Vol.5 No.4 2013

Research paper : Evaluating Uncertainty for the Standardization of Single Cell/Stack Power Generation Performance Tests for SOFC (A. Momma et al.)−254−Synthesiology - English edition Vol.5 No.4 (2013) for “measurement uncertainty.” However, the document fails to draw a clear distinction between the variation of measurements and the uncertainty of measurements; for example, it defines “steady state” as the state in which the variation falls within a given range of measurement uncertainty. The document defines “measurement uncertainty” using a general equation that addresses the correlation between input parameters but does not provide any specific method for calculating uncertainty. (3) The TS on “Single Cell Testing Methods for PEFC,” which was published in 2010,[10] was still a Committee Draft (CD) when we were preparing the draft standard. The CD defined the maximum allowable variation of input quantities and the maximum uncertainty of testing equipment. The final version of the TS recommends that measurement uncertainty be included in the test report, as it is shown in the template provided in one of the Annexes. But again, the document points to the GUM as a guide for specific procedures.[7] Based on the circumstances above, we established the following guiding principles on uncertainty evaluation in the proposed international standard for SOFC performance tests: (a) Use the GUM as the guide for uncertainty evaluation and describe both the evaluation equation and the specific procedures for calculating uncertainty. Uncertainty is obtained by combining the results of a Type A evaluation using statistical methods and a Type B evaluation using any other methods. We propose this approach because it would be difficult to derive the actual steps for evaluating uncertainty from the general uncertainty equation alone; leaving this task to the reviewer would result in a complete lack of consistency in the actual method used. (b) Define the maximum variation of input quantities and the allowable uncertainty of measuring instruments, and perform all measurements in a stable state where input quantities stay within the allowable range. This has the effect of setting approximate allowable values for both Type A and Type B uncertainty evaluations. Moreover, conducting measurements in a stable state eliminates the need to consider any correlation between input quantities when uncertainty is evaluated. (c) Do not impose an unreasonable amount of work on the reviewer. The purpose of the standard is to establish performance test methods for commercial purposes, not to obtain an accurate value of uncertainty. We believe that the goal of uncertainty evaluation would be achieved if the evaluation can show the level of accuracy at which a test operator has conducted the test and produced the results for submission; the test operator should not be expected to deviate substantially from the daily test procedures just to achieve this goal. Based on these principles, we removed those aspects of the general equation of uncertainty evaluation that are believed to have little impact on uncertainty. We also provided a description of the specific procedures involved in uncertainty evaluation, so that the test operator would be able to easily conduct the evaluation. 3.2 Description of uncertainty evaluation in the draft standardWith the above circumstances taken into account, we included the following details concerning uncertainty evaluation in the draft standard: The performance tests incorporated in the draft standard include the rated power test, I-V characteristic test, the effective fuel utilization dependency test, the long-term durability test, the internal impedance test and so on. Because uncertainty evaluations of some of these tests were likely to be very difficult or troublesome, our proposed draft standard only mandated that the uncertainty evaluation be conducted on the results of the rated power tests. Here, the rated power test was defined as the test in which either the current or the voltage is measured while the other is kept at a certain level: it is a single-point test in which all other control parameters are kept constant. An explanation is given below using an example where the voltage is measured while the current is kept constant—the more commonly used of the two methods. Based on the guiding principles above, the draft standard established the maximum allowable variations of input quantities such as the current and the gas flow rate that were applicable to all the tests stipulated in the standard (Table 1). In an actual measurement, each test operator is to set a maximum allowable variation within this defined range, so that he/she would be able to obtain the target uncertainty value of a particular measurand (a quantity subject to measurement). This in effect establishes the allowable range of uncertainty arising from random error. In addition, “stable state” was defined as the state in which the system is stable enough for any input quantity to fall within the tolerance range set by the test operator and the measurement result to meet the target uncertainty level. All measurements were to occur after the test sample was confirmed to have attained the stable state. This would prevent the measurement from being conducted while the system is in a transient response arising from a sudden change in a condition or in various drift states. Consequently, measurement variations would be limited to those due to random noise or the uncontrollable minute variation of


page 30