- Email: [email protected]
Interpreting the rating of cartridge filters
TRIDGE FILTER RATING
Interpreting the Rating of Cartridge Filters By J. M. Bentley and R J. Lloyd Department of Chemical Engineering, Loughborough University of Technology, Ashby Road, Loughborough LEll3TU, UK. This paper was presented at the Filtration Society meeting on Cartridge Filtration, Birmingham, 17 March 1992.
UK,
Cartridge filtration of water is used in many industries. The required purity of water being filtered is often extremely high and the presence of even very low concentrations of fine particles can have a potentially devastating effect on the process for which the water is needed. In such situations it is important for the user to be aware of how the available filters will perform. At present cartridge filters are available with ratings of less than 0.1 pm but there is some confusion as to what these ratings represent. In this paper the rating of fllters Is considered and the inadequacies of the ratings discussed with reference to recent studies. One of the most poorly understood terms in cartridge filtration is the filter rati@). The confusion arises partly because there are absolute and nominal ratings in use. An absolute filter is one which is said to remove virtually all particles (> 99.999999990 /o) larger than the rated pore size(2,3).These tend to be screen filters which remove particles whose dimensions are too large for them to pass through the opening of the filter. A nominally rated filter is one for which it is not possible to assign an absolute rating on the basis of pore size because the pore size is not well defined. Depth filters fall into this category. A nominally rated filter may be said to remove 90-950~ of particles larger than the nominal rating’2). The distinction between the two types of filter is based on how accurately the pore size distribution of the filter media can be defined and measured rather than on the retention efficiency for given particle sizes. As the nominal ratings do not make any specific claim regarding filter performance it is not appropriate to devote too much time to an analysis of such ratings; they simply serve as a guide to the user. However, the extent to which nominal ratings can be used as a guide is limited(4). On the other hand, absolute rated filters are often used in situations where complete removal of all particles above a certain size is desired. Examples are the pharmaceutical industry where organisms must be removed to allow the liquid stream to be cold sterilized, and the semiconductor industry where particles one tenth of the minimum device feature size are thought to be critical to the process(5). It is therefore important to know whether absolute rated filters perform according to their rating. Absolute rating of cartridge filters Absolute ratings of membrane filters were developed according to the requirements of the pharmaceutical industryc6). Membrane filters were rated according to their ability to retain bacteria. Filters which produced sterile effluent when challenged with lo7 Pseudomonas diminuta per square centimetre of filter surface area under standard conditions were considered absolute rated sterilizing ftiters. These filters were accepted as 0.2~ absolute rated, although it is known that the organism is larger than 0.2 pm. Although this method of rating filters is absolute with respect to removal of the Filtration&Separation
July/August
i 992
Pseudomonas diminuta organism, it is clearlyunacceptable as a method of rating a filter as absolute for 0.2 F particles”). To determine by non-destructive means the retention level of every filter manufactured, the bubble point integrity test is used. The pressure required to overcome the capillary forces in the pores of a fully wetted membrane is inversely proportional to the pore diameter. When pressure is applied to the pores of a wetted membrane the liquid from the largest pores will evacuate first. The pressure at which the capillary forces are first overcome is the bubble point value, and this indicates the size of the largest pore. If the bubble point value is known for a membrane which is known to be absolute for Pseudomonas diminuta, then other membranes with the same bubble point value may be assigned the same rating. However, when such a membrane is assigned a rating of 0.2~ and, even worse, when a membrane with a bubble point value twice that of the membrane absolute for Pseudomonas diminuta is assigned a rating of 0.1 pm, it is not possible to say that the absolute rating is equal to the diameter of the largest pore. Discussion of recent work In their work concerned with the rating of membrane filters Roberts et al@) challenged membranes with latex spheres under conditions designed to maximise penetration - by adding surfactant. They found that the efflciency of several membranes at the rated size was considerably less than 100%. They quote results for a double layer of a 0.1 pm rated membrane showing a retention efficiency of 33% with 0.106 pm latex spheres, less than 80% with 0.215pm spheres and 100% with 0.222~ spheres. In fact, of the three 0.1 pm rated membranes tested none retained as much as 10% of the challenge with 0.110 pm spheres when a single layer of membrane was used. Whilst these results do not represent the performance of such membranes in ultra pure water, they do indicate the presence of pores in the membrane which are significantly larger than the rated size. Recent work by Grant(” has shown that the retention efficiency of membrane filters varies with filter loading. A 0.45 m rated membrane filter was found to have an 333
TRIDGE FILTER RATING
I
Retention
efficiency (“I)
Similar results have been noticed by the authors in their work on the transient nature of cartridge filtration(7*s).One filter with a nominal rating of 0.45 ltrn was challenged with a range of particle sizes. In the event of impulses in the upstream concentration it was found that there was a temporary increase in penetration for particles in the range 0.46pm to 0.65 pm. A typical response for the filter is shown in Figs. 2 to 4. The response has been described in terms of a temporary blocking of smaller pores in the event of the very
LRV
90
1
99
2
99.9
3
99.99
4 5
99.999 I
1
Table 1 Relationship between retention effkiency tion value (LpV).
