3.

4.

5.

applications of a material or component.

In the laboratory, materials and components are usually tested

in configurations far different from in-service configurations,

making correlations between accelerated aging tests and in

service performance uncertain.

Recommendations are seldom made as to how the results of stan

dard tests for different materials should be compared with each

other.

Quantitative estimates of time to in-service failure are seldom made. The reason is that most standard durability tests are comparative tests; that is, the experimental procedure only allows for the comparison of the durability of an unknown material or component against the performance of a reference material or component, both exposed to identical stresses,

6. The degradation mechanisms of materials and components

7.

are complex and not well understood so that it is difficult
to design meaningful accelerated tests.

The factors affecting service life are numerous, as indicated in
2
Table 1, and difficult to quantify. Thus, many existing tests do
not include all factors of importance and factors that are included
are seldom related quantitatively to in-service exposure conditions.

Due to the deficiencies in the current technology of durability testing,

ASTM subcommittee E 6.22 was formed in 1974 to provide a more general and

2

Table 1 is located at the end of this report, on page 29.

development of methodologies, the dissemination of knowledge, and the stimulation of research relating to the prediction of the service life of building components, and the demonstration of compliance with durability performance requirements". The work of the subcommittee recently led to the publication of the Recommended Practice for Developing Short-Term Accelerated Tests for Prediction of the Service Life of Building Components and Materials, ASTM E 632-78, which is based upon National Bureau of Standards (NBS) research. The practice outlines the process of developing tests to predict the service life of a building material or component from the results of short-term tests. In making a prediction of the time to failure, the procedure emphasizes the necessity of knowing as much as is practical about the nature of the item and the service conditions, e.g., material degradation mechanisms and in-service exposure conditions. It leaves open, however, the details as to which tests should be used and and how the information should be analyzed.

-

In durability evaluation, the procedures used to measure degradation and
to analyze the durability data are very important, for durability
research is expensive in terms of money, time, equipment and space, and
seldom yields more than a small amount of data. There are two classes
of procedures for determining in-service life deterministic (e.g.,
fracture mechanics) and probabilistic (e.g., reliability). In reality,
the separation is not distinct, for deterministic procedures often use
statistical methods for analyzing their data and probabilistic procedures
often use a materials justification in selecting a life distribution.

blem is addressed in terms of material parameters.

Unfortunately, few,

if any, have been successful in modeling the durability of a material. One possible reason for this failure is that deterministic procedures usually model a material in terms of one failure mechanism whereas, in fact, failures over time are usually the result of several time-dependent failure mechanisms [5]. Also, failure mechanisms which cause time-dependent failures are difficult to identify, for the failure is thought to be initiated at the microscopic or submicroscopic level. The inability to observe the initiation of failure and to predict the time of failure for a stressed material led to the belief that durability research could be analyzed in terms of random events [6, 7]; hence the use of probabilistic procedures. The basis for this is the observation that two or more specimens, which appear to be identical, can have failure times which are several decades apart, even though they are subjected to the same stress and operating conditions. The analysis of such random events is within the domain of probabilistic procedures, especially those which address the problem in terms of the material parameters.

2. THE RECOMMENDED PRACTICE FOR DEVELOPMENT OF ACCELERATED SHORT-TERM TESTS FOR THE PREDICTION OF SERVICE LIFE

The recommended practice for developing short-term tests for the prediction of service life described in ASTM E 632 is summarized by the chart in Figure 1. The chart indicates a sequence of steps which we recommend be undertaken in developing and applying tests for predicting the service lives of innovative building materials or components; or existing ones which are to be used under conditions outside their normal ranges.

Definition; 2) Pre-Testing; 3) Testing; and 4) Interpretation and Reporting of Data. In Part 1, referring to the numbered boxes in the chart, the first step (step 1) is to define the performance requirements to be met by the material or component in service and to set minimum requirements the material or components must meet to be judged serviceable. These criteria provide an objective basis for recognizing when failure has occurred. It should be noted that the failure criteria for a material

can change with the application.

In step 2, if the material or component

is not homogeneous, it should be characterized as thoroughly as possible

in terms of the individual materials contained within it and the inter

faces between the individual materials. This information is important for gaining insights into the possible degradation mechanisms so that the most appropriate tests can be selected. It should be noted that, because of synergistic effects, composites can have durabilities and properties far different from those of the constituents. The critical performance

characteristics are specified in step 3; these characteristics will be used in delineating the limiting condition below which the material or component is deemed unserviceable. In step 4, the expected range of degradation factors, including weathering, biological, stress, incompatibility and use factors, should be identified to help define the conditions to which the material or component is likely to be exposed in service. Synergistic effects between degradation factors can be identified. With this knowledge it may be postulated (step 6) how the degradation processes can be accelerated. If degradation processes can be accelerated without changing the mode of failure, then laboratory test time can be reduced.

ments for the test specimens should be stated (step 7). It must be recognized that much of the knowledge desired may not always be available. In such cases, assumptions based on the best available experience should be made

and recorded.

When Part 1 is completed, Part 2, Pre-testing, can be initiated (step 8

in Figure 1).3 Its purpose is to demonstrate that rapid failures can be

caused by intensifying the degradation factors specified in Part 1. These preliminary experiments provide the background for Part 3 which begins with the establishment of more realistic accelerated aging tests (step 9). These

accelerated tests should be conducted at different stress levels.

At the

same time (step 10), long-term tests under in-service conditions should be initiated. The results of long-term tests provide the most convincing evidence that the results of accelerated aging tests can be extrapolated to in-service conditions. They are important in insuring that second order effects are not causing the failure at low stress levels. If second order effects appear to be causing the failures, then the accelerated test conditions should be reviewed to determine whether factors which accelerate second

order effects are too severe or whether important degradation factors have

been omitted.

If the results of the accelerated tests and the long-term tests are consistent with each other, Part 4, Interpretation and Reporting of Data, can be This includes use of experimental data to predict the course

undertaken.

of degradation under expected in-service conditions (step 13) and to predict

the time at which failure, as defined by the performance criteria, will

3

The figures are located at the end of this report, beginning on page 30.