Characterization of the gas pulse frequency, amplitude and velocity in non-steady dense phase pneuma

Available online at www.sciencedirect.comScienceDirectPARTICUOLOGYEL SEVIER，Particuology 6 (2008) 301-306www.elsevier. com/locate/particCharacterization of the gas pulse frequency, amplitude and velocity innon-steady dense phase pneumatic conveying of powders *Kenneth C. Williams *, Mark G. Jones, Ahmed A. CennaCentre for Bulk Solids and Particulate Technologies, School of Engineering, University of Newcastle, 2308, AustraliaReceived 28 December 2007; accepted 6 March 2008AbstractCurrent modelling techniques for the prediction of conveying line pressure drop in low velocity dense phase pneumatic conveying are largelybased on steady state analyses. Work in this area has been on-going for many years with only marginal improvements in the accuracy of predictionbeing achieved. Experimental and theoretical investigations undertaken by the authors suggest that the flow mechanisms involved in dense phaseconveying are dominated by transient efcts rather than those of steady state and are possibly the principal reasons for the limited improvement inaccuracy. This paper reports on investigaions on the pressure fuctuation behaviour in dense phase pneumatic conveying of powders. The pressurebehaviour of the gas flow in the top section of the pipeline was found to exhibit pulsatile oscillations. In particular, the pulse velocity showedvariation in magnitude while the frequency of the oscillations rarely exceeded 5 Hz. A wavelet analysis using the Daubechie 4 wavelet found thatthe amplitude of the osillations increased along the pipeline. Furthermore, there was significant variation in gas pulse amplitude for diferent typesof particulate material.◎2008 Chinese Society of Particuology and Institute of Process Engineering, Chinese Academy of Sciences. Published by EIsevier B.V.AIll rights reserved.Keywonds: Pneumatic conveying; Powders; Dense phase: Pulsaile fow; Transient flow1. Introductionface between the high velocity gas and the dune is particularlycomplex with the exchange of arflow between the interstitialResearch work in dense phase pneumatic conveying hasgas flow within the dune and the high velocity gas flow over thetended to concentrate on the granular fll-bore slug flow modetop of the dune surface.of dense phase conveying (Konrad, Harrison, Nedderman, &The objective of the initial research is focussed on theDavidson, 1980; Legel & Schwedes, 1984; Pan & Wypych,behaviour and structure of the gas fow over the top of the dunes.1997). This is perhaps not surprising as this mode of flowDhodapkar and Klinzing (1993) investigated pressure fluctua-lends itself to the application of a traditional particle mechanicstions for different modes of flow regimes which included theapproach. Fluidised dense phase conveying of powders (some-fuidised dense phase regime and reported that the frequencytimes known as moving bed or dune type flow) presents a moreof the fuctuations for this flow was rarely above 4 Hz. Anothercomplex scenario in which several mechanisms are observed.important observation made by Dhodapkar and Klinzing is thatThe fuidised dunes can be considered as a continuum where, inthe pressure time trace structure in fluidised dense phase flowhorizontal flow, the dune is supported by the bottom of the pipeshowed irregular oscilating behaviour which was further char-and where the frictional interactions between the dune and theacterised by sharp peaks. To gain relevant information from thesepipe wall induce a retarding force on the flow of material. The toposcillating peaked pressure traces, the choice of analysis toolssurface of the dune is subject to shear due to the relatively highis important, as any significant smoothing of the pressure peaksvelocity flow of gas in the top section of the pipeline. The inter-will reduce the accuracy of the analysis. For instance, a Fourieranalysis consists of breaking up a signal into sine waves of var-ious frequencies while a wavelet analysis is the breaking u* Article selected from 2007 Intermational Symposium on Pneumatic Convey-ofas中国煤化工rsions of the original (oring Technologies, October 18- 20, 2007, Beijing, China.motheFely that signals with sharp'Corresponding author.changYHC N M H Gan iregular wavelet ratherE-mail adress: kenilliams@newcaste edu.au (K.C. Wlliams).1674- 2001/$ - see inside back cover◎2008 Chinese Society of Particuology and Insiute of Process Engineering. Chinese Acadeny of Sciences. Published by Elsevier B.VAII rights reserved.doi:10.1016/jparic.2008.03.000302 .K.C. Wlliams et al. / Paricuology 6 (2008) 301-3069.50m14.57mFILTER147Jm2.59miN |8.22m" FLOW DIRECTIONX 3.99msecondary35.18m6.22m6.07mtansducersS5omm nom bore mld seel pipe8.23mNOTE: Dirtances inficme tnight pipe lnghsFig. 