COMPARISON OF FILTER CLEANING PERFORMANCE BETWEEN VDI AND JIS TESTING RIGS FOR CLEANABLE FABRIC FILTER

Ttle Author(s) Comparson of flter cleanng perfo rgs for cleanable fabrc flter Mao, Nng; Yao, Yupng; Hata, Mtsu Chkao Ctaton Powder Technology, 180(1-2): 109-11 Issue Date 2008-01-14 Type Journal Artcle Text verson publsher URL http://hdl.handle.net/2297/7673 Rght *KURAに 登 録 されているコンテンツの 著 作 権 は, 執 筆 者, 出 版 社 ( 学 協 会 )などが 有 します *KURAに 登 録 されているコンテンツの 利 用 については, 著 作 権 法 に 規 定 されている 私 的 使 用 や 引 用 などの 範 囲 内 で 行 ってください * 著 作 権 法 に 規 定 されている 私 的 使 用 や 引 用 などの 範 囲 を 超 える 利 用 を 行 う 場 合 には, 著 作 権 者 の 許 諾 を 得 てください ただし, 著 作 権 者 から 著 作 権 等 管 理 事 業 者 ( 学 術 著 作 権 協 会, 日 本 著 作 出 版 権 管 理 システムなど)に 権 利 委 託 されているコンテンツの 利 用 手 続 については, 各 著 作 権 等 管 理 事 業 者 に 確 認 してください http://dspace.lb.kanazawa-u.ac.jp/dspac

COMPARISON OF FILTER CLEANING PERFORMANCE BETWEEN VDI AND JIS TESTING RIGS FOR CLEANABLE FABRIC FILTER Nng Mao 1, Yupng Yao 1, Mstuhko Hata 1, Masash Wada 2 and Chkao Kanaoka 2 1 Graduate School of Natural Scence and Technology, Kanazawa Unversty 1 Kakuma-mach, Kanazawa, Ishkawa, 920-1192, Japan 2 Ishkawa Natonal College of Technology 2 Ktacyujo, Tsubata, Kahoku-gun, Ishkawa, 929-0392, Japan ABSTRACT VDI type-1 rg and JIS rg are the two major testng rgs for cleanable fabrc flters. We measured the flter cleanng performance usng these rgs and the results were compared n order to characterze the two testng methods. The flter performance tests showed that the flter cleanng effcency measured wth VDI type-1 rg s hgher than that wth JIS rg. Durng pulse jet cleanng, JIS rg gave a hgher peak pressure and a shorter tme perod of pulse jet compared to VDI type-1 rg. A new fltraton model was appled to the fltraton cycles wth the two rgs and t was shown that the dfference n flter cleanng performance measured by VDI type-1 and JIS rgs can be expressed n terms of the surface cleanng fracton, whch s the rato of cleaned surface area to the total surface area of a flter, and the resdual dust load on the un-cleaned surface after cleanng. INTRODUCTION Fabrc flters are wdely used to remove partculate matters n flue gas wth a hgh collecton effcency. In order to evaluate the fltraton performance, a number of testng rgs have been developed, n whch VDI type-1 rg and JIS rg are the two representatve ones. [1-2] Incomplete flter cleanng has a marked effect on the fltraton performance of fabrc flters where the dust cake s detached unevenly from the flter surface durng the perodcal flter cleanng. The present paper compares the flter cleanng performance measured by VDI type-1 rg and JIS rg and extracts the essental dfference n the testng methods. There are many models for expressng the fltraton process wth patchy cleanng [3-5]. Some prevous models may express the actual dust cleanng pattern on flter surface, but they seem too complcated for practcal use. Therefore, a new model s proposed and appled to evaluate and compare the flter cleanng performance measured by VDI type-1 rg and JIS rg. EXPERIMENTAL SETUPS VDI type-1 rg specfed n VDI/DIN 3926 and JIS rg n JIS Z 8909-1 are used n the present work. They compose of three parts,.e. the dust feedng and dspersng part, the dust collectng and cleanng part, and the montorng and controllng part. The schematc dagrams of VDI type-1 rg and JIS rg are shown, respectvely, n Fgs.1 and 2. Test dust from a two-stage screw dust feeder s dspersed by an ejector. The dust collectng and cleanng part ncludes a vertcal duct and a horzontal duct. The dust laden gas s ntroduced 1

