00428d01 | Pump | Classical Mechanics

of 6

Please download to get full document.

View again

All materials on our website are shared by users. If you have any questions about copyright issues, please report us to resolve them. We are always happy to assist you.
PDF
6 pages
0 downs
5 views
Share
Description
Rudolf Bachert1,2 AUMA Riester GmbH & Co. KG, Aumastrasse 1, 79379 Muellheim, Germany e-mail: rudolf.bachert@auma.com Unsteady Cavitation at the Tongue of the Volute of a Centrifugal Pump The paper deals with unsteady effects of cavitation at the tongue of the volute of a centrifugal pump. For the investigations parts of the volute casing, including the tongue and the hub of the impeller, were made of acrylic glass. Experiments were carried out at a flow rate above optimal value (slight overload
Tags
Transcript
  Rudolf Bachert 1,2 AUMA Riester GmbH & Co. KG,Aumastrasse 1,79379 Muellheim, Germanye-mail: rudolf.bachert@auma.com Bernd Stoffel Former HeadLaboratory for Turbomachinery and Fluid Power,Darmstadt University of Technology,Magdalenenstrasse 4,64289 Darmstadt, Germanye-mail: bernd.stoffel@fst.tu-darmstadt.de Matevž Dular 2 Laboratory for Water and Turbine Machines,University of Ljubljana,Askerceva 6,1000 Ljubljana, Sloveniae-mail: matevz.dular@fs.uni-lj.si Unsteady Cavitation at the Tongueof the Volute of a CentrifugalPump The paper deals with unsteady effects of cavitation at the tongue of the volute of acentrifugal pump. For the investigations parts of the volute casing, including the tongueand the hub of the impeller, were made of acrylic glass. Experiments were carried out at a flow rate above optimal value (slight overload) and at 3% head drop conditions. In thisoperating point there was no cavitation present in the impeller of the pump, hence, thewhole 3% head drop resulted from cavitation on the tongue of the volute. By use of  particle image velocimetry combined with special fluorescent particles it was possible toobtain information about the velocity field outside and inside the cavitating zone. Anadditional camera provided information about the location and extent of cavitation. Theresults imply that cloud cavitation similar to the one seen on single hydrofoils appears onthe tongue. Periodical evolution of cavitation structures, from incipient to developed,with cavitation cloud shedding, is seen during each passing of a blade. The Results implythat greater consideration should be given to the possibility of cavitation appearance onthe tongue of the volute as it is possible that this cavitation location alone causes the 3%head drop. Moreover, the appearance of unsteady cavitation in a higher-pressure region,such as the volute of the pump, can cause severe erosion to the solid surfaces. ͓ DOI: 10.1115/1.4001570 ͔ Keywords: cavitation, particle image velocimetry, centrifugal pump, tongue, volute 1 Introduction To avoid or to reduce the effects of cavitation by design andoperation measures, there is a persistent need of improving thedetailed understanding of the physical phenomena underlying itsharmful effects. It is well known that flow separation often occursin the spiral casing of centrifugal pumps, particularly in the vicin-ity of the tongue, when they are operating at off design conditions,i.e., flow rates lower or higher than the optimal ͓ 1 ͔ . Flow separa-tion, characterized by unsteady shedding of vortices from thetongue, can also be an important source of vibration and noise ͓ 2 ͔ .The other type of instability, common to hydraulic turbomachines,is cavitation, where the pressure locally drops to and below thevapor pressure of the liquid. Occurrence of cavitation in hydraulicmachines leads to problems such as shock waves, noise, and dy-namic effects that lead to decreased equipment performance and,frequently to equipment failure ͓ 3 ͔ . In the case of centrifugalpumps, large attached cavities preferably form on the suction sur-faces of the blades near their leading edge ͓ 1 ͔ .The present study deals with a special case of separated flow ina pump operating at overload conditions, where cavitation at thetongue of the volute casing occurs prior to the cavitation on theblades. In