Deep Bed Filters and High Rate Service

Transcription

1 Deep Bed Filters and High Rate Service R. Rhodes Trussell, Ph.D, P.E. trusselltech.com Ca-Nv AWWA, Sacramento Oct 14, 24

2 Outline 1.Understanding why filtration rate is an important design consideration 2. Understanding the relationship between filter performance, filter rate and filter media design 3. A little bit about the technical constraints to high rate filtration

3 The Bottom Line Deep filters can be operated at substantially higher rates than those customarily used in design The most important risk in a well-operated high rate filter plant is not poor water quality, but difficulties managing recycled flows

4 But first: the importance of chemical conditioning

5 Why the chemistry is important Rapid filtration is accomplished by attachment of the particles to the media, not filtration (i.e. rapid filters do not work by straining or size exclusion) Virtually all particles targeted for removal by filtration are negatively charged and so is the filter media itself Thus the particles and the media are not attracted to each other, in fact they are repelled by each other Thus, for rapid filtration to succeed, the surface chemistry of these target particles must be modified

6 Why make an issue out of the chemistry? Because it s s important to remember that filter rate is not the most important aspect of filter design or operation It s s the chemistry It s s important to keep in mind that no rapid filter will perform well in removing particles if the chemistry is wrong One bad thing about high rate filters is that, when the chemistry is wrong, they produce the same bad water faster

7 Why is filter rate important?

8 Why is filter rate important? The answer is $$$

9 The impact of filter rate on the active filter surface that must be built for a 1 mgd plant 18, 16, Filter area, sf 14, 12, 1, 8, 6, 4, Filter Rate, gpm/sf

12 The impact of filter rate on the active filter surface that must be built for a 1 mgd plant 18, 16, Chicago Filter area, sf 14, 12, 1, 8, 6, 4, Filter Rate, gpm/sf

13 The impact of filter rate on the active filter surface that must be built for a 1 mgd plant Filter area, sf 18, 16, 14, 12, 1, 8, 6, 4, Chicago Most in CA Filter Rate, gpm/sf

14 The impact of filter rate on the active filter surface that must be built for a 1 mgd plant 18, Chicago 16, Filter area, sf 14, 12, 1, 8, Most in CA Prospect, Sydney 6, 4, Filter Rate, gpm/sf

15 The impact of filter rate on the active filter surface that must be built for a 1 mgd plant 18, Chicago 16, Filter area, sf 14, 12, 1, 8, Most in CA Prospect, Sydney 6, 4, Filter Rate, gpm/sf LAAFP

16 The impact of filter rate on the active filter surface that must be built for a 1 mgd plant 18, Chicago 16, Filter area, sf 14, 12, 1, 8, Most in CA Prospect, Sydney 6, LAAFP 4, Filter Rate, gpm/sf Between the rates used at Chicago and Los Angeles, the required filter area changes by more than 3X

17 The impact of filter rate on the active filter surface that must be built for a 1 mgd plant 18, Chicago 16, Filter area, sf 14, 12, 1, 8, Most in CA Prospect, Sydney 6, LAAFP 4, Filter Rate, gpm/sf And filter area is the most expensive part of a conventional water treatment plant

18 The relationship between filter performance, filter rate and filter media design

19 Characterizing a Filter Run NTU Filter run time

20 Characterizing a Filter Run NTU Maturation time, t m

21 Characterizing a Filter Run NTU Maturation time, t m Operating Turbidity, T O

22 Characterizing a Filter Run NTU breakthrough, t b Operating Turbidity, T O Maturation time, t m

23 Characterizing a Filter Run NTU Headloss, inches breakthrough, t b Operating Turbidity, T O Maturation time, t m Run Time, hours

24 Characterizing a Filter Run NTU Headloss, inches ΔH o breakthrough, t b Operating Turbidity, T O Maturation time, t m Run Time, hours

25 Characterizing a Filter Run NTU Headloss, inches breakthrough, t b Operating Turbidity, T O Maturation time, t m design headloss, t h ΔH o Run Time, hours

