An Experimental and Numerical Investigation into Flow ...

AN EXPERIMENTAL AND NUMERICAL INVESTIGATION INTO FLOW PHENOMENA LEADING TO WASTEWATER CENTRIFUGAL PUMP BLOCKAGE

Connolly R.*1, Breen B1. and Delauré Y2. 1Sulzer Pump Solutions Ireland Ltd.,

Wexford, Ireland. 2School of Mechanical and Manufacturing Engineering,

Dublin City University, Ireland. E-mail: [email protected]

ABSTRACT The work covered in this paper had the objective of

investigating key factors in single vane pump impeller blockage using both CFD and experimental analysis. Single vane centrifugal pumps can be found in wastewater applications, where suspended solids, fibers and other flexible material can build up on the pump impeller and cause blockage and failure. It is in the pump manufacturer’s interest to design blockage resistant components while still maintaining hydraulic efficiency. Testing was conducted on a large variety of centrifugal wastewater pumps at different operating points in a purpose built test rig with suitable test material. A pump which had varying blockage performance depending on operating point was chosen for further study with CFD. Transient simulations with the SA-DDES turbulence model were carried out with the same operating conditions as in the experimental analysis. CFD has been used to highlight flow features likely to lead to such blockage, while experiments provided some insight into the significance of certain key parameters. An analysis of the CFD results showed a significant correlation between pump blockage performance and radial velocity components within the fluid domain, specifically in the impeller region. Sensitivity to blockage was found to vary with the head required of the pump. This sensitivity was further explained when applying the hypothesis of the impact of velocity components on the likelihood of pump blockage.

INTRODUCTION

In recent decades there has been significant improvement in access to water sanitation [1]. With increasing global populations, the demand has never been higher for reliability in wastewater systems. Wastewater pumps are integral components of wastewater systems, featuring in collection transportation and treatment processes and as a major contributor to system failures, they are a key concern for reliability [2]. A common cause of waste-water pump failure is related to soft blockage, where fibrous material catches on the impeller and builds up, leading to high motor current and thermal overload. Single vane impellers in both open and closed configurations allow for larger channel size and easier passage of larger solids. Framework agreements such as [3] insist on a minimum solids passage size for installations in their jurisdiction. Many waste-water pump manufacturers offer a single vane impeller in order to meet requirements regarding minimum solids passage and blockage performance.

NOMENCLATURE

Volumetric flow rate Blockage index Impeller blade number

Rag incidence A Cross sectional area length Rotational speed Head Pressure Density Acceleration due to gravity Elevation Velocity

non dimensional cell wall distance Subscripts & Superscripts

radial tangential

Leading edge rotational

An example of a single vane pump hydraulic can be seen in Figure 1. Significant experimental research relating to impeller blockage has been carried out during hydraulic development of new products at Sulzer Pumps. While CFD has been commonly used in the design stage for hydraulic performance, recently experimental research has been increasingly supplemented by CFD in order to further understand how flow features affect impeller blockage.

13th International Conference on Heat Transfer, Fluid Mechanics and Thermodynamics

925

Figurecomponents.

EXPERIMENTALAn

orderaccuratelyfield.

TheMcEvoy,coveringvolumused,

wereintroductioneach the pump.blockagepassed,

=Pumps

wastewaterIn

high passingof recordingmaterialin addition

Infield material.and samplesclothof sanitarywastesamples

Arangeincludedoutletaccuratelybeing

Figure 1. Typicalcomponents.

EXPERIMENTALAn acceptable

order to test a accurately reflectsfield.

The test methodMcEvoy, [4]. Testscovering typical operatingvolumetric flow rateused, as well as at

% and were repeated at eachintroduction of 10

flow rate, basedpump. The average

blockage index (BI).passed, to the

= × 100. Pumps tested

wastewater applicationIn order to supplement speed camera

passing the rag andrecording at 1000fps,

material interacts addition to the forces

In order to reliably conditions,

material. An analysissamples taken

cloth of non-wovensanitary towels

waste-water constituentssamples taken locally

A test rig wasrange of pumps, 1.3kWincluded flow-metersoutlet pipes (or did

urately establishbeing tested. A pinch

Typical single vane

EXPERIMENTAL acceptable and realistic

wide range reflects the conditions

method used in thisTests were repeated

operating rangerate at the Best

at flow rates 30%% respectively.

