Evaluate the impact of EBCT
for in-plant rapid media filters
Scott Summers, CU

Research Approach

Innovative filter modifications and operations will be evaluated with the goal of extending the EBCT to control particulate matter surges, DBPs and DBP precursors. Several approaches are available. One direct method is to decrease filter velocities, HLRs, by reducing the flow rate to a given filter area. Another is to maximize the filter media depth. Both of these approaches will result in improved removal efficiency, but increasing the length will cause more head loss.
Experiments will be run to evaluate the impact of increasing EBCT from 5 (typical) to 15 min to 30 min at different velocities. In phase one bench scale tests will evaluate the control of DBPs and DBP precursors, as particulate removal in filters does not scale up accurately. Surges in particulate matter will be evaluated only at the pilot plant level in phase two.

A controlling factor in the performance of this process is the EBCT. The EBCT in most conventional filters is typically 3 to 9 min, which yields sub-optimal performance, especially under cold temperatures. In addition the amount of biomass in the filters is also important. One option is the use of slow sand filters (SSFs) with EBCTs on the order of 3 to 8 hours. While very effective, SSFs require extensive surface areas (footprint), 50 to100 times that of a conventional media filter, and many currently operating utilities do not have sufficient space for SSF implementation. Another limiting issue is the extensive use of prechlorination, which restricts biomass growth within the filter. Thus, implementation of this technology requires the facility to discontinue pre-chlorination.

Recent work by our team, and others, has shown that increasing the EBCT to the range of 20 to 30 min leads to reduction in DBP precursors or preformed DBPs similar to that in a SSF. For cases in which chlorine residuals are present in the filter influent, then a very thin layer of GAC (< 1 min) effectively removes the free chlorine and allows the biomass to grow in the lower depths of the filter. These changes will allow the biodegradable DBPs (e.g., HAAs) and/or DBP precursors to be effectively biodegraded.

Zearley and Summers (2012) classified pseudo first order rate constants classified into four categories based on three natural breaks at 15%, 50% and 85% steady state removals at 7.9 min of EBCT. The rate constants were classified as: recalcitrant (k' < 0.022 min-1), slow (0.022 min-1 < k' < 0.093 min-1), fast (0.093 min-1 < k' < 0.248 min-1), and very fast (k'' > 0.248 min-1). Using these pseudo first order rate constants as boundaries different zones of performance can be established as a function of the product of EBCT. This is shown in Figure 3-3a.


Servais et al. (2005) have reviewed and summarized the research on biodegradation of OM. Most research has divided the BOM into a fast reacting fraction and a slow reacting fraction. For ozonated waters the fast reacting fraction this can be 50% of the BOM, for raw waters it is usually much lower as most of the fast reacting fraction has degraded in the environment. The model results in Figure 3-3a  show that in the EBCT range of most filters, 3 to 9 min, that 60 to 85 % of the fast reacting BOM is removed, while only 20 to 40 % of the slow biodegrading BOM is removed. Extending the EBCT for the fast fraction only adds an additional 15 to 30 % removal, but for the slow fraction, which makes up most of the  raw water BOM, an additional 40 to 60 % removal maybe achieved. In addition to EBCT, the biomass concentration can also impact the pseudo first order rate constants as shown in Equation 3-2, with higher biomass yielding higher k’ values. Temperature impacts k’ values as well, with increasing temperature yielding increasing k’ values. The impact of biomass and temperature on bioremoval of the fast reacting BOM fraction (k” = 0.003 ml/nmolPO4 min). For biomass concentrations of 150 nmol PO4 (mL bed)-1 the removals at EBCTs of 3 and 9 min are shown on the figure. Under warm temperatures even an EBCT of 3 min will yield more than 60% removal, but at cold temperatures the removal goes down to less than 25%. If a 70% removal is required, then even for the fast reacting BOM fraction, EBCTs greater than 9 min are needed at low temperatures.

