Report No. :11426 Type: (MIS) Title: HANDPUJMPS: TOWARD A SUSTAINABL Author: REYNOLDS, JOHNsoe Ext.: 0 Room: Dept.: OCTOBER 1992 ,5- "I' -',~~~~~~M HANDPUMPS: TOWARD A SUSTAINABLE TECHNOLOGY Research and Development During the Water Supply and Sanitation Decade John Reynolds EfiT UNDP-World Bank Water and Sanitation Program * Copyright 1992 The m iternatona Bank for Reconstruction and DevelopmeniThe World Bank 1818 H Street NW Washington, DC USA First printing October1992. Cover photo by David Kinley (Zambia, 1991). lbs UNDP-Wodd Bank Wate and Saniaion Ppo was organized as a joint eodeavor of the Uied NatI Development Pgramme and tho World Bank and has been one othe primary acts in wordwide eforts to meet fth challenge of provng basic water pply and satio sevic to those most in need n dhe deveoping worK Pares in t1iis ventue are the developing conties tmselv and the mutilatr and biat agecies that fd the Progra's acvi Me UNDP-Wodd Bank Waterand Sanitation Progrm prduces its own pubicatons and also releases wo undr the auspices of the Wodd Bank. The Pgram's publicaons re divided into two seies, a Water and Sanitation Program Report sedes and a Water and Sanitatbi Discuon Paper series. The Prom Report senies presn fomal dscussions of the Progm's opeaons and resch as we as examnatios ofrelevantprojects and ends whthe water and sataton sctor. Pram Reports are subject to omus exunal review by independent authwrities from appropiae The Discusi Paper seies is a ess formal means of communicating lnedy and topwicl obsvaons findigs, and opnions concemng Progrm activities and sector issues. This document has been prepaed and published by the UNDP-Wodd Bank Water nd Sanaon Progam andtheWaterand Sanitato Division,ifnfastcand Urban De Deprmentpa othe fWold Bank. Cops may be obined from the Water and Sanaion Dhisio Materi may be quoted wit pop atributon Any maps that accompany e text hve been paed solely for the convenience of readers e bodrie, and clsdficatio of maps do not imply, an the art of te UNDP-Waod BAn Wate and Sanitaion Prgam, the United Naotins Development Programme, te World Bank Grtp, or any afdiated oganization, any judgenent on the lega or other atus of any ritory, or endoseat or aofanybouny. lb7 fin ndm andconlsosexpressedinthspaperareentely tee of the auhr and shoud not be attributed in any manner to the UNDP-Word Bank Wate and SanitatIon Pam, the United Nations Deveomet Programme, the Word Bank Gmup, or any afiiated n dbaL Contents IN MEMORIAM ..............*............................ v INrRODUCrION v. . .. . ........ ....... vii A:OV'ERN?IEWV ..... ... .. .... I Al.Badcground .. ............. 1 A2. Designing for V[OIM ... S A3. The Research Program .4..........**.. . . 9 A4. Field Trils..................... . . 14 AS. Labowry Tesdng of Handpmnps ...... ............. 15 A6.Conclusionsand Future Research .............. . 16 B: RESEARCH IDM SPECJIC TOPICS ............. 19 B1. Plastic Below-Ground Componets . 19 B2.PlastdccDryBeatings . 33 B3.SeallessPistonsandDiodic Valves. 41 B4.LightweightPtnp Rods for Dire-Acton Pumps. 47 BS. Stresses in Plastic Rising Mains ........... . ... ..53 B6. Ongoing Research: Rising-Main and Pump-Rod Conctors...... 70 C: RESEARCHI THE FIELD. . ........... ........... 79 The GlobalField Trials . s......... 79 Cl.Kenya--TheAfridev Handpump ................ ........ 80 C2.xia -The ndi Mark I and Mak m (VL0M) Handpumps .... 84 C3.Banghdesh -h eTaralrect-Action Pump.4. . 89 D:LABORATORY TESTS OF HANDPUMPS ....... ........ 91 Di. Wavin Handpunmp ................. 93 D2. India-MAiHandpump . ............................. 98 D3. Bestobel Micro Handpmp .. .......................... 102 D4.Abi-ASM HIU nP ............................ 107 D5. Pumpenbose PB-Mk l Handprmp ...................... 111 D6.KnebelHandpunp ........................... 115 D7. AquanontfHandpunp 444444444444444............ 120 PreOSA Report Seris i D8.Aas Copco Modelll Handpu.ap .. .................... 125 D9.AtlasCopco Model 122Handpump ...................... 130 D1O. Aquadv Handpump ............................. 135 Dtl. Afridev Handpump ............................. 140 Appendix:TheLaboratoryTest Regime ....................... 145 REFEURENCES . ..... ....................... jv VNDP-World Ban War ad Saitadon Ptogt In Memoriam During the past decade, effortU to bring watr supply to the rual poor have produced dramati changes around the worl ne cocept of conmuty manam has establhed itself as the most sustinable approach. Techologes have been adapted to widely varying conditions and rigors. And most ipotant for the nillions of poor who lack access to safe water, these technologies have been low cost, and it is now possible for many conunities to constuct and ntan theaiownfacilides. Astisvolumeattests,JomReynoldsplayedapivotalroleimtuisglobaI change guiding the develometof technologies tiroughoutthe WaterDecade. His death, which occured befare te pubLicadon of tis volunme, leaves us, his colleagues, witfi a deep sense of loss. We remember hin for his great contribution to nmeeng the chailenge of pviding safe wae to the hundreds of nilions who lack it, and we dedicate this book, hds last work, to his memnoy. Drawing by Johns Reynooks (untitled,undat ed) ProV=R4PeportSeu*a v Iitroduction Water is a fundamental human need. Pour fifths of our planet is covered by it, but untreated surface water is rarely fit to drink. Yet in most inhabited parts of the world, unpolluted groundwater can be found below the surface. Handpumps provide a cost-effecdve means of access to groundwater, and therefore have an important role to play in delivering safe and susanable water supplies to communities in developing countres. Research and development coordinated under the UNDP-World Bank HAndpumps Project have concentred on the conventional reciprocating pump based on a cylinder and piston, with two nomettn valves to direct the flow and a rod connecting the piston to an operatng handle. Other pumping techniques have been considered, but none offers significant advantages over the reciprocating piston pump without a!'so presenting serious drawbacks. The aim of the research has been to develop designs th,'t tak'. advantage of up-to-date materals and apprpiat manufacturing techniques to produce sturdy and reliable pumps that can be manufactured in the countuies of use anc can be maintained within the limited human and financal resources of user communities. Comnunity Water Supply: The Handpump Optiont was published In 1987 and sumarzed handpump research and development undertaen within the UNDP-World Bank Handpumps Project up to and including 1985. This re,.n deals primaily with subsequent research, both into existing handpumps and specific handpump topics. This report is in four parts: Part A, an overview, describes the context of handpump research and explains why it was necessry for the UNDP-World Bank Handpumps Project to be involved in directly sponsoring research. Part B gives detils of the research undertaken in a number of specific topics of handpump development, principally by the Consumer Research Laboatoy (CRL). Part C summafizes work cared out in Project field trials since the publication of The Handpump Option. Part D summarizes the results of labotory tests on handpumps carried out by CRL since publication of The Handpump Option. Program Report Series Wii Acknowledgments I am dnceroly grateful to all who have contributed to this rport, both for providing the information on which it is based and for deir subsequent help In editing it Speccally, my tns go to staff of: The World Bank, in Washington and in the Regional Offices The Overseas Development Admnitadon MTe Consumer Research Laboratory The InterAction Design Handpump Project The Swiss Centre for Apprate Technology The Open University John Reynolds Colchester, UK, 1991 vwii UNDP-WoeId Ihk Wet. and Sealaton Ptsre. A. Overview A1. Background The handpump is not a new, nor even a recent, invention. Illustadons sunvve from Roman imes of punWps with simple cylinders, pistons, and fLp valves idential in their opeaton to many pumps in use today. Handpumps probably first appeared wel before the Romas, since it is likely that pumps played an important role in the development of large seagoing vessels. And so it may seem sange that we shoud contnue the tchnological development of something that has been around in substantally its present form for thousands of years. But the purpose of handpunp research in the International Water Decade' has not been to reinvent the bandpump. nstead, attention has been ftoused on the development of handPump cocepts and detailed designs that can accommodate the realities of the scale of the demnd for handpumps, pobable workload and condito of use in the field, limitatons on maintenance and repair, and the industral facilities availle in developing countries to manufe whole pumps and spare parts. Of parmout concern has been the developmeat of pumps that can be maintained by the communitie that rely on them. The scale of demand At the icpto of thc Intenational Water Decade, it was estimated that up to two bilLion prsons in nual areas of developing countries were drinking water that threatned thicr healti and wellbeing. Since most developing counties have very limited financial and technical resources, fte nted for water supplies that can be sustaid at low cost is obvious. Groundwater schemes using handpumps can be implented at roughly half the cost of standpipe supplies, and about a quarter of the cost of yard-tap schemes. The handpump thereore has a central role in achieving the goals of the Decade. Moreover, the impotnce of handpumps is no t limited to their impact on the health of the communities they seve. In many cases, vilags without handpumps rely on water sources that can be many hoius away, and 1. hGoe Asmbly of do United N s dodsaed l981-l990as Tho IDatioDrnking Water Supply amd Samitatio D0es beinaftr, e itornatol Water Docado or "d Decade. - - S.- I There are exmples of hadpunps seying communities of 400 or more, but a sigle pump typically provides clean wat for between 100 and 200 pesons. This means that as many as 20 million handpumps cold be need to sads the origina demand, let alone the need for spare pars to mainain pumps and new pumps to replace those that have worn out or corroded. The worling environment Most handpunps have a hard life. Many are in continuous use thoughout the daylight hours for as long as they can survive such treatment. They are exposed to the elements, may be subject to pilfage and vandalism, and can be a convenient scrching post for domestc animals. At least some of the users can be aeected to be unsympathetic in the way they use them, not out of any animosity but simply becae theyjr are unaware of the consequences of sriking the handle repeatedly on its stops, for example. Groundwater its .is often aggressive. Corrosion can cause rapid falure of pumps that might otherwise be very durable. Sand and other suspended solds can cause rapid wear in cylinders and seals. The arrival of a pump is not always welcomed by everyone in the community: pumps may be objects of suspicion and outright hostiy in the minds of those who see them as threats to the traditioal paems of village . Maintenance and repair In the eady years of the Decade, and paricuardy in the global fidd trins during the mid-1980s, it became apparent that maintenance was the single most impon issue in bandpump development. Although conditions vary conadeably amg counties, the limited skills available withmin rual communites of develping counties could not, in genr, be expected to cope with the maint e e mn of most of the handpumps availale at the start of the Decade. Financial and pactcal cosoaints made it certain that centralized maintenance remes achieved, at best, only limited success. The concept of village-level op and maintenance (VLOM) was developed in dic response to this. Initally, the VLOM conept cenrd on the handpump itself, the aim being to provide vffilag the opton of performing mainenane themselves by developing pumps that are: 2 URDP-Wd Ban* Woan Sd Swstae Ptra * Ey mantained by caretks drawn from the user community, with mini- mal sk1ils, a few simple tools, and modest training; Manufacted in th country of use, or at least capable of being manufa- ed there, primarily to ensure the availability of spare parts; * Durable and reliale under field conditions; and * Cost effetve. As the field tials continued, however, it became app t that the issue of maintenance was not confined to the handpump itself. The quality of constuction of the borehole was also criical. Boreholes that failed to screen out 'id and other contmnt ca= conistent pmatue ilue of handpumps that in other ii n would have been reliable. Moreove, more maintence difficulties arose from instutdona or financia srtcomings than from techncal problems with the hardwae. Ihe VWM concept was therefore expanded to include 'software or organitional topic VLOM might now be better understood as standing for 'viDage- lvel opertion and managenent of maintenance," including: * Community choice of when to senice pumps; * Community chice of who will servce pumps; and * Direct payment by the community to the maintainers and repairers. Commu need to be encouaged to dewelp a sense of ownership, and hence of reponsibility, for 'their pumps. With this in mind, it has also been suggested that user comm should be expected to make a financial contribution to te origa cost of the borhole and the pump. Altwhugh most developing countres might be described a lacking in manu resources, there are substantial differe between contries. At th lower end of the scale are counties with little more th embryonic indust failities; raw mate and mana sil tend to be in very sort sly. By contrast oumtries sucb as lndia produce a very wide ang,e of industrial goods, and on an impesive scale; raw mateials and a variety of manuficturing skills are more readily availble. Potnil gl demand for handpumps is in the milions of units. Wit individual o , typicaly translates to demands for tens of husnds of pumps. 'Home mae of this order may not attract the attention of large-scale manu but tbhy prwide exelent opporuties for small and medium- local firms hwvm RReon Seato 3 Swunwy: lSe VLOM concept lb. ooncept of a VWM bmnup has both hadwar and softwmps. lb hadp is oam dset of a *sym that ao iludes tho bodkole, th cmunity rlyinS on the pump, and the availlity of loca aod nationa annufacuring facile. Ea a VLOM bu4ump installatio: _. The borehole must be dosigned and consted inas manner approepist to th pump ad loca codions. Roine mitnac and rpai ca be done by the usor community, which wil decide when to cwaay ut rpairs, who wil do th wodr, and who wll be . Noawesrius parts of th pump mu be durble d liale, and puts sul to woe ms b easy to service and inxpensve. ' As far as posble, the pump must be suitble for maufactl uig exisit loca indusriources, or facite that can be readiy estalished Impoted o are to be used only if tey ae critical to achiving odior VLOM bjcivea sablished manctures are n excluded, since thee is plenty of scope for jon ventures. This sUtIn agues for hndumps that can be manufvturd by methods ta a labor-intesve ter dtan capital-intensive and can also be - . make other products for which there is (or wil be) local demand. Specialize niqe, or those requiting a substantial initial investnent that must be amorzed over a ratively large production nm, are not suitable. The importance of quality contol has been clearly illustated by the lndia Nark II. In a manufctuing enionment where well-developed shop floor sils are at a pmium, quality control is best asred by including it at the desi stage Ti can range from small deails of deign - for example, shaping parts so that they camot be assembled upside down - to choosing proces that demand onscientius production management but rdy leas heavily on the sills of individual operators. It aLso includes desiging jig ad fixr which wiMl autmatically rect wbstadrd comnts durAg manfacte and assenbly. 4 UMWP-Weef Rea* Wa. and &mkatIe. Augmuw A2. DESIGNING FOR VLOM Te VLOM concept encompasses a range of concems, and it has important implications for the handpwnp as an item of manufactured hardware. It is a ignificant challenge to design fr maximum simplicity of maintenance, ase of manuf and high efficiy thougout a long life of constant use, particularly to engineers who are accustomed to working in a relatively sophisticated environment. Reliability versus maintainability A usel &st step is to rognize fte tide-off beween rdebflt and ab W. n ^ dSIhecrucial saoIstomInhiyeettmeftpunpisfimCOfkg rater tha to minime t number of tt breaks down.L Ts a VLOM pump that needs attention ratdvely ftequenly, but can be put right by the local caretaker within a few hours or days, is better than a conventional pump which breaks down les frequnly, but then needs a mobile team to be called out to repair it, for which an average delay of thr months is not unrealistic. Figur Al ilusrat the point. I | -o D 12 24 so 48 so FIpne Al: Relablty vermis uby This reasoning un counter to the classical engineering definition of rebilit as the mean dme between failures (M F). For Pump A, the MBF is about 8 months, for Pump B about 18 months. But if reliability is tk to be the proportion of the totl time for which the pump is fimctoning, Pump A achieves about 99 per cent against less than 85 per cent for Pump B. in reiability engieing, this is defined as availability. Another way of epessing the same idea and more clearly illuminates hie P*ovwD Rqwt &Va 5 dff betwoen the two pumps, is to compar the tot down dme' - for Pump A, the total down time is about 20 days, for Pump B it is over 250 days. The essental difference between Pumps A and B lies in what is needed to maintin tiem. The consequences of failure in a NLOM pump ae diffent by an order of manitde from those for a pump whlch demnands a centalized mantace team. This is not to say that reliability in the sense of MTBF is irelet, but anly Xt it is subsmvient to ease of maintenance. Cleady, oter things being equal, a pump whih breaks down less often may be prefefable to one which breaks down every few weeks. But it has been said that there can be some advantage in pumps which need regular atetin, in that this helps to reinfaore the commitment of the users to the maintenance of the pump and aids in developing and dismin apppriate sildis. This presupposes that the pump meets the requirements for VLOM, however. that it is simple to dismantle and reaemble, and spare parts are cheap and readily available, to ensre that it can be maintained within the limited human and financial resources available to the user community. I I 0 ~~ - I I Po nwpront @ , f Figure A:Fd t....als - a'lsso upfiue Pson w!i 0 6 10 *6 ao 96 so pOMn Of tdol fo Figre A2: Pel tials - a=dd of plump bMures It might be said that the best pump of all is one that never bre down. This is the elusive 'fit and forget' handpump some manar have aimed for, and a few claim to bave achieved. But evidence from both the lbotory and field tdals has been that "fit and forget' is not a readisc option. Pumps ta have been said to have been working for many years without any attentton whatsoever have, on closer examinaton, often been fowd to have been pumping from very shallow sources, or simply litte used. For what8ver reasons - social or cult resevations about the pump, difficulties in using 6 JDP-W,l Ban wow ad S^wk wgrax_ it, problenm such as high iron content in the water produced are possible explanations - the local community has praefd anothe water source. The below-ground asembly The Pfoject field tials showed that three quarters of repairs needed to handpumps were rlated to the below-ground components (Figur A2). Although these reuts may have been influenced by tfie choice of pumps used in the field trals (a elatively high proportion of punps with leather seals, for example), they showed clearly at ease of below-ground repair was an important factor. In many tradtional handpumps, below-ground repairs are far from staightforward. A 40-meter belowgound string using 1.5-inch galvanized steel nsing main and 12-mm stee pump rods will weigh about 200 kg. Full of water - as it will be if it has to be etacted with the footvalve still functioning - the weight will be close to 250 kg. Lifng tace is esstial. Moreover, with conventional eaded joints, a range of heavy tools d considerable sildl ae reqired for dismantling and reassembling the joints in the rising main and pump rods. Many pumps also require the entire pumpstand to be removed from the welLhead to give access to the rising m'in. In the pup cylinder itself, results from both the laboratry and field trials sgges that synthetic seals offer better, or at least more conent, pefmane than adion leather seals. Ihe conventin prooss for the production of the intera parts of pump cylinders was casting in buss or gunetal. However, both the casting process and the required maining prest seiious problems of quality control. The above-Wnd assembly Figure A2 shw tat a substantal number of handpump failures arose from problems witi handles and handle beaing. Bken handles cold geraly be attributed to pump desgner and manucturssmply unng the severity of actual conditis of use, and the sdulion s rela y saightforward: make the handles stronger. Th handc beings, on the other hand, are bound to be subject to a certin amount of wear. Th convenional choic for ndpump bearin has been blaces. But ballaces are expensv, prone to cdon, ca for close tolances in the manufdaeAe of hdir houings and tad to be difficult to replace in the field. Pmgmm Ripn Sale 7 Plain beaings also have a respectable ngneern pedigree. Their usual disadvantage is that they need reguar lubricadon, and this is not a realistic option in a handpump for use in nual areas. VLOM solutions Albert Einstein is quoted as saying tat everything shoud be made as simple as possible, but not simpler. lhis is useu guidance for the VLOM pump designer. Puting the principle of VLOM into pracdoe means desgnng for simplicity, certinly, but not in any sease producing an inferior or second-rate product. Nor is it a quest for a techmology which is "appropriate" in the Schumacher sense; VLOM shoud recognize the existnce of indigenous silf and materials and also of new materials and manufactring technologies where these can play a pat. In a pmcal VLOM hr: dpump: * The basic stuctr compoets of the pump sould neither break nor wear out. Nonwearing parts should last ten year • The weang parts should be readily accessible, requie no spocial skills for seicing, and be inpensive and of consistent high quality, to ensure inecangeabilty. • Th bdowground assembly should be as light as possible so that it can be extracted when necessry, even from deep setting, without the need for lfing tackle. * The inpact of coraron sud be minimized by using matrials which are inherendy corosion-resit. In meeting aU these objectives, the mnufctuin processes and raw materials required should be available in the country of use, or it shold be possible to establish them as slf-supporting, commeal entrprises. As the VLOM concept developed and crystallized, it was ckar that the majority of estblished handpump manuf e were reluctant to sWort the research required to develop practical VLOM pumps. Some were hostile and believed VLOM was unrealis_c. Othe could accept the logic of the idea but fet that the u nties of the market prevented tiem from committing the necesary investment. Unable to fid esting pumps that met their needs, Prqect staff in the field made thir own with local raw matis and mangfiities. T Bangladesh Deepset Mar I, precursor of the Tara, and the Madev, later to evolve into the Afuidev, 8 UAMD-WWU B*Wer ewd Smkiin hvmn are conspicuous examples. Seeing a need which eistg manu e were unable or unwilling to fill, the UNDP-Warld Bank Handpumps Project., in oo on with governmental and nongovernmental donor agencies, set out to coordinate resach into speciAfc topics of handpump design and consrucon. This research was deigned to address direcdy the most important priorities for uig pdacal VLOM pumps. Thughout the program of research, laboratory work - principally carried out by the Consumer Research La y (CRL) and suored by the UNDP, the World Bank, and the Oveseas Development Administration (ODA), UK - has been coordinated with work in the field, mainly in Malawi, Kenya, India, and Bangladed, with parcular reference to the Maltev, Afridev, and Taa handpumps. A3. The Research Program Most of the pumps tesd in tie labotory and in the Project field trials had been of the conventional reciprocaing pistn type, with above-ground lev am opeaon. Oter tpes had been included, howev, suh as hydmuhic pumps (Vergnet, Abi-ASM), recipocating pumps with rotry ation (Volanta, VEW), and sew-type pumps (Mono, Moyno). Although some of thes unconventional pumps offered particular potental advantages, none had shown itself to be especally wel-suitd to acheving the VLOM objectves. For deep well applicatons, it was felt that the msearch effot should refore be aimed primarily at developing conventional, lever-arm, reciprocating piton pumps. For lower lifts, direc action pumps (Ehiopia BPSO, Tam) showed coderbl potential to combine simplicity of manufacture with ease of maintenance, and to avoid many of the problems inherent in suction pumpL he prgram of research was organized in five main topics: • Plastic below-ground components * Plasic and other dry bearing systems * Sealless pistons and diodic valves lightweight pump rods for direct action pumps * Measurement of the es in uPVC sing mains In practice, ese ands of resh were closely intereated, as Figure A3 illuses. h,ws Rqou S.* 9 [14~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~o1 I~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~I .111 ______! 