418 61 1MB
English Pages 170 Year 1996
Practical Guide to the Assessment of the Useful Life of Rubbers
by
Roger P. Brown
O3
Practical Guide to the Assessment of the Useful Life of Rubbers
by Roger P. Brown
Rapra Technology Limited Shawbury, Shrewsbury, Shropshire, SY4 4NR, United Kingdom Telephone: +44 (0)1939 250383 Fax: +44 (0)1939 251118 http://www.rapra.net
First Published 2001 by
Rapra Technology Limited Shawbury, Shrewsbury, Shropshire, SY4 4NR, UK
©2001, Rapra Technology Limited
All rights reserved. Except as permitted under current legislation no part of this publication may be photocopied, reproduced or distributed in any form or by any means or stored in a database or retrieval system, without the prior permission from the copyright holder. A catalogue record for this book is available from the British Library.
ISBN: 1-85957-260-X
Typeset by Rapra Technology Limited Printed and bound by Bookcraft, Midsomer Norton, UK
Acknowledgements
This publication is an output from the Weathering of Elastomers and Sealants project which forms part of the UK government’s Department of Trade and Industry’s Degradation of Materials in Aggressive Environments Programme. I am indebted to Martin Forrest of Rapra Technology for contributing Chapter 5 on Degradation Mechanisms. I am grateful to the members of the Industry Advisory Group who steered the project, particularly the chairman Peter Lewis of TARRC, and to Steve Hawley and Tracy Butler of Rapra for their helpful advice and suggestions.
Practical Guide to the Assessment of the Useful Life of Rubbers
Contents
PART 1 BASICS ................................................................................................... 1 1
Introduction .................................................................................................... 3
2
The Problems .................................................................................................. 5
3
The Choice of Approaches ............................................................................. 7
4
Degradation Agents......................................................................................... 9
5
Degradation Mechanisms .............................................................................. 13 5.1 Introduction ............................................................................................ 13 5.2 Thermo-oxidative degradation ................................................................ 14 5.3 Thermal decomposition .......................................................................... 17 5.4 Radiation degradation ............................................................................ 17 5.5 Ultraviolet light degradation ................................................................... 19 5.6 Ozone degradation .................................................................................. 20
6
Time Dependent Limitations ......................................................................... 23 6.1 Induction period ..................................................................................... 23 6.2 Oxygen diffusion .................................................................................... 23 6.3 Fluid transport ........................................................................................ 26
7
Critical Factors.............................................................................................. 29
8
Parameters to Monitor Degradation ............................................................. 31 8.1 General ................................................................................................... 31
i
Practical Guide to the Assessment of the Useful Life of Rubbers
8.2 Tensile stress-strain properties ................................................................. 33 8.3 Hardness ................................................................................................. 33 8.4 Stress relaxation ...................................................................................... 34 8.5 Set ........................................................................................................... 36 8.5 Dynamic stress-strain properties ............................................................. 37 8.6 Volume change ........................................................................................ 38 8.7 Other properties ...................................................................................... 39 8.8 Functional tests ....................................................................................... 41 8.9 Chemical analysis .................................................................................... 42 9
Preparation of Test Pieces .............................................................................. 43
10 Uncertainty and Application of Statistics ...................................................... 45 PART 2 PRODUCT TESTS AND EXPERIENCE.............................................. 49 11 Simulating Service ......................................................................................... 51 11.1 General ................................................................................................. 51 11.2 Natural environmental exposure ........................................................... 51 11.3 Simulated design life ............................................................................. 54 12 Experience ..................................................................................................... 57 13 Principles of Product Testing ......................................................................... 59 13.1 Introduction .......................................................................................... 59 13.2 When to test products ........................................................................... 59 13.3 Design of product tests .......................................................................... 60 13.4 Examples of test rigs ............................................................................. 62 13.5 Summary ............................................................................................... 66
ii
Contents
PART 3 ACCELERATED TESTS ....................................................................... 67 14 General ......................................................................................................... 69 14.1 Purpose of accelerated tests ................................................................... 69 14.2 Methods of acceleration ........................................................................ 70 15 Fundamental Problems .................................................................................. 73 16 Designing an Accelerated Test Programme .................................................... 75 17 Effect of Temperature .................................................................................... 77 17.1 Low temperature ................................................................................... 77 17.2 Properties at service temperature........................................................... 77 17.3 Thermal expansion ............................................................................... 77 17.4 Heat ageing ........................................................................................... 78 18 Effect of Liquids ............................................................................................ 81 18.1 General procedures ............................................................................... 81 18.2 Standard liquids .................................................................................... 83 18.3 Water .................................................................................................... 84 19 Effect of Gases .............................................................................................. 85 19.1 General ................................................................................................. 85 19.2 Exposure to ozone ................................................................................ 85 19.3 Evaluation of cracking .......................................................................... 87 20 Weathering .................................................................................................... 91 21 Fatigue .......................................................................................................... 95 21.1 General ................................................................................................. 95
iii
Practical Guide to the Assessment of the Useful Life of Rubbers
21.2 Heat build-up tests ................................................................................ 95 21.3 Flex cracking and cut growth tests ........................................................ 97 21.4 Tests in tension ...................................................................................... 98 22.5 Non-standard methods ......................................................................... 99 22 Abrasion ..................................................................................................... 101 22.1 General ............................................................................................... 101 22.2 Types of abrasion test.......................................................................... 102 22.3 Abrasion test conditions ...................................................................... 104 22.4 Abrasion test apparatus ...................................................................... 105 22.5 Expression of abrasion test results ...................................................... 107 23 Other Degradation Agents .......................................................................... 109 23.1 Biological attack ................................................................................. 109 23.2 Ionising radiation ................................................................................ 109 23.3 Electrical stress .................................................................................... 110 24 Service Conditions....................................................................................... 113 24.1 General ............................................................................................... 113 24.2 Temperature ........................................................................................ 113 24.3 Solar irradiation .................................................................................. 114 24.4 Other factors ....................................................................................... 115 25 Prediction Techniques ................................................................................. 117 25.1 General ............................................................................................... 117 25.2 Standardised procedures ..................................................................... 118 25.3 Models for change of parameter with time ......................................... 118 25.4 Environmental degradation tests ......................................................... 120
iv
Contents
25.5 Arrhenius relationship ......................................................................... 121 25.6 Time/temperature shift ........................................................................ 124 25.7 Artificial weathering ........................................................................... 128 25.8 Ionising radiation ................................................................................ 130 25.9 Effect of liquids ................................................................................... 132 25.10 Effect of gases ................................................................................... 133 25.11 Creep and stress relaxation ............................................................... 133 25.12 Set ..................................................................................................... 135 25.13 Fatigue .............................................................................................. 135 25.14 Abrasion ........................................................................................... 137 25.15 Dynamic conditions .......................................................................... 137 26 Limitations and Pitfalls in Accelerated Testing ............................................ 139 26.1 Limitations .......................................................................................... 139 26.2 Pitfalls ................................................................................................. 140 Appendix ........................................................................................................... 143 References ......................................................................................................... 149 Abbreviations .................................................................................................... 151 Index ................................................................................................................. 153
v
Practical Guide to the Assessment of the Useful Life of Rubbers
vi
PART 1 BASICS
1
Practical Guide to the Assessment of the Useful Life of Rubbers
2
Introduction
1
Introduction
After price and delivery time, the most frequently asked question about a product is ‘How long will it last?’ This is usually a very difficult question to answer for rubber products because the expected lifetime is often in tens of years, the service conditions may be complex and there is a scarcity of definitive data. An answer to the question is wanted by both suppliers and users and it is important enough that an enormous amount of effort has been expended on measuring durability over the years and no doubt will continue to be expended. With so much effort having been applied it may seem odd that there is still a shortage of data. This is due to several factors. Firstly, there is a vast matrix of degradation agents, service conditions, properties of importance and different polymers and compounds. Perhaps even more significant is the magnitude of the inherent difficulties of designing tests which can be relied upon to give meaningful predictions of useful life in service. In many cases, as will be expanded on later, the timescales involved are such that accelerated test conditions are essential. Whilst large amounts of durability data are generated by accelerated methods much of it is only useful for quality control purposes and relatively little has been validated as being realistically capable of representing service. Indeed, there is considerable scepticism as to the value of any accelerated data. In truth, much of this scepticism is justified but much valuable information can be gained from accelerated tests if the limitations are recognised and the experimental programme designed with care. The useful life of a product generally means the time until it fails. Strictly, this can only be directly measured by service trials or tests on the complete product. Most assessments of lifetime of rubbers are made by considering some measure of performance, such as tensile strength, and specifying some lower limit for the property which is taken as the end point corresponding to when the material is no longer usable. It will be appreciated that what constitutes failure has to be defined. In the case of catastrophic failure, such as the rupture of an elastic band, this is obvious but in many cases there is no such clear end life. For example, is the end point when a few ozone cracks have appeared or when they have reached 5 mm in length? Broadly, the definition of the end point is the level of a property which is thought to be inadequate for service.
3
Practical Guide to the Assessment of the Useful Life of Rubbers It should also be remembered that lifetime is not necessarily measured in time. For some products it will be thought of as number of cycles of use or, in automotive applications, it might be viewed as number of miles. The object of this publication is to provide practical guidance on assessing the useful service life of elastomers. It covers test procedures and extrapolation techniques together with the inherent limitations and problems. There is a wealth of information which can be applied to help maximise the effectiveness of a durability testing programme, which this guide aims to make readily available. The results of the project studying 40 years of natural ageing of rubber and the accelerated testing programme conducted by Rapra Technology Limited, have been drawn on to indicate the limiting factors for particular materials and test methods. Durability in its broadest sense covers all aspects of irreversible property change which arise with time and use. This includes all types of environmental agent and all aspects of mechanical action that contribute to degradation. This guide seeks to be comprehensive but concentrates on the most common environmental effects and the most important mechanical properties. More details of the test procedures used can be found in text books and the relevant international standards as referenced later in this Guide.
4
2
The Problems
There are two fundamental hurdles to be overcome in assessing the service life of a product: • the uncertainty of, and the variation in, the service conditions, • the timescales for expected lifetime. Simple illustrations of the first point are the difference in climates of Moscow, London and Phoenix, and the different chemicals that might come in contact with an engineering component. The implications are that uncertainty may mean designing for the worst case (or being unexpectedly caught out) and the variety potentially calls for testing (or making extrapolations) under many conditions. With expected lifetimes usually in years and even in tens of years, the timescale to prove the life of the product absolutely by conducting trials under service conditions is often prohibitive. Satisfactory accelerated testing is neither cheap, easy nor even always valid. Both these factors lead to a great practical barrier – proving durability is generally very expensive. The costs and complexity of measuring durability are compounded by the number of properties which may be of interest and the simple fact that they will not all change at the same rate. Further, as will be discussed later, there are effects of geometry and also synergy between degrading agents. When more than one material is involved the scale of the challenge can look daunting. These problems indicate why there is still a shortage of reliable data and also why much durability testing falls short of the breadth and quality necessary to make reliable predictions.
5
Practical Guide to the Assessment of the Useful Life of Rubbers
6
3
The Choice of Approaches
There are essentially three approaches to assessing service life: • real (or simulated) service trials, • the use of experience, • accelerated testing. Few would argue against real service trials being the first choice if the conditions and timescale do not completely rule it out. Even when this approach is not feasible before the launch of a product, it is highly desirable that real life trials are started as early as possible because, at the very least, they can warn of impending disasters in the field. Clearly, if you had experience of the product and application none of this discussion would be relevant. However, indirect experience can be applied – how the same material performed in another application, how other materials performed in the same application, and information generated by other people. The majority of rubber products are probably designed largely on the basis of experience. It is not an inferior approach, rather one that is critically dependent on the quality of the experience and the validity of the way it is applied to the new circumstance. It is probably not going too far to suggest that accelerated testing is the last resort. By this it is meant that the inherent difficulties of accelerated testing dictate that even after the time and expense of the testing the uncertainty in the estimates may be considerable, or in the worst scenario quite wrong. Nevertheless, in a great many circumstances accelerated tests have to be relied on at least in part.
7
Practical Guide to the Assessment of the Useful Life of Rubbers
8
4
Degradation Agents
During its lifetime a rubber will be exposed to one, or usually several, types of degradation due to various agents. The contribution of each of these is often complicated to evaluate but their classification is simply stated as in Table 1. In durability testing, heat as a single agent is generally taken to mean the action of both temperature and atmospheric oxygen. Hence a test for the effect of temperature must consider the availability of oxygen and its diffusion in the test piece. Oxidative degradation is generally considered to be the most serious problem in the use of rubber at high temperature, but also proceeds relatively slowly at ambient temperature. Heat is also very commonly used in testing in combination with other agents, notably light, liquids and gases. Thermo-oxidative degradation Oxidative degradation Crosslinking
Table 1 Degradation agents
Hydrolysis Agent Swelling Temperature Additive depletion Bio-orgamisms Light Fatigue Ionising radiation Creep Humidity Stress relaxation Set AbrasionFluids (gases, liquids, vapours) Adhesive failure Bio-organisms Electrical stress
Type of ageing or effect Thermo-oxidation, additive migration, crosslinking, crosslink loss (reversion) Photo-oxidation Radio-oxidation, crosslinking Hydrolysis Chemical degradation, swelling, additive extraction, cracking Decomposition, mechanical attack
Mechanical stress
Fatigue, creep, stress relaxation, set, abrasion, adhesive failure
Electrical stress
Local rupture
9
Practical Guide to the Assessment of the Useful Life of Rubbers Low temperature (subambient) is not really a degradation agent but produces temporary physical effects on properties, such as stiffening, brittleness and recovery from strain, which may be critical in terms of service performance. Light includes radiation beyond the visible range, particularly in the ultraviolet (UV) region. Short wave UV is the most damaging to polymers, but longer wavelengths may affect pigments and will produce a rise in temperature. A temperature increase accelerates the rate of any chemical reaction and, while most photochemical reactions are not very temperature sensitive, any subsequent chain reactions usually are temperature dependent. Ionising radiation covers X-rays, gamma radiation and the various subatomic particles. It is normally only important in specialist applications. Moisture can produce hydrolysis in some materials and have a plasticising effect. There may be a synergistic effect with other agents. For example, a material resistant to UV alone or to moisture alone may fail when exposed to UV and moisture in combination. Fluids encompass a whole range of chemicals, both gases and liquids, which can come into contact with the material in various ways. Fluids may be absorbed and cause swelling of the rubber, or may extract soluble constituents of the compound, or be a source of pro-oxidant materials (from water, cleaning fluids, etc.), or may have chemical effects. Ozone causes cracking of many rubbers when under strain. Damage caused by bio-organisms is generally insignificant, although rubber seals for sewage pipe can be susceptible in temperate climates and biological attack can sometimes be a serious problem in tropical countries. The effect of the environmental agents outlined above is generally to cause loss of mechanical properties and accelerate the effects of the mechanical stresses listed in Table 1. Fatigue failure resulting from repeated cyclic deformation is possible in many applications and can result in crack propagation or heat build-up. For seals and similar products the set and stress relaxation which develops with prolonged deformation are critical factors to performance. Alternatively, under prolonged stress, creep will occur. In some applications the product will be subject to abrasion, and in specialist areas to electrical stress.
10
Degradation Agents Two degradation agents may have a synergistic effect in that their effect, in combination, is greater or less than the sum of their individual effects. This is clearly very important when multiple agents are present. Simple examples are: perspiration in contact with elasticated garments weakening threads through swelling, which leads to more rapid ageing on cleaning; or light ageing of a stretched product causing surface relaxation, which leads to apparently improved ozone resistance. The rate of degradation by environmental agents is generally increased in the presence of stress. The degradation agent may be applied continuously or intermittently to give cyclic exposure. The exposure cycles may include two or more agents applied alternatively. The effects of cyclic exposure may not be the same as the equivalent dose of continuous exposure.
11
Practical Guide to the Assessment of the Useful Life of Rubbers
12
5
Degradation Mechanisms
5.1 Introduction Polymer degradation is generally taken to mean the irreversible change of chemical or physical properties which is caused by the chemical and/or physical action of agents. Degradation comprises the deterioration of macromolecules caused by bond scissions in the polymer backbone, chemical reactions in the side-chains and intermolecular crosslinking with formation of new chemical bonds between different molecules. There can also be changes to ingredients such as extraction of plasticisers or attack of fillers by acids. From a practical point of view, reversible physical changes are also important. Prime examples are the physical components of set, creep and relaxation, and the stiffening effects of low temperatures. The mechanisms which produce degradation vary with the polymer, additives, the degradation agents present and the exposure level. In many cases there will be more than one mechanism involved and the relative importance will depend on the mix of degradation agents present and the exposure levels. Consequently it is important that at least a general appreciation of the degradation reactions of the polymers being tested is obtained before planning a trial, so that the test conditions can be sensibly chosen. The degradation of rubber compounds is a complex process which is difficult to interpret quantitatively in terms of molecular theory. Therefore, although the advance in analytical techniques has been considerable in recent years, it is still the case that degradation research is mainly carried out by studying the changes in bulk properties. The chemical degradation of rubber essentially takes place via radical reactions, these radicals being initiated by energy sources such as heat and ultraviolet radiation. It is also important to the course of the reaction if oxygen is present or not. Radical reactions are also influenced by species acting as chain transfer agents and terminators. The form of the reaction for a given rubber compound therefore varies according to how the particular polymer and additives behave. The type of radical also plays an important role.
13
Practical Guide to the Assessment of the Useful Life of Rubbers It should be noted that for reactions involving oxygen relatively little of the gas (ca. 1% oxygen absorption) is required, so that degradation occurs even in situations where access of oxygen is restricted. However, at higher temperatures (the reactions are accelerated by temperature) the process can be restricted by the rate at which oxygen can diffuse through the rubber. The dominant oxidative reactions may change with temperature which can restrict the validity of accelerated tests. Antioxidants are added to most rubber compounds to inhibit oxidation and, clearly, if the antioxidant is used up there will be a rapid change in degradation rate. At the same time as degradation is proceeding there may be additional crosslinking taking place which again complicates interpretation of accelerated test results. There are other types of degradation that rubber can undergo, such as degradation by aggressive chemicals (e.g. acids) and physical degradation (e.g. by being swollen by a solvent) but these are outside the scope of this guide. There are a number of reviews of this subject in the literature. One such recent publication was used as the basis of this chapter; it is by Y. Saito and an English translation is given in Reference 1.
5.2 Thermo-oxidative degradation Commonly, in the absence of radiation, the degradation of a rubber takes place by a thermo-oxidative mechanism. In addition to the factors which usually affect a free radical reaction this type of degradation depends upon the oxygen density in the area that it is taking place. The oxygen density in turn depends upon the oxygen solubility, oxygen dispersion rate and the rate at which the oxygen is being used up (i.e. the oxidation rate). The general mechanisms by which thermo-oxidative degradation occur are given below: (a) When a sufficient supply of oxygen is present Initiation
Initiator (I) → I· I· + RH → IH + R·
Chain Reaction
R· + O2 → ROO· ROO· + RH → R· + ROOH
Termination Reaction
2ROO· → deactivated material
where:
14
RH = rubber
Degradation Mechanisms The hydroperoxide (ROOH) generated can itself break down to form two radicals (RO· + OH·) and so the reaction, once started is regarded as being auto-catalytic. Conversely, a hydroperoxide decomposition agent can act as an antioxidant by decomposing the hydroperoxide into harmless non-radical products. Another way in which antioxidants can function is by reacting with free radical sites on the rubber chain (e.g. hindered phenol type antioxidants). These types of antioxidants are called ‘chain breaking’ antioxidants. The free radical source is reduced because the reactive chain radical is eliminated and the antioxidant radical produced is stabilised by internal resonance, i.e. ROO· + AH → ROOH + A· where:
AH is the chain breaking antioxidant
Metal catalysed oxidation can have an important effect on the lifetime of certain rubbers, e.g. natural rubber. Traces (ppm level) of certain transition metals, notably copper, manganese, cobalt and iron, are able to break down hydroperoxides in a redox reaction to give reactive peroxy radicals which then propagate oxidation and so ageing occurs at lower temperatures. They therefore have the opposite effect to a hydroperoxide deactivating antioxidant. The usual mechanism is: ROOH + M++ → ROO· + M+ + H+ ROOH + M+ → RO· + M++ + OHwhere:
M is the transition metal
(b) Where there is no oxygen present Initiation
Initiator (I) → I· I· + RH → IH + R·
Chain Reaction
R· + RH → RRH
Chain Transfer Reaction
R· + R´H → R´· + RH
Beta Scission Reaction
R· → R1 + R2·
Termination Reaction
2R· → deactivated material (crosslinking reaction)
(containing (a form of crossa double linking) bond)
The characteristic feature of both types of reaction ((a) and (b)) is that the initiator (I) is not regenerated as a consequence of any of the reactions, and so the removal of the
15
Practical Guide to the Assessment of the Useful Life of Rubbers energy required to form the initiator (i.e. the heat) will result in the rate of degradation decaying to zero. The rate of oxygen permeation has to be sufficiently high for thermo-oxidative degradation to progress down into the interior of a specimen. It can be the case that a material that has relatively poor oxidation resistance can perform better throughout its bulk than one that has inherently better resistance because of differences in their oxygen permeability. The chemical structure of a rubber has a profound effect on its resistance to degradation. This is due mainly to the effect that the structure has on the ability of carbon or oxygen radicals to abstract a hydrogen atom from the polymer chain, but also on the reactivity of the radicals that are so produced. In general, the hydrogen abstraction by an oxygen radical is very electrophilic and so when an electron absorbing group (e.g. a nitrile group) is present reactivity falls and when a donor group (e.g. a methyl group) is present it increases. Generalised oxidative ageing resistance values (maximum service temperature) for a range of rubber types are given in Table 2.
Table 2 Maximum service temperatures for a range of rubbers Rubber type
Maximum service temperature (°C)
Polyurethane rubber
80
Natural rubber and polyisoprene rubber
80
Butadiene rubber and styrene butadiene rubber
100
Nitrile rubber and polychloroprene rubber
120
Butyl rubber
130
Ethylene propylene rubber
140
Chlorosulphonated rubber
150
Hydrogenated nitrile rubber
160
Acrylic rubber
160
Silicone rubber
225
Fluorocarbon rubber
250
16
Degradation Mechanisms
5.3 Thermal decomposition Thermal decomposition is where radical breakdown occurs in the complete absence of oxygen. In this case the weaker the bond dissociation energies within the polymer backbone the faster the decomposition occurs. Table 3 shows the relationship between bond dissociation energy and thermal decomposition temperature. Rubbers which are halogenated undergo a distinct two stage decomposition mechanism in which dehydrohalogenation occurs first at a relatively low temperature, followed by main chain breakdown. The temperature at which this dehydrohalogenation occurs is often dependent on the thermal stability of structural moieties within the polymer chain. ‘Defect’ structures formed during the polymerisation reaction render the polymer less stable than would be the case if it was a homologue of its monomer. Nitrile rubber undergoes an analogous dehydrocyanation prior to main chain breakdown. The rate of these initial dehydro- reactions depends on the degree of stabilisation of the resulting structures (e.g. by conjugation); the activation enthalpy varies according to polymer structure.
5.4 Radiation degradation When a polymer is subjected to irradiation it is ionised and excited and this leads to the formation of radicals by molecular breakdown, causing breakage of the main chain and crosslinking to take place, and a variety of gases to be released.
Table 3 Thermal decomposition temperatures of a range of rubbers Rubber type
Temperature to achieve 50% decomposition* (°C)
Polyisoprene rubber
323
Nitrile rubber
360
Styrene butadiene rubber
375
Chlorosulphonated rubber
380
Polychloroprene rubber
380
Ethylene propylene rubber
388
Butadiene rubber
407
Silicone rubber
560
* The temperature at which 50% mass loss occurs in a specified time interval
17
Practical Guide to the Assessment of the Useful Life of Rubbers If a molecular structure is depicted as:
-(CH2-CR1R2)n –
then, in general, breakage of the main chain will occur when R1 and R2 do not contain hydrogen and crosslinking occurs when R1 and R2 contain at least one hydrogen. Hence, in most rubbers crosslinking takes place – the exceptions being butyl rubber and fluorocarbon rubber. Scission type degradation is more readily induced when oxygen is present. It is therefore possible to have opposite degradation effects manifest on the surface of a rubber compared to its bulk. The resistance of various rubbers to gamma irradiation (under conditions where no oxidation occurs) is given in Table 4. It can be seen from Table 4, that the chemical structure of the rubber has an important influence on its radiation resistance. For example, compounds containing aromatic rings with a high resonance energy (e.g. styrene butadiene rubber and polyurethane rubber) have the ability to render the radiation energy harmless by transforming it into light and heat. The performance of a rubber with respect to radiation resistance is often different to its response to other mechanisms such as thermo-oxidation; styrene butadiene rubber and polyurethanes having relatively poor resistance to the latter type of degradation.
