Vous êtes sur la page 1sur 9

Desalination 236 (2009) 8593

Efficiency of mechanical pre-treatment on European MBR plants


W. Schiera*, F.-B. Frechena, St. Fischerb
Department of Sanitary and Environmental Engineering DESEE, University of Kassel, Germany email: wernfried.schier@uni-kassel.de b ZENON Membrane Solutions GEs Water & Process Technologies, contec screens, Hilden, Germany
Received 30 June 2007; revised accepted 7 October 2007
a

Abstract In order to ensure a stable and reliable operation of municipal MBR plants, an enhanced mechanical pretreatment of the raw wastewater is essential. Removal of hair, fibrous material and other contraries which can lead to operational problems at the membrane modules is of particular importance. As the number of full scale applications is continuously increasing, now appropriate investigations can be carried out to determine the removal efficiency of different units (screens, sieves) or different combination of units (multi staged system). First results of such investigations on two German MBR plants were already presented [F.-B. Frechen, W. Schier and M. Wett, Water Practice & Technology, Vol. 1, No. 3, IWA Publishing, 2006, doi: 10.2166/WPT.2006057.]. In 2006, DESEE executed further investigations on another three wwtps in The Netherlands, UK and Germany. This paper summarizes the results, thus updates the information available, discusses the present findings and gives an outlook on future developments. Keywords: MBR process; Mechanical pre-treatment; Sieve units; Braid formation capacity (BFC)

1. Introduction Starting already 50 years ago studies were carried out in the USA, Scandinavia and in Germany to fractionate the organic pollutants in raw wastewater in different size fractions. The results came up with around 35% of the organic
*Corresponding author.

matter appearing with a size >100 mm [1]. Out of this fraction a very small portion with a huge impact are hair and fibres that tend to build braids and that way generate operational problems on membrane modules. A recent study at DESEE gave a daily average load of human hair with no more than 47 mg per person corresponding with a daily loss of human hair of 140250 per person [2].

Presented at the International Membrane Science and Technology Conference, IMSTEC 07, 59 November 2007, Sydney, Australia
0011-9164/09/$ See front matter # 2009 Elsevier B.V. All rights reserved. doi:10.1016/j.desal.0000.00.000

86

W. Schier et al. / Desalination 236 (2009) 8593

Additionally fibrous material from clothing and other things of daily use that might be estimated at least in the same dimension has to be taken into consideration. Braids emerge from hair and fibres and this occurrence depends on the concentration of hair and fibres. The general phenomenon of braid forming is not a matter of the used membrane type even it might be conceded that one type may be less sensitive than another. But considering the membranes as a physical barrier where no suspended solids pass through and hair and fibres anyway, it is pretty obvious, that the concentration of hair and fibres will rise within the system and it is just a question of time when braid forming will start. Thus, the removal of hair and fibres prior to the filtration area is an essential task for all submerged MBR systems. 2. Fundamentals In Germany, screens and sieves are defined taking into consideration different aspects of construction and technical equipment like gap size, construction and gap geometry [3]. Basics of screening and sieving are explained in detail in [4,5]. Today there is a wide spectrum of sieves available, applied on European MBR plants and equipped with different gap geometries slit (one-dimensional geometry), hole and mesh (two-dimensional geometries) covering gap sizes ranging from 0.5 to 3.0 mm. Investigations about sieve efficiency concerning the impact of the above mentioned differentiating factors or especially the gap arrangements (horizontal or vertical slit) in the past seldom were conducted, but actually for a short time, since mechanical pre-treatment gets increasing importance, some appropriate projects were carried out [2, 4, 57]. Consolidated findings regarding the removal performance of single units or combination of units can be determined already by short-time investigations.

In contrast, the question whether different membrane systems require different pre-treatment efficiencies, is still in discussion. The verification of suitability of different pre-treatment systems only can be acquired by carrying out long-term investigations lasting for some years. 3. Mechanical pre-treatment on MBR plants current status Considering only wwtps designed for more than 1,000 p.e., in Europe around 150 municipal MBR plants are in use by the end of 2007, serving more than 600,000 m3/d. Table 1 gives an overview about some selected municipal MBR plants in Europe and its pre-treatment systems and that way illustrates the wide range of applications. Stage 1 in most cases employs screens while stage 2 represents an advanced mechanical treatment unit (sieve). 4. Methods Former DESEE investigations determining the removal efficiency of sieves with primary regard to SS and COD took place at German wwtps in Markranstadt and KaarstNordkanal [4]. Sampling was done by applying filtration sacs to get larger sample volumes (around 350 L/d) compared with usual sampling volumes of just 1 or 2 L/day. It is a manual sampling procedure repeated every 1520 min and lasting for 8 h a day. The sampling method and associated results are explained and discussed in detail in [4,5]. In 2006, DESEE carried out further investigations in the exact same manner at the wwtps Varsseveld/The Netherlands, Swanage/UK and Monheim/Germany. Thus, this project covered as well one-staged and two-staged pre-treatment systems as MBR plants equipped with flat sheet membranes and hollow fibre membranes. Based on the former at wwtp Monheim the operation of just two sampling stations was sufficient while