'iiE 24. lIEJd ~iiW}0x.t }~ 41afh:I Plastic dry bearings Plastic bearings have the potenial to be in ndy free from corrosion, ease e manu ances r ed in asated parts, and do ot need to be lubricated. They may wear out considerably more quckly than blces (although rapid wear is not inevitable), but they are much cheaper and easier to replace and could readily be integrated into a vlage-level maintenance gime. The UNDP-World Bank Handpumps Project supported a program of research, based prncpaly on field tests but including a substantil lement of laboratry testing by CRL, to develop plasfic pla bein The tesng included varous types of plascs, and also wooden bearings and torsional rubb bearing. Ofiginally the work centred on the ldev pump, but after 1984 was based on the Afridev and also involved the Swi Centre for Apwppate Technology (SKA7) and DuPont Plastics of Switzerland. The result was the development of the two-part nylon/acetat Af*idev handle bearing ystem- Sealless pistons and diodic valves Nearly one quarter of the pump failures documented in the Project field tials had been atrbutable to problems with pison seas. This research terefore set out to establish the potial for using "solid-stew pists which acheved the futons of conentonal seals and valves by fluidics echiques used in other engineering fields. Although there ae no examples of existng handpumps with *solid-st valves the Voanta and Etdopia BPSO bave shown that a pump can work stsfacoiy withut a flexile seal. Altnati to flexible seals are wdl established in otW fieds of engineering, and they are widely used where a sight loss of fluid is acceptable and the main requirement is for a seal that needs Lle or no maintenane. Solid-sta control of fluids is also well established and extensivy used in applations where mechncl control devices would not be practical. ODA suported reearch by CRL, in consulation with SKAT and the University of Sheffidd, UK, to investiate whether these tchniques could be applied to handpumps. lbe research onfimed that vsolid-sateW pistns could be prctcal, patculry for direct-aon pumps where piston speeds are inherently greater and there is direct feedbck from the pison to the user. Some possble prcal designs were swggested. - R Sf 11 Lightweight pump rods for direct-acdon pumps Direct-action pump benefit firom lightweight, highlacementpump rods. In a conventional lever-um pump, the weight of the rods is tbal by the weight of the handle. In a diet-action pump, however, the user has to lift dirctdy the combined weight of t rods and the column of water in the ising main. By mnmizing the weight of the rods themselves and increng their disp t to reduce the weight of the water column, the liffting forc can be reduced and part of the tal opaing efort ransfrred to the down-stroke, which is ergonomically beneficial. An obvious way to ae the objdtives of light weight and high dispacement is to use sealed tubular rods. Vaious mateials have been tied: uPVC in the Tars and Wavin pumps, polyethylene in the Nira AP8S, and aluminum in the Pek pump. ODA sponsored CRL to inveWgate the availability of these and other potentially suitable mateials, and to test a series of prototype rod confira. The rearch cnfirmed that uPVC, ABS, and polyethylene all had poatial for use as lghtweight pump rods. However, they aU tendd to buckle on the retun stroke, and althoug that did not seriously affect the pumps perormance, it could cwse problems of wear if both the rod and the risi main were of te same materia. Since uPVC and ABS pve comparble perfMance, the best cowomise might therefore be to use ABS rising mains with uPVC tubuar pump rods. Stresses in plastic ring m Most of the potnI benefits arsing fom the ue of plics in the pum cylnder apply equly to the dsing main. Light weight, comrion resistance, and reaively bw cost ae the principle advantaes of plastic mad to sed pipes, and the research o plasic below-ground cmponet envisaged from th start tat plstc pipe woud be ued for fte rig main. Mover, plastic pipe, parady in uPVC, is widdy availble and already manufured in a number of developing countries. While te pottal significance of uPVC pipe to achvg t VLOM objectves has beon appart, doubts about its eliability have persited. In oth applcaions, the prcple strases in te pipe ar generaly te caused by internal preue; like ote pipes, different gades of uPVC pipes ar spefied by the intera prssres thcy withstand. But in a handpump, the intera pressures are iely smal. Instead, the main soures of stress are the tensile loads (a combinatio of the weights of the pipe itself, the cylinder, and the water column) and the reacdon to friction between the piston seal and the cylinder bore. 12 UNDP-Wer Bea Waw aid SaidSa*Pop= While the major statc and quasi-statc forcs in a handpump rising main can be calcuated fairly easily, he dyamic forces are more difficult to predict threcally. The lfe of a pipe in pracie will be deermined not only by the absolute values of the forces involved but also by the way they vary toughout the pumping cycle. Moreover, litle was known about the effet of extea factors such as the type of pumpstand (lever versu ty operation, for example) and the effects of shock loads generated by the handle s n ts stops. A rsearch pogram was therefore initiated at CRL to detennme the actual stresses in handpump isig mains for depths down to 45 meters. The work was sponsored by the UNDP-World Bank Handpumps Project, in coaborion with ODA. A parallel study was undertakn by teracion Design with the suppoct of the government of the Neherlands, for depths down to 100 meotrs. This research has geeated a very large amount of data on the stresses bley to be present in handpump rising mains. It has shown that the absolute levels of stre in the riing main are considerably less than the ultmate tensile stength of uPVC, and that failure is heref more likely to be attributable to fge. Tne test results typically show hihe frquency secondary stress vaiations simpsd on the basic cyce dermined by the delivery and reurn strokes, for both the nrsing main and the pump rod. However, the magnitudes of the ste varaons are such that uPVC pipes used as handpump rin mains, have the potential to offer svice lives of the order of ten year, provided that the pipes are of good quality and that str concentrations are avoided by aWpp a deg d hanling. he rarch included the analysis of samples of uPVC pipes manu in devoping counties, compaing them with the UK-made pipe used for CRL's test. Evidence of lack of qualty control was found in all but one of the pipe samples. Ongoing research: cormectors for rising nains and pump rods Pollowing the t of the pering es in uPVC rising mains and pump rods, resch has begun to evalte and develop easy-to-use nnectors for plasi rising mains and for pump rods. The initial research is being undetken by CRL, with the suo of ODA, GIZ, UNDP, and the World Bank Eventually, the projects wil include assesments in the field. Each of these arh topics is covered in deta in Part B, which explains the background to the resech, mehods used, findings, and conclusions for the individual hPgvw RqAt &rW 13 A4. Field Trials Field tials up to 1985 wer repofted in dei in Conmwky water Ssqly: The Handpwnp Opton. Pret resach in the field since 1985 has bee concentat in Kenya, India and Bangldsh. Kenya - the Afridev handpump Production of the Afiidev pump ben in 1985. A number of dzgn improveens wee mintroduced in 1987 to combat premat wear in the handle bring and internal wear of the rsing main caused by contact with the orginal polyethylene pump rod centralizers. The impovements were incorporated in the Afridev Speification published in dut year. The Afridev was designed for annual replacement of wear-pone parts- baring bushes, valve bobbins, and piston seals. 'This strategy has been geeally successful, and user communities have been able to sustain the pumps in good working order with their own resources. Idia - the India Mark and MA II (VOM) pumps The main obiecves of research in Ida hae been to impe the diability of the Inda Mark II hamdpump and develop a derivative, t India Mark MI tat wi be simpler to maintain. It is estimated that improvents in the India Mark will halve the number of breadowns, with cosequent annual saving in mainenn cost of about $12 million. By deign, most of the ompots of the India Mark H are ngeable with te India Mark Hl. In the India Mark M, existig pump mechandc can cary out over 90 per oent of the repis likely to be required, with the help of the use community and a few tools. Moreover, the Mark H m it possible to train a vie mechanic or blacksmith, eby making the neeay skls available within the user commuity. A number of experintal direct-acon pumps using Afiidev below-grund compornents have also been instaled. 14 VADP-Wl Bea* Wate and S&ndtM hVun I Baxgladesh -dthe Tara direct action pump The T1a is a pump designed from the outset to be suitable for bodh manufacture and village-levl Ienane wiin the very lidtod resoures available in Bangladesh. Mm Tar enjoys very high user accepta, can be installed at a low cost, and can be sustined at a cost of about $0.10 per user per year. It is truly a VLOM pump. Details of field rsah are given in Part C of this repo. AS. Laboratory Testing Inidal rese oncentated on handpumps which were available at the start of the Decade. The objectives were: * To evaluate the major handpump types on the market * To coopert with manufc in the development of design improve- ments and new pump des * To prvide informain to assist d ping country govemnts and donor agenies in choodng bapp iate pmp for par applicatio Compartive tests on handpmps wec daised and caied out by CRL in Hareden, UK (at t time known as Consme' Associatin Testing and Research, or CATR). The oiginal test of 12 pumps was _spoed by the Oveseas Development (UK) and fonned the basis for futier tests caried out under the UNDP-World Bank Th laboutory tet prgram was desned from the outset to provie liable, tive data on pump peformance and enduran, and to assess lily maintenance riements and the pontl for loca manufatr. The oiginal test rm was desd afker etnve consultation,, to refle acta conditins of use in developing comuies, and to treat the pump not simply as a product but as a system that cluded its user. The tege has been refind over the year but its basic pricil remain inta The results for a pump tesd in, say, 1989 can be compared with one tesed in 1981. An impota dement of the program has been to encourae diogue with pump manua to sugges ways in wbich existing pumps might be improved. *vpm R*a &Wm 15 Resilts for pumps tsted up to 1985 bave been published in earlier UNDP- World Bank rport and were sunmarized in Comm_nty Water Suly.: The HandpoW Opton. Laboratory reasts for pumps tsted since 1985 are summaized in Part D of this report. Th response of handpump mto the lboato findings has been mixed. Many manu s have been eag to discuss, and some ultimately to adopt, engineefing improvemnts suggtd by CRL, though only a few have been prepared to acept the challenge posed by the VLOM concept and make fundamental changes to their products. A6. Conclusions and Future Research Handpumps represent a technology which has been understood for thousands of years. Nevertheless, the challenge of developng handpumps which could be maufaud in developing counties and aintained by the communties that rely on them has bee considerabl. For most applications, the management of aintenance within the vil is cucial to the long-term ilitater Supplies. Handpump manuacr have now espoused VLOM and use it liberaly in their advetising, but in the eady years of the Decade, when mauf s were faced with VLOM as a concept ater tha a practical proposition, they were reluctant to take on the reseach and deveom required to put VLOM ideas into practice. Unable to buy off-th-shelf pumps which met thdr needs, Prject staff in the field began to make their own, using matials and manucuing fciliti available to th, and this initiatve was the coneso of the handpump rc and d opment proam coordinated by the UNDP-World Bank Handpumps Project. The consistent strategy behind the researh and development eft has been to combine field testng with work in the laboatory. Tbis hr- enued tht the work has been firmly rooted in the realities of conditions in the field, 1e drawing in expise from a range of disciplines Ihe rsult has been to confirm that VLOM is a pracdcal proposifon. It is exemplified in the Afidev and Tar handpumps, which incpot deign conceps that are fundamental to succesfud, suainable, village-kvi. Some manu have already folow these leads: vions of the Afiidev - the Aquadev from Mono puMPS of the UK - are aready i production. Odhes have made significant changes to 16 UNDP-WwU Bo* Warn ad Snkadem PArW 4 hdeir pumps or have produced new models dedgned to satisfy VLOM requirements. For low lis, direct-action pumps such as the Wavin and ia AF8S are now avaible. A conspicuou success of the progm has been to show how plastc can ase itenance, both in the below-gnd assembly and for handpump bearings. Plasdc part are reMavely cheap and inherently coroon resistet. Their light weight makes it possible for below-rund asmbUes to be widWmwn without lifting tacle. Thrghout the Decade, the research effort has concentiated on what wer seen as the most ignficant outstadig design isues. Other issues remain, howeer, and al research tends to mise new quetns as the aoginal aims are achieved. Opporunities for the next phas of research are smnmized below. Smswy: OppwtuItIes for future handpuap rsearch Risinguis Dvelop VLOM comew system_ (tbis is aked under way wiH to support of ODA and aZ) DIwIap qult sdrds d quality cOutrd pl_ese, ilding dn* ltin, for uPVC pew _macue in ml dsdgn gideliwes fo plsl * i Wn l asism_ Bearings J~~nvestigaft elasonwic mutriuln for use as maldiding beaings develop -c desigsm Cyliner Invstiat design and manufctur of dueasc pumping eleants (a used in tb. Vergu pump) to feduc cost and impr Test ad develop practcd dmpgs for saUess pstons and solid-$ valves Revw, asse_s, nd develop fudhmaiob ofpmecig cyli- Dmu actio Deop iroved densd fo pumd nd risig m pmp c nasotrim ad enamuliafa CI Ames tecni for combding corsios, iding cathoc iute l a. pla c u castng with pbat orndbe Pump tods Deve lop '* ees$-Wrelem coulings pnigiwu Ripn swrim 17 PH LI IIj~~~~~~~~~~~~I)I~~O I CD~~~~~~~~~~~~~O so.,: ~~e~ g plastics command their use In below-gound assemblies. With appropriat dein, plastics also have the potential to smplify maintenance by elimmating the need for tools and miniizg the skils reqired of pump careakr. Spam can be cheap, have long shelf lives, and are less liely to be borrowedO for other purposes because they have no other practica uses. There is good potential for manufacture in the country of use. Although developing countries are unlikely to be able to produce plastic raw matals, the procesing of plasdcs can provide the means to manu e a wide variety of products dth find ready markts. A number of developing ountries already have establised molding and exusion industries. Swuway: The Advantages of latcs t andpums Light weg For asm of pr ad aintnnce Crulon ruesta Most PI= ae not" affed by wsaerd wil notctu to alvani corrsion of other povarswth whic they ame in ontact LOW cost Alth0g a lr inti outl is requird the subsquapit price of t_o Wil be rlavy low in quatt .pas pa_d Vaatilty COnUpi oonompm we sinalost a eany to make an Sim* a em Qualty isaram plhouh lstc proessin reqir es aa mM anM d _pAu th qult of te finihd prod is not drecly dpndo an to dill of fth produti opetr. NIatlalw for local Plasic preoessn is sestablshd in some developing countrie nmafatuet and attactmv to other wantig Ito encourag new idmndutis Developing a design concept The Consumer Research Labatory undertook deveopment of a desig concept for t below-groud assembly of a hApump tat wold exploit the advanta of plastics. The work was jointly funded by ODA, UNDP, and the World Bank. The project did not set out to develop a definitive deign, but rathe to demonate how plaStics oUld te used SuCcsly in tis pcatin and thereby prmvide a lmowege base on which more specific desigs could be developed by manuf and others. 20 Ut(P-Woi*dWB Waue and Saao ?PmVw The research was subsequendy used In the deveopment of the beow-groud components of the Afidev bandpump. Designig for plastis rdly succeeds where the plastic maei I simply substtuted for a traditional material. Th proprides of plast ae distnctly diffeent from those of conventional materials, and t is neceary to dedgn from first principles, not only to realize the advantages that plastics can offer but also to avoid potental shortcomings. Moreove, the design must tae account of the pardculr mueria to be used: the term 'plastic' covers a great variety of materials with a wide range of phyical CRL consulted a number of plastics manufactus. Their constent advice was that the suitability of plasdcs for specific applications was at least as depandent on the deign and the worldng envimn as on the intnsic prpertes of the materis themselves. Hence even the toughet engineering plastcs might fad in a poor desgn, while relaively cheap commodity plastics could be uccsu if the design were good enough. Moreover, the sucoess of particu materals is prdictable to only a limited extent: it is always necessay to make and tet prototypes and modify the design in the ight of test rests. Te project was therefore divided into stages. In the first stage, a wide rnge of design id was expord and evaluated by laboratory and field tests on prowWops mhied from the solid. These included a design by Ken McLeod for a highensity polyethylene (HDPE) piston to be used with a cylinder of stndard uPVC pipe. A more limited slcon of desgns thn went fonrard tO the sooond stage, for wbich molded components wem produced for fter testing in the labatoy. In the ird stae, experimental pro using the molded components we td in the field, and te resut of al the toe and aessent wer brought togeter to produce final recom- mendations and conclusions. The design objectives and consrants are sumnmid on the following page. Stage 1: Testing of the M:leod des A pwotype "Meod cylinder was testod as part of a program of laboratoy tes of te Maldev pump head. TIe peforma of the cylinder at the sa of the was at least compnarable withresutforf convenipumps,with mechanical efficiencies ranging from 55 pecent to 82 peret for depths from 7 metes to 45 metes. After 1,000 howus of endurmnce testing, however, the outfow from the pump was reduced conidebly as a result of wear in the cylinder bore, the sealing edge of the piston, and newa s Swia 21 Swwwy: Dedp Objectves and Coy,uts incAudig th pump rods d dng ma. Mode of opatIon A -i reciproating in suit"abe for mse with the Mdev end dsimlr abmov und m.mblie Capacit A least 0.5 ltoe per cyce Maximu depth 45 1mtes E_du_e At least 10 oilli rversa Cylne and risi uda Standard uPVC pipe for both te rsg wein and the clnde barwl; te complet beowground assembly mus fit a four-inch wenafi . Sals The dedgn must be adaabl to a vaiety of off-tesheif seaing systMS - leate, uetne, nbb, eoc. Footyalv To -simplif main nce and rpair, to footvove dsd be extrce witout th need to remov th rising main, but to design st ails be vible ea a Jneeattl system and repa# T e requirements for village-Ivel oqpetion and m _nnac mu be saisfied i e final design sould nge tho minmum of toola aMd illo. d parts M_aor opomt of dke piand fotaw mud be ideia to rdue ia tod cos, reduce producin Costs, smpli m _intnance and pir, nd eduw tho nmber of es neoedd oa. 1afbnomu caltsda To acommodat umanie, waker user, maimm pump effidency ms o od to lowes opatg spods. T puto and footvalv must throe open ad cloes quckly and effaiciuy at low 13tes of fow. by solids The flow velocity of te vaves sod be ufficit So cla supended polar and priedove of the valve seste. PbUdal forligh-vilmhm niti p ypes may be ptodwed by snail-sae methods, but do pamedom finsheod product must be suta for volume produ n n 22 UAYP-W kI Ba* Watr and Sw_tkm Pmorass the Piston valve. The cylinder barrel was replaced and the piston was repaired, but it wore away rapidly when the te was rearted. This suggested that the combinadon of HDPE and uPVC was not a favorable one, and the test was teminated. The initial CRL designs A basic design was drawn up and adaWd to a range of differt methos for seing the piston in the bore of the cylinder, and these sealing options were compared for both and endurce. ITe main components of the psn and footvalve were identical. Acetal was the maerdal chosen for stressbearing components because of its strength, sffness, resistance to creep,' and stability in water, although some parts were initially produced in uPVC so that they could be solvent welded. Acet is exensively used for snap fits, domestc ad industria water fittings, and other applications which exploit its -springy* cha _cteristics. Both the piston and footvalve consisted of a valve seat and two identica body halves. As the halves came together they engaged both the valve seat and two groves machined in the end of the pump rod. The halves were then secured in position by a collar. The poppet valves were made from acetal, with rubber o-ring seals and mp-in egs deige to limit the valve lift. The valve port diameter was 18 mm. The piston and footvalve were connected by a telescpic link, so that the footvalve could be eacted with the piston. The link was arranged so that it would not interfnewith the normal pumpng action, but woudd come into play only when the psn was widrawn beyond the normal upper point of the pumping stroe. The focUalve was retained in posiion by a barbed, caselated spigot which snaped into a receiver in the boaom of the cyinder. Both the cylnder nd sing main were stndad uPVC pipes of 75 mm outside diameter, the cylinder being Class D pipe and the risig main Cla C (BS3505), to enable the psn to be widrawn for maintenance without extracting the fiing main. Te pump rods were inless steel, with aceta connecton. Like the piston body, these connects consid of identical halves which engged grooves machined in the ends of the pump rods, and wer secured by collars. All scew threads were thereby diminated fom the design. This design approach was intended to be compatible with the Maldev pump, which was under de at the me time The Madev ws designed and made in Mlawi, with ease of servicing and manuflature as primary deign objecte. I Cree is Urns - eS to trn Ihe4s dfaies of plauics nder us.ind stu. Pp'Sm R Swim 23 Four sealing techniques wae evaluated in the first stage: * A proprietary uretan hydrauic u-seal • Cup leathers made by Cieco (India) and Clmax (UK) * A rolling lobe diaphragm seal • Extended tububr piton, without a flexible seal but uset in con on with cloefittng ext polyethylene colars fixed within te ising long and short coUlr were tested) Conpalson ofpwp perfowwe Each prototype was tested for pump performance as part of a complete pump assembly, using a Maldev head as the above-gud purt. The amount of work done on the pump handle was comparod with the usu work done by the pump in raing water, for a range of dept and opeatg seeds. The urethane a-ae, new or pat-wom, worked wdl. The rollng lobe diaphram produced ligh efficencies but considerably lower flow rates ta other designs. Such performance is inet in a seal of this type, in the rolling portion of the diaphragm travels only half the distae of the piston. Te leathe cup seals achieed pessm to the urethane T-9eals. Ih relvely thin Climax seal poducoed a high effency and eaer outflow than e dather thicieco Me al. Th extended tubuar piso wit the short extra cola was Imited in performe by lekge. Better res were obtned from the long extenal collar, but these were still infeiior to the urethane u-seal and the better of the cup leather. Ewance ws For endumance testing, protyp reprsenting the various seal tions were mounted on a mulicsttn te rig. Each cylinder was opamted at 40 cyces per minute at a simulatd head of 45 meters. The target for completon of the st was 2,044 hos, PEsfentIng five miion reversals. Uretane u-s Two samples wee tsted. In one sampe, the pson body broke aftr 1,134 hours and could not be repaired. In the second sample, the piston also broke after 1,173 hours, but was repaired. he test continued to 1,647 hours, when wear in the seal 24 WUJP-WaI Bo* Wa aad SmkiMh Pe had reduced the outlow from the cylinder to a triclde. Wear of both the seals was confined to the lip; the overaU diameter was reduced by approximaty 1.0 mm and 1.5 mm for the two samples. Rolling lobe diaphragm Four samples were tsted, and al filed rapidly when the diaphragms collapsed on the return stroke, causing them to nik up rte than roll. The rolling lobe seal was withdrawn firom the test at this point, and firther investigation was carried out under a separate project (page 33). Leather cup seals hee samples were tested. Two had to be witdrawn when their pistons broke and could not be repired. The third completed the test with relatvy little wear. Extended tubular pistois The prototype with the short extena collar seized after only 124 hours. However, the prototype with only the long external collar failed to complete the test by 120 hours. It was withdrawn because wear of the ola had reduced the outflow to a triclde. Two prototype assemblies were instaled in Malawi. One failed when the legs of the footvalve snap-in featue broke off, mirroring a problem which had been apparent in the laboratory tests. The second survived for 20 months untl the pit broke. * Dign Bcorwnendadow From the results of the laboratory and field test results a series of desi recommeations, summarized on the following page, were prepared as the basis for the second stage of the project. Pso8mu Rpwn Safe 25 Swmay: Desgn Rondatons from Stage 1 Cyinder and rising min 75-mm diamtr uPVC pipe: cylinder, BS350 Class D; rising utain, Clam C PIump rods 12-mm stain sted Piston so A convenDonal a- or cup soal, but this way be either a leather cup waher or a proprietaty synthetic sel of ureane or nitrile rubber. Fiston and footvalve Icdtil copo for al the picpal pss of both th pisto and fotvalve; footvalve snap-in redegd to elimiat excessive stms. Valves Port dimeter 20 mm; molded bobbins fitted with staind tber a-ringg; valve lift ontolled externally. Footive seal Standard rubber oing (iternational Metric, European Metric, or ntiondal inch Sim) M _atials Al stress-bearing parts in acet bbdenance UTb only tool required to dismantle the piston and fwotvalve would be somet likte ra srwdrve to levr th. coampe aparL Ste 2: Laboratory tests on prototypes An improved desgn was proposed combining the improvements idendfied in 1he firstage of laboray and field testng. The pirinpal components of the cylinder wee rationalized to mbumize the nunber of molding tools and the stock of spare part rquired, and to eliminate the uPVC components that had prved unreliable. Each molding was designed for stmplicity of tooling and ease of manufacte. The footvalve snap-in feature was rdeigned to diminate excessive sress, and the bars intended to limit valve lift were removed: many had broken during the tes. The valve oifice diameter was increased to 20 mm. This size was chosen as a compromise between ening that water would s flow through the valves relatively quicldy, to flush pnded solids in the wate and prevent valve seat contamintin, while reducig resistce to flow to a level that would not significantly incrase the applied force at the handle. Figure B1 illustates the relattihip between head loss and valve diameter for a range of pump operting speeds, based on the dimensions of tie expeintal cylinder and the stroke of the MaIdev pump. 26 UNDP-Wed Ban Warw d SaxiaSl Progri Hood loe ()m) 40 -. 40 oycles/minute - 50 cycles/minute 90 20 \\ t0\ 10 16 20 26 so Orifice diemeter (mm) Figure Bl. Head loss verss valve orifice diameter he desg is illustrated in Figure B2, page 29. A set of molding tools suitable for a short production run was manuied, and an initial batch of moldings produced for lbotory tesg. Ilhee variants of the basic design were tested for pump performc, to compare with the resdts obtained in Stage 1: a Halite urethne u-seal * Gaco Ditrile rubber lip seal * Ciecu cup leather Perfom ce as All the seals prvded high levels of pump efficiency, wnth higher efficiencies coresponding to lower opeting speeds. The best results were obtained foMr the cup leather, followed by the nitrile rubber lip seal and the urethane u-seal, but the differences between all the seals were small and unlikely to be significant in practce. For the urethane and leather seals, where comparable results were available from Stage 1, the results for molded Stage 2 pistons and footvaves we consisteny better the required force at the handle of the pump was reduced about 15 percenL hmgmm Rqon S.in 27 Ddwwe tmst Four assemblies were tetd for endurance: • uHaite urethne u-seal * Gaco ni rubber lp sea * Cieco cup leather * A cup leah made in the UK The tes was carried out at 40 cycles per minute at a simuated deph of 45 mer Tbe water was deliberately contaminated with sand and K ladphr to induce exts wear. Ibo target ime was 2,000 hours, repnting about S million everal. Urethane seal The first sed wore out afte 270 hours, reducing the oulow fom te pump to a trickle. A fresh seal was fitted in the original cylinder an lsed for a further 324 hours. However, the cylinder wall remaned in good conditon, with mimal wear, confi g that the seal was the sarfiial element in this ni n. Nitrile nrbber seal The nitrile seal performed well trughout the test, and had th capacity for anothr 1,000 hours of use under the same condit at the end TIe cylinder bore was polished, and about I mm greater in dameer thaln at th ulr Thben results were consistent with CRL tests of the Ta pdup and rest obtined by Lund Univesity of Technolog. Cup leathes Both the cup leathers wore out in 620 and 807 hows fr do UK- J Indian-made ksleas, reewvdy. In both caes, he cylinder bore was bdly scored and worn. pct teftinl rod acnoirs The chosen meod of pump-wrod ten prved to be rlable in the endurance ita, t!t a further tes was cafied ou to asses tir endurance in onse to a repeated shock force of 4,000 Newton (rougly twice the force in the pump rod duing normal pumpig. Connectors immernd in watr lsted signifinly longen dry connectors, but all filed. An intstng result was hat the oinal innt that the load would be saed by two soders in the i s egging a pair of goes in the rod did not seem to be born out in practice. Clearly, in any such desg smal dimena inconsist_ will result in contac being made at one interface before it is made at the odher. However, it had been exWced that elstc avergng in the plasdc coupling woWld 28 UNDP-Wed Bo* Waw ad SI r&m Pv,w I~~~~~Iq I I, 4 - ensue that once appied the load would be shared. In fact, couplings modified so that they had only a sige sioulder survived a smilar number of shocks as tiose with two shoulders. Stage 3: Feld tesin A total of 87 sets of components with a selection of leate, urethane and nitrile rubber seals wene sent to field trial centers in East and West Afkica, India, China, the Philippines, and lTailand. Pictorial instuctions showing how the components should be assembled were also supplied. Reports were received from Kenya, Malawi, Tanzaia, and Ethiopia. Smwwwy: Fied Ted Reports _ staIlalIon M"o detaile ustiuctm woud hxv be helpf; ps should be numbrd to correpod to oisbucio ador sold be =ded i dffeet cor; leathep mis we vay difficdt to aemble. Use satisfatosy xtbctio Very difficult to extract pisons witm w lethe ss ina woiig order (any with worn msoa) Geneal Good desig, but too my pas Wear Urewn and latiher mise wa mr rapidly ta mtnle nbber Th difficulties enoountered when extactg pison with working leather seals resulted from the seal's expanding so that it continued to support the waer column even when the piton was raised beyond the cylinder pipe into the larger-bore rising main. Assembly diffcties were largely related to the feature of the expeimenl design that allowed it to be adapted to a variety of types of seals. The design would be simplified substtially if only one type of seal were used. 30 UNDP-Wed Ba* Waw and S&iade N,guw There was much discussion about the derability of linklng the footave to the piston so that tie footvahre could be exted for inspecdon or repair. If an altrnve means of footvalve extraction could be used, nine components and a solvent welding operation could be onitted. It was also pointed out that the link could be a red disadvantage since there would be no point in extctng a footvalve that was worlang well. Recommendations and conclusions The recommendations and conclusions of the CRL research are sumurmized on the following page. Development of the Afridev below-ground components Many of the ideas developed in this project were subsequently put to use in the Afridev pump. The Afridev was designed to be as simple as possible. and to be capable of manufacu in developing countries. To minimize the stresses in te system, a standard cylinder diameter of 50 mm was adopted, combined with a maximum stroke of about 200 mm and a choice of handle ratios (2:1 or 3:1) to accommodate a range of workng depths. Below ground, the design objectives were to minimie the number and complexity of components and to arrive at a design suitable for high-volume production. The Afdidev cylinder components were deigned and developed with assistrAnce from SKAT and DuPont Plastics. The design carries over the idea of an identical piston and foot vale, molded in aceta, with snap-in legs to rtain the foot valve in a receiver inorporated in the ising main. However, the joint between the moldings is at right angles to the cylinder axis, in the plae of the valve seat, in contrast to the CRL design. The moldings ae pmanenldy joined by spn welding, thereby creating the interior void for the valve, but avoiding the need for exten fseings or retine The valve is of a similar shape to the CRL design, but is molded in rubber to avoid the need for a separate o-fring seal, and to give it sufficient flexdbility to be inserted and removed through the valve ports. The gener wal th ess of the acel moldings is about 4.5 mm, with external ribs 3.5 mm thick. All the molded components are produced in Kenya, and the molds have also been made available to other developing counties While the CRL deign was intened to accept a range of seal tes, the Afridev was de ed from the stat to use a proprietay nitrile abber sed, based on the results of CRL's tests Because the nitrile seal has a reavely lare inside diameter compared to a leather cup washer, it was possible to deces the overail diameter of the piston to suit the 50-mm cylinder bore without reducing the size of the valve. pJ'gam Rqfe swim 31 I t'j J Iii 'I I 4! 