Volume 2 Paper 15

Short Term Testing and Real Time Exposure

James Maxted



Because you are not logged-in to the journal, it is now our policy to display a 'text-only' version of the paper. This version is obtained by extracting the text from the PDF or HTML file, and it is not guaranteed that the text will be a true image of the text of the paper. The text-only version is intended to act as a reference for search engines when they index the site, and it is not designed to be read by humans!

If you wish to view the human-readable version of the paper, then please Register (if you have not already done so) and Login. Registration is completely free.

JCSE Volume 2 Paper 15 ifnotmode(3,'Submitted 13th September 1999') Short Term Testing and Real Time Exposure James Maxted Akzo Nobel Nippon Paint , North Woolwich Road , Silvertown , London . E16 2AP mailto2('James.Maxted','uk.akzonobel.com'); §1 Abstract Three case studies are used to illustrate the difficulties in determining long term corrosion performance and in-service early failure of pre-coated, galvanised steel by the use of laboratory accelerated techniques. Variants of cyclic exposure tests tend to provide better realism in terms of the mode of corrosion, especially at the cut edge of the coated sheet. However, even these do not replicate the exact mode of failure, particularly with regard to the later stages of cut edge corrosion, namely the corrosion of the steel itself. The examples chosen also illustrate the importance of accelerating factors on the initiation and propagation stages of cut edge corrosion, arising from elements such as building design, orientation, and macro / micro-climates. An integrated, corrosion prediction methodology is suggested, combining the response data from fundamental tests, from a broad spectrum of accelerated tests and from a number of outdoor exposure series. These are then linked together in a central model to provide a reliable, predictive guide of performance.   §2 Keywords: Cut edge corrosion, Coil Coating, Accelerated testing, Cyclic corrosion tests, UV exposure, Hot dipped galvanised steel, Chromate free primers, Outdoor exposure, Service life prediction. Introduction §3 Organically pre-coated metal sheeting is a well established product for the construction industry, comprising about 20% of all roof and wall cladding areas in the UK and probably more so in mainland Europe [1]. The corrosion performance of these coated metals is controlled by the combined effects of a galvanic coating, a pretreatment of a controlled metal oxide/ chromic acid rinse, and finally by the application of organic coatings. These organic coatings themselves perform a number of different functions from the point of view of corrosion protection. The primer layer has to be formulated so as to maximise adhesion to both the pretreated substrate and the overlying topcoat. At the same time this coating also has to function as the vehicle in which to carry the inhibitive pigments. Finally, in combination with the topcoat, the coating system functions as an ionic barrier coating (Figure 1). §4 A recent survey of UK installations [2,3] suggested that failures of such materials within their product lifetimes probably amount to no more than 0.25–0.5% of the total area usage of such precoated, materials in the UK. That said, more than 75% of these failures were linked to corrosion problems and specifically those related to critical areas such as the exposed cut edges of sheets in roofing and side cladding situations shown in Figure 2. §5 The issue that is of most concern to the coating’s formulator and user is that of trying to understand and model such corrosion processes. In many cases the type of failure will not have been predicted in the development and quality checks that the same organic coatings were subjected to in the laboratory. §6 Another important concern to be addressed is that of predicting the service life of coatings, particularly in environments where there is no exposure history of the system. The coatings and painted sheets are often supplied to the end user with a guaranty of the time to first maintenance [4]. Similarly, international performance standards are also calling for a minimum level of performance [5] and so some means of determining the coated system’s performance over the broadest range of environmental stresses has to be established. Again neither the traditional test methods nor a limited range of severe, natural exposures can be anything other than an approximation of performance. §7 A further issue to face the coating’s formulator is that of choosing suitable techniques to use in product development. New demands in the coil industry mean the development of more environmentally compliant coatings for the future [6]. This in turn means a move away from traditional pretreatments based on chromic acid, away from primers containing  hexavalent chromate and from thick film, barrier coatings based on polyvinylchloride, all of which have been the mainstay for corrosion protection for so long. To avoid erroneous and untimely conclusions from unreliable tests being made, a new approach to performance assessment has had to be developed alongside the development of the formulations themselves. §8 Several examples are used in this paper to illustrate these issues. The focus is placed upon the performance of zinc galvanised steel at the critical area of sheet cut edge, concentrating on the mode, rather than mechanism of corrosion in the field compared to that generated by accelerated tests. Case Study 1. Agricultural Building Roof, North West England. Observation of Corrosion §9 This building was erected in 1988 on the north west coastal area of the UK and by 1993 cut edge corrosion was observed at the gutter drip edge of the roof sheeting on the westerly facing pitch of the roof (Figure 3). This side of the roof faced towards the Morecambe Bay estuary, about 20 metres away. The other side of the roof showed no such signs of deterioration (Figure 4). §10 The edge defect comprised peel back of the organic coating to a maximum of 7mm up the sheet. On