I
and log reduc-
initial log reduction value (LRV)+ of 2 when challenged with 0.442 pm particles, falling to less than 1 as the loading on the filter was increased. This was carried out under conditions where sieving was the dominant capture mechanism. With 0.527 pm particles the same filter had an initial LRV of 4.5 falling to 2 with loading. There were even measurable concentrations of particles as large as 0.66 l.unpenetrating the filter. Figure 1 shows the kind of response reported by Grant for different particle sizes on a 0.45 pm rated filter. Table 1 shows the relationship between LRV and retention efficiency.
3
12
g
4
5
+\f+(_._.______ _....... _.__._..._..._ .......__-_.-_____._---_._._i
6
7
6
Q
10
11
12
13
Time (min)
Fig. 2 Impulse in upstream concentration. loo-
---1.-._-.-._..___ -_._-_.________ __.___-_. _____--_
....__._
.________ ,__..__-
11. 0
o.oo@s 0.601 o.m15
o.m?
o.ce5
0.003 o.oca
o.OOl 0.0645
I 0.005
-_.._.._
Pore volumes of particles removed Fig. 1. Typical responses of a membrane dlfferent particle sizes (fromGrant@)).
filter challenged
with
There are two points of interest regarding filter ratings. Firstly, particles significantly larger than the rating of the filter are allowed to pass in measurable quantities, showing that there are pores larger than the rated size. In fact, for particles of similar size to the filter rating the retention efficiency is between 99% and !39.9%, well below what would be expected if there were no pores larger than the rated size. Secondly, as the loading on the filter increases the efficiency falls. A model has been developed by Grant describing the change in efficiency with loading. As the loading on the filter increases the smaller pores become blocked, resulting in a redistribution of the flow through larger pores for which the capture efficiency is less. The model and the experimental data are in reasonable agreement. The fall in efficiency is noticed for particles as large as 0.66 w challenging a 0.45 pm rated membrane filter, showing that there are pores significantly larger than the rating. +The log reduction value is given by: LRV = log,,(CJCo), where Ci is the concentration of particles at the filter inlet and Co is the concentration of particles at the filter outlet. 334
m
_.______._
. . . .
ok--?-!. 12345676
’
9
10
’
11
.
.’
12
13
..' 11 12
13
Time (min) Fig. 3. Downstream
3-..
123466769
.
response to upstream impulse.
-.
.
.
.
.
. 10
Time (min) Fig. 4. Penetration
in the event of an upstream Impulse. July/August 1992
Flltratlon & Ssparatlon
TRIDGE FILTER RATING high concentration impulse, with a resultant redistribution of flow through the larger pores. It would be expected that a nominally rated filter would have a wide size distribution with pores significantly larger than the rated size, but the same effect was observed for two membrane filters which were absolute rated 0.4 urn and 0.45 pm. Both membrane filters showed increased penetration at high challenge concentration with 0.46 pm particles, suggesting that there are pores larger than this in the filter media. The above work was carried out using 18 Mf&cm water at a constant flow rate of 10 Urnin. Conclusions
There are inadequacies in the ways in which filters are assigned absolute ratings. Despite the fact that the absolute rating is intended to represent the largest pore size in the filter media, work by three independent groups described in this paper shows that there are pores significantly larger than the rated size in absolute rated filters. It is clear that some changes to the rating system are required, either to measure the pore size more accurately or to base rating on penetration of known sized particles under standard conditions. If pore size is to be used as the basis of a rating system this will not represent the performance of the filter but will be a measure of a physical property of the filter which will indicate the size of particle which is too large to penetrate under any conditions unless the pores are
Filtration 8 Separation
July/August 1992
enlarged by some means such as pulsating flows. Such a system would give rise to a worst case rating. If efficiency under standard conditions is to be used then the user must be aware that under different conditions the filter performance may not match the rating. Either way, some standardization is needed to make the ratings more meaningful. References 1. Sandstedt, H. N. and Weisenberger, J. J. Cartridge Filter Performance and Micron Ratinn. Filtration and Senaration. “’ March/April 1985, p 101. to Filtration System Sizing, Millipore 2. An Introduction Technical Brief. 3. Grant, D. C. and Zahka, J. G. Sieving Capture of Particles by Micronorous Membrane Filters from Clean Liauids. Swiss Contdminution Control, 3 (4a), 1990, ~160. (Presented at ICCCS 1990 Symposium). Filter Performance: Myth and Reality, Practical 4. Juhasz,C. Aspects of Filter Testing, Proc World Filtmtion Congress (IZZ), 1982, p 596. 5. Plotnik, I. Metrology Applied to X-ray Lithography, Solid State Technology, January 1989, p 102. 6. Roberts, K. L., Velazquez, D. J. and Stofer, D. M. Challenging the Membrane Filter Ratings, Semiconductor Znternutionul, November 1990. Mechanisms of Penetration in Cartridge I. Bentley, J.M. Filtration of De-Ionised Water, Ph.D Thesis, LUT, 1991. 8. Bentley, J. M. Dynamic Studies in Membrane Filtration of Ultra-Pure Water, Z.Chem.E. Particle Technology Research Award Meeting, 1992.