W. Schier et al. / Desalination 236 (2009) 8593


Table 1 Pre-treatment systems of selected European MBR plants wwtp p.e. com Membrane type Gap geometry/size (mm) Stage 1 Germany Rodingen Monheim Kaarst-Nordkanal Waldmossingen Seelscheid Woffelsbach Konzen Bergheim-Glessen Porlock/UK Swanage/UK Buxton/UK Ballyclare/UK Brescia/I Schilde/B Guethary/F Rietliau/CH Jinamar/E Arenas de Iguna/E Varsseveld/NL Heenvliet/NL Ootmarsum/NL Oroszlany/HU 3,000 9,700 80,000 2,600 11,500 6,200 9,700 9,000 3,000 23,000 30,000 30,000 46,000 10,000 10,000 22,000 40,000 20,000 23,000 3,330 7,000 40,000 1999 2003 2004 2004 2004 2005 2006 2007 1998 2000 2004 2005 2002 2003 2003 2005 2004 2006 2005 2006 2007 2004 HF HF HF HF FS FS FS HF FS FS HF HF HF HF HF HF HF FS HF FS Tub, ex HF Slit, Slit, Slit, Slit, Slit, Slit, h 3.00 v 5.00 v 5.00 v 3.00 v 3.00 v 6.00 Stage 2

87

Slit, h 0.50 Slit, h 1.00 Mesh 1.00 Slit, h 0.50 Slit, v 3.00 Slit, v 0.50 Slit, v 0.50 Mesh 1.00 Hole 3.00 Hole 2.00 Hole 3.00 Hole 1.00 Hole 2.00 Mesh 1.00 ?, 1.00 Mesh 0.75 Hole 0.80 Hole, 3.00 Hole 0.80 Hole 3.00 ?, 0.75 Hole 1.00

Europe

Hole 6.00 Hole 6.00 Hole 6.00 Slit, h 3.00 Slit, v 2.00 Slit, v 6.00 Slit, v 6.00 Slit, v 5.00 Slit, v 6.00 Slit, v 6.00 Hole 5.00

HF: Hollow fibre; FS: Flat sheet; Tub, ex: Tubular, external; Slit, h/v: Horizontal/vertical.

at wwtp Varsseveld and wwtp Swanage a third one in front of the screen units (S1 in Fig. 1) was necessary to compare the influent qualities of all investigated wwtps. It has to be mentioned that each measurement lasted just for 2 weeks which does not enable any conclusions about the long time effect concerning MBR performance. In the following section the investigated sites, in particular the sieving areas, are introduced. The pre-treatment installations and the membrane equipment of the observed wwtps are compiled in Table 2. Fig. 1 covers flow schemes, sampling points and analysed parameters. Residues of screens and sieves were not analysed. Fig. 2 adds functional principles.

5. Results and discussion Table 3 presents the statistical results. Measurements were intended to be conducted at dry weather flow and only suchlike data were taken into account concerning data evaluation. The last two lines of each table describe either the removal efficiency of the entire pre-treatment system (rem. eff. S1S3) or the removal efficiency of the respective sieve unit (rem. eff. S2S3 and rem. eff. S1S2 respectively). This description cannot be applied to the Varsseveld plant due to operational circumstances explained further below as well as is not possible for the Monheim plant. Reason here

88

W. Schier et al. / Desalination 236 (2009) 8593


Grit chamber/ grease trap

Varsseveld

Screen
Slit,v 5 mm Slit, v 5 mm

Sieve
Hole 0.8 mm hole 0.8 mm

MBR

FC

Swanage

Screen
hole 6 mm Hole 6 mm

Grit chamber/ grease trap

Sieve
Hole 2 2 mm hole mm

MBR / FC
S3

Monheim

Sieve
Slit,h 1 mm Slit, h 1 mm

Grit chamber/ grease trap

MBR

FC

S1

S2

Fig. 1. Sampling positions and analysed parameters at the investigated sites. Table 2 Pre-treatment and membrane installations at the investigated sites wwtp [MLD] Pre-treatment Stage 1 Varsseveld/NL Swanage/UK Monheim/GER 24.2 12.7 6.9 Slit, v/5 mm/step screen Hole/6 mm/band screen Slit, h/1 mm/rotating drum sieve gc gc gc Stage 2 Hole/0.8 mm/drum sieve Hole/2 mm/band sieve Membranes System/supplier HF/Zenon FS/Kubota HF/Zenon (m2) 20,160 15,840 12,320