the surface of the exposed substrate both zinc and steel corrosion could be observed. The zinc corrosion product was seen as a light deposition in a tidemark fashion. At the very edge and, in some cases extending up to 4 mm, the zinc anode had been totally consumed and steel corrosion was occurring. §11 The first response was to review the results of the same coated product on other exposure sites and accelerated tests. In short could this type of failure have been predicted? §12 Saltspray performance, conducted according to ASTM B117 – 90 (Figure 5a), showed only slight edge corrosion after 1000 hours of continuous testing. At the exposed cut edge there was no visible peel back of the coating, rather only very slight blistering (<2/2 according to ISO 4628-2 82) The only other evidence of corrosion was that of dense, compact, white zinc corrosion product accumulated on top of, rather than underneath the coating. Where blisters could be removed, a slightly yellow liquid was released and the substrate underneath found to be a dull metallic colour with little or no visible, white zinc corrosion products. This is similar to that reported by others [7,8]  §13 Like saltspray testing, Prohesion� cyclic testing (conducted according to ASTM G85-94), resulted in little corrosion at 1000 hours of total testing (Figure 5b). However the cyclic testing conditions resulted in primer edge peel and lift rather than blistering at the exposed cut edge. There was also some build-up of solid corrosion product under the coating. This was beginning to form a wedge and so lift the coating further. However the degree of delamination was still slight at the end of test (<2mm from the edge). There was no evidence of anything other than zinc corrosion products, albeit they had a different, more open and voluminous morphology to that produced in the saltspray test. Importantly there was no evidence of red rust and associated steel corrosion products. §14 After 3� years of Scab Corrosion testing [9] (Figure 5c), considerably more corrosion was seen on the panels, particularly at the exposed cut edges. This comprises a heavy deposition of zinc corrosion products underneath the coating and (at the extreme edge) on top it. Red rust was also visible at the edge and below the zinc corrosion layer. In places the organic coating had become embrittled and had peeled back or flaked off. On a macro scale this test shows all the components of corrosion seen on the roof of the building. However the pattern is still not quite the same, particularly in the morphology and quantity of solid, zinc corrosion product formed, which differed considerably from the ‘tide mark effect’ seen in the field. §15 5 years Hook of Holland Exposure performance is shown in Figure 5d. These panels do show the same type of mechanism of failure albeit on reduced scale over a similar time period. Although the pitch was similar at 5� from the horizontal, the test panel didn’t face the prevailing wind direction during exposure and only the overlap rather then the drip edge was corroded. That said the corrosion was characterised by organic coating peel back and flaking, tidemark lines of the zinc and some red rust at the extreme edge. Unlike the corrosion products generated by the Prohesion� test, on this exposure they were not voluminous and did not appear to form a wedge. This infers that other processes were also responsible for the initial loss of adhesion.     §16 Figure 5e shows the corroded edge of the product after 1500 hours testing on a combined Cyclic Fog /UV Exposure test. This new test follows the procedure outlined for testing of other exterior coatings [10,11,12] and which has recently been embodied in the ASTM D5894-96 standard [13]. At the end of test the coating had peeled back and flaked from the edge as seen on natural exposure in Holland and on the building. The zinc corrosion product was not as voluminous as that generated by the Prohesion� test on its own and there was evidence of red rust developing at the extreme panel edge. Discussion §17 It is clear that the older, established tests were not able to predict and or replicate the type of field failure observed. This example evidences the well documented shortcomings of the continuously wet, saltspray test [14] In this test, the conditions used (95-98% relative humidity, 5 % by weight NaCl at pH of 6.5 – 7.5 and 35�C) result in a pattern of corrosion and morphology of corrosion products rarely seen in natural exposures. In particular, the very high concentration of sodium and chloride ions, in conjunction with the constant high relative humidity, result in clearly defined anodic and cathodic sites being set up on the principle of the oxygen concentration cell. The result is that as the pH in the cathodic regions rises, in some instances to as much as 12 or 13 [15], the zinc surface becomes passivated. The high pH would also mean that any corrosion product would likely remain in solution rather than precipitate as solid corrosion product to form a wedge. Furthermore this high cathodic pH can also result in base catalysed hydrolysis and saponification of the organic binder at the interface leading to wet adhesion loss. The overlying zinc corrosion product is predominantly made up of dense plate-like structures of hydrated, basic zinc chloride, Zn5(OH)8Cl2.2H2O (pseudo-hexagonal crystals), rather than the zinc hydroxy carbonates seen on natural exposure [16]. Thus both the anodic products and cathodic mechanisms fall somewhat short of reality and, not surprisingly, do not reproduce or predict the pattern of corrosion seen on the roof. §18 By contrast the wet/dry cycling conditions of the Prohesion� test result in quite a different pattern of corrosion to that of the saltspray. The inclusion of wet and dry cycling periods is now well recognised [17] as significantly influencing the corrosion process. The increase in ionic concentrations during the drying periods results in some solid species being formed as precipitates or evaporites, accumulating as a wedge between the anode and cathode regions. The example here demonstrates