1 ji.  ,fI j.gIL ii olli :1111 It" ii ij.j Ii S. we 8 '! i.j I *"i u ul a = *% Ilij I '11:1 0 III 'ii 1 liii 'I. I **b  I [Iii J ii I ii *iIIi If' I III .1 * 3 I I u11*1 Si I *1 Ii tod couplingm. The rods are *upset" at each end into discs, and then linked togete by a pair of acetal half-couplings secured by snap-on collas. Rolling-lobe diaphragms As a result of the early failures in endurance tests, the rolling-lobe diaphragm was withdrawn from the CRL project. However, because of its potental to eLiminate sliding friction in the cylinder, a sepatate program of development has been undertaken, funded by ODA and using the rig developed by CRL for the original endurance tests. A senes of more refined and developed schemes was devised using multiple diaphragms in opposition to ensure that they remained inflated throughout the pumping cycle. Although the endmance of successive designs has increased, none have been able to achieve results compable with good-qualty, conventional sliding seals. B2. Plastic Dry Bearings A pivodng lever, such as the handle of a handpump, must have bearings. In many deep-well pump designs, a seoond set of bearings is also fitted where the handle is connected to the top of the pump rod. Ile conventional solution is to fit ball races, but these have serious drawbacks in the contet of handpumps. They are expensive, prone to conosion, desged for continuous rotation rather than restricted angular movement, need accurately mahine housings, and tend to be difficult to replace in the field. Somk desi have elmminated at least the second set of bearings by connecting te pump rod to the handle trugh a flexible link: the India Mark 11 is a well-known example. But these desigs have problems of their own, such as the need to ensure that gravity acting on the pump rods will be sufficen to return the piston on the down-strke. Plain beanngs are also widely used across a broad spectrum of enginerin applicati ns. However, most of the materials used require regular, or even consant, lubricatin, which is not practical for a handpump. Phatic bearings have the potential to be inherendy free from corrosion, require less stringent standards of manufcture in aociated parts, and have no need of lubricaton. Ther servce life may well be considerably less than ball races under idl conditons, but because they would be cheap and easy to replae they could readily be inteted into a village-level mainance i. hvgi8 Repo& sed. 33 Acetal and nylon (polyamide) are plastics widely used for other bearing applications. Both offer good dimensional and mechanical sability, good wear resistance and low fricdon, are easily machined and also suitable for injection molding. Nylon has a propenity to absorb water, a potental disadvantage in a bearing for a water pump. Acetal, on the ohr hand, offers low water absorption over a wide temperature range. Other plasics such as PTFE offer very low friction but are much more limited in their mechanical propeti The first experiments Thefimexperimental designs forplastichandpumpbearings were evaluated in the Consumer Research Laboratory and in the Livulezi Valley Project in Malawi, using Maldev pumps. The initial design was a simple cylindrical bush with an integral thrust washer a "top hat bush (Figure B3). Prototypes were machined from solid acetal, but it was recognized from the sta that the bearing should be suitable for injection molding. The same components were used at both the handle fulcrum and the pump rod connection, malkng a total of four bushes per pump. The bushes were used in conjunction with pivot pins machined to a surface finish of 3.2 pm R,, and fitted in housings machined to the same stndard For labotory testng, the bearings were mounted on a purpose-made rig incorporadng a Maldev pump handle. Tbe rig imposed vertical, torsional and lateral loads on the bearings, with one bearing carying a _ _ generaly higher load than the other. The forces on the Fgure B3. Simple "top pump rod were equivalent to pumping water from 45 ha bush meters depth with a cylinder of 50 mm bore, and the operating speed was 60 cycles per minute. The ambient tempure was maintained at approximately 30°C. Wear in the bearings was monitored by meunng the lateal and torsional free play at the end of the handle. The target for completion of the test was 10 million cycles. The bearings were set up in thiree configrations: * fixed reive to the housing, so movement is between bearings and shaft; * fixed rtive to the sbaft, so movement is between bearings and housing; and * free-floating. 34 WADP-Wld Bo* Waer and Sdado Pawn= Summary: Bal Races Vernus Psidc Bearings Bano Plastic beaing Cost Altbough ball macms are widely Plaic bushes can be produced very available, they ar univessy chaply and to close tolerances costly; vey few developing using iecion molding; trensport countries have the specializd and stoag ate also relavely fities rquited to manufe ceap. them so thy must usually be Fldng Ball rac need housings and Fitting can be a quick, nonakillod shafts macbined to clos oerio; accuae machi and toleraces. Fitting needs e and alignme of ote pump com- efectivo quality contrl. poneats is less critical; plastic bushes could be designd to fit doack om. Conosin Ball races are liable to corrosion, Corrosion problem in st4o and bothinh servieo and duing in us woud be elminated. dore Crsion wil cause a ball race to fail long beor it would bv worn out lt _ Old bengs ca be vety dfit Plstic bushes ca be desined for to remove, and nw beang need easy rphan n the fiedd to be teat with cam nd kill in spamaranulkely to be fitting. iorowed for other piupose& these configu osm were intended to determine the optimum arangement for relaive movm t between the bearings and the asociated steel components. The convendonal ngineering view is that the bearing should be fixed relative to its housing, because an incease m b ng area resuting from running the bush on its outside dimeter is likdy to be more than offset by greater wea resdting from an increase in sliding vdocity. The free-floating conigon was also included becau it offered the greatest potential for simplcty in manufacte and fitting. In fac, both the bearings fixed relative to their housings and the feoe-floaing types produced encouraging results, with an average diameter wear of 0.22 mm per million cycles. Wear was substantially constnt thrughout the test (see Figure B4). Higher wear rates were recorded for bushes fixed re ve to the shaft. Pomm Repn SeIe 35 40 - 0 200 400 600 800 1000 Hours Figure B4. Wew h of psic bearb t was clear tht the condidons of the test regim were a good deal more svere t actal conditions in the field, however, for which a wear rate of around 0.1 mm per million cycles was indicated. Assuming that wear up to 1 mm would be acceptable on pumps in the field, this yields an anticipated service life of 10 million cyces: equivalent tO two year of intensive use. The wall thickness of the bearings would allow the pump to continue to operate with clearances greater than 1 mm, but this could lead to undesirable shock loadings on the bushes and other components of the pump. A number of beafings were also installed in pumps in the field. Iidtial resu we favorable but Inconssten here were some indi that these nconsistenci in perfomnce were elated to variations in the surae finish on fte pivot pins The test sites also presented diffent opertng conditions in tem cf water level, pump use, and exposure to direct sunlight and dust. Comparative testing of a variety of bearn types Subswquent research was based on the Afiidev pump, and was canried out in consultion with the Afildev design team in Nairobi A number of differt types of being were compared: * One-piece plastc bushes in a variety of matials * Two-piece plastc bearings in aceal and nylon * Spherical plain beaings * Wooden bushes * Flexurl rubber bushes 36 UND.P-Wod aWn Watr and San Pon Vwn The test method and regime wee mlar to that used for the nitial research. The target for completon of fe test was S million cycles. One-piece plasc bushes Five materials were selected for testing: • Unfilled acetl copolymer (Kematal) * Acetal copolymer incang a lubricant (Railw PV8O) * Unfflled nylon 66 (Zytel) * Nylon 66 incoporating a lubricant (Nylatron GS) * Nylon 66 incorporatng a lubricant and other fillers (Nylaton NSB) All were tested as simple top-hat bushes running against a stainless sel counterface machined to a surface finish of 0.4 pm R.. Two-pece plastic bushes The two-piece design was developed in response to evidence from tstg in the field that the performance of platiNc bushes would be stmangly infuenced by surfce finish on the bearing counterface. The princie of the two-piece bush was that the outer part would be fixed relave to the houing and the inner part fixed elaive to the shaf: thus bothe ng and its countefac can be controUed, and both would be replaced simultaneously. The beaings were deigned with asisance from SKAT and advice from DuPont of Switzerland; the outer part of the bearings were acetal homopolymer incorporating a lubricant (Delnn CL), rwuing on inner bushes of unfiled nylon 66 (Zytl). The molds were manufactredin a workshop in Kenya. Sphedc plain beangs Spherical plan beaings (SPBs) are widey used in speciist applications such as aircia and racing cas. They are often referred to as Rose joints,ahough at name refers to only one of several manufcture. Te beaings chosen for tstig were of a type, said to be maintenance fre in that they required no lubrcaion, in which the inner spherical element and the matching housing were separated by a layer of PTFE-impregnated fabric. The bearings were pressed intD flanged sleeves m ined from sainle steel to enable them to be fitted in the same way as the plasic bearings. Pmgrm Rezt $Wia 37 Wooden bearings Wooden bearings have been used In handpumps in a number of countries. The test bearings were made from Mri wood from Tanzaa, and were untreated. Taking account of work at the University of Dar-es-Salaam and elsewhere, the bushes were insalled so that tie bearing loads were canied predominantly along the grain axis. Fkxural rubber bushes These are commonly used for vehicle suspensions and other applicaions where isolafion of vibration is desirable. There is no sliding contact: the bearing reies on the flexing of tie rubber. To enable off-the-shelf components to be used and to achieve the necessary angular movement, bearings were used in pairs, a total of eight per pump. Rlesults The results of the laboratory tess are summarized in Table B1. The most encouraging results were obtained for the two-piece plastic bushes. The greatest amount of wear was measured on the Delrin outer bush of the most heavily loaded bearing, at 0.05 mm per million cycles. No measurable wear was found on the inner Zytl bushes. For compaison, the best of the one-piece bushes wore at 0.07 mm per million cyclea. The apparent lack of wear in the nylon bushes may have been the result of the beaings' absorbing water, however. Nylon tends to absorb water and epand, and in this case the epansion may have been sufficient to offset any wear. There is some evidence for this: one of the bearings was found to have incrsed in size during the testing. The spherdcal plain bearings developed considerable end float, and the runnig sufaces were scored. They were consdered to have failed. Simila results were obtained from tests in the field. The wooden bearings failed before the wear rate could be established. T.he flexural rubber bushes failed as a result of fatigue in the rubber. A number of one-piece busbes machined from Railko PV80 were also sted in field projects in Malawi and Kenya, using brass and stainless steel pins. The reut showed a high deee of variation, which seemed to be related to the surfce finish on the pins. Pins that had not been machined to a high standard or had deteriorated in use exhibited much higher wear rates. 38 UNP-World Bo* Wawer md S&nad Pogmm Table Bl: Results of Laboratory Tests on Various Bearing Types No. of Fulcrum pin Bearing Maimum cycles temperatue temperatre wear rate (x 10) (°C) (OC) (mm per 10' cycles) One-piece plastic bushes Kematal 3.2 53-82 3844 0.19 Railko VP80 3.2 52-78 36-52 0.17 Zytel 3.5 67-89 41-55 0.34 Nylatron OS 3.5 60-88 46-58 0.69 Nylaton NSB 3.5 35-52 32-41 0.07 Two-piece plasdc bushes . Delri CLZytel 3.7 30-41 30-39 < 0.05 Spherical plain bearingsl 3.2 35-55 28-44 n.a. Wooden bushes 0.17 - not recorded - Flxmal rubber bushes 4.1 37-61 n.a. n.a. a Fulcrum pin and bearing temperatwu reoorded from inner and oute tacks, respectively. Testing at elevated temperatres and humidity The two-piece or composite plastic bearings were subsequently adopted for use in the Afridev pump, with a slot in the inner nylon bush to prent sweflling that might result from water absorption and cause the bearings to seize. However, a number of bearings fitted to pumps in the field recorded wear rates substantially greater hn anticipated. Furte ting was undtake by CRL, jinty funded by ODA, UNDP, and the World Bank, to investigate whether the wear was reltd to the higb. humidity and elevated temperaws found at many instllation sites. The bearings were tested in two stages. In the fist, the dimensio of the bearings were measured for increasing temperes at 99 percent relaive humidity, under no load. The bushes rotated freely at all tempres. At 40°C, the two parts of the bearing could no longer be separated, though they could easily be removed from their housings. At 55'C, it was no longer possible to remove the bearings from their housings, Nrgram Rq&, Seri ig though this may have been the result of the formation of rust inside the housings in addition to the expansion of the bearings themselves. When the temperature returned to 200C and the humidity to 55 percent, the bearings could be removed and dismanded eadly. The bearings were discolored, but the penmanent changes in their dimensions were very small In the second phase, the bearings were installed in three Afiidev pump handles, and wear characterstcs under load at elevated tem s were assessed. Two of the handle assemblies were contained mwthun a controLed environmental chamber; the third remained in the general enviroment at an ambient temperature of about 30"C, as a control. All the bearings were initally bedded-in at an ambient temperature of 301C for 500,000 cycles. The tempeture in the environmental chamber was then increased in increments of 10°C to a maximum of 90XC, running the bearings for 125,000 cycles at each temperture. Bearing tempees of up to 1 10°C were recorded. On average, the higher-loaded bearings ran 18 percent hotter than the ambient temperature. This obseration may prove a useful rule of thumb for estmating maximum bearing temperahte for a given ambient temperatue. No significant wear occurred in the bearings fitted to the control assembly, and relatively little wear was recorded in the bearings tested at elevated tem s. However, in the "hot" bearings, some of the outer bushes split, and it was clear that these problems were associated with residual stresses in the bush around the lug which detmined its position relative to the housing. A number of the inner bushes also aed, but this occurred for both "hote and "cold" bearings and was attibuted to the substantial side loads imposed by the test rig. Overall, the wear rate of the bearings increased at higher temperatures, but the greater wear was not in itself sufficient to explain the ralpl faiures observed in the field Oher fictors that might be significant include hunidity mnside the pumpstad, and sand and dust carried by wind or rain. A small sample of dust from a bearing used in the field was found to contain 67 percent debris from the bushes themselves, 16 percent iron oxide, 5.5 percent zinc oxide, and 2.5 percent sand. The dsgniince of these quantities is not known. Even this small amount of sand combined with a few metal partcles could cause rapid wear. The Afridev pump was subsequently fitted with stinless steel liners in the bearing housings to eliminae corrosion, and the top cover was modified to offer the bearings better protecdon from the elements (see Part C). 40 UEDP-Wo Ban Warr ad Sitaon Progrwa B3. Sealless Pistons and Diodic Valves The idea of 2 piston with no iternal moving part is an atntctve one. Solid-st technology has tnsfmed dectnics and made posble standards of performance and rliability which woud have been completely imdcal using mechanical teniques of tching and flow control. Solid-state contl of fluids is also wel esablised and extenively used in aplicatons where mechanical control devices would not be pactical: inside nuclear power plants, for example The Consumer Research Laboratory, in consultaion with SKAT and the Univerity of Sheffield, UK, undeook to investigate whether thes techniques could be applied to handpumps, both to eliminate flexible piston seals, which have a hmited life, and todevise walves with no movg parts. The CRL research was fimded by ODA. A number of handpumps have appeared in which the seaing of the piston has been achieved by means other than a flexible seal. The Ethiopia BPS0 - an early example of a direct-action handpump - had a piston nachined from a solid block of high-density polyethylene (HDPE). w hie piston was machined to provide a clearance of about 0.5 mm to 1 mm within the cylnder bore, with a seies of grooves forming a simple labyrinth seal. The Volanta pump uses a sell piston consising of a slug of stinl stel which is close fit whin a cylinde of glass-fibe inforced plastic. The fit has to be coUled between 0.15 mm and 0.18 mm. Te Pompe UPM has multiple pitons, one at each joint between the 3-meter secdom of pump rd. Each piston is molded in rgid plastic, with a cupap piston body. All these pumps have conventonal valves, however. In other fields of engeering, labyrinth seals are widey used where Slight loss of fluid is acceptable and the main requirement is for a seal which requires little or no maintenance. They generaly consist of a seies of square.booed. grooves combined with a relatively close fit between the mafting parts. Other fluid control devices are analogou to electronic diodes and banstors. Fluitic diodes are solid stae devices which have marked difference in their forward and reverse flow hacrcs, and fluiic tandstors are solid stae devies which can direct the main flow in a fluidic ciruit by adjusig the flow in much smal control ports. haum Rqenadeeos 41 SKAT research into labyrinth seals for handpump pistons SKAT has supported a number of research prects in this field car-ed out at tie Hohere Technische Leluhnstalt, Brugg-Wmdisch. One of these compared the leakage acr a number of piston shwes with that acoss a plain cylindrical piston of the same diamete. The best result was obained for a piston with five simple labyrinth grooves (Figure BS). Ind" of r,latw IbsJca 100 A. Plain cylindrical piston 5. Cylindrical piston with 5 :ijjiill ~~~~~labyrinth grooves so ll* I C. Internal annular diffuser 40 i f2 fiIflSD. internal diffugser and 3 ~~~~~~ ~~~~~labyrinth srooves 20 :Ij . IIJif. Internal diffuser with 4 .jiI II ~~~~~~labyrinth grooves and sharp JIguRe S. Laskagof psn wih abyrint seb A futher study confirmed that dte reduction m lakage ob ed by adding a simple labyrinth seal was mainained across a rage of piston speeds from zero to 60 cycls per minute. Another study concentrated on tie shape of tie labyrinth groove, comparing squae with round-bottomed and tapered forms. little diffance was found between square and round-botomed grooves, and groove depth also had little effect. However, a conspicuously better result was obtained for a single tapered groove inclined at 6 degreas to the pstn axis. The application of fluidics to handpu The piston in a handpump i essenally a diode, with minimum flow across it on the delivery stroke and fee flow on the return stroke. In a conventional piston with 42 UNDP-W Id Bwa Waw and S _aae huinwn an efficient seal, there should be zero flow on the delivery stoke, and all tie water disptled by the piston will flow across it on the return. A fluidic diode cannot match this pfmance; tere wil always be some leakage, and improving the perfornmnce in te orwad direction must enta soni loss of performce in the rverse directon, md vice vera Moreover, becaus it exploits dynamic effec, the overal perfornunce of a fluidic diode will be directly popdona to the flow rate and hence to piston speed. Neverdel, the best fluidic diodes already in use in ote applications offer diodicity - h atdo of forward to mrve flow - of up to 11, and useful futons can be performed with diodicities of about 5 or 6. The simplest form of fluidic diode is the conical diffusere B6). In the forward direction, the prese loss is reduced due to efficint conversion from dynamic to statc pre in the gradually divergng onical secdon. In the reverse dirion, flow is consticted as the pressure builds up gradually in the conical section, before expanding suddenly in the t The diodicity typically has a value between 2.5 and 3. torwwd OONIQ'J. DIFFU ER VORTEX DIODE Rpg B6. Fluldle dio Mbe vorte diode is a somid devie constng of a spal chambe, the vortex, wi conical diffuse acfting as ports at the center and at a ngnt to te scmll. Forward flow is into the ceter of the spiral and out at the tangential port, and offes ratively litl resistance. Revse flow is restricted by fte conical diffusers and by ihe hqgwn R.pu S.d. 43 water being forced from the outside to the center of the spiral. Diodicides up to 11 m possible, depng on flow rate. A practical design Applying the concept of a conical diffuser to te piston of a pump, th diffuser might be formed as an amulus around the outside of he pston: ben the piStOn and the cylinder wall (Figue B7). This is a very simple shpe to make, hougb some experimentaion may be required to opmize the radius at the lower dge. Tess on an initial pototype also suggested that stabilzing vanes (as shown) may be nec_my. Such a piston might be expected to provide about half the outut of a pisto with a perfct seal at 60 cycles per minute, with the output reducing at lower speeds (Figure p). bow hla reslance r Su o FTgure B7. Diffuser piston Better reuts could be exeted frm a piston Moporn a vortexdiode, particulary if this were combined with a labyrinth sea between the piston and the cylinder wall. Combing the rsults of the rearch ponsored by SKAT with provided by the Universty of Sheffield, there is the prospect of a compledly soid-estae pston capable of providing about 70 peroent of the volume flow of a perfey sal piston at 50 cycles per minute, and about 50 pecent of the volume flow at 30 cydes per 44 UNDP-Wored Bak Wa erad Santto mhgn minute (Figwe . Note tiat the loss of outflow woul be compensat by a reduction in the z*qur effot (and of the stresses in the system) on the delivery stroke of at least equal prpto. A given rate of outflow could therefore be achieved by simply Incai the cylinder diameter. There would be some hydralic friction on the return stro, but this unikely to be greater the mechanical friction present in a coventional anet. fbow index 100 40 t ,. 20____ - Vortex & labyrinth f, , - -- Votex diode O 1- w w -^ ~Arnusu dlltuser 20 s0 40 60 s0 OpwreUnr speed ({o-minute) fgm B8. Potal flow rates for solid e Ih rate of outflow would depend strongly on fte piston velocity. Piston vdlocit ar generly low in lever-arm pumps, as a result both of the geomety of the sytem aind the ineria of the levw a However, te piston velocity i inherently great in direte-acton pumps, and users have a better opportunity to apply a rapid acceleratIon at the sart of the ddivery stoke. Solid-stte pistons migt therefo be particularly suitabk for direct-action pumps. A number of posse design schem were proposed for both single and mutile piston a mnts. An example is shown in Figure B9, which illustra a scud-st pon pm a Vortex diode as the valve and a seies of grooves ating as a labydnth sea. Ihe vorta is open to the outside of the piston to enable it to be molded in plasc. In use, the vortex would be enclosed by the cylinder wall. pf%%Bwn Repu &SW 45 Some expimentatIon wod be requiid to optimize the diameter of the conical diffuser within the body of the pio and the lmance between th piston body and the cylinder bore. . - - SI lrrm~~~~~- ILIvI F1gn B9. Desig for a sogd-tte piston 46 tNP-W.U DW& Wow rd Sad ale Pioua B4. Light-Weight Pump Rods for Direct-Action Pumps Direct-on pumps have much to commend them for water depths to 15 metes or so. A direct-acton pump is one in which the conventidonal operating lever has been replaced by a simple t-handle attached directly to the top of the pump rod. Cleary, this eliminates a good deal of the complexity of a conventional lever arm pumpstand, and pardciularly the bearings, which are susceptible to wear. Direct-action pumps have the potential to be easier to manufacture and to maintain. Merely attaching a t-ar to the pump rod of a conventional pump would result in a very unisfactory direct-acton nump, however. The user would be required to exert a very stg upward pull during the delivery stroke, while the weight of the rods would send the piston crashing down on the retn stroke. In a conventional lever-arm pump, the lever not only provides mechanical advantage, but also reverses the direction of the operating effort, and the weight of the lever helps to countebalance the weight of the rods. To achieve a aisfctory direct action design, it is necessy to transfer part of the effort required on the delivery stroke to the retu stroke. This can be achieved by using lightweight, high-displacement pump rods. On the upward or delivery stroke, the pump user has to overcome the weight of the rods themselves, plus that of the annulus of water around them, which is supported by the piston. To reduce the Lifting force, therefore, it is necesary both to reduce the weight of the rods and to increase their bulk, thereby reducing the cross-sectional area of the annulus of water. Ihe effort that is saved is transferred to the return stroke: as the piston travels downwards, the bulky pump rods displace water fom wthin the cylinder and pump it to the surfice. By lectng pump rods of appropd weit an4 size, the distrution of the total opeting effort between the up- and down-strokes can be controlled. Note that there is no inherent requiement for the pump rod to be less dense than water, however. If the weight is low enough and the disDlacment great enough, the rod will be -buoyant," but this is not a fundamental requirement. The obvious choice for a lightweight, high-dispLacement pump rod is a tu, sealed to prevent water gettng in. Various materials have been tried: the Tara pump uses uPVC pipe, as does the Wavin, whra the Pek pump bad aluminum tubes and the Nira AP8S uses a polyethylene pipe, for example. Fiblss and thin-waUed stainless steel have also been sugested. TMe Consmer Research Labrtory (CRL) undertook to identify and test a selection of mat which might be suitable for use as pump rods in direct-acon pumps. The project was finded by ODA. Pm8nrm Roa Sofa 47 I User preference The first stage of the research ivestgted how users would respond to different arrangements for distribudng the total opeatng effort between the up- and down-strokes. Five configurtions were compared, manging from maximum force on the up-stroke, via equal forces on both parts of the pumping cycle, to maximum force on the down-stroke, but all the configurtions required a total work input equivalent to 50 watts at 30 cycles per minute. Usen ranged from 10 years old to adult. Although avenage work rates turned out to be very similar for all configuraions, the users showed a consstent prefeence for arrangement where either the up and down forces were equal or the down-force was slightly greater an the up-force. Computer model A computer model was deised to examine a wide selection of potential pump mateials and dimensions in terms of their effect on the operation of the pump. Inputs to the model were: - pump rod material * pump rod form, dimensions, length between connections - yhlinder bore and pump stroke cylinder dep below ground * water depth below ground The outputs from the model wee: estimated optng fores on up- and down-stokes s likliood of the pump rod's buckln under compression * stroke rates for a range of work ates a rate of water outflow * effect of leaky rods filing with water The last point was included because the c cs of hollow rods could change markedly if they leak and fill with water, as has been a problem witi some direct-action pumps tested in the laborao. 