Performance improvement of wastewater treatment processes by application of machine learning.

Improving wastewater treatment processes is becoming increasingly important, due to more stringent effluent quality requirements, the need to reduce energy consumption and chemical dosing. This can be achieved by applying artificial intelligence. Machine learning is implemented in two domains: (1) predictive control and (2) advanced analytics. This is currently being piloted at the integrated validation plant of PUB, Singapore's National Water Agency. (1) Primarily, predictive control is applied for optimised nutrient removal. This is obtained by application of a self-learning feedforward algorithm, which uses load prediction and machine learning, fine-tuned with feedback on ammonium effluent. Operational results with predictive control show that the load prediction has an accuracy of ≈88%. It is also shown that an up to ≈15% reduction of aeration amount is achieved compared to conventional control. It is proven that this load prediction-based control leads to stable operation and meeting effluent quality requirements as an autopilot system. (2) Additionally, advanced analytics are being developed for operational support. This is obtained by application of quantile regression neural network modelling for anomaly detection. Preliminary results illustrate the ability to autodetect process and instrument anomalies. These can be used as early warnings to deliver data-driven operational support to process operators.

Fate and occurrence of alkylphenolic compounds in sewage sludges determined by liquid chromatography tandem mass spectrometry.

An analytical method has been developed and applied to determine the concentrations of the nonionic alkylphenol polyethoxylate surfactants and their metabolites, alkylphenoxy carboxylates and alkyphenols, in sewage sludges. The compounds were extracted with methanol/acetone (1:1 v/v) from sludge, and concentrated extracts were cleaned by silica solid-phase extraction prior to determination by liquid chromatography tandem mass spectrometry. The recoveries, determined by spiking sewage sludge at two concentrations, ranged from 51% to 89% with method detection limits from 6 microg kg(-1) to 60 microg kg(-1). The methodology was subsequently applied to sludge samples obtained from a carbonaceous activated sludge plant, a nitrifying/denitrifying activated sludge plant and a nitrifying/ denitrifying activated sludge plant with phosphorus removal. Concentrations of nonylphenolic compounds were two to three times higher than their octyl analogues. Long-chain nonylphenol polyethoxylates (NP3-12EO) ranged from 16 microg kg(-1) to 11754 microg kg(-1). The estrogenic metabolite nonylphenol was present at concentrations ranging from 33 microg kg(-1) to 6696 microg kg(-1).

Treatment and removal strategies for estrogens from wastewater.

Natural and synthetic steroidal estrogens (estrone, 17beta-estradiol and 17alpha-ethinylestradiol) are endocrine disrupters, that are discharged consistently from the sewage treatment works into surface waters, thereby causing endocrine disrupting effects to aquatic organisms at trace concentrations (nanogram per litre). Several years of research have been focused on their fate, behaviour and removal in the environment but primarily in the sewage treatment works which acts as a sink for these compounds. This review attempts to summarize the factors involved in the removal of these chemicals from the sewage treatment works. Biological processes, and to a limited extent physio-chemical properties, play a vital role in the endocrinal deactivation of these compounds. The efficiency of these processes is highly dependent on operating parameters (such as sludge retention time, redox potential, etc) that govern the secondary treatment process of a functional sewage treatment works. Although advanced treatment technologies are available, cost and operational considerations do not make them a sustainable solution.

Determination of steroid estrogens in wastewater by high performance liquid chromatography-tandem mass spectrometry.

This paper discusses the requirement for, and presents an analytical procedure for, the determination of four steroid hormones and a conjugated steroid (estrone-3-sulfate) in wastewaters. The method utilizes LC/MS/MS following solid phase extraction and a two stage clean-up procedure, achieving limits of detection of 0.2 ng l(-1) for estriol, 17beta-estradiol and 17 alpha-ethinylestradiol, and 0.1 ng l(-1) for estrone and the conjugate. The approach demonstrates that using appropriate clean-up and deuterated internal standards, the impact of matrix effects on ionization can be overcome to reliably determine estrogens at environmentally relevant concentrations. The robustness of the method was demonstrated by achieving recoveries of >83% for all steroids in settled sewage and final effluent samples with relative standard deviations of 0.5-12%.

The effectiveness of anaerobic digestion in removing estrogens and nonylphenol ethoxylates.

The fate and behaviour of two groups of endocrine disrupting chemicals, steroid estrogens and nonylphenol ethoxylates, have been evaluated during the anaerobic digestion of primary and mixed sewage sludge under mesophilic and thermophilic conditions. Digestion occurred over six retention times, in laboratory scale reactors, treating sludges collected from a sewage treatment works in the United Kingdom. It has been established that sludge concentrations of both groups of compounds demonstrated temporal variations and that concentrations in mixed sludge were influenced by the presence of waste activated sludge as a result of transformations during aerobic treatment. The biodegradation of total steroid estrogens was >50% during primary sludge digestion with lower removals observed for mixed sludge, which reflected bulk organic solids removal efficiencies. The removal of nonylphenol ethoxylates was greater in mixed sludge digestion (>58%) compared with primary sludge digestion and did not reflect bulk organic removal efficiencies. It is apparent that anaerobic digestion reduces the concentrations of these compounds, and would therefore be expected to confer a degree of protection against exposure and transfer of both groups of compounds to the receiving/re-use environment.

The significance of sample mass in the analysis of steroid estrogens in sewage sludges and the derivation of partition coefficients in wastewaters.

Optimization of an analytical method for determination of steroid estrogens, through minimizing sample size, resulted in recoveries >84%, with relative standard deviations <3% and demonstrated the significance of sample size on method performance. Limits of detection were 2.1-5.3 ng/g. Primary sludges had estrogen concentrations of up to one order of magnitude less than those found in biological sludges (up to 994 ng/g). However, partition coefficients were higher in primary sludges (except estriol), with the most hydrophobic compound (ethinylestradiol) exhibiting the highest Kp value, information which may be of value to those involved in modeling removal during wastewater treatment.