Table 4 Gamma radiation resistance of a range of rubbers Rubber type
Insignificant damage (Radiation dose, Gy)
Butyl rubber
Up to 10,000
Acrylic rubber
Up to 100,000
Silicone rubber
Up to 100,000
Chlorosulphonated rubber
Up to 100,000
Nitrile rubber
Up to 100,000
Fluorocarbon rubber
Up to 100,000
Polychloroprene rubber
Up to 100,000
Styrene butadiene rubber
Up to 500,000
Ethylene propylene rubber
Up to 500,000
Polyurethane rubber
Up to 500,000
18
Degradation Mechanisms Fluorocarbon rubbers also show a distinct difference in their performance, having good thermo-oxidative resistance but poor resistance to radiation. A detailed review of the radiation chemistry of rubbers is given in Reference 2.
5.5 Ultraviolet light degradation This type of degradation can also be referred to as weathering, since a rubber has to be exposed to the elements to be affected by UV light, and photodegradation. The ultraviolet rays of solar light reaching the earth have a wavelength of 290-400 nm, which corresponds to the bond energy of standard organic compounds (99–72 kcal/mol). Chemical reactions are induced when molecules contain specific functional groups that absorb UV light and are excited; the breakdown of these or the energy transfer to other functional groups gives rise to the formation of active species and the initiation of the chemical reaction. In the presence of oxygen, the hydroperoxide groups formed from the initial stages of the degradation, and their carbonyl breakdown groups, act as further initiators. Hence, when there is no oxygen present UV absorbing active species are not reformed in a chain reaction and the total degree of degradation that results is only minimal. It is therefore the case that the area of a rubber that undergoes photodegradation is mainly that which oxygen can diffuse into and so is different from, say, radiation degradation. In the case of an ester group (R1-COO-R2) the R2 undergoes radical attack in thermooxidative degradation, but in photodegradation the group is broken down into R1 and COO-R2. This is due to the high photoreactivity of the carbonyl group. Photodegradation is activated by a variety of active species and this makes an overall description of the process difficult. However, the following three reactions are important in the photodegradation of rubbers: 1) Species such as polynuclear aromatics react with oxygen to produce singlet oxygen. Singlet oxygen reacts with diene rubbers (e.g. natural rubber) to cause degradation, but does not react with saturated rubbers such as ethylene propylene and acrylics. Stabilisers that deactivate singlet oxygen can be used to protect against photodegradation. 2) The breakdown of the peroxides formed during photodegradation in the presence of oxygen is accelerated by light-activated phenyl compounds. This peroxide breakdown assists in the progression of the degradation process by the formation of further activating species (e.g. carbonyl groups). 19
Practical Guide to the Assessment of the Useful Life of Rubbers 3) Carbonyl groups are readily activated by light (they absorb at 290 nm) and this results in main chain scission and the formation of free radicals. Experiments using UV irradiation in the region 340–400 nm, have shown that the crosslink density of sulphur vulcanised rubbers can fall, due to the scission of carbon-carbon and polysulphide bonds. General inorganic fillers can be used in rubbers to screen products from UV light and improve their weathering resistance. Carbon black is particularly effective at improving the UV resistance of a rubber by strongly absorbing UV and protecting the polymer chains. This results in the prime degradation mechanism of carbon black filled rubbers in outdoor environments being thermo-oxidation. It should be borne in mind that a rubber compounded to perform well in the presence of elevated temperatures may not perform well under UV light irradiation. This is because some antioxidants are sensitised by near UV light and can actually promote photooxidation even in the presence of carbon black. The degradation mechanisms of a number of rubbers are dealt with in detail in References 3 to 6.
5.6 Ozone degradation The ozone reaction is electrophilic and ozone therefore attacks where the electron density in a polymer chain is high (e.g. at a double bond site). The reaction is accelerated by the presence of an electron donor group (e.g. a methyl group). Conversely, the presence of an electron accepting group (e.g. a chlorine atom) will decrease reactivity. Certain aromatic amines (e.g. para-phenylene diamines) react very strongly with ozone and they can be used as antiozonant additives in rubber compounds. In the case of saturated rubbers, the ozone reactivity increases with the strength of the electron donor alkyl groups. The ozone reactivity of primary:secondary:tertiary hydrogen is in the ratio 1:13:110. Compounds having a higher electron-donor capacity, such as amines, ethers and thioethers, have a higher rate of reactivity, i.e. polyether rubbers (such as Hydrins (epichlorohydrin based rubbers)) have inferior ozone resistance compared to ethylene propylene rubbers. The key event with ozone degradation is the scission of the polymer chain with crack growth occurring around locally stressed surface flaws. The two characteristic features of the attack are a threshold strain, largely independent of ozone concentration in the absence
20
Degradation Mechanisms of protection, and a rate of crack growth that can be proportional to ozone concentration. Antiozonants can affect one or both of these depending on how they function. The relative ozone resistance of a number of rubbers in the unstabilised state is shown below: Rubbers with very poor resistance
Diene rubbers (e.g. natural rubber, styrene butadiene rubber, nitrile rubber, butadiene rubber)
Rubbers having some resistance
Butyl rubber, Hydrins, hydrogenated nitrile rubber, polythioethers, polychloroprene
Rubbers having good resistance
Acrylics, chlorosulphonated polyethylene, ethylene propylene rubbers, fluorocarbon rubbers, silicone rubbers
The following approaches can be made to improve the ozone resistance of rubbers: 1) Use of stabilisers such as para-phenylene diamines which have a higher reactivity to ozone than the rubber. 2) Use of paraffinic waxes to form a protective barrier on the surface of the rubber. Use of protective paints to protect the rubber surface. 3) Blending ozone resistant polymers (e.g. polyvinyl chloride) with those that have poorer resistance (e.g. nitrile rubber) to prevent crack growth. 4) Modification of rubbers to reduce reactivity with ozone, e.g. hydrogenation of nitrile rubber. Finally, there is a relationship between polymer chain mobility, i.e. glass transition temperature (Tg), and ozone degradation. In the region below Tg of 50–100 °C the rate of crack initiation slows down as the temperature is reduced, but above 100 °C there is no temperature dependence and ozone degradation is more rapid. There is also a critical, or threshold, stress for crack initiation, below which initiation is not observed and above which large numbers of small cracks are formed. Reviews of ozone degradation are given in References 7 to 9.
21
Practical Guide to the Assessment of the Useful Life of Rubbers
22
6
Time Dependent Limitations
6.1 Induction period Oxidising reactions in polymers often show an induction period for ageing whereby relatively little change is seen during this period but then the rate of degradation increases abruptly. It is easy to see how this can happen in rubbers protected by antioxidants in that, with time, the protective additives will be consumed. However, the effect can also be seen in unprotected materials where it is probably due to the time required for hydroperoxide accumulation prior to autoacceleration. The induction period can effectively be a matter of years for an oxidation resistant rubber even in the absence of antioxidant. Well protected general purpose rubbers can also resist changes over prolonged periods at near-ambient temperatures. It is useful to note that for the practical materials used in the Rapra 40 year natural ageing project, none displayed any severe change of degradation rate over this period. Also, the accelerated ageing tests carried out on these and 20 further materials did not produce obvious evidence of induction period effects, although in many cases there were very significant changes in reaction rate with time. The practical importance of an induction period is that it cannot be assumed that lack of change or a very slow rate of change will continue forever and, conversely, the possibility that one exists cannot be ruled out if only single point ageing data are considered.
6.2 Oxygen diffusion An important factor in oxidative ageing is the restriction due to the rate of oxygen diffusion into the rubber, which means that the importance of test piece or product thickness needs to be stressed. It is well known that heat ageing tests are limited by the rate of oxygen diffusion into the test piece. For a natural rubber at 100 °C a figure of 0.8 mm thickness has been given, above which the lack of oxygen reaching the centre of the test piece will cause a significant profile of oxidative effects through the test piece, as illustrated in Figure 1. Clearly, the
23
Practical Guide to the Assessment of the Useful Life of Rubbers commonly used standard test piece thickness of 2 mm is well above this and the changes measured in bulk properties must be influenced by the degradation gradient. Generally, this problem is conveniently ignored. Some tests require much thicker test pieces than 2 mm and inevitably the extent of ageing will be limited by the rate of oxygen diffusion. Hence the results from, for example, compression set tests are likely to be seriously affected. The actual thickness at which a degradation gradient becomes significant will obviously become smaller as the test temperature is raised. It will also depend on the material: faster degrading materials require a greater rate of oxygen supply and hence the limiting thickness will be smaller. For natural ageing at normal ambient temperatures, a much greater thickness should be tolerated before the rate of oxidation far outstrips the rate of oxygen diffusion. However, examination of very old (19th century) bridge bearings by the Malaysian Rubber
Figure 1 Limitations of oxygen diffusion
24
Time Dependent Limitations Producers’ Association (MRPRA) showed that degradation was confined to thin surface layers. The same effect has been observed in other old products and commented on by several workers. The thin surface layer was very hard and had formed a skin which inhibits further ingress of oxygen to the bulk. This skin effect was not seen in the materials aged for 40 years in the Rapra project so possibly it is related to the crude natural rubber formulations used at that time. Lindley and Tao [10] carried out some very interesting ageing experiments on natural rubber blocks and found that a skin was formed (at least in natural rubber) during accelerated ageing at relatively high temperatures. They also predicted that it would take more than 100 years at ambient temperatures for such a skin to form. On balance, it would seem that differential ageing is not a problem under natural conditions but there is little doubt that it will be found in accelerated tests. The limiting factor in oxidative ageing is whether the rate at which oxygen is consumed is greater than the rate at which it can diffuse to the given depth. The depth to which it diffuses is proportional to the square root of time. Whilst the rate of permeation could be estimated from knowing the oxygen permeability coefficient for the material, it is also necessary to know the consumption rate at the temperature of interest to determine whether oxidation will be limited. A degradation profile can be detected by measuring a property, such as micro hardness, as a function of depth after ageing so that the magnitude of any effect from the limitation of oxygen diffusion could be measured for any temperature and material combination. This is obviously an advantageous step to take but carries a cost and time penalty. It should be noted that the effect of a degradation profile on a bulk property will depend on the particular property measured. For example, tear strength will be largely dependent on the (greater) degradation at the surface whereas modulus will be measured as the average of surface and interior values. Kube et al. [11] derived empirical relations for air ageing of hydrogenated nitrile rubber materials incorporating test piece geometry and oxygen diffusion. They showed that thermo-oxidative ageing was directly related to modulus change. Stability under anaerobic conditions (in the bulk of thick components) can be high even for rubbers that undergo rapid thermo-oxidation. However an oxidation resistant rubber such as sulphur cured ethylene propylene rubber can still undergo crosslink changes at elevated temperatures under essentially anaerobic conditions. The practical conclusion is that if comparisons are to be made between accelerated and natural ageing results or attempts made to predict degradation at lower temperatures
25
Practical Guide to the Assessment of the Useful Life of Rubbers from accelerated tests, the existence of degradation gradients is likely to have a significant effect. Also, predictions made from tests on thin test pieces may be misleading if applied to thick products.
6.3 Fluid transport There is a diffusion rate factor when rubbers are exposed to any gas or liquid. Usually absorption of fluid (swelling) is greater than extraction of soluble constituents of the rubber and builds up to an equilibrium condition as shown in Figure 2 (curve A). If extraction is also taking place, a maximum swelling may be reached (curve B). If the absorption of fluid is accompanied by oxidation, volume may continue to increase (curve C). Quite clearly, any measurement of swelling before equilibrium is reached could be misleading. Figure 3 shows absorption curves for two materials differing in maximum absorption (M1 and M2) and absorption rate. If the absorption was measured at a single time point A, both materials would rate as the same. Equally, if there is also extraction, measurements of volume change and other properties, should wait until a final plateau condition is reached. If oxidation, or other chemical reaction, is taking place a rather longer timescale than that for swelling alone will be needed to properly characterise the changes in properties. The time to reach maximum absorption will increase with increased test piece thickness in a manner roughly proportional to the square of thickness. This will need to be taken into account if predictions from results on thin test pieces is applied to thick products. For organic liquids the time to equilibrium is roughly proportional to the viscosity but the rate for water is very slow and in many cases will take months or years. The practical implication is that care must be taken to avoid drawing conclusions from results obtained before equilibrium conditions are reached.
26
Time Dependent Limitations
Figure 2 Time-swelling curves
Figure 3 Swelling rates for two different materials
27
Practical Guide to the Assessment of the Useful Life of Rubbers
28
7
Critical Factors
The service life of a product can be influenced by many factors. These include fatigue failure under repeated stressing, excessive creep or stress relaxation, ozone cracking, excessive change in stiffness due to thermal ageing, and excessive change in a physical property due to the action of chemicals. Usually, there is more than one degradation agent present and there may be several, giving a very complicated situation. In many cases it will be possible to identify the critical degradation agent(s) for the particular application or the particular objective, so that for the purposes of the trial other agents are ignored. Whilst this may limit the trial it will allow a manageable test programme to be designed. As a simple illustration, it is no good spending lots of time evaluating long-term thermal ageing if the chosen material fails by ozone cracking in a couple of months. It must be noted that care should be taken not to overlook an important environmental effect, nor to discount synergistic reactions. For example, product failure can occur through metal ion–catalysed oxidation but this is generally overlooked in an accelerated ageing test; and oxidative ageing can have a deleterious effect on the performance of chemical antiozonants. In a similar manner, whilst many properties will change because of the environment, they will change at different rates and hence to different degrees. For example: fatigue life often deteriorates more rapidly than tensile strength; with many synthetic rubbers hardness and modulus can rise appreciably whilst tensile strength remains virtually the same; some fluids may cause little swelling but may extract protective agents or cause chemical attack. Consequently, it is important to understand what are the most important properties for a given application. As a simple example, there is not much point in compounding for maximum abrasion resistance if failure will occur from cracking at low temperature. Assessing the critical factors is not always easy and many instances of premature failure in service have occurred because a critical factor had been overlooked. Unfortunately, the longer a product is in service the greater the risk of the unusual or unanticipated occurring. Over years there can be a change in operating conditions or the environment.
29
Practical Guide to the Assessment of the Useful Life of Rubbers
30
8
Parameters to Monitor Degradation
8.1 General It is generally accepted that, ideally, a range of properties should be used to monitor ageing and these should be selected to be relevant to service. The use of a range of properties recognises the fact that the degree of change can be property dependent. The ideal is rarely achieved because of cost implications and because many trials are conducted without there being a specific product in mind. In fact, a literature search shows that in practice only a few physical tests are commonly used. The factors which affect the choice of test parameters are: Relevance to service Sensitivity to the degradation agent Relation to fundamental changes at the molecular level General applicability Reproducibility of test results Experimental cost and convenience. As discussed in the previous section, when the trial is directed at some particular service application, parameters critical to that application should have priority. Indeed, it is essential that the most critical parameters are identified if judgements are to be made as to fitness for purpose. Where the trial is directed more generally at assessing the resistance of materials to a particular degradation agent, properties deemed to have general applicability are attractive, but sensitivity to the agent and the relation to fundamental molecular changes will have greater importance. High sensitivity will minimise the exposure time required and demonstrate the worst situation, whilst the ability to relate this to change at the molecular level may allow an understanding of the degradation process taking place. Properties having general applicability to a range of applications are, not surprisingly, those most commonly measured. It should be noted that the advantage of wide applicability is negated if the parameters in question are inferior, for example because of low sensitivity. 31
Practical Guide to the Assessment of the Useful Life of Rubbers Reproducibility of test methods is important and a parameter with distinct advantages in other ways may have to be rejected if it is poor in this respect. When timescales may extend over months and the results are subject to extrapolation, poor reproducibility will result in uncertainty levels which render the conclusions meaningless. Whatever the other considerations may indicate, it is a simple fact of life that in virtually all trials the selection of properties will be affected by the cost and convenience of the experimental requirements. There are enormous differences between different properties in the cost of test piece preparation, testing time, number and size of test pieces, and apparatus requirements. In accelerated exposures the availability of exposure space is very frequently the limiting factor. The following few examples serve to illustrate the interaction of the factors affecting the choice of parameter. Hardness, modulus, tensile strength and elongation at break are usually chosen as having general applicability, whilst stress relaxation in tension is an example of a property which can be related to molecular change. Hardness is obviously a particularly simple and cheap measurement and the test piece takes little space. Both hardness and stress relaxation have the advantage that reproducibility is helped if the same test piece is used after successive ageing periods. For a particular application dynamic stress-strain properties might be very relevant and if Dynamic Mechanical Thermal Analysis (DMTA) is used it provides a lot of data at relatively low cost. Stress relaxation in compression is very relevant to sealing applications but is often substituted by compression set on the grounds (possibly erroneous) that the latter is cheaper. Electrical resistivity might be critical to an application but it can be relatively insensitive to some degradation agents and suffers from poor reproducibility. As a rule, the tests to monitor the amount of degradation are carried out at standard laboratory temperature. Obvious exceptions are low temperature property tests and monitoring continuous stress relaxation. If the service temperature of the product is significantly above ambient there is a strong argument for measuring the chosen properties, for example tensile stress-strain, at that elevated service temperature because the deterioration seen may be greater than that at ambient. The time dependent properties, creep, stress relaxation and set can be measured to monitor degradation, or as degradation tests that add mechanical stress. For details of the test methods used to measure physical properties reference is made to Physical Testing of Rubber [12]. Standard tests have their limitations; most were intended for quality control rather than prediction of service performance and produce arbitrary rather than fundamental measures of the properties. They do have the advantages of making data compatible with others and often have known reproducibility. In many
32
Parameters to Monitor Degradation standard methods the user is encouraged to opt for standard or preferred conditions which may not have relevance to the service conditions of the product. It is then sensible to base the testing on standard methods but to use more relevant conditions of, for example, time, temperature or stress.
8.2 Tensile stress-strain properties In measurement of tensile stress-strain properties, a test piece is stretched to breaking point, and the force and elongation are measured at different stages. For general evaluation, tensile stress-strain properties are almost universally selected as test parameters for ageing, weathering and exposure to liquids. They are also widely used for measurement of heat resistance and assessment of antioxidant protection. They have the advantage of relative simplicity, several useful parameters are obtained from one test and they use a thin test piece so that oxygen and fluid diffusion problems are minimised. Also, tensile stressstrain properties are almost always included in rubber product specifications. Dinzberg and Bond [13] make an interesting case for rejecting tensile strength as a parameter on the grounds that it may increase or decrease and, within limits, it is not very relevant to service. There were certainly cases of increases as well as decreases in the 39 compounds subjected to accelerated ageing at Rapra, with instances of the direction changing with temperature. Several workers have advocated the use of the product of tensile strength and elongation at break [14]. Modulus (stress at a given elongation) is very relevant where the stiffness of the product is important. The determination of tensile stress-strain properties is conducted in accordance with the standard ISO 37 of the International Organization of Standardization (ISO) and the values that can be obtained are illustrated in Figure 4.
8.3 Hardness For polymers, the term hardness refers to a measure of resistance to indentation. The economical use of material, the almost non-destructive nature, cheapness and simplicity make hardness measurements very attractive for monitoring degradation. Because repeatability and reproducibility with durometers (force applied by a spring) can be relatively poor, the dead load methods are preferred. A micro hardness method allows thin test pieces and can also be used to examine degradation as a function of thickness to detect any effect of limitation in oxygen diffusion. Hardness is essentially a measure of stiffness and can be related to the elastic or Young’s modulus of the rubber.
33
Practical Guide to the Assessment of the Useful Life of Rubbers
Figure 4 Typical stress-strain curve
The determination of dead load hardness, including micro hardness, is performed in accordance with ISO 48.
8.4 Stress relaxation Stress relaxation is the measurement of change of stress with time under constant strain. Stress relaxation measurements are normally made in compression or tension with there being a distinction as regards the reason for making the test. The most important type of product in which stress relaxation is a critical parameter is a seal or gasket and, as these usually operate in compression, stress relaxation measurements in compression are used to measure sealing efficiency. Although the importance of measuring decay of sealing force with time has been long appreciated, the widespread use of such tests has been inhibited by the inherent instrumental difficulties. Considerable advances have been made relatively recently in apparatus and the standardisation of test procedures. Stress relaxation measurements can also be used as a general guide to ageing, and it is particularly relaxation due to chemical effects which is then studied. Such measurements are normally made in tension. The standard methods for determination of sealing force are given in ISO 3384. The basic principle of the method is that a test piece is compressed between platens to a constant
34
Parameters to Monitor Degradation strain, and the force exerted by the test piece is measured at intervals by applying a very small additional strain. A way of doing this is by using a jig where a small additional compression results in the top platen being just separated from the body of the jig and the force is then transmitted via a central rod to a force measuring device. Alternatively, the force measuring device can be continuously balancing the force exerted by the test piece. Either disk or ring test pieces can be used, the latter being particularly convenient when tests are conducted with exposure to fluids. The normal practice is for test pieces to be compressed with lubrication of the ends to allow slippage; the preferred compression is 25%. Distinction must be made between the three procedures given in the standard. In procedure A the test piece is compressed at the test temperature and all force measurements are made at that temperature. In procedure B compression and force measurements are made at 23 °C, the test piece being subjected for intervals to the test temperature. In procedure C the compression is applied at 23 °C and force measurements made at the test temperature. With procedure A, there may be difficulty with some designs of apparatus in loading hot, and with procedure B, an apparatus with a large thermal capacity may take a long time to cool. With large test pieces there will be oxygen diffusion limitation effects. Gillen, Celina and Keenan [15] describe measurements using multiple small disks strained in parallel and clearly demonstrated this effect. They also note a large difference between the rate of force decay for standard test pieces and drop in elongation at break, and discuss the separation of physical and chemical effects. Stress relaxation measurements in tension have advantages as a general guide to ageing performance by reducing or eliminating some of the difficulties of conventional mechanical tests. Stress relaxation uses the same test piece throughout and results are more fundamental. Obtaining multi-point ageing data is very time consuming on separate test pieces. In addition, there is uncertainty as to the value of the simpler mechanical properties, such as tensile strength, in relation to service performance, and variability is increased by the use of separate test pieces for each point in the time and temperature sequence. Measurement consists basically of monitoring the stress in a sample whilst subjecting it to an ageing procedure, usually accelerated. There are two variants of the technique: continuous relaxation in which the sample is held stretched throughout the test and intermittent relaxation in which the sample is stretched only periodically for short times to enable measurements to be made. Under suitable conditions when viscous flow is not dominant, the reactions within the rubber molecular network may be related to stress changes. The decay of stress in continuous relaxation measurements provides a measure of the degradative reactions in the network, 35
Practical Guide to the Assessment of the Useful Life of Rubbers whilst intermittent relaxation measures the net effect of both degradative and crosslinking reactions. In the continuous measurement tests, any new networks formed are considered to be in equilibrium with the main network and do not impose any new stress. The intermittent measurements are, in effect, a measure of the change in stiffness with time, with the advantage over the standard tensile measurements that low strains more compatible with service conditions are used. Very thin test pieces eliminate the effect of any limiting rate of oxygen diffusion and the same test piece is used for all measurements at a given temperature. The intermittent technique is attractive but experimentally fairly difficult and special relaxometers are complicated and expensive. Normally, relaxation measurements are made at the ageing temperature. In a more simple test, modulus measurements can be made at ambient temperature whilst ageing takes place at elevated temperature, which enables tests to be carried out with a normal tensile machine. The International Standard, ISO 6914, covers both the continuous and intermittent procedures plus the simplified intermittent method. The term stress relaxation, although commonly used, is not adopted in this standard on the basis that increases in stress, as occur with intermittent tests, cannot be called relaxation. Strip test pieces are used, 1 mm thick to minimise oxygen diffusion effects.
8.5 Set Set is a measure of the recovery after removal of an applied stress or strain. Relatively simple apparatus is required which is probably the main reason why the rubber industry has traditionally paid more attention to set than creep or stress relaxation. Also, it appears at first sight that set is the important parameter when judging sealing efficiency. Set correlates with relaxation only generally and it is actually the force exerted by a seal that usually matters, rather than the amount it would recover if released. However, set is important when movement of the joint opens up a leakage path. Set tests are made in either tension or compression and for their prime use, quality control, the choice of mode can be made according to convenience and the test piece available. If intended to simulate service conditions, e.g. indentation of flooring, the most relevant mode of deformation would be used. Tests can be carried out in which the test piece is subjected to either constant stress or constant strain, but the latter is by far the most widely used. For a constant strain test, the test piece is more or less instantly compressed (stretched) and held at that compression (extension) for a fixed length of time. The test piece is 36
Parameters to Monitor Degradation released and its recovered height measured. It is common practice to measure the recovered height 30 min after release of the test piece but this is an arbitrary time. The term permanent set is sometimes used but if this has any meaning it would be referring to the set remaining after an infinite recovery time. Set is normally expressed as a percentage of the applied deformation, i.e.