48 URP-Wod Bw* Waw nd s&nde P*gmn Market survey Markt research was unden In selected counties to assess the availabilitY of mateials aprim-Aing to the prred ons. * uPVC Widely available: 42 mm-diameer, 2.7-mm wall pipe slected for tsting * ABS Less widely avaiable than uPVC; less sdff, hence a rdativey thick wail secin would be required: 42 mm-dameter, 4-mm wail pipe selected for tesfting * Polypropylene No suitable materials were found • Aluminum Difficu to find in sdtable sizes; tube in the prefer iwze of 44-mm diameter was only available at great cost: 32-mm diameter, 1-mm wal tbe elcted for tesing, in Grade 6063 alloy (BS 1474:1972). • Stnl seel No suitable mateials were found lass-reinforced Difficult to find; a suitable matedal (plasdc) was dentfied, but it was reected due to very high cost * Polyethylene Readily available as condnuous hose: 42-mm diame- ter, 3.7-mm wall pipe wa selected for testing Performance test Prototype ptmp were constucted in ag each of the sdected pump- rod mtteials, ad tested in conjunction with a T1a pumpstand and cylinder. No pump- rod guides were used. lbe pumps were itlled at 20 metes depth in a borehole, with te water at 15 met depth. Joint in the uPVC and ABS pipes wer made using solvent cemt lhe aluminum rods were conected using exy resn. The polywhylene hose was in ed as a .-ngle length with no intmediate connections. The resuts of the tsts are compar in Figue B1. There was ttle to choose between uPVC and ABS. For both mateials, te resuls were in close general agreement with the computer model at 20 and 30 cycles per minute At higher speeds, ineased ficfion suggests ta the pump rods were rubbing on PnW= Avon Sof 49 Outflow (Itrs/min) 60 - - 1 uPVC I 1 -4-' ~~~~ABS tSO--1r j -i ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~....... ................... -*.....Almnu 6o------ -4-X- Aluminium i l j/ / ~~~~~~~~~43- Polyethlylere 40~~~~ ~ ~~~~~ ~~ ~~~~~~~~~~~~~~~~~~~ - ---------- 80' ,_ ___ ... o 60 100 180 200 260 B0o Wbrk Input (W) Fue B10. Pump performance with vaiou rod materias he rising main as a rewlt of the rods buckling under compression. For the aluminum rods, consistey high lvels of efficiecy wer reorded at al opating speeds, suggesting tat the gater r t of the aluminum tube prevented significant contact between the pump rods and the fiing main. The opposite effect was observed for fte polyethylene hose, for which fte measured efficiencies were consistently lower than for othe materials. Enduran tests Pumps i ng each of the rod materials and using Tara pumpstands and cylindes were tsted for 4,000 hours at 32 cycles per minute, simulatng a wat depth of 15 metas and a cylinder immersion of 5 meters. Out-of-line fore were thought to have contributed significantly to the various faures In the case of the polyethylene rods, the out-of-line forces were caused by armanent set n the hose as a result of its having been coiled. i the aluminum rods, the out-of-line forces were transmitted through the handle: the test ng was designed 50 VNP-Wod Bo* Wawr and &Sauinm Prog to Impose slight out-of-line forces during the pumping cycle, typical of those actually applied by people using direc-action pumps. SwIma Lg e pump rods endurance test tiPvc AB3S Ahuinhium Polyuedlui HOUrs Hours Ho= Hours 0 Stat 0 Start 0 start 0 Sta 599 Bottom bmkc. 207 ho 2007 207 2007 20p 00 otiao oo tic tion 2246 Top broken 3080 Top con-c- *db 3522 Top brken 4006 Final 4006 Final 4006 Final 4006 Final - ~~~~~~~mo- ku"spime tic. --1 -h- -------ion tion tio- Each of the pump rods was careflly inspected at the end of the test, and the rig mains were examined interaly. * uPVC Wear was evident on both the pump rod and the rising main, partcuay where lengt of pump rods had been joined. The patten of wea suggested that it had been caused by the pump rod's bucking during the return stroke. The epoxy cementjoint at the top Progim Reporn Swt5 51 of the pump rod, In an area subjected to relatively high stress, was in good condition. * ABS Wear was evident on both the pump rod and the rising main, the pattern suggesng that the pump rod had buckled in an S-shape during the return stroke. The maximum wear was not as severe as for the uPVC pump rods. * Aluminum Very lttle wear was found on either the pump rods or the rising main, suggesting that the aluminum rods had remained rigid throughout the pumping cycle. However, the rods were conroded, and thls had caused a pin hole which had allowed water to enter the pump rod. * Polyethylene Extensive wear was evident on both the pump rod and the rsing main. The pattern of wear suggested that a combination of the original set in the rods and buckling during the reun stoke had been responsible. Paricles of sand were embedded in the pump rods and had caused accelerated wear of the coreponding areas of the rising main. Conclusions AUll fte ba plastic pump rods seemed to buckle on the return stroke, and any pactial design should td account of this. The results for uPVC and ABS were similar, ABS showed a greater tndency to buckle but a lower rate of wear when used with uPVC ising main. An iuesting combination might therefore be uPVC pump rod and ABS risn main. his woud maximi the sffns of the pump rod and minimize problem of lack of stiffnes in the ABS. Polyethylene is pariclarly cheap and natrally buoyant, but it is not as stiff as the other plasdc materials and tends to pick up particles of sand that will inevitably acceera wear. It is normally supplied as a coil so that acomplete length can be cut for use in eacw pump, dinating intermediate connections. However, the set in the pipe which results from coiing it will exacate the tedency to buckle under compresion. Although the aluminum rods were stiff enough not to buckle under compessiln, their low fatigue resistae madehm unable to cope with the flexing which can be expected in a direct acion pump, and thir corodon reistance was hadequate The matial tested was chosen after extensive discussions with the nufur, and was selecte for its combination of mechanical strength and corrosion reistance. Other grades 52 UNDP-WO d Dw* Waw ud Snkadon P_iw of alloy are available with better ftigue restance, and anodizing could improve corrosion resistan, but on the basis of tis research aluminium cannot be recommended as a material for the pump rods of direct-action pumps. B5. Stress in Handpump Rising Mains znd Pump Rods The stresses in the below-ground asemblies of handpumps have been fte subject of considerable debate. Many attempts have been made to assess theoredcaly the relative dgnificance of the quasi-static forces (the "dead weight" of the components and of fte waw), the dynamic forces involved in accelerating the pump rod and the wat column, fricton at the piston seal, shock loads resultng from the pump handle hitting its stops and other possible fatos. Concern centered specfiay on the use of plastic rsing mas in deep well pumps. Plastic, partcularly uPVC, rising mains have a crucial role to play in making handpumps suitble for village-level maintenance. uPVC pipe is widey available, already manufacured in a number of deveoping countries, and cheap. It is light in weight (which makes it possible for below-ground assemblies to be withdrwn for maintenance without the need for lifting tackle) and inherently corroson res4nt uPVC is already used extensively in water supply, and with sucess, but plcations rarely subject the pipe to dgnficant longitudina stress. Existing standards concentrate on the dimensions of the pipe and its ability to withstand presue. To develop an und ing of the potental of uPVC pipe to perform reliably as the rising main of a handpump, a lkowledge of the working streses is sential. ODA, UNDP, and the World Bank jointly sponsored a resch project by the Consumer Reserch lAbty to measure the stresses in a workdng rising main, and the c ondin stresses in the pump rods, for a varety of pump cofigurtions and for different modes of pumping. A paralld program of research was also undaken by InterAction Design, sonsored by the Netherlands Mint for Development Coopeation. The CRL pilot study A pilot study was undertken to establish the best memet positions, test procedures, and technologies. The work was done in CRL's handpump test tower, and all parts of the system were available for access. Two pumpstands were used: an Afidev and an India Mark 1, to compare fixed and flexble connections between the pump hrpm RepSi &m 53 rod and the handle. Tests were made using a solid rising main mount and the rubber cone mount from the Afridev pump. A variety of pumpn techniques were used, ranging frn relatvely dow, full strokes of the handle to fast, short strokes. Extra tests were made for full strokes with the perator gwg the handle to bang against its stops. The rising main on test was 60-mm outside diameter, 3-mm wall Class C uPVC pipe man d in the UK to ISo727 (mateals) and BS 3505/3506 (dimensions). A relatively thin-waled pipe was chosen for the pilot study to ensure that sufflent easion would be obtained in a length of 7 m. =^bE m: Z~~ of StEw ¶lable E3l: CR MaueetoStesIn Rising Maidn Component Measurng device(s) Reference (Figure Bi1) Handle Strain gauge measming force applied by pump opaor 7 Angle displacement ransducer measuring handle movement 8 Pump rod Load cell to measure tensile forces at top of pump rod 6 Vertca dipla ement transducer to measure pump rod movrement 1 Accelerometer measuing accdeton of pump rod 2 Rising main Strin gauge 4-way mounting plate meauing both tensile and bending forces at top of fising main 9-12 Li exeasomet maung etsion and comssn over whole length of the nsing main 5 Strain gauges msming tensile forces at bottom of rising man, immediately above cylinder 3 Pres transducer meuring presure inide rising main imdia y above cylinder 4 A wide range of mesuing insuments was fitted. Outputs from thk devices were recorded on a high-speed, 12-trck UV recorder to enable study of not only the absolute values of the outputs but also their phase ela is. The use of a mi ero as a mianndl data logger was aso considered. It was rejected when it became clear that the available memory would be filled whin a few seconds of pumping, as a result of the fast scanning rate required to capture rasient effects It bad been anficipad that the pilot test would show that some motoring sttons would be puing data of minor significance that could be omitted in subsequent 54 fINP-WwI Bo* Warn and Srk&,'nO Pmgw research for which boreholes would be used. This proved not to be the case, however, and it w apparent not only at a fUlly comprehensive test rig would be required for the borehole tests, but also that a wider range of handpump ons should be included. It was also clear that short-term transient effects were important. Even when pumpng gently i near simple harmon motion, the four-way stain gauges on the mounting plate (Figure B 1: items 9 to 12) detect relatively high frequency variations in sts in the rising main, and other results demons the additonal stresses induced by apid pumping and by banging the handle against its stops. CRL's main test program For the main test program, a wider range of pump configons was tested, including a Volanta pumpstand in addition to the Afridev and India Mark II types, with cylinders of 50 mm bore (Afridev) and 2.5 inches (India Mark II). The combinations are shown in Table B3. Rising main pipe was Class E, 60 mm outside d , 3.9 mm wall The tests were carried out in a borehole at water depths of 25 meters and 45 mets. The Volanta pumpstand was included as an enample of a pump which moves the pump rod in near simple harmonic motion, and without rik of the handle's being banged on its stops. The eimination of shock loads has often been dlaimed as a sgnifiant advantage of flywheel-type bandpumps. Tabe B3: CombInations Tested In the CRL Main Testf m Afridev India Volanta Afridev India Ring Rising Rubber stand stand stand cylinder cylinder main main cone 50 mm 2.5* 25 m* 45 m Ig~~~~~~ J 10v Wate level below ground; cylinders immersed 3 m below water level PMA=iu Repei S.iI. 55 11 ubber cone ~~) assembly 1 Pump rod transducer 2 Acceermeter 3 Strain gauged rising main 3 , 4 Pressue transducer : 14 5 Rif g maini l 6 Pump rod load cel 7 Strain gauged handle 8 Handle isplcee w ~~~~~~rnducer 9 ) Po-way ain gauged 10 ) rising main mounting 11 )pae 12 ) Igue Bll. Test equipment used on India Mak II pump head X6 UNDP-Wod Bnk Wow and Swaih Pmpm For each co on, tests were carried out for ihree pumping techniques: * 40 cycles/minute; maimum strole length without strikng the bump stops * 80 cyclesminute; maxdmum sroke length without striking the bump stops * 80 cycles/minute; maximum stroke lengtli including striing the bump stops For the Volanta pumpstand, tests were carried out at 40 and 80 revolutions per minute. CRL's main test results A vast amount of data was geneated and stored on magnedc tape for subsequent analysis. Figure B12 illustrates part of one the traces: for pump-rod tension in the Af&idev pump, with a solid n- main mount, opeated at 40 cycles per minute fm a depth of 45 meters. The trace is typical of the results and shows a good deal of seodary va speimpo on the pimary pumping cycle. Pump rod tension WkN) 1.68 [ 0 0 4~p 0 1 2 9 4 6 Seconds Jigr B12. A typical result from CRL's tests For most confi, the m stress in the idng main occurred at the bottom, as a combin of tensile ste and hoop stress caused by the iternal pressure. However, the greatest value recorded in the tests was at the top of the iig Progiw Repa Swim 57 main for the India Mark U head with a solid rising main mount, at a stroke rate of 80 cycles per minute and with the handle banging on its stops. The measured maximum load was equivalent to a stress of 7 MPa. This amounts to about 13 percent of the ultimate tensile strength for uPVC (55 MPa at 230C. Even allowing for notch effects in the rising main, which are almost certain to be present in any practical design, it seems unlikely that uPVC pipe of appropriate size and quality will break solely as a result of the levels of stres imposed on it. Failure is more likely to be caused by fatigue, and in the determination of the fatigue Life of uPVC it is the viation in stress rather than the maximum stress which is significant. Figure B13 illusaes the variations in tensile loads in the rising main measured by CRL for the Afidev and India Mark II test configurations. rw purV head Irds Mark Tmo pmo p head Aiming simm Ai~~~~maing ffum :mufle laud wiatiari (kn) tnie foad wledwn (k) = 4r 4^ _ __ 1Rubb0rmon o o A D E F a eo E f A - 25m depth, 40 cyles/minute B = 25m depth, 80 cyclestminute C = 25m depth, 80 cydes/minute strildng bump stops D = 45m depth, 40 cycles/minute B - 45m depth, 80 cycles/minute F = 45m depth, 80 cycles/minute sttikdng bump stops Fgun 1B13. Variations In axial loads In the rising midn These variations in load tnslt into variations in str, shown in Table B4. Note, however, hat these, strus vaaions reabe only to the partcua pipe chosen for CRL's tests, and not to actual Afridev or India Mark II pumps, for which the stress varii will depend on the dimensions of their riig-main pipes. 58 UNP-Wod Bank Water and Saniatai Pmogra Table B4: Variations In Ring Mai Strew Water Pumping Riser Afridev config. India Mk II depth speed mount stress variation config. stress (m) (cyceslmin) (MPa) variation 25 40 solid 1.2 25 40 rubber 1.3 1.9 25 80 solid 1.2 25 80 rubber 1.4 2.7 25 8OBS* solid 1.3 25 8OBS rubber 1.5 2.6 45 40 solid 1.8 2.4 45 40 rubber 1.8 2.5 34 80 solid 2.7 3.5 45 80 rubber 2.8 3.7 45 8OBS Solid 2.8 3.9 45 SOBS rubber 2.9 3.7 B BS = striking the bump stops. CRL analyzed the phase relaonships between the various measure- ments to detmine the sequence of events in a typical pumping cycle: an example, for the Aftidev bead optng at 40 cycle per minute, is shown in Figure B14. At the start of the cycle, the bandle moved down, the piton moved up and the piston valve closed. this caused an xtension of tie pump rod, which then sprung back and oscillated briefly. The pressure in the cylinder reacbed a maxmum as the column of water was accelAed. The rising main shortened as the weight of wat was taken off the footvalve and friction at the piston seal incrased. The system then oscillated at about 6 Hz. On the return stroke, the foot valve closed, the pump rod decelerated and oscillated at about 20 Hz. The oscillations in the pump rod can be seen in Fue B12. During the return stroke, the rsing main lengthened and oscllated at about 2 Hz. CRL's conclusions It is difficult to predict the likely fatige life of the rising main. The fatigue chrceitc of a particular pipe will depend on the formulation of the raw material, the method of manuf ture, the degree of quality control, the addition of stabiizes and lubricant to assist in prcessing, and of fillers and rubbery mateias to improve the impact sgth. hoVugrn Reo Serie 59 On lldi P9il oyl at 40 "b Inln -~~~~~~~~~~~~ I 3~~ gi 0 -- 0. . U°Ti* (secon)1 M~gure B14. Sequence of events In a pumping cycle Creep, log-term elongation of the pipe under continuous load, is unlikly to be a problem provided that pumps are installed with a clearane between the pison at top dead center and the top of the cylinder. The static stres is relatvely small,, and will give rise to a strain of between 0. 1 and 0.2 per cent per year at 201C, equivalent to an extension of 7 mm per year for a 47 nm below-ground string. The maximum stres induced in the pump rods as a result of the tensile loads was 27.8 MPa, considerably greate than the stres induced by acceleration. Since the ultmate tensile stres for stainess steel is around 800 MPa, ther would seem to be little risk of pump-rod breakage. However, sumes will be cocnrtdat the roots of the theds, especially wherethiese are cut rather thanrolled. A typical sts cnetrto facor for cut threads is thre, though considerably bigher figures are possible for badly cut threads,, or where threads are not cut square with the axis of the rod. There is no specific information on the fatgueprorac of threded PUMP rods, but data is availale for steel nuts and bolts. This indicates that the stres solnot exceed 50 MPa for badly cut threadand l50 Mpa forrolledbreads to give a atgue life of ten million cycles. It is therefore reomeddthat rolled threads be used wherever possible,, and particularly for 10-mm pump rods, to ensure, a usatifatory fatigue life. 60UNP-Wo R*1 Dank Warad Sanitatio Pn~mn Other factors in CRL's tes The rubber cone riig maIn mount had the effect of dampng the osillationsin the rsn main, but g their amplitude. The overall effect was geneSay to reduce the maximum sts at the top of the rising main (rticularly witi the India Mark E configuon) though not the variation in stress. The rubber cone also tended to reduce slighldy the output of the pump by shorteing the effective pump stroke. The rotary configuration (Volmta pump head) gave the lowest values for pres above the piston and for deflection of the rising main, but the levels of stress were generally similar to the other configurations. Checking pipe quality Many peoe worlang in handpump projects continue to have serous reservations about uPVC as a rig main material. It is not clear to what extent these reseatios are based on direct expeence of failures in uPVC rising mains, however, as opposed to worries about the poor quality of some pipes manured in developing counies. Tem are indications that at least some diff asia with uPVC rising mains have resulted from ia jointing meods rather than shortcom in the pi temsldves. ODA of the UK and GTZ of Germany are collaboratng in research into conneatr for uPVC rising mains, and agreed to extend this work to include the exinaion of samples of pipes manufacd in developing counties The objectives were to give a generl indication of quality and to suggest a simple qualit control test which could be carried out by manu s. Chemical analysis was carried out by the Open University, UK, and mechanical assessments by CRL. Samples of pipe were obtained from Ethiopia, India (three manfaures), Kenya (two manufacs), Malawi, and Pakis (two manuc). For compaon, the same test were performed on samples of the UK-made pipe used by CRL for their measurements of rising-main stress. All but one of ithe pipes was found to have been inadequately processed, to varyin , lea to iomplete fusing of the uPVC pardcles. However, poor gelation, as thius is nown, can be readily detectd during manuface by a relatvey simple test. It is recommended that manu s Should extrude sing main pipes in spedfic bat , with extra suevon and testng of pipe quality. Below-specification pipe need not be toally reected, but can be set aside for less demanding uses such as PMgMM Repe Sal 61 wastwater disposal. Gelation should be routinely tested using a nonsolvent immersion test in methylene chloride (CH2Cl2) or acetone. The test is easy to perfom, requiring only a lathe to turn a taper on tki- end of a sample of the pipe. Very little intrtAive skill is needed, and the test can be caried out quickly near the extrusion line which mimizes the effect on production. Poor gelaion is generally cuLsed by inadequate die-head tempera during the extrusion process. In extreme cases, the full thiclmess of the pipe may be adversely affected. In less severe cases, the poorly gelated material may only be in the central part of the pipe wall, the inner and outer kins being relatively well consolidated. The quality of solvent-welded joints in poorly gdat pipe is also likely to be adversely affectd, due to solvent penettidon between the inadequately fused uPVC particles. Testing by InterAction Design InterAction Design (IAD) of the Nethelands set out to investigate the stresses in PVC rising mains used for handpumps for depths down to 100 meters. Their tsts were based on the SWN 81 and Volanta pumps, and included a "hybrid' pump combining the Volanta below-ground asembly with the SWN pumpstand. Tests of the SWN pump were canied out for water depths ranging from 20 meters to 96 meters; the Vola and hybrd pumps were tested at 78 meters. A number of other factors were also investated (rable BS). The mentation used is imustated in Figure B15 on page 75. The outputs were prossed by two AT-compatible computers. Strain was measured using groups of four strain gauges glued at 90 degree intevals around the PVC rsing main and pump rod, to permit both axial stain and bending stess to be calculated. At the bottom of the riig main, additional tangential stain guges were applied to compensate for contrctions caused by pressure fluctuations insie the rising main. LAD test results LAD drew a sharp distinon between conventional 'sifF nrsing mains and PVC welasicw -ising mains. The elasticity of conventional steel pipes is slight, but for PVC rising main the effect of the pipe stretching and rlaxing under load is sufficient to reduce considerably the actual stroke of the piston in relaton to the cylinder, and therefe the amount of water delivered. Moreover, the resonant frequency of a typical PVC rising main is sufficiently low that reonance may occur at normal pumping rates. 62 UNDP-World Bo* Wau and Sanl Proim Table DS: Combinations Tested In the IAD Test Program - SWN 81 Volanta Hybrid Waterdpth (m)20 40 60 80 96 78 78 Effect tesd lS ising mstin cnt1;iz -4 SttAndrnr 7. Flywheels 1.5 m and 0.75 m 9^ EpdWnd paump h2nde it 1 1.i Pitnvav f Loss of output Figure B16 is kn from IAD's report, and shows calulated and actal results for te volumetric efficienc of the SWN 81 pump (SO-mm cylinder bore, 165-mm troke), for depths down to 100 meters. Almough the efct if not very signi t at 45 meters depth (the experml results in paricular holding up well, at over 90 percent volumetric effidency), a marked loss of outut is indicated at 100 meers depth. IAD's explanation for this loss of outut is illustrated in Figure B17. At the start of the upward or delivery strolce, the pressue genaed by the nhead" of water in the rising main is gradually tansferred from the rising main to the piston. The main contrcts and the piston and cylinder move upwards together, without any sgnificant movement of the piston relative to the cylinder (Phase 1). Only when all the pressure has been transferred to the piston does it moverelive to the cylinder, opening the footvalve and drawing water into the cylinder (Pase 2). On the return stroke, pressure is gradually tned from the piston to the ring main, cauing the rising main to lengthen. Again, the piston and cylinder move together (Phase 3), with no gnficant relative movement unmtil the ig main has been fully etended (Phase 4). The effecdve stroke of the piston is heirere reduced by the aetsiO of the rising main. Pram RqkW &aia E6 I tes.01mfro supeen F , 1 / wlt~*eiht drthe pm o ern ReAe#trin g to tin raxdiard xolm dklspedoeutiet of be flOest ZK.2 Mise cenrall unde the fotWe ZVO sftran emS.B I esemt dfitted n theD cyIndr abt$ the piseaton.ba to masurethe reOf pipston2 dfmisplaemen ROO *tas V,bt oaen aggnI Px C dotube unOde the oUISW Ofef 11~~~~~~~~~. 