335
@TRACTSOFREFEREEDPAPERS Recent Developments in OII Filter Test Methods Neuentwicklungen bel Olfllter-Testmethoden van B. M. Verdegan, K. McBroom u n d L. L i e bm ann Standardisierte Filtertestmethoden erleichtern den Vergleich verschiedener Filterprodukte. Der Filtertest mit mehreren Durchgangen stellt die Grundlage f~r zahlreiche (~lfilter-Testnormen dar, dessen Reproduzierbarkeit dutch bessere Spezifikation der Testbedingungen und Verbesserungen in den Partikelz~hlmethoden /aufend verbesserf worden /st. Partikelzahlung /st h~Jufig darer verantwortlich, dab die Ergebnisse yon e/nero Labor zum anderen variieren. Aufgrunddessen wird seit kurzem in den USA eine Kalibrierungsmethode zur Partikelzahlung, ANSII(NFPA) T2.9.6R1-1990, angewendet. Mit H/Ire der neuen Methode /st es mOglich, international anerkannte Definitionen des
Meters anzuwenden, die bei der a/ten Methode der AC Feinteststaubkalibrierung nicht zuganglich waren. Die PartikelgrOf~endefinitionen sind jedoch bei qten beiden Methoden unterschiedfich. Die bestehenden Normen sind geringfi)gigen Anderungen zu unterziehen, um sie der neuen Kafibrierungsmethode anzugleichen. Andere M(~glichkeiten der Verbesserung yon Reproduzierbarkeit und Vermittlung von Testergebnissen sind ebenfalls erSrtert. (5 sn., 4 abb., 15 ref)
D6veloppements r(~cents dens les m~thodes d'essels des flltres b hulle p a r B. M. Verdegan, K McBroom et L. L i e bm ann Des methodes d'essais standard/sees de filtres facilitent la comparaison de differents appareils. Le test multipass constitue la base pour plusieurs standards de filtres a huile et sa reproductibilite a (~t~ continOment amdlioree par de meilleures spdcifications des conditions d'essai et par un raffinement des methodes de comptage des particules. Ce dernier est use cause importante de variabilit~ dans les essais inter-laboratoires. Dans ce contexte, ANSII(NFPA) T2.g.6RI-1990, une m~thode de calibration du comptage utilisant des spheres de latex en suspension dans I'huile, a etd adoptc~e par les USA.
Cette methode permet de s'adapter aux definitions du metre accept(~es au plan international mais que I'on ne peut atteindre par la vie/lie m~,thode de la calibre t/on AC avec fine poussii~re. Cependant, les definitions de la granulom~.trie des deux mdthodes different. Des modifications mineures aux standards existants s'imposent pour les harmon/set avec la nouvelle methode de calibration. On discute ~galement d'autres moyens d'ameliorer la reproductibilitd et la presentation des rdsultats d'essaL (Sp., 4 fig., 15ref.)
Desarrollos corrlentes en los re&fades de probsr flltros pars acelte p a r B. M. Verdegan, If. McBroom y L. L i e b m a n n M~todos estdndares de prober filtros facilitan comparaci(~n de distintos productos de filtro. La prueba mu/tipaso de filtro es /a base de muchos estdndares de prueba de fi/tros pars ace/re, y su reproducibilidad se ha rnejorado por major especificaciOn de la manera de hacer /as pruebas y mejores m~tods de contar particulas. Contar particulas es la mayor fuente de variabilidad entre laboratorios. Por repuesta, recientemente se ha aprobado en los EE UU el m6todo de calibraciSn de confer particu/as ANSII (NFPA) T2.9.6RI-1990, m~todo qua emp/ea esferas de latex suspend/des en aceite. El nuevo
~
~P
m~todo demuestra relacibn con definiciones aprobadas internacionalmente de/metro, que no ocurre con el viejo mdtodo AC de calibraci~n con po/vo fins de prueba. Sin embargo,/as clef/sic/ones de tamer, s de particulas son distintas con los dos m~todos. Pequer}as modificaciones son necesarias con estdndares cordentes pars arrnonizar/es con el nuevo mdtodo de calibraciOn. Se discuten otras maneras de mejorar reproducibilidad y presenter resultados de prueba. (5 p~gs., 4 figs., 15 rets.)
Interpreting lhe Rating of Cartridge Filters Interpretation der Nennlelstung van Kerzenflltarn v a n J. M. Be nt l e y u n d P. J. Lloyd
Die Kerzenfiltration von Wasser /st in zahlreichen Industriezweigen Oblich. Die erforderliche Reinheit des zu filtrierenden Wassers ist oft aufJerordentlich hoch, und des Vorhandensein selbst sehr geringer Konzentrationen feiner Partikel kann u . U . verheerende Auswirkungen auf des Verfahren ausOben, fbr des des Wasser benOtigt wird. In derartigen Situationen ist es wichtig, dab der Benutzer ~ber die Leistungsdaten der verfOgbaren Filter genau informiert /st. Zur Zeit gibt es Kerzenfilter mit Nennleistungen yon wen/get sis 0,1 I~m: es /st ledoch oft nicht ganz klar, was diese
Werte eigentlich beinhalten. Im vorliegenden Referat wird die Festsetzung von Filternennleistungswerten dargelegt und die Unzul~nglichkeiten der Nennleistungsbestimmung unter Bezugnahme auf neue Untersuchungen erOrtert, (3 sn., 4 ebb., I tab., 8 ref)
Interpretation des r~sultats des flltres b cartouches p a r J. M. Bentley et P. J. Lloyd La filtration d'eau sur cartouche est utilisee dans beaucoup d'industries. La puret~ requise de reau filtr(~e est sourest tr~s ~lew~e et la pr6sence de concentrations m~me tres basses de fines particules peut avoir un effet dd.vastateur considerable sur le procdd~ pour lequel I'eau est necessaire. Dens de relies conditions, fl est important pour I'utilisateur de savoir comment les filtres potentials se oomporteront. Pour le moment, il y a des filtres a cartouches avec des seuils de coupure de 0.114m mais il existe quelque contusion quest (~ ce clue reprdsentent ces seuils. Dens cet article, on
considere les seuils des filtres et on discute les inad~quations de ces parametres par rapport ~ des recherches rdcentes. (3 p., 4 fig., I tab., 8 r~f.)
