Ferric Chloride Weld Procedure Testing of Steel Weldments

TWI

Published in: UK Corrosion and Eurocorr '94. Proceedings, International Conference, Bournemouth, UK, 31 Oct.-3 Nov.1994. Publ: London SW1Y 5DB, UK; The Institute of Materials; 1994. Vol.3. pp.51-60.

Abstract

Key variables in the ASTM G48 ferric chloride pitting corrosion test^[1] have been studied, with emphasis on ways of making the test more reproducible for weld procedure qualification of duplex stainless steels. The make up of the test solution, the specimen size and preparation method, and the pass/fail criterion were considered. The results of the experimental programme were used to develop a recommended test procedure, which includes addition of Na₂EDTA.2H₂O to the test solution, specimens with test face dimensions of 25x50mm and sides ground to a 1200 grit finish, and a failure criterion of ≥20mg weight loss or visual identification of pitting.

1. Introduction

Over the past decade, ferritic/austenitic duplex stainless steels have been proposed increasingly as candidate materials for a range of applications, notably offshore. This is a reflection of their attractive general/pitting corrosion and stress corrosion cracking resistance, and high strength to weight ratio. However, the performance of duplex stainless steels in corrosive service is often controlled by weldment properties and consequently, emphasis has been placed on assessment of corrosion resistance during weld procedure qualification (WPQ). The most popular means of assessing the corrosion resistance of duplex weldments for WPQ purposes has been by immersion of a sample in ferric chloride solution, generally following the ASTM G48 method A, which is designed to address the pitting corrosion resistance of stainless steels in chloride environments.^[1] Whilst this test is not intended to be representative of any particular service environment, it is rapid, and the aggressive 6wt% aqueous solution of FeCl₃ has been considered similar to the environment which may develop as a result of corrosion within an occluded pit.^[2] The outcome of the test is simply determined by whether the specimen shows pitting after a specified exposure period at a constant temperature. By repeated exposure at progressively higher temperatures, a critical pitting temperature (CPT) may be determined, and used as a measure of relative pitting resistance. The CPT has been used successfully as a means of ranking materials in order of resistance to pitting attack. Use of this method as a quality control test, for WPQ purposes, is achieved by choosing a required minimum CPT and testing individual weldments at that temperature. The occurrence of pitting can thus be used as a pass/fail criterion. Such a test is very convenient for WPQ purposes as it is relatively quick and simple to perform.

However, the ASTM G48 Method A provides only outline details for testing. There is no widely accepted test technique for high alloy materials^[3] and, in particular, weldments, which has led to difficulty in specifying appropriate test temperatures and repeated difficulty in passing the test, even when all other indications (eg proof strength, phase balance, microexamination, hardness etc) suggest that the weld is of a suitable quality for service.

To address these difficulties and to develop a standard test technique suitable for weldments in duplex stainless steels, a project was initiated with the sponsorship of a group of companies including parent duplex stainless steel and welding consumable manufacturers, fabricators, contractors, oil/gas companies and a regulatory body.

2. Experimental programme

A questionnaire regarding current WPQ practices was sent to over 70 companies with a potential interest in the use of duplex stainless steels. The information obtained from this survey was used to identify the key variables which must be specified in a standardised reproducible ferric chloride test method. These were as follows: (i) the make up of the test solution, (ii) the preparation of the test piece sides and the test face, and (iii) the pass/fail criterion.

A number of other important variables were identified, including: (iv) the test piece dimensions, (v) the method of test piece cleaning after exposure and (vi) the test duration.

The first three variables were investigated by carrying out a series of ferric chloride immersion tests, on (i) 10mm thick 22%Cr duplex stainless steel plate meeting UNS S31803, (ii) a test weld in the same material, and (iii) similar thickness plate in 25%Cr superduplex stainless steel meeting UNS S32750 (Table 1). Welding was by the mechanised gas tungsten arc (GTA) process throughout, using a consumable wire meeting AWS A5.8 type ER 2209 (Table 1). Appropriate values for the second three variables were proposed as a result of Sponsor Group discussion, to reflect the results of the experimental programme and current industrial practice.