Slit, h/v: Horizontal/vertical; gc: Grit chamber; HF: Hollow fibre; FS: Flat sheet.

was that a sampling point downstream the grit chamber was not part of the sampling concept. At Varsseveld and Swanage each sieve device is positioned behind a screen and a grit chamber so that the pollutant content of the wastewater is already decreased twice. This differs from Monheim facility where the influent for the sieve

(S1) represents the influent quality of the entire plant. However, it can be noticed that influent concentrations at Monheim are rather low compared with those of Varsseveld and Swanage. This applies to all analysed parameters. On the other hand the influent quality of the Varsseveld and the Swanage wastewater (S1) is

Fig. 2. Functional principles of sieve units (left: Varsseveld, middle: Swanage; right: Monheim).

W. Schier et al. / Desalination 236 (2009) 8593


Table 3 Statistical data Data SS S1 Varsseveld/NL: drum Quantity Min (mg/L) Mean (mg/L) cv (%) Max (mg/L) rem. eff. S1S3 rem. eff. S2S3 (%) S2 S3 COD S1 S2 9 688 1.126 36 1.820 S3 TN S1 S2 9 60.8 71.5 6 76.5 S3 TP S1 S2 8 12.60 15.69 12 18.10 S3

89

sieve (hole; 0.80 mm) 9 217 270 197 760 273 333 266 825 10 11 14 7 301 398 316 921 20

579 746 10 846 27

75.5 82.2 5 88.7

61.5 68.7 6 73.4 4

13.60 15.24 6 16.60

12.60 15.03 10 17.20 4

Swanage/U.K.: band sieve (hole; 2.00 mm) Quantity 8 Min (mg/L) 439 380 371 Mean (mg/L) 704 568 524 cv (%) 23 23 21 Max (mg/L) 977 802 661 rem. eff. S1S3 (%) 23 rem. eff. S2S3 (%) 7

725 925 19 1.260

6 568 840 22 1.172

543 805 21 1.107 13 4

65.9 84.6 12 98.6

7 68.2 80.1 9 91.2

65.0 77.4 9 88.1 8 3

10.90 15.35 24 22.80

6 11.50 13.93 19 19.50

11.40 13.48 17 18.30 11 3

Monheim/GER: drum sieve (slit, h; 1.00 mm) Quantity 7 Min (mg/L) 174 141 509 Mean (mg/L) 486 312 619 cv (%) 39 27 14 Max (mg/L) 829 426 766 rem. eff. S1S3 (%) 32

7 471 564 14 678 9

39.0 58.4 21 74.8

7 39.1 56.9 22 74.3 3

6.16 8.95 24 12.90

7 6.04 8.47 22 11.70 5

approximately at the same level with the exception of parameter SS. Influent concentrations of SS in Swanage are at least twice as high as in Varsseveld and also clearly higher than in Monheim. This might be caused by Swanage coastal location in Southern UK with lots of sandy beaches and sandy soil entering the sewer system. The circumstances require a very accurate and careful grit chamber design assuming that a 2 mm sieve device downstream the grit chamber will not remove many of those sandy matters that may pass the grit chamber. The latter could be observed during the sampling by the way of applying filtration sacs.

Another important aspect concerns the operation concept regarding the chemical cleaning of the membrane moduls at the Varsseveld plant. There a chemical maintenance clean on air is performed regularly. For these purposes filtration chambers are emptied and the sewage of the filtration chambers is discharged into the effluent area of the grit chamber. Thus, the sewage is sieved before entering the biology to get out rests of suspended solids like hair and fibres that may badly influence the filtration process. This operation mode has the consequence that the wastewater load treated in the sieves is definitely larger than the influent load to the screens

90
Table 4 Removal efficiencies wwtp Pre-treatment Stage 1 Sieve unit Varsseveld/NL Swanage/UK Monheim/GER

W. Schier et al. / Desalination 236 (2009) 8593

Parameter Stage 2 gc gc gc Hole/0.8 mm/drum sieve Hole/2 mm/band sieve SS 20% 7% 32% COD 27% 4% 9% TN 4% 3% 3% TP 4% 3% 5%

slit, v/5 mm/step screen Hole/6 mm/band screen Slit, h/1 mm/rotating drum sieve

Entire pre-treatment system Swanage/UK Hole/6 mm/band screen Slit, h/v: Horizontal/vertical; gc: Grit chamber.