further the effect of cyclic conditions in the formation of solid corrosion products that can mechanically damage the film [18]. This process, allowing free access of oxygen and CO2, together with the use of a more dilute electrolyte fog (0.35wt% (NH4)2SO4 and 0.05 wt % NaCl) will favour the build-up of different corrosion products from the salt spray; predominantly the complex salt of zinc hydroxy sulphato chloro hydrate; Zn12(OH)15(SO4)3Cl3 (H2O) 5. [19]. Equally the cyclic test conditions favour the formation of compact rust layers consisting of �-FeOOH and g-FeOOH similar to that found on outdoor exposure testing [19,20]. That said, the standard test conditions used were not able to produce the less voluminous corrosion products or the degree of steel corrosion as seen on the cut edge of the roof. It is likely that the salts used in the test still overwhelm the buffering effect of surface moisture containing bicarbonate ions which would otherwise lead to the formation of basic zinc carbonates such as Zn5(OH)6(CO3)2. §19 Even though the test represents a more realistic one than the continuous salt fog , the overall performance on this test would still not predict the extent of cut edge corrosion and particularly anode depletion leading to steel corrosion, seen in the field. §20 Both the Scab corrosion test and the outdoor tests at the Hook of Holland exposure site demonstrate the same type of corrosion performance as seen on the roof. The former test is the more aggressive. The Dutch exposure was the closest in terms of the amount of solid corrosion products and the mode of coating delamination and degradation although this occurred to a smaller extent and in the area of panel overlap rather than at the free drip edge. The mode of failure appeared to be the same although it is significantly less in degree (2 mm total creep against 7 mm for the roof in the same time period). This suggests that the corrosion process be accelerated by a number of different factors. In the case of the west facing roof pitch of the building it may be the combination of more frequent and rapid rates of wetting and drying and the build up of debris and dirt at the edge that lie behind the cause of the failure. §21 For those panels exposed at the Hook of Holland the features and processes involved in corrosion of overlap regions may also provide a clue to the drip edge problem. The accelerating influences in the former instance could be longer wet times (because of entrapped water), crevice corrosion and even differential aeration cells. It is possible that the same factors may be operating on the very exposed and wet conditions on the gutter edge of the roof. §22  Of the laboratory tests the combined cyclic fog/ UV exposure test was the only one which began to demonstrate the same pattern of performance as seen on roof. In many ways the edge corrosion appeared similar in basic morphology and, in particular, less voluminous than that produced by the Prohesion� test. It was also the only accelerated technique, which caused the topcoat to loose gloss and become matt in a similar fashion to that seen on natural exposure. Loss of gloss is a common feature related to photolysis and/or photo-oxidation of the surface layers of polyester coatings. The reduction in gloss renders the surface more hydrophilic, lowering the critical RH for the onset of corrosion. The same processes can also produce microcracking and micropores, further facilitating the uptake of moisture and thereby aggressive ions [21]. That said, the test still did not give the same degree of steel corrosion (visible as red rust) as seen on the roof, even after completing 1500 hours of testing. It is possible that a modified wet/dry regime, to encourage more rapid drying and a different concentration and combination of electrolytes may better replicate the failure. §23 Thus it is clear that this type of edge corrosion failure could not be anticipated by the accelerated tests employed. However, the example does demonstrate the importance of introducing combined stresses into the accelerated corrosion protocols, rather than just using those of a continuously wet nature. It would seem that wetting and drying of the panel together with photodegradation are some of the more important ones to utilise. Case Study 2. Packaging Factory, South West England. Observation of Corrosion §24 This building was erected in 1987 in the Bristol area and by 1991 showed the type of overlap edge corrosion shown in Figure 6a and 6b. The following points were noted regarding the occurrence of this phenomenon and were felt to be significant with respect to the nature of this corrosion: §25 a)     Corrosion was limited to the roof only and occurred in the trough valleys of the profile. §26 b)     The sheet gauge used on the roof was heavier (1.1mm) than the side cladding (0.7mm) §27 c)     Corrosion was limited to the south west facing pitch main roof. Neither the north east pitch nor roofs of smaller outbuildings showed a similar problem. See Figure 6b. §28 d)     Corrosion was limited to strip overlap areas. The drip edges of the same sheets at the gutter were corrosion free. §29 e)     In the troughs there were significant deposits of extract from the factory ovens. When wet this material formed a sludge or poultice over the coated steel in places up to 0.5 – 0.75 mm thick. §30 A detailed investigation of the defect was undertaken both in-situ and back in the laboratory. This lead to the identification of a number of factors which differed between material that had failed prematurely and that which was in a good condition. These findings are summarised in Table 1. Discussion §31 Again, none of the traditional laboratory testing techniques predicted that this type of failure would occur. Salt spray [22] and high humidity [23] testing showed the coated sheets to be giving the level of performance anticipated for the product. Even the severe, accelerated, outdoor scab corrosion test [9] indicated that the pre-coated coil performance was in line with expectations. Therefore this example demonstrates further the dangers of