Argentum Solutions, Inc.
Sterling guidance on corrosion and materials degradation
| Potential-pH Diagrams |
 |
|
| Intelligent Tools |
 |
|
 |
|
 |
|
| Corrosion Calculator |
 |
|
| Corrosion Economics Estimator |
 |
|
|
David C. Silverman |
||||||||||||||||
|
Features Useful for Interpreting Results Two characteristics identified as being important in the early development of this technology were mass change with time during both the soaking and drying portions of the test. These characteristics were expected to provide information about the diffusion of components into and out of the non-metallic material (A. O. Fisher and C. N. Carpenter, Elastomeric Linings, in Process Industries Corrosion, B. J. (Moniz and W. I. Pollock, ed.), p. 589, NACE, Houston, TX). During the early development of the test, the belief was maintained that curve-fitting of this type of information to the diffusion equation shown above would provide information on diffusion coefficients. This simple concept, though, is only correct if one component is diffusing. In complex environments and with complex nonmetallic materials the mass change with time is a complex function of the rates of components from the environment migrating into the nonmetallic and components originally in the nonmetallic material migrating into the environment. Later development suggested that more information would be required for a proper first-pass screening. SEQEXPERT uses the parameters shown in the table below to make its prediction from the Sequential Immersion Test. While these parameters may not be the only ones that can be used, they are the features that provided consistency between observation and prediction during training of the artificial neural network. SEQEXPERT requires that these features be used together. This requirement must be kept in mind when reading the additional background information on the meanings of the individual variables.
The following two figures 1. Mass fraction change during soak and dry cycles The mass (weight) fraction change during soaking is the final mass (weight) before the drying stage minus the initial mass (weight) divided by the initial mass (weight). This value is a measure of two quantities, the uptake by the nonmetallic sample of components in the environment and the loss of material leached from the non-metallic sample to the environment. This number could be negative if more mass leaves the material than enters it. The mass (weight) fraction change after drying is the final mass (weight) after drying minus the initial mass (weight) before soaking divided by the initial mass (weight). Assuming that the components that migrate into the nonmetallic material will be removed with the drying phase, this mass fraction is a measure of the total amount of material that was leached from the nonmetallic sample during the soaking cycle. This quantity can be negative if material was leached during the soak cycle. Such an observation should always be considered as a warning flag of questionable behavior.In theory, if the tested sample forces one-dimensional diffusion (in Cartesian coordinates) the diffusion coefficient and expected mass change at infinite time could be estimated for the soak portion of the test using the figures above. But, such a solution is only valid if one component is diffusing into the sample. As discussed in the section Sorption and Diffusion Influence on Sample Size and Test Duration from a practical standpoint, the test should be designed so that the final mass change provides a good estimate with a desired error of the mass change expected when migration is no longer occurring or when migration into the specimen equals migration out of the specimen. No good method exists to predict an estimate for test duration for the drying portion. Unfortunately, no good-rule-of-thumb exists to provide guidance as to what maximum overall change in mass during soaking would be acceptable for all situations. Some people have used a maximum of 10% gain during soaking for a lining and 15% (or even 20%) for a gasket as rules-of-thumb but those numbers should not be treated as fact. The final use dictates acceptability. If the final application is a lining, such properties as tensile strength and adhesion to the metallic or plastic superstructure need to be considered. If the final application is a gasket, such properties as creep relaxation, compressibility, and sealing ability need to be considered. The final mass change after drying is a function of the amount of the material that originally migrated into the nonmetallic material during the soak cycle that migrated out and by the amount of material (e.g. plasticizer, filler, etc.) originally in the nonmetallic material that migrated out during the soak cycle. The optimum result would be a final mass change of zero. Such a case would be consistent with no loss of material during soaking and elimination of all external material during drying. But, a cautionary note is in order. Though rare, such an outcome is also consistent with another scenario, any residual material that migrated into the specimen virtually offsetting any loss of original material during soaking. A positive mass change very likely has two causes, either material that migrated into the specimen during the soak portion became somehow bound or that material has such low volatility that it cannot migrate out during the drying portion. Either way, it is retained material that might affect performance. A negative mass change usually means that material originally present in the non-metallic material has migrated out during the soak portion of the test. Large negative changes can be accompanied by mechanical changes as reflected in a large change in hardness. The ambiguities mentioned above are reasons that mass change cannot be considered as the sole variable for chemical compatibility. 