ed either fully or partially complete. 11 the responses . e questIOnnaIres: . TABLE 1 .ho ddition, the artIcle dIscusses a Locations of operating slow sand .n. slow sand filtration research facilities that responded to the survey 't at Logan, Utah, and compares City and State 1hf1cility with dfiltr ation faclhtles. The faclhty at Ashland, Wis. Pawhuska, Okla. 3 Auburn. N.Y. Peekskill , N.Y. 1 n has a designed capacity of 75 m / d Denver, Colo. Port Henry, N.Y. rovide long-term operating data Gilbertsville, N.Y. cranton, Kan. I contribute to an overall evaluatIOn Hamden. Conn. Short vi lle, N.Y. he performance of slow sand filters. Honolulu. Hawaii Stayton. Ore. Ilion , N.Y. Stilwell. Okla. t article evaluates the Logan facility Kelley Island. Ohio uperior. Wis. "h reference to being of opera t- Lane ville, Ind. Waipio, Hawaii :treat ment plant faclhtles. Little Falls. N.Y. Wat onville, Calif. Lyndonville. N.Y. Waverl y. N.Y. ach McIndoe Falls, Vt. We t Hartford , Conn. Figure 1. Populations of communities Ogdensburg. N.Y. We tfield , Mas . The survey questionnaire was de- served by operating slow Parker. Kan. med to collect information plants overall physical charactenstlcs of iv idual facilities, including age, ca- 'ity, and served; the char- eristic q uah ty of mfl uen t and effl uen t ler; the of the facilities; the rating practIces; and the costs of at men t. To ensure a reasonably high percent- of return, the questionnaire was An aerial view of the Kassler slow-rate sand filtration plant shows the six basins with one basin emptied for . anized simply and required only scraping. This plant has served the Denver Water Department for more than 80 years. aightforward responses. Its of the survey Geographic distribution of slow sand 'on plants. Table 1 lists the locations Figure 2. Ages of operating slow sand Figure 3. sources used at facilities that responded to the ques- filtration plants operating slow sand filter plants mnaire. Most of these facilities are ated in the eastern United States. 100.0 100 The Application and Effectiveness 8.0 80.0 trty-seven percent of the filtration 60.0 6.0 SO.O 5.0 nts in the survey are located in New 4.0 40.0 3.0 Jrk. No other state reported more than 30.0 2.0 • of Slow Sand Filtration o low sand filtration facilities. 20.0 characteristics of slow sand 1.0 mon facilities. Survey results indi- 10.0 0.8 8.0 • ted that most existing slow sand " 0.6 c:", 6.0 0.5 in the United States ration facilities serve communities of 5.0 ! 0.4 ;; ;; 4.0 • 0.3 'er than 10000 persons (Figure 1). 3.0 Lloyd A. Slezak and Ronald Sims C. me 31 percent of the facilities serve 0.2 2.0 pu lations of fewer than 1000 persons. 0.10 me fa cilities serving larger popula- 1.0 0.08 m use slow sand filters in parallel 0.8 0.06 • A survey of 27 slow sand filtration plants in the United States indicated that most of these 0.6 0.05 There is currently no known ource of hrapid sand filters that were installed 0.5 004 plants are currently serving communities of fewer than 10000 persons, are more than 50 information concerning application of . er the construction of the slow sand 0.4 0.03 0.3 years old, and are effective and inexpensive to operate. A slow sand filtration research facility slow sand filtration in the United tat h' er plants. Data collected on plant 0.02 in Logan, Utah, was compared with the operating plants to determine if locally available, However, information t of Pacities. as expected, closely paralleled 0.2 unsieved sand achieved similar results. The 75-m 3/ d research facility performed well in design and operating tic . epOpu lation served. The Logan, Utah, 99,8 99 98 9590 80706050403020 10 5 2 1 removing turbidity, coliform , and particles of a size representative of Giardia cysts. slow sand facilities now in use le n ._ earch faci lity has a plant capacity of 99.8 99 98 95 90 80 70 60504030 20 10 5 Percent Exceeding MCL tial for eval uating the feasibility of ,n)lfd, which is significantly less than Percent Exceeding