School of Water, Energy and Environment (SWEE)

Permanent URI for this community

https://dspace.lib.cranfield.ac.uk/handle/1826/8737

Browse

Browsing School of Water, Energy and Environment (SWEE) by Course name "PhD"

Now showing 1 - 4 of 4
  • Thumbnail Image
    ItemOpen Access
    A comparison of floating and sunken media biological aerated filters (BAF)
    (1997-03) Mann, Allan; Stephenson, Tom
    The aim of this experimental work was to directly compare the performance of two types of media support in pilot scale biological aerated filter (BAF) reactors. The two media types were identical in shape and size except one was sunken with a relative density of 1.05 and the other floating with a relative density of 0.92 and made of polypropylene. Empty bed tracer studies were initially undertaken to ascertain the hydraulic characteristics of the media types under different process conditions. Almost ideal plug-flow was seen without aeration but with aeration some mixing and by-passing was seen which increased with higher aeration rates. Aerator design and positioning had little effect on the flow and that the sunken media would perform best in downflow and the floating media in upflow during biological treatment. Two methods of start-up were employed during unsteady state analysis, activated sludge seeding and the use of the process liquid (settled domestic sewage) at the operational flowrate. Both methods showed a similar total start-up time of 28 d. At steady state the floating media removed 78% suspended solids (SS) and 75% soluble chemical oxygen demand (sCOD) compared to 66% and 68% respectively in the sunken media. At high flowrates and during shock loadings of up to 1.5 times the nominal flowrate the floating media again showed a better performance than the sunken media. With increasing shock loadings the recovery time increased with a corresponding decrease in solids and soluble COD removal rates. At the maximum SS loading of 1.397 kg m'3 d'1 ( 1.403 kg m"3 d'1 sCOD) only 35% (30% sCOD) removal was seen in the sunken media compared to 60% (40% sCOD) in the floating media. The steady state results were used to produce an empirical model relating effluent soluble COD to influent COD and reactor height. Assuming plug flow the profiles showed a first order rate of reaction. A first order empirical model was then established based on that used for trickling filters, which produced constants dependent on the media type. During tertiary treatment low ammonia loadings of 0.2 kg m'3 d"1 restricted autotrophic growth and led to preferential growth of heterotrophic bacteria. At ammonia loadings of up to 1.16 kg m'3 d"1 nitrification was rapid. Below 5 °C nitrification was minimal but almost full nitrification was achieved at approximately 10 °C. Though backwashing was carried out only on a weekly basis, overwashing appeared to have caused a gradual reduction in treatment efficiency. During nitrification the sunken media was more efficient at ammonia removal than the floating media, indicating that the autotrophic bacteria prefer the more open structure of the sunken media. Reactor profiles indicated a zero to half order reaction.
  • Thumbnail Image
    ItemOpen Access
    Minimising particulate passage during ripening of direct filters
    (1996-12) Colton, Jason; Fitzpatrick, C. S. B.
    A significant consideration in forward planning for water treatment works design and operation concerns the effectiveness of filtration plant in providing a barrier to particulates in the low micron size range, including Cryptosporidium oocysts. The filter ripening period has long been identified as a cause for concern with respect to particulate passage into the filtrate, this work has shown that up to 54% of all particles that pass into supply during a 48 hour run, do so in the first hour of operation The aim of this study was to investigate methods of reducing particulate passage into the filtrate during ripening and identify optimums. Thus strengthening the filtration barrier to Cryptosporidium oocysts. The work was carried out on a large direct filtration pilot plant treating an upland lake source. Variables investigated included, filter media type and size, backwash regimes and durations, start-up strategies such as slow-start, and filtration rate. The two filter media types and configurations that gave best removal were in beds of 0.5-1.0mm silica sand and 2m beds of 2.0-3.35mm quartz sand. A combined air/water backwash, at rates to achieve collapse-pulsing, reduced the numbers of particles in the filtrate during ripening and increased run lengths when compared to water only wash and air followed by water wash. Optimum backwash durations identified for the 0.5-1.0mm sarid and 2.0-3.35mm sand were 4 and 3 minutes respectively. The effect of start-up strategies on the numbers of particles in the filtrate during ripening was found to be variable. Slow start (both 1/2 hour and 1 hour) performed well under normal conditions experienced at the plant but performance was adversly affected by deteriorating raw water conditions. Delayed start was not as effective as slow start under normal conditions but was much more effective during periods of poorer raw water quality. Higher filtration rates (10 and 20m/h) resulted in much greater numbers of particles in the filtrate during ripening, thus fast-start was not a viable option. Cost benefit analysis of the technical options showed backwash optimisation and delayed start to be attractive retro-fit options in terms of cost effective reduction of particulate passage during filter ripening.
  • Thumbnail Image
    ItemOpen Access
    Upflow and downflow biological aerated filters (BAFs) for wastewater treatment
    (1999-03) Mendoza-Espinosa, Leopoldo Guillermo; Stephenson, Tom