2. Final curvature of mass change profiles The final curvature of the mass fraction curves during the soaking and drying portions of the test are qualitative indicators of proximity of the final measured change in mass to that change in mass expected if equilibrium or some type of steady state had been achieved. These variables are included to account for the fact that the elapsed experimental time may not be long enough for the first two variables to accurately reflect the final change in mass at steady state or equilibrium. For example, if the curve corresponding to the soak cycle is continuing to increase at a reasonable rate when that portion of the test is terminated, the mass fraction used for the first variable could be a serious underestimate of the true value. If the curve is leveling out, then the value used could be a reasonable estimate. The interpretation of the curvature has to be somewhat qualitative. The proper method would be to examine at least the first and second derivatives of the mass change versus time curve (velocity and acceleration). But, a rigorous mathematical treatment of these curves is impossible because (1) the mass change is semi-quantitative at best, (2) a limited number of data points are usually obtained because of the labor-intensive nature of the intermediate weighing, and (3) for each derivative there is a loss of one place of accuracy, one significant digit. The decision of whether or not the curve is leveling becomes easier the more curves that are examined. The figures above demonstrate two extremes.3. Sign of mass change for soak and dry cycles These variables were discovered to be important to SEQEXPERT being able to "learn" the "patterns" to make predictions from the Sequential Immersion Test. Though the mass fractions are written with sign, the assumption is that the sign itself is important as an independent indicator of performance. For example, a negative sign on the mass change after drying could suggest a loss of initial material such as plasticizer. This result tends not to be a favorable indicator of long-term performance. The result might suggest a loss of flexibility of the non-metallic material resulting in cracking or loss of adhesion. A positive value at the end of the drying cycle may mean that some absorption was irreversible again suggesting that mechanical properties might be affected.The six variables outlined up to this point reflect the chemical interaction between the environment and the non-metallic material. Mechanical changes are also important. One can have very small chemical effects and a very large mechanical effect and vice versa. As this test is to provide a first pass screening mostly for rubbers and possibly thermoplastics, the concept is to use a simple mechanical test that can provide some insight into general mechanical degradation. For this reason, the change in hardness was included in the test protocol. Rubbers tend to be tested on the Durometer hardness scale (ASTM D2240) and rigid plastics on the Barcol hardness scale (ASTM D2538). A large increase in hardness may be caused by a loss of plasticizer or irreversible reaction of environmental components with the polymer creating stiffening. Such an increase may be accompanied by a negative overall change in mass. A large decrease in hardness may be caused by a partial dissolution of the polymer chains into environment components that have migrated into the material. Such a decrease may be accompanied by a large increase in absorption during the soak portion with positive change in overall mass. The procedure is to obtain the hardness of the non-metallic material using the appropriate scale for the material being tested prior initial immersion in the fluid. Usually, about ten readings are taken. The hardness is again measured at the conclusion of the drying portion of the test. Again, about ten readings are taken. One rule-of-thumb is that a change of about six units is significant enough to raise a warning flag. But, since this measure of mechanical properties is very approximate, any change must be interpreted in conjunction with the change in mass. No one variable stands alone. Several examples of how these parameters are derived and the ability of SEQEXPERT to use them are in D.C. Silverman, "Artificial Neural Network Predictions of Degradation of Nonmetallic Lining Materials from Laboratory Tests", Corrosion, Vol. 50, p. 411, 19941  (514k).
and D. C. Silverman, "Corrosion Prediction from Laboratory
Tests Using Artificial Neural Networks", Paper #048, presented at the 12th
International Corrosion Congress, Houston, TX, September, 1993.
Previous Page: Sorption and Diffusion Effect on Sample Size and Test Duration Next Page: SEQEXPERT (the intelligent prediction tool) 1 © NACE International publication and year shown in citation above. All rights reserved. Displayed with permission from NACE International, Houston, TX (http://www.nace.org). Published in Corrosion, in the month and year shown in the citation above. | ||||||||||||||||
|
David C. Silverman, Ph.D. - Primary Consultant |
){kind=link}
){kind=link}