1" belw IiwOQIU R toa measueet axia dirpectiont glof theos outsideOflor OR M abslowutehrsue tROO dus. pisseydWIn the fylude VROA diitgal Ob In eoxt diecr,tionn gctd to peump rOdt0 i W lil l l rbelwr tho pmp fmd bhupwkv e B Tet equipment ued wtl ten W N_ Ilunox adlon pl n t _ _~~~~~~~~rActw to uthe re In tli dlsotbn. d to otlf * bw t _nlu _ va m~~mwt i th foot VoloP _ _- PI~~~~VD otan ZO pine oxital lyeln luM the 0 ot *XEf _6z # l td mek Wtw tbo RBCdon O bIP. m dw the lld a R t tfan xus d*Motont of to P i 01 m ],Jll _1l OK0 mZnotic tfUld r,ztrone lyr OredI the polndlw, rwotl~~~~~~above thbe pistond Xld_to 196-]>~~~~W4 li t0ota sInkth dooot w cmoo t , Fh I%_B5 Teg e1Rme_ ined wthal th _Oly 81 te Inthumpde, tW VM-W Bik Wa _ and &uuak p r pmdInt ol -~ ~ ~ ~ ~ ~ ~ ~ ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~- lt } 1|1--1g1 l8L~~~~~~~~~-I { it 41 C[soII a~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ approximations. At hgher speeds, however, dynamic effects added ("-nsiderably to the total stress. The most significant dynamic effects were pressure fluctuations caused by resonance of the water column and rising main. At the start of the upstroke, the water immediately above the piston is compressed. The pese is loclized by the inerda of the system, and the rising main expands to accommodate it (water being virtually inompreble). This preu front moves up the ystem at a speed deWmined by the dimensions of the pipe and its elasticity, about 400 metessecond for typical rising main pipes, and is reflected back down at the top. After a few pumping cycles, a complex pattern of pressure waves is set up within the rising main, as newly produced waves intect with reflected waves. Resonance ocous when the waves reinforce one another, that is, when a refleced wave arrves at the moment a new wave is produced. This depends on the length of the rising main, as ilustated in Figure B18. a. 1t + SWSpn , 4 2 i= 0 20 40 s w m igure BIS. Resonant frequency vers length of Mngs mmt At oher frquenci, the effect of intctions of the pressure waves will still be preseit, appearg typicaly as prese fluctuations at aproximatey the resonant frequen supimposed on the basic pressure cycle. IAD analyzed this effect in detail and got good aweement betwee d theiory and thdr actual results, both for mets of preSse and of the resultant es in te rising main and pump rod. 66 IrP-Wo * Bo* Wand Santaox Prgun lAD found that epeiments with the valves of the Volanta pump (halving the valve lift from 10 mm to 5 mm and reducing the mass from 80 g to 25 g) had no material effect on the strains and stss in the rn main or pump rod. LAD developed formulas relatng the stresses in the rising main and pump rods to the physical c cs of the pump and conditions of use. These formulas have been used to produce the figures shown in Table B6. Table B6: Varaons In Rising Main Stress Stress variatios for pupn rae: Water Pump depth 20 40 60 80 100 (m) (cyclswinute) SWN 81 35 0.8 0.9 1.0 1.0 1.1 55 1.3 1.5 1.7 1.9 2.5 73 1.8 2.1 2.5 3.0 2.4 Volanta 73 1.2 1.4 1.7 2.0 1.6 IAD's conclusions and recommendatios • Significant, differeces exis between pumps with PVC risin mains and conventional dsing mains, because of the geate elasticity of PVC. • The most likely camse of failure of pump rods and PVC risg mains is fatigue. * The pump yield will be reduced by the loss of effedve piston stroke that results from a PVC rising main strching under load. * At low frequencies, the stre and strain flucta in the rsing main and pump rods can be accurately predicted by quasi-stac approimaions. * At pumping fequencies etean about half the resonant frequency of the ising main, dynamic efects due to interfeing pressure waves in the rising main become X The reductio in output from sgt of the tising main and the sres in the g main and pump rod can be reduced by maximizing the rato of pump stroke to piston diameter. Paogrm Repn Seaf 67 IAD advises that the str fluctatons in PVC nsing mains should not exceed 5 MPa, and in stainless steel pump rods should not exceed 50 MPa. Stress concentratins should be mimized by makdng changes in section (at joints and couplings, for example) as smooth as possible. Joints must be accurately aligned. Rolled threads in pump rods are prefenred to cut threads. Rising main centraize will minimize the bending stress fluctuations caused * by swingng and saking, though they will not prevent buckling when compressive stresses occur. Allowance must be made for long-term creep, particularly in deep insalations, to prevent the piston toning out" in the cylinder. Overall conclusions and recommendations The sress vanatis measured by CRL and IAD confirm that uPVC pipe has te capability to perform adequately as a handpump rising main, provided the pipe is of good quality and the overall design tames account of the specific propertes of uPVC. The mawimum values of sftr established in this research are well within the maximum ensile sgth of uPVC pipe. The p nat cause of failure is therefore more likely to be fatgue. Pipe quality is important: the fatigue life of poor-quality pipe may be as litfle as one dsxth that of good-quality pipe. The test on samples of actual pipes suggest hat a large proporion of c ally available pipe may be of less than atactory qaty, and that pipe made in the developed world cannot be assumed to be superior to pipe made in developing countries. However, relatiey simple techniques are available fr monitoring pipe quality during manuture: specifiers should insst that these tecniqe be used and a sasactory level of qualit achieved in pipe intended for use as handpump ring main. Featur of handpump design which will tend to maimize the effecdveness and fatigue life of a handpump with a uPVC tising main include: * A relatively small pisto diameter. The dynamic forces in a handpump are dependent an the square of the piston diameter: a 30 percent reduction in pistn diameter wil more than halve the dynamic forces, and hence the stresses. * A rtively long stroke. This offsets the effect of the relatively small piston and ensures ta the loss of stroke wbich results from the sgtething and contraction of the ig is a relatively small prporion of the total stroke. E0 UNDP-W.d Bk Watwr wad SaatI P8wn * Rolled, raher tan cut, thrads on the pump rods. Care must be takn to ensure ta the theads are accurately aligned. It i aso impornt to mnmiz extema damage to the pipe and the pump rods during insallaton and repair. Any dent or sratch can act as a focus for str concentraton, from which a faigue crck may propagate and eventually cause failure. This research has confirmed that an unsympatheati pumping technique, banging the handle on its stops at tie ends of the stroke, will have the effect of increasing the strees in the system. The increases are modest however, and there is no evidence that this could be the predominant cause of the failure of plastic rising mains in the field. Similarly, there is no evidence that rotary opeating systems (which have no fixed stops) offer significant reductions in stress. Pigure B19 illustrates the likely relationship betwn fatigue life and ste variation for uPVC rising mains, for a range of stress variations typical of those measured in the CRL and IAD test programs. The beavier shading indicates the likely fatfigue life of good-qualit pipe; the lighter shadig rpr pipe of inferio quality. Str vwarletlon (MP,),_,_ ____ _ 10 4~~~~4 1 6 10 20 so Hue B19. Estiatd fatigue fife of uPVC rin main hvwm Reon spin d9 B6. Ongoing Research: Rising Main and Pump-Rod Comectors Following the reseach into the stresses in fte below-ground assemblies of handpumps described in sect BS, CRL has undeak further eh and development into the design of VLOM comnectors for plastic rising mains, sponsored jointly by ODA and GTZ, and connectors for pump rods, sponsored by ODA, UNDP, and the World BanL This resech is ongoing, and the information in this report therefore represent te interim reults Rising main connectors This is approject in several phas, intmded to develop easy-to-use couplings for plastic nrsing mains. The research has progressed to the point where a number of poteal connector concepts have been identfied, market research into suitable adheives ha been carried out, and a number of adhesives have been test. The next phase of the wor will conist of tie manufacte of protoype molded connector for tstg in the laborato and the field. The design objectives are summaized on the following page. The research hu been based on connectors sitable for DIN 8062 (BS3505) uPVC pipe, 63-mm outide diameter, 53-mm inside diameter, used as a handpump rising main at settig down to 50 meters. A limit of 85 mm was placed on the outside diameter of the connector to allow for a 6-mm secuity rope and provide some sak when fitted in a 100-mm borehole casing. Selecdon of deigns for tsng Initally, ten dei conceps were devised, from which six were selected for further evaluation: * a theaded connector, with female-threaded collars solvent-cemented to the pipe and a loose male-eaded couling * an entic coupliw , lying on the fiction generated between winfewrgng ecentrs, engaed and diseged by twisting * a split colla coupling, using collas solvent-cented to the pipe locked togeher by snap- ectenal split half-collars 70 UNDP-W$4 Dan Waw ad SanIt Pan. SwnMiy: Desig objectives for risn min connectors * Simpb to use, needing no tools to asemble or dimnle, and an obvis mechanism that canot be assembled incorcy * Should fail safe, by leakage ather than bpikage, so that faflure does not cause the ising m to fail down the boreole, but should not leak significantly until the point of failure is reached. * Suitable for manufictue in developing counties: designed for maximum local mateal and processin contt If plastics are used, they shold be iection molded or extruded. * Target cost about $5 per assembly ian medium-scale production * Should incorporate, or be capabe of inorpoating, a rising main cenralizer * Should be desgned to mini points of stess o tion, especially where notch-siive Materials such as uPVC ae used * Ieally the compoet ould be supplied factory-assembled to the pipe, avoiding on-ste assembly * Must be self-cleaning and have adequate flow characteristics * Should be desigped to outlast tie rsiog main pipe and be capable of witatadg typied storag antd hadin conditions sh as exposure to ultaviolet radiation and accidental abue * two versions of bayonet couplings, with lugs in one part engaging L-aed recesses in the second part, engaged and disengaged by tisting • a compression coupling relying on friction generated by compressing rubber rings onto the outside of the rising main pipe The relative merits of these designs are compared in Table B7. Models were made of these six designs to give a clearer idea of what each would be lie to make and use. The eccentric coupling and the two types of bayonet copling were rejected at this stage. The eccentric coupling would require very close teances and might therefore be susceptible to contmiati from partces of dirt or sand, and coud tempt mtenance staff to use pipe wrenches which would damage the rising main. The bayonet conntors would require complex molds, and it would be diffiult to make the couplings easy to oprte without makng them insecure in sevice. Three designs were therefore selected for further development, leading to prototype manufacture and testing. Progra Repo Ssl 71 Tbe 37: C _d o nector DeSis 1 2 3 4&5 6 Connector type: Threaded Eccentic Split Bayonet Compresion coupling coupig collar couplig couaplinag Easy to assemble - + 0 + - Can be fastened without tools - + + + Durable in unddiledhands - 0 + + Does not rely solely onfiction + . + + - Require little fiwrter deveopment + - 0 - 0 coaa sificial ldement + + + Key: + Posite attribute - Negaive attribute o Neutrl or Inde-e-minae • spl-coll couping • thede couping * compession colig Of tese, it was recognizd tbat the compresion joint had a number of po il drawbacks, but it also had the unique advantage of acceptng the bare end of the ising main pp It could therefre be rtofited in the field, for example, withot solvent cement or other adhesives. Howev fiuter i a reveald that it would not be posile to make a prctical compreionjint capable of wih i the tesile loads in a riin main, and the desg was therefore widrawnL 72 UNDPWWMidB Beak W.dSxkl WV= Adhesives for uPVC pipe Adhesives for uPVC are genally based on volatle solvents (typically Cyclohexanone or tetrahydrofuran) which attack the uPVC and form a weld when the solvent evapomtes. CRL was unable to obtain data on the shiear strength of solvent-welded joints from adhesive manuctues This inmation is crucial in the design of couplings, however, to determine the area of the bond which in tum detmines the dimensions of the coupling. CRL therefore carried out a seties of simple tests to detemine the shear strength of solvent-welded joints in uPVC. The resuts revealed a wide variation whch coud not readily be expWlained in terms of differences in the area of the joints or the conditions under which they were made. However, it was clear that joints left to cure for 72 hours were apprximately twice as strong as joints left to cure for 24 hours, and that simply incng the bonded area of the joint did not necessrily result in an increase in strength. CRL also surveyed 65 adhesive manu s to identify sources of adhesives which were not based on solvents. Only two were able to offer products which they considered would meet the requirements: the frs was a rubber-toughened epoxy and the second an acrylic adhesve. Both were two-part adhesives which have to be mixed immedialy prior to use. Joints made with these adhesives were tsted alongside solvent-welded joints, using values of force variation drawn from CRL's meas ts Of the stress in handpump rising mains (secion BS of this rport). Both the nonsolvent joints and one of the solvent-cemented joints performed well The seond solvent-cemented joint failed apidly, however, even though the same procedures were used to make all the joint Examinton of the failed jomt showed poor * tact between the pipe and the connector, with only a small portion of the total area bonded scurely. Since similar failures are known to occur in the field, CRL feels that this result may have much wider significance. The problem appears to be related to the formation of solid or "gelled" mateial in the adhesive, as a result of solvent evaportion, priao to the joint being brought together. Gelled adhesve cannot bond effectively with the uPVC, and gaps are therefore created within thejoint. Because the solvents are very volatile, and some time must elapse between the application of the adheive and bringing the joint together, a cetain amount of geltion is ievitable. The problem will be exacerbated by high ambient tempeas, and virtaly impossible to detect because there is no extera evidence of it once the joint has been made. Phgmm Rpoft Sai 73 Ii testing ofproiotype rsing main connecors Prototypes of the two remaining coector designs were produced and ubjected to an enduance test regime based on foc measured in CRL's tests on uPVC rig mains (secton BS). The target for complon of the test was 115,200 cycles. For each joint, components to be bonded to the pipes were made in uPVC and ote components in acetal, which was chosen for its combination of strength, fatigue resstance and negligible wat absoption. Spt collar The first sample failed rapidly (after 3,387 cycles) at the coupling solvent cemented joint betwe the socket and the pipe. Examinaon revealed that only 50 per cent of the available area had been bonded. A second sample lasted somewhat longer (9,829 cycles), but failed in a similar way. In this case, bonding haa ocurred over about 75 percent of the avaiable area The vaiance in these two joints prepared with car, under identica conditions highlights the inherent difficulties in achieving consistent solvent- welded joints. A fiter test sample was prepared, using ep adhesive to secure the collars, which required a small modificaion to the outer half-collars. Tbis lasted 90,870 cycles, when filure occurted in the pipe immediately above the epoxy jiont. There were no signs of failure in the acetal outer half-collars. Threaded Fllowing the epeience with the split collar coupling, the coupling threaded connector was attached to the pipe with epoxy adhesive. The first sample was tightened by hand and failed after only S0 cycles. It was remade using strap wrenches and completed the full 115,200 cyles. The test was continued to failre, wich occurred after 124,000 cycles due to shear separation in the epoxy adhesive. Examination revealed that castrhic failure of the pipe would have occurred shordy theter. There were no signs of filue in any of the coupling components. Fina designs for both join bave been drawn up with a view to the manufcture of molded proto for tesig in the laboatoy and at selected sites in the field. Ibese designs are ilusated in Figue B20. Note for the split collar coupling that o lip or saps are shw for secwing the outer half clars: it is intended that in a production veion of this coupling, the half coUars would have an integra reining feature. 74 UND?-WovI Ban* Wawe and Swanidg. ?.vw Split colar couplina: Threaded coupling: FP. MO0. Seected connco duu P *Vwn E.p. Swln 75 Pump-rod connectors This b an ongoing project to compare exit designs for puip-rod coOnectors which need no tools, and to develop improved designs. Existing connet designs that wre evalated ar: a Afridev hook-and-eye conctor, in mild sto d stlem ad * Indian forged connector (from Inalsa), in mild el and sinless staee * Afidev pin-and-pae connct, in carbon steel and snle std * Aquadev act connector (ftom Mono Pumps) These were compared by setting up five comectors of each type In a rod string. The teSt conss of lifting a weight equivalt to a 45-metr water column, loweig it to allow the sing to colapse slghtly. Testng was carrd out at 40 cles per minute, and the target for the ts was 10 million cycles. Each rod string wa connected to the rg using one of t Afridev hanga connecto. A sing d I thrae dmilar or four dissmilar break occurred. Results are illustrated in Fgure B21. A D _ 1 2 3 4 8 a 7 S 0 10 Millions of owes A - Atrldev hook-and-eye connotor; stainless st"l 8 - Afrldev hook-and-eye conneotor; mild stee O * Inalsa forged oonnector; stainless sto D - Inalsa forged connector; oarbon steel E a Atridev pin-nate connector; stainles ste F - Afrldev pin-end-plate connetor; carbon stee Q - Mono Pumps Aquadev aetal connector lgure B21. Enduance tests on rod connetors 76 rUND-WWd Ban Waser and Sa _sadoa PrqV= AMdev AU the avanfAd mild steel conneco iled in less tan 1.2 million hookand.e. cycleL Four of the stailess stee comecto filed within 4 million _onnecton cyceqs. Most broke at the joint between the hook and the pump rod, probaly due to stress concentratin induced by welding. A few broke withn the bent hook secton. CRL has recommended that this connectr be modified so that the hook is formed direcdy from the rod, ediminat the need to weld on the hooL 1Aw foed All the test samples completed the allotted 10 million cycles without cometoaw failure, and with raively litle wear. AM , plu.. Mos of the test sples completed the allotted 10 million cycles. Two ed-pkt of the sinle stel connectors filed, one at the joint between the plate conectoW nd the pump rod, the other at the joint between the pin and the plate. Aquadw Four of the smples filed within 2.4 million cycles, due to breakage of acea the aetl This connector bas subsequently bnen modified by Mono _ounKtP R PumPs Prgrmn Rw Sot 77 C. Handpump Research in the Field Field tials in 17 countries-nine in Africa, seven in Asia and one in South Ameima - were coordinaed by the UNDP-World Bank Handpumps Project. Over 2,500 insallaions rersnting about 70 different pump types were monitored. The field trials wer teported in detai i Commwdty Water Supply: The Handpwnp Opfon. SLOW, ~~~~~~~~~~so . s- 100w 9" 10 Fgure Cl. Global field tris Since 1986, acdve research on the Afridev, India Mark H, and Tarn handpumps has condnued in Kenya, India, and Bagadesh. ftOgram Rqwn Sato 7 Cl. Kenya - The Afridev Handpump The Afidev handpump was developed fom the Maldev concept, and was deigned from the st with ease of ainance as a top prionty. The inital field expiments with plasdc below-ground assemblies and plastic dry bearings were caried out in the Livulese Valley Project in Malawi, usn the Maldev pump, and this early exeiece was used in planning the resech into these topics described in Part B of this report Production of the Afiidev pump began in Kenya in 1985. The main problems encountered with the first pumps were errtic wear of the two-part handle bearings and interr'al damage to the rising main where the polyethylene pump-rod guides rubbed it. The eratic wear of the handle bearings was not easily explained. Some bearings wore slowly, suggestng potential lives of wei over one year, others had to be replaced in a few weels. A number of experiments were carried out to compare various potential bearing materials and to assess the significance of high temperature and humidity (see Part B). But the most likely causes of high wear rates seemed to be distortion caused by corosion inside the bearing housings, and the ingress of conrosion products and of wind- and rain-born dust, partieularly in the handle fulcrum bearings. Modification were introduced in 1987: • Stainles steel liners were fitted to the housingp for the handle and rod hanger beainps. * The pumphead cover was extended to prevent the ingress of extena dust. * A stainless steel plunger rod was fitted to stabilize the piston in the cylinder. * The polyethylene rod guides were replaced by nitrile rubber guides. These modifications were also incorporated in the Afiidev Specification published by SKAT in 1987. Field testing of modified pumps The pumps were tested in the Kwale Water and Sanitation Project (KWSP), with assistance from the Kenya Water f Health Organision (KWAHO). Kwale is a oasta area southwest of Mombasa, characteed by high temperatures and high humidity. Aggresive groundwater is common. Nieteen pump insIlations were monitored, with static water vels ranging from 13 meters to 45 meters and pump settings from 16 meters 0 WtVP-WoI Bo* Wa ad Swfadm Prgm to 50 mdet (igure C2). All the field tials were caried out whfle the pumps were providing water to tfie local comm s, and the monitoring procedures were threfo deged to obtain field data with minimum diupdon to normal use. The results as tiey stood at March 1990 are shonm in Table Cl. 40 so> Faguire C2. SWLs ad cylinder setting S The plastic handle bearings performed wrell, and in most cases lasted over one year. Fulcrum besrings tended to wear more apdly than rod banger bearngs: this wasepced since thie fulcrum berig carry generally higher loads and have to resist an side lod applied to te handle. The stainless steel bearin housI'w liners wee successful in eliminating corrosIon. The nitrile rubbae pump-nrd cetiers were a markd imovems t over the yearlier p et yp Wear was conen d on the centralher bar than the isin main. The uPVC riig mains the sample were up to four years old. There were no instances of rising mains falling under load, eve in ring mais tht had been badly worn interly by the polyethylene pump-rod centralizers. CylHndes performed well A few footvalve laks were encountered on early pumps, due to eessive flash on the valve bobbin moldings which was eiminated by impoving the molds. The stainless steel plung rod elimated instability in earlier cylid rsat allowed the piton body to contact the cylinder wall. Piston seas wore reatively slowly, as the results for volumetric efficiency confirm, but a few showed a undency to stetch or swel after several months of use, and on two pumps the seals roled off the piston while it was being exaed for maintenance. Wear on the valve bobbins was very low. nPoW Rqei Sr. 81 Table Cl: Afrldev HIdpumps In KWSP, Kenya Minimum Minimum Age of bearing eMfl- rising SWL^ Setftng Water lifeb ciency main (m) (m) pH (months) (%) (months) Mrima camp 23 30 7.2 lip 85 45 Kigombero 1 45 50 5.8 lop 89 39 Kigombero 2 31 39 5.7 21p 64 36 Kigombero 3 26 32 5.9 19* 82 33 Milanani 2 13 16 7.7 35' 96 35 Mwaebe 14 17 6.7 27* 94 51 Mrima Ndooni 1 18 26 6.4 24' 98 21 PeraniPrimary 20 32 7.0 238 96 26 Msambweni Pofice 16 23 7.1 27* 97 27 Kivuleni 2 21 37 5.8 1s5 96 40 KR3ulu 1 23 33 7.3 22* 96 32 Mivumoni Primary22 32 5.3 18' 98 33 NdengwaPrimary 26 39 5.3 23' 99 26 Mtitd 1 22 32 5.6 n.a. 98 39 Mowahande 3 22 29 6.8 280 98 29 Nguluku Mwalimu 26 38 6.8 19* 95 19 Nguluku Mwainzandi 41 50 5.8 198 81 19 Vwivwini 18 28 5.6 11' 94 33 Mkaa Nguluku 30 44 6.7 20W n.a. 22 a Static water level b F = fulcrum bearing wore out fist; H = banger bearing wore out first; B = both wore out at the same time; * - beaings did not need to be replaced. c Volumetric efficiency Corosdon of the glvanized mild stee pump rods was considerAble, in some cases after only six months, and tnded to be concentrated at the hook-and-eye coupings. In one saple ahook connector broke after being weakbned by coroon. Te hot-dip galvanizing on the pumpheads was very effective m restng corronion. ITe rubber ridIg main centrafzim, fitted to center the rising main in the borehole, showed a tendency to "migate along the pipe, in some cases allowing contact between the rising main and the borehole casing. 82 UJDP-WOdd B* Waer sad Swafdon Progr,m Maintenance Throughout the trial, pumps were progressively handed over to tfieir user communites, and by March 1990 all but he of the monitored pumps had been tnsfered. To follow up the monitoig process, however, a questionnaire was ntroduced to obtain relevant data when spare parts were purchased from stores. The uestionnaire covered the following points: * why spare parts were being purchased * which parts were required * the curent condition of the pump being repaired * whether an attempt had been made to repair the pump before purchasing spare parts * the extent to which the community was siadsfied with the pump It emerged that communites were generally satisfied with pumps, though there re complaints abounpleasant taste effects caused by corrosion of the mild steel pump rods. There were signs that rutine replacement of parts (planned at intrvals of one year) was not taking place. Rather, users were tending to carry out maintenance when problems became apparent: when the pump failed to produce water or when it stated squealkng, for example. In some cases, spare parts were being replaced unnaly because faults had been incorecdy diagnosed. It was also clear that ther was a need to decentiilize spare-parts distribution, as some users had to travel conideable distances to the exiWg distribution center. Recommendations For the pump: • Stailess steel pump rods should be used where the groundwar is aggressive. This will increa the cost of the Afiidev pump by around 40 percent, but the long- term costs will be lower and the water qualit will be improved; moreover, the improvement in the tast of the water will encourage people to use it in preference to traditional water sources. * Piston seas and valve bobbins wear out in a "fail-safe mode, but not handle bearings. There is evence that users cannot be relied on to change the bearings yearly. A visual or audible waming built into the bearings would alert users that the bearings are almost wom out Pmgiw -n &Cwo 83 * A method of locating the iin main centralizers on the pipe would be beneficial. h shiould be cosered in the ongoLag resrh into detachable rising main connectos (see sectiB6, page --). For mainance: a Futer information b needed to comfim early indications that annual maintnce is not peformed by users. If it is not, a strategy to ensure that wearing parts will be replaced before the basic strucure of the pump is damaged should be developed and icporated into Waining procedures. Other topics Othr topics included in the Kenya field tdals wee: * The development of low cost concrete pedstals designed to avoid contamination of the aquifer by the seepage of water aroumd the base of the pump. * An analysis of user prences regading apron design and water point facilites such as basins for washing clodeL * Field tet of an Insllat maml and a schodued int card for the Aftidev pump. C2. India -The ldia Mark I and Mark H (VLOM) HIandpumps The dia Mak has an admiable record of community serwice dating back to the late 1970s. By 1983, more than 800,000 India Mark H pumps had been Insd in India; in 1989 the total reached 1.3 milLion pumps serving an esdmated 360 million persom in rural and peiurban areas. The basic design predates the VLOM concept, however, and the Ini Mark IH rlies heavily on a centalized maintenance me., In 1983, the UNDP-Wodd Bank Hnpumps Proect set up a field teng project in C ,mbatoe, in southem India, workn with the Naional Drnkdng Water 84 fJN-Wod Bak Wae ad S_kle h,m8 nA Mision of the Government of ndia, the State Government of Tamil Nadu, UNICEP, and Rchardson & Crddas (1972) Ltd. The objectives were: * to ma imprvemts to the lndia Mark II pump, ineae the mean tme between failures (MTBF), and simplify man; d * to document consumptian of spare parts and i tions for maintenance and Coimbatorei b by revely deep water tables and hugh use of handpumps. Nearly 80 handpumps were included in the project, of which 48 wene standrd India Mark II pumps and 32 were fitted with expermeal components. The range of static water leves recorded when the pumps we illed between 1984 and 1986 (fom 4.6 meters to 38 meters) s illusted in Fue C3, which also shows the cylinder sefflngs (anging from 21 meter to 49 meters). The populations served by indivdual pumps varied from 100 to 560. M a t sted an the pumps included • different types of pisto seals * pump rods with difern materiah and types of cotinp • PVC rising mains w variou types of coectors • PVC cyinders * the additon of a bottom intak pipe * h incoan of asand tAP * a c handle beaing * 50-mm, brss-lined, cast iron cylinde * pump-rod centrlzer * special tools * different platfm design In addition, a VLOM dervative of the Mark II, the India Mark MI, was developed. This model uses an open-topped cylinder and 2.5-inch galvanized pipe fbr the rig main, to enable the piston to be withdrawn for maintance without extctig the fing main. An intemediate plate was also added to the pump head to give easier acces to the below-ground assembly. Oher modi ons npd in the India Mark Im include: * a mechanism to lift the valve gide when the piston is coeted t the foovve for exraction, dumping the colum of water when the foovalve is raised a few millimets * a conical receiver and o-ring seal for the footvalve PWm Rpet Sea 85 Depth tn) 20 H:l -WL 10 ~~~~~~~~~~~~~~~~~~~~~11Se?tting Instelletlonf In order of 8WL gue C3. SWI ad sefting depths - nitrile mbber piston sea her than leather • a two-pieoe upper valve, to elimina failures due to disconnecdon of the threaded joint, and a two-piece foot valve * a euaro dban round beag ng housin on the handle, to incrase igidity and mnim distordon during welding and increase the sevice life of the bearings a deeper water tank, to eliminate splashing during fast pumping * a reducdon in the overall pumpstand height, to reduce banging of the handle on its lower stop Several of these difictions were also tsted on India Mark H pumps. Comparing the India Mark I and Mark m (VLOM) pumps On average, repais to India Mark m pumps tookjust one third of the ime needed to carry out similar repairs to Mark HI pumps. The differences are illustrated in Figur C4 and CS. The Mark H pump requires a minimum of four semisdlled workers with a mobile van and special tools to repaxr the below-ground components. By contast, a mech, carying all tie necessary tools on a motorbike, can extract the piston and footvalve of tie Mark m with the asistance of the pump caretaker or another member of the us community. A mobile team is required only for the relatively infreqt repLacement of the rising main or cylinder body. The work of the mobile team was reduced by a factor of 10, and most repair work on the Mark m was shared between mechanics equipped with two-wheeler and local caretaks, as illustrae in Figure CS. 86 UNDP-Wd Baa* Warn and SaaiMtl P'gr= Ii~~~~~~~~f 2J 10 8 ^ 4 r ~- I:Il'2Z q I iL8 13 I X 1 ~Il XM1 |~~ ~ l I, III 1x [,. Pu~~~~~~. PuWl replaed. The larger-diameter ising main of the Mark m pump reduced abrason between the pump rods and the inide of the rising main, helping to keep the galvanizing intact and increaing the lfWe of both fte rising main and tie pump rods. Aother factor was t the rising main was less prone to damage from wrenches and clamps, which can cause corrosion, since it did not have to be removed to service the piston or footvalve. F-'.