International NACE PaperLaboratory Test Methods For Offshore Coatings - A Review Of A Round Robin Study

2011

Laboratory Test Methods for Offshore Coatings - A Review of a Round Robin Study

Michael Winter

International Paint LLC., Protective Coatings 6001 Antoine Drive Houston, TX 77091

USA

ABSTRACT

In 2006 a Round Robin study was conducted on a number of laboratory test methods listed in NACE TM0104-2004, TM0204-2004, TM0304-2004 and TM0404-2004 in order to establish the reproducibility of the test methods. The results from this study indicated that significant differences in results could be achieved between different laboratories testing the same coating systems.

This paper discusses the results from this study, including statistical analysis of the data generated, gives guidance on how laboratory test results should be interpreted and provides some suggestions for future test method development.

Key Words: Offshore Coatings, Laboratory Test Methods, Round Robin Study

©2011 by NACE International. Requests for permission to publish this manuscript in any form, in part or in whole, must be in writing to NACEInternational, Publications Division, 1440 South Creek Drive, Houston, Texas 77084. The material presented and the views expressed in this paper aresolely those of the author(s) and are not necessarily endorsed by the Association.

1

Paper No.

11041

INTRODUCTION

In 2006 a Round Robin Study was carried out to evaluate the reproducibility of various NACE standard laboratory test methods for Offshore Coatings. The test methods studied were taken from NACE Standards TM0104,1 TM0204,2 TM03043 and TM 0404.4

These test methods are referenced in a NACE Standard Practice for the Corrosion Control of Offshore Structures by Protective Coatings SP0108-2008. The main purpose of the Round Robin study was to aid committee members in setting Recommended Acceptance Criteria for five particular test methods (Rust Creepage Resistance, Water Immersion, Cathodic Disbondment, Dimensional Stability and Hot/Wet cycling). The concern among committee members on these test methods was due to a lack of history of carrying out the tests in different laboratories. For the other test methods referenced in the Recommended Practice (Edge Retention, Thermal Cycling, Impact Strength, Aging Stability and Thick Film Cracking) there was agreement among Committee members on the repeatability of the test method and on the Recommended Acceptance Criteria.

ROUND ROBIN METHODOLOGY

To conduct the Round Robin study, 17 different paint systems were all applied by KTA Tator in Pittsburgh. The test panels were then sent to the seven laboratories that agreed to participate in the study and tested according to whether the systems were designed for atmospheric or ballast tank exposure (see Table 1)

Table 1

Paint Systems Tested in Study

Category Coating System Supplier 1 Supplier 2 Supplier 3 Atmospheric OZ/E/PU A11 A12 A13 Atmospheric IOZ/E/PU A21 A22 A23 Atmospheric E/E/PU1 A31 A32 A33 Atmospheric E/E/PU2 A41 A42 A43 Ballast tank E/E/E1 B11 B12 B13 Ballast tank E/E/E2 B21 B22 Ballast tank E/E/E3 B31

OZ = Organic zinc, E = epoxy, PU = polyurethane, IOZ = inorganic zinc

ANALYSIS OF DATA

In order to determine the consistency of the data generated in the Round Robin study, the results have been analyzed using a Method Repeatability and Reproducibility (Method R&R) approach. R&R study software was employed to conduct the analysis. 5 The analysis is conducted using each individual data point developed during the testing – where 4 panels are measured, each measurement is included – the summary data tables included in this paper only contain the average of data generated by each laboratory. This analysis enables us to evaluate the overall variation in results and look at the variation that is attributable to differences within a single laboratory (“repeatability”) and the variability that is attributable to differences between laboratories (“reproducibility”). The analysis ends up assigning values to Repeatability, Reproducibility and Overall Repeatability and Reproducibility.

An effective measurement system has an R&R of < 10%, while an R&R value of >30% indicates that the measurement system needs improvement. An R&R value between 10% and 30% may be satisfactory, but depends on the specific circumstances of the method.