gc

Hole/2 mm/band sieve

23%

13%

8%

11%

and therefore the removal efficiency of the entire pre-treatment system can not be determined. The effect can be easily seen by comparing mean data concerning SS and COD between S1 and S2 in Table 3. Table 4 compiles the removal efficiencies either of the investigated single sieve units or of the entire pre-treatment system. The respective pre-treatment units contributing to the removal data in the same line are highlighted. Based on Tables 3 and 4, the following statements and comments can be given: Effect of gap size: hole with 0.80 mm gap size versus hole with 2 mm gap size. Table 5 is an excerpt from Table 4 considering only SS removal of the two second-stage sieve units with hole geometry in Varsseveld and Swanage. It is not surprising that the unit with the smaller gap size is more effective than the one with the bigger gaps. Additionally it has to be kept in mind, that the SS load in Swanage (gap size of 2 mm) was almost two times higher than in Varsseveld, the poor efficiency is not due to the fact that efficiency rate might be affected by low load. General remark including all aspects like gap size, gap geometry, influent qualities and

Table 5 Hole with 0.80 mm gap size versus hole with 2 mm gap size wwtp Varsseveld Swanage Sieve unit Hole (0.80 mm) Hole (2 mm) SS 20% 7%

the positioning within the pre-treatment system. It was reported earlier [2] that a mesh sieve with a gap size of 0.75 mm was observed removing SS in a magnitude roughly twice as high compared with a horizontal slit sieve of the same gap size. Hence, it was concluded, that, comparing equal gap sizes, two-dimensional sieve units are more effective than one-dimensional sieve units. Today this seems to be common knowledge. Against this background the presented investigation carries out a contrary fact (Table 6) worth to be discussed more detailed. Positioning: SS > 5 mm are removed in Varsseveld by the screen unit and that way downgrade the numerical efficiency of the subsequent sieve unit. In Monheim the first treatment unit is the sieve itself and hence removes both SS > 5 mm as well as SS < 5 mm which deeply influences the mass balances. There it

W. Schier et al. / Desalination 236 (2009) 8593


Table 6 SS removal of selected sieve units wwtp Varsseveld Sieve unit Hole (0.80 mm) SS 20% Comment

91

Monheim

Slit, h (1 mm)

32%

Determination includes recycled sludge; removal efficiencies of screen and grit chamber are unconsidered Untreated raw water

can assume, that also a sieve unit like in Varsseveld would yield larger numerical removal effiencies when operated as the first treatment unit. Sieve influent quality: The Varsseveld sieve is by comparison with the Monheim sieve on the one side polluted with clearly less suspended solids native of raw water. On the other side the numerical efficiency is additionally influenced by recycling biological sludge out of the filtration chambers to the upstream side of the sieve unit. Out of this sludge only contraries are removed but not the vast majority of biological flocs which exist clearly smaller than 0.8 mm. But this fraction is mathematically taken into account both on the upstream and on the downstream side. Thus, it has to be reasoned that comparing sieve units needs to consider the different conditions of sieve influent qualities. View on the entire pre-treatment systems concerning parameter SS. Besides the efficiencies of single units, in general of course it is necessary to regard the overall efficiency of the wastewater pre-treatment. Table 7 shows the removal efficiencies of entire pre-treatment systems concerning SS. This determination could only be carried out in Swanage. The appropriate sieve unit showed the lowest numerical removal efficiency compared with the other two sites. But like in Varsseveld also in Swanage the removal percentage of the sieve unit can be expected on a higher level if the influent wouldnt contain a rather large fraction of fine sands that pass both the grit chamber and the sieve respectively, which can

Table 7 SS removal of entire pre-treatment systems wwtp SS Measured Varsseveld Swanage Monheim ? ? 20% 23% 32% ? 40% Estimated 4050% SS influent load low High sandy content SS influent load high Comment

be seen from the fact that SS influent concentration in Swanage is about 1.5 times higher than in Monheim and about 3 times higher than in Varsseveld.. Then removal efficiency would increase beyond 23% in view of braid forming pollutants. One way to improve operating conditions in Swanage is to reduce the gap size of the sieve unit; more promising is an upgrade of the existing grit chamber. In Varsseveld the evaluation of the pretreatment effectiveness is affected by the addition of sludge from the emptied filtration chambers. It can be estimated that, without sludge sieving and calculating the efficiencies of a 5 mm step screen and a standard grit chamber like measured earlier on other wwtps, SS removal in the range of 4050% would be likely. In Monheim, the effectiveness of the grit chamber on the SS content was not measured. Total SS removal can also be assumed to end up in the same range of around 40%.