relying on short term testing to predict real time performance. It also illustrates other variables and influences to consider when relating short term, accelerated testing results to premature in-service failure. Some act as confirmation of those influences already discussed, but there are others that require consideration: §32 a)     The importance of mixed electrolytes and their respective concentrations are a significant influence on the mode of corrosion. This has been argued for a long time [24] and the results from Table 1 clearly support the use of sulphate rich electrolyte solutions. It is interesting to compare the measured ppm levels of sulphates, and chlorides with those commonly used in the saltspray and  Prohesion� tests. The levels of chloride ion present at the overlap are two orders of magnitude less than that employed in the saltspray (50,000 ppm of NaCl). However the Prohesion� test deposits quantities of NaCl and (NH4)2SO4  (500 ppm and 3500 ppm respectively) almost identical to that measured in the uncorroded overlaps and 2 –3 times lower than in the problem areas. §33 b)     The effect of the high cathode: anode ratio significantly effected the rate of anode depletion. Our own work with accelerated tests supports this, although the picture is less clear cut with the thinnest gauges. The general trend for the tests was similar at gauges between 0.5 – 1.3 mm, even though the mechanisms of corrosion clearly were not. On the roof this ratio was ~27:1 compared with ~10:1 for the previous example. Whatever the specific process it is clear that the larger the cathode area the harder the anode is driven at the cut edge. The PUCAT test devised by Walters[25] can provide a means of assessing the coating’s response as this factor is varied by increasing the effective cathodic area. §34 c)     The comparatively low Tg (+15�C) of the topcoat exacerbated the degree of permanent deposition of the factory extract. This resulted in a rapid and significant build-up of debris at the cut edge and immediately below it forming a poultice. This, in turn acted rather like a sponge, to trap and accelerate the concentration of aggressive ions. Current accelerated tests have no means of determining the effects of either coating Tg or poultices on the overall corrosion performance. §35 d)     Both the low pitch of the roof (<10�) and the design of the sheet overlap and fixings combined to cause ponding and lengthen the wet times at the edge. Additionally no sealant had been used at the overlap, so allowing moisture to be trapped underneath. This will promote the set up of differential aeration conditions, setting up conditions which favour crevice corrosion. These may also have contributed to the premature corrosion at the cut edge. Whilst recognised as significant influences on the corrosion process [26], these influences are not routinely exerted into tests in the laboratory §36 In addition to incorporating combined or co-acting stresses during the accelerated testing, this example argues for careful selection of electrolytes and concentrations. It also underlines the significant influence of building design and construction which can result in a coating failing prematurely. Case Study 3. Comparison of Exposure Results at Two Different European Sites. Observation of Corrosion §37 A final example from the field illustrates the influence of location and elevation of the exposure on the corrosion resistance of coil coated systems. The effect is quite an obvious one to discuss but it is important to demonstrate just what sort of differences can be seen even with the same coating system. In this example the coating was applied under commercial coating conditions on hot dipped galvanised steel (0.5 mm gauge), galvanised with 250 g/m2 of zinc. The organic coating comprised a primer and topcoat .The former was an epoxy melamine based primer with strontium chromate as the inhibitive pigment. The latter was a polyvinylidene flouride coating; considered to be the most durable generic type used commonly for coil coatings. Panels of this system were also exposed in the accelerated environments detailed in Table 2. §38 The overall patterns for the edge corrosion of each of these tests is shown in Figures 7a – 7h inclusive. The principal issue is again, not in the details of the corrosion mechanism, but rather the mode of corrosion generated by each exposure and to what extent the short term tests replicate this. §39 From the photographs a number of features are worth commenting on: §40 a)     There are quite distinct differences between the natural exposure panels on the basis of both geographic location and orientation. The exposures on the 90 �, north-facing orientation at Hendaye gave the greatest degree of degradation. However the degradation here is characterised by significant face blistering and, at the cut edge, the build up of voluminous zinc corrosion product overlying as well as undermining the organic coating. In places the coating has embrittled and flaked but there is surprisingly little evidence of steel corrosion /red rust considering the severity of the exposure. §41 b)     At the Hook of Holland site on 90�, north facing exposure the same material shows quite a different corrosion mechanism and seems to follow the more usual edge disbondment, peel and flake off of the organic coating. There is considerably less zinc corrosion product and some bare, presumably passivated zinc exposed. There are traces of red rust at the extreme cut edge. §42 c)      The exposures on the 45�, south facing orientation at Hendaye correspond closely with those at the Hook of Holland at 45�, both showing the beginnings of edge peel after 5 years. §43 d)     The pattern of corrosion of the system exposed at Hendaye on 90�, north facing elevation is best replicated by the scab corrosion test and to a much lesser extent the saltspray. As before the other natural exposures seem better replicated by the laboratory cyclic tests and in particular the Prohesion� and Cyclic salt fog/ UV exposure test. Discussion §44 The results show clearly that the outdoor