Entender Is clasificaclbn de liltros de cartucho J. M. Bentley y P. J. Lloyd Se emplea filtracibn de cartucho en muchas industrias. La pureza necesaria de/ague filtrada es muchas veces muy alta, y la presencia de hasta concentraciones muy bajas de particulas fines puede dejarse sentir devastadoramente en el process que emplea el agua. En tales situaciones cobra importancia que el usario conoce el rend/relents de filtros disponibles. AI presente hay fittros de cartucho con clasificaci6n de mends de 0.1 p.m, pero la signihcacibn de esta no es segura. En este articulo se trata de la
t
clasificacibn de liltros, con la insuficiencia de esta respects de estudios recientes. (3 pdgs., 4 figs., I tabs., 8 refs,)
Effects of Dust Properties on Gas Cleaning Using Rigid Ceramic Filters Auswirkungen van Staebeigenscheften euf Gasrelnigung unter Verwendung harter Keramildllter u o n D. Koch, W. C h ~ m g , J. P. K. S e v i l l e um~ R. Cleft
Proben von harten Keramikfiltermedien wurden feiner Vergasungskohle und Calciumcarbonatstaub ausgesetzt, urn sowohl den Kuchenwiderstand sis such den Krattaufwand zu messen, der zur LOsung des Kuchens vom Filtermedium erfordertich war. Be/Vergasungskohle scheint der Kuchenwiderstand vonder FliJchengeschwin. digkeit unabhiJngig zu sein und sich proportions/zur fliJchenmi4Bi~en Kuchenbelastung zu verhaiten, w~hrend der zur LOsung des Kuchens ertorderhche Kraftaufwand ann~hernd unabh~ngig yon der Belastung /st. Be/ Calciumcarbonat stieg der Kuchenwiderstand im Verh~ltnis zur Breite der Gr~Benvertei/ung an, w~hrend der
zur Kuchenl(~sung ertorderliche Kraftaufwand etwas abnahm. Erhdhte Partikelgr~Be und KuchenlOckengrad lighten zu einer Verminderung yon Kuchenwiderstand und Kraftaufwand. Da erhOhte PartikelgrSBe jedoch me/starts in Verbindung mit vermindertem KuchenlOckengrad auftritt, /st die Skala der gemeinhin beobachteten Kuchenwiderstands- und LOsungskraftwerte nicht so breit wie zu erwarten wiire. (5 sn., 4 abb., 20 ref)
Effet des proprl6t6s de Is pousslbre sur I'&puratlon des gaz Iors de I'emplol de tlltres rlgldes en c&ramlque p a r D. Koch, W. CTwung, J. P. If. Seville et R. Cli~L Des media de filtration rig/des on cdramique ont 6t6 test~s pour des poussibres de charbon fin de gaz6ification et de carbonate de calcium; on a mesurd la r6sistance du g~}teau et la tension requise pour di~tacher la g~teau de media. Pour la poussi~re de oharbon, la r(~sistance du gtiteau semble ind6pendante de la vitesse faciale et proportionnelle ~ la charge superficielle, tandis que la press/on pour d~tacher /e g~teau est pratiquement ind6pendante de la charge. Pour le carbonate de calcium, un e/argissement de la distribution granulom~trique a augmentd la r~sistance du gMeau
mais a r6duit Idgbrement le stress de ddcollage. L'accroissement de la tail/e des particules et de Is porosit~ ont diminu~ et la r(~sistance du g~teeu et /e stress de d~charge. Cependant, ~tant donn~ qu'un accroissement de la faille des partieules semble sourest i)tre accornpagnd par une chute de la porositY, la plage commun{~ment observ~e des r6sistances de gz~teau et des pressions de d6col/age est plus ~troite qu'on pourrait le pr~voir. (5 p., 4 fig., 20 r6f.)
Electos de proprledades de polvo en el lavamlento de gases empleando flltros cer6mica rigids par D. Koch, W. Cheung, J. P. K Seville y R. Cleft Se has tratado trozos de medios cerdmica rigida con polvos de carbonizado fins de gasificador y de carbonato de calcio, y se hen mad/do la resistencia de torts y la fuerza necesaria pars separarla de/medio. Con carbonizado de origen gasificador parece que resistencia de torts es independiente de velocidad a la cara, y guarda proporcion con la oarga de torta por unidao de arcs; mientras, la fuerza de quitar 18 torta es mas o mends independiente de la carga. Con carbonato de calcio aumento de la game de distribucKJn de tama?los aumente la resistencia de torts pars reduce un poco la fuerza
326
necesaria pars quitar la torts. Aumento de tamat}o de particula y fraction de vacios de /a torts reducen tents la resistencia de torta corns la fuerza de separarla. Sin embargo, puesto que aumento de tamat}o particula parece acompet}ar uns disminucibn de fraccidn de ratios de torts, la game de resistent/as de torta y fuerzas de separaci6n es rues estrecha qua se espera. (5 p&gs., 4 figs., 20 refs.)