2

RESULTS

Rust Creep Evaluation

A 3.5 inch vertical scribe was made on each specimen (4 replicates). The specimens were then placed into exposure per ASTM D5894.6 Synthetic seawater was used as the electrolyte. Flow rate of the prohesion cabinet was measured per ASTM G85,7 and the test was run for 12 weeks (6 cycles). After exposure, loose material was removed along the scribe using various techniques among the participants. Twelve perpendicular measurements were made along the scribe and average creep, r, was calculated.

Results for rust creep are shown in Table 2. As expected, the zinc systems (A11-A13 & A21-23) out performed the non-zinc systems, with the exception of A12. Considerable differences in scribe creep were obtained by different laboratories on the same system. The Method R&R analysis gave a Repeatability of 27.8%, a Reproducibility of 39.6% and an overall R&R of 48.4%, which is poor. Contributing factors to this low level of repeatability and reproducibility are as follows:

Inaccurate measurements: The measurements range from 0-5 mm, with many of them being < 1mm. These are small values, and a few millimeters of variation results in a high percentage variation, i.e., 2 mm is 100% greater than 1 mm. In cases where creep was a few tenths of a millimeter, which is nearly impossible to measure accurately, 0.2 mm is 100% greater than 0.1 mm. It might be preferable to employ the rating scale per ASTM D1654 8 in which numerical ratings (0-10) are provided for various measurement ranges. Table 3 contains the data after converting to D1654 numerical rating. It would be interesting to perform the R & R analysis on this data set.

Method of removal of loose material around the scribe: Several different methods were used, including picking with a knife, power washing, and scraping with a spatula. The different degrees of force used to remove the loose material may result in varying measurements of the scribe creep.

Variations in corrosivity of cabinets: Variations in the cabinets among laboratories could also have lead to the high R & R.

Table 2 Average Scribe Creep in mm

System Lab 1 Lab 2 Lab 3 Lab 4 Lab 5 Lab 6 Lab 7 A11 0.1 0.2 1.5 0.2 0.0 0.6 1.1 A12 1.5 3.3 2.6 3.2 2.2 4.6 4.4 A13 0.3 0.6 1.3 0.2 0.2 0.5 1.0 A21 0.1 0.7 0.6 0.3 0.0 1.2 0.9 A22 0.1 0.4 1.0 0.5 0.1 1.3 2.8 A23 0.0 0.4 0.3 0.7 0.0 0.4 0.8 A31 2.6 2.7 2.5 4.4 2.3 4.6 3.9 A32 0.5 2.1 1.8 2.5 0.6 3.0 2.2 A33 0.3 1.6 1.4 2.4 1.3 2.3 2.1 A41 1.7 2.7 1.0 2.5 2.0 3.1 2.7 A42 0.3 1.7 1.5 1.9 1.0 2.3 2.0 A43 2.9 4.0 3.6 4.0 3.7 6.9 4.7

3

Table 3 Average Scribe Creep per ASTM D1645

System Lab 1 Lab 2 Lab 3 Lab 4 Lab 5 Lab 6 Lab 7 A11 10 10 7 10 10 9 8 A12 7 5 6 5 6 5 5 A13 10 9 8 10 10 10 8 A21 10 9 9 10 10 8 9 A22 10 10 10 10 10 8 6 A23 10 10 10 9 10 10 9 A31 6 6 6 5 6 5 5 A32 10 6 7 6 10 6 6 A33 10 7 7 6 7 6 6 A41 7 6 7 6 6 5 6 A42 10 7 7 7 7 6 7 A43 6 5 5 5 5 4 5

Hot/Wet Cycling Evaluation

All sides of the steel panels including the edges were coated with the specific coating system. After 1 week of curing a 90 mm long and 2 mm wide scribe was introduced down to the bare steel in the middle of the panel of one side. The panels were then exposed to cyclic salt fog (ASTM G 85-A5), 3 hours wet condition at room temperature followed by 3 hours dry condition at 60 ºC (140 º F). After 336 cycles all loose coating around the scribe was removed and the rust creep calculated based on 12 readings along the scribe.

Three laboratories have done a comparison test of 5 different systems, see Table 4.

Table 4 Rust Creep from a 2 mm Wide Scribe on 5 Different Coating Systems.