Set = where:
t0 − tr t0 − ts
× 100%
t0 is the original thickness of the material tr is the thickness after recovery ts is the compressed thickness
Compression set methods are given in ISO 815 and tension set in ISO 2285. Both cover only constant strain. In ISO 815 disk test pieces are used and the compression is made between very smooth platens which are lubricated. Hence, the compression is made with some attempt at perfect slippage. Fairly obviously, the degree of slip and the test piece shape factor can affect the measured values of set. Tension set measurements can use strip, dumbbell or ring test pieces. The arbitrary conditions specified are intended for quality control. If set is to be used as a measure of performance, it is necessary to test under conditions relevant to service which may, for example, involve recovery at the test temperature and measuring set as a function of time and/or temperature.
8.5 Dynamic stress-strain properties Dynamic properties are measured by mechanical tests in which the rubber is subjected to a deformation pattern from which the cyclic stress-strain behaviour is calculated. This does not include cyclic tests in which the main objective is to fatigue the rubber. Dynamic properties are more relevant than the more usual quasi-static stress-strain tests for any application where the dynamic response is important. The dynamic modulus at low strain may not undergo the same proportionate change as the quasi-static tensile modulus. Dynamic properties are not measured as frequently as they should be simply because of high apparatus costs. However, the introduction of DMTA equipment has greatly improved the ease of dynamic property measurement. These analysers exist in many forms but are essentially relatively small bench instruments, which use small test pieces and can be programmed to measure damping and dynamic moduli as a function of temperature and frequency. Apart from their 37
Practical Guide to the Assessment of the Useful Life of Rubbers importance for measuring the dynamic properties where these are relevant to service, they allow the generation of a large quantity of data over ranges of temperature and frequency, extremely efficiently. They can be used effectively to obtain the modulus even if the application is not dynamic. However, like intermittent stress relaxation, strength properties are not obtained. Another valuable use is to obtain glass transition temperatures. The dynamic moduli are related by: E = E′ + iE′′ where:
E′ is in phase or storage modulus E′′ is out of phase or loss modulus E is complex dynamic modulus i is the mathematical symbol for the imaginary number the square root of -1
and
Tan δ = E′′/E′
Although the apparatus is expensive, DMTA produces a lot of data in a short time and is hence a very efficient way of generating information on dynamic properties. Furthermore, it has the advantage of being non-destructive and uses a small test piece. Dynamic properties have not been very widely used to monitor ageing but this is mostly because of the apparatus cost, and the potential advantages of DMTA are very clear. It should, however, be noted that the strain attainable in dynamic analysers is limited and for measurements under service conditions, particularly for testing products, larger servo hydraulic test machines are preferred. ISO 4664 is a guide to measuring dynamic properties of rubbers.
8.6 Volume change In the case of exposure to liquids, the change of volume is universally used as the basic guide to the resistance of a rubber. The more liquid absorbed the greater the effect on mechanical properties as well as the obvious effect on dimensions. However, more specific information will be gained by also monitoring other physical parameters. The measurement of volume change is included in ISO 1817 for exposure to liquids and is also considered further in Section 18.
38
Parameters to Monitor Degradation
8.7 Other properties As mentioned earlier, the properties which should be measured to monitor ageing resistance are those of relevance to the intended application. However, in practice only a few of the most common physical tests are used, largely because of cost restrictions. Electrical properties are only of relevance for such applications as cables and antistatic products, but then they may be the most important factors. Those most likely to be needed are: Resistance or resistivity Power factor and permittivity Electric strength. It is possible that electrical properties could be useful for more general evaluation of degradation, but experimental difficulties and, often, poor reproducibility, has not made them attractive. Methods for determining resistivity of insulating materials are based on standard IEC 93 of the International Electrotechnical Committee, and insulation resistance of rubbers is covered by ISO 2951. For conducting and antistatic rubbers ISO 1853 is used for measuring resistivity and ISO 2878 for resistance. The general method for determining electric strength is IEC 243 and methods for power factor and permittivity are often based on IEC 250. Properties such as friction, gas permeability and thermal conductivity are only likely to have significant importance in rather specialised applications and it is not surprising that they are very rarely measured in ageing studies. Determination of frictional properties is covered by ISO 15113, gas permeability by ISO 2782, water vapour permeability by ISO 2528, vapour permeability generally by ISO 6179 and thermal conductivity by BS 874. The effect of degradation agents on low temperature behaviour must be relevant in many applications but is virtually never measured. It would seem that deterioration of low temperature resistance as a result of ageing has not often proved to be a problem. However, leaching of plasticisers would raise brittleness temperature and oxidative ageing of natural rubber can actually enhance resistance to crystallisation. Determination of low temperature properties for rubbers is covered by ISO 2921 (recovery), ISO1432 (stiffness) and ISO 812 (brittleness).
39
Practical Guide to the Assessment of the Useful Life of Rubbers In contrast to stress relaxation and set which are often proposed for measuring ageing (although less often actually used), creep is rarely measured on rubbers. This is simply a reflection of the limited number of applications where creep is an important factor and, particularly, the small number of cases where chemical ageing is expected to seriously affect creep performance. Notable exceptions are anti-vibration mountings and bearings. Creep measurement is covered by ISO 8013. Abrasion and fatigue are important properties for a great many products and will be significantly affected by ageing. The reasons why they are not often measured in ageing trials is, firstly, the relatively high cost and time factors involved and also because the standard laboratory methods are not renowned for correlating well with service. It can, hence, be argued that an indication of ageing effects can be deduced from simpler mechanical properties. However, they are very important mechanical degradation agents and are covered in Sections 21 and 22. The reason why tear strength is very little used to monitor ageing is certainly because tensile strength will serve perfectly well. Compression stress-strain properties are, surprisingly, much less often measured on rubbers than tensile modulus or hardness and they are even less often used to monitor ageing. There is, however, a good practical reason in that the relatively bulky test pieces required will be subject to limitations of oxygen diffusion in any accelerated tests and changes should be reflected by tensile modulus measurements. Similarly, shear stress-strain is very rarely used for monitoring ageing. Tear strength measurement is covered by ISO 34, which includes the small Delft test piece. Compression stress-strain is covered by ISO 7743 and shear by ISO 1827. For bonded components there is the need to monitor bond integrity. There are several standard adhesion tests for polymers: ISO 813, 90 degree peel of rubber to metal; ISO 814, direct tension of rubber to metal; ISO 1827, shear of rubber to metal; ISO 36, 180 degree peel of rubber to fabric; ISO 4637, direct tension method for rubber to fabric and ISO 2411 for coated fabrics. There are also specialised methods for tyre cord in ISO 4647 and ISO 5603. In practice, these quasi-static tests are limited in that they may not adequately distinguish between a good and a very good bond. Generally, it is preferable to devise an appropriate fatigue test or, in some cases to use creep or impact. If a proven non-destructive method for bond assessment is available this has obvious advantages. Unvulcanised compounds are a special case where effects of ageing can be assessed by using the processability tests such as plasticity, viscosity and cure metering. Logically,
40
Parameters to Monitor Degradation the methods used would be those applied in the factory for quality control of that compound. Visual inspection is always worth including as it costs little and is, of course, particularly important when strained test pieces are exposed to detect ozone cracking. Instrumental measurements of colour or gloss are not normally useful for rubbers. A somewhat indirect approach to investigating resistance to ageing is to use differential scanning calorimetry (DSC) to measure the induction time of oxidation. In DSC the differential in heat input to maintain the test sample and a reference sample at the same temperature is monitored. The sample is heated to a given temperature under nitrogen then the atmosphere is change to oxygen and the temperature maintained. After an induction time while antioxidant is being used up, the onset of oxidation is seen as an exotherm. This method finds value for evaluating antioxidants in plastics but has been applied to rubbers only rarely. Procedures for predicting the thermal endurance of polymers have been evaluated. Budrugeac [16] examined the use of differential thermal analysis (DTA), DSC and chemiluminescence to obtain activation energies from isothermal thermal analysis data. He found large discrepancies between activation energy values obtained from these analytical methods and the conventional procedures of the standard IEC 216. Another indirect method to monitor ageing is to measure the uptake of oxygen. At one time this gained some popularity but appears to be little used nowadays. It has to be assumed that all the oxygen absorbed is being used to cause degradation and correlation with mechanical property change is generally only successful with very similar compounds. Wise et al. [17] demonstrated the very high sensitivity of the technique and were able to measure uptake at service temperatures and relate this to oxygen uptake and mechanical properties at normal ageing temperatures.
8.8 Functional tests There are clear advantages in obtaining direct evidence of performance by exposing the total product in ageing tests. This is particularly so in cases involving complex degrading environments and critical applications. Unfortunately this is rarely possible, particularly for accelerated tests, due to limitations of exposure space and costs. When it is possible, it is better to use functional tests on the product to assess the environmental effects, rather than the standard methods for materials. Increasingly, product specifications include such tests but in many cases it would be necessary to devise methods for the product in question (see Section 12).
41
Practical Guide to the Assessment of the Useful Life of Rubbers
8.9 Chemical analysis It should be appreciated that, in the previous section, the physical tests discussed are being used to measure the effect of chemical changes. Monitoring chemical changes allows more direct measurement of the fundamental mechanisms causing degradation and is essential for studies aimed at understanding the degradation processes. Chemical analysis methods also generally require quite small samples and in many cases are relatively quick to perform. Hence, they also potentially offer more cost-effective ways of monitoring changes during ageing trials. However, the instrumentation required can be very sophisticated and is available in relatively few laboratories. The measurement of chemical change does not give a direct indication of the effect on performance, which is what is generally wanted by design engineers and for compliance with performance specifications. For these reasons, and because of the lack of established correlations with physical property changes, chemical analysis methods are far less frequently used than physical testing methods. The analysis carried out in the Rapra project on 40 year aged samples found, using the more usual techniques at least, that the results of analysis could not correlate with physical property changes and, hence, this was not a viable approach to monitoring ageing.
42
9
Preparation of Test Pieces
The mixing of the rubber compound, the method of preparation of test pieces (or products) and their subsequent history will all influence the material properties. Consequently, it is important for any testing that all the preparation and conditioning procedures are defined and carefully controlled. The time and temperature of cure require especial consideration as a given cure may not be optimum for all properties and there may be a significant difference between the optimum cure and that used in practice in production. It may be worthwhile conducting preliminary trials using several cure conditions. There are three particular considerations for durability tests: Any errors could be particularly expensive, at least over time The timescale of testing is normally very long Variability of results is magnified by any extrapolation procedure. It is obvious that an error in mixing discovered months into an ageing programme will not have trivial consequences. For short ageing periods it is good practice to test both aged and unaged samples together to minimise variability. For long ageing periods this would mean holding the unaged samples for times over which degradation even at ambient conditions may not be negligible. If the test pieces exhibit the best possible repeatability it will help to minimise the uncertainty in the predictions made. Most tests will be made on products or standard test pieces. However, it is worth considering the influence of test piece size. Firstly, differing cure cycles will be necessary to reach the same state in different thickness mouldings. Apart from cure effects, there is a general tendency for larger test pieces to fail sooner than small ones, because of the greater chance of flaws. For example, this might be seen in the area used in an electrical breakdown test, and there is also evidence of this effect for tensile dumbbells. Thick sections being flexed may simply experience higher strain. Clearly, quite the reverse is true for other parameters, for example, the time dependent effects discussed in Section 6 and the amount of material to wear away in abrasion situations.
43
Practical Guide to the Assessment of the Useful Life of Rubbers
44
10
Uncertainty and Application of Statistics
The uncertainty of a measurement is always a very important consideration but it is especially so with durability tests. In addition to the basic physical test methods, there are the complications of exposure conditions, tests spanning long times and the process of extrapolating results to make predictions. The combination of these factors will inevitably lead to large uncertainties that can very easily be of such a magnitude that any conclusions are meaningless. There are two terms used to describe precision: repeatability refers to within-laboratory variations; reproducibility refers to variation between laboratories. All test results are subject to uncertainty arising from the instruments, the procedure and material variability. It is good practice to make estimates of the uncertainty for the test methods used to monitor degradation using established procedures, and this is considered essential in accredited laboratories. Guidance on the estimation of uncertainty is given in the ISO ‘Guide to the expression of uncertainty in measurement’. Good indications of the order of repeatability to be expected can be obtained from the precision statements found in some test method standards, although the uncertainty in each case depends upon the instruments and procedures in the particular laboratory and the particular material in question. Every precaution should be taken to minimise the basic uncertainties arising from instrument calibration and test procedure. Often, the uncertainty due to material variability is the dominating factor and strict control should be applied to the production of materials and test pieces. The number of replicates used should be maximised within the limits of time and cost available. Further uncertainty arises from the environmental exposure. For accelerated tests, repeatability can be estimated from exposure of replicate test pieces and minimised by control of the exposure conditions. Particular points to consider are spacial variation in temperature, mean temperature and air flow in ovens. In accelerated weathering apparatus, spacial variation and variation with time of light sources can be very significant. When the test programme spans long times, the situation is more akin to reproducibility conditions than to repeatability, even though the measurements are made in the same laboratory. The increase in uncertainty can be illustrated by comparing the repeatability and reproducibility figures given in precision statements.
45
Practical Guide to the Assessment of the Useful Life of Rubbers In natural exposures there will be variations in conditions which are completely uncontrollable and will lead to very large differences between different occasions and places. It then becomes meaningless to attempt to estimate uncertainties. Comparing accelerated results obtained in different laboratories with different apparatus (particularly for artificial weathering) is similar in some ways, because of the large variations which are often found due to poor reproducibility. It is always advantageous to include a known reference material if one is available. Differences are generally much reduced if the ranking of materials is considered rather than absolute values. When extrapolation of measured data is carried out all the uncertainties become magnified, increasing as the degree of extrapolation is increased. Inevitably, extrapolations from accelerated tests to normal ambient conditions will be subject to enormous uncertainties, which is why the general advice for temperature is to extrapolate to 30–40 °C beyond the last data point at the very most. Additional to this, but generally not quantifiable, is the uncertainty of the validity of the extrapolation model. Confidence limits of the regression line from which predictions are made can be calculated. For example, a procedure is given in IEC 216 (‘Guide for the determination of thermal endurance properties of electrical insulating materials’) for an Arrhenius plot. Statistical principles should be applied to the planning and analysis of all test programmes. Details of commonly needed techniques are given in BS 903 Part 2, ‘Guide to the application of statistics to rubber testing’. The basics of design of experiments are included and the standard has a bibliography. In ageing trials the most important considerations are the number of test pieces, exposure times and levels of the degradation agent, together with ensuring that the test pieces used are representative. Ideally, the number of experimental points should be maximised but in practice cost considerations lead to restrictions. Techniques such as the application of factorial designs to experiments are normally only applicable where various compounding changes are being evaluated, or if several levels of more than one environmental agent are included in the experiment. The application of statistical techniques generally requires somewhat tedious calculations. The analysis of results to make predictions is likely to involve curve fitting and the application of relations specific to a particular predictive model. This is most easily achieved with software packages such as SigmaStat (general statistics) and TableCurve (curve fitting) from Science Software UK. In the absence of such software many useful functions are included in spreadsheet programmes such as Microsoft Excel. For the
46
Uncertainty and Application of Statistics Arrhenius and WLF (Williams, Landel and Ferry) relations (see Section 25) simple software is available at Rapra. More sophisticated software for the Arrhenius relation has been developed for use with IEC 216. The practical conclusions from this Section are that for durability trials: • Extra attention must be paid to minimising uncertainties but, even with all precautions, uncertainty of results is likely to be high. • Caution has to exercised with extrapolation because uncertainties are magnified. • Poor reproducibility can result in large differences between laboratories: consideration of the ranking of materials or comparison with a reference is often the best approach. • Results from natural exposures are subject to the large variations in natural conditions.
47
Practical Guide to the Assessment of the Useful Life of Rubbers
48
PART 2 PRODUCT TESTS AND EXPERIENCE
49
Practical Guide to the Assessment of the Useful Life of Rubbers
50
Experience
11
Simulating Service
11.1 General Putting new materials or designs into service to prove their suitability is generally not viable for environmental factors because of the risks and the timescales involved. However, extended real trials are sometimes used for critical products, normally after exhaustive laboratory testing has been carried out to give high confidence in the chosen materials. When, after various trials, a product has been put into service it can be very valuable to undertake condition monitoring. By examination or testing of the product at intervals it may be possible to detect potentially serious problems before they occur and to make estimates of remaining life. For rubbers, a restriction is the relatively few good nondestructive tests generally available. Simulating service conditions avoids all the risks of using real service and offers the possibility of moderate acceleration by simulating the worst conditions possible. However, the timescales will still be long and in many cases it is difficult, if not near impossible, to simulate real conditions accurately. It may be noted that one form of acceleration applicable to some products is to simply use the item more. Clearly, simulated service trials are most attractive where the expected lifetimes are relatively modest and the conditions to be simulated are not too complicated. The term simulated service trials implies that all factors present in service are considered (e.g. mechanical stress and environment) and some form of rig devised which subjects the product to these factors. The principles of product testing are considered in Section 12.
11.2 Natural environmental exposure For environmental resistance, another approach that can be considered as simulated service is so called natural exposure – the exposure of materials or products to natural environmental conditions. Such exposures, to what is commonly called weathering, provide information on the durability of the material without the effects of such factors as fatigue and abrasion. 51
Practical Guide to the Assessment of the Useful Life of Rubbers Natural exposure to any of the environmental agents can be carried out. For example, if the product is to spend its life in water at 70 °C then exposure to water at 70 °C can be considered natural ageing. Exposure to water at 80 °C could be called natural ageing at worst possible conditions or with a safety factor. Natural exposure defined in this way is carried out by adapting the standard methods for air ageing and exposure to liquids as there are no specific natural exposure standards for these agents. A particular type of natural exposure for rubbers is performed to assess ozone resistance. Again, there are no specific international test methods and the principles of the standards for accelerated exposure are followed. Commonly, products (rather than test pieces) are exposed and strain applied to simulate that in service. The degree of strain is then likely to be less precisely defined than in laboratory tests and the geometry of the straining needs to be chosen carefully. It will be appreciated that natural exposure for assessing ozone resistance will also include exposure to other agents, for example UV light, and is essentially a form of weathering test. As such, it is subject to the general problem of natural weathering whereby the conditions vary from place to place and season to season. For example, different results may be obtained in the UK in summer and winter with wax protected materials simply because of the effect of the temperature difference on wax solubility and diffusion rate. Natural exposure is most commonly thought of in terms of weathering – the prevailing conditions of temperature, sunlight, rain and possibly pollution. Standard procedures for rubbers exist in ISO 4665. This standard refers to the equivalent standards for plastics for apparatus and procedures, giving only details particular to rubbers. For natural weathering the relevant plastics standard is ISO 877 which also includes exposure under glass and ‘accelerated natural’ exposure using Fresnel mirrors to concentrate the sunlight. Further relevant standards are ISO 9845-1 giving reference to solar spectral irradiance, ISO 61725 on analytical expression for daily solar profiles and ISO 9370 on instrumental determination of radiant exposure in weathering tests. A good, up-to-date account of weathering tests on polymers is given in Handbook of Polymer Testing [18]. To cover the influence of various climatic conditions, studies are commonly carried out in relation to different classes of climate such as temperate, desert and wet tropical. Clearly, there are many levels of severity within each class. The position and angle of the test pieces to the sun (for both radiation and temperature) is important. A wide range of exposure methods and angles are available at most test stations. The more conventional and widely used are:
52
Simulating Service Vertical; samples are exposed at 90° to the horizontal and usually facing south, but may face east, west, or north, depending on test specifications. 45° facing the equator; samples are mounted so as to be exposed at a 45° angle to the horizontal and facing the equator. Horizontal; samples are positioned to expose the test surface to the weather on a horizontal plane. More advanced facilities allow a variable angle to track the sun. The method of mounting the test pieces has significant effects, in particular the kind of backing of any enclosure raising the temperatures achieved. Most often test pieces are exposed unstrained but strain should be applied if it occurs in service. Because UV light is not generally considered to be an important factor with black rubbers, there is nothing like the same amount of weathering data available as there is for plastics. The large difference between exposure sites has been well established for plastics and it has been demonstrated that the differences in severity do not necessarily follow the actual irradiance levels and average temperatures, but can be influenced by other local climatic factors, such as being near the sea, the proximity of forests or urban pollution. For example, exposure to a tropical maritime climate, a so-called moist tradewind climate, has been found to produce particularly pronounced changes to important properties of rubber articles. For temperate climates, acceleration can be achieved by exposure in parts of the world with more severe conditions of UV and temperature, but care has to be taken because spectral distribution as well as amounts of precipitation can be significantly different. Exposure can be made in conditions where UV light and rain are excluded, which is essentially measuring the effects of temperature, oxygen and humidity. This is relevant to storage of products and to cases where the product is used indoors. The Rapra 40 year ageing trials were carried out in these conditions and demonstrated that performances in climates of differing severity do not always correlate with the climate. Exposure under marine conditions is a particular case which can be especially complex. Apart from irradiance levels and temperatures being different, where the salt water has significant effect, the frequency and duration of immersion can be critical. A standard for marine exposure of plastics is being developed as ISO 15314. It is essential that all the conditions of exposure are recorded so that the results can be put in perspective. Preferably, the environmental parameters of temperature and radiation should
53
Practical Guide to the Assessment of the Useful Life of Rubbers be monitored continuously so that appropriate forms of average, maxima and minima can be derived. Some guidance is given in the standards and text book referenced above. The important considerations in natural weathering tests can be summarised as: Because of natural variability, the results of any trial pertain only to the particular conditions at the time. All meteorological parameters should be recorded to assist in understanding the relevance of the results and their comparison with other data. The aspect of exposure is important in respect of the irradiance, temperature and the effect of rain. The construction of exposure racks will influence the effects of the environmental factors such as temperature of the test pieces. Whether or not test pieces are strained can be critically important. Natural exposure can only be used when the timescale is viable. The standard procedures are particularly suited to products that are not subject to significant dynamic strain. Also, natural (as well as accelerated) ageing can be useful in conjunction with accelerated mechanical tests on products as explained in Section 12.
11.3 Simulated design life Estimates of service life are usually made either by carrying out natural or simulated trials or, most commonly, by performing accelerated tests with extrapolation to predict performance at longer times under less severe conditions. An alternative approach is to subject the product to environmental exposures which equate to a chosen design life and then to assess performance by real or simulated service tests. The exposures usually have to involve accelerated procedures and can be composed of several environmental agents applied simultaneously or sequentially. These are circumstances in which this approach can be particularly useful. For example, where the end performance assessment can be made by operating the product, and where a safety critical product is to be subjected to abnormally harsh conditions (either for the environmental exposure and/or the end assessment). It is also attractive for particularly complex products. Clearly, this approach begs the question as to how accurately the lifetime exposure can be simulated, and makes the assumption that a valid extrapolation procedure is known
54
Simulating Service and that the necessary input data has been determined. This apparent chicken and egg situation can be resolved by arguing that the data needed may be known to a sufficient approximation, perhaps with a safety factor, for the most critical material in the product. Also, there need be no data on how the actual product manufactured from the material performs after ageing as this is determined by the end performance assessment. The process can be illustrated by a very simplistic example. It can be claimed that from accelerated ageing test results, reaction energies (via the Arrhenius relation) for mechanical properties of a rubber are known at least approximately. From this, the product could be given a simulated lifetime of, say, 10 years by heat ageing and then actually operated under service conditions. Using a tyre as an example, it might be found that although the mechanical properties after a simulated 10 years were reasonable they deteriorated very rapidly in heavy service, because the additives incorporated to protect against high running temperatures had been depleted. Hence one would be advised not to use tyres stored for long periods even if they seemed intact. A more complex sequence could be applied, for example, to nuclear plant where sequential applications of temperature and radiation doses to simulate different operating phases could be made. For some products it would be appropriate to apply mechanical fatigue simultaneously with the ageing. A potential problem to consider is that, with several materials making up a product, using the reaction energy for the most critical material could result in another material being over-aged, which is consequently the element to fail. Simulating service life in this manner uses the same basic concepts as the more usual accelerated testing trials, but the differences are the way in which the accelerated procedures and predictions are applied and in the concept of applying service tests after exposure.
55
Practical Guide to the Assessment of the Useful Life of Rubbers
56
12
Experience
As suggested earlier, it is probable that the majority of rubber products are designed largely on the basis of experience. A great many product specifications do not include real performance tests, but are based on general material properties using tests which do not yield fundamental results. This is particularly so for environmental resistance where the requirements specified are normally single point under somewhat arbitrary accelerated conditions. Products made to such specifications are satisfactory in practice, essentially because experience over many years has shown that materials meeting the minimum requirements work, not because any scientifically based predictions have been made. Indeed, the data obtained using the specification tests alone would be inadequate for any extrapolation to service conditions. The problem arises when significantly different materials are introduced, new products are envisaged or the conditions of service change. Then the experience pertaining to an established specification is no longer valid. However, in such a changed situation there are several sources of experience which can be usefully tapped: Data generated by others General knowledge of materials Knowledge of the same material in other applications Knowledge of how other materials performed in similar applications Predicting from known performance to different conditions Comparison of new material with one proven. It is always worthwhile to search the literature for data generated by others before embarking on a durability trial. Even if no totally suitable data is available there may be information which can add to, or help substantiate, your own results. Examples of substantial published data are the Rapra 40 year natural ageing trials [19], the accelerated testing programme conducted in association with those trials [20, 21] and a Swedish collection of ageing data [22].