~~ Wk r r S I-s loc " w 'S plow soft _boM sdehl I LPL IuI i p I BIgu C6. Parb Fepld per yewr, I1di MA II and Mw* m PMPS Aotha 1 of this resh wa to compa t 6 of vaious types of piston seal, pariculaly leater versus nitrile rubber seals. Of fte leaher seals tesed, vegetble-tanned lastd longer ftan senuchrome- or chrome-tanned ses, but ntile rubber se out-performed em all. The aveze life expecny for nitrile seals was in excess of 2.5 years, assuming 25 m depth and seven hours of use per day. With leathr piston seals, it bas been necessary to rlace the brass cylinder liners at intenals of four or five years, due to abrasion of the liner from sand prticles embedded in the leather. The nite eals showed a markedly lower tendency to embed sand particles, however, which is conent with results obtained both in other field resch proects and bbora tests. It is therefore esimated ta,with nitile seals, the bra liner can be expected to last at least seven to eight years. The average Life of handle bearings was 2.8 years. 88 UDP-WorI Be Waw wd SwaIwam Prigaw Conclusions and recommendations Minor modifications to existing ndia Mark II handpumps oould improve reliability substandally. The costs of modific2 ion would be fully offset by reduced maintenance costs in less than two years. The design modifications which have been shown to be successful in the Coimbatore project have therefore been incorporated in Indian Standard IS:9301. An Indian Standard for the Mark m handpump has also been published, IS:13056, 1991. The Mark m design dramaticaly reduces the dependence on a mobile team for the majority of repairs. A mechnic with a two-wheeler can carry out 90 percent of repair with as istance from a user or local caretaker. The additional capital cost of the Mark m over the Markl will be more than offset by the lower maintenance costs in less than three years. It is recommended that a National Standard be prepared for the Mark mI handpump. To make possible greater self-reliance in user communities, research should concentrate on ways to simplify manance requiements even fiuther and ways to build capability at village level to carry out maintenance and repais. C3. Bangladesh - The Tara Direct-Action Pump The Tua is a smple, low-lift, direct-acti pump designed for both m.anufcture and village-level maintenance within the vy Limited resources available in Bangladesh. The development of the Tar is fuly desuibed in The Tara Handpwnp: he Birth of a Star. Work has included both field ialsand laboraty tests, and by the end of 1988 it was esimated that 5,000 Tamn handpumrps bad been installed toughout Bangladesh. The Bangladesh government's Department of Public Health Engineering (DPHE plans to install up to 60,000 Taras by the end of 1993. A sample of 148 Tara pumps have been monitored since their insalladon in the mid-1980s. In addition, UNICEF, DPHE, and the UNDP-World Bank Handpumps Project have extended the monitoring to include up to 2,000 additional pumps. The most significant findings of this research and development effort are smmnarized below. Pr*giw Repon Serw 89 * A cheap, effeive, and rdiable dir-cton pump can be produced using standard uPVC pipes for both he rsn main and pup rods. Intemal pump-rod couplings minimie wear between the pump rods and the inside of the rising main since the asembled pump rod has a smooth swfue and a constant dimneter. * Me Tam uses stndard uPVC pipe as the cylinder, withcut mny special processing. Howev, cylnder pipes are seleted from normal production for good intea finish and minimum variation of wal thins. The life of a cylnder in the field has been estmated conseraively at 12 years under typical Bangdesh conditions. * Various piton seals bav been trid, ncludig some synthetic matials. For a period, a itie eal was used, but tie dght tolerances needed for efficient opeation were with the pipe ailable locally. Leatier has the abilit to onf to a wider rne of cylinder bore sizes, and subsequent attention has theefore centeted an opimizing the dimensi and improving quality control of leath seL a Simple rubber flap valve, for which spe part can be cut from old inner tubes, have pwved sucsful Nylon has replaced aluminum as the material for the pisto. Th foot valve b made from higb-density polyetiylene using rudimentary injecd moli equipment. The Tam is a true VLOM pup which enjoys very high user acceptan. It can be llod at low cost and swtained at a cost of about $0.10 per user per year. N JfDP-Wed Ba* Waw and lnWlen PmV D. Laboratory Tests he Comer Rach Laboratory (CRL), bad in Harpenden, UK, bas been testing bandpuimps ince 1977. These laboat e wer never intended to be a substitute for tesdng in the field, but rther to pvide reliable and comparadive information under controWlled labomtory conditions that could be used in conjunction with field data The results of tests carried out up to and including 1985 have been published in previous World Bank Technical Papers and Applied Reearch and Technology Notes, and were incorpoated in Commwdty Water Sppy: The Hanpwn* p Option. This report summ the laboratoy's findings for eleven handpumps tested between 1986 and 1989: • Wavin: a direct-action pump from the Netherlands designed to work at depths of 10 meters to 15 meters * India-Mali: a verson from Mali of the well-established India Mark Il deep-well iandpump * Bestobel Micro: a deep-well handpump from Zambia * Ab1-ASM: a hydraulic pump combining a pumpstand manufactured in Cote D'Ivoire and a pumping element made in France; no longer in production. It has been included because of continuing interest in pumps using this principle of oeation. • Punpenboese: a derivative of the India Mark II made in Germany • Kuebel: a cable-operated handpump from Denmark designed for depths to 25 meters * Aquamont: a deep-well handpump from the UK * Atlas Copeo 111: a deep-well pump employing a stetching rubber hose as the pumping element, from Sweden * Atlas Copco 122: a deep-well handpump combining the pumpstand of the Atlas Copco 111 with a conventional piston and cylinder P&am Ripn &u*w 91 * Aquadev: the first pump based on the Afdidev concept in full-scale production, made by Mono pumps in tfie UK * Afrldev: a deep-well handpump designed with particular stres on ease of maintenance and repair, from Kenya All pumps were tesed in accordance with procedures developed by CRL in consultation with the UNDP-World Bank Handpumps Project and others. Test procedures are summarized in the Appendix. band fiat hexagon lfiug pipe pipe tools spanners keys tackle clamps wrenches saw threading sealants/ lubricants skUsld labourers tools adheasives worker Key to symbold used for installation, maienance, and repair 92 UNDP-World Bok Waw and Sakadon P)ogun Dl. Wavin handpump The Wavin pump used stdard uPVC pipe for the rising main and as a lightweight pump rod. Tbe pumpstand and handle were fabricatd ftom steel =sectons, galvanized, wi a polyethylene bush for the emergig pump rod. The piston ad foowtvve were plsic, with rubber valve balls. e piston seal was ramie cord. Spare cord and valve balls were supplied atched to the dip tube, the intntion being that in the event of failure spar would be readily availale when the * 4 pump was removed from the borehole. Te piston acted directy on the bore of the rg main: there was no separate cylinder. Oa 1 | l The sections of the pump 15 '~rod and rising main were joined using threaded ______________________________________ connectors epoxy- cemented to the pipe. Ihe test sample was kn from al initi production batch of pumps. Results obtned from initial inspection,.design assessment and perormance tests were used to refine the deign before embarking on the endurance test The Wavin is a direct-acton pump intended to operate at water depths between 10 meter and 15 meters. It was desgned to be easy to manufacture, operate, and maintin. It wa tested in 1986; the CRL refce is A9113. nvgrm Reupo Soa 93 Mmufachuer Wavin Oveeas b.v. P 0 Box 158, 7700 AD Dedemsvaart, Neterlands Price (October 1990) Complete for instalaton at 25 m depth (te mamum deign depth), in lots of 50: $530 each Weights Pumpstand: 7.3 kg Rising main, per meter: 0.6 kg CyLinder assembly: 1.4 kg Pump rod, per meer: 0.2 kg Dimensions Maximum handle height: 1,365 mm Nominal cylinder bore: 45.2 mm Minimum handle height: 585 mm Maximum stroke: 785 mm Spout height: 325 mm Maximum diameter of below- ground assembly: 59 mm Abov-ground assembly.: Steel fabrication, basic machining, galvnizing Below-gound assembly: uPVC ectrusion and fabrication, basic machining, rubber molding Good quality control is enta to make satactory joints for both the rising main and the pump rod, but in general the pump is potentially suitable for manufactre in developing counties with established pipe ersion capabilities. h11talton and major repu_ The lightness and simplicity of the pump make it easy to istall or extact, with no need for special tools or Ti den lfting tackle. Some care was needed to avoid damage to the uPVC pipe and joints. R~~~~~~~~ Routine mitnne_______ Very straightford. Maintece of the pumpstand is limited to replacement of the top bush, an easy task. J The piston can be extacted trough the nsing main by simply removing the top bush in the pumpsand. Intallation and manee nf AD installon and maintenance manual was provided, in nglish. It was well illustad and included details of the constuctio of a suitable welhed apron and com ive instins for insftalltin and maintenance of all below-ground pars. ITe Latter included 94 UNDP-World DBank War ad Samwon Amomm vuious means of renewing the poim of dh ising main used as the cynder, to aopensate for wear. Resistance to con Generally good; the spout could be wasily modified to pret fad c ntamii, if requird. Reistance to abuse The handle may be suweptible to impact s in t tends to dse to its uppmost position as a result of the buoyancy of the pump rods. Pup perorance_____________ The rame cord pistn seal tended to be sff, but afte bedding-in for 24 hour, the worc WN_N '0"0 required to achieve a given outflow was ouw reducedby approximately 30 percet. In the course of the endurance tes, a number o _ _ _ - of modifications were made to the pump in consultaton with the m. IThe * -_ I 7 / 7 results of the tes of pump peformnce after Z enduanc e terefore not stctly compble to r- f- - - with those for the new pump. Since the pump after enduranoe is mor e of the 15 fnal design, the results ater endan are ilusated. The pump produd wat at a r o - of 18 liters per minute for a wo nput of 75 -' was, from a depth of 15 metes_. he at efficiency was distidncy better at lower a so f 1| u 1 Vs0 opraing speeds, the best result beig m 70 percent for 20 cycles per minute at 15 meten depth. Endurance The pump did not break durng t enduranc b est Howevm , the pump seized svel times when paries of uPVC lodged bee tepisto sea and the cylinder. In the fis failure the uPVC debs appeared to have boe caused by shp edges on the pison body rubbin against the cylinder bor In with f the the plow body was modified, and the problem with is appart can did not recur. s 5 o '5 Test record (fures set in Itagc type) Hours: 754 2,000 3,772 4,014 I' I 1 1 1 Start Pmp very st0f Inspection Pison skzed Piston siezed to opemte; ex- In cylinder in cylider cessive uPVC I debhs in outlow Final inspection During the second phase of the test, when sand and Kieselguhr had been added to the water, the pump seized a second time. Once again, particles of uPVC debris were packed around the piston seal. Howeer, in this case the debris had not resulted from contact between the piston body and the cylinder, but seemed to bave been generatd by the pump rods rubbing ainst the rising main. In consultaion with the manuf, a filter was added. The pump seized again, however, and the test was stopped at this point. The final insection confirmed that the debis had originated from contact between the PumP rod and rising main in the first 2 meters or so of below-ground assembly beneath the pumpstand. Abuse tests The pumpstand body was undamaged by side impacts. Because of the tendency of the handle to rise whle unattded, a side impact test was caried out on the protrdng handle. The bandle was slightly bent by lateal imnpacts but could be straihtened by hand and continued to be serviceable. The handle was tested by repeatedly bangng agaist its lower stop, from which it emerged undamaged. A imilar test of the upper Limit of handle movemet was not carried out, since it would be most unlikely that the handle would ever strike the upper stop in actual use. Verdct The Wavin pump incorporaed some intesting innovations, pmrcpally the use of epoxy cement forjoints in uPVC pipes and of ramii: =ord as a pisn seaL The pump completed the endurance test with no failures in the rods or rising main. The ramie cord appeared to be a durable sel, but it was uceptible to a in by debris, resulting in sezr of the piston. 96 UNDP-Wodd Buk Waer d Sauorn PMn, JI If the pump's duility can be improved, It has tie potential to be eay to opeate, insll, maimn, and rpair, and thefore to stf the i -quirements for village level operon and maintenace i devdoping coumties. Some developing countdes would also have the capabilit of m ,anuf ng the pump. Further lIfomnatlon A number of modificahios bave been introduced following the CR1 tests: Piston and footvalve. These are stll based on similar components, but are no longer identical. The footvalve has a rubber seaing ring, in contrast to the rameh cord rings on the piston. Piump-rod guides. The collars on the pump-rod connectors have been modified to act as 8sp or guides between the pump rod and the inside of the rising main. These are molded in polyacetal and are fitted at every pump-rod couplin at 2.5-meter intvals. Handle. The polyethylene buffer block beneath the handle tee-bar has been replaced by a steel plate. The conntr by which the handle is attached to the pump rod is now molded in polyacetal rather than PVC. In addition to these mo Ions, an altem ative pumpstand, made of cast iron and galvanied, is now available. Pro8ft= - Ser. 97 D2. India-Mali Handpump Construction The India-Mali followed closely the established India Mark II design. The pumpstand was fabricated from seel, with a chain and quadrant linldng the handle to the top of the pump rod. The handle fulcrum bearings were proprietary bal races. The pump was supplied with cylinders of 68 mm, 80 mm and 105 mm diameter, to suit both the depths and the requred rates of delvery for different isl.:ulations. The 105-mm cylnder was intended for use at a maximum depth of 20 meters. Cylinders of all thuee sizes were simila in construcdon, with branes of stinless steel pipe and cast iron end caps. The pistons and footvalves were castg gunmetal, and the pistons were fitted with leaher cup washers. The pump was desgned to use galvanied stel rng maun of 1.25-inch nomin bore. The size of the pump rod should be chosen to suit the depth at which the pump i to be used. The India-Mai is a verson of the familiar India Mk II deep-wel handpump, made in Mali and intended for use at depths between 7 and 45 meters. The pump was tested in 1986; the CRL reference is A9123. 98 WJDP-Wrd Ba* Wawed and SakW= Pop B.P. 68, Sikso, Mali Price (November 1990) Complete for installation at 30 metes depth, in lots of SO: price per pump, $685 with galvanized pump rods and rsn main; $1,385 with saless stel pump rods and ring main. Weghs Pumpstand assembly: 41.7 kg Risiag main per meter. 2.9 kg Cylinder assembly: 7.3 kg (not suid) Maxdmum handle height 1,117 mm Cylinder bore: 68 80 105 mm Minimum handle height 420mm Maimum stroke: 100 100 100 mm Handle length: 1,055 mm Volume per cycei 3$3 503 866 ml Angular movement: 42 deg Maximum diameter Handle ratio: 7.7: 1 of below und Spout height: 445 mm assembly: 100 110 115 mm Above-ound assemWy: Stee gabrication, geal machkting Belowgund assembly: Iron and gunmetl fudry, gene mahinng, galvanizing, leatdr cai Tne pump is suitable for manufe in many devloping countries, although those without established foundry fclties would have to find another way to produce the piston, foowalve, and cylinder end caps Rigorous quality cotr is sential, particularly to ensure tat the handle bearn housings e accurately machined and propely aligned, and to provide an adequate itenal finish on te cylnder barre. nsaaon and a3jor repa* Liftig take wilbe essential unle plastc rhe t_ galvamzed steel isi main can be used. In other Wi n respects both inllation and extaction should be 1 straightforard. Routne m A single 19-mm spannr wi fit all the fseing in the oE punpstand. Replacing the handle fulcrum bearings requires a high degree of sdll and undersandI to __ ensure that the new beings are not damaged. - R se, 99 Instalatlon and ainaon The test amples we supplid with no normaion on ista m , orrepair. Without this inormation, inllation and maintenance In the field could be cried out only by weI-tained, epenced pesonneL Resta to coo Cood. n would be difult to puish sck and stones into the pout. The desgn of the pumpstand allw for the well casing to prtrude above ground level, ensuring that the well will not be con by surfce water. Resitance to abuse Good. The pumpstand body was generally robust Howevr, the short handle stroke may mean that users will be more lily to bang the hande on its stops at the ends of the stroke, which may damae the pump in the longer term. Pinp perormance; The pump was tested with each of the cylinder Wib ea. cyminor sizes supplied. The res iustrated (right) are for the 63-mm cylinder used for the ,/ endurance test I II' 16 With the 68-mm cylinder, the pump was 1// capable of povidng wate at ately 9 16 iters per minute from a depth of 45 mees, fi i /_ / for an input of 100 was The eo ficdency 4 varied fom 45 peroent at 7 meters to approximaey 68 percent at 45 meters depth. Similar results were obtaned afer the endurance test. Ihe rate of delvery was 8 maintained, and in most cases the requred ________ work inputs were dightly lower, nggsting a so 3o 2-O smal reducto in fiion. - k hOpt cw Teakage Insignificant for the 68-mm cylinder, at all depths tested. For tie 80-mm and 105-mm cylindrs, some leakge was observed around grub scrws in the lower cylinder end cas, though there was no leakage from the footvalves hmselves. 100 UDP-Wod Ban Wa r ad Sabden iopws EndUrance Tet ncord Hours: 2,081 4,123 I . I Start Inspcton Fna inspec- and volume tion and full checke test The pump compleed the endurance test with no failures. In the fin inpection, a good deal of wear was evide in the pumpstand. he handle beari were in good condition, but the rod guide was badly wozn, and there was wear of the chain and the pumpstand body wber the handle rubbed on it as a reslt of ms l of the pivot holes during manufcue Below ground, there was a good deal of corrsion, particulary at the joint between the pisto and the pump rod. he cylinder bore was poishd but only lightly scored, and the cup ses were found to be still serviceab. The footvave seal was worn but sdll serviceable. Abmn tests Th pump was unamaged by sde ixpact on fthe body and by latal Impacts on the handle, and no damage resuled from repeady banging the handle agait its stops. A tong, genwously pxoportned desige , of which this derivative manu edin Mali proved to be surdy and reiale. However, its poental for appication is limited by serious drawbac I maut u and mai Manu e req latively well- developed industrial and enneing skMatenance also demands a high degree of skill, and lifing tackle is reired unl a plac ising main can be used. The requirements for both manufaing and maintenance ould be eased by appriate modifications. P~oo ie Sere 101 D23. Bestobell Micro Handpump The pumpstand and t-shaped handle were fabricated from steel. The handle bearings were proprietary open-ball /mes. A single spanner fits all the fastenings in thepupatand. t The manufacturen envis- asged that the rising m and pump rods would _~ * _nonnally be purchased locally, and these were not supplied. The pump was deSigned t use uPVC ris- that pmain, and provided L _ w | s~~~~~~~~~~~~~ht pip of suficet dia I nmete (at least 2 inches) is _ w used, the piston can be _ widrawn for mainten- ance or rpair without removing the isnmai from the well. HovzYer, __ 9 it is sdll necessary to L_______I extract the entire below-ground assembly to gain acce to the foot- valve. The pump was deigned for 12-mm se rods with threaded connectors. Ihe cylinder is a leng of heavy-duty, 2-inch uPVC pipe. The piston and footave are also uPVC. The initial piston sea w6e leather, but fte were rplaced by a molded rubber s during the teL The Besobel Micros a deep-wel force pump degned for a maimum depth of 60 mees. The test samples were nufu In Zambia, and the pump was tested in 1986; the CRL rerence is A9170. 102 VWP-Wodd Bak Waw ad Sanion PbiSrm Manufacturer BestobeUl Zambia I f'-ted 875 Zambia Road, P 0 Box 230003, Ndola, Zambia Price (November 1990) Complete for Installation at 30 m depth, in lots of 50: $36S per pump. Weights Pumpstand assembly: 20.8 kg Cylinder assembly: 3.8 kg Maximum handle 4ight: 1,350 mm Nominal cylnder bore: 48 mm Minimum handle height: 550 mm Maximum stk: 152 mwr Handle length: 520 mm Nominal volume per cycle: 275 ml Angular movement: AOW deg Maximum diameter of below- Eandle ratio: 6: 1 ground components: 75 mm Spout height: 470 mm Above-ground assembly: Steel fabriato, genera machining Blow-gound assembly: Cekeral machining, cewcutting, ubber molding Manufacture of the pumpstand could be underaen in many developing counties Below ground, the piston and footvalve could be injection molded or exuded for volume production. The cylinder assembly is potentially sitable for manufacture in some developing countries, though care will be needed in cutting the screw theads in the cylinder barel hntaaon and mjor repaim Lifting tackle would not be required if uPVC n Tl 1 main were used. f rsing main of at least 2-inch diameter is used, the piston can be withdrawn for maitenncewithout e=tng thie sigmain. Routine mhenance Only one spanner is required for all the n pteni Y in the pumpstand. Care is required in replacing handle bearin. *~vw Rapi QS..10 intalon ad maitnac Information Instuctions In English wae supplied with the pumps. They were comprehensive and useful, thougb a presentation relying more on Mlustaons than text would be an advantage. Redstane to natio Poor. Care is neoded to provide an adequate seal at the base to ensure that the wel is not contamnint by surtace water. It would be easy to push sdeks and stones into the spout. Resistance to abuse Although the pump was, in general, fairly robust, the handle fulcrum and the pumpstand baseplate are susceptible to damage from prsstent abuse or severe impacts. Pump performnce The pump was sted with both the orginal sat tCo leate seals and with a molded rubber seal o fitted midway through the endurance test when * I - - the original piston filied (see Endurance, below). Substantially better resuts were 14 obained from the rbber sed, and thee are shown in the diagm. When new, the pump - - V provided water at apy 9.5 liters per minute from a depth of 45 meters, for an input /jI of 100 watts. The efiency varied from around 45 pecent at 7 mete to over 70 pecet at 45 meters depth 8 . -. _m At the end of the endurance test, the 4 - M performance of the pump at 7 and 25 meters .. depth had detriorated sligy but had o 90 40 oo eo Io t2 140 *o improved at 45 meters depth, with a tpicl Wok Input (W) efficiency of 75 percent. Leakage buignificant at all depths tested. Endurce (failums are shown in italhs) (wordgraf) No failures vee appaent in the first 2,000 hour stage of the test, using clean water. Howeer, when the cylinder was dismanted for ipecion, the piston body was found to have broken between the two le r seals. The piston was repaced by a piston of the 104 WNP-Word Bank Waar a7d 8wioa PgV same design. The pump failed again after 47 hours of tie secotd phase of the test, when sand and Kweslgur were added to the water. Sand embedded in the leather seals caused severe wear to the cylinder bore. 'r,t record (failures set In Italc type) Lo=: 2,138 2,185 2,425 4,257 I Statt Inspection and I Beadpgs colapsed Fina inpection and volume flow check at connection of pump performanmce Piston bnken and I handles to top of test one seal msing pamp rd I Severe wear of cylinder andfootvave lekdng due to connatin by Uand new piston and Soda footvalve ftted After discussion with the manufaurer, a new piston was made using a rubber seal provided by Bestobell, and the footvalve was replaced with a propdietary 'Socla" valve. Tbis new piston and footvalve completed the test and were in good working order at final inspectioL Considerble wear was observed in the inspecn after 2,000 hours in t diecast zinc alloy beaings at the connecon between the handle and the top of the pump rod. After 2,425 hours, the bearings collapsed. The second test sample was supplied with bauL races at this point. They were fitted for the remainder of the test and were in good condition at final inspection. Abuse tests The hndle was not damaged by side impacts, but a heavy impact on the pumpstand body produced distortion of the baseplate and a crack in one of the welds. When the handle was repeatedly banged agunst its stps, the fulcrum bracket ftured at the weld to the body of the pumpstand after 32,666 cycles. Verdict A design with good potential for manufacture, and repair in developing counties, but it rqires cderale additonal development before it could be considered for large-scale application. Prq8vw Rqiv Se5f 105 Addiionl ifo Tbe ma ur has subs quentiy s gtheed th pumpstand, and the baseplate now made fom a dsili piece of stedl 106 WJDP-Wor Bek Waw and Se&*aIe hw' D4. ABI-ASM Iadpp Cosrion .he ABI-ASM is a hybrid handpump workdng on ~~~~~~hydrad pitinciples. Two polyethylene hoses are the ol.y connections beween the pumpstand and 'cyinder assemblies. The pumpstand is part cast iron and part bricated steeL The main handle bearings are molded nylon, with a proprietary spherical plain bearing at the connection between the bandli and the piston rod. The primary cylinder iside the pumpstand is drawn brass tbbe brazed to a cat gunmetal baseplate; the piston is also gunmetl, fitted with three leather zing seals. By _- _ _ _ _ ___________________-__________ lifing the hadle fuIly, the primary cylinder is automaticaly replenied from the delivery water. Below ground, the moto of the primary piston s tansmitted via a polyethylene command hose to a flexible rubber tube encased in a stnles steel vessel. ITe alterang dilaton and contraction of the rubber membrane pumps water to the surfce via a polyethylene Te ABI-ASM consists of an aboveound assembly manued in Ce d'Ivoire and a cylinder assembly mIn Frane An a ver of the pump was tested in 1983/84, and the results pulished in UNDP Preoct Managemt Report No. 3 (World Bank Technical Paper No. 19). A number of m fat to the pump were implemented P* V% Rqwn Sato 107 subsequently, both above and below ground, and this second test was carried out in 1986; the CRL reerence is A9127. Production of the pump ceased while the laboratory test was under way, and the tstig was stopped at that point. CRL's report does not therefore contain results for the user trial, abuse tests, or some parts of the engineering asessment. Manufacturer Groupement Abidjan Industrie Z.I. de Vridi, 01 BP 343 Abidjan 01, CMte d'Ivoire Weighs Pumpstand assembly: 62.5 kg Cylinder assembly: 9.4 kg Maximum bandle height: 1,075 mm Primary cylinder bore: 60 mm Minimum handle height 155 mm Maximum stroke: 120 mm Handle lgthh: 750 mm Nomira volume per cycle: 340 ml Angular movement: 80 deg Maximum diameter of below- Handle ratio: 8.7: 1 ground assembly: 91 mm Spout height: 390 mm Installation,n maitnance,and repair were not assessed. Insaiklaio iand m intenance Ifo Vnatlon A manual, handwritten in French, covered both inshlaion and mainteonane of the pump. t was clear and neat though a presentation relying more on illustatons than on tet would be more appropriae for use in developing countdes. Resistance to contamintonand abuse were not assssed. is UNP-Wodd ak Wa ad Saniation Pram Np pertonance The pump required relatvely hig levs of Abi-SM hW941M work Input to acieve even modest outflows, at all deps. Por a depth of 45 meters, an input 1 of 100 watts produced a flow of less than S litersper minute Efficiencyvaried from less 14 -- than 10 percent at 7 meters to around , I_/ 32 percent at 45 meersdpth. 