environment, even in the two locations chosen here, is neither uniform in the mode or, quite probably, the mechanism of corrosion. Equally as interesting is that all the various accelerated tests seem to have relevance depending on location, with the possible exception of the salt fog test. §45 Table 3 shows the principal climatological differences between the two sites [27]. The severe marine site at Hendaye is obviously the more aggressive of the two, evidenced by the corrosion rates of zinc and steel. The reason for this difference is most likely found in the combination of high Cl- levels, higher average annual temperatures and radiation levels. §46 Being sheltered and north facing, the 90� exposed  panels at Hendaye are likely to experience the longest wet times of any natural exposure, especially at the horizontal cut edge. It may be postulated that at this edge, conditions above critical RH will persist and ionic concentrations build most rapidly. Corrosion products will be more formed more rapidly, will be denser (possibly containing more of the hydrated zinc hydroxy chloride (Zn5(OH)8Cl2. 2H2O)), and more strongly isolating. §47 A likely consequence will be that the zinc is less able to function sacrificially, hence anodic depletion will stall, evidenced by the lack of further coating peel back and red rust. §48  Being outdoors and sheltered 90� and facing north, the scab corrosion test conditions are clearly very similar to this, albeit that the greater Cl- content in the spray appears to accelerate the degradation. Saltspray testing takes this a step further still, but fails to produce the same, more open corrosion product morphology. §49 At other angles of exposure on the same site the effects of more rapid, wet/dry cycling, direct sunlight and greater temperature cycling would explain why such exposures are better replicated by the cyclic tests with lower chloride concentrations and periods of UV irradiation. The resultant, thinner, less dense and thus more permeable corrosion products are more likely to contain more hydrated zinc hydroxy carbonate salts. The mode of edge corrosion in this case and replication of it is shown in Figures 8 and 9 respectively. §50 This case demonstrates the dangers of assuming that a more severe marine environment like Hendaye just serves to accelerate the rate of corrosion. Clearly the mode, rate and mechanism can differ profoundly between geographic location and elevation. This underlines the importance of understanding not only the macro and micro climates of chosen test sites, but also the specific climate of any building before offering advice or guaranties of performance. An Integrated Model for Corrosion Performance Prediction. §51 The case studies used here illustrate and reinforce the fact that the nature of corrosion in the environment where pre-coated steels are used is complex and variable. They also serve to emphasise the influence of product and application design and the accelerating effects of atmospheric contaminants and pollutants. Clearly these all contribute towards the overall corrosion performance of the product in a combined rather than isolated fashion. §52 This being the case it is unlikely any single test will be capable of reproducing the corrosion process in its many forms and simultaneously take into account the various design and orientation influences. The demand for a generally applicable corrosion test is therefore, as Funke put it ‘rather like the demand for a medicine, which cures all ills’. [28] §53 This dilemma can be resolved by two possible approaches. The first is to use laboratory proving tests which do not necessarily accurately model atmospheric attack mechanisms but which are consistently reproducible. They must obviously also  produce an adequate acceleration of corrosion. Such an approach is ideal for quality control and ranking evaluations. Considerable work has been undertaken by both the SSPC and CSCT in this area, utilising rank correlation statistics to determine the reliability of an such approach. [29,30,31]. The key to this approach is to preserve the outdoor, ranked performance of known systems in a reliable range of accelerated tests. Having done so, the performance of an untried, new product can then be assessed and ranked with confidence within this matrix. §54 The other approach supported here is based on one suggested for the automotive industry [32]. This entails the integration of information relating to the corrosion process derived from three sources; firstly, that from standard, natural exposure sites, secondly from a broad spectrum of laboratory accelerated performance tests and finally, from measuring system’s response to the fundamental processes involved in corrosion. §55 Implicit in this methodology is the recognition that any particular accelerated or outdoor test will often overstress particular elements of the corrosion process. The advantage of this approach is that it both recognises and exploits these emphasised stresses. By integrating the responses from all three sources into a central model, it then becomes possible to both predict service life in untested environments and to understand and rectify early failures. §56 Information from the field is predominantly a question of careful observation of any failure mode, in particular how it is initiated and how it proceeds. This should be  supported by characterisation of the process using in-situ, scanning techniques such as SRET, potential mapping or Kelvin probe analyses. Such techniques and the information about the real time processes are then utilised to validate the other two approaches in the methodology. A broad range of exposure sites with detailed information on their climatological characteristics helps to put the corrosion modes recognised into the context of the environment of exposure. §57 The accelerated tests chosen, are done so as to provide a means of assessing the sensitivity of materials to one or more of the specific degradation influences outlined below.  