57
Practical Guide to the Assessment of the Useful Life of Rubbers The rubber technologist should have a good idea of the general environmental resistance of materials, which will at least enable the potential candidate materials to be short listed and clearly unsuitable materials eliminated. There is clearly uncertainty in relying on knowledge of the same material in different applications. To what extent it proves useful (successful) will depend on the closeness of the different circumstances and on the validity of the reasoning used to predict to the new circumstance. If performance is known at one set of conditions it may be a relatively modest step to predict performance under different conditions. As an illustration, if performance has been proven at say 60 °C, the rule of thumb that reaction rates for rubbers will approximately double for each 10 °C rise in temperature, might be applied to estimate performance at 70 °C. Knowledge of the performance of one material can be utilised by making relatively shortterm tests comparing the new material with the established material. In many cases it is not unreasonable to take the comparative results as an indication of likely service behaviour. Analysis of previous failures is a powerful source of experience which helps particularly with determination of critical agents, parameters and synergistic effects. Unfortunately, accounts of product failures are, perhaps not surprisingly, infrequently made public. However, a compendium of failures has been compiled which details real life case studies of a range of rubber and plastic products [23]. This book is particularly instructive in illustrating how mistakes and overlooking a particular factor can cause disaster. A recent Rapra Review Report [24] covers long-term and accelerated ageing tests on rubbers and contains a large number of references to literature relevant to assessing useful life. The message is that existing knowledge and its possible transfer to new circumstances should be considered. In many cases this will be no more uncertain than predictions from extensive accelerated trials.
58
13
Principles of Product Testing
13.1 Introduction The great majority of durability tests (indeed all types of test) are made on test pieces rather than on the complete product. There are several reasons for this, not least the cost of developing specialised product tests and, for durability trials, the exposure space required. This section examines the principles of when tests should be made on the product as opposed to test pieces and considers the approaches to developing such product tests. The basic argument for product tests is that the proof of the pudding is in eating the pudding and not tasting the quality of the ingredients. This applies to durability as much as to any other parameter. Testing the materials is valuable in its own right and, if you specify the materials and then put them together correctly, the product should work. There are, however, limitations to this philosophy because putting the product together may change the nature of the ingredients, and the behaviour of the ingredients individually does not tell you how the product will behave as a whole. To predict product performance from material properties, additionally requires protocols and design rules to relate the properties of materials to service. There are in fact three possibilities as, in addition to testing the materials or the product, in many cases tests can be made on test pieces cut from the product. This has the advantage that the material properties measured are those that relate to the material as processed in the factory rather than to those on test pieces prepared under laboratory conditions. The only disadvantage is the limitations to obtaining suitable test pieces from many products.
13.2 When to test products If our knowledge of the properties and behaviour of polymers, and hence our design rules, were such that we could predict the performance of the product accurately from tests on laboratory test pieces then product testing would be rarely needed. Tests which yield fundamental data in a form that is independent of test conditions and test piece geometry are unfortunately rather rare. Hence, the fact that our understanding is far from perfect means that there will often be a need to test the whole product since this is the only way to be sure that it will perform satisfactorily. 59
Practical Guide to the Assessment of the Useful Life of Rubbers It follows that the main reason for testing the product is to establish fitness for purpose, i.e. performance testing. Durability testing to estimate service life is performance testing. It was observed in Section 11.1 that putting products into service to prove their suitability as regards environmental factors is generally not viable. However, no simulation test will reproduce service perfectly and proving in the field will always give the greatest confidence. It is worth noting again that even when accelerated tests are used for estimating durability, the exposure of prototype products to natural service does mean that you will know of any unforeseen failures before they occur in products sold. It is generally very clear when it is desirable to test the whole product. What is usually much more difficult is to weigh up the risks and the information gained against the costs of testing. Arguably, when tests are made to estimate the service life of a particular product they should always be made on that product. However, it has to be accepted that this will frequently be impossible in practice. By definition, trials for simulated service life as discussed in Section 11.3 always require the use of products. Where test pieces are being used for quality control or to estimate the performance of the product it is always preferable to use pieces cut from the product, but again this is often impracticable. The cost of a basic accelerated heat ageing programme using test pieces is by no means insignificant as it involves a series of temperatures with, usually, several properties being measured after a number of exposure times at each temperature. Tests on products can involve much higher costs. It can be extremely difficult and/or expensive to devise tests to simulate performance adequately and justification for the investment will be in proportion to the importance of the product in risk and/or sales terms. Service conditions are almost inevitably complex and include mechanical and environmental stresses over an extended time period. There is clearly much skill involved in designing rigs and test schedules which give maximum information at minimum cost. In practice there is a danger of spending very large amounts and still not getting the simulation accurate enough, although most commonly, the pressure is to under-design the apparatus and curtail the programme to cut costs.
13.3 Design of product tests There are legions of product tests, about as many as there are products, and the trend is for more product specifications to include performance tests on the product. Almost by definition, product tests are devised to suit a particular part and application. A fair
60
Principles of Product Testing number have been standardised but even then many show their ad hoc origins. The range of sophistication goes from the extremely simple to the very complicated. Thus it is only possible to deal here with some of the principles and look briefly at a few examples. Generally, one cannot simulate full service conditions in one test and it is necessary to restrict the scope to one bit at a time, or things get excessively complicated. For example, loading that varies with time of use, plus abrasion from external sources and environmental ageing in general. Car tyres have a complicated pattern of cornering, sports surfaces are subject to the actions of different sports and pipes in the North Sea have almost unpredictable wave motions superimposed. The usual starting point is to look at how the object is stressed in service and simulate it – flex a belt, compress a mounting, pressurise a hose, etc. The more complicated the service stresses the more complicated the simulation will become – a mounting may be compressed and sheared, which has to be tested simultaneously, whereas a conveyor belt may be flexed and impacted which can be done sequentially. It is worth noting that sometimes a product test will give a more valuable assessment of quality for the same testing cost as needed for test pieces. This would be true, for example, for compression of a simple mounting because the cost of moulding test pieces would be little different from the value of the mounting and the testing costs would be equal. Compressing the mounting would actually be cheaper than cutting standard compression test pieces from it. More often than not an action is repeated in service many times and the test is made to do likewise, i.e. we move to having a fatigue test. The frequency of repeated action can usually be higher than in service, and hence a degree of acceleration is achieved, but care must be taken that the frequency is not such as to cause behaviour which would not occur in practice. It should be remembered that a fatigue test will only compound any inadequacy in the choice of stresses imposed and their amplitude. In real life the stressing of a product, perhaps repeated, takes place in what might be termed an aggressive atmosphere. This means exposure to any of the environmental agents listed earlier or some combination. This is where the test design can get especially complicated. All the problems of ageing tests apply, the uncertainty and variation in service conditions, the timescales involved, the validation of any accelerating process and the extrapolation to service conditions. To add environmental effects will escalate the costs and the uncertainty of truly matching service rises steeply.
61
Practical Guide to the Assessment of the Useful Life of Rubbers To help lessen the complications of product testing, a fairly common approach is to test mechanical properties on a relatively simple product test rig but to make degradation studies on the material or materials. It can then be argued that a given degree of deterioration in the mechanical properties will result in a proportional reduction in the product performance. Alternatively, the product can be subjected to an ageing process and subsequently tested for mechanical performance – this is the procedure used in so called simulated design life (see Section 11.3). The limited range of application of any product test rig often means that very few are built. This, together with the costs and technical difficulties, results in many cases in the validity and reproducibility of the test being inadequately investigated. If tests are included in national or international performance specifications before proper evaluation, problems of interpretation and differences in results are likely to arise.
13.4 Examples of test rigs The following examples are far from exhaustive but serve to illustrate some of the possible approaches. They demonstrate the variety, and give some indication of the levels of complexity, of product test rigs in use. Even when quite simple, product rigs tend to be large and limited in their application, making them a relatively expensive proposition.
Seal testing rig Seals are mounted in a jig as in service and pressurised cyclically with the service fluid. The jig is contained in an enclosure so that elevated or subnormal temperatures can be applied. This is an example of a fairly comprehensive, composite test which includes both mechanical and environmental stressing. The degree of acceleration can be varied for both mechanical and environmental factors. The same rig can be designed to accommodate different seals.
Ramp friction test (Figure 5) An operator walks on a ramp the angle of which can be varied. Either the floor surface on the ramp or the shoes can be the product under investigation, the other being a reference material. The angle at which the person slips with a given shoe/floor surface combination is measured and is related to the coefficient of friction. Wet, dry, the effects of cleaning, abrasion, ageing and contamination can be investigated.
62
Principles of Product Testing
Figure 5 Ramp friction test
This is a case of a single property being measured as a function of different conditions. It is clearly specific to floors or footwear.
North Sea tank Service conditions are simulated by placing the product in a tank of sea water at controlled temperature. In a test on the insulation of riser pipes, hot oil is circulated through the pipe at the same time. Such a test is an example of a fairly simple environmental test bed for specific circumstances. The basic structure of such a rig is applicable to a considerable range of products and can involve acceleration by using more severe conditions than service.
63
Practical Guide to the Assessment of the Useful Life of Rubbers
Stressing frame (Figure 6) The product is subjected to a predetermined mechanical stressing pattern by means of, for example, hydraulic cylinders. Strength, stiffness, creep, etc., can be monitored with suitable instrumentation. This is a multi-purpose mechanical test bed where the basic frame can be adapted to test virtually any product. The scope of a rig varies considerably depending on its size and the sophistication of the loading and measuring instrumentation. Fatigue can be incorporated but adding environmental effects is rare.
Sports surface test rigs (Figure 7) The performance characteristics of sports surfaces – ball bounce, energy absorption, friction, spike resistance, etc., – are measured at ambient conditions on separate rigs. Most of the rigs in use have been developed specifically for synthetic surfaces, and sometimes for specific sports.
Figure 6 Stressing frame
64
Principles of Product Testing
Figure 7 Sports surface test rigs a) Impact absorption; b) Artificial athlete; c) Pendulum friction; d) Indentation
This approach entails taking one property at a time without superimposing any ageing. Where environmental effects must be accounted for, ageing is carried out separately and the rig tests repeated.
Tyre test machine (Figure 8) The loaded tyre is run against a revolving drum at given velocities, and the integrity of construction and heat build-up noted. This is a fairly simple fatigue simulation rig which is unusual in that the environmental effect of temperature is self-generated.
65
Practical Guide to the Assessment of the Useful Life of Rubbers
Figure 8 Tyre test machine
13.5 Summary The performance of a product is best demonstrated by tests on the complete product. The limitations to this are the difficulties and the cost of devising satisfactory product tests. In consequence, the decision to use a product test has to be made by weighing up the risks against the costs and effort required. When using test pieces to give an indication of product performance, it is preferable to cut test pieces from the product. The cost and complexity of product tests increases sharply with any increase in the scope and number of parameters included. Because of costs, technical difficulties and few rigs being built, the validity and reproducibility of product tests are often not fully evaluated.
66
PART 3 ACCELERATED TESTS
67
Practical Guide to the Assessment of the Useful Life of Rubbers
68
14
General
14.1 Purpose of accelerated tests This Guide is concerned with assessing the useful life of rubbers and accelerated tests are an extremely important means of doing this. However, the results of accelerated tests are not always used to directly predict service life. Indeed, the majority of accelerated tests are carried out for quality control purposes, to show conformity with a specification or to make a comparison of materials. It is important that the purpose of any accelerated test programme is clearly established because this has a large bearing on the design of the trials and the interpretation of the results. As a generality, tests where the objective is to directly make predictions from the results are a great deal more difficult to design and will be considerably more costly than tests for the other purposes. Where the purpose is simply to make a comparison or to check against a specification the value of the results are less critically dependent on the relation of the test conditions to service. This does not mean that the test conditions are unimportant for comparison or quality checks, but that their relevance to service need only be established in general terms and not proven rigorously. The essential requirement for comparison is that the test conditions are not such that they give a distorted view of relative performance. For quality control it is particularly important that the test procedures and conditions are standardised and reproducible. When the results of an accelerated ageing programme are intended to be used to make predictions of lifetime, it is clearly essential that rules about extrapolation from the test conditions to those of service are known and have been verified, such that they can be used with confidence. For comparisons and quality checks the purpose can often be achieved with relatively short timescales and relatively few experimental points. Tests aimed at predicting lifetime will generally need much longer timescales and considerably more experimental points. The purpose of the trial also affects the choice of degradation agents and the parameters used to monitor degradation. For comparison and quality control purposes, agents are 69
Practical Guide to the Assessment of the Useful Life of Rubbers normally considered in isolation. For prediction purposes multiple agents are likely to be realistic as regards service. Multiple agents will make extrapolation rules rather more complicated, but even if exposures are made one at a time their synergy has to be considered. The parameters measured in trials to predict lifetime must be those critical to service, but in many instances of comparison or quality checks the choice of parameter can be heavily influenced by experimental convenience. One particular reason for carrying out accelerated ageing trials is to estimate shelf life. Often this reduces to a relatively simple case because all agents other than temperature can be eliminated. Usually, a general property such as elongation at break is chosen as the monitoring parameter and a minimum acceptable value agreed. Whatever minimum acceptable value is specified, at the end of the storage period the product starts service with a disadvantage. A potential problem is, particularly if service involves elevated temperatures, that the rate of degradation in service will be larger than expected after storage, perhaps because of depletion of antioxidant. This emphasises the value of using real or simulated service tests. There is a British Standard, BS 3574, which gives specifications for the controlled storage and inspection of rubber products.
14.2 Methods of acceleration The basic concept of accelerated testing is to increase the levels of the degradation agents. This is easily envisaged for environmental factors, for example raising the temperature or the intensity of radiation. The main concern is whether the same degradative processes occur at the elevated levels as would happen at those pertaining to service. This is often overlooked or ignored in the case of temperature but is perhaps more apparent for such factors as increasing concentration of a chemical or the level of a stress. The alternative to increasing the level of the degradation agent is to increase the frequency at which it is applied. This can be achieved in various ways. For temperature, a high level may be applied continuously where in service it is variable. In weathering an artificial light source can run 24 hours a day as opposed to something like half that for the sun. For chemical resistance, the exposure may be continuous whereas it is intermittent in service. It will be appreciated that in all these cases there is the danger that continuous exposure could result in different effects than those which occur in the exposure pattern of service. In mechanical testing, the frequency of stress application can be increased, although heating effects and time for relaxation processes have to be considered. For some products it is appropriate to simply use them more often, for example where in service the use is intermittent or there is normally downtime.
70
Accelerated Tests - General Another approach, which is not in itself acceleration, is to monitor degradation to a level less than that considered the end point and extrapolate in time. This can also be done when an accelerated level of the degradation agent has been used. When acceleration is produced by increasing the level of the degradation agent, it is usually achieved by applying a constant (elevated) level. It is also feasible to raise the level of the agent in steps so that test pieces are exposed for set times to progressively increasing levels of the agent. Products are not generally subjected to this pattern of exposure in use and it therefore becomes much more difficult to relate results to constant service level conditions. However, failures can be achieved in relatively short times. It has been said that although this cumulative damage approach is much talked about nobody uses it. A further variation is to increase the level of the agent in a continuous linear manner but this does not appear to have been applied to polymers.
71
Practical Guide to the Assessment of the Useful Life of Rubbers
72
15
Fundamental Problems
There are a number of intrinsic problems with accelerated tests used to assess lifetime. Firstly, the general problems in assessing service life mentioned in Section 2 apply. Regardless of how you go about making assessments, service conditions are not constant and, with long expected lifetimes, either the degree of extrapolation is very large or even accelerated tests need to be carried out over considerable timescales. The problems particular to accelerated tests are related to the extrapolation process. It was stated earlier that it is essential that rules about extrapolation from the test conditions to those of service are known and have been verified such that they can be used with confidence. In practice this is only an ideal as extrapolation procedures have not generally been comprehensively validated and almost certainly will not give accurate predictions in all cases. The only choice is to use the best techniques available and apply them with caution. Any relation between degradation, acceleration and time is only likely to be valid for a limited range of acceleration because the degradation mechanism may change with the level of the agent. The best results will be obtained with the lowest acceleration levels but at the cost of longer test times. Where multiple degradation agents are applied, the extrapolation rules for each have to be combined in such a way that synergistic actions are accounted for. Such procedures are likely to be very complicated and add greatly to the uncertainty. Regardless of the validity of the extrapolation procedure, the intrinsic experimental uncertainty of the measurements will be magnified as the degree of extrapolation increases so that predictions will always be very far from precise. In addition, there are difficulties associated with determining the critical degradation agents and the critical properties for the application as discussed earlier. As almost everything said about predicting lifetime from accelerated tests is negative it is reasonable to ask ‘Why are such tests and predictions carried out?’ A good answer is given by Andersson [22]. It is the same reason that one actually takes pains to predict the weather. With pure guesses, a frequency of correctly predicting weather of ca. 50% can be attained. With ‘qualified’ guesses, it is possible by guessing that the weather will be
73
Practical Guide to the Assessment of the Useful Life of Rubbers the same tomorrow as it is today, to attain a frequency of correctly predicting the weather of ca. 67%. But if the present weather situation is instead analysed, if continuous analyses and calculations are made and a model is created, it is possible to attain a correct answer frequency of about 80%. Thus it is a matter of striving to improve confidence in the estimates of how long a product will last.
74
16
Designing an Accelerated Test Programme
An accelerated test programme to assess useful life will represent an appreciable investment in time and money and deserves careful and systematic planning. The design of the test programme is likely to be an iterative process but a number of steps in the process can be identified. 1.
At the outset define clearly the purpose and objectives of the trial.
2.
Specify the number of years of life which need to be predicted and the acceptable uncertainty. In all probability the uncertainty can only be stated in general terms commensurate with how critical failure in service would be.
3.
Identify the critical degradation agents.
4.
This leads to deciding on which acceleration levels are viable. In practice, the acceleration levels are often limited by the available timescales for testing.
5.
Decide whether whole products or test pieces (or a combination) are to be exposed to the degradation agents.
6.
Consider synergistic effects and whether combined ageing will be used. This should lead to it being possible to establish the limitations of the trial in relation to service conditions and to estimate whether the original objectives can be met.
7.
Decide on the critical parameters to be monitored and the test methods to measure them.
8.
Select the analysis (extrapolation) model(s) which will be used.
9.
Decide on the detailed test conditions (number of levels of the agent(s), number of times, etc.), which will yield the number of test pieces (products) needed.
10. Consider the logistics of the programme and plan the timing and sequence of exposures and tests. 11. Specify test piece or product preparation.
75
Practical Guide to the Assessment of the Useful Life of Rubbers Experience has shown that there is a high chance that the resultant test programme will be too large or complicated to be viable. The process then has to be repeated making compromises on either the original objectives or the level of uncertainty.
76
17
Effect of Temperature
17.1 Low temperature As regards ageing, the usual concerns are with long life times and exposure to ambient or higher temperatures. However, low temperatures cause stiffening and eventually brittleness which in some applications can be the prime cause of failure. The effect is measured using, for example, the standard short-term tests referenced in Section 8.7. Additionally, the longer term effect of low temperature may need to be considered for rubbers that crystallise at particular subambient temperatures. This involves monitoring the stiffening effect over a timescale of weeks. In principle, any of the standard low temperature tests can be used to study crystallisation by conditioning at the low temperature for longer periods than is usual. In fact, most of the standard methods include a clause to the effect that the method can be used in this way. In the temperature retraction test it is suggested that larger degrees of applied elongation should be used as crystallisation is more rapid in the strained state. There are two methods standardised particularly for measuring crystallisation effects. Hardness is used in ISO 3387 and a compression set method in ISO 6471. Whatever methods are used, exposure can be continued until an equilibrium level has been reached. This is not really an accelerated test.
17.2 Properties at service temperature In environmental exposure tests, the changes in properties are normally monitored at ambient temperature. This is often the case even when the accelerated tests are extrapolated back to an elevated service temperature, rather than to normal ambient. Temperature has a short-term effect on properties and where the service temperature is elevated the properties should be monitored at that temperature.
17.3 Thermal expansion Thermal expansion and contraction are reversible short-term effects of temperature which may be very important in some applications. Repeated temperature cycling can be 77
Practical Guide to the Assessment of the Useful Life of Rubbers considered as fatigue in the long-term. Clearly, a temperature cycling test would be devised in this case, but there are no standard methods. The accelerating factors can be the temperature range and/or the frequency of cycling.
17.4 Heat ageing Heat ageing tests are carried out for two distinct purposes. They can be intended to measure changes in the rubber at the elevated service temperature, or be used as an accelerated test to estimate the degree of change which would take place over longer times at lower temperatures. Here, we are concerned with the second purpose. The basic procedure is to expose test pieces in an oven at the selected ageing temperatures. Almost exclusively, the exposures are made in air at atmospheric pressure. The standard procedures are given in ISO 188. This standard separates the two purposes for which ageing tests are carried out and carries a warning of the dangers and difficulties of predicting room temperature performance from accelerated tests. Either single chamber or multi-cell types of oven can be used. The advantage of a multicell oven is that by placing one compound only in each cell there is no danger of migration of plasticisers, antidegradants, etc., from one material to another. Single chamber ovens are generally more versatile and less expensive but only very similar compounds should be heated together. Because oxygen is used up in the ageing process it is important that an air flow is maintained and that the test pieces are exposed to air on all sides. With either type of oven, there must be a steady flow of air through the oven, giving between 3 and 10 complete changes per hour. This means that general purpose laboratory ovens are not suitable for rubber ageing tests. The air velocity will also affect the rate of ageing and two types of oven are specified in the standard. In the first, the air velocity is low and determined by the air exchange rate, whilst in the second forced circulation of air gives a velocity over the test pieces of between 0.5 and 1.5 m/s. Either type may in principle be either single chamber or multicell construction, but cabinet ovens without an internal fan are likely to have uneven temperature distribution. If the higher air velocity oven is used the rate of ageing can be higher, because of increased oxidation and volatilisation of plasticisers and antioxidants. It should be noted that previously the standard allowed cabinet ovens with an internal fan and these are probably the most common in practice. Test pieces must not occupy more than 10% of the free oven space and should be separated by at least 10 mm. It is essential that ovens do not contain any copper or copper alloys which can accelerate ageing in rubbers. 78
Effect of Temperature The temperature of the oven must be carefully controlled. A tolerance of ±1 °C is specified for up to 100 °C and ±2 °C above that. Preferably, the tolerance would be ±0.5 °C. It must be appreciated that the temperature will not be completely uniform throughout the oven space and should be measured at a series of positions to define the useable volume. The reproducibility of heat ageing tests has been found to be relatively poor. It is probable that much of the variability is due to lack of control of the oven and exposure parameters (for example, putting test pieces too close together or too close to the walls). For tests intended to make predictions to lower temperatures, exposures have to be made at a series of temperatures with a number of times at each temperature. The more temperatures and times that are used the better because uncertainty will be reduced. The minimum number of temperatures is 3, but this is rarely adequate and 5 is a more generally accepted number. The lowest temperature should be chosen such that the threshold value (see Section 25) is reached in a time in excess of 1,000 hours. The minimum number of times at each temperature is 5, but more are preferable and will be essential if the shape of the graph of change of property with time is complicated. The spacing of times is usually linear for oxidative ageing. The number and type of test pieces exposed for each measurement point will depend on the property being measured. Usually the number specified in the relevant test method standard is chosen but, again, the more the better. ISO 188 also includes a method for ageing in oxygen at elevated pressure. In principle, this should aid the diffusion of oxygen and hence alleviate problems of degradation profiles through the test piece. However, the increased acceleration produced is considered likely to decrease the correlation with natural ageing and the method is not nowadays recommended. The use of air at elevated pressure is also sometimes used and would be expected to be intermediate in severity but this method is also not in favour. The other extreme would be to age in vacuum, but this is not common practice because it would rarely be relevant to service.
