Toggle Main Menu Toggle Search

Open Access padlockePrints

Designing experiments for maximum information from cyclic oxidation tests and their statistical analysis using half Normal plots

Lookup NU author(s): Dr Shirley ColemanORCiD


Full text for this publication is not currently held within this repository. Alternative links are provided below where available.


Cyclic oxidation testing at elevated temperatures requires careful experimental design and the adoption of standard procedures to ensure reliable data. This is a major aim of the 'COTEST' research programme. Further, as such tests are both time consuming and costly, in terms of human effort, to take measurements over a large number of cycles, it is important to gain maximum information from a minimum number of tests (trials). This search for standardisation of cyclic oxidation conditions leads to a series of tests to determine the relative effects of cyclic parameters on the oxidation process. Following a review of the available literature, databases and the experience of partners to the COTEST project, the most influential parameters, upper dwell temperature (oxidation temperature) and time (hot time), lower dwell time (cold time) and environment, were investigated in partners' laboratories. It was decided to test upper dwell temperature at 3 levels, at and equidistant from a reference temperature; to test upper dwell time at a reference, a higher and a lower time; to test lower dwell time at a reference and a higher time and wet and dry environments. Thus an experiment, consisting of nine trials, was designed according to statistical criteria. The results of the trial were analysed statistically, to test the main linear and quadratic effects of upper dwell temperature and hot time and the main effects of lower dwell time (cold time) and environment. The nine trials are a quarter fraction of the 36 possible combinations of parameter levels that could have been studied. The results have been analysed by half Normal plots as there are only 2 degrees of freedom for the experimental error variance, which is rather low for a standard analysis of variance. Half Normal plots give a visual indication of which factors are statistically significant. In this experiment each trial has 3 replications, and the data are analysed in terms of mean mass change, oxidation kinetics (up to the onset of spall), the onset of spalling, spallation rate and also the variation between the 3 replications. The method is illustrated by cyclic oxidation of alloy 800 undertaken as part of the COTEST project. © 2006 WILEY-VCH Verlag GmbH & Co. KGaA.

Publication metadata

Author(s): Coleman SY, Nicholls JR

Publication type: Article

Publication status: Published

Journal: Materials and Corrosion

Year: 2006

Volume: 57

Issue: 1

Pages: 22-30

Print publication date: 01/01/2006

ISSN (print): 0947-5117

ISSN (electronic): 1521-4176

Publisher: Wiley - VCH Verlag GmbH & Co. KGaA


DOI: 10.1002/maco.200503891


Altmetrics provided by Altmetric