Chemistry Dept., Universidade Nova de Lisboa, Caparica, Portugal.
Water Sci Technol. 2011;63(7):1381-8. doi: 10.2166/wst.2011.195.
The application of membrane bioreactors (MBR) for wastewater treatment is growing worldwide due to their compactness and high effluent quality. However, membrane fouling, mostly associated to biological products, can reduce MBR performance. Therefore, it is important to monitor MBRs as close to real-time as possible to accelerate control actions for maximal biological and membrane performance. 2D-fluorescence spectroscopy is a promising on-line tool to simultaneously monitor wastewater treatment efficiency and the formation of potential biological fouling agents. In this study, 2D-fluorescence data obtained from the wastewater and the permeate of a MBR was successfully modelled using projection to latent structures (PLS) to monitor variations in the influent and effluent total chemical oxygen demand (COD). Analysis of the results also indicated that humic acids and proteins highly contributed to the measured COD in both streams. Nevertheless, this approach was not valid for other performance parameters of the MBR system (such as influent and effluent ammonia and phosphorus), which is usually characterised through a high number of analytical and operating parameters. Principal component analysis (PCA) was thus used to find possible correlations between these parameters, in an attempt to reduce the analytical effort required for full MBR characterisation and to reduce the time frame necessary to obtain monitoring results. The 3 first principal components, capturing 57% of the variance, indicated and confirmed expected relationships between the assessed parameters. However, this approach alone could not provide robust enough correlations to enable the elimination of parameters for process description (PCA loadings ≤ 0.5). Nevertheless, it is possible that the information captured by 2D-fluorescence spectroscopy could replace some of the analytical and operating parameters, since this technique was able to successfully describe influent and effluent total COD. It is thus proposed that combined modelling of 2D-fluorescence data and selected performance/operating parameters should be further explored for efficient MBR monitoring aiming at rapid process control.
膜生物反应器（MBR）因其紧凑性和高出水水质在全球范围内得到越来越多的应用。然而，膜污染主要与生物产物有关，会降低 MBR 的性能。因此，尽可能实时监测 MBR 以加速控制措施以实现最大的生物和膜性能非常重要。二维荧光光谱是一种很有前途的在线工具，可以同时监测废水处理效率和潜在生物污垢形成剂的形成。在这项研究中，使用偏最小二乘法（PLS）成功地对从 MBR 的废水和渗透物中获得的二维荧光数据进行建模，以监测进水和出水总化学需氧量（COD）的变化。结果分析还表明，腐殖酸和蛋白质对两个流道中的测量 COD 有很大贡献。然而，这种方法不适用于 MBR 系统的其他性能参数（如进水和出水氨氮和磷），这些参数通常通过大量分析和操作参数来描述。因此，使用主成分分析（PCA）来寻找这些参数之间可能的相关性，试图减少全面 MBR 特性所需的分析工作，并减少获得监测结果所需的时间框架。前 3 个主成分，捕获了 57%的方差，表明并证实了评估参数之间的预期关系。然而，这种方法本身无法提供足够稳健的相关性，无法用于过程描述（PCA 加载值≤0.5）。尽管如此，二维荧光光谱所捕获的信息有可能替代一些分析和操作参数，因为该技术能够成功描述进水和出水总 COD。因此，建议进一步探索二维荧光数据和选定的性能/操作参数的组合建模，以实现高效的 MBR 监测，从而实现快速过程控制。