79
Practical Guide to the Assessment of the Useful Life of Rubbers
80
18
Effect of Liquids
18.1 General procedures Tests in which rubbers are exposed to liquids are often called swelling tests, simply because the resulting change in volume of the test pieces is by far the most commonly used measure of the effect of the liquid. The term oil ageing is also sometimes used because standard grades of mineral oil are the liquids most often specified. Volume change is a very good measure of the general resistance of a rubber to a liquid. A high degree of swelling clearly indicates that the material is not suitable for use in that environment, although in certain applications, such as seals, a negative swelling (extraction) could be equally bad as regards service. In addition, the degree of swelling can be related to the state of cure and the crosslink density estimated, although this is not normally a consideration in ageing tests. The usual form of test involves test pieces immersed in the liquid held in a glass container. The standard procedures are given in ISO 1817. The volume of liquid should be at least 15 times the combined volume of the test pieces, which should be completely immersed and with all surfaces freely exposed, for example, by hanging on hooks. The containers are held at the required test temperature by placing them in a single cabinet oven. For tests at temperatures considerably below the boiling point of the liquid, the container is simply stoppered, but for temperatures near the boiling point the container is fitted with a reflux condenser to minimise evaporation. To simulate service conditions, tests may be needed with exposure being on one side of the test piece only. This is achieved by using a simple jig in which the test piece forms one end of a cylindrical container. The standard does not include cases of partial immersion (other than one side) nor immersion under pressure, when special rigs have to be developed. Tests are usually made on unstrained test pieces but, as for heat ageing, tests such as compression set and stress relaxation may also be required. After immersion, test pieces are cooled to room temperature which is best done by transferring them to a fresh portion of the test liquid. Surplus test liquid must be removed
81
Practical Guide to the Assessment of the Useful Life of Rubbers from the surface but no evaporation should be allowed. For very volatile liquids the test piece is rapidly transferred to a weighing bottle. To measure volume change, simple sheet test pieces of about 1–3 cm3 are exposed. The volume change is usually measured by weighing in air and weighing submerged in water, before and after immersion in the liquid. The change is calculated on the basis that volume is proportional to the weight in air minus the weight in water. Care must be taken to exclude air bubbles when weighing in water and this is aided by adding a trace of detergent and/or quickly dipping the test piece in ethanol before weighing. If the rubber is less dense than water, a sinker has to be used and the calculation is slightly more complicated. An alternative way of obtaining volume change is to measure the dimensions of the test piece before and after immersion. A particularly neat way of doing this is to project a magnified image of a small rhomboid test piece and measure the lengths of the diagonals. By assuming that swelling is isotropic, volume change is obtained from:
⎡ AB ⎞ V = ⎢⎛ ⎣⎝ ab ⎠ where:
3/ 2
⎤
− 1⎥ × 100%
⎦
V is volume change A and B are lengths of diagonals after swelling a and b are lengths of diagonals before swelling
For change in other properties test pieces complying with the relevant standard method are exposed. The most commonly used properties are tensile stress-strain and microhardness and these are covered in ISO 1817. If volume change is measured after drying the test piece, the amount of material extracted from the rubber by the liquid can be estimated. Alternatively, it can be estimated by evaporating off the immersion liquid and weighing any residue. Procedures are given in ISO 1817 but they are not very accurate. For the single sided exposure test the standard procedure is to record the change in weight expressed as mass per unit area. Where exposure to the liquid in service is intermittent is will be of interest to know the effect on properties after drying. ISO 1817 also includes procedures for measuring tensile stress-strain and hardness after a drying period. Whether or not results before or after drying are wanted depends on the application and possibly both will be of relevance.
82
Effect of Liquids Cyclic exposure to liquids is not commonly used but may have application to model a service condition, particularly if significant oxidation will occur during the dry periods. It is standard practice to measure the cross section of tensile dumbbells before immersion but to mark the gauge length after immersion. This is largely for experimental convenience but is should be noted that the calculated results for strength and modulus are, to say the least, arbitrary. When the effect of a liquid is purely physical and it is possible to continue the immersion until equilibrium absorption is reached, then actually no acceleration is involved. If chemical reactions are taking place, including the effect of temperature, the situation is similar to that for heat ageing and generally exposures will be needed at a series of temperatures and, perhaps, concentrations of the test liquid. The choice of exposure times is complicated when more than one reaction is taking place as the rates will be different. Generally, it is best if equilibrium absorption is reached relatively quickly in comparison to chemical changes and its effects treated separately from the subsequent chemical changes. This is not unlike the situation for physical and chemical stress relaxation. There is a further complication if extraction also takes place at a slower rate than absorption.
18.2 Standard liquids Although for any particular application the liquid(s) relevant to service should be used, for testing it is common practice to use standard liquids representative of the types of liquid to which the product should be resistant. This approach is obviously advantageous in quality control and for inter-laboratory comparison as commercial liquids are not always well defined. It is also very useful for ageing studies aimed at a general characterisation of a material. However, if a standard liquid is used care must be taken to consider the possible effect of an apparently minor additive or ingredient. A good example is the variety in formulations of fuels containing alcohol. Consideration also has to be given to concentration when a solution is used. It would be extremely dangerous to assume that the degradation effect and concentration are related in a simple way. The most commonly used standard liquids are defined in ISO 1817. The oils in particular need to be obtained from a certified source.
83
Practical Guide to the Assessment of the Useful Life of Rubbers
18.3 Water The permeation of water into rubber is very slow compared to organic liquids and using the normal 2 mm test piece a very long time is required to reach equilibrium. Because of this a method is standardised in BS 903 Part 18 in which the rubber is cut into small particles by cutting, rasping or grinding. This results in a large ratio of surface area to volume. The particles are exposed to water vapour at a set humidity and hence the results equate to that humidity. The measured equilibrium absorption is substantially the same as the equilibrium absorption that would be obtained by immersion in an aqueous solution which would maintain the test humidity. As this implies, the equilibrium water absorption is reduced if the water is not pure. Hence, exposure to aqueous solutions should be made at the concentration of interest. As the humidity approaches 100% even small amounts of a salt have a significant effect on equilibrium absorption. Apart from the complications of the approach using ground particles, it is impractical if properties of the rubber are to be measured after immersion. Hence, it is usually necessary to suffer long exposure times using ISO 1817 or to make extrapolations from below equilibrium. As an illustration, it has been found that the effect of water absorption on compression stress relaxation at ambient temperature are not noticed until upwards of one year of exposure. Considering that polymers such as polyurethanes can be liable to hydrolysis, it is a little surprising that there are no general International Standard methods for exposure to moist heat or steam. Where moisture is a consideration, tests could be made in the same manner as heat ageing tests but using an injection type humidity cabinet with both temperature and humidity controlled. Dry heat ageing could also be carried out to isolate the effect of humidity. Steam at 100 °C or above would only be relevant in a few applications and a special exposure jig would need to be devised or may be found in the product specification. It would be expected that tests involving moisture at elevated temperatures would be sensitive to the amount of air present and the design of the test should take this into account.
84
19
Effect of Gases
19.1 General Leaving aside air (which has been considered under heat ageing) and ozone very little testing is carried out on the effect of gases, which doubtless reflects the relatively small number of applications where gases are important. There are no general standard methods and a specialised test would need to be devised for any particular product and gas. Usually, the approach involves using an exposure jig or chamber through which the gas is circulated. One difficulty is in safely disposing of the gas if it is toxic or an explosion risk. The same considerations as for liquids apply in respect of equilibrium absorption, there being both physical, chemical and temperature effects. An extra consideration where high pressures are involved is the possibility of explosive decompression due to absorbed gas rupturing the rubber when pressure is released.
19.2 Exposure to ozone Although ozone exists in only very small quantities in the atmosphere, it is important because less than 1 pphm (part per hundred million) can severely attack non-resistant rubbers in the strained condition. Indeed, ozone attack can be the major factor as regards lifetime when rubber is exposed to the atmosphere. The effect of ozone is to produce clearly visible and mechanically very damaging cracking of the rubber surface. Laboratory tests consist of exposing test pieces in a closed, non-illuminated cabinet through which ozonised air at a known concentration is passed. The apparatus and procedures are standardised in ISO 1431. Part 1 covers static exposure, which is the usual method, Part 2 covers dynamic exposure where cyclic strain is applied to the test piece and Part 3 specifies methods for measuring the ozone concentration. Factors such as temperature, ozone gas velocity and distribution of ozone in the chamber need to be controlled. The concentrations used are much higher than those existing naturally in most parts of the world. They generally range from 25 to 200 pphm with 50 pphm being the most common, which represents a considerable degree of acceleration. Occasionally, very much
85
Practical Guide to the Assessment of the Useful Life of Rubbers higher concentrations are specified for special purposes. Concentrations lower than 25 pphm are extremely difficult to control accurately. Ozone only attacks rubber in the strained condition although with the less resistant rubbers the ‘threshold strain’ for attack may be very low. The basic test piece specified in ISO 1431 is a thin strip held in tension between clamps made of a material, such as aluminium, which does not decompose ozone. A variation of this is to add tab ends to the strip to facilitate gripping and the T50 dumbbell is also specified. This has the advantage of small size when cutting from products, and mounting strained test pieces is made particularly easy by hooking the ends over suitable frames. It can be more convenient with extrusions to wrap them around a mandrel, but the resulting strain is generally less well defined than it is with strips. It is usually desirable to expose test pieces at a number of different strains. The small T50 test pieces are economical when this is the case, but in theory at least it would be advantageous to have a form of test piece which covered several strains simultaneously. Such test pieces have been developed but are not commonly used. The areas where a test piece is attached to clamps and cut edges are preferential sites for cracking. It is generally good practice to coat clamped areas with an ozone resistant paint, such as one based on Hypalon, but cut edges are best left. Clamps, even when made of material such as aluminium, should be ‘soaked’ in ozone prior to use. Any pattern or flaws on the test piece surface will also tend to act as stress raisers and show preferential cracking. Test pieces are conditioned in the strained state to allow protective additives that must form a surface bloom to become effective. The usual conditioning period is between 48 and 96 hours and the test pieces should be kept in the dark and in an ozone-free atmosphere. For this treatment to be effective, the test piece surface must not, of course, be touched in the course of subsequent handling. It has been very reasonably argued that this conditioning procedure is artificial, as in service the conditions of temperature and installation may not allow a bloom to form. For specification or quality control purposes a single strain level is used, commonly 20%, but for evaluating the resistance of a material a series of strains should be used. Temperature does affect the rate of ozone cracking but it cannot be said simply that higher temperatures accelerate the effect. The blooming characteristics of different waxes can make a rise in temperature increase or decrease ozone resistance. Above about 70 °C, all ozone is destroyed. In the current major standards, 40 °C is specified, just a small degree of acceleration above ambient and practically the lowest level which can be controlled without cooling in many cabinets. 86
Effect of Gases It has been shown that the humidity of the ozonised air can affect the rate of ozone attack, but generally any significant change is restricted to very high humidities and testing at below 65% is usual. The procedure is that test pieces are placed in the chamber at the required concentration and inspected at intervals. Most specifications give a set exposure period. However, for proper evaluation of a material test pieces should be examined at a series of times, such that data can be obtained on the relationship between strain and time to appearance of cracks. ISO 1431 requires examination to be carried out with a lens of x7 magnification but, unfortunately, any examination of cracks is to some extent subjective. Because many products are subjected to cyclic strain in service and because protective wax coatings, which are easily removed by mechanical contact, cannot withstand cycling, there is much logic in using a dynamic exposure test. The standard method is given in ISO 1431. Part 2 involves cycling strip (or T50) test pieces in tension at 0.5 Hz. The low frequency is used so that there is little contribution from fatigue mechanisms. Either continuous cycling or a sequence of dynamic cycles and periods of static strain can be used, but there is no consensus as to which sequence correlates best with particular service applications. It has been shown that T50 test pieces can be successfully used in a dynamic exposure test with complete fracture as the criterion of failure. This obviously results in a longer test but the means of assessment is much easier and not subjective. Rupture is commonly used as the failure criterion in fatigue tests but has not been adopted widely for ozone attack. Another way of getting round the subjectivity of observing cracks is to measure the relaxation of stress in the strained test piece as ozone attack proceeds. Despite having been suggested many years ago, this has not as yet been widely accepted, probably because of the extra apparatus, expense and complexity. Also, it does not work well with isolated cracks near the threshold strain.
19.3 Evaluation of cracking When only a single strain and exposure period has been used, the result is simply expressed as either cracking or no cracking. The degree of cracking can be described and a number of arbitrary scales have been used, but they are all very subjective. The most widely used is the 0-3 scale where 0 is no cracking, 1 is cracks only seen under magnification, 2 is very small cracks and 3 anything worse. Even this simple rating scheme falls down when there are one or two large cracks only.
87
Practical Guide to the Assessment of the Useful Life of Rubbers If regular inspections are made, a better approach is to record the time until the first appearance of cracks. This becomes even more advantageous when a number of strains are used. It is then possible to observe the relationship between time to cracking and applied strain. In some cases, a linear plot will show the existence of a limiting threshold strain (L) as shown in Figure 9. For some rubbers a log-log plot will yield a straight line but it is very dangerous to extrapolate this to longer times. The first criterion for describing a material as ozone resistant is total freedom from cracking. Therefore, the higher the threshold strain after a given exposure period, or the higher the limiting threshold strain if this exists, or the longer the time before cracks appear at a given strain, the better is the ozone resistance. However, when materials with relatively low ozone resistance are being compared such that cracking is inevitable during service life, then the severity of cracking is important. Very small cracks may be of little consequence apart from a cosmetic point of view. This is usually the case when thick sections of rubber are involved and cracking is confined to the surface. The way in which the severity of cracking is related to strain is not simple. The usual trend is shown in Figure 10; by definition there being no cracks below the threshold strain (T) for any given exposure period. A few cracks, often large, are found at strains slightly above the threshold and the cracks will become more numerous and smaller at progressively higher strains. It is quite possible for the cracks at very high strains to be so
Figure 9 Threshold strain
88
Effect of Gases small as to be invisible to the naked eye. As exposure time increases numerous very small cracks may coalesce to form larger but relatively shallow cracks. Hence, a non-resistant rubber at high strains could be more suitable than a ‘better’ resistance rubber just above its threshold strain. This illustrates the futility of protecting a rubber such that it will just pass a single strain and period standard test, when it will exhibit large cracks in service.
Figure 10 Relation between crack size, crack density and strain Curve A = average crack length; Curve B = average crack length with coalescence; Curve C = crack density
89
Practical Guide to the Assessment of the Useful Life of Rubbers
90
20
Weathering
Weathering is taken to mean exposure to the atmosphere, which includes sunlight, temperature, precipitation and any pollutants present. Accelerated weathering is based on using an artificial light source in a cabinet in which the humidity, water spray and temperature can also be controlled. In the case of rubbers, this means that the important degradation agent ozone is not included. The core feature of an accelerated weathering apparatus is the light source which, ideally would provide an accurate simulation of sunlight over the complete spectrum. Until about 1960 the only source available with a sufficient level of irradiance was the carbon arc. Although improvements were made, even the best carbon arcs are a poor match for sunlight and are considered by most people to be unsatisfactory. Despite this, they are still used in some quarters for plastics. With filtering to cut out the radiation below 300 nm and to reduce the infrared, xenon lamps provide a quite good simulation of sunlight across the spectrum. They are now considered the preferred source by most people. Unfortunately, xenon lamp apparatus is very expensive and the exposure temperature is usually rather high. Fluorescent tubes can reproduce the UV part of the spectrum quite well but do cover higher wavelengths. As UV is generally thought to be the important part of the spectrum as regards polymer degradation they have become widely used. Their popularity is greatly helped by relatively low cost and low heat generation. The UV-B 40 lamp is generally preferred. More recently, metal halide lamps have been introduced that give a reasonable simulation of sunlight and offer very high irradiance with modest heat generation. They have mostly been employed in large solar simulation systems. Xenon lamps can be filtered to approximate the solar spectrum under glass and a fluorescent lamp, the UV-B 351, can be used to simulate this in the UV region. The temperature of exposed samples is dependent on both the air temperature in the cabinet and the absorbance of direct radiation. Temperature is usually measured with a black panel thermometer which gives the surface temperature of a perfectly absorbing material. White panel thermometers are also commonly used which measure the other
91
Practical Guide to the Assessment of the Useful Life of Rubbers extreme. The actual temperature reached by a test piece depends on the material and its colour. It will also depend on the air temperature and velocity, so that both the air and black panel temperatures should be controlled. Most cabinets incorporate a device whereby the test pieces can be sprayed with water at intervals for various periods of time. Distilled water is normally used to avoid any effect of impurities. Humidity of the air between water cycles can also be controlled. It will be appreciated that the actual conditions on the test pieces can be somewhat complicated because of the interactions of the heating, cooling and moisture. For plastics at least, water in contact with the material, particularly dew, increases UV degradation because the dissolved oxygen is more active than the oxygen in air for photo-oxidation. The effect of light on rubbers has generally been considered to be much less important than it is for plastics. With any degradation being restricted to the surface layer it is thought only to be of consequence for thin walled articles and coated fabrics, plus coloured materials. The result is that weathering tests on rubbers are carried out relatively infrequently. This is probably a complacent view as back in 1970 Angert and Dubok [25] reported significant weathering effects on both black and white filled compounds. Recent experience at Rapra is that artificial weathering in fluorescent tube apparatus at 45 °C produces quite large changes in many compounds. Clearly, this will include a contribution due to temperature, and an attempt was made to isolate the purely weathering effect. This was achieved by comparing the results of accelerated ageing tests and applying a WLF shift, so that inevitably the uncertainties were very large. In many cases the observed changes on weathering could be accounted for by the effect of temperature, but in at least as many cases there was evidence that the effect of weathering was significant and, also, that in other cases the effects of weathering and temperature were in opposite directions. Despite the uncertainty, the evidence was sufficient to suggest that the effects of weathering should not be ignored. The international standard for weathering of rubbers is ISO 4665 which relies heavily on, and references, the ISO methods for plastics. ISO 4665 essentially gives only the requirements for such things as conditioning and expression of results which are particular to rubber. The plastics methods are given in ISO 4892-1 (‘General guidance’), 4892-2 (‘Xenon lamp’), 4892-3 (‘Fluorescent tubes’) and 4892-4 (‘Carbon arc’). These refer to ISO 4582 for determination of the changes in properties after exposure. Other useful standards are ISO 9370, Instrumental determination of radiant exposure in weathering tests, and International Commission on Illumination (CIE) publication 851989, ‘Solar spectral irradiance’.
92
Weathering Part 1 of ISO 4665 gives a lot of guidance on conducting accelerated weathering tests and the parts for specific exposure apparatus recommend particular conditions. However, it will be appreciated that a considerable variety of exposure conditions are possible and these have to be selected with regard to the specific aims of the trial. Further information and guidance on weathering tests is given by Kockott in Handbook of Polymer Testing [18].
93
Practical Guide to the Assessment of the Useful Life of Rubbers
94
21
Fatigue
21.1 General Fatigue is taken here to mean the application of repeated cyclic deformation which results in a change in stiffness, a loss of mechanical strength and ultimately rupture. The mechanisms which may contribute to breakdown include thermal degradation, oxidation and attack by ozone as well as the propagation of cracks by tearing. The manner of degradation will vary according to the geometry, the type of stressing and the environmental conditions. In rubber testing it is normal to distinguish between two types of fatigue test: Tests in which the prime aim is to cause heating of the rubber by the stressing process – heat build-up tests. Tests in which the prime aim is to induce and/or propagate cracks without significant heating – flex cracking or cut growth tests. The first type is applicable to applications where heat build-up may be a problem, such as in tyres. The second type is by far the most widely applicable as most applications are not supposed to involve significant fatigue heating.
21.2 Heat build-up tests Heat build-up tests use a relative bulky test piece deformed in compression and/or shear so that a large temperature rise can be generated. As well as the temperature rise the change in stiffness can be measured and, ultimately, the point at which rupture of the test piece occurs. The apparatus used is commonly called a flexometer, which is confusing as it brings to mind flex cracking and cut growth. A number of flexometers have been developed, all of which are somewhat arbitrary in their deformation cycles and the geometry of the test piece. Currently, ISO 4666 specifies
95
Practical Guide to the Assessment of the Useful Life of Rubbers two sets of apparatus, the rotary flexometer and the compression flexometer. Part 1 of the standard gives the basic principles of this type of test. The compression flexometer is essentially the same as the Goodrich flexometer that has been specified in the American Society for Testing and Materials standard ASTM D623 for many years. The rotary flexometer is similar to the Firestone flexometer of ASTM D623 in that they both superimpose a dynamic shear deformation onto a static compression deformation, but the actions are different. British Standards have not adopted the rotary flexometer as it is so vaguely specified in both ISO and ASTM as to be thought not worthy of standardisation. There is now a Japanese proposal to add a new type of compression flexometer which operates with constant stress cycles. The vagueness of the ASTM and ISO methods for rotary flexometers reflects the arbitrary nature of these tests. The first part of ISO 4666 makes a reasonable attempt to describe the basic principles of fatigue testing and to give guidance on the interpretation of results obtained using particular apparatus and test conditions. Attention is drawn to the care necessary in measuring temperature rise and the fact that the result depends on where the temperature is measured and on the test piece geometry. It recommends testing at a series of strain or stress levels because a comparison of rubbers at one level only can be misleading. The standard also mentions the measurement of creep and set in the test piece after periods of dynamic cycling. In truth, results obtained under a particular set of conditions are quite arbitrary and have no significance outside of the conditions used. Consequently, it is difficult to contemplate using the standard methods to make attempts at predicting service life rather than simply comparing materials. Most heat build-up tests apply a fixed pre-stress or strain partly because it is necessary to hold the rubber in place without bonding of the test piece. The amount of pre-stress or strain will affect the fatigue life; in particular, the fatigue life is appreciably shortened if the cyclic deformation passes near to or through zero strain. The pre-stress or strain can in fact also be a constant stress or a constant strain. In principle a test can be made with cycles of either constant strain amplitude or constant force amplitude. With constant strain, the resultant stresses are greater for higher stiffness rubbers, so that these are stressed more highly and, other things being equal, will develop more heat. On the other hand, with constant stress a stiff rubber will deform less and consequently tend to give a better fatigue performance. Consequently, to avoid conflicting results it is necessary to choose conditions which correspond with those met in service. It is also possible to use cycles of constant energy which is quite commonly the situation met by such products as dampers and shock absorbers, but this is more difficult from the point of view of apparatus.
96
Fatigue
21.3 Flex cracking and cut growth tests The vast majority of flex cracking tests strain the test piece in flexure; this represents the mode of deformation experienced in service by such important products as tyres, belting and footwear. A variety of flex tests have been used, but many that were once popular are now rarely seen. Despite the obvious logic of using flexure tests there are serious disadvantages. The principal problem is that it is difficult to control the degree of bending. For example, this may vary with the modulus of the rubber and, because the fatigue life of the rubber will be sensitive to the magnitude of the applied strain, misleading results may be obtained. The more nearly the deformation produced in the laboratory test reproduces that experienced in service the better should be the hope of correlation. However, most products are subjected to a complex pattern of straining. The alternative approach is to use a simple but reproducible mode of deformation such as pure tension. Flexing has been produced in various ways, including linking test pieces to form a belt and running it over pulleys; attaching strips to the periphery of a rotating wheel so that they are bent against a fixed roller; fixing a strip in two clamps which move towards each other so that the strip is bent into a loop, and bending over a rod. The most widely known test equipment is the De Mattia apparatus which is standardised in ISO 132. This uses a strip test piece with a transverse groove fixed in two clamps which move towards each other. The maximum surface strain is somewhat indeterminate. The Ross flexing machine bends the test piece through 90° over a rod and the maximum surface strain is rather more controlled than in other flexing devices (ASTM D1052). Even more important than the control of maximum strain in a flexing cycle is the control of minimum strain because cracking is particularly severe if this is zero. In all flexing machines the strain is deliberately intended to be zero, but only in the Ross apparatus is this achieved precisely and in a reproducible manner. Also, in all bending methods, the maximum strain depends on the thickness of the test piece and hence this must be closely controlled. ISO 132 now includes the methods for determining crack initiation and for growth which were previously in separate standards. There is a significant difference between the two tests. This is illustrated by the fact that natural rubber fairly quickly develops fine cracks in a flex cracking test but is relatively resistant to the further growth of these cracks or of a purposely made cut, whereas styrene butadiene rubber shows just the opposite behaviour. Both methods use the same test conditions, the essential difference being that for cut
97
Practical Guide to the Assessment of the Useful Life of Rubbers growth a cut is made through the bottom of the groove in the test piece before flexing is started. In flex cracking tests, one of the most difficult problems is how to assess the degree of cracking. Visual examination is the only really feasible procedure and inevitably the assessment is subjective and operator dependent. Unfortunately, the pattern of cracking in a De Mattia test varies with the type of rubber and is likely to start at the edges of the test piece, although this can be virtually eliminated by putting a radius on the edges. The procedure specified in ISO 132 is based primarily on the length of the largest crack present at any stage and the depth of the crack is ignored. Other methods have been suggested but are generally too complicated to be acceptable. Judging on the time to the first appearance of cracks gives only a single point measurement and is liable to be more variable than taking the times to a series of grades of increasing severity. In the cut growth method, a 2 mm cut is made through the rubber with a tool of specified geometry. The length of this cut is measured at intervals and the number of cycles for it to increase by 2 mm, 6 mm and 10 mm is deduced. ASTM methods have additional apparatus including the Ross machine. This has been used (at a rather slow speed) in the UK for footwear but, considering its advantages, it is perhaps surprising that it is not more widely specified.
21.4 Tests in tension All of the bending methods are to some extent arbitrary as to the degree of strain used and, in most tests, neither maximum nor minimum strains are well defined. By cycling in simple tension, strains can be reproduced more easily and a range of strains and prestrains can be readily realised with one apparatus. A standard procedure for fatigue in tension was developed from MRPRA work on the concept of tearing energy and is given in ISO 6943. Test can be made at a number of extensions and compounds can be compared in terms of fatigue life at the same strain or at the same strain energy. In the latter case, absolute comparisons can be made on compounds of different modulus. When comparing different rubbers, it is necessary to test at a number of strains or to define the severity of conditions which will occur in service, because with the number of variables involved the ranking order may vary with the maximum strain employed. Fatigue life is influenced by the environmental conditions under which the test is carried out, in particular temperature, oxygen and ozone levels. In practice most testing is done in the standard laboratory atmosphere.