this case the standard threshold level of 3% head drop ͓ 4 ͔ due to cavitation does not come from cavitation in the impellerbut from the cavitation on the tongue of the volute. Three reasonsthat have not yet been thoroughly investigated are probably re-sponsible for this phenomenon.1. The pressure rise generated by the impeller decreases moreand more with an increase in flow rate ͑ see Fig. 5 ͒ . Thismeans that the static pressure in the vicinity of the tongue atoverload conditions is lower, making the system more proneto cavitation.2.At specific flow rates q ͑ q=Q / Q opt ͒ bigger than optimal ͑ q Ͼ 1 ͒ , the direction of the absolute velocity at the blade trail-ing edge deviates from the angle of zero incidence of thetongue shape. This leads to an incidence angle, which in-creases with increasing flow rate. Because the flow angle ismostly larger than the angle of zero incidence of the tongue,the stagnation point shifts from the leading edge of thetongue to a location on the tongue side oriented toward theimpeller ͑ inner side ͒ . A low pressure zone develops on theopposite side, which is oriented toward the volute dischargeand cavitation forms ͑ Fig. 1 ͒ .3. In addition, the passing blades of the impeller create a peri-odically varying velocity and pressure field in the vicinity of the tongue. This unsteady flow field is prone to cavitationand has a distinct effect on various cavitation phenomena atthe tongue. For example, it is well known that unsteadycavitation is the most erosive type of cavitation ͓ 5 ͔ , hence, itis not surprising that often local erosion damage is found atthe tongue of centrifugal pumps after they have been oper-ated at overload conditions for elongated periods of time ͓ 1 ͔ .The investigations of velocity field in the vicinity of or insidecavitation is a complex task. Different investigations with the aidof probe measurements were performed in the past ͓ 6 ͔ . With thismethod the velocity can be measured only in one point at a time,hence, only an average velocity field can be determined. Theproblem can be solved by particle image velocimetry ͑ PIV ͒ as firstshowed by Zhang et al. ͓ 7 ͔ . They employed PIV to examine theturbulent flow in the wake of an open partial cavity. Laberteauxand Ceccio ͓ 8 ͔ measured velocities in developed cavitating flowwith vapor cloud separation. However, they were unable to obtainthe information about velocity field inside the cavitation itself.The problem they faced was overexposure of the particles addedto the water ͑ they were not visible since the vapor structure re- 1 Corresponding author. 2 Previous address: Laboratory for Turbomachinery and Fluid Power, DarmstadtUniversity of Technology, Magdalenenstr. 4, 64289 Darmstadt, Germany.Contributed by the Fluids Engineering Division of ASME for publication in theJ OURNAL OF F LUIDS E NGINEERING . Manuscript received October 15, 2009; final manu-script received March 23, 2010; published online May 19, 2010. Assoc. Editor:Olivier Coutier-Delgosha. Journal of Fluids Engineering JUNE 2010, Vol. 132 / 061301-1Copyright © 2010 by ASME Downloaded 27 Jul 2010 to 193.51.48.34. Redistribution subject to ASME license or copyright; see http://www.asme.org/terms/Terms_Use.cfm  flects much more light ͒ so the velocity field inside the cavitationstructure could not be measured. Recently, a promising methodbased on X-ray absorption and phase-contrast enhancement,where the velocity of both liquid and vapor phase could be mea-sured, was introduced by Vabre et al. ͓ 9 ͔ . The method applied inthe present study, which allows flow velocity measurements out-side and also inside the vapor structure is a combination of PIVand fluorescent particles. The technique was already successfullyapplied to simpler geometries such as hydrofoils ͓ 10,11 ͔ .During the experiment, the pump was operating at overloadflow rate and cavitation condition leading to 3% head drop. Visualaccess to the whole flow field enabled confirmation that no cavi-tation is present within the impeller. Also, detailed observations of the cavitation structures on the tongue of the volute casing weremade.The main aim of the study was to measure the