26 Characterizing a Filter Run NTU Headloss, inches breakthrough, t b Operating Turbidity, T O Maturation time, t m design headloss, t h ΔH o breakthrough, t b Run Time, hours

27 Plot of Mintz Concept Optimum Depth Turbidity Breakthru t b t h Design Headloss Depth of Filter Media

32 Plot of Mintz Concept Optimum Depth Turbidity Breakthru t b Region of Possible Operation t h Design Headloss Depth of Filter Media

33 Let s s look at some real data to verify Mintz concept

34 Pilot Data on Owens River Water, et al, 1977; V = 13.5 gpm/sf; ; 1.55 mm anthracite] [Stolarik, et al, 1977; 2 2 Turbidity Breakthru t b, h t b t h t h, h Design Headloss Depth of Media, m

35 Pilot Data on Owens River Water, et al, 1977; V = 13.5 gpm/sf; ; 1.55 mm anthracite] [Stolarik, et al, 1977; 2 2 Turbidity Breakthru t b, h t b t h t h, h Design Headloss Depth of Media, m

36 Pilot Data on Owens River Water [Stolarik,, et al, 1977; V = 13.5 gpm/sf; 1.55 mm anthracite] 2 2 Turbidity Breakthru t b, h t h t Mintz b Optimum t h, h Design Headloss Depth of Media, m 81

37 Pilot Data on Owens River Water [Stolarik,, et al, 1977; V = 13.5 gpm/sf; 1.55 mm anthracite] 2 2 Turbidity Breakthru t b, h t h t Design of b LAAFP t h, h Design Headloss Depth of Media, m 81

38 Pilot Data on Bull Run Water, 1991; 1 gpm/sf,, 1.5 mm anthracite] [Kreft, 1991; Turbidity Breakthru t b, h Depth of Media, in t h, h Design Headloss

39 Pilot Data on Bull Run Water, 1991; 1 gpm/sf,, 1.5 mm anthracite] [Kreft, 1991; Turbidity Breakthru t b, h Depth of Media, in t h, h Design Headloss

40 Pilot Data on Bull Run Water, 1991; 1 gpm/sf,, 1.5 mm anthracite] [Kreft, 1991; Turbidity Breakthru 38 t b, h L opt 123 in Depth of Media, in t h, h Design Headloss

41 Pilot Data on Bull Run Water, 1991; 15 gpm/sf,, 1.5 mm anthracite] [Kreft, 1991; Turbidity Breakthru t b, h t h, h Design Headloss Depth of Media, in.

42 Pilot Data on Bull Run Water, 1991; 15 gpm/sf,, 1.5 mm anthracite] [Kreft, 1991; Turbidity Breakthru t b, h L opt 15 in t h, h Design Headloss Depth of Media, in. 45

43 Pilot Data on Bull Run Water, 1991; 1 & 15 gpm/sf,, 1.5 mm anthracite] [Kreft, 1991; Turbidity Breakthru z t b, h Depth of Media, in t h, h Design Headloss

44 Pilot Data on Bull Run Water, 1991; 1 & 15 gpm/sf,, 1.5 mm anthracite] [Kreft, 1991; Turbidity Breakthru t b, h Increasing the filter rate: z 1) Modestly reduces the time to breakthrough Depth of Media, in t h, h Design Headloss

45 Pilot Data on Bull Run Water, 1991; 1 & 15 gpm/sf,, 1.5 mm anthracite] [Kreft, 1991; Turbidity Breakthru t b, h t h, h Increasing the filter rate: z 1) Modestly reduces Depth the time of Media, to breakthrough in. and 2) Substantially reduces the time to headloss Design Headloss

46 Pilot Data on Bull Run Water, 1991; 1 & 15 gpm/sf,, 1.5 mm anthracite] [Kreft, 1991; Turbidity Breakthru t b, h t h, h Increasing the filter rate: z 1) Modestly reduces Depth the time of Media, to breakthrough in. and 2) Substantially reduces the time to headloss 3) Resulting in a lower optimum depth and shorter runs Design Headloss