98
Fatigue ISO 6943 specifies two different types of test pieces, rings and dumbbells, which correspond to the geometries used on commercially available apparatus. There is, in principle, little difference between the two forms of test piece, but dumbbells are necessary for studying directional effects. They are also easier to cut from sheet than rings, but normally a specially moulded sheet is required such that the dumbbells have a raised bar across each tab end to aid location and gripping. There are no gripping problems with rings as they are placed over rollers and the roller separation is a direct measure of strain. The dumbbells specified are the same as those for tensile stress-strain tests except that the preferred thickness is 1.5 mm. The ring is also the same as the tensile ring which means that the bulk of the two types of test piece (rings and dumbbells) are different. The range of frequency specified is between 1 and 5 Hz and the standard only covers strain cycles passing through zero, although pre-strains could be applied. It is suggested that at least five test pieces should be tested at each strain and that usually it is desirable to test at a number of maximum strains. The strain on ring test pieces is calculated on the internal diameter. The test is continued until complete failure of the test piece occurs and then the number of cycles is recorded. During the course of a test the stress-strain relationship of the test piece will change and there will also be a degree of set. It is recommended that both these parameters are measured at intervals and the results reported as well as the fatigue life. The results can be presented in graphical form as log (fatigue life) against strain, log (stress energy density) or log (stress). An annex gives explanatory notes including a section on interpretation of results, which introduces the concept of a fatigue limit.
22.5 Non-standard methods The standardised methods for fatigue, both for heat build-up and crack growth, are generally based on very old designs of mechanical equipment. Far more comprehensive fatigue conditions can be achieved by using a servo hydraulic test machine, which would allow various loading patterns and frequencies, and enable stress-strain relations to be recorded. Such machines, if of sufficient capacity, can also be used to fatigue complete products. A more modern apparatus, the so called Tear Analyser, has been developed in Germany for cut growth measurements [26]. A 1 mm lateral cut is made into a strip test piece which is then cycled through a pre-determined strain regime. A video camera is used to measure cut growth and the test conditions can be varied widely to match service requirements in terms of frequency, temperature and strain cycles. Rate of cut growth
99
Practical Guide to the Assessment of the Useful Life of Rubbers can be plotted against a number of parameters and excellent correlation with tyre endurance is claimed.
100
22
Abrasion
22.1 General Abrasion can be defined as the loss of material from a surface due to frictional forces and is most often the result of two surfaces being rubbed together. Abrasion resistance is then the resistance to wear resulting from mechanical action on the surface. The mechanisms by which wear of rubber occurs when it is in moving contact with another material are complex, but the principal factors involved are cutting and fatigue. It is possible to categorise wear mechanisms in various ways and commonly distinction is made between abrasive wear, fatigue wear and adhesive wear; additionally, wear by roll formation is sometimes considered as a separate mechanism: Abrasive wear is caused by sharp asperities cutting the rubber, which is rather at odds with the general definition of abrasion given above, which specifies friction; Fatigue wear is caused by particles of rubber being detached as a result of dynamic stressing on a localised scale; Adhesive wear is the transfer of rubber to another surface as a result of adhesive forces between the two surfaces; and Wear by roll formation is where there is progressive tearing of a layer of rubber which forms a roll. There can also be corrosive wear due to direct chemical attack on the surface, and the term erosive wear is sometimes used for the action of particles in a liquid stream. In any particular wear situation more than one mechanism is usually involved but one may predominate. Abrasive wear requires hard, sharp cutting edges and high friction. Fatigue wear occurs with rough but blunt surfaces and does not need high friction. Adhesive wear is much less common but can occur on smooth surfaces. Roll formation requires high friction and relatively poor tear strength. Roll formation results in a characteristic abrasion pattern of ridges and grooves at right angles to the direction of movement.
101
Practical Guide to the Assessment of the Useful Life of Rubbers Abrasive wear and roll formation result in much more rapid wear than fatigue processes. The mechanism and hence the rate of wear can change, perhaps quite suddenly, with the conditions such as contact pressure, speed and temperature. In any practical circumstance the mechanisms may be complex and critically dependent on the conditions. Consequently, the critical factor as regards testing is that the test conditions must essentially reproduce the service conditions if a good correlation is to be obtained. Even a comparison between two rubbers may be invalid if the dominant mechanism is different in test and service. The range of conditions encountered in tyres, for example, are so complex that they cannot be matched in a single test. It follows that there cannot be a universal standard abrasion test method for rubber and the test method and test conditions have to be chosen to suit the end application. Also, great care has to be taken if the test is intended to provide a significant degree of acceleration.
22.2 Types of abrasion test A great many abrasion testing machines have been devised and several standardised at national level for use with rubbers. The majority of rubber tests involve a relatively sharp abradant and were originally devised for use with tyre tread materials. One distinction between test types is that some use a loose powder abradant and others use a solid abradant. A loose abrasive powder can be used rather in the manner of a shot-blasting machine as a logical way to simulate the action of sand or similar abradants impinging on the rubber in service. A loose abradant can also be used between two sliding surfaces. Conveyor belts or tank linings are examples of products subject to abrasion by loose materials. A car tyre is an example of a situation where there is a combination of abrasion against a solid rough abradant (the road) together with a free flowing abradant in the form of grit particles. This situation can occur in testing as a result of the deliberate introduction of a powder or the generation of wear debris from the solid abradant. The majority of wear situations involve the rubber moving in contact with another solid material, and solid abradants can consist of almost anything. Another distinction is on the basis of the geometry by which the test piece and the abradant can be rubbed together. A great many geometries are possible and some common configurations are shown in Figure 11. In type (a) the test piece is moved backwards and
102
Abrasion forwards linearly against a sheet of abradant (or alternatively a strip of abradant could be moved past a stationary test piece); in (b) the abradant is a rotating disk with the test piece held against it; in (c) both test pieces are in the form of wheels either of which could be the driven member; in (d) the abrasive wheel is driven by a flat rotating test piece and in (e) both the test piece and the abradant are rotating in opposite directions.
Figure 11 Abrasion test geometries
103
Practical Guide to the Assessment of the Useful Life of Rubbers
22.3 Abrasion test conditions The choice of abradant should be made primarily to give the best correlation with service but in practice is often chosen largely for reasons of convenience. In laboratory tests the most common are abrasive wheels (vitreous or resilient), abrasive papers or cloth and metal ‘knives’. The usual abrasive wheels and papers really only relate to situations where cutting abrasion predominates. Materials such as textiles and smooth metal plates have been found to give good results for some applications, but smooth materials abrade relatively slowly and if conditions are accelerated give rise to excessive heat build-up. With any geometry involving a fixed abradant there is relative movement or slip between the abradant and the test piece, and the degree of slip is a critical factor in determining the wear rate. In Figure 11(a) there is 100% slip and the rate of slipping is the same as the rate of movement between abradant and test piece, whereas in (c) the degree of slip can be varied by changing the angle between the wheels. In (b), (d) and (e) the rate of slip will depend on the distance of the test piece from the centre. In all cases the rate depends on the speed of the driven member. An increase in the rate of slip will increase the heat generated and hence the temperature. If the abrasion is unidirectional, abrasion patterns will develop which can markedly affect subsequent abrasion loss. The contact pressure between the test piece and abradant is another critical factor in determining wear rate. Under some conditions, wear rate may be approximately proportional to pressure, but abrupt changes will occur if the abrasion mechanism changes with changing pressure. For example, a change can occur because of a large rise in temperature. Rather than consider contact pressure and degree of slip separately, it has been proposed that the power consumed in moving the rubber over the abradant should be used as a measure of the severity of an abrasion test. The power used will depend on the friction between the surfaces and will determine the rate of temperature rise. An important difference between apparatus of, for example, type (a) and type (d) of Figure 11 is that in the former case, the test piece is continuously and totally in contact with the abradant and there is no chance for the heat generated at the contact surface to be dissipated. Any change in the nature of the contacting surfaces will affect the rate of wear and this includes changes in the abradant and the test piece as the wear process proceeds. There can also be changes due to the presence of another material between the contacting surfaces, either due to deliberate addition, accidental contamination or debris from the abradant and debris from the test piece. 104
Abrasion Introduction of a particulate material between the contacting surfaces can be carried out to simulate service, as for example a car tyre running on a dusty road. Similarly, a lubricant such as water can be introduced. Relatively few types of apparatus are capable of operating with these conditions. It is common practice to remove wear debris by continuously brushing the test piece or by the use of air jets, in which case care must be taken to ensure that the air supply is not contaminated with oil or water from the compressor. Clogging or smearing of the abradant with rubber is a common problem with abrasive wheels and papers, and its occurrence will invalidate the test. It is normally caused by a high temperature at the contact surfaces and, although the problem can sometimes be reduced by introducing a powder between the surfaces, it should be treated as an indication that the test conditions are not suitable. If high temperatures are to be realised in service, a test method in which new abradant is continually used should be chosen. If correlation between laboratory tests and service is to be obtained, the test conditions must be chosen extremely carefully to match those found in the product application. Although temperature has a large effect on wear rate and is one of the important factors in obtaining correlation between laboratory and service, it is extremely difficult to control the temperature during testing. Clearly, it is the temperature of the contacting surfaces which is of importance rather than the ambient temperature, and the surface temperature reached is dependent on several experimental factors as discussed above.
22.4 Abrasion test apparatus The only method currently standardised internationally is the rotating cylindrical drum device, more commonly known as the DIN (Deutscher Institut für Normung) abrader, which is specified in ISO 4649. A disc test piece in a suitable holder is traversed across a rotating drum covered with a sheet of the abradant. In this way, there is a relatively large area of abradant, each part of which is passed over in turn by the test piece so that wear of the abradant is uniform and relatively slow. In the standard method, there is no provision for changing conditions from those specified but the abradant and the load on the test piece could be changed. The degree of slip is 100% and it would be inconvenient to test in the presence of a lubricant. Although not versatile, the method is very convenient and rapid and well suited to quality control. A large number of types of abrasion test apparatus have been developed and several standardised nationally. Some of those of most significance to rubber testing are listed in Table 5 with brief comments.
105
Practical Guide to the Assessment of the Useful Life of Rubbers Table 5 Abrasion test apparatus Equipment name
Description
Akron
Wheel on wheel geometry, notable for ability to vary degree of slip by relative angle of wheels
DuPont (Grasselli)
Pair of small, flat faced moulded test pieces on paper disk
FKK wear tester
Very recent introduction from Japan using constant frictional force
Frick Taber
Abrasive wheels on disk test piece with additional flow of abrasive powder. Noted for simulating wear of flooring
Laboratory Abrasion Tester 100 (System Dr Grosch)
Sophisticated computer controlled apparatus allowing variation of several parameters. Wheel on disk geometry
Lambourn (Dunlop)
Both test piece and abrasive wheel are driven, slip being produced by eddy current braking
Improved Lambourn
Significantly improved design
Martindale
Disk test piece on cloth abrasive disk. The pattern of relative movement forms a Lissajous figure giving multiple direction wear
NBS (Footwear Abrader)
Small square test piece in contact with a revolving drum covered with abrasive paper. Particularly used for footwear compounds
Pico
Disk test piece revolved in contact with a pair of tungsten knives with a uniform flow of dusting powder
Rotating cylindrical drum (DIN, Conti)
Small disk test piece traverses rotating cylinder covered with abrasive paper, which gives a large abradant/test piece area. Standardised in ISO 4649
Rotating cylindrical mill
A number of designs involving test pieces (usually disks) together with particulate abrasive being tumbled inside a hollow rotating cylinder. The action simulates the action of free flowing abrasive materials
Schiefer (WIRA)
The test piece and abradant are two disks giving the geometry of Figure 11(e). The movement produces multi-directional abrasion. Various abradants may be used including serrated metal surfaces
Taber
A pair of abrasive wheels are in contact with a driven flat disk test piece as in Figure 11(d). The force on the wheels and the nature of the abradant is readily varied and the test can be carried out in the presence of liquids
106
Abrasion
22.5 Expression of abrasion test results Because of the difficulty in ensuring reproducibility of the abradant and/or test conditions, it is common practice to use a reference material against which results on the test material can be normalised. Some workers prefer to use a standard rubber to normalise or certify the abradant. When the abradant is certified in this way by one source only, this does not then enable corrections to be made for machine variations and ageing of the abradant. The alternative, or additional, approach is to refer results on the test material to the results obtained at the same time on a reference rubber with the objective of eliminating variability due to differences between nominally identical machines and abradants. In abrasion testing of rubber the terms Abrasion Resistance Index and Relative Volume Loss are used. The Abrasion Resistance Index is the ratio of the loss in volume of a standard rubber to the loss in volume of a test rubber, measured under the same specified conditions and expressed as a percentage. Relative Volume Loss is the loss in volume of a test rubber after being subjected to abrasion by an abradant, in a test which has been standardised using a standard rubber tested under the same conditions. In standard abrasion tests, it is usually mass loss which is the parameter measured although in certain cases the change in test piece thickness is more convenient. Because it is the amount of material lost which matters, it is usual to convert the mass loss to volume loss by dividing by the density. For example, the volume loss can be expressed as the loss per unit distance travelled over the abradant per 1000 revolutions of the apparatus. A less usual practice is to express the result as loss per unit energy consumed in causing abrasion. The volume loss may also be calculated per unit surface area to give a Specific Wear Rate. Whatever measure is used to represent the loss, the rate of wear may not be constant because of inhomogeneity of the test piece, change in the test piece surface or gradual change in the nature of the abradant. To investigate the test piece effects, wear rate can be plotted against number of cycles or distant travelled. An abrasion resistance can be quoted, calculated as the reciprocal of volume loss. If a reference material has been used to normalise the abradant the Relative Volume Loss is calculated from:
Vrelative =
Vtest × Vconst Vref
107
Practical Guide to the Assessment of the Useful Life of Rubbers where:
Vtest is the volume loss of the test rubber Vconst is the volume loss of the reference rubber as defined in the standard Vref is the measured volume loss of the reference rubber
If the volume loss is to be normalised to a reference material the Abrasion Resistance Index (ARI) is calculated from:
ARI =
Vref Vtest
× 100
None of these measures of abrasion resistance are fundamental properties of the rubber but relate only to the specific conditions of the test. If abrasion volume loss is measured as a function of such parameters as speed, temperature, degree of slip, energy consumed, etc., it is possible to perform a multiple correlation analysis and obtain a composite measure of abrasion resistance. This is not possible or is at least extremely tedious with most abrasion apparatus but can be achieved automatically with the Laboratory Abrasion Tester 100 (System Dr Grosch).
108
23
Other Degradation Agents
23.1 Biological attack Although rubbers and/or additives can prove attractive to living organisms, attack in service is relatively rare. Nevertheless, there can be serious problems in tropical countries and there has been considerable concern that products such as rubber pipe seals are susceptible in temperate climates. The other side to the coin of attack by living organisms, particularly microorganisms, is that sometimes it is welcome. Increasingly, biodegradable polymers are being introduced. These are specially formulated so that they are broken down relatively quickly by microorganisms and hence their disposal after use causes no environmental problems. Although measurement of the effects of exposure to living organisms is likely to involve physical properties the exposure itself is biological and is most unlikely to be carried out in a polymer laboratory. Normally, the exposure will be entrusted to experts in specialist laboratories. ISO 846 covers resistance of plastics to fungi and bacteria and could be applied to rubbers. British Standard, BS 7874, covers microbiological degradation of pipe joint rings. The concept of acceleration is perhaps difficult to apply to biological exposure: the procedure will involve making the conditions most suitable for attack, using a narrow temperature range and added nutrients; whereas under other conditions, such as absence of the organism, there will be no attack.
23.2 Ionising radiation The intensity of ionising radiation at the earth’s surface is not high enough to significantly affect rubbers. Hence radiation exposure tests are only required in connection with applications in nuclear plant and possibly where radiation is used to induce crosslinking. Ionising radiation includes gamma rays, X-rays, electrons (beta), neutrons, alpha particles, etc. Each of the types of ionising radiation has a characteristic way of interacting with matter and transferring its energy. Alpha radiation has the least penetrating power and its effects
109
Practical Guide to the Assessment of the Useful Life of Rubbers are limited to the surface layers of a material, so it only needs to be considered when a surface is contaminated by an alpha emitter. Beta radiation has a range of up to a centimetre or two, whilst X-rays, gamma radiation and neutrons are very penetrating. Most often, accelerated tests are carried out using gamma radiation from an isotope source or an electron beam from an accelerator. Radiation from nuclear reactors can also be used but will be a mixed radiation which may or may not be suitable for the simulation. The penetration of an electron beam is inherently limited which means that only relatively thin samples can be treated. Hence, gamma irradiation is the more versatile technique. With thin samples, such that penetration limits are not a problem, there are conversion factors to equate approximately the various radiations and energies to an equivalent gamma dose. The unit of radiation dose is the Gray (which is 100 times larger than the earlier unit, the Rad). In accelerated tests the dose rate might be up to 10 Gy/s whereas in service rates are often below 1 mGy/s. Particularly at higher levels of exposure, the effects can depend on dose rate and it is good practice to use at least two rates to detect any rate dependence. It should be noted that temperature can have a significant effect on the degradation as it controls both the rate of oxygen diffusion into the material and the rates of reaction of the products of the irradiation. The diffusion of oxygen is a limiting factor (as it is with heat ageing) and oxidation is directly connected to the dose rate effect. The specialist equipment and expertise needed to carry out irradiation tests is clearly not generally available and the limited interest means that there has not been wide-scale standardisation of test procedures. However, there is an international guide to determining the effects of ionising radiation on insulating materials given in the four parts of IEC 544, and also a recommended practice for exposure of polymeric materials to various types of radiation in ASTM E1027. Although the exposure to radiation needs specialised knowledge and test facilities, the methods used to monitor changes are the same as for any other ageing test.
23.3 Electrical stress Clearly, electrical stress is only of concern in electrical applications of rubbers and generally restricted to incidences of high voltage or current. Like exposure to radiation, the interest is rather specialised and there are no general standard test methods for evaluating durability of polymers over long times. The general approach is to simulate the service stress and the standard methods for such parameters as breakdown voltage or tracking may be useful starting points. Acceleration
110
Other Degradation Agents can be achieved by raising the voltage or current, as relevant, or increasing the frequency of application. Apart from the direct electrical stress, heating due to the current or dielectric heating will result in thermal ageing. Conversely, thermal or UV ageing will promote tracking and may reduce breakdown strength. A particular area of concern is partial discharges occurring within or at the surface of insulation. If there are local flaws or voids, the local electrical stress can exceed the breakdown level of the material at that point. With an AC voltage, partial discharges occur at each half cycle and the ‘electrical fatigue’ causes degradation of the insulation and a conducting path develops, often referred to as treeing. Water treeing arises with insulated cables in wet conditions where water diffuses into the insulation and, under particular conditions, forms fine channels. Accelerated test methods have been developed by a CIGRE (International Council on Large Electrical Systems) Task Force [27].
111
Practical Guide to the Assessment of the Useful Life of Rubbers
112
24
Service Conditions
24.1 General One of the fundamental problems of assessing service life is the uncertainty of and the variation in service conditions. Unless the intended application is very specific it is generally necessary to refer to typical, average or worst case conditions. A common cause of failure is due to a product having been designed for or tested in particular conditions and then used in harsher conditions. However, always designing for the worst case scenario could be prohibitively expensive. For general meteorological data it is worth noting that there is lot of information available from responsible organisations such as the UK Meteorological Office, some of it free via their web sites.
24.2 Temperature For products intended for operation at elevated temperatures it would be expected that the temperatures would be known. Where the operating temperature cycles, the maximum might be used or an equivalent temperature dose estimated on the basis of the Arrhenius relation. For operation indoors at ambient temperature it is usual practice to take 20 or 23 °C, largely because these are seen as the normal temperature in laboratories. The actual temperature in factories, warehouses and homes could clearly be somewhat different, particularly in different parts of the world. Outdoor temperatures are less easy to quantify, not only because of different climatic conditions in different places but because temperature is very dependent on the degree to which the product is exposed to sunlight, whether or not it is enclosed and its colour. In temperate climates it is common to again take 20 or 23 °C as average and this is probably sufficiently accurate in many cases. However, omitting to take account of the temperature of products used or stored in sunlight when selecting materials, can lead to unexpected rates of deterioration or failure. 113
Practical Guide to the Assessment of the Useful Life of Rubbers Conversely, the temperatures reached during weathering tests, both natural and accelerated, need to be considered when assessing the results. Because of the great variation in practical conditions and the fact that accurate surface temperature measurement is not easy, there is some spread in reported figures for natural exposure. Any estimate of temperatures likely to be reached is approximate but the following can be taken as a useful guide, temperatures are in °C: Ambient 26 Ambient 34
Black sample 50 67
White sample 33 46
If 50 mm of insulating backing is used a black sample may reach 70 or even 80 °C and a white sample above 50 °C. Temperatures under glass, such as inside a car, can exceed 100 °C and this is simulated in trials by mounting test pieces in special black box devices. The effect of sample colour is illustrated by the following list relating maximum temperature (°C) likely to be reached, to colour at an ambient temperature of 26 °C: White 33 Green 43
Yellow 38 Grey 47
Red 40 Brown 49
Blue 41 Black 50
24.3 Solar irradiation Warnings are often given that acceleration factors for relating artificial light sources with service are meaningless because of both the variation in solar irradiation and variation in spectral distribution. Regardless of this, acceleration factors are estimated, and indeed have to be if any extrapolation from accelerated tests is to be made. CIE Publication No. 85 provides data on solar spectral irradiance for typical atmospheric conditions. A condensed version of a table for maximum global irradiance at the equator is given in ISO 4892-1. Reference solar spectral irradiance can be found in ISO 9845-1 and analytical expressions for daily solar profiles are given in IEC 61725, but this sort of data cannot generally be used to provide simplistic average acceleration factors. It can, however, be noted that both total irradiation and UV content vary with the location, the time of year, the atmospheric conditions and the angle of the sun. There does not seem to be one definitive collection of measured data for various locations worldwide, although quite a lot have been collected by Wypych [28]. Figures for total
114
Service Conditions irradiation cannot be sensibly used because it is necessary to work with the irradiation at the more important UV wavelengths to make comparisons from the artificial light sources. Figures were obtained for the 295-385 nm band of 280 and 333 MJ/m2/year for Florida and Arizona, respectively. Combining these with figures given by Davis and Sims [29] for total irradiance in London and Phoenix (75 and 175, kcal/cm2/year, respectively) and making several assumptions, rough acceleration factors were calculated for artificial UV exposure: Acceleration factors for: UVA 340 lamp at 0.7 W/m2 at 340 nm Xenon lamp at 0.55 W/m2 at 340 nm
Arizona 6.0 4.7
Florida 7.1 5.6
London 14.0 11.0
Taking a typical total xenon irradiance of 1,000 W/m2, and comparing it to a typical quoted figure of 3.5 GJ/m2 for the UK gives an acceleration factor of 9. In calculating exposure times, adjustment needs to be made for light/dark cycles in artificial weathering apparatus.
24.4 Other factors For most indoor applications humidity and moisture can be ignored, although it is possible that in some cases an abnormally high level in service can be predicted. Out of doors it is quite impossible to suggest any typical level and if moisture is expected to be a problem the worst case should be considered. The worst case may well be intermittent precipitation and drying. With fluids generally, the particular chemical(s) of importance need to be identified and tests based on total immersion. The exceptions are if service is known to involve intermittent or one sided exposure, which can be simulated. Pollutants are a special case of chemical exposure. Ozone levels in the atmosphere vary considerably, but in most places are very low, no more than a few pphm. Higher levels are found in certain locations around the world and indoors, if there is a source in a poorly ventilated area. Data is collected in some countries which could be used for reference, but it is not common to apply acceleration factors to laboratory ozone tests.
115
Practical Guide to the Assessment of the Useful Life of Rubbers
116
25
Prediction Techniques
25.1 General An indirect indication of service life is obtained simply by comparison of the performance of materials under given test conditions, the one which shows the smaller change being deemed to perform better. If one material is a ‘standard’ with known service performance an estimate can be made of the other material’s expected performance. This is in fact a dangerous assumption, particularly with accelerated tests, because the differences seen under the test conditions may not be similar to the differences in practice. To make a direct estimate of service life it is necessary to apply some form of extrapolation technique to measured data. For tests made under unaccelerated conditions it is only a matter of extrapolating to longer times, which means obtaining a function for the change of the parameter(s) of interest with time. By definition an accelerated test requires that the degrading agent or agents is present at a higher dose than that to be seen in service. The general procedure is to measure the degree of degradation by monitoring changes in selected properties of interest as a function of time of exposure to the degrading agent and, unless there is previous knowledge, it is necessary to carry out tests at a number of levels of the agent. There are then two stages to modelling the degradation process: i)
Obtaining a function for the change of the parameter(s) of interest with time.
ii) Obtaining a function for the rate of change of the parameter(s) with the level of the degrading agent. Using these relationships the change in the property on exposure to longer times and lower levels of the degrading agent can be predicted. Clearly, the success of the process is critically dependent on the validity of the models used. Whilst a number of models applicable to polymers have been known for a long time, they are in practice relatively infrequently applied and the majority of accelerated durability tests carried out are used on a comparative basis only. As discussed earlier, there are a number of reasons for this, not least that there is a lack of evidence for the universal validity of the models and the
117
Practical Guide to the Assessment of the Useful Life of Rubbers behaviour found for many materials is very complicated. It is also a fact that the generation of data over sufficient times and levels of agent is an extremely time consuming and expensive process.