1. Schematic of 130m pipeline used in the cement meal pneumatic conveying tests.than with a smooth sinusoid. Li (2002) analysed the pressuresolids mass gain into the receival bin were measured via a seriesfluctuations for 3.5 mm polyethylene pellets in horizontal pneu-of load cells to determine the mass flow rate through the pipeline.matic conveying pipe using wavelet analysis. More recently in2006, Pahk, Sahin, and Klinzing (2006) utilised wavelet analy-3. Analysissis to investigate the variance and approximation of the pressuresignal for 3.3 mm polyester pellets, 3.9 mm polystyrene pelletsA comparison of the pressure in the blow tank, the secondaryand 4.6 mm polyolefin pellets.The aim of the research detailed in this paper was to inves-air injection point and the pulsatile pressure at 85.1 m from thetigate the transient pulsatile gas flow behaviour associated withstart of the pipeline (pressure transducer location T5), at 86.1 mthe dense phase conveying powders in a pneumatic conveying(T6) and at 87.1 m (T7) can be seen in Fig. 2. During each test,system. In particular, analysis of the structure of the conveyinginitially there is a surge of product and pressure through theair pressure trace over the top of the dense phase particulate layerpipeline when the bulk material fow is initiated. Subsequently,was conducted, namely, the frequency, velocity and amplitudethe conveying conditions dampen to steady state conditions,which is characterised by relatively constant pressure in theof the gas pulses.blow tank. It is clearly apparent that fluctuations in the pres-sure occurred in the secondary air injection point and theT5- -T72. Experimental programpressure transducer locations. It is the pressure fluctuations ofT5- -T7 in the steady state region that is the initial focus of theFluidised dense phase pneumatic conveying tests were ini-gas flow analysis. The final stage in the conveying test occurstially conducted with cement meal (average particle diameter,when the height of the material in the blow tank reaches a criti-dso= 11 μm, particle density, Pp = 3000 kg/m3 and loose pouredcal level. At this critical level, the pressure fuctuations becomebulk density, Pbl = 930 kg/m3) through a 130-m test pipeline (seelarger and more erratic which is indicative of the emptying cycleFig. 1) with the ratio between the solids mass fow and the airof the pneumatic conveying test.mass flow (m*) in the range of 51-70 and a total pressure dropIn investigating the structure of the pulsatile pressure fuc-range of217- 387 kPa. The cement meal was fed into the pipelinetuations within the steady state period, a closer examination ofusing a top discharge configuration in a 1-m3 blow tank with thethe test data was undertaken for each of the conveying tests; anair mass flow rate controlled via a bank of sonic nozzles (chokedexample for a test with pressure transducer locations T5 andT7flow array). A series of pressure transducers approximately 1 mis shown in Fig. 3. The pressure structure investigation showedapart were placed in the longest straight section of pipe. Subse-that for each test and each pressure transducer location, Ti, therequent testing on flyash and alumina was conducted in a 176-mwas indication of a fluctuation period (Tn). Also observed was apipeline.lower and upper bound of the pressure fluctuations, from whichFor each test, the pressure in the blow tank, the pressure atamaj中国煤化工e determined. It was also .the secondary air injection point and at selected locations alongappar(Otp(Ti- -Tj) of the pres-the pipeline were measured via Barksdale 0-100psi pressuresureCNM H Geach subsequent pessuretransducers, with the data being recorded at a frequency betweentransducer location, as is shown in Fig. 3 for Stp(Tr5- -T7). .20 Hz and 50 Hz, depending on the number of transducers usedWavelet analysis represented the next logical step: a win-during each test. The solids mass loss from the blow tank and thedowing technique with variable-sized regions. Wavelet analysisK.C. Wlliams et al 1 Particuology 6 (2008) 301- -306303500 ,Tolal (primary) ar pressure. Secondary air pressure (4 m)T5 (85.1 m)400T6 (86.1 m) .start of materialT7 (87.1 m)owi 300以MSteadystate200Pre-periodpressurisalionEmptyingand itialCyclesurge10030050000Time (s)Fig. 2. Example of the dfference in pressure behaviour betwcen the blow tank and the pulsatile behaviour in the pipelie.allows the use of long time intervals where we want more pre-been conducted and shows the signal being