Continuous Measurement of Dissolved Sulphide in Sewers

Highlights -- Grab sampling for wastewater sulfide is problematic due to high in-sewer variability -Continuous measurement of dissolved H2S and pH in conjunction allows high-resolution dissolved sulfide data to be obtained -Continuous measurement is quite doable even on large catchment scale Background The city of Aarhus, Denmark, with its 400,000 PE and its 1,632 km of sewers, experiences odor and corrosion problems. These have intensified over the last decades as wastewater treatment has been centralized, combined systems converted to separate ones, and water consumption has declined. The water utility, Aarhus Vand, is taking a proactive approach to manage septicity issues, part of which is monitoring for septicity and modelling with the sewer process model Mega-WATS. Measuring sulfide in sewers is difficult as there are many possible locations and because sulfide is inherently difficult to measure. The common approach is to grab water samples at known hotspots and use test kits to analyze them on-site or analyze conserved samples at a lab (Matias et al., 2017). H2S gas in the sewer air can also be monitored and water phase dissolved sulfide estimated here from (Vollertsen et al., 2015). These approaches have their pros and cons. Test kits for dissolved sulfide are difficult to use accurately, and total sulfide as analyzed at a lab includes metal-bound sulfides present in the wastewater. A downside of grab sampling is that it inherently assumes little variation in sewer sulfide levels. Sewer air H2S gas monitoring overcomes the issue of variable sulfide levels, but it is challenging to deduct dissolved sulfide levels here from. Recently, sensors have come on the market which can quantify dissolved sulfide in-situ (Despot et al., 2021): UV-Vis sensors that spectroscopically identifies the HS- ion was the first on the market (Sutherland-Stacey et al., 2008), and a Clark-type electrochemical sensor that measures the partial pressure of the H2S molecule followed a decade or so later. Both sensor types hence only measure a fraction of dissolved sulfide (H2S + HS- + S2-) which must be calculated from the measured signal in combination with the wastewater pH. Knowing that wastewater pH in sewers varies, a continuous monitoring of dissolved sulfide hence requires continuous measurement of both pH and one of the dissolved sulfide fractions. The importance of including pH measurement cannot be emphasized too strongly. For example, a sensor measuring only H2S-gas (pKa = 7.04 at 20°C): At pH 6.5 a measured dissolved H2S-gas concentration of 1 mg/L corresponds to 1.3 mg/L of dissolved sulfide in the water. At pH 7.0 it corresponds to 1.9 mg/L, at pH 7.5 to 3.9 mg/L, at pH 8.0 to 10.1 mg/L, and at pH 8.5 it corresponds to 29.8 mg/L. In any network, pH will vary rapidly with time, often one pH unit, sometimes more. Especially at discharge points of force mains, pH is prone to quite some variation. Hence a dissolved H2S-gas measurement without the corresponding pH cannot be used to quantify wastewater total sulfide. In this presentation, a catchment-wide measuring campaign of pH and dissolved H2S gas (using SulfiLoggerTM) is presented, and benefits and issues related to the approach discussed. Finally, the applicability for calibrating a sewer process model, Mega-WATS, is touched on. Methods A Mega-WATS model was set up for Aarhus and the uncalibrated model used to select 14 monitoring locations where sulfide was expected. In addition, 3 locations were selected based on known odor issues, but where Mega-WATS did not indicate any sulfide. Six sets of SulfiLoggerTM and pH meters were used and moved between locations. The locations where sulfide was monitored (with and without success in terms of consistent measuring data) are shown in Figure 1. The two sensors, a Sulfilogger for dissolved H2S-gas and a pH meter, were hung into the sewer so both would be submerged as much time as possible (Figure 2). However, force main discharges and gravity sewers in the upper end of the network only carry water intermittently, and some sensors were hence only intermittently submerged. Results Of the 17 locations, 13 yielded good quality data on both dissolved H2S and pH for at least 3 consecutive dry weather days (Figure 1). The 4 remaining locations failed due to low flow. Three of these were those selected because of known odor issues but for which Mega-WATS predicted zero dissolved sulfide. Figure 3 gives an example of a very good dataset. The top graph shows the raw signal of dissolved H2S-gas and pH, indicating that the first varied between 0 and 1 mg/L while pH varied between 7 and 8. Temperature also varied, which plays a role when assessing the release of H2S-gas to the urban atmosphere. The bottom graph shows dissolved sulfide calculated from combining dissolved H2S and pH, showing that dissolved sulfide varied significantly more than what would be expected from the dissolved H2S alone. The results furthermore show the risk of relying on grab sampling, as it would be problematic to capture the high dissolved sulfide variability. Besides such good quality data as shown in Figure 3, there were numerous issues during the campaign which compromised the data. Keeping the sensors submerged was one, while drifting of the pH sensor was another. Especially intermittent submersion of the pH sensor seemed to compromise the data quality. Figure 4 gives an example where data behave acceptable in the beginning of the monitoring period, albeit drifting somewhat towards lower pH and higher dissolved H2S-gas. On June 1st, the sensor was maintained, and the signal changed nature significantly. It was most likely clogged by some debris, and the first part of the monitoring series hence probably not valid. Reasonable data was obtained for a week or so, upon which pH during the night became clearly untrustworthy, with values slowly drifting downwards, reaching as low as pH 5. Assessing data quality upon monitoring, and probably more important: Actively maintaining the sensors during a monitoring campaign, is hence paramount when conducting continuously measurements of dissolved sulfide in real sewer networks. Conclusions While significant effort is needed to continuously monitor sewers for dissolved sulfide, the benefits are also significant. It allows going from a qualitative assessment of the sulfide in the sewer network, as can be obtained by wastewater grab sampling or sewer air H2S-gas measurement, to quantitative data showing the full variability of the sulfide phenomenon. This allows calibrating a sewer process model to a much higher certainty and to tailor solutions which takes sulfide variability into account.