nto the top of the vertcal duct, and the dust n the gas s collected on the testng flter nstalled at the nlet of the horzontal duct. The flter pressure s recorded by a pressure transducer (VALCOM, VPRN-A4). The fltraton-cleanng cycle controlled by ether a prescrbed flter pressure or tme nterval s repeated at a gven fltraton velocty. When the flter pressure reaches a prescrbed value (1000 Pa n the present work), flter cleanng s carred out wth compressed ar of 500 kpa-gauge through a solenod valve connected to a 2.5 L compressed ar reservor. Otherwse, flter cleanng s carred out wth a preset tme nterval (5 s n the present work) for a large number of cycles. The downstream dust concentraton s measured by samplng the test partcles wth HEPA flters. The JIS rg s small n dmenson compared to VDI type-1 rg. Snce all the dust laden gas passes through the test flter n JIS rg, t s called as total-flow type. However, n VDI type-1 rg, only a small fracton of gas passes through the test flter and a larger fracton of gas leaves from the rg at the bottom of the vertcal duct. Therefore, VDI type-1 rg s called as part-flow type rg. The test flter used n JIS rg s rectangular n shape wth the sze of 300 mm (wdth) 300 mm (length), whereas the flter s crcular wth 140mm n dameter n VDI type-1 rg. RESULTS AND DISCUSSIONS Flter performance test The flter performance test s conducted usng JIS and VDI type-1 rgs. The testng procedure s as follows: Step 1: Vrgn performance test perod: 30 fltraton-cleanng cycles wth a prescrbed cleanng pressure drop (1000 Pa) Step 2: Agng condton perod: 5000 cycles wth flter cleanng at gven nterval of 5 s Step 3: Stablzng perod: 10 fltraton-cleanng cycles wth the prescrbed cleanng pressure drop (1000 Pa) Step 4: Performance test perod: 30 fltraton-cleanng cycles wth the prescrbed cleanng pressure drop (1000 Pa) The testng condtons are lsted n Table 1. The pressure drop after flter cleanng s referred to as resdual pressure drop, whch s an mportant parameter for the evaluaton of fabrc fltraton performance. The testng result of resdual pressure drop s shown n Fg.3. At the end of the test, the test flter s removed and weghed to obtan the resdual dust load (Fg.4). In Fg.3, the VDI type-1 rg and JIS rg gve nearly the same resdual pressure drop n the frst 30 cycles of Step 1 before the flter agng condton. However, after the agng condton, the resdual pressure drop n the last 30 cycles of Step 4 wth JIS rg s much hgher than that wth VDI type-1 rg, ndcatng that the flter cleanng effcency wth JIS rg s consderably lower than that wth VDI type-1 rg. The dfference n the cleanng effcency between the rgs s also clearly seen from the resdual dust load at the end of performance test shown n Fg.4. The development of resdual pressure drop s nfluenced by not only the flter cleanng effcency but also the number of pulse jet cleanng. For a new flter before the agng, the partcles collected on the flter are relatvely easy to be cleaned so that the dfference n flter cleanng 2

