoelastic
fluids to various shearing and extensional deformations are presented by Dealy [22].
Mathematical descriptions of extensional flows were presented, for example, by Meissner et al
[79] and Walters [15].
For tensile (simple or uniaxial) extension where ε is the true strain rate and ε is the true strain,
defined as
ε = ln( l / l o) (A1.8)
then in rectangular coordinates:
v1 = ε x1 (A1.9)
v2 = - ½ ε x2 (A1.10)
v3 = - ½ ε x3 (A1.11)
where ε ≥ 0.
Alternatively, in cylindrical coordinates:
vz = ε z (A1.12)
vr = - ½ ε r (A1.13)
where ε ≥ 0.
A1.5 Material properties
Following the notation presented by Dealy [22], and prepared by the Nomenclature Committee
of the Society of Rheology, for start-up flow in tensile (simple) extension at constant (Hencky)
strain rate ε the following definitions are given. Equivalent expressions for cessation of steady
tensile extension, tensile creep, tensile recoil and tensile step strain are presented by Dealy [22].
60
A1.5.1 Net tensile stress
The net tensile stress σE is defined by
σE = σ11 - σ22 = σ11 - σ33 = σzz - σrr (A1.14)
where σii is a stress tensor in either rectangular or axisymmetric co-ordinates. It has units of
pascal (equivalent to N/m2). The tensile stress growth function is indicated by σE+ where the +
indicates start-up rather than cessation of flow.
A1.5.2 Tensile stress growth coefficient
The tensile stress growth coefficient ηE+ is defined by:
ηE+(t, ε ) = σE/ ε (A1.15)
where t is time. It has units of pascal.seconds (Pa.s).
A1.5.3 Tensile viscosity
The tensile viscosity ηE is defined by:
ηE(t, ε ) = lim [ηE+(t, ε )] (A1.16)

t →∞
It is the limiting tensile stress growth coefficient value and represents an equilibrium
extensional viscosity if a steady value is achieved. However for materials that do not exhibit a
steady state behaviour the use of an ‘equilibrium extensional viscosity’ such as this is
obviously not appropriate.
61
Appendix A2 Analysis of extensional rheometer testing
A2.1 Definitions and terminology
ν - velocity of separation of the ends of the specimen, m.s-1
ω - angular velocity of drum, rad.s-1
r - drum radius, m
t - time, s
lo - original specimen length, m
l - specimen length at time t, m
Ao - original specimen cross-sectional area, m2
A - specimen cross-sectional area at time t, m2
F - force, N
ε - Hencky strain
ε - Hencky strain rate, s-1
σE - net tensile stress, Pa
ηE+ - tensile stress growth coefficient, Pa.s
The Hencky strain and strain rate and the tensile stress growth coefficient are defined in
Appendix A1.
A2.2 Analysis of extensional flow
Rearranging the Hencky strain
æ l ö
ε = ln çç ÷÷ (A2.1)
è lo ø
yields
l = l o e (ε ) (A2.2)
or, for constant Hencky strain rate ε ,
ε = ε t (A2.3)
where t is time. Thus
62
l = l o e (εt ) (A2.4)
Assuming conservation of volume of the specimen then
l oAo = l A (A2.5)
The tensile stress, expressed as the ratio of the force to area,
F
σE = (A2.6)
A
can thus be rewritten, using Equation A2.5, as
Fl
σE = (A2.7)
Ao l o
Substitution for l using Equations A2.2 and A2.4 yields

( )
Fe ε
σE = (A2.8)
Ao
or
F e (ε t )
σE = (A2.9)
Ao
respectively. Thus the tensile stress growth coefficient, the ratio of tensile stress to Hencky
strain rate, is given by
+ F e (ε )
ηE = (A2.10)
A o ε
or
+ F e (ε t )
ηE = (A2.11)
A o ε
A2.3 Analysis of the deformation in extensional rheometers: Type A
The following is an analysis of testing carried out using the NPL rheometer (Type A in
Figure 1). This analysis does not assume steady state behaviour. Analysis of flow in other
configurations can be similarly derived. For the specimen the effective length is given by the
separation between the clamps which in the case of the NPL instrument is between the fixed
clamp and the point at which a line from the fixed clamp meets the drum at a tangent ( l o in
Figure 1). The effective specimen length in the test region remains unchanged during the test as
63
the specimen is wound onto the drum. For the extensional rheometer using a single rotating
drum the instantaneous velocity of separation of the ends of the specimen, ν, is given by
ν = ωr (A2.12)
where ω is the angular speed of the rotating clamp and r is its radius. The velocity of separation
can also be written as
ν = ∂ l /∂t (A2.13)
Thus, by using the definition of Hencky strain rate
ε = 1/ l × ∂ l /∂t (A2.14)
the strain rate in the specimen is given by
ωr
ε& = (A2.15)
lo
This assumes that the length l o is constant. The specimen length will change with force due
to the compliance of the force transducer but this was considered to be minimal for the
configuration used.
Integrating Equation A2.15 with respect to time yields the strain

t
ε = ò ε dt (A2.16)
0
Thus
ω rt
ε = (A2.17)
lo
Thus the tensile stress is given, using Equation A2.8, by