July/August 1992
Filtration & Separation
Interpreting the Rating of Cartridge Filters By J. M. Bentley and R J. Lloyd Department of Chemical Engineering, Loughborough University of Technology, Ashby Road, Loughborough LEll3TU, UK. This paper was presented at the Filtration Society meeting on Cartridge Filtration, Birmingham, 17 March 1992.
UK,
Cartridge filtration of water is used in many industries. The required purity of water being filtered is often extremely high and the presence of even very low concentrations of fine particles can have a potentially devastating effect on the process for which the water is needed. In such situations it is important for the user to be aware of how the available filters will perform. At present cartridge filters are available with ratings of less than 0.1 pm but there is some confusion as to what these ratings represent. In this paper the rating of fllters Is considered and the inadequacies of the ratings discussed with reference to recent studies. One of the most poorly understood terms in cartridge filtration is the filter rati@). The confusion arises partly because there are absolute and nominal ratings in use. An absolute filter is one which is said to remove virtually all particles (> 99.999999990 /o) larger than the rated pore size(2,3).These tend to be screen filters which remove particles whose dimensions are too large for them to pass through the opening of the filter. A nominally rated filter is one for which it is not possible to assign an absolute rating on the basis of pore size because the pore size is not well defined. Depth filters fall into this category. A nominally rated filter may be said to remove 90-950~ of particles larger than the nominal rating’2). The distinction between the two types of filter is based on how accurately the pore size distribution of the filter media can be defined and measured rather than on the retention efficiency for given particle sizes. As the nominal ratings do not make any specific claim regarding filter performance it is not appropriate to devote too much time to an analysis of such ratings; they simply serve as a guide to the user. However, the extent to which nominal ratings can be used as a guide is limited(4). On the other hand, absolute rated filters are often used in situations where complete removal of all particles above a certain size is desired. Examples are the pharmaceutical industry where organisms must be removed to allow the liquid stream to be cold sterilized, and the semiconductor industry where particles one tenth of the minimum device feature size are thought to be critical to the process(5). It is therefore important to know whether absolute rated filters perform according to their rating. Absolute rating of cartridge filters Absolute ratings of membrane filters were developed according to the requirements of the pharmaceutical industryc6). Membrane filters were rated according to their ability to retain bacteria. Filters which produced sterile effluent when challenged with lo7 Pseudomonas diminuta per square centimetre of filter surface area under standard conditions were considered absolute rated sterilizing ftiters. These filters were accepted as 0.2~ absolute rated, although it is known that the organism is larger than 0.2 pm. Although this method of rating filters is absolute with respect to removal of the Filtration&Separation
July/August
i 992
Pseudomonas diminuta organism, it is clearlyunacceptable as a method of rating a filter as absolute for 0.2 F particles”). To determine by non-destructive means the retention level of every filter manufactured, the bubble point integrity test is used. The pressure required to overcome the capillary forces in the pores of a fully wetted membrane is inversely proportional to the pore diameter. When pressure is applied to the pores of a wetted membrane the liquid from the largest pores will evacuate first. The pressure at which the capillary forces are first overcome is the bubble point value, and this indicates the size of the largest pore. If the bubble point value is known for a membrane which is known to be absolute for Pseudomonas diminuta, then other membranes with the same bubble point value may be assigned the same rating. However, when such a membrane is assigned a rating of 0.2~ and, even worse, when a membrane with a bubble point value twice that of the membrane absolute for Pseudomonas diminuta is assigned a rating of 0.1 pm, it is not possible to say that the absolute rating is equal to the diameter of the largest pore. Discussion of recent work In their work concerned with the rating of membrane filters Roberts et al@) challenged membranes with latex spheres under conditions designed to maximise penetration - by adding surfactant. They found that the efflciency of several membranes at the rated size was considerably less than 100%. They quote results for a double layer of a 0.1 pm rated membrane showing a retention efficiency of 33% with 0.106 pm latex spheres, less than 80% with 0.215pm spheres and 100% with 0.222~ spheres. In fact, of the three 0.1 pm rated membranes tested none retained as much as 10% of the challenge with 0.110 pm spheres when a single layer of membrane was used. Whilst these results do not represent the performance of such membranes in ultra pure water, they do indicate the presence of pores in the membrane which are significantly larger than the rated size. Recent work by Grant(” has shown that the retention efficiency of membrane filters varies with filter loading. A 0.45 m rated membrane filter was found to have an 333
TRIDGE FILTER RATING
I
Retention
efficiency (“I)
Similar results have been noticed by the authors in their work on the transient nature of cartridge filtration(7*s).One filter with a nominal rating of 0.45 ltrn was challenged with a range of particle sizes. In the event of impulses in the upstream concentration it was found that there was a temporary increase in penetration for particles in the range 0.46pm to 0.65 pm. A typical response for the filter is shown in Figs. 2 to 4. The response has been described in terms of a temporary blocking of smaller pores in the event of the very
LRV
90
1
99
2
99.9
3
99.99
4 5
99.999 I
1
Table 1 Relationship between retention effkiency tion value (LpV).
I
and log reduc-
initial log reduction value (LRV)+ of 2 when challenged with 0.442 pm particles, falling to less than 1 as the loading on the filter was increased. This was carried out under conditions where sieving was the dominant capture mechanism. With 0.527 pm particles the same filter had an initial LRV of 4.5 falling to 2 with loading. There were even measurable concentrations of particles as large as 0.66 l.unpenetrating the filter. Figure 1 shows the kind of response reported by Grant for different particle sizes on a 0.45 pm rated filter. Table 1 shows the relationship between LRV and retention efficiency.