The 5 Coating Systems are Evaluated by 3 Different Laboratories

System Lab 1 Lab 2 Lab 3 B11 2.1 3.2 4.9 B12 1.4 2.2 3.4 B13 5.3 7.8 8.1 B21 2.1 3.2 3.7 B22 1.6 2.6 3.1

4

Figure 1: Rust creep values taken from Table 4 Figure 1 shows that the rust creep values reported from the three laboratories are significantly

different. For the single coating systems the rust creep values are increasing when moving from laboratory 1 to laboratory 2 and again from laboratory 2 to laboratory 3. The Method R&R analysis gave a Repeatability value of 50%, a Reproducibility value of 43% and an Overall R&R of 66%.

An explanation for the poor R&R might be found in the limitations of temperature control in the salt spray chamber. As indicated in Figure 2 below, the cooling curve shows a moderate decrease, however, room temperature (20ºC) is never reached after the high temperature set point, 60ºC. The lowest temperature reached in this test is about 40ºC and it indicates that the cooling capacity has been too low. A salt spray chamber running with a large number of panels requires much more cooling than a chamber with a less panels because of differences in the total heat capacity of the panels. Therefore if a laboratory conducted the test using a full chamber it may not expose the panels to the same thermal cycling gradients as a laboratory which tested in a virtually empty chamber. A way to improve the repeatability would be to apply accelerated cooling after the high temperature cycle to reach room temperature very fast.

5

Figure 2: Room temperature (20ºC) is never reached after the high temperature cycle (60ºC).

Cathodic Disbondment

Six laboratories participated in the cathodic disbondment test study using 5 ballast water tank coating systems according to the NACE TM 0104 Section 7. In this test method, ASTM G 8 9 was modified in the electrolyte (synthetic seawater (ASTM D 1141)10), impressed voltage (-1.0 to -1.1 V D.C. (SCE)), holiday diameter (3.18 mm / 0.125 in.), number of holidays (two: one holiday for each side of the test panel) and test duration (12 weeks).

The results of this test are summarized in Table 5. Participating laboratories reported a good deal of data scattering for each panel. In addition, order of cathodic disbondment value for five coating systems showed very little consistency between laboratories.

Table 5 Cathodic Disbondment Results in the Round Robin Test.

Panel # Coating Systems

Lab-a Lab-b Lab-c Lab-d Lab-e Lab-f Av.,[mm] Av.,[mm] Av.,[mm] Av.,[mm] Av.,[mm] Av.,[mm]

B11 E/E/E1 3.7 1.8 3.3 3.0 4.5 1.5 B12 E/E/E1 4.5 2.4 0.5 1.6 0.4 4.0 B13 E/E/E1 13.0* 11.2 17.0 14.4 12.8* 18.1 B21 E/E/E2 2.8 1.9 1.0 7.9 6.4 5.5 B22 E/E/E2 3.6* 0.6 0.0 0.0 0.0 0.0

*Blistering was observed.

A Method R&R analysis of the results gives a Repeatability of 18.6%, a Reproducibility of 14% and an overall R&R of 23.3%.

Due to the data variation that was seen, the Round Robin study was rerun, with some additional specifications added to the test method:

6

Rerun Round Robin Test

The following conditions and equipments were specified in order to improve the repeatability and reliability of the test result;

1) The holiday shall be drilled with a flat head drill (end mill).

2) Use impressed current only.

3) Use of an air sparger not to exceed the saturated oxygen content at the test temperature.

4) The synthetic seawater electrolyte shall be replaced every four weeks.

5) Potential will keep -1.00 V.

6) Counter electrode (anode) can be Titanium, Platinum Copper or Graphite rod.

7) The anode shall be isolated from the electrolyte to prevent the contamination of the test specimens. A fritted glass tube to hold the anode as described in the test standard ASTM G 95.11

8) Record temperature where the test is conducted.

9) Use a pocket knife to remove the loose paint.

The rerun study was conducted by four laboratories using the revised test method.

The results of the rerun round robin test for cathodic disbondment test are summarized in Table 6. There was still some data scattering for each panel between the participant laboratories.

Table 6 Cathodic Disbondment Results in the Rerun Round Robin Test.