æ ω rt ö
çç ÷÷
è lo ø
Fe
σE = (A2.18)
Ao
and the tensile stress growth coefficient, using Equation A2.10, by
æ ω rt ö
çç ÷÷
è lo ø
F lo e
ηE + = (A2.19)
Ao ω r
A2.4 Analysis: Type B
For the experimental configuration in which there are two sets of rotating clamps, each
having the same angular speed ω (Type B, Figure 1), then the analysis presented for type A is
valid except that
64
ν = 2ω r (A2.20)
Thus the strain rate is given by
2ω r
ε& = (A2.21)
lo
and the strain by
2ω rt
ε = (A2.22)
lo
The net tensile stress is given by equation (A2.9) and the tensile stress growth coefficient by
equation (A2.11), substituting for strain rate as appropriate.
A2.5 Analysis: Type C
In the case of a single translating clamp (Type C, Figure 1), to obtain a constant strain rate ε
then the speed of the clamp V, where
∂ l
V = (A2.23)
∂ t
is given, using equation (A2.4), by
V = l o ε& e ( )
ε t
(A2.24)
where l o is the original specimen length. For a specified constant strain rate, the strain is
given by equation (A2.3), the net tensile stress by equation (A2.9) and the tensile stress
growth coefficient by equation (A2.11).
A2.6 Analysis: Type D
In the case of using two clamps moving with the same speed (Type D, Figure 1), to obtain a
constant strain rate ε then the speed of the clamps V, where
∂l
2V = (A2.25)
∂t
is given, using equation (A2.4), as
l 0ε& e (εt )
V = (A2.26)
2
where l o is the original specimen length. For a specified constant strain rate, the strain is
given by equation (A2.3), the net tensile stress by equation (A2.9) and the tensile stress
growth coefficient by equation (A2.11).
65
Appendix A3 Analysis of the uncertainties of measurement of

transient extensional viscosity
A3.1 Introduction
Due to the considerable stretching that the specimen normally undergoes during testing,
particularly on reaching high strain values, there is a very significant decrease in the cross-
sectional area of the specimen, accompanied by a similarly large decrease in the measured
force. As a consequence there is a significant increase in the uncertainty of the measurements.
To understand and quantify the level of confidence that one can put in the accuracy of the
results obtained, it is important that the uncertainties in the measurements, in particular that
due to the accuracy and resolution of the force measurement, are known.
An example analysis is presented of the uncertainties in transient extensional viscosity

measurements for an extensional rheometer having a single rotating clamp and a fixed clamp,
Type A. The equations can be used to calculate the uncertainties in the derived tensile stress
growth coefficient values, given the uncertainties in each of the components of the
measurement. Similar analyses can be carried out for other instrument configurations by
following the principles presented. For further details see [6].
A3.2 Uncertainty analysis
Following a more rigorous approach, used for example by Kandil (2), the combined
uncertainty uc(y) of the measurand y (the quantity to be measured) can be determined from the
partial derivatives of the function and the uncertainties in the parameters. Assuming that
individual uncertainty sources are uncorrelated, the combined uncertainty uc(y) can be
computed using the root sum squares:
m
u c ( y) = å [c u ( x )]
i =1
i i
2
(A3.1)
where ci is the sensitivity coefficient (partial derivative) associated with an input quantity xi
and u(xi) is the uncertainty in that quantity.
The combined uncertainty uc(y) corresponds to one standard deviation and therefore has an
associated confidence level of approximately 68%. Assuming a normal distribution then it is
assumed that an expanded uncertainty U for 95% confidence level can be determined using a
coverage factor of 2 (i.e. equivalent to 2 standard deviations). The relative uncertainty is the
ratio of the uncertainty in the parameter to the value of the parameter.
66
A3.3 Uncertainty in strain rate
The strain rate ε in the specimen is given (Appendix 2) by
ωr
ε& = (A3.2)
lo
where ω is the angular speed of the winding drum, r is the radius of the winding drum and
l o is the original specimen length. The sensitivity coefficients, or partial derivatives, of the
strain rate are thus
∂ε& r
= (A3.3)
∂ω l o
∂ε& ω
= (A3.4)
∂r l o
and
∂ε& −ωr
= (A3.5)
∂l o l o 2
In what follows, a subscripted u denotes an uncertainty in the subscripted quantity. Thus,