25.2 Standardised procedures International and national standards for predicting lifetimes of polymeric materials are mostly conspicuous by their absence. The only aspect to have been standardised is the application of the Arrhenius relationship (see Section 25.5) to evaluating accelerated test results involving the effect of temperature. IEC 216 is a guide to evaluating the thermal endurance of electrical insulating materials; ISO 2578 applies to the determination of time/temperature limits after prolonged exposure to heat for plastics. In both cases the emphasis is more on finding maximum service temperatures rather than extrapolating to normal ambient temperature. Use of the Arrhenius relationship has also been standardised for rubbers in ISO 11346. This standard is currently being revised to include the WLF model for time-temperature superposition.
25.3 Models for change of parameter with time The change in a parameter with time may take several forms, which may vary with the level of the degrading agent as well as with the parameter chosen. A common pitfall is using too great an acceleration resulting in a change in the degradation mechanism and, hence, the rate of change. The difference in the degree of change with different monitoring parameters should also be emphasised – the best practice is to use properties of direct relevance to service. Some possible forms of the change of parameter with time are shown in Figure 12. The easiest form to handle, a linear relationship, is unfortunately not found frequently because of the complicating effect of several factors interacting. There may be an induction period, due for example to protective additives, or an initial non-linear portion while equilibrium conditions are reached. A chemical reaction may produce a linear change whilst a physical effect may be logarithmic and the two may occur together. An autocatalytic reaction will show an increasing rate after a period of time. The cyclic trace shown in Figure 12 is a real example of the changes in the modulus of a rubber after ageing at an elevated temperature, with firstly the effect of more curing, then softening and finally the rubber becomes brittle. Generally, a complex form for the change of property with time indicates that more than one reaction is taking place.
118
Prediction Techniques
Figure 12 Property change with time
A general relation for the change of a property (P) with time is sometimes quoted:
⎛ t⎞ P = P0 ⎜1 − ⎟ ⎝ tf ⎠ where:
k
P is the property at time t P0 is the property at time = 0 t is time tf is the time at which P = 0, and k is a constant
This is said to describe a property which reduces from P0 to 0 over time tf. If the final value of P is not zero this value can be added to the right-hand side of the equation. when:
k = 1 the change of a property with time is linear k > 1 it degrades rapidly to start and slows down k < 1 degradation is more rapid at the end
119
Practical Guide to the Assessment of the Useful Life of Rubbers This is essentially a power law and experience with the results of the Rapra accelerated ageing programme is that it cannot be universally fitted to ageing results. Also, it could not be expected to describe changes in compression set or stress relaxation. In some cases it may be possible to transform a curved plot to linear form, for example by taking logarithms. In other cases a relatively simple relation can be found to fit, such as the power law mentioned above or the square root of time. With composite curves it may be justifiable for the end purpose requirements to deal only with one portion, for example by ignoring what happens before an equilibrium condition is reached. The cyclic rubber modulus example was dealt with by considering the fairly linear softening section, on the basis that curing would not happen at normal ambient temperature and useful life had been exceeded when brittleness set in. It is common practice when similar materials are being compared to ignore the shape of the curve and to take the time for the property to reach some percentage, say 50%, of its initial value. This may be expedient but is clearly less satisfactory than modelling the curve and could be extremely misleading if materials with substantially different curves are compared. If a time/temperature shift method is used to model the effect of temperature (see Section 25.6) no function to describe the change of property with time need be assumed.
25.4 Environmental degradation tests When the form of the change in a parameter with time has been established and a suitable measure to represent that form selected, the relation to the level of the degradation agent is needed to allow extrapolation to the service level. Generally, measurements need to be made at several agent levels to establish a model with reasonable confidence. Typically five levels are considered satisfactory. However, it should be noted that when extrapolation is to be made over several decades of time the uncertainty of the prediction will be large, even if the measured data looked very consistent. Estimates of uncertainty should always be made. It is feasible to make an empirical fit to a graph of change against agent level although it can be dangerous to do this with no theoretical justification. In cases with multiple degrading agents, and hence a complicated relation, it could be the only option, but normally an established form with theoretical justification is fitted if possible. A number of models which have been used are considered below. Before applying any model it is essential to have confidence that the input data is valid. There are repeated warnings in the literature of data being invalid because it is obtained at such accelerated levels of the degradation agent that reactions occur which are not seen at lower levels.
120
Prediction Techniques
25.5 Arrhenius relationship The best known and most widely used model is the Arrhenius relationship which in particular is applied to the permanent effects of temperature as the degrading agent. The Arrhenius relationship is:
−E ⎞ K(T) = A exp⎛ ⎝ RT ⎠ Thus ln K(T) = where:
-E + C RT
K(T) is the reaction rate for the process E is the reaction energy R is the gas constant T is absolute temperature A and C are constants
A plot of ln K(T) against 1/T should yield a straight line with slope E/R. The Arrhenius relation is generally the first choice to apply to the effects of temperature, but no general rule can be given for the measure of reaction rate (change of parameter with time) to be used with it. Very frequently the time taken to reach a given percentage of the initial value is chosen. In the example shown in Figure 13 the property parameter has been plotted against time at three temperatures, and the reaction rate taken as the time for the property to reach a given threshold value or end of life criterion (y1). In Figure 14 the log of reaction rate (time to threshold value) has been plotted against the reciprocal of absolute temperature to give the Arrhenius plot. The best fit to the Arrhenius plot can be found by the least squares method and extrapolated to find the time (tu) to the threshold value at a temperature of interest (Tu). To obtain an estimate of the maximum temperature of use, extrapolate the line to a specified reaction rate or time to reach a threshold value. 20,000 or 100,000 hours and 50% change as the threshold value are commonly used for establishing a general maximum temperature of use.
121
Practical Guide to the Assessment of the Useful Life of Rubbers
Figure 13 Change of property with time at three different temperatures
Figure 14 Arrhenius plot
122
Prediction Techniques ISO 2578 calculates a Temperature Index (TI), which is the temperature at which the chosen threshold is reached in (usually) 20,000 hours. The Relative Temperature Index (RTI) is a comparative value with a reference material. The HIC is the halving interval: the temperature change needed to halve the time to the end point from the TI. Other terms used in connection with the maximum service temperature are the UL (Underwriters Laboratories Inc.) index; and MCUT (maximum continuous use temperature) which is usually based on 100,000 hours. One criticism of MCUT (and other similar measures) is that the changes in properties are measured at ambient temperature rather than at the operating temperature. It is interesting to note that if the activation energy for degradation of a particular material is known, predictions can be made by measuring the change in a property at one temperature only. The activation energy is obtained by dividing the slope of the line by the gas constant in the Arrhenius plot. Measured activation energies are sometimes found in the literature and could be used to make approximate estimates from a single point measurement. By assuming a conservative value for activation energy, some measure of safety can be built in. The rule of thumb that reaction rate doubles for each 10 °C rise in temperature is a particularly crude way of applying this principle. The inherent weakness of the Arrhenius approach is in the assumptions which are made. The relation describes a simple chemical reaction whereas in practice the reactions are likely to be complex. It is assumed that the reactions at the service temperature are the same as those at the testing temperatures, that the activation energy is independent of temperature and that the chemical changes relate directly to the physical properties measured. If any of these are not true the relation will be invalid. Discussion of the problems and some examples are given by Le Huy and Evrard [30]. The problems were certainly evident in the Rapra accelerated test programme where, with less ageing resistant compounds, it appeared that different reactions occurred at higher temperatures. Also, in many cases the shapes of the curves of property change with time were complex and varied with temperature, indicating that more than one reaction was taking place. In these circumstances the success of applying the Arrhenius relation will depend on the measure of property change with time that is chosen. When a form of the change of parameter (X) with time other than linear is proposed, a power law is usually tried first: f(X) = Xn Combining this with Arrhenius gives a relation of the form:
123
Practical Guide to the Assessment of the Useful Life of Rubbers
⎡ At n
X = X0 exp ⎢−
⎣
n
−E ⎞ ⎤ ⎝ RT ⎠ ⎥⎦
exp⎛
There are occasions when the Arrhenius equation does not give a straight line and hence there is clear indication that predictions from it will not be valid. An alternative expression which has improved the line in certain cases is:
ln K (T ) = ln K0 + where:
B(T0 − T ) 10
K0 is the reaction rate at a reference temperature T0 B is a constant
Imposed stress will alter the rate of degradation and, with theoretical justification, relations have been proposed which predict that the log of failure time will be proportional to stress as well as to the inverse of temperature. The form of relation is:
t = t0 exp ⎡
⎢⎣
where:
E − sσ ⎤ kT
⎥⎦
t is time to failure t0 is atomic vibration period (10-13 s) E is activation energy s is structure coefficient σ is stress k is Boltzman’s constant T is absolute temperature
This implies that the characteristic activation energy is reduced by the applied stress.
25.6 Time/temperature shift An alternative to constructing the Arrhenius plot is to shift the plots of parameter against time along the time axis to construct a master curve. This approach has its origins in producing master curves of such physical effects as creep at various temperatures. However, it can be applied successfully to chemical ageing reactions. It is based on the principle of time and temperature superposition - a change in
124
Prediction Techniques temperature being equivalent to a change in rate. Essentially, the method consists of gradually shifting plots of a property against time determined at different temperatures to the plot at a selected reference temperature until the curves partially overlap. Use is made of the WLF equation to perform a time/temperature superposition:
Log( aT ) =
c1 (T − T0 )
c2 + (T − T0 )
In this expression, aT is the shift factor of an isotherm determined at temperature T, in relation to the isotherm at the reference temperature T0, and c1 and c2 are two adjustable coefficients dependent upon the material. The results for each property and each exposure temperature are plotted as a function of time with the results from different temperatures plotted as different lines. Typically, a log time X-axis is used. Taking the reference temperature as fixed, the lines corresponding to each of the other temperatures are slid in turn along the X-axis until the best possible overlap with the line at the reference temperature is obtained (see Figure 15). In this way, a ‘master curve’ is constructed at the reference temperature which simulates how the material would behave over a much wider timescale than was possible to realise by direct experimentation. The amount by which each line at the non-reference temperatures is moved (movement in the
Figure 15 Principle of constructing master curve
125
Practical Guide to the Assessment of the Useful Life of Rubbers positive direction is movement towards longer times and vice versa) is the shift factor (or, if the X-axis is a logarithmic scale, the log of the shift factor). By definition, when T = T0, log(aT) = 0 and there is no shift to apply. A plot is made of the value of log(aT) for each temperature against the corresponding temperature value as shown in Figure 16 (sometimes absolute temperature is used, although mathematically this is unnecessary since temperature differences are in fact used for the shift). Standard curve fitting techniques are used to determine the best fit for the WLF equation to give values of the coefficients c1 and c2. Alternatively, in the absence of curve fitting software the equation can be re-written in a straight line form and then the same linear regression method, as for the Arrhenius method, used to find the coefficients. The straight line form is given by: u = r.v + w where:
u = 1/[log(aT)] v = 1/(T-T0) r and w are coefficients
Figure 16 Shift factor plotted against temperature
126
Prediction Techniques Having found the coefficients r and w the WLF coefficients can be found from: c1 = 1/w; c2 = r/w To obtain an estimate of lifetime, use the WLF equation to determine the shift factor from the reference temperature to the temperature of interest. Apply that shift factor to each of the points on the master curve to obtain the required property/time curve and read the time to reach the threshold value. To obtain an estimate of the maximum temperature of use, extrapolate the line to a specified reaction rate or time to reach the threshold value. 20,000 hours is commonly used as the time for establishing a general maximum temperature of use. This technique has the advantage that no particular measure of the reaction rate has to be chosen nor any form assumed for the change of parameter with time, but it can only be used if the curves at different temperatures are of the same form. In principle, other relationships between the shift factors and temperature could be fitted on an empirical basis but, with no theoretical justification, particular caution would be advised with extrapolation. There is a problem because of the inherent discontinuity in the WLF equation. The form of the equation is such that if, in the denominator, the best fit estimate for c2 is equal to (T – T0) at a particular value of T, the expression for the shift factor reaches a discontinuity. At temperatures lower than this ‘critical’ temperature the log (shift factor) changes from being large negative to large positive and then decreases to zero as T is lowered further. The effect of this is that for certain compounds, the extrapolated temperature is in the critical region. This leads either to abnormally long times being predicted (millions of years or more) if the temperature was just above critical, or abnormally short times (fractions of a second) if the temperature was just below the critical point. In these circumstances a modified approach to the shift factor can be used and the formal WLF equation abandoned. The shift factor concept is still used but for these situations an Arrhenius equation is fitted to the shift factor [note NOT to the log(shift factor)]: aT = P x exp(Q/RT) where:
P and Q are coefficients found by best fit calculation R is the gas constant T is the absolute temperature
This has the advantage that it has no discontinuities and so a smooth temperature transition is assured.
127
Practical Guide to the Assessment of the Useful Life of Rubbers Comparisons of Arrhenius and WLF have not been found in the literature. Rapra experience of using both is that, although Arrhenius is mathematically simpler, with computer help WLF is easier to use because of there being no need to specify a measure of reaction rate nor to make any assumptions when interpolating between points. The WLF approach is also more versatile in that it is relatively easy to produce predictions in terms of time to reach an end point and as change in a given time. With Arrhenius this necessitates re-doing the calculation completely with a different measure of reaction rate. Predictions made by the two techniques for the Rapra accelerated ageing programme were generally similar with good quality raw data but, not surprisingly, differences could occur with complex or inconsistent ageing curves. Also, because of the inherent discontinuity in the WLF relation, anomalies did occur in extrapolation to ambient temperature unless the modification mentioned above was applied. Examples of the practical applications of the Arrhenius and WLF relations are given in the Appendix.
25.7 Artificial weathering Weathering is clearly a more complicated case than heat ageing alone because there are temperature effects added to the light and probably other agents such as moisture, ozone, etc., as well. Not surprisingly there is no very widely accepted relationship equivalent to Arrhenius. The result is that many workers have developed empirical relations which are usually only shown to be applicable to a narrow range of materials and conditions. An attempt can be made to combine the various effects. In a number of cases at least, the rate of degradation can be considered as proportional to radiation dose: X = X0 + bD where:
D is radiation dose b is a constant X is the property after receiving dose D X0 is the initial property
This can be combined with the Arrhenius relation for temperature effects to give a relation of the form: X = X0 D exp(-E/RT)
128
Prediction Techniques If there is a synergistic effect between the radiation and the temperature then a second exponential term can be introduced. Likewise, a further term can be added to cover the effect of moisture, which will also be temperature dependent. Other, non-linear, proposals for the relation between property and radiation dosage include a power law and an exponential relationship: X = X0 + bDn or X = X0 + a exp(D) where:
a is a constant
Another form used is: K = 10b(D-a) where:
K is the ratio of properties X /X0 and the constant ‘a’ represents an induction period before degradation starts, which is commonly found with plastics.
Note that radiation dosage has been used, although this is substituted by time of exposure by many workers. In the Rapra accelerated weathering programme, a variety of shapes for the curves of property change with time were found. Most could be represented by a power law or exponential relationship but a few were apparently complex. The property changes measured were the sum of the effects of UV light and temperature (and possibly moisture). In the complementary accelerated heat ageing trials, some of the curves of property change with time were far from simple and it would be expected that this would be reflected in the weathering results. The relationship of degradation to light intensity is certainly nearer to linear than is the case with temperature. The implied timescale would add considerably to the effort needed to collect data at a series of acceleration levels. Also, only very recent weathering apparatus would be capable of operating over a range of irradiance levels. The result in practice is that it is extremely uncommon for more than one irradiance level to be used and degradation is assumed to be proportional to irradiance. By this is meant irradiance of
129
Practical Guide to the Assessment of the Useful Life of Rubbers the same spectral distribution. Different distributions, particularly in the UV regions, will inevitably produce different results. A general approach is to start with a relation of the form [31]: X = f(y1 + y2 + y3 … + yn) where:
X is the property y1, etc., represent the various factors or agents which may cause degradation, e.g. dosage, intensity, spectral distribution, temperature, humidity.
Regression analysis techniques are then used to find the significant agents and produce a model for the particular data in question. This is essentially an empirical approach. Alternatively, a mathematical form for the data could be found by curve fitting without consideration of the effect of individual agents. Extrapolation is then particularly dangerous. This form of approach is probably the only way to deal with multiple agents, for example where contaminants are added to the normal weathering factors of light, heat and water, and it can be extended to include synergistic effects. However, this quickly becomes very complicated and attempts at such analysis are not common. In many weathering trials results are only available for one set of conditions and strictly no extrapolation can then be made for temperature. The temperature conditions in weathering cabinets can be quite high and if the temperature of service was more modest the predictions could be very misleading. If, additionally, heat ageing results are available then these can be combined with the weathering results. A further factor which accounts for a great deal of the difficulties of correlating accelerated and natural weathering is the differences in spectral distribution of the light sources. The degree of degradation is critically dependent on wavelength, but natural sunlight varies in intensity and spectral distribution, and several different lamps are used in weathering cabinets. This is not taken into account in the usual models, but in principle could be incorporated into a multi-factor regression analysis.
25.8 Ionising radiation The assumption can be made that degradation is independent of the dose rate for ionising radiation. However, acceleration levels can be very high and this is a prime reason why in practice it is often found that the effect of a given dose decreases with increased dose rate. The limiting factor is the rate of oxygen diffusion. Recommended practice is to test at two or more dose rate levels to determine the magnitude of this effect. IEC 61244-1 explains techniques for monitoring diffusion limited oxidation.
130
Prediction Techniques When heat and radiation are considered together the two effects will be additive. However, there can also be a synergistic factor [32]. Burnay [33] has developed a predictive model, which is based on the use of the superposition technique to determine thermal and dose rate shift factors relative to a master curve of compression set versus time. The relation between the shift factors and environmental parameters of temperature and dose rate are given by:
⎡ Ea ⎛ 1 1 ⎞ ⎤ ⎜ T − T⎟⎥ R ⎠⎦ ⎣ ⎝ ref
aT = exp ⎢
where
Ea = C1 - C2log10D aR = 1 + C3Dχ
where:
aT and aR are the thermal and dose rate shift factors, respectively Ea is the activation energy Tref and T are the reference and service temperatures (Kelvin), respectively D is the dose rate C1, C2, C3 and x are empirical constants
Values of the thermal and dose rate shift factors are found using the equations at the temperature (T) and dose rates of interest. These factors are then used to determine the behaviour in service by shifting the service timescale (ts) by:
ts = where:
tm aT aR
tm is the time on the master curve at T + Tref and D = 0
An interesting observation in this work was that the effect of radiation was such as to prevent the uptake of water in the seals tested. Three methods used to extrapolate from high dose rates to those more typical of service are given in IEC 61244-2. It has been observed [34] that gamma radiation reacts with oxygen to form ozone which can attack the surface of the test piece. The result was an increase in modulus at the surface and cracking if the test piece was strained.
131
Practical Guide to the Assessment of the Useful Life of Rubbers
25.9 Effect of liquids The case of liquids is simpler than weathering in that it is essentially a two agent situation, liquid and temperature. However it is generally necessary to take account of the rate of diffusion of the liquid into the material, which may be slow in relation to the timescale of an accelerated test. Also, it is necessary to consider that there may be physical change (swelling) of the polymer as well as chemical degradation. In fact, with rubbers the swelling effect is frequently the more important factor. Clearly it is advantageous to work with thin test pieces such that equilibrium is obtained quickly, but this is not always possible and extrapolation to thicker products may be needed. When equilibrium absorption is attained well within the time of the experiment the situation is similar to heat ageing. The form of change with time of the property used to monitor degradation has to be modelled or a degree of degradation specified. Then it is sensible to use an Arrhenius relation to account for temperature change. In many practical circumstances chemical degradation can be considered negligible and only the effect of physical swelling considered. Diffusion in the unsteady state before equilibrium is described by Fick’s 2nd law of diffusion:
dC dt where:
= D
d 2C dx 2
C is volume t is time D is diffusion coefficient x is thickness
As a general rule the time to equilibrium is proportional to the square of thickness. When the diffusion coefficient is known a suggested estimation of the time to reach equilibrium to a depth b is: t = b2/2D Both the diffusion coefficient and the solubility coefficient vary with temperature in accordance with an Arrhenius relationship. The diffusion coefficient increases with temperature, but the solubility coefficient increases for gases and decreases for vapours. For a full treatment of absorption, a text on mass transport should be consulted.
132
Prediction Techniques The degradation curve is likely to show a marked change in shape for a property measured up to, and past, equilibrium absorption. This is very noticeable in compression stress relaxation results for some rubbers in water at ambient temperature – a rise in modulus can be seen due to swelling at times greater than one year. In the case of water, the amount of water absorbed at equilibrium is dependent on humidity. In some, but by no means all, cases the relation may be linear. The relation needs to be known if performance at different humidities is to be estimated. Again, absorption will change with temperature. It becomes apparent that to transpose data from different humidities, temperatures and thicknesses and varying stages of approach to equilibrium can be very involved. Further, whilst the transport relationships apply to the uptake of fluid, the effect of a fluid on properties at times below equilibrium can never be simple because the concentration varies with thickness.
25.10 Effect of gases The treatment of gases and vapours is generally similar to that of liquids. A particular case for rubbers is resistance to ozone. The results of accelerated tests are almost exclusively used to make comparisons between materials rather than to attempt prediction of time to crack under ambient concentrations. If results are obtained of time to cracking as a function of applied strain, there is often an apparent threshold strain below which there is no cracking or the time to cracking is very long. (A limiting threshold strain is one below which no cracking occurs even at very long times.) In prolonged natural exposures there can be less evidence of a limiting threshold strain than in accelerated tests and hence this simple approach to prediction can be dangerous. If results are obtained as a function of ozone concentration, in theory it should be possible to make extrapolations to ambient conditions, by empirically fitting a relation to the curve of concentration against time to cracking/crack growth rate. For natural rubbers there has been evidence that the relation is broadly linear.
25.11 Creep and stress relaxation When a polymer is subjected to a stress or strain the observed stiffness changes; an increase in strain is seen in a creep test and a reduction of stress in a relaxation test. The
133
Practical Guide to the Assessment of the Useful Life of Rubbers changes will be composed of physical and chemical effects: physical processes dominate at lower temperatures and shorter times, chemical or degradation effects dominate at higher temperatures and longer times. If we consider the case of tests at ambient temperatures and moderate times then the effects measured are principally physical. Creep and stress relaxation tests under these conditions may need extrapolating to longer times. Ignoring any possible degradation effects, it is commonly found with rubbers that a plot of modulus against log of time will yield a linear relationship which makes extrapolation very easy. It should be emphasised that it is generally dangerous to extrapolate in this way greater than one decade in time. At longer times other processes may come into play, including in particular the effects of chemical ageing. The total relaxation can be expressed as:
⎛t⎞ δσ = A log⎜ ⎟ + B(t − t0 ) σ0 ⎝ t0 ⎠ where:
σ0 is the reference stress at time t0 δσ is the change in stress at time t A and B are the physical and chemical relaxation rates
Creep or stress relaxation tests may be made at a series of different stresses or strains to create a family of curves. A relation where strain is a function of stress and time will exist between the curves and this relationship can be represented as a surface in three dimensions. Computer techniques allow this surface function to be generated, including smoothing of the experimental data. When creep or stress relaxation tests are carried out at elevated temperatures, ageing effects are generally present. Often, after the exponential physical change, the chemical effect is linear. Extrapolation to lower temperatures and longer times may be made using the techniques discussed earlier, but is made complicated by the need to separate the physical and chemical effects. In the practical application of seals it should be remembered that the performance will be affected by changes in temperature, as sealing force will be lowered if temperature is reduced. If fluid is present the sealing force may be temporarily increased through swelling.