progressively bro-cise low-frequency information, and shorter regions where weken into five parts with the high frequency parts being removedwant high-frequency information. Like Fourier analysis consistsat each stage from d1 to d4 and finally a4 being a low fre-of breaking up a signal into sine waves of various frequencies, quency approximation of the trace. The a4 decomposition waswavelet analysis is the breaking up of a signal into shifted andsubsequently analysed to determine the pressure fuctuationscaled versions of the original (or mother) wavelet. The pressureamplitude.curves were plotted and a visual comparison of the wave shapewith the shape of the different mother wavelets, available in the4. General behaviour of pressure fuctuation structureWavelet Toolbox of Matlab7, was conducted to find a waveletwhich when shifted and scaled would best describe the pressureFrom the experimental program results, it was found thatrace.there was generally a 2-kPa pressure shift between the trans-The Daubechie db4 wavelet was found to be representative ofducer pressure at T5 and T7 with no significant variation inthe pressure trace and with a 4th level decomposition of the sig-pressure pulse period or major amplitude between transducernal, different parts of the trace frequency range were separated.traces. Consequently, the results detailed for the cement mealFor instance, consider Fig. 4, the 4th level decomposition has tests represent the average pressure fluctuation behaviour from807570Aaxm雷65{i60下引58中国煤化工可MHCNMHG2602626263264265 26667269 270Fig. 3. Pulsaile pessre fucutios at transducer locations 85.1 m and 87.1 m along the pipeline within the steady state conveying period.04K.C. Wlliamns et al. / Particwology 6 (2008) 301-306Decomposition机level4:5=a4+d4+d3. d2 d1.50080001500200008 250600 3500 4000 4500Fig. 4. Wavelet decomposition of signal at 86.1 m of a flyash test (m* =46.2).the pressure- time traces. Initially, the aim of this analysis wasThe superficial gas velocity was calculated using the ideal gasto establish some generalised behaviour of pressure pulse veloc-law at the average pressure drop values determined at T5-T7.ity, frequency and amplitude of the pressure fuctuations of theEven though there are major gas velocity fuctuations, the aver-cement meal with regards to variations in conveying parameters.age velocity from the pressure fuctuation data are similar tothe superficial gas velocity. The significance of the relationshipshown in Fig. 5 is that the average pressure fuctuation veloc-4.I. Velocityity can be represented by the superficial gas velocity. However,more analysis is required to determine the fluctuation in pulseFirstly, a basic analysis of the time delay difference was cal-velocity for each specific conveying condition.culated for nine different peak and trough pressure values withthe maximum, minimum and average time delay determined.4.2. FrequencyThe maximum, minimum and average pressure pulse velocitywas determined from the time delay and compared with theInitially, the frequency (Ff) of these pressure fuctuations wereaverage superficial gas velocity with the results shown in Fig. 5.determined from 10 different pulse periods (i.e.fp = 1/Tp) with20NOTE: abscissa and1ordinate error bars indicateminimum and maximumξ 12velocities84十中国煤化工MYTHCNMHG0Superficial gas velocity at T5-T7 location (m/s)Fig. 5. Comparison of TS-17 velocity behaviour of the pressure pulses and the superficial gas velocity.K.C. Wllians et al. /Paricology 6 (2008) 301- -30605.4 r.2+..8 tNOTE: abscissa error bars are themaxima and minima pressure values..4 tordinate error bars are the standard0.2-deviation of the frequency data.0 +0406(8010120140T5-T7 Average pressure (kPa)Fig. 6. Comparison of frequency of pesure pulses with pesure along the TS-T7 locations in the pipeline.mthe results compared with the average pressure across the trans-ducers as shown in Fig. 6. Although there is some increase in. 51.40|pulse frequency between the tests below 40kPa and the tests56.11greater than 60 kPa displayed in Fig. 6. Subsequent analysis ofthe pressure trace undertaken using a Fourier transform (e.g.- + 56.59Fig. 7) has shown that the there is no frequency above 5 Hz with米- 65.15some dominant frequencies between the 0Hz and 2Hz range,- 66.83which is in basic agreement with the work of Dhodapkar and- +- 69.71Klinzing (1993).B0i28486Pleline length (m)4.3. AmplitudeFig. 8. Results from wavelet. based analysis of pessre fucuation amplitudefor cement meal.Similarly to the wavelet analysis conducted by Pahk et al.