using equation A3.1, the combined uncertainty in strain rate uε is given by the root sum of
the squares of the product of the partial derivatives and the corresponding uncertainties
2 2 2
æ r ö æω ö æ − ωr ö
(uε& ) 2
= çç uω ÷÷ + çç u r ÷÷ + ç 2 u l o
çl
÷
÷
(A3.6)
èlo ø è lo ø è o ø
and the combined relative uncertainty in strain rate is given, by dividing through by ε , by
2 2 2 2
æ uε& ö æ u ö æ u ö æ ul ö
ç ÷ = ç ω ÷ + ç r ÷ + çç o ÷÷ (A3.7)
è ε& ø è ω ø è r ø è lo ø
A3.4 Uncertainty in strain
The strain ε is given (Appendix 2) by
ω rt
ε = (A3.8)
lo
where t is time. Thus, in a similar fashion to that for strain rate, the combined relative
uncertainty in strain is given by
2 2 2 2 2
æ uε ö æ u ö æ u ö æ u ö æ ul ö
ç ÷ = ç ω ÷ + ç r ÷ + ç t ÷ + çç o ÷÷ (A3.9)
èε ø è ω ø è r ø è t ø è lo ø
67
A3.5 Uncertainty in transient extensional viscosity
Given the expression for transient extensional viscosity ηE (Appendix 2):

æωrt ö
ç ÷
è lo ø
Fl e
ηE = o (A3.10)
Aoω r
then the sensitivity coefficients ci are given by

æω r t ö
ç ÷
∂η E l e è lo ø
= o (A3.11)
dF Aoω r
æω r t ö
ç ÷
∂η E − Fl o e è lo ø
= (A3.12)
∂A o Ao ω r
2
æ æωrt ö
çç ÷÷
æωrt ö
çç ÷÷ ö
∂η E Fl o ç - e ÷
è lo ø è lo ø
ω te
= ç + ÷ (A3.13)
∂r Ao ω ç r 2 l or ÷
è ø
æ æωrt ö
çç ÷÷
æωrt ö
çç ÷÷ ö
Fl o ç - e ÷
è lo ø è lo ø
∂η E rte
= ç + ÷ (A3.14)
∂ω Ao r ç ω 2 l oω ÷
è ø
æ æωrt ö
çç ÷÷ ö
F ç ÷
æϖ r t ö è lo ø
∂η E çç
è lo ø
÷÷
l oω r t e
= çe + ÷
∂lo Aoω r ç − l 2o ÷
è ø
(A3.15)
æ æωrt ö
çç ÷÷ ö
F ç æϖ r t ö
çç
è lo ø
÷÷
ω rte
è lo ø
÷
= çe − ÷
Ao ω r ç lo ÷
è ø
and
æωrt ö
çç ÷÷
∂η E
è lo ø
Fe
= (A3.16)
∂t Ao
The combined relative uncertainty in the tensile stress growth coefficient values is then
evaluated as
2 2 2 2 2 2 2
æ u ηE ö æ ∂η u ö æ ∂η u ö æ ∂η u ö æ ∂η u ö æ ∂η u ö æ ∂η u ö
ç ÷ = ç E F ÷ + ç E Ao ÷ + ç E r ÷ + ç E ω ÷ + ç E lo ÷ +ç E t ÷
çη ÷ ç ∂F η ÷ ç ∂A η ÷ ç ∂r η ÷ ç ∂ω η ÷ ç ∂l η ÷ ç ∂t η ÷
è E ø è E ø è o E ø è E ø è E ø è o E ø è E ø
(A3.17)
where the partial derivatives are given in Equations A3.11 to A3.16.
68
A3.6 Evaluation of uncertainty in transient extensional viscosity

measurements
The values of the terms in the above uncertainty budgets, except for the force terms, are
relatively easy to determine. The difficulty with the force term is that the force decreases
considerably at high strain values and thus the relative uncertainty in the force will increase
significantly as the test progresses. To determine the effect of its contribution to the combined
uncertainty in the determination of the transient extensional viscosity values further analysis,
presented below, is required.
The force F supported by the specimen during testing is given by
F =σ A (A3.18)
where A is the cross-sectional area of the specimen and σ is the tensile stress. Using the
definition of transient extensional viscosity ηE+ (or tensile stress growth coefficient),
Appendix 1, then
σ = ηE + ε (A3.19)
Thus the force
F = ηE + ε A (A3.20)
or, using Equations A2.2, A2.5 and A2.17 of Appendix 2,