Table 1 Specifications covering the materials used

Tests were performed on coupons cut from the appropriate plate or weldment. The coupon width was 25mm (parallel to the length of the weld where appropriate), the thickness was equal to the original plate thickness and lengths of 25, 50 and 80mm were employed. The test face and sides were prepared to the appropriate finish and the coupons were then degreased and weighed prior to exposure. Each specimen was placed at an angle of approximately 45° in a beaker of the test solution (100g of FeCl₃.6H₂0 in 900ml of water) at the chosen starting temperature (15°C and 20°C were employed) and left for 23 hours. The coupons were then removed from the solution, cleaned (either by scrubbing with a nylon bristle brush or in an ultrasonic bath), dried, re-weighed and examined visually, under a low power binocular microscope, for signs of pitting. Probing of the specimen surface, to open up pits, was not allowed, except in tests where two independent operators were used, one of whom was allowed to use a sharpened steel tool. After examination, each specimen was re-immersed in a beaker of fresh solution, at a temperature 2.5°C higher than the previous one. This procedure was repeated for each specimen until it had lost a significant amount of weight (usually >100mg) and showed visible pitting. CPT values were obtained, for each specimen, by both visual identification of pits and weight loss measurements. A summary of the test results is given in Table 2.

Table 2 Summary of test results

Firstly, the make up of the test solution has been the subject of some debate in recent years due to the variation of solution pH and precipitation from solution which occurs at temperatures above approximately 45°C. One way of stabilising the solution is to add a complexing agent. The International Institute of Welding is pursuing the approach of making an addition of 20g/l Na₂EDTA.2H₂O to the solution and has shown that this prevents precipitation up to the boiling point.^[4] The use of this addition was therefore investigated. The need to employ reagent grade chemicals and water of controlled quality (eg ASTM D1193-91 Type IV^[5]) was considered essential to solution reproducibility.

Secondly, the effect of specimen surface finish was investigated by preparing a series of systematically varying test pieces with the test face (ie the weld root) (i) ground flat, (ii) brushed and pickled (in 20%HNO₃ + 10% HF in water), or (iii) degreased but otherwise as-welded. The test piece sides were (i) rough ground on a 60 grit rotating abrasive belt, (ii) ground to a 600 grit finish, (iii) ground to a 1200 grit finish, (iv) manually polished to a 1µm finish, or (v) electropolished.

Thirdly, the pass/fail criterion was assessed by monitoring the weight of the specimen and by carrying out a visual inspection of the specimen surfaces after each test exposure. In some tests, two independent operators performed separate inspections.

3. Results and discussion

3.1 Pass/fail criterion

In order to assess the test data in a meaningful way, it was appropriate to consider first the most suitable pass/fail criterion. Figure 1 shows a typical plot of measured weight loss against the temperature of the ferric chloride solution, for a specimen given multiple exposures. Pitting corrosion is associated with rapid increase in weight loss with increasing temperature, eg as seen on the right hand side of the curve in Fig.1, whilst small weight losses such as shown on the left hand side of Fig.1 may be a result of oxide or inclusion dissolution, after which the material surface passivates and becomes stable to further attack. Occasionally a small weight gain was measured (typically <1mg) at the lower exposure temperatures.

The simplest means of assessing a ferric chloride test is by visual examination, with positive visual evidence of pitting constituting failure. This may be performed with confidence for parent material samples with a high degree of surface finish, as even very small pits (associated with only a few milligrams weight loss) can be reliably identified optically, eg with the aid of a low power binocular microscope. However, as pitting progresses, it does so predominantly by subsurface corrosion, often creating a large cavity with only a very small opening to the surface (eg Fig.2). The observation of such features on the surface of a weld which is irregular and, perhaps, in the as-welded condition, has adherent oxides, is much more difficult than for parent material. This difficulty is compounded by the fact that pitting tends to initiate in any geometric notches or crevices, such as are formed by the weld toe. One way of increasing detection rates for pits on a weld surface is to allow probing with a sharpened pin although this may cause surface damage which can be mistaken for pitting.

In the experience of the Sponsor Group, as-welded samples frequently show no visual sign of pitting even for substantial weight loss which could not be a result of oxide/inclusion dissolution alone. Consequently, a weight loss pass/fail criterion was considered particularly desirable for welds, thus removing the subjectivity involved in visual examination. However, difficulty exists in defining a weight loss criterion which will be widely accepted by all interested parties. A balance must be achieved between setting the level (i) high enough to be confident that pitting has initiated, (ii) low enough so that specimens with significant pitting do not pass the test and (iii) not so high that the temperature for the onset of pitting is greatly exceeded before the specified weight loss is achieved. Clearly, no 'correct' figure exists.