MCL Slow sand filtration for providing Figure S. Average filtered water tur- biologically active top layer of sand is applications of this technology. f 1:0 . tof the operating slow sand filtration Figure 4. Average raw water turbid- bidities at operating slow sand filter potable water has been used since the scraped away. type of information should be use u ffl' Iities. nineteenth century. However, this tech- ities at operating slow sand filter plants Becau e small communities in the consulting engineers, regulatory °ur_ Figure 2 shows the distribution of the nology is not now as widely applied in plants United States cannot take advantage of cials, and plant operators. For Pan .of the surveyed slow sand filtration the United States as rapid sand filtra- the economies-of-scale associated with pose of obtaining this information. nd IIties. Some 56 percent of facilities TABLE 2 tion. Slow sand filters are operated at S3 rapid sand filtration, slow sand filtration attempt wa made to locate all slOW 'e rnore than 50 years old, but there has Filter component depths at operating slow sand filtration facilities filtration rates between and ch 0.1 0.4 may be a feasible, economical alternative filtration facilities identified . n recent construction activity asso- m3/ h/ m2 of filter area, whereas rapid ::irt Coefficient of ·for treatment of small and rural com- state regulatory agency. A alt ed with slow sand filtration. Figure Filter Mean Depth Variation Number of sand filtration uses filtration rates be- munity supplies. Accord- was designed and sent to these act Component cnl perce"t Responses 3 marY ·a °sh ows that from 25 to 50 years ago tween 5 and 15 m / h/ m2• 1 Slow sand ingly, the research described in this in an effort to compile a th e rate of construction of slow sand Water 173 4 25 filters treat low-turbidity waters « 10 article was conducted to investigate the of slow sand filtration I eft tration facilities declined. Sand medium 84 30 24 ntu) for weeks or months before accumu- es Support media 56 72 25 limitations and potential benefits of United States . Forty-seven faclhtl. nal. Most slow sand filtration lated materials clog the top layer of using slow sand filters for small com- identified and were sent questIOn re re- . we llitles use lakes or reservoirs as raw sand, at which time 1 to 2 cm of the munity systems. Twenty-seven questionnaires ater sources (Figure 3). A few use

38 MANAGEMENT AND OPERATIONS JOURNAL A CCMBER 1984 L.A. SLEZAK & R.C . SIMS 39 rivers or streams and groundwater. __,_ --, ead for the process. Table 2 h the reported values for each or . pant · reservOIrs reported th depths. T hese data show large ca..... 10.0 TABLE 3 measures are somet' at al,... The range of sand depths 9.0 L U Imes req . 8.0 Sand characteristics at operating slow sand f iltration facilities .ogan, tah, research f . 38 to 183 cm. 7.0 flver as a raw wate aCth 6.0 Effective Uniformity Sand '1 " r source uniformity of filter medium All f aCl 1tIes reported . 1000.0 5.0 Size (D )· IO Coefficient Cost n to have a significant effect on 800.0 • Facility mm ity removal efficiencies D60t/DIO Sand Source dollars/ tOil . 19ure perfo rmance. Table 3 itemizes 600.0 4.0 Ashland. Wis. 0.26 1.6 to questions related to the 500.0 Beach 10.00 400.0 3.0 Auburn. N.Y. 0.3 1.5 Supplier-sieved • 4.69 ics of sand used at the facil- 300.0 Denver. Colo. Q.34 2.76 River-sieved facil reported very uniform Gilber·tsville. N.Y. 0.40 2.25 Supplier-sieved itie 200.0 2.0 Hamden. Conn. 0.3 2.2 Supplier and mo t had effective sizes in Honolulu. Hawaii 0.45-0.55 1.2-1. 