æ −ωrt ö
ç ÷
+ ç l ÷
F = η E ε Ao e è o ø
(A3.21)
Assuming that the transient extensional viscosity as a function of time can be fitted
reasonably well using a power law model (see Figure 8 of reference 1) then the equation of a
straight line fit to the data is of the form
log(ηE + ) = m log(t ) + log ηE ,o + ( ) (A3.22)
where ηE + = ηE ,o + at t = 1 s and ηE ,o + and m are constants. Equation A3.22 can be rewritten as
+ +
η E = η E ,o t m (A3.23)
The force over the test duration is thus approximated by

æ −ωrt ö
ç ÷
+ ç l ÷
F = η E ,o t ε Ao e m è o ø
(A3.24)
69
Given the values of ηE ,o + and m, the force F can be determined as a function of time and,
assuming that all strain rate data fall on the same master curve, also as a function of strain if
one ignores any additional strain hardening phenomenon. Thus the relative uncertainty in the
force F, given the decreasing magnitude of force at high strains, can be reasonably accurately
determined. However the combined uncertainty in transient extensional viscosity values must
be quoted as a function of time rather than as a single value.
Analysis is presented based on the case of a high density polyethylene HDPE (HGH000) at
150 °C and a strain rate of ≈ 1 s-1, Table 1 and Figure 10. The transient extensional viscosity
values are plotted along with 95% confidence limits derived using Equation A3.17 and a
coverage factor of 2, Figure 24. It clearly illustrates the significant increase in measurement
uncertainties at high strains due, predominantly, to the increase in the uncertainty in the
measurement of force resulting from the significant reduction in the cross-sectional area of
the specimen at high strain values. Expanded uncertainties representing 95% confidence
limits were estimated to be of the order of ± 20% at a strain of 3.3, ± 50% at a strain of 4.8
and ± 100% at a strain of 5.7. The initial decrease in uncertainty, up to a strain of 1, is due to
the low forces being generated in the specimen at the start of the test resulting in a high
relative uncertainty for force.
Table 1: Example values and uncertainties in the extensional testing parameters
ηE ,o + = 2.4 x105 Pa.s and m = 0.6 estimated for an HDPE (HGH000) at 150 oC at a strain rate
of ≈ 1 s-1.
Quantity Symbol Units Type Probability Divisor Quantity Quantity Standard Relative
distribution value range, ± uncertainty uncertainty
x ux ux/x
Initial specimen d1 m A Normal 1 0.0033 0.0001 0.0001 0.030
diameter 1
Initial specimen d2 m A Normal 1 0.0033 0.0001 0.0001 0.030
diameter 2
Force F N B Rectangular variable 0.016 0.0092 variable
3
Effective specimen lo m B Rectangular 0.1 0.003 0.0017 0.017
3
length
Angular speed ω rad/s B Rectangular 5.0 0.015 0.0087 0.0017
3
Drum radius r m B Rectangular 0.02 0.00075 0.00043 0.022
3
Time t s B Rectangular variable 0.024 0.014 variable
3
2 -6 -7 -7
Initial area* Ao m A Normal 1 8.6 x 10 3.7 x 10 3.7 x 10 0.043
Note: * Indicates calculated values based on the assumption of independent orthogonal measurements of the
specimen.
The standard uncertainty is obtained by dividing the quantity range by the divisor.
70
Appendix A4 Effect of viscous dissipation in the specimen

during stretching on measurements
It is assumed, for the purposes of this calculation, that the force exerted on the specimen F for
the duration of the test is constant and is equal to the maximum force exerted on the specimen.
Furthermore the specimen is of initial length l o, initial radius R o and undergoes a strain ε in a
Type C or D configuration (Figure 1). The final specimen length is given (Equation A2.2) by
l = l o eε (A4.1)
Thus the work done WF is given by
WF = (l o e ε − l o ) F (A4.2)
The work done per unit volume of the specimen is thus
(l o e ε − l o ) F
wF = (A4.3)
π Ro 2 l o
Assuming the thermal behaviour
wθ = ρ C p ∆θ (A4.4)
where ρ is the density, Cp is the specific heat capacity and ∆θ is the temperature rise in the
specimen then, given the energy balance
wF = wθ,
and thus
(l o e ε − l o ) F
∆θ = (A4.5)
( ρC p )(π Ro l o )
2
Using the values based on the HDPE (HGH000), Figures 5,