Panel # Coating Systems

Lab-a Lab-b Lab-c Lab-e Av.,[mm] Av.,[mm] Av.,[mm] Av.,[mm]

B11 E/E/E1 1.0 2.5 3.0 0.0 B12 E/E/E1 0.9 3.1 4.0 5.9 B13 E/E/E1 25.0 18.0* 18.0 23.6* B21 E/E/E2 4.8 2.6 5.0 6.1 B31 E/E/E2 2.2 0.2 3.0 0.0

*Blistering was observed.

The Method R&R data for CD rerun test gave a Repeatability of 15.6%, a Reproducibility of 9% and an overall R&R of 18%

The rerun results showed a slight improvement in both Repeatability and Reproducibility, although the rerun study was based on only four laboratories.

Immersion Evaluation

The sea water immersion test is conducted by immersing test panels in synthetic sea water at 40 ± 2 ºC for 12 weeks. The Test Method offers two alternatives for evaluation – a pull off adhesion test or a wet disbondment test. The pull off adhesion test is conducted using ASTM D4541 12 within 8 hours of removing the test panels from sea water. The wet disbondment test is conducted by making a 0.125 in holiday using a flat head drill in the center of each side of the test panels prior to immersion.

7

After immersion, loose coating around the holiday is removed using a pocket knife and the resulting radius of disbondment is evaluated using 4 diagonal measurements. In both test methods, any blistering is also noted.

In the round robin study, only the wet disbondment method was evaluated by all laboratories. The results from 5 coating systems tested in 4 different laboratories are recorded in Table 7.

Table 7 Average Disbondment in mm for Immersion Test

System Lab 1 Lab 2 Lab 3 Lab 4 B11 0.6 2.1 6.6 4.25 B12 1.5 1.8 5.9 2.35 B13 8.8 6.7* 9.2* 6.38* B21 0.0 2.5 7.1 2.0 B22 0.0 2.2* 5.4* 2.75*

* = Panels blistered in test

A Method R&R analysis on this data gave a Repeatability of 39.5%, a Reproducibility of 64.3 and an overall R&R of 75.5%, indicating very poor correlation of results both within and between laboratories.

Due to the variation in results that was seen, this test was rerun with 6 participating laboratories and 5 coating systems. The results are recorded below in table 8.

Table 8 Average Disbondment in mm for Rerun Immersion Study

System Lab 1 Lab 2 Lab 3 Lab 4 Lab 5 Lab 6 B11 2.1 6 2.9 0.4 11.0 0.0 B12 1.6 6 3.1 0.1 10.2 2.0 B13 18 14 16 20.3 25.8 11.2 B22 3.6 8 4.3 0.0 11.5 2.5 B31 2.5 5 1.4 0.2 12.0 0.0

A Method R&R analysis for the rerun study gave a Repeatability to 23.7%, a Reproducibility of 56.7% and an overall R&R of 61.4%. While the results were slightly better for Repeatability, Reproducibility and overall R&R than the previous study, there is still a very high degree of variation and poor correlation of results.

The most likely reasons for the differences that were seen between laboratories are variability in the test conditions itself and variability in the disbondment measurements.

It was noted during discussions that most, but not all, laboratories routinely conduct water immersions tests using air sparging to keep up the dissolved oxygen level. The method and degree of sparging varies between labs and this item is not addressed in the test method.

The disbondment evaluation measurement will be affected by the type and sharpness of the pocket knife being used and the force that is applied by the operator to the knife. It is perhaps not surprising to see significant differences between laboratories in this case.

8

Dimensional Stability Evaluation

The dimensional stability test is conducted by applying the coating system to be tested to polyethylene panels. The coating film is removed from the panel and allowed to cure for one week before testing. The film is measured for length, width, thickness and mass prior to immersion in sea water at 40ºC for 12 weeks. After removal from the sea water, the samples are dried with a towel and then immediately measured for length, width and mass. The test is conducted in triplicate and the average is taken.

In the round robin test, five systems were tested in six different laboratories. Many labs reported difficulties handling the samples – free films of epoxy coatings are notoriously brittle – and also some labs had difficulty conducting the length measurements. As a result, for the analysis of the test data we have used the results from four laboratories for the length measurements and from 5 laboratories for the weight measurements.