7 Beach-sieved of 0.1 to 0.5 mm. Many facilities 100.0 Ilion. N.Y. 0.2-0.4 <3.5 Supplier-sieved 9.25 sand that was sieved by 80.0 Kelley Island. Ohio Supplier-sieved 63.00 Costs varied from slightly 60.0 1.0 Lanesville. Ind. Supplier 4.60 50.0 • 0.9 Little Falls. N.Y. Supplier-washed th an $4/ton to $63/ ton. Many 40.0 • 0.8 Lyndonville. N.Y. Supplier-sieved 4.35 tie also reported that filter sand, 30.0 0.7 Pawhu ka, Okla. 0.25-0.35 2.5-3.5 Supplier-sieved • 0.6 purcha d, ha never been replaced 20.0 Port Henry. N.Y. Supplier-sieved u e the and is recycled. The sand o.? cranton. Kan. Supplier-sieved in the Logan facility is the least 0.4 Shortsville. N.Y. 0.35-0.45 1.7 Supplier-sieved 5.00 10.0 • Stayton. Ore. 1.2-1.4 1.4-1.6 Supplier-sieved rm and has a smaller effective 8.0 03 Stilwell, Okla. River-sieved 45.00 the plant is used to evaluate 6.0 Superior. Wis. 0.45-0.55 1.35-1 .70 Supplier-sieved 28.95 5.0 performance of cheaper, locally 4.0 Waipio. Hawaii Beach-sieved 20.00 0.2 Watsonville, Calif. 0.3 2.3 hIe unsieved sand. 3.0 Waverly, N.Y. 0.12 3.5 Supplier 4.00 The slow-rate sand filtration 'onal characteristics. The primary 2.0 West Hartford, Conn. Supplier·sieved 10.30 roll parameter in slow sand plant at Logan. UIIl". able Westfield . Mass. 0.3 2.3 Native. washed 0.1 tion is filtration rate. Filtration 1.0 99.99 99.8 9998 95 90 80 706050403020 10 5 consIsts of a control building, til 99.99 99.8 9998 95 90 80 706050403020 10 5 2 of thrqugh which 10 percent of sand will pass Percen1 Exceed,"!! MCl two-cell filter, and a storage: at slow sand facilities are less than Percent Exceeding MCl t Slze of sieve openmg through which 60 percent of sand will pass at rapid sand facilities. Filtration Figure 7. Coliforms in filtered waters presedimentation tank. Figure 6. Coliforms in raw waters at ypical rapid sand filtration rate.s, at operating slow sand filter plants operating slow sand filter plants that .about 50 percent of the fa 'Ii out chemical treatment, results 10 TABLE 4 had mfluent turbidity level at m eptable levels of filter performance. Filter cycle durations at operating slow sand filtration facilities than twice the maximum contami it ionally , lower filtration rates at Mean Length of Coefficient of level (MCL) of 1.0 ntu. Figure 5 Filtt:r Cycle Variation Number of that fewer than 15 percent of th Season days percent Responses had effluent turbidity levels in Spring 42 73 21 the MCL. Turbidity of raw water Summer 43 63 21 Logan facility varies from 1.0 to 15.0 Fan 46 70 21 and the effluent turbidity is conCli· ...."t .. Winter 60 72 22 less than 1.0 ntu. Figures 6 and 7 show coliform As with turbidity, all plant r TABLE 5 good performance. Figure 6 ho Comparison of slow sand filtration facilities surveyed with Logan, Utah, facility more than 20 percent of the facili . Value for Logan, Utah, Mean Value From influent coliform levels in Parameter Facility Facilities Surveyed 100/ 100 mL. Figure 7 indicat Figure 9. Season of longest filter cycle percent of the facilities reported Figure 8. Filtration rates at operating duration at operating slow sand filter Plant capacity mVd 75 39000 slow sand filter plants-m3/h/ m2 Filtration rate-mVh/ m2 0.20 0.15 coliform levels of 1/ 100 mL or I plants Influent turbidity-ntu 5 4 Data related . to other water q Effluent turbidity-ntu 0.28 0.65 parameters, such as organic and. Sand depth-em 30 33 ganic species, were not requested 10 e-up view of the experimental Water depth-em 122 173 ·rate sand filter shows the two· Sand effective size-mm 0.18 0.38 questionnaire. Responses to qu Sand uniformity coefficient 4.4 2.3 concerning sampling requiremen filter at top of the photo and the Treatment cost-dollars/ J 000 gal 0.15 0.20 dicated that most of these analy age -sedimentation tank at required only a few times a year. om. facility indicated that ma b- TABLE 6 nese and iron caused operational pro . and fi ltration facilities result in Total coliform analysis at Logan, Utah, research facility lems. However, it was not I onged durations of the filter cycle. whether these metals were remo re shows the distribution offiltra- Effluent Influent Effluent .rate used at slow sand filtration Sampling Temperature Total Coliforms Total Coliforms Design charaderistiCS. Slow s:: iI· Date °c number/ IOO mL number/ l 00 mL tration is characterized by a. I hties. Figure 8 also shows that most nt mai ntain filtration rates of less Oct. 1, 1983 10 50 o simplicity of design compared I 2 Oct. 21 . 