Thirty-one laboratories measured biochemical methane potential (BMP) of six substrates in a project aimed at improving BMP measurement quality. Laboratories used their established measurement procedures, with limited standardization. As in earlier studies, reproducibility was quantified, but here laboratories came from a wide geographic range, and the accuracy of BMP measurements was assessed by comparison with 6 experienced "reference" laboratories. Measurement reproducibility between the remaining "evaluation" laboratories was poor in the first test; relative reproducibility standard deviation was 21-33% for four substrates. Values improved to 9-19% in the second test following review of a detailed analysis of submitted data, troubleshooting meetings, and method modifications by some laboratories. Reference laboratories performed better, with reproducibility standard deviation of 5-12%. Repeatability within laboratories was also better for the reference group, with relative standard deviation of 2-4% vs. 3-7% for evaluation laboratories. Evaluation laboratory results showed a persistent negative bias of 5-16% compared to reference laboratories. Application of cellulose validation criteria improved reproducibility but not accuracy. Most of the extreme BMP values were associated with abnormal methane production curves, and a subjective visual evaluation proved valuable as a screening tool. The cause of measurement error was generally unknown, although data processing and volatile solids measurement errors caused some extreme results, and changes in inoculum source improved performance in two cases. There were no consistent differences among the diverse measurement methods and the sources of error probably vary among laboratories. Consistently accurate BMP measurement is possible for laboratories that follow established guidelines and develop experience.
Hafner et al. (Sun,) studied this question.