velocity field inthe vicinity of the tongue of the volute casing to check the hy-pothesis of the high angle of attack on the tongue of the spiralcasing and the unsteady flow conditions in the region. For thispurpose, PIV was applied at the tongue of a centrifugal pump.Results show that the velocity direction at the blade trailingedge indeed deviates from the angle of zero incidence of thetongue shape, oscillates significantly and that this indeed causesappearance of cavitation on the tongue that results in the 3% headdrop of the pump. 2 Experimental Set-Up Experiments were set up in a cavitation tunnel at the Laboratoryfor Turbomachinery and Fluid Power in the Darmstadt Universityof Technology. 2.1 The Pump and the Test Rig. The radial impeller of thetest pump ͑ Fig. 2 ͒ has a three-dimensional blade geometry typicalfor standard centrifugal pumps produced by pump manufacturers.It has six blades, an outer diameter of 260 mm, and the specificspeed is n s =26 min −1 .n s = ␻  ͱ  Q ͱ  4 h 3 ͑ 1 ͒ where ␻  ͑ in min −1 ͒ is the rotational frequency, Q ͑ in m 3 / s ͒ is theflow rate at best efficiency point and, h is the corresponding head ͑ in m ͒ .The rotational frequency of the impeller was 2000 min −1 dur-ing the experiments for which the volume flow rate Q at the bestefficiency point ͑ Q=Q opt ; q=1.0 ͒ is 143 m 3 / h. To provide op-tical access to the impeller blades, the front shroud of the impellerwas made of acrylic glass. Experiments were performed at over-load conditions at q=1.17, where the value of net positive suctionhead ͑ NPSH ͒ was held constant at NPSH=NPSH 3% =4 m.NPSH=p i − p v ␳  · g+c i2 2g ͑ 2 ͒ where p i is the static pressure at pump inlet, p v is the vapor pres-sure, and c i is the absolute velocity at pump inlet.In Fig. 3, the cross-section through the pump is shown. Thesuction nozzle, the suction sided cover plate of the casing and thevolute with the tongue were made of transparent acrylic glass.This enabled optical access to the outlet of the impeller to thetongue and to the volute discharge in the axial direction. This wasalso the direction for the digital camera sight, which was used forthe velocity measurements. The laser light sheet is brought inperpendicularly to the axis of rotation through a window in theouter casing wall.The closed test loop used for the investigations on the testpump is shown in Fig. 4. The test rig and the test pump are Fig. 1 Schematic representation of the flow field in the vicinityof the volute tongueFig. 2 Impeller of the test pump with transparent front shroudFig. 3 Cross-section through the pump installation with notedview point of cameras 061301-2 /  Vol. 132, JUNE 2010 Transactions of the ASME Downloaded 27 Jul 2010 to 193.51.48.34. Redistribution subject to ASME license or copyright; see http://www.asme.org/terms/Terms_Use.cfm  equipped with the necessary instrumentation to determine the op-erating conditions ͑ rotational frequency n, shaft torque T, flow rateQ, and pressures at the pump inlet and outlet p s1 and p s2 ͒ of thetest pump.The flow rate was measured with an inductive flow meter Fis-cher and Porter type D10D, DN 150 with an uncertainty of  Ϯ 3.5 m 3 / h. The system pressures p s1 and p s2 were measuredwith sensors JPB-type 304 at four pressure taps on the pipe andthen averaged. The uncertainty of the measurements was Ϯ 20 mbar. The fluid temperature was measured by Jumo PT100sensor with the uncertainty of  Ϯ 1.2 K. The rotating frequencywas determined by a proximity sensor mounted on the pumpshaft. The torque on the shaft was measured with the SteigerMo-hilo 02FE with the uncertainty of  Ϯ 1 Nm. Considering the com-bination of inaccuracies of pressure, flow rate, rotation frequency,and temperature measurements, the pump operating conditioncould be determined within Ϯ 1.7% of the measured value with aconfidence level of 95% ͓ 12 ͔ .The quality of the water was monitored by measuring the gascontent by an apparatus based on the Van Slyke method with anuncertainty of 1%. The water used for the experiments was almostsaturated with gasses ͑ more than 50 mg of dissolved and undis-solved gas per liter of water ͒ so that the effects of the