1983 8.4 40 1 water treatment . ypl f 1 0.25 m:l/h/m • The Logan, Utah, o y 3 2 Nov. 1, 1983 14.4 30 o slow sand fil tration untts con It F.;t Wa operated at 0.20 m / h/ m . Figure 11. Cost of treatment at oper- Jan. 27, 1984 2.4 12 o Figure 10. Numbers of personnel depth of graded gravel,. called U f : ter cycle durations should be ex- ating slow sand filter plants-dol- Jan. 31. 1984 1.2 10 1 em ployed at opera ti ng slow sand fil ter Feb. 3. 1984 1.1 4 1 media, that separates ftlter sant ftl to have significant effects on the lars/l 000 gal Feb. 29. 1984 2.3 4 1 the underdrains; (2) a depth t IVe cost and ease of operations. plants Of po r sand (which may vary beCause:r h required at the end of ing operations); and (3) a depth 'd iv' fIlter cycle, is a labor-intensive lty above the filter sand that pro I d . It i usually performed with shovels, although a few facilities JouRN L.A. SLEZAK & R.C . SIMS 41 40 MANAGEMENT AND OPERATIONS 1984 have developed custom methods using power tractors. Table 4 shows mean Raw water inlet fi ltration because of excessive tological and influent water quality • Operating slow sand filtration facil- values of reported filter cycle durations f?r each season. Two facilities reported rmr==G------__ expressed. The Logan facility's conditions on the performance of slow ities are serving primarily smaller com- fIlter cycles of one year and one facility tr atment cost, based on a 10- sand filtration facilities. munities with populations of fewer than reported 6-month cycles, except during ign life, was estimated to be The research facility was evaluated 10000 persons. WInter. These three plants are not in- gal. for its effectiveness in removing turbid- • Most slow sand facilities are more cluded in the data summarized in Table ity, coliforms, and particles. The results than 50 years old; however, there has 4 and Figure 9. Figure 9 shows which Supernatant of turbidity analyses from the field-scale been a relatively significant amount of aus small communities do not facility are shown in Figure 13. The slow sand filter construction during the seasons .were reported as yielding the e longest fIlter cycles. Generally, Figure 9 economic resources vertical lines in Figure 13 represent the last 25 years. and Table 4 indicate that winter yields 'anced water treatment processes, It occurrence of filter scraping. The facility • Most slow sand facilities operate at propriate to evaluate slow sand consistently produced water with tur- filtration rates of less than 0.25 m3/h/m2• longer filter cycles, although there are Some of the plants reported as an bidities below the MCL of 1.0 ntu. • The longest filter cycle durations that fIlter cycle durations were controlled hod for such communities. SInce The total coliform analyses for the occur during the winter months at most regular maintenance procedures; i.e., Baflle around -t slow sand installations sn:all Logan research facility are listed in slow sand facilities. perimeter fIlters were scraped on a regular basis to prevent lTluni ties, the propose? Table 6. Coliform removal by the facility • Slow sand facilities are inexpensive short Circuiting regardless of the development of head ustifi d. Most operatIng faclhtles was good, even with a sand media uni- to operate because few personnel are loss. Other facilities allowed filters to J rted successful performance without formity coefficient of 4.4 (Table 5). required and there is only a minimal run until a certain head loss developed 'v chemical treatment other than Particle data were collected during need for chemicals. Filter medium The plants had few compo- one complete filter cycle. Figure 14 shows • The slow sand filtration research or until adequate flows could not be support maintained. The filters that are regularly tsand seemed to be relatively easy to the distribution of particle size in the facility at Logan, Utah, can be considered