Ideally these should be applied at a number of different levels in order to generate a response surface for the coating. §58 a)     Mixed electrolytes , pH and concentration, buffering effects §59 b)     Influence of wetting and drying rates and cycle durations §60 c)     Influence of photodegradation of the organic coating §61 d)     Influence of design –overlaps, drip ends, scratches and bends, composite panels. §62 e)     Influence of orientation and exposure angle (90�, 45�, 5� from the horizontal). §63 f)       Influence of temperature fluctuations (-25�C� + 70�C) §64 g)     Influence of panel moisture/wet time and  RH §65 h)     Influence of substrate thickness §66 Clearly this list is not exhaustive but in an industrial development laboratory some balance between time, resource and depth of study has to be struck. A number of these influences are already covered in existing assessment standards. What is necessary is to utilise a broad range of testing techniques, each one emphasising one or more of the particular parameters listed above.  Like the jigsaw puzzle it is the combination of the responses of the system to each individual influence that enables the formulator to gain the overall picture. Each piece on its own is not only uninformative but has the potential to be misleading. §67 The purpose of the fundamental studies are to link up the coated system’s macro corrosion performance characteristics with its response to electrochemical and physical stresses, which have been identified and characterised on examples in the field. These will be the essential chemical and physical mechanisms lying at the heart of the corrosion processes. The techniques  should be employed to characterise the basic adhesion, barrier and inhibitive properties of the coating under a variety of different conditions. The best tools to use here are likely to be those that evaluate the coating in-situ rather than as a free-film [33]. Electro chemical  impedance spectroscopy (EIS) has been successfully used to measure ionic barrier properties, water uptake rates through the coating and wet adhesion at the substrate interface[34]. The protective action of the coating and the efficiency of the inhibition process at the cut edge has also been be studied with EIS , atomic absorption spectroscopy and scanning reference electrode techniques [35]. Measurements of microhardness and adhesion at temperatures around and exceeding the Tg of the coating will also be of value. §68 These fundamental, mechanistic responses then need to be integrated with the responses of the system to the accelerated stresses/artificial environments of laboratory tests and the information relating to real life failure modes and service conditions.  This is done through the central predictive model as shown in Figure 10. Case Study 4: Development of Chromate Free Primer Systems. Description of the Programme §69 The integrated testing methodology suggested is particularly suited to the determination of service life performance of new, chromate free systems. Typically traditional, accelerated techniques discriminate against such products and this last example is an illustration of how the proposed methodology can be used to avoid falsely condemning promising materials. In this project, fundamental and exterior evaluations are still in progress, but the accelerated testing protocols and the coating’s responses are worth reviewing. §70 All the coated materials in this example were factory applied under standard commercial conditions. Their general compositions are shown in Table 4a. §71 Panels were against a selection of accelerated tests, the specifications of which are shown in Table 4b. Results and Discussion §72 On accelerated testing the major point of degradation, as expected, is the exposed cut edge. Edge creep performance, summarised in terms of degree and mode for the various accelerated methods, is shown in Table 5. The key to interpretation of the results is now not the performance against a particular specified, standard test such as ASTM B117, but rather performance of the material on application of a particular stress or set of conditions. More work needs to be done in the resolving each procedure into its component stresses. However, by way of illustration, some tentative conclusions on edge creep performance of the chromate and chromate-free primers can be deduced. §73 a)     Wet Adhesion and effect of Base Catalysed Hydrolysis – System A shows the better resistance, especially in terms of degree of edge delamination. §74 b)     Wet Adhesion and Influence of High RH under condensing and immersed conditions – Good osmotic resistance and wet adhesion is demonstrated by both System A and B. §75 c)     Influence of cycling conditions – These produce edge peel and corrosion product wedges in Prohesion� and Cyclic Fog/UV Exposure. The cyclic PUCAT test does not seem sufficiently progressed and the cycling conditions of CCT-1 are masked by the high NaCl concentrations. At this stage System A demonstrates more resistance to the stresses built up by the cycling conditions than System B. §76 d)     Influence of increased cathode and anode ratio. At 750 hours the Cyclic PUCAT test did not produce a high degree of anodic depletion and red rust was not yet observed. This suggests that both primers provide good protection for the galvanised substrate and that both exhibit good wet adhesion and inhibitive properties. System A shows some edge peel and System B blistering as a response to this applied stress.  §77 e)     Influence of UV Degradation. Whilst the topcoat and its degradation remain the same for both systems the increased effects of wetting and ionic transport through the film are exerting a slightly more delamination effect on System A than System B. Both show the start of steel corrosion at the cut edge. §78 f)      Influence of electrolytes- System A and System B respond similarly to the NaCl and (NH4)2SO4 electrolyte mix used in Prohesion�, Cyclic PUCAT and Cyclic Fog/UV exposure tests. §79 Using this response data coupled with an understanding of the specific, applied stresses and influences involved, a combined picture of accelerated performance can be built up. With further evaluations and the required integration of responses from the field and fundamental studies it will be possible to confidently construct a comparative model of predictive performance of this chromate free system. Summary and Conclusions §80 The objective of this paper has not been to promote any single, new test to replace salt fog or other existing laboratory based accelerated techniques. It is clear that whilst the adoption of cyclic tests such as ASTM D5894-96 represent a move in the right direction they are unlikely to be able to reproduce all aspects of real time corrosion performance. §81 Rather the purpose has been to support, on the evidence of real time, field exposures, the adoption of a more coherent approach to predicting both early failure and long term service life. This methodology entails the integration of response data from three different, but complimentary approaches, into a central predictive model. No one of them stands on its own as an indicator of performance, contrary to the traditional method of corrosion assessment and prediction. §82 In all but sheltered marine environments, the edge corrosion process for zinc galvanised, organically pre-coated steel appears to follow a common mode. It is characterised by peel back and embrittlement of the organic coating , by relatively thin layers of zinc corrosion product and  by an apparently high rate of anodic reaction and consequently corrosion of the steel to form red rust. The process  is initiated by adhesion loss (electrochemical or mechanical disbondment) of the organic coating and is accelerated by a number of factors not necessarily related to the coating itself. These may include the substrate gauge, the nature of the electrolytes and any contaminant, the rate and duration of wet and dry and temperature cycles. Propagation can be similarly accelerated by external influences such as the orientation and attitude of the coated sheet on the structure (hence wet and dry characteristics and resultant stresses).To date most, common accelerated tests do not reproduce this phenomenon. Acknowledgements §83 The author wishes to thank the management of Akzo Nobel Nippon Paint Ltd, for their permission to use their data and Dr Scantlebury, Dr R Howard and Dr A Darwin of the corrosion protection centre at UMIST for their help and co-operation in certain aspects of corrosion protection assessment. References §84 1.     Metal Cladding and Roofing Mfgs. Association. Statistics.1991 2.       Ryan,P.A., Wolstenholme,R.P, Howell,D.M., ‘The Durability of Cladding, A State of The Art Report’. Thomas Telford, W.S.Atkins 1994. ISBN 07277 20120 3.      Cox,R.N., Kempster,J.A., Bassi,R., ‘A Survey of the Performance of Organic Caoted Metal Roof Sheeting’, BRE Report. Watford 1993. 4.      BS 5427 .’Code of Practice for  Performance and Loading Criteria for Profiled Sheeting in Building’. Clause 19.BSi.1996. 5.      Pr EN10169-2. ‘Continuously Organic Coated Steel Flat products Part 2 – Products for Building Applications’ .Draft Version. BSi, 1995. 6.      Asher, J.M.B., ECCA General Meeting, Conf. Trans. 1998. ‘Future Ctgs, The Changing Face of the Industry’. 7.       Dickie,R.A., Smith,A.G., Chemtech. Vol.10 , 31, 1980 8.      Skerry, B.S., Simpson,C.H., Corrosion, Vol 49, 8 ,1993 9.      SS 11 72 11, ‘Accelerated Outdoor Test By Intermittent Spraying of a salt solution (Scab Test)’. Draft , Swedish Standards Institute, 1992. 10.     Skerry,B.S.,Simpson,C.H.,Ray,C.J., J.Prot.Ctgs.& Linings.,Vol.8, 5, 1991. 11.     Skerry,B.S., Alavi,A., Lindgren,K.I., J. Coat.Tech.,Vol.60, 765, Oct. 1988. 12.    Gardner,G., J. Protect. Ctgs.& Linings, 50-62, Sept.1998 13.    ASTM D 5894-96. ‘Standard Practice for Cyclic Fog/UV Exposure of Painted Metal’, ASTM Annual Book of Standards, Volume 6.02, Paint –Tests for Formulated Products and Applied Coatings. ASTM 1996.  14.    Appleman,B.R., Campbell,P.G., J. Coat.Tech.,Vol.54, 686 ,March 1982. 15.    Shrier,L.L, ‘Corrosion’, Shrier (ed), 2nd Edn, UK Newnes-Butterworths, London. 1976 , 1-138. 16.    Pereira,D., Almeida,E., Figueiredo,O., Prog. In Org.Ctgs., 17, 175 –189, 1989 17.    Stratmann,M., Bohenkamp,K.,Ramchandran,T.,Corrosion Sci., 27, 905, 1987. 18.    Standish,J.V., Ind.Eng.Chem.Prod.Res.Dev., 24, 357-361, 1985 . 19.    Lyon,S.B.,Thompson,G.E.,Johnson,J.B.,Wood,G.C.,Ferguson,J.M., Corrosion NACE, Vol.43, 12, 719-726, Dec. 1987. 20.    Graedel,T.E., Frankenthal,R.P., J.Electrochem.Soc., 137 , 2385, 1990. 21.    Skerry,B.S., Simpson,C.H., Procs. ‘Corrosion ‘91’, Paper 412, NACE Annual Conf. 1991, Mar. 11-15. 22.    ASTM B117-97, ‘Standard Practice for Operating Salt Spray Testing Apparatus’, ASTM Annual Book of Standards, Volume 03.02, Wear and Erosion; Metal Corrosion. ASTM 1998.  23.    BS 3900: Part F2: April 1973, ‘Determination of Resistance to Humidity Under Continuous Conditions’, BSi, London, 1973. 24.    Harrison,J.B., Tickle, T.C.K., J. Oil & Col. Chem. Ass. 45 , 571, 1962 25.    Walter,G.W., Corrosion Sci., Vol 35 , No. 5-8. 1391 – 1404. 1993 26.    Lyon, S.B.,Lyon,J.D., Johnson,J.B., ‘Realism in Cyclic Cabinet Corrosion Testing of Coil Coated Steel for BUILDING Applications’, Cyclic Corrosion Testing, ASTM STP 1238, Haynes,G.S., Tellefsan,K.,(eds),ASTM Philadelphia. 1994. 27.    Almeida,E., Santos,D., ECCA Conf. Trans. ECCA General Meeting May 25-28, 1997, ‘ECCA Own Programme, Its First Few Years’. 28.    Funke, W.,  J. Oil & Col. Chem. Ass. 67, 3, 71 ,1984. 29.    Boocock,S.K., J. Protect. Ctgs.& Linings . 51-58, Oct.1994. 30.    Cleveland Society for Ctgs. Technol. Tech. Com., J. Coat.Tech. 66, 837, 49-67, Oct.1994. 31.    Cleveland Society for Ctgs. Technol. Tech. Com., J. Coat.Tech. 68, 858, 47-61, July 1996. 32.    Dickie,R.A., J. Coat.Tech., 64, 809, 61-63 , June 1992. 33.    Funke,W.,et al, J. Coat.Tech. 58, 741, 79-86, Oct. 1986. 34.    Deflorian,F., Fedrizzi,L.,J.Adh.Sci.Tech., 13, 5, 629-645, 1999 35.    