12 - The cause of the high operating effort was 1 thought to be eacesive friction in the primary 8- piton. Experience with otier pumps of this type suggests that this friction would decrese a in as the pump bedded in' with use. 45m _ 2 0 100 200 S00 400 Work biPUt (W) Endurace (faur set In Itcs) Test record EHouw: 139 338 481 984 1s633 2,007 I I I I I I Start Handl Pd,nwy Handle Piston PIon Incn heaigs pison eangs hamger hanger and volume worn o seal womn ou beaing ot flow check worn out joumals had seared I polshed worn; off pamping I iled dalent Pmary split piston seal worn out The handle jowuals were suppLed in a poor conditim. They were pitted with rus and covered with a patchy coating of paint, and their poor surfce led to prnmatmue failure. Afer lei and polishing in the workshop, the second set of bearings lasted substantaly longer, ahough wear rtes were still high. Prguw Repea3e 109 Premature failure of the piston seals was caused by a sharp edge on the replenishing hole in the side of the primary cylinder. The piston banger bearing wore apidly and had to be replaced. Tbis was an I at application for a plain spherical bearing because of the high loads being applied while the bearing was rotating. The spigot canying the piton was poorly fitted in the hande, witfi clearance of approximatly 0.1 mm. The forces were therefore takn by the end weld. The weld was of poor quality and not to the ful depth. After 2,000 hours, the rate of outflow from the pump was greatly reduced: the pumping element was found to be sfit. The endurance test was disconinued at this stage. Verdit Interest persits in pumps wodring on this pinciple because the connection between the above- and below-groud parts i simply a pair of flexible hoses, which makes the pumps pottally very easy to install and maintain. However, the pumping element is a spciaized item that would have to be imported, and as ths example showed, sound desip and good quality control will be essential if a sitable pumpstand is to be manufactured in a developing countty. 110 UAYDP-Wrd Bak WaiW d S=Uae hqiw D5. Pumpenboese PB-Mk It Handpump __________________________ Constructon The pb-Mk I reained al the essential feames of the - . 1 - India Mark in hg above- / 7 _ qml % z ~~~~ground parts, including ffi />oX~~~~~~~~~glie sbm pumpstand > I I 4r _ ~~~he pvwty ball-race - > _ | (ahandle bears, and tfe u _ , -j u chain and quadat Te cylinder was not of the type normay fotmd on the India . -. _ _ _ w | Mk II, however. The sam- . $ . ple bad a sinle sel cylinder bafrel of nominl intemal diameter 63 mm, with end caps also mad of s 2 W , ~~stabiless steel, seded to the ^ _ ~~~~bane by rubber o-nngs and beld in pilac by tawo sWa- ~~~~~~~I= V" stas Th pst e ~~~~~was altso stain] slee4, withi 8 rubber valvesnd . minog mamn and pump rods wwer available in either -__.____ galvanized or stainless steel. At dme of testing, manufac- tur were also worbng on a polyetylene g main. The P em pb-Mk I1 deep-wel andpump is a derivative of the well-established India Mark II design, and is intded for use at depths between 15 meters and 50 metera The test samles were made in Germany, and the pump was tested in 1987; the CRL refnc is A9101. Pr'eiwsr Rq.ei S.f. 11I n~~~~~~~~~~~ Price (October 1990) Complete for insta on at meters, in lots of 50: with galvanized rods and rising main, $759; with stainless steel rods and rising main, $1,390. Manufacturer Pumpenboese KG Raiffeisen Str 2, D-3006 Burgwedel 1, PO Box 1250 Germany Weights Pumpstand assembly: 48.2 kg Rising main, per meter: 3.2 kg Cylinder assembly: 3.3 kg Pump rod, per meter: 0.9 kg Di menls ion s Maximum handle height: 1,190 mm Nominal cylinder bore: 63 mm Minimum handle height 415 mm Maximum stroke: 104 mm Handle length: 1,175 mm Nominal volumeicycle: 324 ml Anglar movement 44 deg Maximum diameter of below- Handle ratio: 7.7: 1 ground components: 73 mm Spout height 455 mm Above-ground assembly: Steel fabrt, maching, galvanidzing, electoplaing Below-ground assembly: Pressing, rolling, drawing and geneal fabication of staindles steel, rubber molding The manufacture of the cylinder requir sophiscatd facilities and rigorous quality control, and has little potal for manufactue in developing countries. Some developing counties may have sufficiendy developed industrial resources to make the pumpstand, though care must be taken with the dimersona accuracy of the bering housings, sindle and the loting holes in the pump head. Installatio and m*jor repairs The below-gound assembly is heavy, and liHftg tackle will be required in deep settings, for both insllafion r and below-ground repain. The design is not suited to A villagelevel maDtenance. o Y tf n nf Routinemanenc Replacing the bandle bearigs requires a high degree of skill toDense that the new beaings and the spindle are not damaged. All the fastenings on the pumpstand I___ can be removed using 17-mm and 19-mm spannes. I 112 UJDP-Wodd Jk War nd Sandadoen hro8m Instaaidon and man Iformatio A manual was supplied for each tt sample dealing with instlla and maintenance. The insallation instrucdons were comprehensive and well illuawd but could be confusing. There was no cear division or heading for the methods of insalling each type of rising main. A quick-eference fold-out page was provided for pump part numbering and a list of tools with codes used in the text Instctions for maintenance wene Limited to an enginering diagam of the cylinder and instctions for disassembly of the pump cylinder. Reistance to con aon Fair. It would be difficultto inset sdi and stonesinto the spout, but the design of the spout allowed it to be easily blocked, with a consequent risk of contaminton. The retining bolts for the inspecdon hatch are easily pilfered which could give direct access to the weU and lead to possible pollution. Redstace to abuse Fair. ITe pumpstand is generaly sturdy, in spite of some deparues from the Indian Standard. None of the fastnings on the pumpstand were pilfer resistant. Pump performance When new, the pump was capable of providing u pb4* u1 hWr water at a rate of aoi y 9.5 li per miute fom a depth of 45 meters (see i - - illustron, right). The effidey vaied from about 40 percent at 7 mets to appMoximately 14 -_ 67 percent at 45 meters depth. In the enduce test the outflow from the OO pump was reduced markedly when the stainer 0 - - - dogged with ctaminants suspended in the or water. When thie shaine was removred, 8 _>_ however, the pump provided more water ta TM in the original nce sts, with a - conseuent improvement in efficiency. Considerable ovality bad been notd in the _ cylinder at the st of the test, and at te end 4 go 40 D soD a0 to *O it was nodceable thbat this ad resulted in Wrk u (WI Ueven wea of pistn the seal Insinificant at all dept tested. P gwm R5pe Swia 113 Endmuce Tet record Hours 2,105 4,106 _~~~~ I Start lnspon Final inspec- and volume tion and fudl check pformance test There were no failum in the endurance test4 although the output of the pump fell off substantially toward the end of the test as a rPut of clogging of the strainer. Te pisn sead was wom to the point wher contact had been made between the cylinder and the melal pl suppg the seal, with some scratching of the cylinder as a result. Howevert both the cylinder and the seal were still serviceable at the end of the test Abuse tests The pump was undamagd by side impacts. In the test where the handle was suc1k repeatedly on its stops, some crakng appeared at the lower edge of the handle exit hole in the pumpstnd. 1T has been noted on some other India Mark II pumps tsd in the same way. Verdict e Pumpnk H pved to be as strong and reliable as other Inda ik U datives tsted by CRL H e, e lbory felt that the potential for its aplcdion was Emited by ma_ce and uacuing drwbacks With its stain ste cylinder, it was only suitable for manufacture in deeloped countries, and inence demands a bigh degree of skill and knowledge. The strainer suplied could not be recommended for applcations where sand or silt might be present Father Iht it Pumpenboese subsequently made a change to fte diameter of the metal plate support the piston seal From June 1987, al sppoct plates have ben 61 mm in diameter her than 62 mm, to avoid damage to the cylinder wal as the seal wears An inunal stiffener has been added to tho sucton tube to prevent ditorin of the footvalve seating face during 114 UNDP-Wed Rai* Wwff ad SantW*W hwgwm D6. Knebel Handpump Tne pumpstand, water tank, and head were all rmanufatured *fom gal- vanized steel. The plastic-sheathed sted ope- ating cable ran through a u _ j1%69 guide bush in the base of the pump head, around a quadrant anched to the handle, and then through the hollow square secton of handle. The rsng main was sup- plied in one continuous length of high densty polyethylene (HDPE) hose. Te cylinder was fitted with an inflatable reinforced rubber expan- .W sion collar connected by a small-bore plastic tbe to a inside thepumpstand. The top secin of the cylinder was formed from stainless _*. } e i steel pipe and was seal to the lowr balf of the ______________ 1J ti ctylinder by a rubber "M-rng. The lower half of the cylinder was formed from stinless steel, with a nominal bore of 64 mm. The cylinder emplojyd a forced rng return for the piston. Th piston and footvalves were rubber, sealng against pforatd stanl stel cones Te propietuy piston sea was formed from plastic. This handpump firom Knbe Drl of Demk is deaped for depths downs to 25 meters. t is outwardly similar to thexIda Mark I, but uses a cable rther than a pump rod, with FrO8A= Aeoi Si. 115 the cylinder anchored in the borehole by an inflatable collar. The pump was tested in 1989; the CRL reference is A9204. Manufacturer Knebel Dri A/S Industlvej 20, DK-8550 Ryomgaard, Denmark Price (October 1990) Complete for installation at 30 meters depth, in lots of 50: $1,095 per pump weigs Pumpstand assembly: 45.4 kg Rising main, per meter. 0.5 kg Cylinder assembly for Opating cable, per meter: 0.2 kg 4.5 inch borehole: 6.5 kg Cylinder ambly with adaptor for 6.25 inch borehole: 9.8 kg Maximum handle height: 1,450 mm Nominal cylinder bore: 64 mm Minimum handle height: 360 mm Maximum stroke: 108 mm Maxdmum handle length: 1,510 mm Nominal maximum volume/cycle: 347 ml Angular movement: 47 deg Maximum diameter of below- Maximum handle ratio: 10.1: 1 ground aembly 100 mm Spout height: 450 mm Above-ground assembly: Stee fabricatio, galzing, geal machining, plastic moldings Beow-ground usemblr. Stas teel fabricaton, plastic pipe extuon, wie inning, nibber/plas mouding, general machining, sing manufacture, crimping Only the basic components of the pumpstand are likely to be suitable for manufcture in developing counti The cylinder is a complex assembly reL -ng a high degree of skill and costly facilites. nstallton and major repais Lifting tackle is not required for instllation, though 'i )% liftng the rig main ful of wate is diffiult Some exerinea necessary to inflat the colar and to D K K ension the oprat cable corntly. 116 UNDP-Worl Ban* Wata ad S=anIato P*ugiw Routie aineac The handle beaings are difficult to replace and not 'f )P suited to typical conditions in the field. Two people ae requred to tenion the cable. The pump is not suitle for a VLOM maintenance regime. . __ Insdtalaton and manteac noaiOD A looseleaf & binder containing information on pare parts, insllaion and maintenance was ulied with each pump. The instucdons for instton were almost enirely pictorial with only a few words of English. Each step of installation was numbered and deptd in clear, uncomplicatd digms, so ta the instllion int ons could be followed by persons wih little liteacy. The instructions also gave deails of a suitabe pump surround and soakaway. Maintnance details were also given in a pictorial format dealing with the chedkng and replacment of the handle bearings. Resstane to otni Poor. Atkough it w ould be difficult to pushsti orstones Intthespout it could easly be blocked off to build up water in the water ank with a consequent risk of contamination. The inspection batch which houses the valve for the expanion element was positioned low on the mpstand: this hatch also gives direct access to the well for chloination but may lead to conta on of the well by surfac water. Resstance to abuse Poor. The mts and bolts of the pumpstand Knebd hwul were unprotcted from pilferage. Although the pumpatand was generally sturdy, the leverage i - availableat the handle could make the pump susceptible to damage from side impacts. Pbmp per4orace t4 1-4 - At the request of the manufater, hei peformance tes were carried out in a t borehole rather than the CRL test tower. / When new, the pump produced over 12 liters / 7 per minute fiom a depth of 25 mters, for an - -in iput of 75 watts At 15 meters ad 2me25 rs and a stoe rate of 50 strokes per minute, o water was forced up trough the sue bush 0 a o 75 U into the pump head and was wasted. k7pun W efficiency vared from 45 percnt ai 7 meters to 66 percent at 25 meters depth. PtVwx Ripe Satu 117 At the end of the endurace tes, the pump produced no output for stroke rates less than 50 cycles per minute, due to a badly worn pion sa. Insignificant at all dephs tested. Endurace (faiures set In talc) Test record Hours: 889 2,000 2,206 3,783 3,s2 4,010 I I I I I I Start Inspection and Rs ma Final inspecdon t volume flow I perforated I and full perform- | check I huap rod ancetest Operadag Opeaing In cyflider cable broken cabe brakex sheaed he opating cable proved to be unreliable, breaking twice in the course of the test Ile inflatable collar was found to be deflated at both the 2,000-hour inpection and the final spection at the end of the test, with consdemble abion having resulted from contact with the wel casig. ITe riing main wor though as a result of contact with the eAing cable At the end of the tes, the piston sa was wom down to its supportplate, with the cyinder bore badly scored a a resu Abuse tests Tne body of the pumpstand was undamaed by side impact Lateral impacts on the handleprouced a offset which restct the bhae's movement, preventng a full strolk. The pump was undamaged by repeated impacts of the handle against its stops, however. Verdict Overall, CRL concluded that the pump was geneally well made but had a number of design flaws that led to problems In the tesL Tbe design of the cylinder and the choice of materials did not lend temselves to manufcture in developing countries. Mintenance problems also made the pump unsitable for VLOM pplications. A number of design changes would thefore be needed befor the pump could be consdered for community use in rural war supply prgams 118 URDP-Wed Ba* Waw and Sau*aden hgouwn Fer foo Folowing the CRL test, the mauf has introduced a number of modifa , including: Cwle. The standed opeating cable has been replaced by a szle- stand 6 mm stainle stod wire which will not rouate under tension, connected to a standed cable within the pumpstand. Rising main. The HDPE hose is now alwed to cool before being made into rolls for traport, to ensue that there is no permanent seL Handle bearings. The design has been changed to make it easier to remove and rlace the bearigs. Inspection hatch. The inspection hatch has been eliminated. The air valve has been repositioned win the pump head with access via the top cover. Handle. This has been reinforced close to the beaing housing to en any bending due to side loads will not impair the opertion of the pump. P*Avm RP4W S.O 119 D7. Aquamont Handpump ConStrUction The pumpstand and handle were galvanized steel and could either be mounted on a protruding well casing or bolted to a concrete plinth. The main handle bearings were taper roller bearings, with a needle roller beaing at the connection to the pumprod. As a a 10 ~~~~~~~~~~direct result of problems in the endurance test - and following a further test at CRL - the manur replaced this needle roller bearing with a more simple and relible hardened steel bush and stainless pin that were evaluated in a separate teL _ _ F ffi fi Below ground, the cylinder gwas formed from glass-reinforced epomy (lsJ2~ 8 resin, with a coating of PTFE on the bore. Te acetal piston was fitœ- with rubber seals: th'e. orginal was Ditdle rubbet, I but it was replaced with Hytrel during the tesL The cylinder and pumpstand were teaded to suit 1.25-inch rising main, and the pump rods were 10-mm stinles steeL Te Aquamot is a conventon deep-wel force pump, inutded by the manu in the U.K. to be suitable for depths down to 90 metres. It was tested in 1988; the CRL reference i A9108. 120 UADP-WId Bak WOW and S &adi ProVw Price (October 1990) Comple for use at 30 mets, in lots of SO: $-?- per pump Manufacturer Eurafric T lding Company Ltd. Cunard Building, Water Street, liverpool, L3 1HR, UK Weights Pumpstand assembly: 54.7 kg Rising main, per meter: 1.2 kg Cylinder assembly: 3.4 kg Pump rod, per meter: 0.7 kg Dimensions Maximum handle height: 1,350 mm Nominal cyinder bore: SO mm Minimum handle height 430 mm Maimum stroke: 225 mm Hindle length: 1,074 mm Nominal maximum volume/cycle: 442 ml Angular movemet: 61 deg Maxdmum diameter of below- Handle Ratio: 5.5: 1 ground assembly: 82.5 mm Spout height: 430 mm Above-ground assembly Steel fabrication, general machning, galvanzing Below-ground assembly: Plastic moldings, glass-fiber-reinforced epoxy resin forming, general machining Only the abov-ground assembly would be suitable for maufcture in developing countries. Instaaon and mnaor repairs Te weight of below-ground components mak lifting I C tacle essential. Major repair could not be carried out at village level and would require a centralized or j_- ftft regional maintenance ngime. I_ __n Routine ntace It would not be practical to replace the handle beings under field conditions, and a complete new handle will be required when the bearngs are worn out. ITe cylinder must be extcted to service the piston and footvalve. Instalaon and mai_e nance oo Suppled in French and English. Clear and well-illustated, but with no instrucons or schedules for maintenance. PAWX RePon Sera 121 Resstance to contaminatio Poor. The design of the spout allowed it to be easily blocked, witfi a consequent risk of contamination. The well-head seal was uns t when the pump was histaled on a concrete pliUnth. Resstance to abuse Poor. The pumpstand was generally surdy, but the handle was not Nuts and bolts could be pflfered easily. Pnimp perforace The cylinders supplied for testing varied In rformance, due to dimensional incondsten- n d' ro cies. Using tie cylinder selected for endurance o ___ - - - testing, the pump was capable of providing water at over 10 liters per minute from a depth ji of 45 meters. for an input of 100 wats (see 14 illustration, right). The efficiency varied from 1' ,/ 54 percent at 7 meters depth to over 80 percent A at 45 meters. _2 Tle perfonnance was retsted at the end of the '1i 4,000-hour endurance test. The rate of outflow -- was unchanged, but the required work int 1 -p was slightly reduced, with a consequent small o - _ -J-mm _ improvement in efficiency. This suggests a 46M reduction in fticton as a result of bedding-in. o, O 10 400 80 100 O 140 HO Leakage Work InUt bW Insignificant at all depths tested. Endurace (failures set In Italc) Test reord Hours: 2,000 2,216 4004 Start Inpecdon: Lowerpiston Final inspoo Beadng at handl- seal rolled off, tion to-pwnp-rod con- jwnxed cylnder necion fai 122 U?0P-Wod Bmk Waw e and SanIta PndWV Te principal problem encountered in the endurance test was failure of the bearing at the handle-to-pumprod connecdon. Replacg this bearing by a more suitable arrangement was one of several design improvements suggesed by CRL. Abuse tests The pump stood up to side impacts with only minor damage. In a te where the bandle was repeatedly struck against its stops, the bandle broke when one of the piston seals jammed in the cylinder. Verdict The Aquamont pump achieved very good efficiencies for cylinders that were within specification, and showed a small improvement in efficiency at the end of the endurance test, but Its overall perfumance was marred by lack of quality control. If these problems can be overcome the pump may prove to be efficient and rliable in service. The materials used in the cylinder make it generally unsuitable for manufacture in developing countres, however. The new arrangement at the connection between the handle and the pump rod, with a hardened steel bush and a stainles steel pin, has been tested separately (see Further infomation, below). ther nf Following the original test by CRL, the manufcurer aranged for the labomtoy to evaluate a number of altentve designs (including the use of plastics) for the handle bearngs, and particularly for the connection between the handle and the pump rod. The most suc ul amngeent retained the orignal taper rolU for the fulcrum bearings, with a plain, hardened, proprietay bush and stainless steel pin at the pump-rod connecton. This was tsted using a severe regime onally developed for testing Afidev handle bearings, and the bearings were sdll in workig order after 10 million reversals. This new bearing arrangement was therore adopted immediately in place of the origial needle roll bearings. Other improvements introduced by the man er foLowing the CRL tests: Improved quality control procedures so that pump heads should never be splied with incorrect nuts, washers, bearing spindles, etc. Welding to be checked for acceptabiit and any roughness or sharp edges in the galvanisng should be coted before the prduct leaves the factoy. Rising main flange adapter to be tped after galvanizng to avoid the threads being clogged with zinc. hopm R5er S.f 123 Rubber gWt to be inaeased In thickn to tak up any distortions In the base the pump. Packaging of the pumps will nomay be to the cusom' requirents so if no fork lift was available the pumps would be sent in small light cates to make them easy to handle. Pump rods to be packed on a tmber base with timber noggis for fork lift access. Me maximum weight of one bundle is 80 kg unless the customer ackowledge receipt of heaver loads. pHmpitand height to be inacrased by mounting it on a corete plinth above th user area Handle to be stregheed. 124 UDP-WaF1BwE Wawr d Swaltkat wmm D8. Atlas Copco Model 111 Handpump Tno pumpsd was fabricated from galvanizd steel and was designed to be l , / !. z ! | n . t built into a concrete plinth. The handle was deigned to apply the necessary amount of pre-ten- s|oning on the rubber hose; it was formed from lvanized solid steel bar and was exceptiona/ly long and heavy. Two handle lengths were availae, 1,300 mm or 1,600 mm, to suit different depths. The handle bearings were bonded rubber with no sliding contacL The rsing man was fomd from stain- less steel Connec- dons between the three meter legths were made using rope threads sealed with a rubber -fring. Tbes sectio ooud be aembled and tightn by band to prodWuce a seal at up to 45 metes stauc water head. TIe pump rod was also frmed fiom stainle st with connectons simila to those of the dig maiL Pastic gudes were suppled for use with the pump rod. The perforated hrgaw Ripe &re 125 the pump rod. The pefoated pumping element surround was farcated from stainless sted and provided a small amount of pre-tensio to the rubber hose. Atas copoo offer two unconentonal deep-well handpumps, models 111 and 122. The 122 uses a convendona cylinder, whereas the 111 has the unique rubber hose pumping element first see on the Petro handpumP. Ts exploits the reduction in volume of a stretched rubber hose to provide the pumpig action. The Adas Copoo pumps were tested in 1988; the CRL reference is A9136. Manufacturer Atlas Copco Energy AB S-105 23, Stocdiolm, Sweden Price (October 1990) Complete for insallaton at 30 m depth, in lots of 50: $2,110 per pump, including stainle steel riser pipes. Weghs Pumpstand assembly: 49.5 kg Rising main, per meter. 1.3 kg Cylinder assembly: 11.0 kg Dimensions (with 1.6-meter handle) Maximum bandle height 1,460 mm Spout height 565 mm Minimum handle height 210 mm Maximum diameter of below- Overall handle length: 1,835 mm grund assembly: 77 mm Angular movement 450 Handle ratio: 8.2:1 Mamfatu I I Aboverundmbly: Steel fbrication, genera maching, galvaizing, rubber mding, aluminum casting Below-round aembly: Specalit rubber brication, stainless steel fabrication, plastic molding, aluminum foing The pump has limited scope for manufacture in a developing country. Both the pumping element and stainle steel pipes reqre speilt s and facitie bftallon and miJor repairs Installation and remval of the belowgound assembly is quick and easy with the pipe clamp and lfting tool spplied with the pump. Dismantlig the cylinder is likly to be tme-consming, howve. _ _ _ _ _ 126 UDMP-Wod Bao Waw and Santa*e ?Prqr Routinemiteac Te handle fulaum unit must be replaced as a complete unit, and two people will be needed to cope with the J very heavy handle. A good tool kit was supplied, but the pump is not wel-suited to vge-level Instalatio and ma ane nfo n A comprehensive manual was supplied with the pump,, induding directions for building a astable platfom, full insallaton instructions, a spare parts list, and a schedule for weekdy and monthly inspecons. The instrtons were well illustrated, though some refeece to the text would also be necessary to assemble the pump cormcfly. Resta to tion Acceptable. It would be difficult to pushi sticls or stones into the spout, but it could be blocked off to build up water inside the pump, witlh a consequent risk of contmination. Resistane to abuse Good. The pump is sturdy, and the fstening are protctd from pilferage. Pnp peformnc The pump demanded relatively high levels of Adaooo 'I' mNwuri force to achieve a reasonable outflow. It was tsted with both the handles supplied: the best - resl were achwved wt the longer handle Ii (see Muf1lo, right). For a dept of s 45 meters, an input of 100 wat produced an i/ outflow of apmaty7.S lits per minute. The dEcn varied firom 20 percent at *- 7metes to oximately 56parcent at 45 metrs, contastng with 17 percent to 2 around 44 percent using the shor handle. | if After the endurance test, the available stroke E 7 am_ had inacreased, suggestng greater freedom of 46m movement in the handle bearings. An input of 4. i ....i .I 100 watts produced an outflow of9.5 lie per 0 m 6WO minute from 45 meters dept k inlut (W Iasignifcant at al depths tested. Prmream Rqwoi S& 127 f Endurance Test record Hours: 2,006 4,000 I -I Stt Inpecdon and Final insec- volume flow chec; tion and full guide bushes badly performance worn test her were no failures in the endurance test, but the acetal guide bushes in the pump stand were found to be badly womn at the 2,000- hour inspecdon and were replaced by polypropylene busheL In the fna inspection, the rubber in the handle bearings was found to be extruding from the beanng housing, though the beaing wer sdll fumconal. The polypropylene guide bushes fitted after 2,000 hours were in good condition. The ends of the cylinder were badly coroded. Abuse 1s The punpstand body was undamaged by sde impacts; the handle was slightly distted but was stil funcional. No damae resdted from repeatedly banging the handle on its stop. Verdict The Atlas Copoo Ill handpump was a heavily bult pump of geous proportions and resstant to abuse. The pump was easy to instal using the clamps and wrenches supplied, but it was difficut to dimande the below-ground components. It should be possible for the pumpsnd and other above-ground components to be manufactured in many developing countries, aloug the below-gound compnents are unliky to be suitable for manufacture in develping countries at the present ime. The pump proved to be reiable during the endurance phase and provided all pumps of this type are sWlid with the polypropylene split guide bushes the pump should give good service in the field. FJre foaion Tbe man r bas intrduced a number of modifica foowing the tesing by CR1L: PMp stand The split guide bushes are now manuf red om polypropylene. 128 UJDP-Wed Ba* Waw and Saaia*_ Pw Handle A ng web bas been inId in all hadles snce the bemnning of 1987. The leara betwee the web and the top cap has been adjuswted correctly. Webblng stap A sathetc fiber sap is now standard. Pmp rod and rising main To ense that the o-rngs will not be diwlaced duing ssembly, they can be lubicated with a little petroleum jelly. A note about to Xt effect wil be added to the insalaton manual, and petroleum jely will be supplied with the pump. Prp= R Sofa 12r D9. Adas Copco Model 122 Handpump Constmucton Te pumpstd was fabricatd from gal- vanized steel and was designWd to be bilt into a concrete pliuth. The bandle was formed from i ^ f ~~~~~~~~11~CZ4 galvzed steel bar. It was exceptionally long and heavy, act- ing as a counter- I ~~~~~~~ ~~weight to the below-ground com- IP. c5' 1 i ponents. Two handle S Ica~~~kgf were sp plied, 1.3 meters and 1.6 meters, to sut different depths. The handle was mounted on rubber torsion bushes, elimi_ating sliding contact. The connection to the pump rod was a * webbing sra. -. rul~~~~~~~~~7e risng main was .______________ _ 42.4 mm outside diameter stainless steMe, with connecos made by rope thueads formed in the pipe and saed by rubber o-rings. The joints coud be dgened by hand to prduce saiactory joints at up to 45 met satic head. The pump rod was 10.8-mm diameter sainless steel The cylinder was sainless aseel, and Inpaed a thWd va for empWng the rising main when witdsawibg the pump from the well he piston and footvalve were plastc and the piston 10als rubbra. 