for the large percentage 'ign. Many of the facilities were from influent and in the effluent of the field- typical of other operating facilities. -to 100 years old, indicating long scale facility. The data shown are aver- o,f faclhtles reporting seasonally equal Acknowledgments ign lives. Most plants treat water at a ages from all of the data collected during fIlter cycles (Figure 9), Some plants Figure 12. Cross section of field-scale facility reported equal winter and summer cycles :tof less than $0.10/ 1000 gal. Assum- the filter cycle. The data indicate that The authors thank Gary S. Logsdon of water use of 378 Li d per capita (100 two-log (99 percen t) red uctions of particle the US Environmental Protection Agency tha,t were longer than the cycles for 100 spring and fall. Inlluent :¢), a cost to the individual user of counts occurred for the smallest sizes office in Cincinnati, Ohio, for his review o Effluent ,... :out $O .O l/d is indicated. Generally, detected and three-log (99.9 percent) to of this manuscript. V. Dean Adams of . Postchlorination for disinfection of 1000000 _ Inll'*ll filter .was practiced by 93 percent surv y did not provide any evidence four-log (99.99 percent) reductions were the Utah Water Research Laboratory c:::::J EHI'*II of the faclhtles studied in the survey. :ainst using slow sand filtration for achieved for larger sizes. Figure 15 shows donated the slow sand filter tanks used Prechlorination was reported at 22 per- 100000 . all communities with appropriate the results of particle counts grouped by in this research, and the secretarial staff rater sou rces. time of day for the 7 -12-,um size range. at the Division of Environmental Engi- cent of plants. Most plants using 10 prechlonnatlOn reported that it served The facili ty at Logan, Utah, is intended The 95 percent confidence intervals for neering and the Utah Water Research 10000 solely as a disinfection measure. One provide long-term operational data on the means are plotted with each of the Laboratory at Utah State University plant, however, cited prechlorination w sa nd filtration under controlled data points. located and established communications nditio s. The facility is discussed for The results of the particle count with the operators of the slow sand oxidation of influent manganese and 1000 Iron and for enhanced removal of 0.02 to 'e purpose of comparing it with oper- analyses show that the field-scale facility filtration facilities discussed in this O.4-,um-sized clay particles. Three plants ing sl w sand filtration facilities, as removed significant quantities of par- article. '!presented by the survey responses. If ticles sized from 2.4 ,urn upward. The reported using copper sulfate as an algae 100 Reference control .measure in prefiltration storage :Ie Logan facility is typical of other size range of 7 -12 ,urn is considered to be rating facilities, more confidence can representative of the size of Giardia Two facilities aerated water 1. HursMAN , L. & WOOD, W.E, Slow Sand prior to fIltration, and one plant adjusted 10 placed in the long-term data collected cysts. Consistent three-log reductions Filtration. World Health Organization, effluen.t pH for distribution system TOm th facility. were observed in this study. It should be Geneva, Switzerland (1974). The facility at Logan consists of two The facility at Logan, Utah noted that many small-sized particles 1n Itration cells, each with a filtration 0.1 1 passed through the filter bed. Currently, ?Id not use chlorine disinfection Feb. Mar. Apr. May June July Aug. eaof9.3 m2• Each cell is an independent It, was a research facility and did not the relationship between particle counts About the authors: Time of Year It ration unit so that controlled experi- and the potential for the presence of discharge to public supplies. However Particle Size Range- I'm Lloyd A. Slezak is an Figure 13. Turbidity data from field- can be conducted. A cross section pathogens is unknown. If the ratio of reduced