each flow rate 10 blockage samples.based on the percentage

average of these(BI). The BI is

total number

tested were of standardapplication, as would

supplement andcamera was used

nd blocking. T1000fps, provides

with the fluidforces which act

reliably carry out it was necessary

analysis of blockagestaken for classification.woven synthetic fibres,

towels and disposableconstituents has locally [5]. was designed and

1.3kW to 90kW.meters and pressure

did you meanestablish the duty point

pinch valve wa

vane waste-water

realistic test method of pumps with

ditions experienced

this study wasrepeated over three different

range of wastewaterBest Efficiency30% above and

respectively. Three with each test

samples. A scorepercentage of samples

three flow rates defined as the

number of samples

standard type usedwould be seen in the

and further inform to record theThe camera, which

provides some insightfluid and the pump

act upon it. out a test that is

necessary to selectblockages was carried

classification. Most blockagesfibres, typically

disposable domestic wipes. found results

and built in order90kW. The test hardware

pressure transducersmean something

point at whichwas located between

water pump hydraulic

method was neededwith material

experienced by pumps in

was based on thatdifferent flow

wastewater pumps. Efficiency Point,

and below this value, independent

test run involvingscore was assigned

samples passedrates gave the pumpthe ratio of samples

samples introduced,

used in municipalthe field.

inform the research,the hydraulic which was capable

insight into how thepump impeller surface

is representativeselect a suitable

carried out on three blockages contained

typically the outer casingwipes. A study

results consistent

order to test a largehardware on the

transducers on both inlet else) in order

which the pumps between the pressure

hydraulic

needed in that in the

that of rates The was

value, tests

ving the assigned for

passed by pump

samples introduced, ,

municipal

research, a both

capable the test surface

representative of test sites

contained casing

study of with

large tank

inlet and order to

were pressure

tappingPerspexpointvisualwithcamerabehaviourflow,

EXPERIMENTAL

prominentpumpresearchlargewithrates.comethetheprocesseithermovedleadingFigurethesuctiononFigureacrosssurfacesimpellerpressurethanpinningin

observation

tapping and thePerspex windowspoint of the testvisual inspectionwith the pump camera in the blockagebehaviour of eachflow, low flow and

EXPERIMENTAL

McEvoy [4]prominent typepump in wastewaterresearch on the large number ofwith the high speedrates. When passingcome through thethe impeller trailingthe volute via theprocess of leadingeither side of themoved outwardsleading edge blockageFigure 2 showsthe suction surfacesuction and pressureon entering the Figure 2. [i]. If across the streamlinessurfaces then it impeller surface,pressure and suctionthan the impellerpinning the rag

equilibrium, it

Figure 2: Stagesobservation

the flow-meter towindows were located

test pump and inspection and recording

suction streamblockage test rig

each of the shortand best efficiency

EXPERIMENTAL RESULTS

[4], states that leading of blockage

tewater application. contributing factorsof observations

speed camera equipmentpassing through

the inlet, alongtrailing edge, aroundthe discharge after

leading edge blockagethe impeller leading

outwards quickly enough.blockage was observed

how the rag surface via the inletpressure surfaces,

pump can havethe rag is found

streamlines towards is likely to block,

surface, it is alreadysuction surface

impeller rotational on either sideit is likely to remain

Stages of rag

Inlet

to maintain thelocated directly

at the tank sides.recording of rag behaviour

stream and impeller.rig was used to

shortlisted sampleefficiency point.

RESULTS

leading edge found in this

application. It was decidedfactors to leading

observations were madeequipment at the the pump, the

along the impeller around the leadingafter one or two

blockage was theleading edge as enough. The actionobserved and recreated is seen to enter

inlet. With the flowsurfaces, it was noted

have an impactfound to be simultaneously

towards both the pressureblock, [ii]. As the

already being streamed by the fluid flow,

rotational speed, [iii].side of the impeller

remain stuck in

rag blockage based

flow

the desired flow below the couplingsides. This permitted

behaviour as it interactsimpeller. The high

to compare blockagesample materials

blockage is thethis type of centrifugal

decided to focusleading edge blockage.made of pump blockage

the specified testthe rag was observed

suction surface,leading edge andtwo rotations. A

the rag being a result of notaction of consistent

recreated in 3Denter the pump towards

flow split betweennoted that rag orientation

ct on blockage,simultaneously distributed

pressure and the rag approachesstreamed alongflow, which is

[iii]. If pressure impeller leading edge

in this position

based on experimental

flow rate. coupling

permitted interacts

high speed blockage at high

the most centrifugal focus this

blockage. A blockage test flow

observed to surface, past

and out of A typical caught

not being consistent 3D CAD.