effcency between VDI type-1 rg and JIS rg would not cause any dfference n the resdual pressure drop. However, once the flter s aged by applyng 5000 cycles of fltraton cycles, the partcles on the flter become more dffcult to be detached and therefore even a small dfference n the flter cleanng effcency leads to a bg dfference n the resdual pressure drop for the last 30 cycles. Pressure drop durng pulse jet cleanng In order to compare the flter cleanng performance between VDI type-1 rg and JIS rg, the tme change n pressure upstream and downstream of the flter durng the pulse jet cleanng s measured. Fg.5 shows the pressure evoluton durng the pulse jet cleanng as well as the pressure drop across the flter. Because of the nature of part-flow employed n VDI type-1 rg, the pressures on both sdes n VDI type-1 rg are lower than those n JIS rg. The maxmum pressure drop s hgher for JIS rg at a shorter tme perod of pulse jet cleanng. The flter cleanng s a complcated phenomenon, whch s nfluenced by many factors such as the force actng on a flter, the cleanng ar volume, the peak pressure, and the pattern of cleanng pressure buldup. Durng the flter cleanng, the dust cake detaches from the flter surface at the nstant that the force reaches the maxmum and a fracton of removed dust wll re-attach on the flter surface at the end of flter cleanng because of the reversed gas flow toward the flter surface. The flter cleanng effcency s therefore affected by the combnaton of these detachment and re-attachment mechansms. The detachment of partcles s enhanced wth the peak pressure and the re-attachment s reduced wth ncreasng the tme perod of pulse jet. The dfferences n confguraton of the pulse jet system employed n VDI type-1 rg and JIS rg result n the dfferent flter cleanng performance. In VDI type-1 rg, the pulse jets are ejected through a nozzle wth 3mm n dameter toward the downstream sde of the test flter at a dstance of 625 mm downstream of the flter. The rg volume of the clean sde s 13 L. However, n the JIS rg, the nozzle wth 9 mm n dameter faces downward n the horzontal duct. The flter cleanng s performed by the arflow reflected on the duct wall. The dstance between the nozzle and the flter s 150 mm and the rg volume of the clean sde s 29 L. The dfferences n the confguraton of two rgs make t more dffcult to compare the flter cleanng performance. APPLICATION OF NEW MODEL FOR FILTER CLEANING PERFORMANCE EVALUATION A model for fabrc flter Kanaoka and Yao [6] recently studed the fltraton phenomenon wth a pulse jet non-woven fabrc flter. They found that the evoluton of pressure drop of a flat type fabrc flter could be dvded nto three stages n a fltraton cycle, where the ncreasng rate of pressure drop gradually ncreases n the frst stage and then decreases n the second stage, and fnally t s almost a constant n the thrd stage as shown n Fg.6. Based on ths study, we propose a new model, dvdng the fabrc fltraton process nto three stages,.e. depth fltraton stage, transton fltraton stage and surface fltraton stage. After flter cleanng, the resdual dust cake remans on a porton of flter surface, whereas on 3

a porton of flter surface, not only the dust cake s completely removed but also a part of dust nsde the flter s also cleaned. Therefore, the depth fltraton takes place on the cleaned surface just after the flter cleanng untl the dust load reaches a maxmum retenton n the flter. Snce the fltraton velocty through the cleaned surface s much hgher than that through the un-cleaned surface, we assume that there s arflow only through the cleaned surface. Once the dust begns to accumulate on the cleaned surface, whch s the end of depth fltraton stage, the second stage of transton fltraton stage begns. Durng the transton stage, the surface fltraton on the un-cleaned surface also plays a role for dust collecton. The local fltraton velocty vares over the flter surface because of the non-unformty n the dust cake dstrbuton. As tme passes, the non-unformty wll gradually dsappear because of self-equalzng mechansm for parallel path flows. The dfference n local fltraton velocty wll then dmnsh and ths wll lead to unform deposton of flter cake, whch s the begnnng of surface fltraton stage. Surface cleanng fracton and resdual dust load on un-cleaned surface In order to express the non-unformty of resdual dust cake dstrbuton after flter cleanng n a smpler manner, the present model ntroduces two parameters,.e. the surface cleanng fracton, f, whch s the rato of cleaned surface area to the total surface area of a flter, and the resdual dust load on un-cleaned surface after cleanng, W u0. Here cleaned mples that the dust cake accumulated on the flter s completely removed and yet the dust nsde the flter s also partally cleaned. The resdual dust dstrbutes homogenously on the un-cleaned surface. Both W and f change wth the number of fltraton cycle. Accordng to our model the pressure drop development durng the transton and surface fltraton stages can be expressed as a functon of, and t. In order to obtan the surface cleanng fracton,, and the resdual dust load on the un-cleaned surface after cleanng,, two ponts on the expermental curve of f W u 0 f 1 W u 0, 1 Δ p s at t ( t = τ and t = τ + ψ ) are utlzed, where the duraton of depth fltraton perod and u0 τ s ψ s the duraton of transton and surface fltraton perods for the -th cycle. At t = τ, Eq.(1) can be obtaned snce R uτ, >> Rc τ,. u Δ p τ, = μ Rc τ, (1) f At t = τ + ψ, the pressure can be approxmately expressed by: 1 Δpmax = μu( Rc τ, + K(1 f 1) Wu 0, 1 + Kc0uψ ) (2) If we assume that Eq.(1) holds for a short tme perod, Δt, n the begnnng of transton fltraton, then we have: u u Δp = + Δ τ + Δt, μ Rcτ, Kc0 t (3) f 1 f 1 where the second term n the bracket s the ncrease n resstance due to the partcles accumulated on cleaned surface. Subtractng Eq.(1) from Eq.(3) and lettng Δt 0, we have: Δp 2 τ +Δt, Δpτ, dδp μkc0u lm = = Δt 0 Δt (4) dt f τ, 2 1 4