suspended Figure 14. Average particle size eli - environmental engi- scale facility (vertical lines indicate a filtration cell is shown in Figure 12. inert particles to pathogens remains the materials. tribution in influent and effluent from neer, specializing in durat£on offilter cycle) \n elevated presedimentation tank, with same through filtration, these data indi- . of slow sand filtration. Figure 10 field-scale facility systems, 3,l,h detention time, was provided and cate that significant reduction in patho- Indicates the number of personnel em- 1000000 -±.... Inlluent with Strand Asso- :an be optionally included in the treat- gens can be achieved with slow sand pl,oyed at the slow sand filtration facil- _!'- Effluent ciates Inc., 910 W. 1 t 1 ent fl ow sheet. Logan River water is filtration. ... Wingra Drive, Madi- ities surveyed. These facilities are oper- 1 1 1 1 1 as a raw water source. The Logan able with only a few personnel and low t i $ Filter cycle durations lasted from two .. son, WI53715. While Ver i a watershed conduit for a to six weeks. The shortest filter cycle labor reduce Survey 100000 * * * conducting the research described in this * * 'OUnta in canyon. The filtration facility was associated with an attempt to scrape In general, indicated that article, he was a graduate research assis- located at the mouth of this canyon. a minimum amount of sand to initiate a highly skilled pers0!lnel are not employed tant at the Utah Water Research Labora- able 5 com pares performance and de- new filter cycle. Many other factors that tory in Logan. A member of A WWA and at slow sand fIltratIOn facilities, another parameters in the Logan facility factor contributing to low costs. 10000 were not controllable in the field-scale WPCF, he has' a master's degree from ." thav erages of reported val ues of those facility probably affected the length of Utah State University, Logan, and a The questionnaire requested an as- .l3 rameters collected from the survey. sessment of overall treatment cost per filter cycles. Influent turbidity, algae bachelor's degree in civil engineeringfrom able 5 indicates that the Logan facility growth, and temperature are all expected the University of Wisconsin, Madison. 1000 gallons. Many plants did not re- 1000 i of a typical operating to have an effect on filter cycle duration. Ronald C. Sims is head of the Division of spond to this request because of lack of 2 Figure 15. Number of 7-12-,um-sized .ac1 ltty. T he sand used in this facility is, The limited data collected from the field- Environmental Engineeringat Utah State appropriate records or unwillingness to 22222 2 2 f particles coun ted from field-scale less uniform and of a smaller compIle. a cost estimate. Figure 11 +----$---t---1--$--*--+-- scale facility thus far do not allow University, UMC 82, Logan, UT 84322. study, grouped by time of day (tit. : echve size than those reported in speculation on which of these variables ·summarIzes responses indicating that He specializes in thefields of water supply 100 marks indicate 95 percent confidetltl facilities because the Logan most strongly influences filter cycle and hazardous waste treatment and con- most plants achieve treatment for less E intervals for the means; nu"'befS than $0.10/ 1000 gal. One plant was .. i being used to evaluate the durations. ducts research at the Utah Water Research 8 indicate subsets of means that aff lCacy of using sands of lower than reported to have a treatment cost of Laboratory. Currently, he is working with statistically the same at 0.05 level 0/ quality. This facility will Conclusions $6.50/ 1000 gal, and dissatisfaction with Time of Day private industry and the state of Utah to significance) d info rmation that can be useful in The following conclusions were develop slow sand filtration systems for 42 MANAGEMENT AND OPERATIONS aluating the effects of varying clima- reached as a result of this research: small communities.

DECEMBER 1984 L.A, SLEZAK & R.e. SIMS 43