towards between the orientation

blockage, as per distributed

suction approaches the

along both is slower

forces edge are

position [iv].

experimental


926

Figuredimensionalised

Average

relativeThis influenceincrease

The

Figure 2 Averageddimensionalised flow

Average results

relative flow rate and can be explained

influence the blockage.increase in proportion

The volumetric

velocity overand trailingpump stronglyrag out of

The tangential. If the

into the tangentialit is less likelyit is fullyvolute region.from purely

The opposite effect The area

and trailingflow ejectionthe impeller.an effect smaller gap.hydrodynamicis difficult

The lengthinterior partthe volutecirculatingleading edgeimpeller canto a rag more likelytrailing end

Averaged blockageflow rates

results indicatedand Blockage

explained by a numberblockage. The anti

proportion to:

volumetric flow rateover the impeller

trailing edge strongly influenceof the impeller

tangential flow velocitythe flow transporting

volute past thetangential velocity than

likely to be caughtfully transported

region. This howeverpurely geometric

effect should occur of the gap

trailing edge: the smallerejection velocity

impeller. The number as an increase

gap. however,hydrodynamic performance,

difficult to qualify. length of the rag,

part of the impeller,volute it finds itself

circulating around the edge of the impeller.can catch up with

wrapping itselflikely to occur for

end is kept back

blockage performance

indicated that a correlation Index does exist

number of keyanti-clogging performance

rate or the averageimpeller opening betwe

. High flowinfluences the speed

region into thevelocity of the

transporting the ragthe trailing edge

than the speed ofcaught up by the into the rotating

however is difficult consideration.

occur by increasing between the

smaller this area which helps to

number of blades increase in is likelyhowever, also impacts

performance, but the

rag, : as the ragimpeller, past the

itself in a flow volute at lower

impeller. If thiswith the rag, eventually

itself around thefor a longer ragback by interaction

performance at multiple

correlation betweenexist (see Figure

key factors shownperformance should

average radial between its leading

flow rate through of ejection of

the pump volute.the impeller opening

rag as it is ejectededge is at a higherof the leading edge,the impeller before

rotating flow ofdifficult to quantify

consideration.

increasing: the impeller leadingarea the stronger

to clear the rag may also

likely to lead impacts on the general

effect in this

rag flows out ofthe trailing edgeflow which maylower speed thanthis is the case

eventually leadingthe impeller. Thisrag in particular

interaction with the

non-

between Figure 3). shown to

hould

flow leading

through the of the

volute. opening ejected higher edge,

before f the

quantify

leading stronger the

rag off have to a

general case

of the e into

may be than the case the leading This is

if its inlet

clogginganddefined

pumpdoand

pump

dependencealso(seecomplexorienta

wall orwall byfriction

The rotationalimpellerto catchthe impellerperformancevelocitythe antidifficultvolute effects statisticalmay fail

To summarize

clogging shouldand inversely todefined as follows

An average

pump families testeddo confirm thatand blockage index

Figure 3 Ragpump families

Although

dependence of also be recognised(see Figure 4-5complex interactionsorientation at pump

0

20

40

60

80

100

120

1.00

Bloc

kage

Inde

x

or the wear plate.by pressure or

friction against the wallrotational speed

impeller for a fixed inletcatch the rag. An

impeller can performance affectinvelocity at the exit from

anti-clogging difficult to quantify

and impeller on the flow.

statistical analysis focussingfail to show strong

summarize it can beshould increase in

to . A non-dimensionalfollows:

=

average of the experimentaltested in this

that an inverse correlationindex [Figure 4]

Rag Incidence vs.

ensemble averages the Blockage

recognised that there5). Blockage Index

interactions of rag pump entry.