dδp Snce dt τ, dδp = dt max,, f 1 can be calculated by: 2 dδp f 1 = μ Kc 0u (5) dt max, Now we have f -1, can then be calculated by Eq.(3) : W u 0, 1 W u0, 1 = Δp max 1/ 2 f 1Δpτ, Kc0uψ μu K( 1 f 1) (6) Evaluaton on flter cleanng performance by the model Utlzng the flter performance test results wth the VDI type-1 and JIS rgs, f and W u 0 are calculated and plotted n Fgs. 7 and 8. Fg.7 shows there s no sgnfcant dfference n f by the rgs employed. However, n Fg.8, obtaned wth JIS rg s consderably hgher than that wth VDI type-1 rg for the last 30 cycles after the agng, ndcatng that the dust cleanng effcency s hgher for the VDI type-1 rg. Ths s n accordance wth the results shown n Fgs.3 and 4. Therefore, the dfference n flter cleanng performance can be expressed n terms of and by our model. W u0 f W u 0 Resdual pressure drop vs. outer resdual dust load Resdual dust load conssts of two parts,.e. nner resdual load and outer resdual dust load. The outer resdual dust load s determned by and as: W out f W u 0 = W 0 /(1 f ) (7) The outer resdual dust load s obtaned from the calculaton results n Fgs.7 and 8 and t s plotted wth resdual pressure drop together n Fg.9. It s shown that the resdual pressure drop ncreases wth the outer resdual dust load n the same way for the two rgs ether before agng or after agng. The compatblty of the VDI type-1 rg and JIS rg s verfed by the model. u CONCLUSIONS 1. VDI type-1 rg gves a hgher flter cleanng performance than JIS rg. 2. Durng the pulse jet cleanng, the peak pressure measured wth JIS rg s hgher than that wth VDI type-1 rg, and the tme perod of pulse jet wth JIS rg s shorter than that wth the VDI type-1 rg. 3. The dfference n flter cleanng performance measured by the VDI type-1 and JIS rgs can be expressed n terms of and accordng to our model. f W u 0 NOMENCLATURE c 0 : Inlet dust concentraton (kg/m 3 ) f : Surface cleanng fracton (-) K : Specfc resstance of accumulated dust on flter (m/kg) 5