(2006), it was also found that the Daubechie 4 mother wavelet and were in a straight section of pipeline away from any otherbest represented the peaked pressure time trace. An algorithm connections (Fig. 1).was developed which determined the peak and trough compo-More recently, amplitude fuctuation analysis in a 176-nents of the trace from which the average pressure amplitudeof m pipeline using flyash (dso= 14 pum，Pp= 2530kg/m',the gas pulses was calculated. Ullising the wavelet/algorithmPb1=810kg/m3) and alumina (dso = 78 μm, Pp = 3300kg/m',method, further analysis was conducted to determine if therePbl = 1050kg/m*) has also shown an increase in pressure fuctu-was any change in average pulse amplitude with changes in ation with an increase in distance from the start of the pipeline,pipeline length. Fig. 8 shows the results of this analysis which as shown in Figs. 9 and 10. The increase in pulse amplitude withclearly indicates that the pressure pulse amplitude increase inpipe length suggests that the gas expansion (or gas velocity)length from 81.5 m to 85 m. However some of the tests at 87 mwithin the pipeline is a major factor. Also noticeable is a bulkshow a decrease in amplitude from the previous transducer and material effect on the magnitude of the pressure pulse ampli-then generally increased in amplitude to the 88 m transducertude as there is a signifcant increase between the Flyash resultslocation. The reason for this decrease in pressure amplitude is (1.8- 4.4 kPa at 40 m and 6.5- 9.0kPa at 130 m) when comparednot immediately clear as the pressure trace at the 87 m and 88m to the Alumina results (6.5 9.5 kPa at 40m and 25 -32kPa atlocations were similar in peaked structure with the other traces140 m).m°+ 34.2-+37.- - 46.2女58.9. 72.9中国煤化工*- 85.1MHCNMH G。1315015Pieline length(m)Frequency (Hz)Fig. 9. Results from wavclet based analysis of pressure fuctuation amplitudeFig. 7. Fourier transform analysis example of the pressure- time trace at T6.for flyash in a 176-m pipeline.306K.C. Williams et al. / Particuology 6 (2008) 301- -30640 rn°quency approximation of the trace. From the wavelet analysis,crucial information on the amplitude of the trace was deter-30，mined which indicated that pressure fuctuation levels increase曼量z0with pipeline length and vary, depending on material type. Thisanalysis suggests that the gas expansion is a crucial factor in the- - 63.70-◆-70.7increase in amplitude of the pressure pulses through the pipeline.The resultof this analysis has given greater insight into the natureof the gas flow over the top of the dune surface, which repre-7090 110130 150sents the initial stage in understanding the complex interfacePipe length (m)between the high velocity gas and both the material dune andFig. 10. Results from wavelel-based analysis of pessre fuctuation amplitudethe interstitial gas flow within the dune.for alumina in a 176-m pipeline.References5. SummaryDhodapkar, s. V, & Klinzing, G. E. (1993). Pessre fuctuations in pneumaticconveying. Powder Technology, 74. 179-195.Analysis was conducted on the pressure fuctuation behaviourKonad, K., Harison, D, Nedderman, R. M, & Davidson, J. F (1980). Predic-of the gas stream above the moving dunes of powders in densetion of the pressure drop for horizontal dense phase pneumatic conveying ofphase pneumatic conveying. The pipeline used were 130 m andparticles. In Proceedings of the fjfh intemnational conference on pneumaticfransport of solids in pipes (Pneumotransport 5) (pp 225- 244).176 m long with an intemal diameter of 53 mm. The struc-Legel, D, & Schwedes, J. (1984). Investigation of pneumatic conveying of plugsture of the gas fuctuations showed features of pulsatile motion,of cohesionless bulk solids in borizontal pipes. Bulk Solids Handling, 4(2),frequency and amplitude. From the time delay analysis, a rela-399-405.tionship between the average pulse velocity and the superficial L,H.(2002). Aplication of wavelet muli resolution alysistoessre fuctugas velocity was shown. However, the variations in gas pulseations of gas-solid two-phase flow in a horizontal pipe. Powder Technology,125, 61-73.velocity and frequency suggested that the pressure pulses inPahk, J. B.. Sahin, I,. & Klinzing, G. E. (2006). Pressure fuctuations in asessnsgfluidised dense phase conveying are dominated by dynamicflow regimes in pneumatic conveying of polymer pellets. In Proceedings forrather than rhythmic oscillations. Consequently, a decomposi-ChoPs 05.tion of the pressure trace was conducted using a Daubechie db4 Pan, R., & Wypych,P W.(1997). Pessure drop and slug velocity in low-velocitywavelet-based analysis, which provided an excellent low fre-pneumatic conveying of bulk solids. Powder Technology, 94, 123 -132.中国煤化工MYHCNMHG