Identification of oil contamination in process water using fluorescence excitation emission matrix (FEEM) and parallel factor analysis (PARAFAC).

Fuel oil is widely used within Eskom, a power generation company in South Africa. Eskom's coal-fired power stations use up to 30,000 L of fuel oil per hour during a cold start-up, a consequence of which results in oil leaks to the dams. Oil contamination in water treatment plants causes irreversible membrane fouling, requiring costly replacement. This research work focused on the development of a rapid method for the identification of low concentrations of the water-soluble oil component fraction of crude fuel oil. For the developed method, known volumes of the water-soluble fraction of crude oil were spiked into various matrices of process water. FEEMs were collected using the patented HORIBA Aqualog spectrometer and data were modelled with PARAFAC. The results were well described with a four-component model, which included an oil component and three natural organic matter components, with a split-half validation match of 90%. The oil component was verified using linear regression of the PARAFAC component scores yielding an R2 value of 0.98. From the scores, a qualitative pass/fail test was developed such that process water can be analysed and subjected to the model to indicate the presence of oil contamination beyond a damaging threshold.

Seasonal variations of natural organic matter (NOM) in surface water supplied to two coal-fired power stations.

Global issues such as pollution and global warming have resulted in changes in water characteristics over the past 20 years. Natural organic matter (NOM) which is a major component in water systems has shown an increase globally. This increase in NOM concentration has negatively affected both water treatment processes and drinking water quality. It is subsequently critical to understand the seasonal variations and composition of NOM to be able to address issues related to NOM. In this study, techniques such as ultraviolet-visible spectroscopy, total organic carbon and liquid chromatography-organic carbon detection (LC-OCD) were used for characterisation and quantification of NOM. Two coal-fired power stations were selected for this study with each power station receiving water from a different source, i.e. power station A receives water from the Vaal River and power station B from the Nkomati River. Results from this study demonstrated that composition and concentration of NOM from these two water sources varied seasonally. Characterisation of NOM using the LC-OCD indicated that the different fractions of NOM, i.e. low molecular weight neutrals, low molecular weight acids, building blocks, humic substances and biopolymers, varied seasonally. The dissolved organic carbon concentration and specific ultraviolet absorbance values of the raw water at both power stations showed an increment amid the wet seasons and a decrease amid the dry seasons.