Tabulated below are the average results for length and mass change.

Table 9 Length Change in Dimensional Stability Test

System Lab 1 - % Change

Lab 2 - % Change

Lab 3 - % Change

Lab 4 - % Change

B11 -0.82 -0.79 0.00 0.00 B12 -2.56 -2.11 -2.60 -1.33 B13 -1.24 -1.68 -1.73 0.00 B21 -0.63 -0.74 -1.30 0.00 B22 -1.29 -1.06 -0.87 0.00

Table 10

Mass Change in Dimensional Stability Test

System Lab 1 - % Change

Lab 2 - % Change

Lab 3 - % Change

Lab 4 - % Change

Lab 5 - % Change

B11 -0.35 -0.72 -0.22 0.75 -0.67 B12 -4.6 -4.99 -4.39 -2.60 -3.73 B13 -1.56 -3.23 -0.33 0.70 -3.27 B21 4.14 -0.30 -0.08 0.84 -0.41 B22 6.5 -0.78 0.87 1.88 -3.33

The averages of the three replicate samples masks significant variation seen within a laboratory between the replicates. When the individual sample values are subjected to a method R&R analysis, the following results are obtained:

Table 11 R&R Data for Dimensional Stability Test

Method % Repeatability % Reproducibility Overall R&R Length change 56.5% 42.6% 70.7% Weight change 32.7% 47.0% 57.2%

The data analysis shows an unacceptably large variation in both repeatability (data variation within a laboratory) and in reproducibility (data variation between laboratories).

9

Further Statistical Study of Data

Due to the generally rather poor Method R&R results for the tests being studied, it was postulated that the test results may show more correlation if the analysis was done based on relative rankings of coatings in a particular test between laboratories. While Method R&R data is looking at the Repeatability and Reproducibility of an actual number generated by a test, a ranking analysis will establish if a test will consistently identify the best and worst coatings, irrespective of the actual value achieved.

Although the Round Robin study was not set up to evaluate coatings using the ranking approach, an attempt has been made using Fleiss Kappa statistics and Kendalls Coefficient of Concordance to determine if indeed that test methods can show consistent ranking of coating systems.

The ranking data for each test and the Kendalls Coefficient of Concordance are shown in Table 12. Note that = signs are used where coating systems received exactly the same rust creepage value from a laboratory and there exhibited equivalent performance.

Table 12 Rust Creepage Test Relative Rankings

Coating System/Ranking

Lab A11 A12 A13 A21 A22 A23 A31 A32 A33 A41 A42 A43 A =2 9 =5 =2 =2 1 11 8 =5 10 =5 12 B 1 11 4 5 =2 =2 =9 8 6 =9 7 12 C =7 11 5 2 =3 1 10 9 6 =3 =7 12 D 1= 10 1= 3 4 5 12 =8 7 =8 6 11 E =1 10 5 =1 4 =1 11 6 8 9 7 12 F 3 =10 2 4 5 1 =10 8 =6 9 =6 12 G 4 11 3 2 9 1 10 7 6 8 5 12

Kendalls coefficient: 0.853, P = 0.00

These values indicate a positive correlation between rankings made by different laboratories. The kappa values that were obtained indicated that the laboratories can distinguish fairly well between the very good and very bad performing coatings but are not consistent in differentiating between coatings of similar performance. (Note: the practice of ranking a number of products equally may be having a detrimental effect).

Table 13 Hot/Wet Test Relative Rankings

Coating System/Rankings

Lab B11 B12 B13 B21 B22 A 3 1 5 4 2 B =3 1 5 =3 2 C 4 2 5 3 1

Kendalls coefficient: 0.92, P = 0.026

There is a limited data set here, but the rankings of the very best versus the very worst are quite consistent.

10

Table 14 Cathodic Disbondment Test Relative Rankings

Coating System/Rankings

Lab B11 B12 B13 B21 B22 A 3 2 5 4 1 B 3 2 5 4 1 C 4 2 5 3 1 D 2 4 5 3 1 E 3 4 5 1 2 F 2 3 5 4 1

Kendalls coefficient: 0.744, P = 0.0013

There is complete agreement on the worst performing system and good agreement on the best performing system, but laboratories struggle to consistently agree on ranking of the intermediate performing products.