Zin,I.M.,Howard,R.L., Badger,S.J.,Lyon,S.B., ‘The Mode of Action of Chromate Inhibitor in Epoxy Primer on Galvanised Steel’ Prog. in Org. Ctgs, in press. Figures §85 Figure 1 : Section Through  Edge of Pre- Coated Steel Sheet (Not to scale) §86 Figure 2 : Failure modes on Precoated Steel .Source [2] §87 Figure 3 : Drip Edge, West Pitch §88 Figure 4 : Drip Edge, East Pitch §89 Figure 5a : Saltspray 1000 hours §90 Figure 5b : Prohesion� 1000 hours §91 Figure 5c : 3 � Years Scab Corrosion Test §92 Figure 5d : Hook of Holland  5 Years §93 Figure 5e :  Cyclic Fog/UV Exposure Test,   1500  hours total exposure §94 Figure 5 : Accelerated Testing : Edge Corrosion Patterns (x 8 magnification) §95 Figure 6a  Roof South West Pitch, Overlap Edge §96 Figure 6b  Roof North East, Overlap Edge §97 Table 1 :  Characteristics of  Different Roof Pitches Measurement South West Pitch North East Pitch Organic Coating Thickness 36-40 micrometres 37-42 micrometres Galvanised Layer Thickness 18-21 micrometres 18-21 micrometres Metal Gauge 1.05 milimetres 1.05 milimetres Extent of Red Rust (from edge) 30-35 milimetres none Extent of Coating Peelback (from edge) 40-45 milimetres none Cross Hatch Adhesion 100% 100% Pencil Hardness (Faber Castell) F minimum F minimum Coin Scratch Good Good Solvent Resistance(methylethyl ketone) 40-45 double rubs 29-45 double rubs pH at the overlap edge 6-7 6-7 Cl - ppm(w/w)     Drip Edge 40 60 Top of Overlap 880 400 Bottom of Overlap 1200 730 NO3- ppm(w/w)     Drip Edge <10 <15 Top of Overlap <70 <50 Bottom of Overlap <60 <70 SO42- ppm(w/w)     Drip Edge 220 410   6800 4000 Bottom of Overlap 9500 1300 §98 Figure 7a. Saltspray Test - 1000 hours Continuous Fog §99 Figure 7b. Prohesion� Test 1000 hours wet/dry cycling §100 Figure 7c. Cyclic Wet Dry/ UV Exposure Test 1500 hours  §101 Figure 7d. SCAB Corrosion Test 5 years Exposure §102 Figure 7 : Case Study 3 : Accelerated Exposure Tests  (x 0.54) §103 Figure 7e  Hook of Holland, 90� North Facing 5 Years §104 Figure 7f  Hook of Holland, 5� South Facing  5 Years §105 Figure 7g  Hendaye, France, 90� North Facing  5 Years §106 Figure 7h  Hendaye, France, 5� South Facing  5 Years §107 Figure 7 : Case Study 3 : Natural Exposure Tests . (x 0.54) §108 Table 2 :  Case Study 3 : Corrosion Test Details 1)      Saltspray according to ASTM B117 –97 2)      Prohesion� according to ASTM B117 –97 3)      Cyclic Fog / uv Exposure according to ASTM D5894 – 96 4)      Scab Corrosion according to SS 11 72 11 5)      Hook of Holland 90�, North Facing Natural Exposure (ECCA T19.5.2) 6)      Hook of Holland 5�, South Facing Natural Exposure (ECCA T19.5.3) 7)      Hendaye, France.  90�, North Facing Natural Exposure (ECCA T19.5.2) 8)      Hendaye, France 5�, South Facing Natural Exposure (ECCA T19.5.3)   Table 3 : Climatological Data Comparison 1995-1996 (Source Reference 27) Characteristic � Hook of Holland Hendaye Location  51� 59' N  43� 28' N Average Annual Temperature (�C) 10 15 Average Annual Relative Humidity (%) 82 77 Global Radiation (hJcm-2 ) 3820 4612 Average Annual rain mm 58 117 Average annual Wind Speed 5 - Predominant Wind Direction SW W Average annual Deposition Rate     SO2 (mg.m-3) 12 - Cl- (mg.day-1.m-2) - 163 Steel Corrosion Rate         (g.m-2.yr-1) 390-410 420-580 Zinc Corrosion Rate         (g.m-2.yr-1) 7-10 16-25   UV Category RUV 2 RUV 3 Corrosivity Category C3 C4   §109 Figure 8 :  Cut Edge Corrosion Mode (x 8 magnification)    b) Edge peel /Zn Corrosion    c)  Fe Corrosion  Exposure    a) Initiation   d)  Coating detachment §110 Figure 9 :  Reproducing Cut Edge Corrosion (x 8 magnification) Scab Corrosion  Test  3 Yrs Hook of Holland 6 Yrs Scab Corrosion  Test Exposure Prohesion�  Test 1000 hrs Saltspray  Test 1000 hours §111 Figure 10 : Corrosion Prediction Methodology §112 Table 4a  Case Study 4 : Systems Evaluated System A: Chromate-free Polyester primer on HDG: Substrate: HDG, 0.5 mm gauge, Zn coating thickness 17-24 microns  (250gsm) Pre-treatment : Bonderite 1303 /Parcolene 62.(Chromic acid based) Primer : No chromate pigments, 6-8 microns. Polyester/amino binder Topcoat : Brown polyester/amino topcoat of 18-20 microns dry film thickness System B: Chromated Polyester primer on HDG : Substrate : HDG, 0.5 mm gauge , Zn coating thickness 20 microns  (250gsm) Pre-treatment : Bonderite 1303 /Parcolene 62. (Chromic acid based) Primer : Strontium chromate containing, 6-8 microns. Polyester/amino binder Topcoat : Brown polyester/amino topcoat of 18-20 microns dry film thickness Table 4b Case Study 4 : Accelerated Corrosion Test Detail 1) Saltspray according to ASTM B117 –97 2) Prohesion� according to ASTM B117 –97 3) Condensing Humidity according to BS 3900 : Part F2 4) Total Water Immersion according to ASTM D870 –92 5) Cyclic Fog / UV Exposure according to ASTM D5894 – 96 6) Scab Corrosion according to SS 11 72 11, exposed as ECCA T19.5.2 6) Cyclic PUCAT            1 hr immersion in 3500ppm (NH4)2SO4 ,500ppm NaCl  solution.              1 hr drying at ambient             200mm2 coupled SS Cathode coupled to 30mm x 0.5mm Coated Edge 7) CCT – 1 Cyclic Corrosion Cabinet (Automotive Test CCT - A)            Cycle 1 : 4 hours  5% NaCl Fog @ 35� C                Cycle 2 : 2 hours  drying  @ 60� C            Cycle 3 : 30 mins. drying @ 40�C            Cycle 4 : 2 hours condensing humidity @ 50� C, >95%RH §113 Table 5 : Case Study 4  Evaluations – Edge Corrosion Results Test System Edge Corrosion (mm from edge) Coating Peel Back (mm from edge)  Edge Blistering (ISO 4628-2 82) Quantity/Size  Zinc   Corr.  Red Rust Saltspray 1000hrs A 4 - 6 0 2 / 4 Overlying Dense None   B 4 - 9 0 3/4 Overlying Dense None Prohesion 1000 hrs A 1 - 1.5 2/2 Wedge, Dense Very Slight   B 2 - 2.5 2 - 2.5 3/2 Wedge, Dense Very Slight                1000 hrs A 0 None 2/2 None None   B 0 None 2/2 None None Water Soak 1000 hrs A 0 None 0 Very Slight None   B 0 - 0.5 0 - 0.5 0 Very Slight None Cyclic Fog / UV  1320 hrs A 3 - 6 3 – 6 0 Wedge, Dense Some on edge   B 3 - 5 3 – 5 0 Wedge, Dense Some on edge Cyclic PUCAT  750 hrs A 1.5 1 2/2 Slight None   B 1 - 2 None 3/4 Moderate None CCT- 1     937hrs A 6 - 10 None Obscured Very dense  Overlying None   B 9 - 12 None Obscured Very dense  Overlying None Scab Corrosion  6700 hrs A 1-2 None 2/2 Overlying Dense Very Slight   B 1 - 1.5 None 2/2 Overlying Dense Very Slight