Our experimental system has been designed using these modeling results and recent work by our team, and others, that has shown that increasing the EBCT to the range of 20 to 30 min leads to reduction in DBP precursors or preformed DBPs similar to that in a SSF. In phase one experiments will be run at the bench scale to evaluate the impact of increasing EBCT from 5 (baseline) to 15 min to 30 min at different velocities. To facilitate the pre-acclimated media will be utilized in the columns. Three source waters with different BOM concentrations will be used and using jacketed columns the temperature will be controlled at 8, 18 and 22 C. Since many utilities practice pre-chlorination, side-by-side comparisons of biofiltration treatment with and without prechlorination will be made. For cases in which chlorine residuals are present in the filter influent, then a very thin layer of GAC (< 1 min) will be added to remove the free chlorine and allow the biomass to grow in the lower depths of the filter. The influent and effluent will be sampled weekly for TOC, DBP and DBP precursors. In addition to the regulated TTHM and HAA9, nitrogenous DBPs and other unregulated DBPs will be assessed (HANs, chloropicrin, chloral hydrate, 1,1-dichloro-2-propanone and 1,1,1-trichloro-2-propanone). In addition biomass measures will be taken from the filter media.

In phase two of the study both DBP removal by the biofilters and surges in particulate matter will be evaluated only at the pilot plant level. The UC-B pilot plant consists of two parallel trains operated at a flow rate of 1.5 gpm (2 gpm design) per train. Source water is sent to the raw water tank, from which water is pumped to the treatment trains. Each train consists of rapid mix, three stage tapered flocculation, and sedimentation. Rapid mixing of the coagulant and any preoxidants are achieved via a static mixer through three chemical dose ports prior to the mixer. Vertical turbines are used for flocculation. Tube settlers can be added to aid sedimentation in the baffled sedimentation basin. The settled water can be pumped or flow by gravity to four parallel filter columns. The filter columns are 8 feet tall with a diameter of 3.0 inches. The pilot plant can be configured to run either train with 1, 2, 3, or all 4 filters. At a flow rate of 1.5 gpm the average detention times are approximately 0.5 hours for flocculation and 1.5 hours for sedimentation. Normal flow through the filters at 0.25 gpm with the excess water overflowing to waste, maintains a constant head. Chemical dose points include the rapid mix (RM), mid-flocculation (MF), and post-sedimentation (PS). All chemicals are dosed with cartridge pumps. Online turbidimeters allow for the monitoring of the raw, settled and filtered waters.

The pilot plants will be run with a local surface water. Currently, both pilot plants are set up at the City of Boulder’s Betasso Water Treatment plant where they can treat one of two source waters; a reservoir and a high mountain stream. The advantage of this location is that in the winter and spring the water temperature is cold, less than 10 C, and natural surges in water quality occur in the spring and summer with TOC concentrations increasing from a baseline of 2 mg/L to above 12 mg/L. Other available locations include a) the Boulder Reservoir which is fed by local runoff and Western Slope water and has stable water quality year round, and b) site at the UC-B where Boulder Creek water, also with naturally occurring water quality surges, is available or a groundwater is available.

Three anthracite/sand filters of equal depth will be run in parallel at EBCTs of 5 (control), 15 and 30 min and sampled for turbidity by an online system. The latter two filters will also be sampled by hand at 5 min EBCT. The filters will be run at steady state for one month to allow for bioacclimation. After one month, samples for turbidity, TOC, DBP and DBP precursors will be taken weekly for one month from the influent and the five effluent ports, in addition to the online turbidity measurements. In addition biomass measures will be taken from the filter media.

After the baseline is established, the source water particle concentration will be increased for one day through the addition of collected particles from the source water and the impact on turbidity measured. Some grab samples TOC, DBP and DBP precursors will be taken. Based on the results of this first run we will assess other operating conditions such as perturbations in the coagulation process and various stop-start operation conditions. The specifics of these runs will be based on the results from the first run. We may want to evaluate the impact of different influent turbidity levels, different media types and configuration, “bump” backwashing the filters prior to re-startup, increases in HLR to simulate one filter of a bank being taken out of service for backwashing. E. coli challenges are also planned for these latter runs.