3
12
g
4
5
+\f+(_._.______ _....... _.__._..._..._ .......__-_.-_____._---_._._i
6
7
6
Q
10
11
12
13
Time (min)
Fig. 2 Impulse in upstream concentration. loo-
---1.-._-.-._..___ -_._-_.________ __.___-_. _____--_
....__._
.________ ,__..__-
11. 0
o.oo@s 0.601 o.m15
o.m?
o.ce5
0.003 o.oca
o.OOl 0.0645
I 0.005
-_.._.._
Pore volumes of particles removed Fig. 1. Typical responses of a membrane dlfferent particle sizes (fromGrant@)).
filter challenged
with
There are two points of interest regarding filter ratings. Firstly, particles significantly larger than the rating of the filter are allowed to pass in measurable quantities, showing that there are pores larger than the rated size. In fact, for particles of similar size to the filter rating the retention efficiency is between 99% and !39.9%, well below what would be expected if there were no pores larger than the rated size. Secondly, as the loading on the filter increases the efficiency falls. A model has been developed by Grant describing the change in efficiency with loading. As the loading on the filter increases the smaller pores become blocked, resulting in a redistribution of the flow through larger pores for which the capture efficiency is less. The model and the experimental data are in reasonable agreement. The fall in efficiency is noticed for particles as large as 0.66 w challenging a 0.45 pm rated membrane filter, showing that there are pores significantly larger than the rating. +The log reduction value is given by: LRV = log,,(CJCo), where Ci is the concentration of particles at the filter inlet and Co is the concentration of particles at the filter outlet. 334
m
_.______._
. . . .
ok--?-!. 12345676
’
9
10
’
11
.
.’
12
13
..' 11 12
13
Time (min) Fig. 3. Downstream
3-..
123466769
.
response to upstream impulse.
-.
.
.
.
.
. 10
Time (min) Fig. 4. Penetration
in the event of an upstream Impulse. July/August 1992
Flltratlon & Ssparatlon
TRIDGE FILTER RATING high concentration impulse, with a resultant redistribution of flow through the larger pores. It would be expected that a nominally rated filter would have a wide size distribution with pores significantly larger than the rated size, but the same effect was observed for two membrane filters which were absolute rated 0.4 urn and 0.45 pm. Both membrane filters showed increased penetration at high challenge concentration with 0.46 pm particles, suggesting that there are pores larger than this in the filter media. The above work was carried out using 18 Mf&cm water at a constant flow rate of 10 Urnin. Conclusions
There are inadequacies in the ways in which filters are assigned absolute ratings. Despite the fact that the absolute rating is intended to represent the largest pore size in the filter media, work by three independent groups described in this paper shows that there are pores significantly larger than the rated size in absolute rated filters. It is clear that some changes to the rating system are required, either to measure the pore size more accurately or to base rating on penetration of known sized particles under standard conditions. If pore size is to be used as the basis of a rating system this will not represent the performance of the filter but will be a measure of a physical property of the filter which will indicate the size of particle which is too large to penetrate under any conditions unless the pores are
Filtration 8 Separation
July/August 1992
enlarged by some means such as pulsating flows. Such a system would give rise to a worst case rating. If efficiency under standard conditions is to be used then the user must be aware that under different conditions the filter performance may not match the rating. Either way, some standardization is needed to make the ratings more meaningful. References 1. Sandstedt, H. N. and Weisenberger, J. J. Cartridge Filter Performance and Micron Ratinn. Filtration and Senaration. “’ March/April 1985, p 101. to Filtration System Sizing, Millipore 2. An Introduction Technical Brief. 3. Grant, D. C. and Zahka, J. G. Sieving Capture of Particles by Micronorous Membrane Filters from Clean Liauids. Swiss Contdminution Control, 3 (4a), 1990, ~160. (Presented at ICCCS 1990 Symposium). Filter Performance: Myth and Reality, Practical 4. Juhasz,C. Aspects of Filter Testing, Proc World Filtmtion Congress (IZZ), 1982, p 596. 5. Plotnik, I. Metrology Applied to X-ray Lithography, Solid State Technology, January 1989, p 102. 6. Roberts, K. L., Velazquez, D. J. and Stofer, D. M. Challenging the Membrane Filter Ratings, Semiconductor Znternutionul, November 1990. Mechanisms of Penetration in Cartridge I. Bentley, J.M. Filtration of De-Ionised Water, Ph.D Thesis, LUT, 1991. 8. Bentley, J. M. Dynamic Studies in Membrane Filtration of Ultra-Pure Water, Z.Chem.E. Particle Technology Research Award Meeting, 1992.