A study was made of the correlation between measured weight loss when pitting was first observed visually and the location of pitting. Figure 3 summarises the results of this work, and illustrates a distinct difference between identification of pits developing on flat, prepared surfaces and those developing on a weld root or cap with the as-deposited profile intact. Two thirds of the specimens which showed the first visual sign of pitting on a flat surface did so for a weight loss of ≤50mg and one third did so for weight loss of ≤ 20mg. For attack on surfaces with the weld profile intact, pitting was typically identified only when the weight loss was ≥ 90mg and independent operators often made different visual assessments of individual specimens. In some instances, one operator reported pits two temperature steps before the other. The lowest weight loss associated with visible pitting was 4mg (on a flat side face) and the highest stable weight loss (ie which was recorded after a particular exposure but did not increase during the next exposure at higher temperature) was 12mg. It was concluded that reliable visual observation of small pits on a weld surface was particularly difficult but that pitting had actually developed on a sample by the time the weight loss reached some 20mg. Therefore, 20mg was chosen as a suitable pass/fail weight loss criterion for welds of the type studied. However, it should be noted that this is regarded as an upper limit, ie pit initiation occurs below 20mg weight loss, not at 20mg weight loss. It was also agreed by the Sponsor Group that visual examination should be employed, with identification of pits before 20mg weight loss constituting failure, but that this is unlikely for many weld types. However, it was recognised that for pickled welds of good profile, visual identification of pitting might be made at less than 20mg weight loss.

A number of specific weld types were identified as potentially being unsuitable for the use of this criterion, namely, (i) welds made with a root pass having higher pitting resistance than the cap or filler, so that pitting on the specimen sides, ie outside of the weld root, should not constitute a test failure, (ii) welds in cast material, where dissolution of inclusions may give rise to substantial weight loss without necessarily initiating pitting, and (iii) heavily oxidised welds, where loss of oxide might give weight loss above 20mg. Whilst these special cases were noted, the need for a single pass/fail criterion was recognised. It was felt that item (i) could be addressed by using visual assessment alone, if the customer agreed that this was appropriate, and that items (ii) and (iii), should be covered by the 20mg criterion, as a considerable weight loss in the ferric chloride test must be a cause for concern, because it may represent an event which could give rise to pit initiation even if it does not positively indicate that pitting has occurred.

3.2 Make-up of test solution

The use of Na₂EDTA.2H₂O in the test solution was assessed by comparison of tests in solutions with and without the addition. Table 3 lists the CPTs obtained from these tests. The data show that for pitting at lower temperatures (ie for 22%Cr duplex stainless steel), the two solutions gave very similar results, whereas at higher temperatures (ie for 25%Cr superduplex stainless steel) the solution containing Na₂EDTA.2H₂O gave lower CPT values, ie the solution was apparently more aggressive when the addition was made. The addition was effective in preventing precipitation above 50°C, which is consistent with the more corrosive nature of the modified solution at higher temperatures, and in agreement with previous findings.^[4] It was concluded that the addition should be recommended to prevent precipitation and enhance the reproducibility of the test.

Table 3 Comparison of tests with and without Na₂EDTA.2H₂O

3.3 Test piece preparation

The results from test pieces with various surface finishes are summarised in Tables 4 and 5. Firstly, for the test face, comparison of the various finishes showed CPTs in the order brushed and pickled > as-welded > ground flat. The difference between the highest and lowest measured CPTs was on average around 5°C, indicating a distinct bias of the test results based on weldment surface finish. It was decided by the Sponsor Group that the test face surface finish must be specified in any reproducible procedure and that the most appropriate condition would be 'as-prepared for service', except that degreasing would be allowed.

Table 4 Effect of test face preparation on test outcome

The preparation of the test piece sides was also found to play an important part in the behaviour during testing. Pitting on the sides of the specimen is an undesirable test outcome, as (i) this material is not exposed in service, (ii) such attack may provide electrochemical protection of the test face, and (iii) it prevents assessment of test face pitting on the basis of weight loss. As the smoothness of the surface finish on the sides was increased, it was generally observed that the incidence of pitting on the specimen sides decreased. Therefore, roughly ground sides often pitted in preference to or at the same temperature as the test face, whereas a 1µm manual polish or electropolish gave greater resistance to pitting, often similar to or better than that of the test face. However, the maximum corrosion current capability of typical electropolishing equipment requires preparation of several individual 'patches' on the specimen sides. Where these patches overlapped, pitting often initiated, suggesting 'defects' in the surface finish and that this approach is only suitable for very small samples.