134
Prediction Techniques
25.12 Set Set, most commonly compression set, is closely related to creep and stress relaxation but the expression of results is such that ultimately it must asymptotically approach 100%. A function found useful at Rapra to model compression set (CS) is:
CS =
where:
100 b t⎞ ⎛ 1+ ⎝ a⎠
t is time a and b are constants for a particular material
25.13 Fatigue Fatigue can be defined as the decrease in load bearing capacity with time under load. Under constant load conditions this is termed static fatigue or creep rupture, and under cyclic or intermittent load, dynamic fatigue. Creep rupture is not often encountered with rubbers. The most simplistic approach to making predictions of fatigue life is to carry out tests at a series of stress or strain levels, and construct a curve of cycles or time to failure against stress or strain. A limiting stress or strain below which the fatigue life is very long may be found as illustrated in Figure 17, or extrapolation can be made to lower stresses by finding a relation to fit the curve. Work at the MRPRA led to the following expression for fatigue life of a strip in tension:
N = where:
G
(2 KW )
n
×
1 C0
N is fatigue life in cycles to failure G is cut growth constant K is a function of the extension ratio W is strain energy per unit volume C0 is initial depth of cut (or intrinsic flaw) n is the strain exponent dependent upon the nature of the polymer
135
Practical Guide to the Assessment of the Useful Life of Rubbers
Figure 17 Fatigue life Hence, at a constant value of K a plot of log(N) against log(W) will have a slope n. The value of n has been found to be about 1.5 for a natural rubber tyre tread and 3 for a styrene butadiene rubber tread. If no artificial cut is introduced then C0 is the effective size of a naturally occurring flaw. The strain energy density, W, can be found from the area under the stress/strain curve for the test piece and is strain dependent. The fatigue life is independent of the specimen geometry when expressed in these terms. At low strains, the equation does not adequately describe the fatigue behaviour, and there is a fatigue limit corresponding to tearing energy below which there is virtually no cut growth and fatigue life becomes very long. It will also be appreciated that at longer times and with higher temperatures there will be ageing effects which generally reduce fatigue resistance. Similarly, cracking may be induced more quickly if tests are made in the presence of ozone. A particular difficulty with making predictions from fatigue data is that the spread of failure times is generally large and not usually normally distributed. In consequence, the
136
Prediction Techniques measured data have relatively large uncertainties which are magnified considerably by any extrapolation process. Attempting to predict heat build-up in different geometries involves solving equations for heat flow. Sae-oui et al. [35] have developed a finite element method to predict equilibrium temperature using a combination of non-linear stress analysis and thermal analysis.
25.14 Abrasion Laboratory abrasion tests are notorious for not correlating with service. The abrasion process is complicated and the rate of abrasion is very dependent on the particular conditions. Laboratory tests cannot often properly represent the conditions in service and hence correlation is difficult if not impossible to find. That does not mean to say that correlations never exist but these will be application and test specific. The usual approach is to seek a correlation for the particular circumstances and to test materials on a comparative basis, rather than to predict service wear rate in absolute terms. With an apparatus which allows tests to be readily made over a range of test conditions, such as pressure and slip angle, it may be possible to establish empirical relationships.
25.15 Dynamic conditions All the treatments discussed above have been concerned with static conditions, i.e. where in the accelerated tests the level of the degrading agents has been held constant throughout one exposure, and any extrapolation to service implicitly assumes that conditions there will also be constant. In real life however it is much more likely that service conditions will be cyclic. Generally therefore, further approximations have to be made. Most commonly the worst situation is assumed, for example with temperature the reaction rate will be something like doubled for a 10 °C rise and lower temperatures will have relatively insignificant effect. With natural weathering the conditions change daily and geographically and this is the basic reason why light dosages should be used rather than time. With temperature it is theoretically possible to estimate an equivalent ‘dose’, i.e. the temperature which represents the mean of temperatures encountered, duly weighted for their degradation effects. With fluids, contact can be intermittent with drying out possible. The total chemical effect is likely to be less for less contact but there may be effects of expansion and contraction. Such conditions would best be modelled in the accelerated experiment.
137
Practical Guide to the Assessment of the Useful Life of Rubbers In some simple cases an additive approach can be successful. The service life is divided into stages, for example moderate exposure for one year, severe exposure for one month and low exposure for five years. The predicted effects for the three periods can be summed and the condition of the product at the end, and hence residual lifetime, estimated. If it is assumed that degradation is proportional to the time of exposure and that the damage from different exposure conditions can be cumulatively added then the so called Miner’s rule could be used:
1 tf where:
138
⎡⎛ ti ⎞ ⎤ ⎟ / t fi ⎥ i =1 ⎣ tot ⎠ ⎦
i=n
=
∑ ⎢⎜⎝ t
tf is lifetime at various conditions tfi is failure time for condition i n is number of conditions ti is exposure time at condition i ttot is total exposure time
26
Limitations and Pitfalls in Accelerated Testing
26.1 Limitations From the foregoing it is clear that prediction from accelerated tests is at best a hazardous procedure. To minimise the limitations in any particular case it is essential to design the accelerated trials to simplify and ease as far as possible the prediction process. Of the limitations the most important can be summarised as: 1.
Statistical uncertainty due to quality and number of test results.
2.
Quality of accelerated data, in terms of test conditions, being sufficiently valid to relate to service conditions.
3.
Validity of extrapolation procedure.
The first is a matter of minimising variability, maximising the quantity of data and minimising the degree of extrapolation needed. For all but very modest extrapolations the uncertainty will inevitably be significant. The quality of accelerated data in relation to service is dependent on a considerable number of factors, many of which have been discussed in this Guide, and must be systematically addressed when designing the trials. The validity of the extrapolation model is likely to be better the more proven the procedures used and the smaller the degree of extrapolation. There are two critical aspects to quality of data from accelerated tests: The degree to which the test parameters and the degradation agents match what is critical in service Whether the acceleration introduces reactions and behaviour that do not occur in service.
139
Practical Guide to the Assessment of the Useful Life of Rubbers The match with service conditions is limited by how accurately what happens in service is known and by practical difficulties in reproducing complicated conditions in test procedures. Quality in terms of the validity of the acceleration becomes interwoven with the validity of the model. Generally, it is not unreasonable to argue that if the data shows a good fit to the model it is valid. However, this provides absolutely no evidence that the model itself is valid, nor that the data is valid or even relevant when applied to the intended service conditions.
26.2 Pitfalls The following list provides a summary of factors that can detract from the value of, or even invalidate, accelerated test data. 1.
The programme is curtailed such that an adequate evaluation is not possible.
2.
The parameters used to monitor degradation are not the ones critical to service performance.
3.
The degradation agents used are not the ones critical to service performance.
4.
The severity of the conditions in service are wrongly estimated.
5.
Synergistic effects not investigated are important in service.
6.
The test piece history is different to that of the product.
7.
The test conditions cause a distorted view of relative performance.
8.
The parameters used are not sensitive to the degradation mechanism.
9.
The degradation mechanism in test is different to that in service (e.g. through use of too high temperature, or wrong wavelengths).
10. Plasticisers, etc., may migrate between samples exposed together. 11. The effect of oxygen diffusion is not accounted for. 12. If equilibrium liquid absorption is not reached the comparison is invalidated.
140
Limitations and Pitfalls in Accelerated Testing 13. The assumption about the relationship between degradation and liquid concentration is incorrect. 14. Liquid/gas used is not truly representative of service. 15. The effect of drying out/cyclic exposure in service is not accounted for. 16. Small quantities of oxygen mask the true effect of a gas. 17. Insufficient test pieces, times and exposure levels are used, so that reasonable uncertainty levels cannot be determined. 18. The extrapolation procedure is not valid. 19. The significance of results is overestimated, e.g. using comparative results to make unjustified predictions.
141
Practical Guide to the Assessment of the Useful Life of Rubbers
142
APPENDIX
EXAMPLES OF THE APPLICATION OF THE ARRHENIUS AND WLF RELATIONS The practical application of the Arrhenius and WLF relations is illustrated here by their use in the analysis of the hardness changes during accelerating ageing of two compounds, B and R. The hardness of each compound was plotted against time for different exposure temperatures (Figures A1 and A2). Compound B showed relatively complicated behaviour with the shape of the hardness-time plot changing with temperature, whilst compound R showed a relatively consistent increase of hardness with time. Hardness Compound B 70°C
80°C
90°C
100°C
100
Hardness (Micro-IRHD)
90
80
70
60 0
30
60
90
120
150
180
Heat Ageing Period (Days)
Figure A1 The hardness of compound B plotted against time at different exposure temperatures
143
Practical Guide to the Assessment of the Useful Life of Rubbers Hardness Compound R 70°C
80°C
90°C
100°C
100
Hardness (Micro-IRHD)
90
80
70
60
50 0
30
60
90
120
150
180
Heat Ageing Period (Days)
Figure A2 The hardness of compound R plotted against time at different exposure temperatures This is reflected in the WLF master curves (Figures A3 and A4) where R gives a much smoother fit than B. The best predictions from these for change in hardness after 40 years at 40 °C are: B = 49% and R = 82%. After long-term (40-year) exposure in a hot dry climate (not averaging as high as 40oC) these materials actually changed by 24% and 55%, respectively. Arrhenius plots were constructed with the end point taken as 24% and 55%, respectively (which required judicious extrapolation for some curves) and are shown in Figures A5 and A6. These yielded predictions of 7 years and 6 years at 40 °C for B and R, respectively and 24 and 16 years at 23 °C. Considering the shapes of the hardness-time curves for compound B, this end point would be difficult to justify and it is perhaps remarkable that sensible predictions were obtained. The Arrhenius plot for B in particular is clearly not a perfect straight line. If the 70 °C point is ignored the prediction becomes 21 years at 40 °C and 124 years at 23 °C, whilst if the 100 °C point is ignored these figures are 3.5 and 9 years, respectively. This is not the direction one would expect and is probably an artifact of the end point used.
144
Appendix 5HIHUHQFH7HPSHUDWXUH & &
&
&
&
+DUGQHVV0LFUR,5+'
7LPH0RQWKV
Figure A3 WLF temperature shifted hardness – compound B
Reference Temperature = 23°C 100°C
70°C
80°C
90°C
100 95
Hardness (Micro-IRHD)
90 85 80 75 70 65 60 55 50 1
10
100
1,000
Time (Months)
Figure A4 WLF temperature shifted hardness – compound R
145
Practical Guide to the Assessment of the Useful Life of Rubbers
Figure A5 Arrhenius plot for change in hardness with time – compound B
Figure A6 Arrhenius plot for change in hardness with time – compound R
146
Appendix The effect of ignoring points for compound R is less drastic, reflecting the better straight line seen in the Arrhenius plot. The corresponding figures are 6 and 17 years ignoring the 70 °C point and 4.5 and 11 years ignoring the 100 °C point. Had it not been for an interest in making direct comparisons with long-term exposures, it would have been more sensible to have chosen rather lower end points. If an end point of 10% change is taken for compound B, the scatter on the Arrhenius plot is increased and the predictions are 1.5 years at 40 °C and 4 years at 23 °C. However, if the 100 °C point is ignored these become 5 years at 40 °C and 35 years at 23 °C; with linear extrapolation this is equivalent to 12 years and 84 years to reach 24% change at 40 °C and 23 °C, respectively. Taking an end point of 20% change for compound R also increases the scatter on the Arrhenius plot and yields predictions of 2 years at 40 °C and 9 years at 23 °C. Ignoring the 100 °C point yields predictions of 3.5 years at 40 °C and 20 years at 23 °C; with linear extrapolation this is equivalent to 10 years and 55 years to reach 55% change at 40 °C and 23 °C, respectively. Where the Arrhenius plot is not linear there is justification in ignoring the highest temperature as not being representative of reactions at lower temperatures. With this selective use of the data very reasonable predictions in comparison to natural exposure are obtained. In summary, several observations can be made: All the predictions showed the same trend as in natural exposure but tended to overestimate the rate of change. Considering the uncertainty in actual temperature on natural exposure the WLF predictions could be said to be good for both compounds, in spite of the shapes of the property-time curves for compound B. There is considerable scope in the selection of the end point for Arrhenius analysis which can yield different predictions. This is particularly the case if the shape of property-time curves is complex. Rejection of data where the Arrhenius plot is not linear will make large differences to the predictions and may be justifiable on the grounds of different reactions taking place. (Data can, of course, also be rejected for WLF analysis.) With selective use of the data good predictions were obtained for both compounds by the Arrhenius approach.
147
Practical Guide to the Assessment of the Useful Life of Rubbers
148
References 1.
Y. Saito, International Polymer Science and Technology, 1995, 22, (12), T/47.
2.
G.G.A. Bohm and J.O. Tveekrem, Rubber Chemistry and Technology, 1982, 55, 575.
3.
J. Boruta and A. Petrujova, International Polymer Science and Technology, 1987, 14, (10), T/59.
4.
J.A. Bousquet and J.P. Fouassier, European Polymer Journal, 1987, 23, (5), 367.
5.
M.A. Rodrigues and M.A. De Paoli, European Polymer Journal, 1985, 21, (1), 15.
6.
M. Nowakowska, Polymer Photochemistry, 1985, 6, (4), 303.
7.
R.W. Keller, Rubber Chemistry and Technology, 1985, 58, 637.
8.
P.M. Lewis, Polymer Degradation and Stability, 1986, 15, 33.
9.
R.W. Layer and R.P Lattimer, Rubber Chemistry and Technology, 1990, 63, 426.
10. P.B. Lindley and S.C. Tao, Plastics and Rubber: Materials and Applications, 1977, 2, (2), 82-8. 11. O. Kube, T. Hofer and E. Files, ACS Rubber Division 157th Meeting, Dallas, April 4-6, 2000, Paper No. 39. 12. R.P. Brown, Physical Testing of Rubber, Chapman and Hall, London, 1996. 13. B. Dinzburg and R. Bond, Rubber World, 1990, 201, (4), 20. 14. E.W. Bergstrom, Elastomerics, 1977, 109, 3, 21. 15. K.T. Gillen, M. Celina and M.R. Keenan, Rubber Chemistry and Technology, 1999, 73, (2), 265. 16. P. Budrugeac, Polymer Degradation and Stability, 2000, 68, 289. 17. J. Wise, K.T. Gillen and R.L. Clough, Polymer Degradation and Stability, 1995, 49, 403. 18. R.P. Brown (Ed.), Handbook of Polymer Testing, Marcel Dekker, New York, 1999. 19. R.P. Brown and T. Butler, Natural Ageing of Rubber, Rapra Technology, Shrewsbury, 2000.
149
Practical Guide to the Assessment of the Useful Life of Rubbers 20. R.P. Brown et al., Ageing of Rubber, Accelerated Heat Ageing Test Results, Rapra Technology, Shrewsbury, 2001. 21. R.P. Brown, T. Butler and S.W. Hawley, Ageing of Rubber, Accelerated Weathering and Ozone Test Results, Rapra Technology, Shrewsbury, 2001. 22. C. Andersson, Lifespan of Rubber Materials and Thermoplastic Elastomers in Air, Water and Oil, The Swedish Institute for Fibre and Polymer Research, Molndal, 1999. 23. D.C. Wright, Failure of Plastic and Rubber Products – Causes, Effects, and Case Studies Involving Degradation, Rapra Technology, Shrewsbury, 2001. 24. R.P. Brown, M.J. Forrest and G. Soulagnet, Rapra Review Report 110, Longterm and Accelerated Ageing Tests on Rubbers, Rapra Technology, Shrewsbury, 2000. 25. L.G. Angert and N.N. Dubok, Soviet Rubber Technology, 1970, 11, (November), 29. 26. U. Eisele, S. Kelbch and H-W. Engels, Kautschuk und Gummi Kunststoffe, 1992, 45, (12), 1064. 27. CIGRE Working Group 15.06.05, Electra, 1996, 167, 59. 28. G. Wypych, Handbook of Material Weathering, ChemTec Publishing, Ontario, Canada, 1996. 29. A. Davis and D. Sims, Weathering of Polymers, Applied Science, London, 1983. 30. M. Le Huy and G. Evrard, Angewandte Makromolekulare Chemie, 1998, 261/ 262, 135. 31. S.H. Hamid and W.H. Prichard, Journal of Applied Polymer Science, 1991, 43, 651. 32. M. Ito, International Polymer Science and Technology, 1987, 14, (7), T38. 33. S.G. Burnay, OPERA 89 Conference, Lyon, France, 1989, Vol.2, 561. 34. R.L. Clough and K.T. Gillen, Journal of Polymer Science Part A, 1989, 27, 2313. 35. P. Sae-oui, P.K. Freakley and P.S.Oubridge, Plastics, Rubber and Composites, Processing and Applications, 1999, 28, (2), 69.
150
Abbreviations
AC
Alternating current
ARI
Abrasion resistance index
ASTM
American Society for Testing and Materials
BS
British Standard
CIE
Commission Internationale de L’Éclairage (International Commission on Illumination)
CIGRE
International Council on Large Electrical Systems
DIN
Deutscher Institut für Normung
DMTA
Dynamic mechanical thermal analysis
DSC
Differential scanning calorimetry
DTA
Differential thermal analysis
HIC
Halving interval
IEC
International Electrotechnical Commission
IRHD
International rubber hardness degree
ISO
International Organization for Standardization
MCUT
Maximum continuous use temperature
MRPRA
Malaysian Rubber Producers’ Association
pphm
Parts per hundred million
RTI
Relative temperature index
TI
Temperature index
UL
Underwriters Laboratories Inc.
UV
Ultraviolet
WLF
Williams, Landel and Ferry
151
Practical Guide to the Assessment of the Useful Life of Rubbers
152
Index
A Abrasion 9, 10, 29, 40, 43, 51, 61, 62, 101, 137 Abrasion Resistance Index 107, 108 Abrasion test apparatus 105 Abrasion wear 104 Abrasive wear 101, 102 Absorption of fluid 26 Accelerated testing 3, 4, 5, 7, 23, 25, 29, 32, 33, 35, 45, 46, 51, 52, 54, 55, 57, 58, 60, 61, 69, 70, 73, 75, 77, 78, 79, 86, 91, 92, 93, 110, 117, 123, 129, 132, 133, 139 Accelerated weathering 130 Activation energy 41, 123, 124, 131 Additive depletion 9, 26, 29, 39, 55, 70, 81, 83, 140 Adhesive failure 9, 40 Adhesive wear 101 Air velocity 78, 92 Alpha radiation 109 Anaerobic condition 25 Antioxidant 14, 15, 20, 23, 33, 41, 70 Antiozonant 20, 29 Arrhenius 46, 47, 55, 113, 118, 121, 123, 124, 126, 127, 128, 132 Artificial weathering 91, 92, 115, 128
B Biological attack 109 Bio-organisms 9, 10, 109 Black panel thermometer 91 Bond scission 15, 17, 18, 20
Bond scissions 13 Breakdown voltage 110
C Carbon arc 91, 92 Chemical analysis 42 Compression stress-strain 40, 61 Conditioning 86 Crack evaluation 87 Creep 9, 10, 29, 32, 36, 40, 64, 96, 133, 134 Crosslinking 9, 13, 15, 17, 20, 25, 36, 109 Cure conditions 43
D De Mattia apparatus 97, 98 Dew 92 DIN abrader 105 DMTA 32, 37, 38 DSC 41 Dynamic stress-strain 37, 38, 87
E Effect of gases 85, 133 Effect of liquids 81, 115, 132 Effect of sample colour 114 Electrical properties 39, 43, 46 Electrical resistivity 32 Electrical stress 9, 10, 110 Electron beam 110
153
Practical Guide to the Assessment of the Useful Life of Rubbers Elongation at break 32, 33, 35, 70 Equilibrium absorption 26, 85, 132, 133, 140 Equilibrium swelling 26, 83, 84 Erosive wear 101 Experience 7, 51, 57
L Limiting threshold strain 133 Low temperature 10, 13, 29, 32, 39, 62, 77
M F Fatigue 9, 10, 29, 40, 51, 55, 61, 64, 78, 87, 95, 96, 97, 98, 99, 101, 135 Fatigue wear 101 Fick’s 2nd law 132 Flex cracking 95, 97 Flexometer 95, 96 Fluid diffusion 26 Fluid transport 26, 33 Fluorescent tubes 91, 92 Fresnel mirrors 52 Frictional properties 39, 62
G Gamma radiation 18, 110, 131 Gas permeability 39 Global irradiance 114
H Halving interval 123 Hardness 25, 29, 32, 33, 40, 77, 82 Heat ageing 9, 17, 23, 29, 33, 46, 55, 60, 78, 83, 84, 110, 111, 118, 128, 129, 130 Heat build-up 65, 95, 104, 137 Hydrolysis 9, 10, 84
Marine exposure 53, 61, 63 Master curve 125, 131 Maximum continuous use temperature 121, 123, 127 Maximum service temperature 16 Metal halide lamps 91 Miner’s rule 138 Modulus 25, 29, 32, 33, 36, 37, 38, 40, 118 Multiple agents 5, 9, 11, 29, 52, 54, 55, 61, 70, 73, 75, 78, 84, 86, 91, 92, 128, 130, 131, 132, 133
N Natural environmental exposure 25, 46, 47, 51, 91, 114, 128, 133, 137 North Sea tank 63
O Oven 78 Oxidative degradation 9, 14, 16, 18, 19, 23, 25, 26, 29, 39, 79, 83, 92, 95, 130 Oxygen absorption 14, 41 Oxygen diffusion 9, 14, 16, 19, 23, 25, 33, 35, 36, 40, 79, 110, 130, 140 Ozone degradation 10, 20, 29, 85, 86, 95, 115, 131, 136 Ozone resistance 20, 21, 52, 88, 133
I Induction period 23, 41, 118, 129 Ionising radiation 10, 18, 55, 109, 130
154
P Photodegradation 19, 20, 92
Index Processability test 40 Product testing 41, 51, 52, 54, 55, 57, 59, 62, 66
R Radiation degradation 17 Radiation resistance 18 Ramp friction test rig 62 Reaction rate 10, 14, 20, 23, 24, 58, 83, 110, 121, 123, 124, 137 Relative Temperature Index 123 Relative Volume Loss 107 Repeatability 33, 43, 45 Reproducibility 31, 32, 33, 39, 45, 46, 47, 62, 66, 69, 79, 83, 107 Roll formation 101 Ross flexing machine 97, 98
S Seal testing rig 62 Sealing force 34, 36, 134 Service life 5, 29, 31, 32, 35, 51, 54, 55, 58, 60, 69, 73, 75, 85, 88, 113, 117, 137 Service trials 3, 7, 51, 54, 55, 60 Servo hydraulic test machine 99 Set 9, 10, 24, 32, 36, 37, 40, 77, 81, 96, 99, 120, 131, 135 Shear 40, 61, 95 Shelf life 70 Shift factor 124, 125, 126, 127, 131 Simulated design life 54 Software packages 46, 126, 134 Solar irradiation 19, 52, 53, 70, 91, 113, 114 Specific Wear Rate 107 Spectral distribution 130 Spectral irradiance 92, 128 Sports surface test rig 64
Standard liquids 83 Statistics 46 Stress relaxation 9, 10, 29, 32, 34, 35, 36, 38, 40, 81, 84, 87, 120, 133, 134 Stressing frame 64 Surface skin 25 Swelling 9, 10, 26, 29, 81, 132, 134 Synergy 5, 10, 11, 29, 58, 73, 75, 129, 131
T Tear Analyser 99 Tear strength 25, 40 Temperature Index 123 Tensile strength 29, 32, 33, 35, 40 Tensile stress-strain 32, 33, 82, 99 Test piece geometry 5, 25, 35, 37, 40, 43, 59, 75, 82, 83, 86, 95, 99, 105, 136 Test rigs 62 Thermal decomposition 17, 95 Thermal expansion 77 Thermo-oxidative degradation 9, 14, 20, 25 Threshold strain 20, 86, 87, 88, 89, 133 Threshold stress 21 Time/temperature shift 118, 120, 124 Tracking 110 Transition metals 15, 29, 78 Treeing 111 Tyre test machine 65
U Uncertainty 5, 43, 45, 46, 61, 73, 76, 139, 140, 141 Useful life 3 UV light 10, 13, 19, 45, 52, 53, 91, 92, 111, 115, 129
155
Practical Guide to the Assessment of the Useful Life of Rubbers
V Volume change 26, 38, 81, 82, 107
W Water 26, 39, 52, 84, 91, 92, 105, 131, 133 Water treeing 111 Wear by roll formation 101 White panel thermometer 91 WLF 47, 92, 118, 125, 126, 127, 128
X Xenon lamp 91, 92, 115
156
O3
Ra pra Te chnology Lim it e d Rapra Technology is the leading independent international organisation with over 80 years of experience providing technology, information and consultancy on all aspects of rubber and plastics. The company has extensive processing, testing and analytical facilities housed in over 7,500 sq. metres of laboratory and production space. It provides testing to a range of national and international standards and offers UKAS accredited airborne hazard monitoring and analytical services. Rapra also undertakes commercially focused innovative research projects through multi-client participation. Its expertise is disseminated through reports, training and software products including computerised knowledge-based systems. Rapra publishes and sells books, journals, technological and business surveys, conference proceedings and trade directories. Rapra Abstracts is the world’s most comprehensive database of commercial and technical information on rubbers and plastics.
Shawbury, Shrewsbury, Shropshire SY4 4NR, UK Telephone: +44 (0)1939 250383 Fax: +44 (0)1939 251118 Email: [email protected] http://www.rapra.net