47 Pilot Data on Bull Run Water [Kreft,, 1991; 1 & 15 gpm/sf,, 1.5 mm anthracite] Turbidity Breakthru t b, h t h, h But this comparison based on filter run time isn t t really an accurate z portrayal of efficiency because the filter running at 15 gpm/sf Depth produces of Media, 5% in. more water in the same time period. What happens if we compare both filters on the basis of gallons/sf-run?` Design Headloss

48 Filter Run Volume to Brkthru Pilot Data on Bull Run Water, 1991; 1 & 15 gpm/sf [Kreft, 1991; z FRV b gal/sf 7, 6, 5, 4, 3, 2, 1, gpm/sf,, 1.5 mm anthracite] Depth of Media, in. 7, 6, 5, 4, 3, 2, 1, FRV h gal/sf Here are the same data replotted with run time converted to filter run volume (FRV), expressed as gallons/sf. Filter Run Volume to Hdloss

49 Pilot Data on Bull Run Water, 1991; 1 & 15 gpm/sf [Kreft, 1991; gpm/sf,, 1.5 mm anthracite] Filter Run Volume to Brkthru FRV b gal/sf 7, 6, 5, 4, 3, 2, 1, , 6, 5, 4, 3, 2, 1, FRV h gal/sf Filter Run Volume to Hdloss Depth of Media, in. Note that now we do a bit better at 15 gpm/sf (yellow) than we did at 1 gpm/sf (white)

50 Filter Run Volume to Brkthru Pilot Data on Bull Run Water, 1991; 1 & 15 gpm/sf [Kreft, 1991; z FRV b gal/sf 7, 6, 5, 4, 3, 2, 1, gpm/sf,, 1.5 mm anthracite] Depth of Media, in. 7, 6, 5, 4, 3, 2, 1, FRV h gal/sf Note that now we do a bit better at 15 gpm/sf (yellow) than we did at 1 gpm/sf (white) Filter Run Volume to Hdloss

51 So we get more through the filter at higher rates even though the run time is shorter Is there anything we can do to get a longer run time if we want that? Remember the runtime decreases at the higher rate because the time to design headloss decreases

52 Can we do anything about the headloss? The higher headloss at higher rates is almost all a result of increased clean bed headloss Clean bed headloss is very sensitive to media diameter Thus increasing media diameter will result in a longer time to design headloss

53 But increasing media diameter might affect turbidity What is the impact of media diameter and filter rate on effluent turbidity?

54 Data from the Cedar River three diameters, three filter rates.12.1 [L/d = 13 in all cases] ntu d=1. d=1.5 d= Filter rate, gpm/sf Here are data from work done on the Cedar River (the water supply for Seattle) that show the effect of media diameter and filter rate.

55 Data from the Cedar River three diameters, three filter rates.12.1 [L/d = 13 in all cases] ntu d=1. d=1.5 d= Filter rate, gpm/sf Diameters of 1 to 2 mm and rates of 5 to 15 gpm/sf are shown

56 Data from the Cedar River three diameters, three filter rates.12.1 [L/d = 13 in all cases] ntu d=1. d=1.5 d= Filter rate, gpm/sf Conclusion: Increasing either the rate or the media diameter does result in some degradation in effluent turbidity

57 Can we compensate by making the filter media deeper?

58 The Iwasaki Equation suggests it should be pretty easy Ln[C L /C o ] = -λl- Where λ = filter coefficient

59 Turbidity Removal Ln[C L /C o ] Iwasaki: Ln[C L /C o ] =!"L Media Depth, Ft

60 Data Gathered by DWP [Owens River, C o = 11 ntu,, V = 15 gpm/sf, d m = 1.55 mm] Turbidity Removal Ln[C L /C o ] Iwasaki: Ln[C L /C o ] =!"L Media Depth, Ft The data gathered by DWP show an improvement with depth. But the improvement achieved has diminishing returns