Table 15 Water Immersion Test Relative Rankings

Coating System/Rankings

Lab B11 B12 B13 B21 B22 A 4 2 5 1 3 B 3 4 5 =1 =1 C 2 1 5 4 3 D 3 2 5 4 1 E 1 2 4 5 3

Kendalls coefficient: 0.458, P = 0.057

The results of this test indicate borderline statistical significance. There was reasonable agreement on the worst coating, but no agreement on the other rankings.

Table 16 Dimensional Stability (Linear) Test Relative Rankings

Coating System/Rankings

Lab B11 B12 B13 B21 B22 A 2 5 3 1 4 B =1 5 =1 =1 =1 C 2 5 4 1 3 D 1 5 4 3 2 E 2 4 5 1 3

Kendalls coefficient: 0.738, P = 0.005

11

Table 17 Dimensional Stability (Weight) Relative Rankings

Lab B11 B12 B13 B21 B22 A 2 5 3 1 4 B 1 4 2 3 5 C 2 5 4 1 3 D 2 5 3 1 4 E 2 5 4 1 3 F 2 5 1 3 4

Kendalls coefficient: 0.722, P = 0.002

The dimensional stability rankings indicate some ability to differentiate the very best from the very worst in this test, but laboratories struggle to consistently rank systems of similar performance.

CONCLUSIONS

The NACE standards try to address a range of properties that are required for coating performance, as opposed to those standards/methods that have concentrated on only one coating property, and in this regard they are a welcome step forward. In addressing some of these properties, new test methods have been introduced. As these tests have been used on more occasions and in more laboratories, it has become apparent that the repeatability and reproducibility for certain methods is not as good as it should be – this has been especially highlighted by the round robin study.

There are multiple reasons for this variation in results, including, but not limited to, different interpretations of the test methods, different options in the methods which yield different results, test methods not being exactly followed and inherent variability in the methods themselves (especially those involving a lot of human operations). The high (>30%) Repeatability numbers generated in Hot/Wet cycling, Immersion and Dimensional stability tests means that even within the same laboratory the test results aren’t consistent and the variability cannot be blamed on different laboratories interpretation or test methodology. In the case of the dimensional stability test the difficulty in conducting the measurements combined with the variability observed makes the test virtually worthless.

When these test methods are taken and used as comparative studies in a single laboratory, the relevance and usefulness of the data generated is much greater than when data generated in different laboratories is compared. The problem is not necessarily the methods themselves, but the fact that they are referenced in NACE SP0108-2008 together with a set of recommended acceptance criteria (with the exception of dimensional stability, where no criteria could be agreed). Since different labs can generate significantly different test values when conducting ostensibly the same test on the same coating system, Coating Suppliers will inevitably continue to test the same system in different labs until a study results in the “right” figure that can be quoted to the customer. This does not do anyone any good– the supplier keeps testing the same paint to get the right test data and the customer ends up with the same coating anyway.

This situation is not unique to these particular test methods – many coating test methods suffer from the same issue. For example, ASTM D5894-05 6 (from which the rust creepage test is derived) states:

“this practice is best used to compare the relative performance of materials tested at the same time in the same exposure device. Because of possible variability between the same type of exposure devices, it is not recommended to compare the amount of degradation in materials exposed for the same duration at separate times, or in separate devices running the same condition. This practice should not be used to establish a “pass/fail” approval of materials after a specific period of exposure

12

unless performance comparisons are made relative to a control material exposed simultaneously, or the variability in the test is rigorously quantified so that statistically significant pass/fail judgements can be made.”

The R&R analysis has shown conclusively that the actual test value generated in the test methods discussed in this paper is of little relevance. The Ranking analysis has shown that certain methods do have the capability to reproducibly rank a set of coating systems in order of performance (although, it must be said, that this has not been correlated to actual field performance), which is the way that coating companies use these tests when developing new coatings.

