Quantitative Fr 13 Failure Modelling of Uv Irradiation for Potable Water Production – Demonstrated with Escherichia Coli

Abstract

Steady-state ultraviolet (UV) irradiation for potable water production is becoming an important global alternative to traditional disinfection by chlorination. Failure of UV to reduce the number of viable contaminant pathogens however can lead to enduring health legacies (with or without fatalities). To better understand vulnerability of UV operations to failure, the probabilistic Fr 13 risk framework of Davey and co-workers1 is applied for the first time in this thesis. Fr 13 is predicated on underlying chemical engineering unit-operations. It is based on the hypothesis that naturally occurring, chance (stochastic) fluctuations about the value of ‘set’ process parameters can unexpectedly combine and accumulate in one direction and leverage significant change across a binary ‘failure– not failure’ boundary. Process failures can result from the accumulation of these fluctuations within an apparent steady-state process itself. That is to say, even with good design and operation of plant, there can be unexpected (surprise and sudden) occasional failures without ‘human error’ or ‘faulty fittings’. Importantly, the impact of these naturally occurring random fluctuations is not accounted for explicitly in traditional chemical engineering. Here, the Fr 13 risk framework is applied for the first time to quantitatively assess operations of logically increasing complexity, namely, a laminar flow-through UV reactor, with turbulent flow in a concentric annular-reactor, both with and without suspended solids present (Davey, Abdul-Halim and Lewis, 2012; Davey and Abdul-Halim, 2013; Abdul-Halim and Davey, 2015; 2016), and; a two-step ‘global’ risk model of combined rapid-sand-filtration and UV irradiation (SF-UV) (Abdul-Halim and Davey, 2017). The work is illustrated with extensive independent data for the survival of viable Escherichia coli - a pathogenic species of faecal bacteria widely used as an indicator for health risk. A logical and step-wise approach was implemented as a research strategy. UV reactor unit-operations models are first synthesized and developed. A failure factor is defined in terms of the design reduction and actual reduction in viable E. coli contaminants. UV reactor operation is simulated using a refined Monte Carlo (with Latin Hypercube) sampling of UV lamp intensity (I), suspended solids concentrations [conc] and water flow (Q). A preliminary Fr 13 failure simulation of a single UV reactor unit-operation (one-step), developed for both simplified laminar flow and turbulent flow models, showed vulnerability to failure with unwanted survival of E. coli of, respectively, 0.4 % and 16 %, averaged over the long term, of all apparently successful steady-state continuous operations. A practical tolerance, as a design margin of safety, of 10 % was assumed. Results from applied ‘second-tier’ studies to assess re-design to improve UV operation reliability and safety and to reduce vulnerability to Fr 13 failure showed that any increased costs to improve control and reduce fluctuations in raw feed-water flow, together with reductions in UV lamp fluence, would be readily justified. The Fr 13 analysis was shown to be an advance on alternate risk assessments because it produced all possible practical UV outcomes, including failures. A more developed and practically realistic model for UV irradiation for potable water production was then synthesized to investigate the impact of the presence of suspended solids (SS) (median particle size 23 μm) as UV shielding and UV absorbing agents, on overall UV efficacy. This resulted in, respectively, some 32.1 % and 43.7 %, of apparent successful operations could unexpectedly fail over the long term due, respectively, to combined impact of random fluctuations in feed-water flow (Q), lamp intensity (I0) and shielding and absorption of UV by SS [conc]. This translated to four (4) failures each calendar month (the comparison rate without suspended solids was two (2) failures per month). Results highlighted that the efficacy of UV irradiation decreased with the presence of SS to 1-log10 reduction, compared with a 4.35-log10 reduction without solids present in the raw feed-water. An unexpected outcome was that UV failure is highly significantly dependent on naturally occurring fluctuations in the raw feed-water flow, and not on fluctuations in the concentration of solids in the feed-water. It was found that the initial presence of solids significantly reduced the practically achievable reductions in viable bacterial contaminants in the annular reactor, but that fluctuations in concentration of solids in the feed-water did not meaningfully impact overall vulnerability of UV efficacy. This finding pointed to a pre-treatment that would be necessary to remove suspended solids prior to the UV reactor, and; the necessity to improve control in feed-water flow to reduce fluctuations. The original synthesis was extended therefore for the first time to include a rapid sand-filter (SF) for pre-treatment of the raw feed-water flow to the UV reactor, and; a Fr 13 risk assessment on both the SF, and sequential, integrated rapid sand-filtration and UV reactor (SF-UV). For the global two-step SF-UV results showed vulnerability to failure of some 40.4 % in overall operations over the long term with a safety margin (tolerance) of 10 %. Pre-treatment with SF removed SS with a mean of 1-log10 reduction (90 %). Subsequently, an overall removal of viable E. coli from the integrated SF-UV reactor was a 3-log10 reduction (99.9 %). This is because the efficacy of UV light to penetrate and inactivate viable E. coli, and other pathogens, is not inhibited by SS in the UV reactor. This showed that the physical removal of E. coli was accomplished by a properly functioning SF and subsequently disinfection was done by UV irradiation to inactivate viable E. coli in the water. Because the Regulatory standard for potable water is a 4-log10 reduction, it was concluded that flocculation and sedimentation prior to SF was needed to exploit these findings. Flocculation is a mixing process to increase particle size from submicroscopic microfloc to visible suspended particles prior to sedimentation and SF. This research will aid understanding of factors that contribute to UV failure and increase confidence in UV operations. It is original, and not incremental, work. Findings will be of immediate interest to risk analysts, water processors and designers of UV reactors for potable water production.