61 Let s s look at some other data

62 Data Gathered at Seattle:Effect of Filter Rate and Media Depth [Cedar River, C o =.2 to.3 ntu, monomedia d m = 1.25 mm] Effluent.12 Turbidity.1 ntu Media Depth, in. V=8 V=12 V=16 V=2

63 Data Gathered at Seattle:Effect of Filter Rate and Media Depth [Cedar River, C o =.2 to.3 ntu, monomedia d m = 1.25 mm] Effluent.12 Turbidity.1 ntu Media Depth, in. V=8 V=12 V=16 V=2 Depth does make a difference. At a depth of 12 inches the performance at 2 gpm/sf is about the same as the performance At 8 gpm/sf

64 Data Gathered at Seattle :Effect of Filter Rate and Media Depth [Cedar River, C o =.2 to.3 ntu, monomedia d m = 1.5 mm] Effluent.12 Turbidity.1 ntu Media Depth, in. V=8 V=12 V=16 V=2

65 Data Gathered at Seattle :Effect of Filter Rate and Media Depth Effluent Turbidity ntu [Cedar River, C o =.2 to.3 ntu, dual media d m = 2/1 mm] Media Depth, in. V=8 V=12 V=16 V=2

66 Data Gathered at Seattle :Effect of Filter Rate and Media Depth Effluent Turbidity ntu [Cedar River, C o =.2 to.3 ntu, dual media d m = 2/1 mm] Conclusion: increasing media depth compensates Media Depth, well in. for increases in filter rate V=8 V=12 V=16 V=

67 Data Gathered at Seattle: Comparing all 3 Media at 2 gpm/sf Effluent Turbidity ntu [Cedar River, C o =.2 to.3 ntu,, various media, V = 2 gpm/sf] Focusing on media diameter Media Depth, in mm 1.5 mm 2/1 mm

68 Comparing all 3 Media at 2 gpm/sf: Looking at L/d Ratio [Cedar River, C o =.2 to.3 ntu,, various media, V = 2 gpm/sf] Effluent Turbidity ntu Media L/d Ratio 1.25 mm 1.5 mm 2/1 mm

69 Comparing all 3 Media at 2 gpm/sf: Looking at L/d Ratio [Cedar River, C o =.2 to.3 ntu,, various media, V = 2 gpm/sf] Effluent Turbidity ntu Media L/d Ratio 1.25 mm 1.5 mm 2/1 mm Conclusion: increasing media depth compensates well for increases in media diameter, provided if we keep L/d constant

70 Conclusion from this data Higher rates can be achieved by using deeper filters and larger diameter media

71 Limits of high rate filtration What is the upper limit in media depth? Don t t know. Max. used in full-scale design so far is ~1 in. Max. used in pilot is ~ 17 in. (14 ft.) Above 48 in., recommend both air/water and surface wash What is the upper limit in media size? Again, don t t know Upper limit tested in pilots is 2 mm Required depth increases roughly in proportion to the diameter

72 Limits of high rate filtration What is the upper limit for filtration rate? Depends on media design The highest rate plant in service is LAAFP 6 mgd 13.5 gpm/sf 1,5 mm, 6 in. depth Next highest I ve I worked on is Prospect in Sydney 9 mgd 1 gpm/sf 1.55 mm, 6 in. depth Highest rate in pilot test ~ 4 gpm/sf (LADWP) Several successful tests at 15 to 2 gpm/sf

73 Limits of high rate filtration Success with high rates requires the use of deeper media and often larger diameter media as well To have a reasonable chance of success I recommend both pilot studies and studies with a large-scale prototype. I don t t know what DHS will say before me, but I suspect they ll want to see testing as well This was done in both Los Angeles and Sydney

74 finis

75 Treatment of Bull Run Water O3=1.5, 81=1, ferric 2.5 mg/l UFRV, gal/sf Turbidity, ntu 2 min. flocculation 5 min + In Line In Line