tensilestrength of the water were reduced to the minimum possible level.This condition is necessary since the variations in water qualitycan greatly influence the cavitation behavior ͓ 13 ͔ .Integral characteristics of the pump ͑ q-H diagram at NPSH=NPSH 3% =4 m on the left and NPSH-H diagram at q=1.17 onthe right ͒ are show in Fig. 5. The operating point in which theexperiments were performed is also indicated.Figure 6 shows the intake of the pump during operation atoverload conditions ͑ q=1.17 ͒ and at NPSH=NPSH 3% =4 m. Thecamera position for the image in Fig. 6 is schematically presentedin Fig. 3. The intake and the cover plate of the casing are bothmade out of transparent acrylic glass but since they are perpen-dicular to each other the light is refracted. In the bottom of theimage in Fig. 6, we see the suction eye of the impeller and in theupper part we see the outer part of the blades with the bladetrailing edges. As already mentioned, almost no cavitation can beseen on the blades. Some bubbles are present but their srcin isthe cavitation in the gap between the housing and the impeller. Soit can be concluded that no cavitation is present on the blades of the impeller at the investigated conditions of operation. 2.2 Velocity Measurements. For the velocity measurements,the region of interest extended over the tongue of the volute. Theillumination was provided by a vertical laser light sheet ͑ Nd-YAG-Laser ͒ approximately 1 mm thick and perpendicular to theobservation axis. The position of the light sheet was at half widthof the tongue, 24 mm from the front acrylic glass wall. Specialfluorescent tracer particles ͑ PMMA-RhB-Partikel-G029, mean di-ameter of 10 ␮  s with standard deviation of 0.25 ␮  s ͒ were addedto the water. The particles receive light from the laser at a wave-length of 532 nm ͑ green spectrum ͒ and emit light at a wavelengthof 590 nm ͑ yellow spectrum ͒ . By fitting the CCD cameras with anappropriate light filter ͑ that filters the visible light but lets the lightin yellow spectrum trough ͒ it is possible to get suitable images of the tracer particles for the PIV analysis. Since the camera records Fig. 4 Test loop for pump testsFig. 5 Integral characteristics of the pump „ q-H diagram atNPSH=4 m—left and NPSH-H diagram at q=1.17—right … Fig. 6 View into the impeller at the investigated operatingpoint „ q=1.17, NPSH=4 m … Journal of Fluids Engineering JUNE 2010, Vol. 132 / 061301-3 Downloaded 27 Jul 2010 to 193.51.48.34. Redistribution subject to ASME license or copyright; see http://www.asme.org/terms/Terms_Use.cfm  only the light in the yellow spectrum, the cavitation structure isfiltered out of that image and tracer particles inside it can also bedetected. Despite less particles are detected inside the cavity avelocity field could be determined if the position of the light sheetwas close enough to the front observation window ͑ a more de-tailed description of the method can be found in previous paper bypresent authors ͓ 10 ͔͒ .The camera and the laser shots with a duration of 10 ns weretriggered by the control unit of the PIV-system. For the PIV analy-sis, two images with 30 ␮  s time delay were recorded. Using thestandard DANTEC DYNAMICSTUDIO software a cross correlationof the two images of tracer particles was made. The image size is1280 ϫ 860 pixels, corresponding to 85.9 ϫ 57.7 mm 2 ͑ 6.7 ϫ 10 −2 mm / pixel ͒ . A low-pass Gaussian subpixel interpolationwas used for the determination of the correlation peak. The size of the interrogation area was 16 ϫ 16 pixels; the overlapping was50% leading to distance of approximately 0.5 mm between thevectors. About 1% of the vectors in the region of interest wererecognized as invalid “bad” vectors and were substituted. Thedominant error in PIV measurements is usually the bias intro-duced by the subpixel peak finding algorithm. Taking all the un-certainties into consideration, we estimate a nominal particle im-age displacement error of about 0.1 pixels. Considering thevariations of the velocity magnitudes ͑ up to 30 m/s ͒ , i.e., particleimage displacement ͑ up to 1 mm or about 12 pixels ͒ , in the mea-sured sections, the accuracy of the instantaneous velocity can beestimated to less than 1% ͓ 14 ͔ . 