Table 5 Effect of specimen side preparation on test outcome   

None of the finishes examined gave immunity from pitting on the sides, so a compromise was sought between minimising pitting and having an acceptable specimen preparation time. A 1200 grit finish was chosen as most appropriate. It was noted that pitting on the specimen sides was a potential outcome of any test. The need to specify which specimen faces were under test (eg root, cap) was also recognised, so that attack of any non-test face (usually the sides and perhaps also the cap) would invalidate the test and a re-test should be performed.

3.4 Other factors

Tests were performed on specimens with different dimensions, to investigate the effect of the weld area:parent material area ratio. No significant variation in test outcome was noted for different specimen sizes. It was agreed by the Sponsor Group that 25x50mm was an appropriate test face dimension but that the only relevant limitation on specimen size was that it should be such that all other testing requirements could be met. In particular, the need for weighing to an accuracy of better than ± 1mg, and the 200g maximum weight limitation of many laboratory balances, suggest that 200g would be a suitable specimen upper limit in most instances. However, if a higher capacity balance with suitable accuracy is available, this limit will increase accordingly.

Comparison of scrubbing and ultrasonic cleaning after exposure indicated that scrubbing helped to open pits in some cases but little difference in weight loss was noted for the two techniques. Scrubbing was chosen as most appropriate, by the Sponsor Group, as it was the simplest of the two cleaning techniques.

The test duration has also been a subject of debate, as extended exposure might allow for more pronounced pit formation, thus simplifying visual pit identification.^[3] However, data supplied by one of the project sponsors indicated that no distinct advantage was noted in extending the duration to 72 hours for tests on duplex stainless steels. Therefore, the 24 hour exposure was adopted as most appropriate for WPQ purposes.

3.5 Recommended test procedure

The test procedure derived from the experimental work was checked by means of a Round Robin exercise and incorporated into a document which describes a recommended procedure for carrying out ferric chloride tests on weldments in duplex stainless steels, with a bias towards usage for WPQ purposes.^[6] The Sponsor Group decided to make this document available to all interested parties and copies are now available from TWI. It is hoped that this procedure will help to standardise test methods used, allow more confident seting of required CPT levels and improve the ability of the test to differentiate between acceptable and unacceptable weldments.

4. Summary

A Group Sponsored Project was performed to investigate ways of standardising the ferric chloride test and thereby improve its reproducibility for weld procedure qualification purposes. A questionnaire was used to identify key areas of variability in the test, which were examined in a series of ferric chloride test and the results were used to write a document describing a recommended test procedure, which is now freely available. The following are the primary features of the recommended procedure.

1. The test specimen should have sides prepared to a 1200 grit finish and the test face should be 'as-prepared for service'.
2. The test solution should be made with reagent grade chemicals, controlled purity water and an addition of Na₂EDTA.2H₂O.
3. The test should be assessed primarily on the basis of weight loss, with ≥20mg constituting failure. The Sponsor Group agreed that a visual examination was appropriate and that visual identification of pits should also represent failure.

5. Acknowledgements

The members of the Sponsor Group, namely Avesta-Sandvik Tube, British Gas plc, Chevron UK Ltd, Creusot-Loire Industrie, Det Norsk Veritas, Lincoln Smitweld BV, Press Offshore Ltd, Redpath Engineering Services Ltd, AB Sandvik Steel, Shell UK Exploration and Production, Soudometal SA, Statoil, Stena Offshore Ltd and Weir Materials Ltd, are thanked for financial support and technical contributions to the project.

MTS Ltd and OIS plc, Oilfield Division, are gratefully acknowledged for participating in the Round Robin.

W Martin, P E Kendall, M J Bennett, M R Howe and C S Hardy are thanked for performing the experimental work, and R N Gunn and Dr T G Gooch are acknowledged for their technical assistance.

6. References

1. ASTM G48-92: "Standard test methods for pitting and crevice corrosion resistance of stainless steels and related alloys by the use of ferric chloride solution".
2. Schofield M J and Kane R D: "Critical review of corrosion test methods for duplex stainless steels", Corrosion '90, NACE, 1990, Paper 572.
3. Corbett R A: "Problems in utilising ASTM G48 to evaluate high alloy stainless steels", Corrosion '92, NACE, 1992, Paper 298.
4. Lefebvre J and Potty A: "Experience with CPT determination by means of ASTM G48A: Attempt to improve it", IIW Doc II-C-898-92.
5. ASTM D 1193-91: "Standard specification for reagent water".
6. TWI Document 5632/19/93: "Recommended practice for pitting corrosion testing of duplex stainless steel weldments by the use of ferric chloride solution", June 1993.