1.50 2.00Rag Incidence

plate. It may beor exposed to shearwall surface.

speed of the impellerinlet flow rate, increase in the also alter

affecting the ratio offrom the impeller

performance. quantify without takingimpeller geometries

This would suggestfocussing on

strong correlation.

be expected that proportion to

dimensional rag

=ℎ

experimental results study at threecorrelation between

4] exist.

vs. Blockage

averages suggestBlockage Index on Rag

there is significantIndex does not

and wear plate

2.00 2.50Rag Incidence

be pinned againstshear stresses

impeller : the fasterrate, the more likely

the rotational speed the hydrodynamicof radial to tangential

impeller which is This however

taking into account and details ofsuggest that a the rotational

correlation.

that the likelihoodto , and possiblyrag index can then

results of three differentthree different flow

between rag incidence

Index for 3 different

suggest a non-negligiRag Incidence,

significant scatter in thenot account for

plate or variance

3.00 3.50

against the due to

faster the likely it is

speed of hydrodynamic

tangential is key to

however is account the

of their a purely

rotational speed

likelihood of possibly

then be

different flow rates incidence

different

negligible it must the data

for more variance in rag

Family A

Family B

Family C


927

Figureat each

It

rag movesfluid,it canOncebe movedimpeller

NUMERICALA

experimentalrotatingchallengesflowssteadyin 1999centrifugalcomputationalfluctuationstudythe interactionThis Allmarasused

A

wall and foundpressuremodelproximityare splitused minimumand velocitypump

0.00

1.00

2.00

3.00

4.00

5.00

6.00

Rag

Inci

denc

e

Figure 4 Blockageeach test flow rate

It can be hypothesisedmoves through

uid, it can be interceptedcan be carried to

Once it has becomemoved easily as

impeller rotation ensure

NUMERICAL METHODA numerical

experimental resultsrotating equipmentchallenges. The typeflows which are highlysteady models are

1999 on the highlyntrifugal pump,

computational resultsfluctuation by as study it is important

interaction with means having

Allmaras Detached in this study in

A hexahedral mesh adjacent mesh

of 300 onfound that a similarpressure coefficientmodel [9]. The meshproximity refinement

split 8 times. in the inflation

minimum of 6 impellervelocity values

pump head with Eq.8.

=

0.00

1.00

2.00

3.00

4.00

5.00

6.00

0 20

Blockage Index vs.rate

hypothesised fromthrough the pump driven

intercepted by theto a radial position

become trapped onas forces drivenensure adhesion

METHOD approach was

results and interpretationsequipment by CFD

type of pump consideredhighly time dependent

are generally nothighly transient

pump, while results which

much as 30%.important that the transient

with the moving having to rely on a

Detached Eddy Simulationin order to better

mesh which containedmesh characterised

on the impellersimilar resulted

coefficient predictionsmesh contains

refinement at the impeller 4 prism layers

inflation layer. Theimpeller revolutions

values over the lastEq.8.

−+ −

20 40

Blockage Index

vs. Rag Incidence

from the above driven by the throughthe impeller leading

position outside ofon the leading edge

driven by both theadhesion to the impeller.

was adopted interpretations. The

CFD presents considered in this

dependent so thatnot suitable. Kaupert

pressure field DeSouza [7]

had pump 30%. In the case

transient processes impeller are correctly sliding mesh

Simulation turbulencebetter resolve turbulent

contained ~65 characterised by

impeller [8]. A reviewresulted in accurate

predictions with an identical 4mm cells in

impeller in fourlayers with a growth

The pump wasrevolutions and the

last rotation were

− +−

2

60 80

Blockage Index

Incidence for full data

results that asthrough flow ofleading edge beforeof the leading edge.edge the rag is

the fluid inertiaimpeller.

to help confirmThe pumps and other

several specificthis study exhibits

that simpler quasiKaupert [6] reported

field within a 7 blade[7] demonstrated

periodic pressurecase pump blockage

processes characterisingcorrectly captured. model. A Spalart

turbulence model wasturbulent vortices.

× 105 cells with of 116,

review of literatureaccurate prediction

identical turbulencein open areas

four steps where growth rate of 1.2was simulated forthe average pressure

were used to calculate

100

data set

as the of the before edge. is not

inertia and

confirm other

specific exhibits

quasi-reported

blade demonstrated

pressure blockage

characterising captured.

Spalart-was also

vortices.

with a of 6

literature has prediction of

turbulence with cells

1.2 are for a

pressure calculate

(1)

withresulhave

FigureResults

experimentalCFDrespectively.analysiscasequantitative

NUMERICAL

additionplanesblade

Planes.Positions

workthe

120

The results forwith head and flowresults of this canhave been normalised

Figure 5 ComputationalResults

The comparison

experimental curveCFD at -30%,respectively. Generallyanalysis is ≤10%.case as a qualitativequantitative study.