RECOMMENDATIONS

Some specific recommendations on the use of these test methods are as follows:

Rust Creepage test: This test had a marginal rating on repeatability of the test result in a single lab and poor reproducibility between labs. It did have a fairly good ranking correlation, so the test is capable of distinguishing between the best and the worst systems on a reproducible basis. The test is, therefore, useful for the comparison of coating performance, but would require improvements in order for a Recommended Acceptance Criteria value to be assigned.

Hot/Wet Cycling test: This test had very poor repeatability and reproducibility results (interestingly, intralaboratory repeatability is worse than interlaboratory reproducibility, suggesting certain test variables may be out of control). The ranking results suggest that it may differentiate between the very best and very worst performing systems, but based on a very limited data set. The test certainly cannot be used to set Acceptance Criteria – significant improvements in the method would be required to achieve this.

Cathodic disbondment test: This test gave the best repeatability and reproducibility ratings of any of the test methods that were studied. The testing laboratories consistently agreed on the worst performing paint and had good agreement on the best performing paints, but struggled to consistently rank the intermediate performing materials. The rerun test which was conducted using a more highly specified test method gave some improvement in the R&R values. There is still a quite wide spread in the individual values obtained for the same paint by the different testing laboratories. Overall, the test differentiates well between the best and worst performers, but struggles to produce consistent values or reproducibly rank similarly performing paints.

Water immersion test: This test was also subject to a rerun evaluation. The initial test gave extremely poor repeatability and reproducibility ratings and the rerun study was slightly better. The ranking correlation indicates borderline statistical significance, with the only reasonable agreement being on the worst coating. At best, this test is capable is screening out poor performing coatings but cannot rank in any meaningful way coatings of intermediate or good performance. Any acceptance criteria would have to be set very at a high disbondment value so that it just screens out a poor performing coating.

Dimensional stability test: This test gave very poor R&R values and also a number of laboratories reported significant difficulties in running this test. The ranking results were somewhat better, but still the recommendation is that this test method is scrapped from the standards.

13

ACKNOWLEDGMENT

The author would like to acknowledge the help and assistance given by Claus Weinell, Dave Allerton and Tomohiro Tanabe during the preparation of this paper.

REFERENCES 1 NACE StandardTM0104-2004 “Offshore Platform Ballast Water Tank Coating System Evaluation”

2 NACE Standard TM0204-2004 “Exterior Protective Coatings for seawater Immersion Service”

3 NACE Standard TM0304-2004 “Offshore Platform Atmospheric and Splash Zone Maintenance Coating System Evaluation”

4 NACE Standard TM0404-2004 “Offshore Platform Atmospheric and Splash Zone New Construction Coating System Evaluation.”

5 AIAG Measurement Systems, Analysis Reference Manual, 2nd Edition

6 ASTM D5894-05 “Standard Practice for Cyclic Salt Fog/UV Exposure of Painted Metal, (Alternating Exposures in a Fog/Dry Cabinet and a UV/Condensation Cabinet).”

7 ASTM G85-09 “Standard Practice for Modified Salt Spray (Fog) Testing

8 ASTM D1654-08 “Standard Test Method for the Evaluation of Painted or Coated Specimens Subjected to Corrosive Environments”

9 ASTM G8-96 “Standard Test Methods for Cathodic Disbonding of Pipeline Coatings”

10 ASTM D1141-98 (2008) “Standard Practice for the Preparation of Substitute Ocean Water”

11 ASTM G95-07 “Standard Test Method for Cathodic Disbondment Test of Pipeline Coatings (Attached Cell Method)”

12 ASTM D4541-09 “Standard Test Method for Pull-Off Strength of Coatings using Portable Adhesion Testers”

14

www.international-pc.com protectivecoatings@akzonobel.com

and and all product names mentioned in this publication are trademarks of, or licensed to, AkzoNobel. © AKZONOBEL 2010.

International Protective Coatings has used its best endeavours to ensure that the information contained in this publication is correct at the time of printing. Please contact your local International Protective Coatings representative if you have any questions.

Unless otherwise agreed by us in writing, any contact to purchase products referred to in this brochure and any advice which we give in connection with the supply of products are subject to our standard conditions of sale and the provisions of the relevant product data sheet. A

N02

801_

1412

09G

4_O

TC_C

ON

04/1

1