p: Flter pressure drop (Pa) R : Resstance of flter (1/m) t : Fltraton tme (s) u : Fltraton velocty (m/s) W : Dust load per unt surface area of flter (kg/m 2 ) μ: Gas vscosty (Pa s) τ : Duraton of depth fltraton perod (s) Ψ: Duraton of transton and surface fltraton perods (s) Subscrpt c : Cleaned surface : Number of fltraton cycle max : Maxmum out : Out s : Transton and surface fltraton stages u : Un-cleaned surface τ : At the end of depth fltraton REFERENCES 1. VDI/DIN 3926 Testng of Cleanable Flter Meda 2. JIS Z 8909-1 Testng methods of flter meda for dust collecton 3. W. Duo, N.F. Krkby, J.P.K. Sevlle, R. Clft, Patchy Cleanng of Rgd Gas Ftlers Ⅰ. A Probablstc Model, Chemcal Engneerng Scence, 52 (1) (1997) 141-151. 4. W. Duo, J.P.K. Sevlle, N.F. Krkby, H. Buchele, C.K. Cheung, Patchy Cleanng of Rgd Gas Flters Ⅱ. Experments and Model Valdaton, Chemcal Engneerng Scence, 52(1) 1997 153-164. 5. A. Dttler, G. Kasper, Smulaton of Operatonal Behavor of Patchly Regenerated, Rgd Gas Cleanng Flter Meda, Chemcal Engneerng and Processng, 38 (1999) 321-327. 6. C. Kanaoka, Y.P. Yao, Tme Dependency of the Pressure Drop n a Flat Type Pulse Jet Fabrc Flter, Kagaku Kogaku Ronbunshu, 29(2) 2003 267-271. n Japanese 6

Fg. 1 Schematc dagram of VDI type-1 rg Fg. 2 Schematc dagram of JIS rg Resdual pressure drop [Pa] 400 300 200 100 0 after agng JIS rg VDI rg before agng 0 10 20 30 Number of fltraton cycle [-] Fg.3 Resdual pressure drop 7

Resdual dut load [kg/m 2 ] 0.13 0.125 0.12 0.115 0.11 0.105 JIS VDI Fg.4 Resdual dust load at the end of performance test Downstream (or upstream) pressure [kpa-gauge] 16 12 8 4 0-4 -8 Pressure drop(jis) Pressure drop(vdi) Pulse pressure: 500kPa-gauge Downstream Upstream Downstream Upstream 0 1000 2000 Tme [ms] JIS VDI 6 2-2 -6-10 Pressure drop [kpa] Fg.5 Pressure drop development durng pulse jet cleanng 8

1 m u 0 Ru τ 1 f 1 ( τ + ψ, Δpτ + ψ ) Rc τ f 1 Pressure drop, p Ⅰ Ⅱ ( τ, Δpτ ) Ⅲ Ⅰ: Depth fltraton Ⅱ: Transton fltraton Ⅲ: Surface fltraton Tme, t Fg. 6 Illustraton on a fltraton cycle of fabrc flter f [-] 0.7 0.6 0.5 0.4 0.3 0.2 0.1 0 JIS before agng VDI after agng 0 10 20 30 Number of fltraton cycle [-] Fg.7 Surface cleanng fracton 9

Wu0 [kg/m 2 ] 0.08 0.07 0.06 0.05 0.04 0.03 0.02 0.01 0 after agng before agng JIS VDI 0 10 20 Number of fltraton cycle [-] 30 Fg.8 Resdual dust load on un-cleaned surface Resdual pressure drop [Pa] 400 300 200 100 JIS VDI VDI(after agng) VDI(before agng) JIS(before agng) JIS(after agng) 0 0.02 0.04 0.06 0.08 W u0 /(1-f ) [kg/m 2 ] Fg. 9 Resdual pressure drop vs. outer resdual dust load 10

Table 1 Testng condtons Fabrc flter Materal: [-] PE snged Structure: [-] Felt Partcle materal: [-] Pural NF Dust Mass medan dameter: * [μm] 5.8 Densty: [kg/m 3 ] 600-800 Dust concentraton: [g/m 3 ] 5 Fltraton velocty: [m/s] 0.033 Testng condtons Pulse jet pressure: [kpa-gauge] 500 Flter pressure before cleanng: [Pa] 1000 * Mass medan dameter was measured wth Sympatec HELOS 11