Evaluation of testing methods for fuels using signal/noise ratio as a statistical criterion

✍ Scribed by T. Fujimori; K. Ishikawa

• Publisher: Elsevier Science
• Year: 1972
• Tongue: English
• Weight: 529 KB
• Volume: 51
• Category: Article
• ISSN: 0016-2361
• DOI: 10.1016/0016-2361(72)90090-7

✦ Synopsis

The authors deal with problems of the evaluation or selection of testing methods for fuels. Signal/noise ratio, suggested by Taguchi', is shown to be a useful statistical criterion for the above purposes. This criterion is composed of two parts: one, the difference between different samples, and the other, precision of the testing methods. It can, therefore, be regarded as an index of the detection capability of the testing method. For fuel-testing methods it is an especially useful criterion because the tests used are frequently 'model tests', for the results of which there are no true theoretical values: e.g. proximate analysis. In such a case, neither accuracy nor precision can be an effective criterion, because the result depends upon details of the procedure. Four examples are examined to illustrate the usefulness and application of the signal/noise ratio for fueltesting methods.