NUMERICAL RESULTS Results were

addition to the planes were insertedblade and volute

Figure 6 LePlanes. Right, Positions

Observations

work found thatthe coincident locations

0

0.2

0.4

0.6

0.8

1

1.2

1.4

1.6

1.8

0

H*

for computationalflow from the experimcan be seen in Figure

normalised to best

Computational Validation

comparison of the curve shows that30%, , andGenerally an

≤10%. Direct comparisonqualitative analysis

study.

RESULTS AND

were post processed inlet pressure

inserted in the domainvolute at each flow

Left, Projection Cross Section

Observations made duringthat leading edge

locations of the

0.5

computational head wereexperimental pumpFigure 5, where efficiency point

idation against

computationalthat the predicted

and +30% is acceptable level

comparison is notanalysis is being carried

AND DISCUSSION

processed in ANSYSpressure plane a number

domain to studyflow rate [Figure 6

Projection view of ImpellerSection of Flow Domain

during the experimentaledge blockage was

the area where analysis

1q*

were then comparedpump curve. The

where head and flowpoint where ∗ =

Experimental

computational curve versuspredicted pump head

is 4%, 11% andlevel for quantitative

not as importantcarried out rather

DISCUSSION

ANSYS CFD post.number of other analysisstudy the fluid around

6].

Impeller and 4 AnalysisDomain at 4 Impeller

experimental part was most prevalentanalysis Planes

1.5

Experimental

SA-DDES

compared The

flow = .

Experimental

versus the head from

and 13% quantitative

important in this rather than a

post. In analysis

around the

Analysis Impeller

of this prevalent at lanes 4 and

2

Experimental

SA-DDES


928

5 can be seen in Figure 6. Plane 2 was chosen in order to analyse the flow at the pump inlet, while plane 3 was chosen to examine why no blockage was seen over this area of the pump.

The CFD results were analysed in terms of the components

of flow in the radial and tangential directions with respect to the axis of rotation of the impeller:

= −+

++

(2)

=+

++

(3)

This made it possible to compare the speed of rotation of

the flow with the circumferential velocity of the impeller leading edge . . while taking account of the radial flow. It can be assumed that > . . would increase the risk of blockage as the impeller can catch up with the rag. A higher velocity in the radial direction however should have the opposite effect by helping to flush the rag from the inner part of the impeller.

Figure 7 on Plane 4 at Q*=0.68(left) and Q*=1.11(right)

An analysis and on the selected planes does

provide some useful information. Comparing flow predicted at high and low pump flow rates over Plane 4 where blockage does occur (Figure 7,8), there appears to be a relationship between the ratio of radial to tangential planar velocity and the probability of blockage occurring. This is based on an analysis of the flow between the leading and trailing edge of the impeller. Figure 7 shows a much higher value of radial velocity at high flow rate compared to low flow rate. An area of high radial velocity can clearly be seen at the impeller leading edge where the relative flow incident angle is higher, as described by Paresh [10]. This essentially means that as the flow rate changes the fluid incidence angle relative to the impeller leading edge also changes. Conversely, as Q*

increases, tangential velocity decreases, [Figure 6-5]. This is in line with [11] and also the experimental results. One hypothesis for an inverse relationship between Q* and is that there is less through-flow at low flow rates but very similar impeller rotational speed, causing a higher level of recirculation. As the fluid will remain for a longer time in the domain at low flow, there is a longer time for the impeller to impart energy on to the fluid. Also as separation on the pressure surface is less likely to happen at low flow rates [12], the fluid may be more likely to stay attached to the blade allowing more transfer of angular momentum. The higher levels of recirculating flow can only serve to increase the probability of blockage due to the fact that the impeller is exposed to the rag for a longer time while rotating with the impeller. This is particularly true if the trailing part of a longer rag is still within the leading edge region of the impeller, slowing its ejection into the outer part of the volute. The impeller leading edge is then more likely to catch up with the rag.

Figure 8 Tangential Velocity on Plane 4 at Q*=0.68(left)

and Q*=1.11(right) Figure 8 shows a much higher incidence of tangential

velocity at low flow rate compared to high flow rate. Given that as the fluid is more inclined to recirculate the rag has a higher probability of impacting the leading edge and blocking the impeller. The low level of tangential velocity existing at the leading edge at Q*=1.11 where high radial velocity is present in Figure 6-3, makes it possible to infer that the flow is almost purely radial.