335
@TRACTSOFREFEREEDPAPERS Recent Developments in OII Filter Test Methods Neuentwicklungen bel Olfllter-Testmethoden van B. M. Verdegan, K. McBroom u n d L. L i e bm ann Standardisierte Filtertestmethoden erleichtern den Vergleich verschiedener Filterprodukte. Der Filtertest mit mehreren Durchgangen stellt die Grundlage f~r zahlreiche (~lfilter-Testnormen dar, dessen Reproduzierbarkeit dutch bessere Spezifikation der Testbedingungen und Verbesserungen in den Partikelz~hlmethoden /aufend verbesserf worden /st. Partikelzahlung /st h~Jufig darer verantwortlich, dab die Ergebnisse yon e/nero Labor zum anderen variieren. Aufgrunddessen wird seit kurzem in den USA eine Kalibrierungsmethode zur Partikelzahlung, ANSII(NFPA) T2.9.6R1-1990, angewendet. Mit H/Ire der neuen Methode /st es mOglich, international anerkannte Definitionen des
Meters anzuwenden, die bei der a/ten Methode der AC Feinteststaubkalibrierung nicht zuganglich waren. Die PartikelgrOf~endefinitionen sind jedoch bei qten beiden Methoden unterschiedfich. Die bestehenden Normen sind geringfi)gigen Anderungen zu unterziehen, um sie der neuen Kafibrierungsmethode anzugleichen. Andere M(~glichkeiten der Verbesserung yon Reproduzierbarkeit und Vermittlung von Testergebnissen sind ebenfalls erSrtert. (5 sn., 4 abb., 15 ref)
D6veloppements r(~cents dens les m~thodes d'essels des flltres b hulle p a r B. M. Verdegan, K McBroom et L. L i e bm ann Des methodes d'essais standard/sees de filtres facilitent la comparaison de differents appareils. Le test multipass constitue la base pour plusieurs standards de filtres a huile et sa reproductibilite a (~t~ continOment amdlioree par de meilleures spdcifications des conditions d'essai et par un raffinement des methodes de comptage des particules. Ce dernier est use cause importante de variabilit~ dans les essais inter-laboratoires. Dans ce contexte, ANSII(NFPA) T2.g.6RI-1990, une m~thode de calibration du comptage utilisant des spheres de latex en suspension dans I'huile, a etd adoptc~e par les USA.
Cette methode permet de s'adapter aux definitions du metre accept(~es au plan international mais que I'on ne peut atteindre par la vie/lie m~,thode de la calibre t/on AC avec fine poussii~re. Cependant, les definitions de la granulom~.trie des deux mdthodes different. Des modifications mineures aux standards existants s'imposent pour les harmon/set avec la nouvelle methode de calibration. On discute ~galement d'autres moyens d'ameliorer la reproductibilitd et la presentation des rdsultats d'essaL (Sp., 4 fig., 15ref.)
Desarrollos corrlentes en los re&fades de probsr flltros pars acelte p a r B. M. Verdegan, If. McBroom y L. L i e b m a n n M~todos estdndares de prober filtros facilitan comparaci(~n de distintos productos de filtro. La prueba mu/tipaso de filtro es /a base de muchos estdndares de prueba de fi/tros pars ace/re, y su reproducibilidad se ha rnejorado por major especificaciOn de la manera de hacer /as pruebas y mejores m~tods de contar particulas. Contar particulas es la mayor fuente de variabilidad entre laboratorios. Por repuesta, recientemente se ha aprobado en los EE UU el m6todo de calibraciSn de confer particu/as ANSII (NFPA) T2.9.6RI-1990, m~todo qua emp/ea esferas de latex suspend/des en aceite. El nuevo
~
~P
m~todo demuestra relacibn con definiciones aprobadas internacionalmente de/metro, que no ocurre con el viejo mdtodo AC de calibraci~n con po/vo fins de prueba. Sin embargo,/as clef/sic/ones de tamer, s de particulas son distintas con los dos m~todos. Pequer}as modificaciones son necesarias con estdndares cordentes pars arrnonizar/es con el nuevo mdtodo de calibraciOn. Se discuten otras maneras de mejorar reproducibilidad y presenter resultados de prueba. (5 p~gs., 4 figs., 15 rets.)
Interpreting lhe Rating of Cartridge Filters Interpretation der Nennlelstung van Kerzenflltarn v a n J. M. Be nt l e y u n d P. J. Lloyd
Die Kerzenfiltration von Wasser /st in zahlreichen Industriezweigen Oblich. Die erforderliche Reinheit des zu filtrierenden Wassers ist oft aufJerordentlich hoch, und des Vorhandensein selbst sehr geringer Konzentrationen feiner Partikel kann u . U . verheerende Auswirkungen auf des Verfahren ausOben, fbr des des Wasser benOtigt wird. In derartigen Situationen ist es wichtig, dab der Benutzer ~ber die Leistungsdaten der verfOgbaren Filter genau informiert /st. Zur Zeit gibt es Kerzenfilter mit Nennleistungen yon wen/get sis 0,1 I~m: es /st ledoch oft nicht ganz klar, was diese
Werte eigentlich beinhalten. Im vorliegenden Referat wird die Festsetzung von Filternennleistungswerten dargelegt und die Unzul~nglichkeiten der Nennleistungsbestimmung unter Bezugnahme auf neue Untersuchungen erOrtert, (3 sn., 4 ebb., I tab., 8 ref)
Interpretation des r~sultats des flltres b cartouches p a r J. M. Bentley et P. J. Lloyd La filtration d'eau sur cartouche est utilisee dans beaucoup d'industries. La puret~ requise de reau filtr(~e est sourest tr~s ~lew~e et la pr6sence de concentrations m~me tres basses de fines particules peut avoir un effet dd.vastateur considerable sur le procdd~ pour lequel I'eau est necessaire. Dens de relies conditions, fl est important pour I'utilisateur de savoir comment les filtres potentials se oomporteront. Pour le moment, il y a des filtres a cartouches avec des seuils de coupure de 0.114m mais il existe quelque contusion quest (~ ce clue reprdsentent ces seuils. Dens cet article, on
considere les seuils des filtres et on discute les inad~quations de ces parametres par rapport ~ des recherches rdcentes. (3 p., 4 fig., I tab., 8 r~f.)
