Uncertainty test design

Corrosion inhibitor research requires accelerated testing of candidate inhibitors such as REM compounds, before field trial of the most successful candidates. A challenging issue is that an inappropriately designed accelerated inhibitor test could introduce major uncertainties to its results. Acceleration of a corrosion test is usually achieved through the enhancement of the aggressiveness of the test environment in order to intensify major corrosion controlling factors. The identification of major environmental factors that may control the thermodynamics, kinetics and mechanism of a corrosion process is, therefore, the first step in inhibitor test design. 

The introduction of automated scanning systems was a great leap forward in the development. That way, the uncertainties of manual probe guidance were eliminated. Usually, these systems were designed for high-frequency surface tests and followed the outer profile of the surface with a probe that could be moved in several axes. A continuous 100 % scan became possible and, as a result, the documentation of the tests with stripchart recorders suggested itself. Now for the first time, wheel testing became retraceable. 

Although a torsion test is simple to carry out, it is not commonly accepted as an integral part of a material specification furthermore, few torsion data exist in handbooks. If, as is usually the case, the design needs to be based on tensile data, then a criterion of elastic failure has to be invoked, and this introduces some uncertainty in the calculated yield pressure (8). 

Uncertainty about a material s properties, along with a questionable applicability of the simple analysis techniques generally used, provides justification for extensive end use testing of plastic products before approving them in a particular application. As the use of more FEA methods becomes common in plastic design, the ability of FEAs will be simplified in understanding the behavior and the nature of plastics. 

The designer must be aware that as the degree of anisotropy increases, the number of constants or moduli required to describe the material increases with isotropic construction one could use the usual independent constants to describe the mechanical response of materials, namely, Young s modulus and Poisson s ratio (Chapter 2). With no prior experience or available data for a particular product design, uncertainty of material properties along with questionable applicability of the simple analysis techniques generally used require end use testing of molded products before final approval of its performance is determined. 

The results of activation analysis are subject to well known and common analytical sources of uncertainty, as well as method specific uncertainties, e.g. summarized by Greenberg (1997), and also in Section 2.2. In order for INAA experiments to measure differences in induced activity, i.e. differences due to heterogeneity in the amount of analyte in a given test portion, the experimental procedure is designed to allow only the following uncertainties to be part of the result ... 

This is determined by applying a suitable design stress factor (factor of safety) to the maximum stress that the material could be expected to withstand without failure under standard test conditions. The design stress factor allows for any uncertainty in the design methods, the loading, the quality of the materials, and the workmanship. 

Data adequacy The key study was well designed and conducted and documented a lack of effects on heart and lung parameters as well as clinical chemistry. Pharmacokinetic data were also collected. The compound was without adverse effects when tested as a component of metered-dose inhalers on patients with COPD. Animal studies covered acute, subchronic, and chronic exposure durations and addressed systemic toxicity as well as neurotoxicity, reproductive and developmental effects, cardiac sensitization, genotoxicity, and carcinogenicity. The values are supported by a study with rats in which no effects were observed during a 4-h exposure to 81,000 ppm. Adjustment of the 81,000 ppm concentration by an interspecies and intraspecies uncertainty factors of 3 each, for a total of 10, results in essentially the same value (8,100 ppm) as that from the human study. ... 

Simple illustrations of the variety of service conditions are the difference in climates of Moscow, London and Phoenix, or the different chemicals that might come in contact with an engineering component. The implications are that uncertainty may mean designing for the worst case (or being unexpectedly caught out) and the variation potentially calls for testing (or making extrapolations) under many conditions. 

Intakes that exceed the ADI will not necessarily result in any adverse effect because the uncertainty factors are designed to be conservative. In practice it is probable that most people could exceed the ADI by a considerable margin before suffering any harm. Nevertheless, the probability that an individual will suffer harm (risk) increases once the ADI is exceeded and so this must be balanced against the costs of control. Conversely, the level of risk below the ADI is never quite zero because there is always a residual risk that relates to the lack of absolute certainty in the methods used for toxicological testing. In some cases no adverse end-point can be identified, such as for many naturally-occurring compounds that are widespread in foods. In such cases an ADI Not Specified (ADI NS) is allocated. 

The term JT A Hi in equations 8.1 and 8.3 is frequently very small compared to the uncertainty in the determination of A/frcp and in many instances can safely be neglected. This should of course be tested by performing blank experiments under normal operating conditions. For example, the enthalpy associated with breaking an ampule (independently from the contribution from vaporization effects) can be determined by breaking ampules partially filled with the calorimetric solvent in the calorimetric solvent. For many systems this contribution is negligible, provided that a well-designed breaker mechanism and ampules ensure that the dissipation of heat is reduced to a minimum. The importance of vaporization effects can be evaluated as described by Vanderzee [129]. 

The estimation of purely experimental uncertainty is essential for testing the adequacy of a model. The material in Chapter 3 and especially in Figure 3.1 suggests one of the important principles of experimental design the purely experimental uncertainty can be obtained only by setting all of the controlled factors at fixed levels and replicating the experiment. 

If we assume the model y, = 0 + r, , the data and uncertainties are as shown in Figure 8.1. We can test this model for lack of fit because there is replication in the experimental design which allows an estimate of the purely experimental uncertainty (with two degrees of freedom). 

However, even against this background of uncertainty and apparent imprecision, in some cases it is possible to measure airborne asbestos with acceptable precision through replication for a reasonable price. Since there is no exposure standard, "clean" must be defined by comparing indoor and outdoor levels. A statistical comparison of indoor versus outdoor measurements that is significantly different from zero indicates that the indoor space is not clean. Tests may be designed chat either compare the average of... 

---