What is notable, especially in the high flow condition, is

the occurrence of two areas of high tangential velocity [Figure 8]. The region of high on the suction surface of the impeller is caused by the flow following the curvature of the impeller and the flow incident angle. The other area of high tangential velocity is caused by the impeller accelerating the fluid, due to centrifugal force, tangentially towards the volute outlet. Studying other positions of the impeller relative to the volute shows similar occurrences [Figure 9].


929

FigureQ*=1.

Comparing

positionsincreasecontributeleadingexperimentalwhichthe leadradialleadingfield,impeller.planelow flowrag positionradialleading

FigureQ*=1.11(right)

Considering

informvelocityquantitativeflow area furtherhigh planeshown

Figure 9 RadialQ*=1.11(right) for

Comparing the

positions between increase in proportioncontribute to the ragleading edge at experimental observationswhich clearly show

leading edge andradial velocity componentsleading edge. It isfield, as it is a factorimpeller. Higher radialplane 2, at the pump

flow rate [Figureposition or orientation

radial flow will helpleading edge once

Figure 10 Q*=1.11(right)

Considering

inform the overallvelocity componentsquantitative description

rate, most importantly of the pump

further investigate and low flow

plane 4, 11mm inshown in [Figure 6

Radial Velocity at for each of the four

the radial velocity the two pump

proportion to the flowrag being more high flow

observations madeshow that rather than

and blocking, thecomponents has

is important tofactor in how the

radial velocitypump inlet at highFigure 10]. It can

orientation whenhelp to move it has passed

at pump inlet

both contouroverall behaviour

components in thedescription of how

importantly at with the highest

investigate the relationshipflow rates, a rake

in advance of 6-10].

Q*=0.68(left)four impeller positions

velocity at differentpump flow rates clearly

flow rate [Figuremore quickly ejected

rates. This made with the high

than being caughtthe rag when subjected

has now been movedto also consider

the rag is presentedvelocity can be observed

high flow rate, whencan be argued thatwhen entering the

move it outwards through the inlet.

inlet plane at Q*=0.68(left)

contour plots and iso of the flow.

the domain provideshow these components

the impeller highest probability

relationship between line was imposed

of the impeller

Q*=0.68(left) and positions studied.

different impellerclearly confirms

Figure 9]. This shouldejected away from

is confirmedhigh speed camera

caught either sidesubjected to higher

moved clear ofconsider the inlet presented to the pump

observed at analysiswhen compared

that regardless ofthe pump, a strong past the impeller

inlet.

Q*=0.68(left) and

iso-surfaces h A comparison

provides a bettercomponents vary

leading edge,probability of blocking.

between and imposed onto analysis

impeller leading edge,

studied.

impeller confirms its

should from the

confirmed by camera side of higher of the flow pump

analysis mpared to

of the strong

impeller

and

helps comparison of

better with

edge, the blocking. To

at analysis edge, as

transformationcomparable

positionandclearrate

highermadebothblockagethanvelocityimportantto before

Edge

Edge

Figure 11 RakeA rotation

transformation comparable velocity

Plotting theposition shows theand high flow clear that israte by as much

is the dominanthigher than made at impellerboth componentsblockage may bethan at impellervelocity of theimportant consider

assess how quicklybefore it can be

Figure 12 Edge Rotational

Figure 13 Edge Rotational

Rake line on planerotation matrix was

on the rake linevelocity profiles at

the profiles ofthe difference

rates at impelleris the largest

much as a factor ofdominant component

. Also notableimpeller position 4 [

components is higher at be more likely

impeller position the leading edge,

consideration as it quickly the impeller ejected into the

, on ImpellerRotational Speed, at

, on ImpellerRotational Speed, at

plane 4 at impellerwas used to perform

line in order toat each of the

of and difference between each

impeller position velocity componentof 2. Conversely,

component with a magnitudenotable is when the

[Figure 13] the both flow rates.

likely to occur at 4. The instantaneous

edge, , at each can be compared

impeller can catchthe volute.

