Statistical Interpretation of Semi‐Continuous Anaerobic Digestion Experiments on the Laboratory Scale

Abstract

The importance of thorough experimental design, implementation, and interpretation of laboratory experiments is known but rarely considered in continuous anaerobic digestion experiments. Especially in the evaluation of small effects on gas yields, e.g., by substrate disintegration, the usual methods such as the use of single digesters or the mere comparison of mean values often reach their limits. A proper statistical interpretation will be demonstrated on four examples of semi‐continuous laboratory‐scale anaerobic digestion experiments that investigated the effects of substrate disintegration on biogas yields. Power analysis was performed for one exemplary experimental setup. In this particular case, the power analysis showed that, with digester duplicates per treatment, a minimum 6 % difference in biogas yield could be significantly detected.

Full-Scale Assessment of Ultrasonic Sewage Sludge Pretreatment Using a Novel Double-Tube Reactor

“…To ensure compliance with the t -test requirements (i.e., normal distribution of data and homogeneity of variance), data sets were previously analyzed using the Shapiro-Wilk test and the F-test, respectively. The additional requirement of independent measurement data (control vs test digester) was met through the selected experimental setup …”

“…Statistical Analysis. According to the decision diagram for statistical tests in anaerobic digestion experiments, 46 differences between the control and the test digester were assessed using a Student's t-test, with p values <0.05 indicating significant differences and p values <0.01 indicating highly significant differences. To ensure compliance with the ttest requirements (i.e., normal distribution of data and homogeneity of variance), data sets were previously analyzed using the Shapiro-Wilk test and the F-test, respectively.…”

“…The additional requirement of independent measurement data (control vs test digester) was met through the selected experimental setup. 46…”

The Influence of Pressure-Swing Conditioning Pre-Treatment of Cattle Manure on Methane Production

“…All phases were analyzed separately with ANOVA and post hoc test by means of SPSS statistics Version 20 (IBM, Chicago, IL, USA). Further details about the selection of statistical tests are described by Hofmann et al (2016) [18]. In order to meet the requirements of variance analysis with regard to their applicability, the experimental data were checked for normal distribution using the preliminary significance tests (Kolmogrov-Smirnov and Shapiro-Wilk).…”

“…For the statistical evaluation of the semi-continuous test period, the distribution of the measured methane production is illustrated with the help of boxplots. This allows presumptions on the result of every single phase and on the fulfillment of the assumptions for statistical tests according to Hofmann et al (2016) [18]. Figure 2 shows the boxplots of the biogas yields for the pre-phase and three following hydraulic retention times.…”