2.3 Cavitation Image Capturing. Additionally to the PIVmeasurements another CCD camera ͑ SensiCam with sensor CCD-Interline Progressive Scan ͒ was used to capture images of cavita-tion in various parts of the pump. This gave a better perspective of the flow conditions in the system. Images were captured at 8 bitresolution. The size of captured image was 860 ϫ 1280 pixels.The illumination was provided by a stroboscopic light triggeredby the proximity sensor. 3 Results 3.1 Visualization. In Fig. 7, the cavitation at the tongue of thevolute is shown.The left image shows the view through the window in the cas-ing radially inwards to the axis of rotation ͑ this window was pri-marily installed for laser illumination for PIV measurements ͒ . Inthis image the movement of the impeller is from bottom to up.The leading edge of the tongue is located horizontally a bit belowthe middle of the image ͑ marked by a black line in the image ͒ . Atthe moment when the image was taken, the blade trailing edge had just passed the tongue leading edge and is also visible through thetransparent tongue ͑ marked by a black line in the image ͒ . Theright image shows the appearance of cavitation on the tonguefrom the side view ͑ through the transparent front cover of thecasing—the viewpoint, which was also used for the PIV measure-ments ͒ . Again, the blade trailing edge can be seen as it just passesthe tongue leading edge.The appearance of cavitation closely resembles that on a singlehydrofoil in a cavitation tunnel ͓ 10 ͔ . The essential difference isthat in the case of a single hydrofoil, the cavitation dynamics isdominated by periodical, self-excited cloud generation—the fre-quency of which is related to the flow velocity and the Strouhalnumber ͓ 15 ͔ . In the case of cavitation on a tongue of the pump,the pressure gradient from the suction side to the pressure side of the blade in combination with the wake of the blade instead causethe periodically varying of flow conditions at the tongue ͓ 2 ͔ . Thisresults in periodical cavitation cloud generation with a frequencythat corresponds to the blade passing frequency, which is illus-trated in Fig. 8. The tongue was made out of transparent materialand is hard to see in the images. Therefore, it is highlighted by ablack line in the first image. The series of images shows onetransition of the blade, where the whole cycle of the cavitationevolution can be seen.In the first image the blade just moved out of the view of thecamera. We can see a larger cavitation structure on the tongue,somewhat downstream of its leading edge. As will be shown laterfrom the PIV results, the angle of incidence of the flow on thetongue is closest to optimal at this instant—no attached cavitationon the leading edge can be seen but a cloud generated at previousblade passing remains present.Cavitation is barely visible in image No. 2. Actually only aseparated cavitation cloud with relatively low volume fraction of vapor ͑ hence, hardly visible ͒ can be sensed further downstream of the leading edge of the tongue.As the blade starts to approach the tongue the flow conditionsbecome less and less optimal ͑ image No. 3 ͒ . A large attachedcavitation structure grows on the leading edge of the tongue.Meanwhile the separated cloud has left the region of interest.Image No. 4 shows the growth of the attached cavity while itsseparation from the tongue can be seen in image No. 5. The shape Fig. 7 Cavitation on the tongue of the volute „ front view—leftand side view—right … Fig. 8 A sequence showing evolution of the cavitation struc-ture as the blade passes the tongue 061301-4 /  Vol. 132, JUNE 2010 Transactions of the ASME Downloaded 27 Jul 2010 to 193.51.48.34. Redistribution subject to ASME license or copyright; see http://www.asme.org/terms/Terms_Use.cfm
Related Search
We Need Your Support
Thank you for visiting our website and your interest in our free products and services. We are nonprofit website to share and download documents. To the running of this website, we need your help to support us.

Thanks to everyone for your continued support.

No, Thanks