Impeller Rakeat Impeller Position

Impeller Rakeat Impeller Position

impeller position 1perform a rotatto consistently 4 impeller positions

at each impellereach velocity type

1, [Figure 12component at high

Conversely, at low flowmagnitude 3 to the same comparisonthe difference between

rates. This indicatesat impeller position

instantaneous rotationaleach position

compared to andcatch up with

Rake Line and LeadingPosition 1

Rake Line and LeadingPosition 4

1

rotational consistently study

positions.

impeller type at low

12]. It is high flow flow rate 5 times

comparison is between

indicates that position 1 rotational

position is an and the rag

Leading

Leading


930

It can be suggested that as is the largest velocity component at high flow rates including when compared to the impeller rotational speed, the rag can move out of the way more quickly than the impeller can catch it [Figure 12 and Figure 13]. Also if is greater than then the impeller should never catch up with the rag. As this has not been observed over the entire course of testing it can be concluded that is not significant a factor for blockage performance as .

CONCLUSION In this study a comparison of experimental and computational analyses has shown that most pumps perform better at resisting soft blockage at high flow rates compared to low flow rates. A new test rig was designed and a new test material was chosen in order to compare pumps of different sizes and at different design points. An analysis of factors which may contribute to blockage resulted in a dimensionless number which could in future be used to describe how probable the occurrence of pump blockage is with a view to informing future designs. Based on these results an observational analysis was performed with high speed photography at each flow rate with all test pumps to inform on the possible causes of soft blockages. The CFD analysis was performed to attempt to describe the processes responsible for leading edge blockage, the most common type of blockage observed. The study focussed on a comparison of radial and tangential flow components over 5 planes along the axis of the impeller. Results suggested that the higher radial velocity obtained at higher pump flow rates helps flush the rag as it flows from the inlet sum through the impeller radially, away from the leading edge before contact can occur. The higher tangential component observed at low flow rates also compound the risk of blockage by increasing the likelihood of the flow to carry the rag towards the leading edge. It was also suggested that impeller position in the volute can have an effect on blockage performance. This may not be significant in physical testing, as the average inlet velocity at Q*=1.11 is 3.87 ⁄ , compared with a leading edge velocity of 4.7 ⁄ . By the time the rag has travel 40 mm through the inlet of the pump the impeller has already rotated by 90.4˚. REFERENCES [1] JMP, “WHO/UNICEF Joint Monitoring Programme for

Water Supply and Sanitation,” October 2016. [Online]. Available: http://www.wssinfo.org/definitions-methods/. [Accessed October 2016].

[2] H. e. a. Korving, "Statistical Modelling of the Serviceability of Sewage Pumps," Journal of Hydraulic Engineering, vol. 132, no. 10, pp. 1076-1085, 2006.

[3] Health Research Inc, "Policies for the Design, Review, and Approval of Plans and Specifications for Wastewater Collction and Treatment Facilities," Albany, NY, 2004.

[4] R. McEvoy, "Innovative Development of Single-Blade-Impeller Hydraulics for Wastewater Application," Limerick, 2011.

[5] R.-L. Mitchell, Investigations into wastewater composition focusing on nonwoven wet wipes, Berlin: Technical Transactions, 2017.

[6] T. S. Kevin A Kaupert, "The Unsteady Pressure Field in a High Specific Speed Centrifugal Pump Impeller," Journal of Fluids Engineering, pp. 621-632, 1999.

[7] A. N. R. M. B. de Souza, "A numerical investigation of the Consant Velocity Volute Design Approach as Applied to the Single Blade Impeller," Journal of Fluids Engineering, p. vol. 132, 2010.

[8] Y. Delauré, "CFD Modelling of Turbulent Incompressible Flow for Pumping/Mixing Applications," Dublin, 2016.

[9] R. Nichols, "Turbulence Models and Their Application to Complex Flows," University of Alabama, Birmingham, Rev 4.01.

[10] O. M. Paresh Girdhar, Practical Centrifugal Pumps, Elsevier, 2011.

[11] F.-K. B. H. D. J. Feng, "Unsteady Flow Visualisation at Part-Load Conditions of a Radial Diffuser Pump: by PIV and CFD," Journal of Visualization, pp. 65-72, 2009.

[12] C. B. J. N. P. Rikke K. Byskov, "Flow in a Centrifugal pump Impeller at Design and Off-Design Conditions - PartII: Large Eddy Simulations," vol. 125, 2003.

[13] J. F. Gülich, Centrifugal Pumps, Heidelberg: 2008, 2008.


931

An Experimental and Numerical Investigation into Flow ...

Documents

Transcript of An Experimental and Numerical Investigation into Flow ...