Entender Is clasificaclbn de liltros de cartucho J. M. Bentley y P. J. Lloyd Se emplea filtracibn de cartucho en muchas industrias. La pureza necesaria de/ague filtrada es muchas veces muy alta, y la presencia de hasta concentraciones muy bajas de particulas fines puede dejarse sentir devastadoramente en el process que emplea el agua. En tales situaciones cobra importancia que el usario conoce el rend/relents de filtros disponibles. AI presente hay fittros de cartucho con clasificaci6n de mends de 0.1 p.m, pero la signihcacibn de esta no es segura. En este articulo se trata de la
t
clasificacibn de liltros, con la insuficiencia de esta respects de estudios recientes. (3 pdgs., 4 figs., I tabs., 8 refs,)
Effects of Dust Properties on Gas Cleaning Using Rigid Ceramic Filters Auswirkungen van Staebeigenscheften euf Gasrelnigung unter Verwendung harter Keramildllter u o n D. Koch, W. C h ~ m g , J. P. K. S e v i l l e um~ R. Cleft
Proben von harten Keramikfiltermedien wurden feiner Vergasungskohle und Calciumcarbonatstaub ausgesetzt, urn sowohl den Kuchenwiderstand sis such den Krattaufwand zu messen, der zur LOsung des Kuchens vom Filtermedium erfordertich war. Be/Vergasungskohle scheint der Kuchenwiderstand vonder FliJchengeschwin. digkeit unabhiJngig zu sein und sich proportions/zur fliJchenmi4Bi~en Kuchenbelastung zu verhaiten, w~hrend der zur LOsung des Kuchens ertorderhche Kraftaufwand ann~hernd unabh~ngig yon der Belastung /st. Be/ Calciumcarbonat stieg der Kuchenwiderstand im Verh~ltnis zur Breite der Gr~Benvertei/ung an, w~hrend der
zur Kuchenl(~sung ertorderliche Kraftaufwand etwas abnahm. Erhdhte Partikelgr~Be und KuchenlOckengrad lighten zu einer Verminderung yon Kuchenwiderstand und Kraftaufwand. Da erhOhte PartikelgrSBe jedoch me/starts in Verbindung mit vermindertem KuchenlOckengrad auftritt, /st die Skala der gemeinhin beobachteten Kuchenwiderstands- und LOsungskraftwerte nicht so breit wie zu erwarten wiire. (5 sn., 4 abb., 20 ref)
Effet des proprl6t6s de Is pousslbre sur I'&puratlon des gaz Iors de I'emplol de tlltres rlgldes en c&ramlque p a r D. Koch, W. CTwung, J. P. If. Seville et R. Cli~L Des media de filtration rig/des on cdramique ont 6t6 test~s pour des poussibres de charbon fin de gaz6ification et de carbonate de calcium; on a mesurd la r6sistance du g~}teau et la tension requise pour di~tacher la g~teau de media. Pour la poussi~re de oharbon, la r(~sistance du gtiteau semble ind6pendante de la vitesse faciale et proportionnelle ~ la charge superficielle, tandis que la press/on pour d~tacher /e g~teau est pratiquement ind6pendante de la charge. Pour le carbonate de calcium, un e/argissement de la distribution granulom~trique a augmentd la r~sistance du gMeau
mais a r6duit Idgbrement le stress de ddcollage. L'accroissement de la tail/e des particules et de Is porosit~ ont diminu~ et la r(~sistance du g~teeu et /e stress de d~charge. Cependant, ~tant donn~ qu'un accroissement de la faille des partieules semble sourest i)tre accornpagnd par une chute de la porositY, la plage commun{~ment observ~e des r6sistances de gz~teau et des pressions de d6col/age est plus ~troite qu'on pourrait le pr~voir. (5 p., 4 fig., 20 r6f.)
Electos de proprledades de polvo en el lavamlento de gases empleando flltros cer6mica rigids par D. Koch, W. Cheung, J. P. K Seville y R. Cleft Se has tratado trozos de medios cerdmica rigida con polvos de carbonizado fins de gasificador y de carbonato de calcio, y se hen mad/do la resistencia de torts y la fuerza necesaria pars separarla de/medio. Con carbonizado de origen gasificador parece que resistencia de torts es independiente de velocidad a la cara, y guarda proporcion con la oarga de torta por unidao de arcs; mientras, la fuerza de quitar 18 torta es mas o mends independiente de la carga. Con carbonato de calcio aumento de la game de distribucKJn de tama?los aumente la resistencia de torts pars reduce un poco la fuerza
326
necesaria pars quitar la torts. Aumento de tamat}o de particula y fraction de vacios de /a torts reducen tents la resistencia de torta corns la fuerza de separarla. Sin embargo, puesto que aumento de tamat}o particula parece acompet}ar uns disminucibn de fraccidn de ratios de torts, la game de resistent/as de torta y fuerzas de separaci6n es rues estrecha qua se espera. (5 p&gs., 4 figs., 20 refs.)
July/August 1992
Filtration & Separation