F=1N
l o = 0.1 m
R o = 1.5 x 10-3 m
ε=3
ρ = 900 kg/m3
Cp = 2300 J/(kg.K)
then the temperature rise ∆θ is estimated to be ≈ 1.3 oC.
This estimate assumes that the force was constant over the duration of the test, that all the
energy was dissipated as heat and that no heat was dissipated from the specimen. These
71
assumptions are all invalid. For example, Figure 5 clearly shows that although the peak force
was approximately 1 N the average force would have been lower. Also no account of the
energy stored elastically in the specimen or heat dissipated from the specimen was made. All
of these factors would result in a reduction of the temperature increase. For a HDPE
(HGH000) the shear flow temperature dependence was estimated to be less than 1 %/oC, and
the extensional flow properties were of the same order [74]. As the dissipation would be
evenly distributed through the specimen the effect on measurements is considered to be
negligible.
Although a C or D type configuration was used in this analysis, for ease of visualisation of the
problem and simplicity of the analysis, the predicted temperature rise will be the same for
other configurations where the same strain rate, strain and force values are used.
For other materials and testing conditions the temperature rise may not be so insignificant. A
more accurate calculation of the temperature rise can be carried out by summing the work
done, i.e. force x displacement for increments in specimen length, over the test duration using
the force and length equations, Equations A3.24 and A4.1 respectively.
72

Extensional Flow Properties of Polymers Using Stretching Flow Methods

Uploaded by

Copyright:

Available Formats

Extensional Flow Properties of Polymers Using Stretching Flow Methods

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Extensional Flow Properties of Polymers Using Stretching Flow Methods

Uploaded by

Copyright:

Available Formats

Measurement Good Practice Guide No.

Extensional Flow Properties of Polymers using

Martin Rides and Crispin Allen

In many forming processes the polymer melt undergoes predominantly extensional

National Physical Laboratory

3. General measurement considerations ............................................................................3

5. Stretching flow methods ..................................................................................................8

7. Experimental data for polymer melts...........................................................................33

8. Estimate of uncertainties in extensional testing ..........................................................39

• presents an assessment of the effect of uncertainties on the measured extensional

It illustrates the measurement of extensional flow properties with reference to results

In many plastics forming processes the melt undergoes predominantly extensional

In comparison, stretching flow methods can be used to generate quantitatively accurate

In summary, this Guide:

• defines approved terminology and presents definitions (Section 4)

• describes important considerations for stretching flow measurements (Section 5)

• describes practical considerations for measuring extensional flow behaviour, including

3. General measurement considerations

3.1 Extensional flow behaviour of polymer melts

In conclusion, to fully understand the flow behaviour of polymer melts in extension it is

3.2 Extensional flow stretching methods

1 For definitions see Section 4.

3.3 Industrial needs

3.4 Advantages and disadvantages of stretching flow methods

A summary of the main advantages and disadvantages of stretching flow methods is

• Strain and strain rate range

• Closely mimics many processing methods

4.2 Elongation ratio

4.3 Hencky strain

4.4 Hencky strain rates

Hencky strain rate ε (s-1): is given by

where t is time. It is independent of the original length l o .

4.5 Net tensile stress

The net tensile stress σE (Pa): in tensile (simple) extension it is defined by

σE = σ11 - σ22 = σ11 - σ33 = σzz - σrr (4)

where σii is a stress tensor in rectangular or axisymmetric coordinates.

4.6 Tensile stress growth coefficient

ηE+(t, ε ) = σE/ ε (5)

4.7 Tensile viscosity

The tensile viscosity ηE (Pa.s) is defined by

5. Stretching flow methods

5.2 Description of stretching flow methods

The basic principle behind stretching flow measurements is to subject a specimen to a

Type A: Single rotating clamp and a fixed clamp.

Type B: Two rotating clamps.

Type D: Two moving (i.e. translating, non-rotating) clamps.

In summarising the current state of measurement capability, extensional flow measurements

5.3 Limitations in using stretching flow methods

The application of stretching techniques to the characterisation of materials is limited by the

5.4 Measurement considerations

The following Sections contain detailed discussions of various aspects of extensional

5.4.1 Specimen preparation

It is accepted that the preparation of specimens is of critical importance for extensional

5.4.2.2 Translating or single rotating clamps (Types A, C and D)

Everage et al [60] used a modified Rheometrics Mechanical Spectrometer (rotational

5.4.2.3 Counter-rotating clamps (Type B)

5.4.3 Temperature control

Meissner et al [42] commented that temperature uniformity is essential, particularly for

5.4.4 Specimen support

Hencky strain rate ε (s-1): is given by

ηE+(t, ε ) = σE/ ε (5)

The strain rate ε is given by