92

W. Schier et al. / Desalination 236 (2009) 8593

In general, also including experiences from other plants, it can be stated that concerning necessary removal for braid formation prevention, there is no significant difference between hollow fibre and flat sheet systems, if there is a difference at all. General remark on COD, nitrogen and phosphorus removal. Expecting SS removal of around 50% at maximum and considering data of Table 3 and results of former investigations it can be noted that the removal efficiency of an entire one or two staged pre-treatment system covering sieve units does not exceed the removal rates of conventional mechanical treatment systems covering a primary sedimentation. COD removal does not badly affect subsequent denitrification. Biological treatment is slightly relieved by removal of nitrogen and phosphorus of around 10% at maximum. 6. Conclusions Mechanical pre-treatment of wastewater is essential and crucial for the MBR process. There is still discussion ongoing on how to design the optimal pre-treatment system. This is mainly due to the fact that still today there is relatively poor knowledge about the ability of different pre-treatment units mainly sieves and entire pre-treatment systems to modify the raw wastewater in a way that avoids braid formation at the membrane modules. The results presented confirm that, applying smaller gap sizes, increasing removal efficiencies concerning SS are measured. Moreover, in view of future applications it can be expected that more and more two-dimensional gap geometries will be employed and with this, removal performance of sieve units will increase. The above described investigations are another step on the way to close this lack of knowledge, and they promote a view that regards not only nominal gap sizes but also takes into account aspects of one-dimensional and

two-dimensional gap geometries, one-stage or two-stage pre-treatment and removal efficiencies of single units and entire pre-treatment systems.

Outlook It must be stated that analysing SS is not properly the focus but a temporal solution. The real objective is to determine the amount of hair and fibrous material, which are responsible for braid forming, and its removal by mechanical pre-treatment. In this context, further investigations at DESEE are underway to quantify specifically the amount of hair and fibrous material and other unfavourable contents. Very important, our batch tests so far have revealed that braid forming occurs rather quickly under any turbulent surrounding conditions (like the biological zone) even without any installations (like membrane moduls). Braid forming as a phenomenon thus basically has nothing to do with membranes. It depends on the length of hair and fibres and the energy density of the mixing environment. DESEE now is about to define a suitable method to measure the braid formation capacity BFC of wastewater that would allow to quantify the efficiency of different pre-treatment systems really result oriented, that is to say regarding braid formation. From this it can be concluded that a promising concept can be established by dividing the mechanical treatment concept into a conventional pre-treatment area including screen and grit chamber/grease trap and a post-treatment area consisting of a sieve unit (Fig. 3). It can be assumed that this concept supersedes the discussion of always smaller gap sizes. With this concept, braid will be formed within the biology, attracting single hair and fibres, and will easily be removed by the sieve, thus assuring a trouble-free filtration process. Moreover, it

W. Schier et al. / Desalination 236 (2009) 8593

93

[3]

[4] Fig. 3. Future design of mechanical treatment.

is one more argument to vote for the design of separate filtration chambers.

[5]

References
[1] H. degaard, Optimised particle separation in the primary step of wastewater treatment. Water Science and Technology, 37 (10) (1998) 4353. [2] F.-B. Frechen, W. Schier and C. Linden, Pretreatment of Municipal MBR Applications, 4th IWA International Membranes Conference 2007

[6]

[7]

Harrogate/UK, Conference Proceedings, ISBN 978-1-86194-127-5. German Standard DIN 19569-2:2002-12, Wastewater treatment plants Principles for the design of structures and technical equipment Part 2: specific principles for the equipment for separating and thickening of solids, DIN Deutsches Institut fur Normung e.V., Beuth Verlag, Berlin, (2002). F.-B. Frechen, W. Schier and M. Wett, Pre-treatment of Municipal MBR Applications in Germany Current Status and Treatment Efficiency Water Practice & Technology, Vol. 1, No. 3, IWA Publishing, 2006, doi: 10.2166/WPT.2006057. W. Schier, Mechanische Abwasservorbehandlung bei kommunalen Membranbelebungsanlagen; 4. DWA/ DVGW-Membrantage, 2006, Kassel, Germany. STOWA (2002) MBR for Municipal Wastewater Treatment, Supplementary report with side studies, Hageman Fulfilment, Zwijndrecht, ISBN 90.5773.167.3. B. Rusten, H. degaard, Evaluation and testing of fine mesh sieve technologies for primary treatment of municipal wastewater, Water Sci. Technol., 54 (10) (2006) 3138.

Vous aimerez peut-être aussi