    Investigations on two small pilot-scale biological aerated filters operating in upflow and downflow mode using identical media were undertaken in order to compare their performance for the removal of carbonaceous matter and ammonia, calibration of empirical models for suspended solids, soluble and total COD and their resistance to organic and hydraulic shock loadings. The initial existence of a plenum chamber under the media bed caused severe reduction in active volume in the downflow reactor. Elimination of the plenum chamber by means of the addition of media solved the problem and both reactors presented plug-flow conditions. Their performance for carbonaceous matter and ammonia removal was compared. With the exception of ammonia, the removal efficiencies for the parameters studied in both types of reactor were very similar. The downflow reactor achieved a greater removal of ammonia than the upflow BAF. Autrotrophic activity was located at the bottom of the reactor in the downflow column and at the top in the upflow unit. Had the running conditions been set identically for both columns, the difference in performance between the two reactors may have been due to the impact of the backwashing on the biological population, effectively washing-out the autotrophs. The verification of a simple empirical model for soluble COD removal for a different type of media than the previously reported was conducted. The new values for the media constant and the overall process constant of the system suggested that overall, the performance of the system was dictated by its media type/liquid flow characteristics. The constants from 4 empirical models were used to predict the removal of SS, sCOD and tCOD in a small BAF (0.07 m3 empty bed volume) and a large BAF (0.69 m3 empty bed volume). The model for soluble COD that included parameters such as gas and liquid velocity provided values closer to the actual observed values. The model for suspended solids and tCOD removal (based on a rapid gravity filters model) was efficient in predicting the overall removal rates of SS and tCOD in both reactors. When the models were used for scaling-up results, only the simplest model that did not include parameters such as gas and liquid velocity was effective. It was demonstrated that a small pilot-scale BAF was not capable of absorbing high peak organic shock loads when the organic loads were increased from a normal load of 1.2-1.4 kg sCOD m-3 d-1 to shock loads between 5.1- 7.3 kg sCOD m-3 d-1, based on empty bed volumes. Nevertheless, the effect on the biomass was limited as normal performance resumed very quickly. Increases in the hydraulic velocity from 0.7-1.0 m h' 1 to 1.5-2.9 m h" 1 had little affect on BAF performance in terms of soluble COD removal. Soluble COD effluent concentrations of <40 mg I-1 (-16 mg 1-1 BOD5) were detected thus, meeting most discharge consents in UK sewage works. The recovery of the reactor in both cases was fast, resuming normal performance within 60 min after the application of the shock load.
  • Thumbnail Image
    ItemOpen Access
    Using particle monitors to minimise Cryptosporidium risk
    (2002-07) Hamilton, Paul; Parsons, Simon
    Over the last two decades, there bas been much interest in whether particle counters hold any significant benefit over conventional nephelometric turbidimeters in monitoring potable water treatment processes. Southern Water, which supplies drinking water to two million customers living in Kent, Sussex, Hampshire and the Isle of Wight first used particle counters at one of its works in 1992. This study presents the key results of a three-year study, conducted in conjunction with Cranfield University to find the most beneficial use of these monitors, so that a sensible investment can be made. This study comprised a series of monitoring trials, conducted at different ground and surface water treatment works. In many instances,. there was a strong similarity between turbidity and particle count trends, effectively making one of monitors redundant However, particle counters were shown to be beneficial in three ways: (a) they could be more sensitive to changes in water quality at low turbidities (below 0.1 NTU), (b) they could be more sensitive to changes associated with larger particle sizes and ( c) they could also provide useful information on particle size distribution. This issue of monitor sensitivity bas been analysed using a regression model built from experimental data For a given water sample, this model predicts how many more times sensitive particle counters will be, in detecting changes in water quality, compared to nephelometric turbidimeters. This indicated that whereas turbidimeters typically 'flat-line' at low values, particle counters are frequently more sensitive and so can be used as a fine-tuning optimisation tool below 0.1 NTU. However, this sensitivity is also proportional to the particle size distribution of the sample; particle counters are more suited to samples containing a high proportion of large particles (> 1 0µm). This explains why particle counters are not always 'more sensitive' below 0.1 NTU. Although no links could be found between particle counts (and turbidity) and Cryptosporidium oocysts, it appears that if oocysts are present in the raw water then inferior particle removal across a treatment process can lead to increased risk. Cryptosporidium oocysts were found even in very low turbidity (<0.1 NTU) treated water samples. This shows the need for fine-tuning treatment processes below 0.1 NTU and highlights a potential optimisation role for particle counters. The study concludes, however, that particle counters are best used as an optional process research/optimisation tool only: turbidimeters remain the preferred monitor for process control Indeed, the study finds no overwhelming evidence to justify the permanent installation of particle counters at treatment works. However, an increased use of portable particle counters in optimisation work is recommended. Consideration is given to other practical concerns such as where and how to use particle counters and what parameters to measure. The value of particle counters' sizing ability has also been assessed. In addition to sensitivity modelling, particle size distribution data revealed a large difference in the volume of particles passed by two sludge treatment plants. The study concludes that, where particle counters are used, there may be some value in monitoring particle size distribution using a particle size ratio or a similar statistic.