130 UZDPWelBakWaaniSanalnPqw Atlas Copoo offrs two deep-wei handpumps,, models 111 and 122. Te 122 is a corventional foe pump with a piston and cylinder and isnded for use at depths to 6C metes. The pump was tested in 1988; the CRL refeece is A9136. Manufactuer Adas Copoo Enery AL S-105-23, Stockhlm, Sweden Price (October 1990) Complete for Insallatdon at 30 m depth, in lots of 50: $2,220 per pump, including stainless steel riser pipes and pump rods. Pumpstand assembly: 49.5 kg Rising main, per metew 1.3 kg Cylinder assembly: 7.6 kg DimeusIoos (using long hane) Maximum handle height: 1,640 mm Nominal cylinder bore: 65 mm Minimum handle heig 120 mm Maximum stroke: 185 mm Handle overal length: 1,835 mm Nominal maxiinum volume/cycle: 614 ml Angular movement: 580 Maximum diameter of below- Handle ratio: 8.2: 1 ground assembly: 75 mm Spout height: 565 mm Above-gound asembly Steel bcaton, genr machining, galvanizing, rubber moldn, aluminium catng Below-groud ainembly: less steel fabricaton, genl mnhining, plastic moldi The manufture of te stainless seel cylinder and ris main pipes require specialis sdkils and cilities. However, the pumpstand bas some potental for manufte in deveoping counies nsaualou and major repas nsllation and emoval of tie below-oud asembly t was quick and easy usg the pipe amp and lifting tool supplid with the pump, although worldng on the pump-rod joints was rahr awkward. Prhimn PRqie Soft 131 Routine a nce The handle fAdlrum unit must be replaced as a comploet unit, ad two people would be needed to cope with the very heavy handle. A good tool kdt was supplied, but the pump was not well-suited to village level n1 Installation and iee I n fmation A corehenve manual was supplied with the pump, including dictions for building a suitable platform, ful Isalon instructions, a spare-pa list, and a schedule for weeldy and monthly insoptons. The instrucons were wel illustrated, but needed some clarification in a few areas. Resstance to coni on Acceptable. It would be difficult to push sticks or stones into the spout, but it could be blocked off easily to build up water in the pumpstand, with a consequent rsk of contamination. Resstance to abuse Good. Tbe pump was sturdy, and the astenn were protece from pifage. Pump perfomace ______2 The pump was tsted with both the handles supplied. The results were similar, though olw _t_tu slightly better for the 1.6m handle (see 1 8 =i Illustration, right) than for the 1.3-meter w handle. Te pump produced an outputofjut over 8 Le per minute from a depth of4S - meters for m input of 100 watts. Efficiency 14-_ ied frr about 25 peent at 7 met to ovrer 60 per4ent at 45 meters depth. 12 _ After the endumance test, a dight ipvement li o in efficiency was measured, suggesting a r JM reduction in friction. However, because the * - um _ + hlandle bearing had lost some of its resilience, - m j more effort was needed than befre on the o, " oo m M go return stroW Sigficant leakg, varyig from 6 liters per minute at 45 met head to 36 liters per minute at 7 meters head, was observed from the footvalve. 132 URDP-WeI Ban Water and &nIa*n Pmpw Euduraxce Test reord (failures set In Italic type) Hom: 1,560 2,006 2,103 2,377 3,7M 4,009 3,801 - '' -1'I --- I I I Start Webbing lnspcn on & Handle Pn Finl in- qp and volume foeetvw bwkex valw spection pled flow check seized due around seid and perform- out of tos a4 beadug hut ance test bottom . lIdelesguhr housng due to conneclion At the 2,000-hour stage the webbing strap was replaced at the manuf's request with a strap of synttfiber. After the addition of sand and kilguhr, the piston valve was prone to seizure. After the handle assembly broke, the manufacturer provided a modified verion with a web over the top of the bearing housing and a modified top cap which did not at properly. In the final inspecton, the handle bearings and the webbing str were found to be wor but still fincioning. The footvalve was still leakig but functional, and the sand trap was half full of sand and keiselguhr. The piston seals were sast. The pump rod cenoiier in the rsing main was badly wom. Abus tes The pumpstand body was unamagd by ide impacts; the handle was slightly distorted but still functionaL No damage resulted from repeately banging the handle on its stops. Verdict The Atlas Copoo 122 was a heavily built pump of generous proports that gave it good reistance to abuse. The unusual pipe conncons made it easr install, maintain, and repair than many pumps with conventional threaded pipe joints, but it was still not well suited to village-level mainanc. Tbe pumptand could be produced in many developing countries, although the cylinder and ring main would have to be imported. The pump was unreliable in the enduroe test, and micatn caied out by the mansufact during ting would need to be implemented to prevent failure in the field. The usable life of the bearing when woing at 45 meters water head was short For Proerm RSpoilerk1 133 sandy applcations the dearance on the valve stems would need to be imcrased to prevent premture Sfaiure. Additlonal Infomation Modificatons have subsequently been introduced by Atlas Copco: Handle. A web has been incrpraedin the handle and the clearance between the web and the top cap has been adjusted corecty. Webbing stap. A synthetic fiber strap is now standard. M,om baring. The rubber type has pred to be inadequate in certain clhmates and has therefore been replaced by sealed bai races. Rising main Joint A litde petroleum jeUy is nded to ensure ta the t -rins will not be dispaced during assembly. A note to this effect will be added to the instructions and petroleum jely will be supplied with the rig main pipes. Clinder. The deances in the footvalve and the piston valve have been increased to eliminate stickng problems in sandy water. The sand tap has been left out. The lower p=ston seal has been reversed so it will sweep the cylinder walls cean on the return strole in sndy water. This lower seal can also be used as a sp part and replacement for the uper sea Footave le. The dein of the sealing sfac in the footvalve has been improved. Centralism. When necessary, centralizers can be deivered in other materials or with bushing to derea wear. If any centlie Is worn through, the plastc bushing on the conecting nut gives protection undl the centizer is replaced. 134 UNDP-WIBa=k Wir andSankaAPr D1O. Aquadev Handpump The pumpstatd and adjustable tbar handle were steel, procd by pai"vated zinc decto- plating. Both the handle fulm and Lr X _ fpum an gro used identical two part nylon and polyacetl plain beinpg , with stainless Stee pins. Thersn ;: ma iin was uPVC with solvent-cementd joints, and was suspended from the pumptand by a compressed rubber cone, with a oi polypopylene rope to retdeve the rising mahn in the event of a brea . The l-mm pump rods, available in carbon or stainless st, had upset ends and plastic couplngs which also acted as rod centalizers. .__ _ _ __ The 50-mm intnal diameter cylinder was a secon of rising main lined with a sainless seel tUb he piston and footvalve bodies used idi components,a injection molded in polyaceta and welded. The valve bobbins were molded rubber. The plston and footvalve were concted togete by a nylon line to enable the footvalve to be acted with the piston through the sig main. Ihe Aquadev, made by Mono Pumps of the UK, was the first bandpump from an establhed manfaurer to be based on the Afridev coept simce it was placed in the public domain. It is a deep-well force pump, deed to siable for manuf e, ^vW"3 R4pe SeriS 135 installation, maintnce and repair within tie existing reources of many developing counties. The pump was tsted in 1989; the CRL reference is A9218. Manufacturer Mono Pumps Limted, MENCA Divison Cromwel Trig EstaBt, Cromwel Road, redbuIry, Stockport, Cheshire SK6 3RF, United Kingdom Prce (October 1990) Complet for use at 30 m depth, in lots of 50, with sainless steel pump rods: $880 per pump. Weghts Pumptnd assembly: 54.7 kg Rising main, per meter: 1.2 kg Cylinder asembly: 3.4 kg Pump rod, per meter: 0.7 kg Dnions (with handle adjusted to mid-point) Maximum handle height: 1,416 mm Nominl cyEnder bore: 50 mm Minimum and height: 497 mm Maxmum pump stroke: 225 mm Handle length: 921 mm Nominal volume per cycle: 442 ml Angular movement 600 Maimum diameter of below- Handle ratio: 4.1: 1 pound assembly: 82.5 mm Spout height: 760 mm Manfatuing Above-rond assembl: Sted fabication, general m ning, plahsic molding, rubber g,~~ a molding electroplating Below-gud assembly: Plastic molding, rubber molding, stin seel fbicat Developing countim with skils and filities in geer al engig and plstics procesing would be able to manufiaue the pump. Strict quality contrl wil be needed to ensure locally produced uPVC pipe meet international stndards nstallation and maJor repair Liftg taclde is not required, but tg off the ropes suppotng the ring main requie many hands The short rods supplied with the pumps avoid the need to cut a thead oan the top of the pup rod. ____n_ n rmsth - maintenanc A dngle spanner is suppiod which fits all the asenings on the pumpstnd, though a 19-mm spanner would also be required to relase the lock-nut on th pump rod. I The handle beaigs can be relaced in minuteL 136 UJWP-WblBWatu ad S=&anaio gi Imtaftoon and e IfatIon A manual was supplied, in Engb. The insallaon tuc s rled ather heavily on an un dersnding of the text, but the mainteume infrmation was presented on a single encapsulated sheet and littl reence was needed to the text accompanying each of the 23 line drawings. Resistan to cot Fair - the base of the puwasnd not flat and could allow uwface water to penetate, although the design allows the boehole casing to extend above the plinth level iside the pumpstand. Ie spout could be bloced off to build up water in the water tank, with a risk of contamiaion. Re _stane to abuse Good - the pump was sturdy,, and all the ftnings on the pumpstand except the flnge bolts were p from pilfernge. n p With a cylinder of 50 mm diameter and a reatively long strk, te Aquadev could lift re qatites of wate, ev from deep A%Mdw hUUp stting. I'This demanded crrespondingly igh levels of work input, however. For an input of : - - - 100 watts, the test results incat (rh) that . te pump would be capable of prvidn water is at over 11 HtL per minute *iom a deK of // 45 meters T efficiency varied from around 16 4 50 percet at 7 met to about 80 percent at 45 meters depth. 14 After the endmuance test, the pump _IJ / 7/ dlivered more wate per stroke when 12 new, but at the expene of an increase in work input. Overall efficien did not cdmp 26m significantly..-46 _ o so to MO 200 Iacge VWItlpkut (W) Insint at all depths tsted. ProVp Rqw ep va 137 Endurace Test record (filur set n Inbllc type) Hours: 2,013 4,006 I I Start Volume Final inecion flow check and performance test ne pump completed the endurance test without incident. In the Snal inspection, the reining lug had broken off one of the handle fcrum bearings and caused a rcumferntial crw The hanger barng wer atsfactory. Both the handle fulcrum pin and the hanger pin were badly coroded on the threads and captive nuts. The pumpstand was badly croded inside the water lank, and the handle showed signs of rus where the surfice had been damaged. The pump-rod connectors were worn and allowed free play in the rod string. Inside the cylinder, the piston and footvalve were in good condito, but the nylon line attaching the footvalve to the piston was frayed and broke when it was used to attempt to witdraw the footave. Abse tests The pump was undamaged by sie impacts on the body of the pumpstand and laterd impacts on the handle. In the test where the hande was repeatedy struck on its stops, a weld on the handle secming the handle top yoke to the forks broke. The hande was rplaced with a handle welded on the inside angle of the yoke as well as on the outside. The second handle also failed, this time at the base of the handle tube 3 mm in front of the welded area A third bandle completed the test without further incident. Verdict The Aquadev performed well in the tests, gave good effidency and completed hie endurance tst without incident. However, the pump's low resisDce to corrosion will need to be impved to ensure the logevity of the pump. The handle also requires strengtng to improve ince to abuse. T.he pump is suitable for manufictur in a number of developi countries, and its ease of maintenance should ensure its inclusion in VLOM-based projects. 138 UNDP-Word Ban Wawr axd San&aPrq8n Fre nfration Pump rods. All pumps are now suppled wi less tel pump rods. Onih. The extena finish of the pmpstand is now phosphate and chromate treatment to Britih Standard 3189, fowed by two coats of r-fini nt. Ihe inside of the punpatand is protcted by a biminous coting to Bdtish Stadard 3416. Footvalve retievaL . The nylon connectng cord is no longea fitted. Instead, each pump is supplied with a sepante fishing tool for reievthe footvalve. Rsng main. Pipes are upplid in 2.8 meter lengths to maiain strighte in transt. IDStrctions. A pictor leaflet enlated in plastic is now suppLied with each pump. The main lngage is Englisb, but p ion bas been made for local language subtitles. Locad _m e. Disussons are under way to set up jint manufIng venus in geia, Malawi, Keya, Tzan and Ehopia Prop= Repn Sw& 139 DlI. Afridev Handpump _ _._____J_________ Construction The pumpstand was galvanized steel, with a t-bar handle adjustable for length to suit different water depths. Identical two-part nylon and acetal plain '?~ |bearings, running in stainless steel housings, _. -<¢> were used for both the - D bhandle fulcrum and the pump rod hanger. All routine maintenance on the pumpstand could be carried out using a single spanner, wbich was sWplied. The rising main was uPVC pipe with solvent cemented couplings, and was suspended from the pumpstand by a compressed rubber cone. The pump rods had special joints which eliminated the need for tools for assembly or dismantling. The joints incorporated rod cen- tralizers. The cylinder was a length of the riig main pipe, fitted with a brass liner: this gave it an internal diameter of 50 mm, and meant that the pistn could be extrcted without distubing the rising maim The piston and footvalve were molded from acetal, spun-welded, and were fitted wih rubber valves and seals. The safety rope would enable a broken rising main to be recovered. 140 UNDP-WId Ba* Waia ad Sanka*In= The Afrldev is a deep-well handpump for water depths down to 45 meters, developed from the Maldev handpump. The design is in the public domain, and aims to demonste the VLOM concepts of easy, low-cost maintenance and suitabiit for manufcture in developing countra. The pump was tesd in 1989, with samples supplied from Kenya; the CRL reference is A9218. Manufacturer East Afncan Foundry Works (K) Ltd PO Box 48624, Nairobi, Kenya Price (October 1990) Complete for installaion at 30 m depth, in lots of 50: $643 with galvnized rods, $978 with stai seel rods, per pump, ex-works Kenya. Export prices may be somewhat lower due to govemment subsidies. Weghs Pumpstand assembly: 47.2 kg Rising main, per meter: 1.4 kg Cylinder assembly: 6.1 kg Pump rod, per meter: 0.6 kg Dimensions (with handle adjusted to midpoint) Maximum handle height 1,285 mm Nominal cylinder bore: 50 mm Minimum handle height: 350 mm Maximum stoke: 234 mm Handle length: 1,120 mm Nominal volume per cycle: 459 ml Angular movement: 630 Maximum diameter of below- Hanl ratio: 4.1:1 ground assembly: 95 mm Spout height: 465 mm Above-ound assembly: Steel fabrication (including stnes), generl machinig, plastics and rubber molding, galvang Below-ground assembly: Plasdcs extrusion and molding, rubber molding, steel fabrication Developing counties with kils and facilites in general egneering and plastics procing will be able to manuficwure the pump. Strict quality control will be needed to ensure that locally produced uPVC pipe meets intnaonal s rds. nstllaion and major repairs Lifling tacle is not reqired, but tying off the ropes p aSupring the riing main requres many hands. In the latest design, the need to cut a thd in top of the pump rod has been elminad (see Further Information, c_n_*__ p. 138). hotiwm Re a Soda 141 Routine oiena_c A single spanner is supplied which fits all the fastenings on the pumpstand. The handle bearinG can be replaced in minutes. nstallatio and e fotlo The instuctions sWlied with tie test samples were poor, with a number of errors and ambiguities and a lack of clear diagams. It would have been difficult for a person with litde liteacy or with litte knowledge of the English language to install the pump. Resistance to o mination Good, although the spout could readily be blocked off to build up water inside the pump, with a consequent risk of contaminaon. Resitac to abuse Good. The pumpstand was robust and all the fastenings on the pumpstand except the flange bolts are protected from pilferage. PUp pformneidw With a cylinder of 50-mm diameter and a relatively long stroke, the Afidev can lift large C o ot "), quantities of water, ever. from deep settings. I This demanded cdigy high levels of I work input, however. For an input of 100 ,e - - watts, the test results indicat that the pump would provide about 11 liters of water per mnute from 45 meters depth. Tbe dEiency . - 2 vaied from 35 per cent at 7 meters to app¢oxi- / mately 70 per cent at 45 meter. After the endurance test fiction in the pump 10 - was reduced, with a consequent impovement I6m in efficiency to around 75 per ent at 45m s m i depth. 160 200 280 0 80 m m0 V4rk kOut (W) Leakage There was no dsgnficant leaage at 25 meters or 45 metes depth At 7 meters depth, leakage from the footvalve was measured at 57 ml/minute. 142 UDP-Woild Ba* Wawr and SaiatiPon _m Enduanee Test record Hours: 2,003 4,009 I Start Inection Finalinpection and volume don and ful per- check formance test There were no failures during the endumance test. However, the endurance te certainly contributed to the subsequent pump rod failure during the handle shock test (see Abuse tests, below). In the final inspection, the handle bushes were worn and some were very difficult to spate. There was some evidence of corrosion at the stainless steel liner/sleeve interfaces. Superficial rusting was also found in areas of the pumpstand and handle where the galvanized coatng had been wom away. The pump rods were severely conroded. The rsing main was badly worn where the rod centralizers had been in contact with it The piston and footvalve were in good worldng order. Abune tests The pump was undamaged by side impacts on the body of the pumpstand and on the hadle. in the test where the handle was repeatedly struck on its stops, the pump rod broke about half way through the test at one of the hook-and-eye connectors. The joint was replaced and the test was completed without further incident. Verdict Ovea, the Afidev pump performed well in the labortory tests. It completed the endurance test without failure, though there was evidence of potential problems related to rising main wear and corrosion. The pump was suitable for manufacture in many developing countries and the inherent ease of maintenance should ensure its inclusion in VLOM-based projects. Changes in the pump rod jointing method, corrosion protection and centralizers will need to be considered to ensure long term reliability. The instuction manual must be clarified and changed to include simple line drawings and to minimize the accompanying text. hftVm Repor Sesie 143 Fwrthw fo A number of cges have been introduced to the Afidev pump since the CRL test was caried out, principaUly: Rod hanger. A clamp-type hanger is now used, whicd accepts a plain 10 mm pump rod and secures it with a 16 mm bolt. Tis has eliminated the need to thed the end of the pump rod during installation. Rod centsaler: molded rubber centzilizers are now used intead of polyethylene, and have performed succesf y in the field: wear takes place on the centralizers rather than the rising main. Footvalve: the footvalve fitting has been modified to elminate any risk of the footvalve's jamming on the piston. Ilstructions: the instructions have been completely revised, although it is acknowledged that there is still scope for further iovement. A taining video on insallaton is also planned. Pmp rods: stainless steel pump rods are now recommended for applications where groundwater is less than pH 6.0. Other comments from UNDP-World Bank project staff: Manufe of Afxidev pumps is under way in Kenya, India, Pakistan, and England, and i planned for Malawi, Ethiopia, and Madgwar. Corrosion of the bearing housing liners noted in the lboratory test has not been appat on pumps in the field, even in severe conditions. The pump incorporates overflow holes just above the spout to prevent a build-up of water if the spout is blocked. The two partsof the bearing assemblies are imtended to be difficult to seprte, to ensure that bearings are always fitted and replaced as pairs. The minimum romme operaing depth is 10 meters. 144 UNDP-Wod NW* Waw rd S onL8 P?opw APPENDIX. CRL's Handpump Test Program he Consume Reseach Labor y bas boe involved with handpump tsting and development since 1977, for the Oveseas Development Administration, the World Banl, donor agencies, and manufcuers. CRL's approach reflect over 30 years of expeence in the feld of ae tg of a vey wide range of pducts. Te remaindr of is Appendix describe the bandpump t progam in de, including the labotoy's condItons of aceptn of handpumps for tesng. CRL's faciitfies for testing and developmt of bandpumps include: * a bandpump sing towe • 4 blind borehol, tie deepestofwhich8s 1 m, with varable water levels * cylinder testg rig with 8 testsaions * beaig tet rgs in dambers * large wor fadities For furier inforation, you may contt: Ken J. Mills OBE Consumer Resach Laboary Harpeden Rise Harende Herts AI5 31J United Kingdom CRL hnpump test procedures 1. Description The test samples (a minimum of two complete samples) oWld be of the manfactur's normal output. Wherever possible, sample pumps ould be obtained tugb independent_Z p =-magencie 1.1 Manufacturer or agency: name and addres of pump manu and/or supplying agency hugvxw Rqun s&a 14S 1.2 Pump model and tpe: manufcer's model reference, deep or shallow we3l tpe, free discharg or delivery lift 1.3 Cost 2. Inspectdon 2.1 Pakging: e t of bow wel the packaging has stood up to traoi to the lboy, and of the suiability of the pakaging for tansport in develing counties. 2.2 Coiaon of pumps: Whetr coeived in working orde, summary of dects on delivery. Te pumps will be dismad and insected for visible defects. 2.3 lirure: whether suppied with pump or obined from otber source. Ae n of carity, accumcy ad useuness. 3. Weights and measures 3.1 Wdghts of icipal compnents pumpstand inclung handle, cylihder, ring main per mest legth pump rods per meter length (inudng couphin) 3.2 PincpW cylinder dimensions: nomial bore and stroke, nominal volume per stok, usable cylinder length 3.3 Cylinder bore dmeter. m ured at ten pints along two r axes pral with the cylider axis, to check for ova and taper. The surface roughne of the bore is meaurd at thee points alog axes palel to the cylinder axis. 3.4 Egnomic measurements. handle and spout heiht, angular movement and vedociy ratio of bandle, descriptio of exit water pattern. Pumps designed to be ntad an pinths wil be mounted at the man turer's ommended heighL Where this informatn is not available. pums will be iaed so tat the midpoint of handle orF-ation is as lose as possible to 0.9 metes from floor level, subject to a maxium pout beigt of 0.6 meter 146 UiDP-WbrdBakW&ad SaaltuI S w 4. Pump perormance 4.1 Volume flow, wok input and eficiency: ma ts of volume flow, work input and th efficiency of tie pump can be bind 5 a single tet prcdwu. Strin gauges win be atahed to the pump handle to measure the applied forces, and a roty potentometeror opical transduoer fixed to the body of the pump will measure angular movement. A m is progammed to record the daft from the tansducers and calculate the work done by the opeo on the pump. The computer cmpa this work done on the pump wi the work don by the pump in aising water (the product of the wet of wat raised and the head) to calulate the efficienc of the pump. The tansducers are cabra at the start of each test; the calibration procedure is built into the computer program. Shallow-wel pumps are tested at 7 metes head. Deep-well pumps are nomaly tsted at 3 heads - 15 meters, 25 meters, and 45 meters, but other depths can be accommo. For each head, tie pump is sted at three opeating speeds, normally 30, 40, and 50 strokes or revolutions per minute. The stro rate will be alred if necay, and act rates agreed with the manufactr before testing. For each t, the computer will plot appled force on the handle aginst bandle displacemen 4.2 L1eaa the volume of water lekg pat the footval is mead for tie same heads as the tests of volume flow, work ipu, and efficiency. 5. User tal A small number of users will be recruited. They will be women and children, of vaious heigts and ages. Al users are give an opporuity to become fami with the pump and to find their prefered metod of opeaon. They are asked to fill a 10-liter coniner in their own e The number of stks or rvolutions and the time ta are noted. Deep-well pumps are normaly se at 20 meters depth 5.1 User comments: each user wil be asked to fill out a short quesionnaire to record their opinions of the ease of opeation of the pump. 5.2 Observaon of users: methodical obseatons of the reationss between pumps and peopwle are made, reinfored by selective wideo recrdings. 5.3 Water exit pattern cmments ae made if the water pattern leaving the spout woud not be suitable for narow-neckd vessl Pmgram Repa Seda 147 6. Enduracetest Two stages of 2,000 hours with a check test of volume flow after the first stage. The pumps are mechaically driven, normally at 40 strokes or revolutions per minute, or at the speed most approprate to pump desig. The handla of the reciproing pumps are driven In simple harmonic motion imposing no shock loads. Ihe simulated depth is 45 meter for deep-well band pumps unless otherwise agreed with the manu. 6.1 Stage 1: Clean bard water, approxhnate pH 7.2, followed by volume flow check. 6.2 Stage 2: Hard water with one gam per liter of Kleselguhr, masimum paticle size 7.5 um, and one gram per lite of fine sharp quartz sand, pardcle size between 75 pm and 500 pmTis Is followed by a performance test at three speeds at the endmance test head to compre with the rets obtained inidally and obtain data on changescaused by wear. For stage two the wat is agitated, and frequent chk made to ensure the correct concentratill of contaminants is maitained. 6.3 Failure Repot any fure is examined and an assessment made of the probable cause - use of mate, design, inadequate qualit control, or poor manufae, for example. Sugestions for desig improvemets or manuing changes are made. 7. Abuse tests 7.1 Handle shock loadin tes whe applable: control shocks are applied to the handle sops using Impacts detmined by using a human opra whre the handle is allowed to tavel with the normal level of effort on to the stops Both depwel and shll ow-well pumps are tstd at a head of 7 meters, since for deep-well pumps fte user i more lily to bit te handle on is stops when the pump is used at relatively shallow depth. The ot will be carried out i the normal endurance strolk speed at a rate apriate for the type of pump. The test consists of 96,000 shocks for force pumps and 72,000 shock for scion pumps, or until pump failur 7.2 Impacts on pumpstand: using a pendulum, impat in stes of 100 Joules to a maximum of 500 Joules, on the center of the pumpstand. 7.3 Impac^. on handle: using a pendulum, impacts in steps of 50 Joues to a maximum of 200 Joules, on the center of the pump handle. 148 N-W orI BakWo*Ww and &San I 8. Engineering assessment The pumps are dismantled. Each component examined to evaluate the matrial, the method of manufe, the degee of kill required, and the sandard of workmanship, to form the basis of the alsement of suitability for manuface in developing countries, 8.1 Meias, mac methods, etc: ide a of materials and manufng proceses used for each component of the pump. Assesmet of fitness for purpose of chosen mateials and process. 8.2 Suiabity for manufctw in developing countries: summary of manucfct proesses required, and assessments of the degree of skill demanded for each process 8.3 Ease of installation, maintenance and repair. assssment of techniques, sdils, and equipment required. In assessing ease of instalation, maince, and repair, CRL considers the degree of echnial compewne requied by the design and construcion of the pump, and whether the pump could be repaired using indigenous mateals. 8.4 Restance to con and abus: assesment of snita seaing of boti pmstand and weliead. Assesment of resstanc to ddiberae abuse, pilfeage, impacts by domestc animals etc. in a1ssing santay sealing, CRL coniders the resistnce of the p_mpstand to acciental or delt co, and the ih ood of contaminatio of the weJl by face water. Resistance to abuse includes both the probale sscepibiliofthepumpsand to ats (fromdomestcaimals, forample), pilfrage, or vandism. 8.5 Potential afet hazards: assessment of potential dangers - of finger traps, inseu fsenins, and projecions, for example - to both punp users and bystanders. 8.6 Suggested deg improve: suggesons for ioveents in eitier pump desig or mauf e, at minimal cost CRL hopes these will sdmulate a responae from pump manuf 9. Verdict A dwrt summary of the main good and bad feaur of the pump and its perfomnce. Comment on suitability for manufacture in devdoping countries. Prop= RpAl &Sed 149 10. Reprting 10.1 TM. fis terdm repo contan al infoation pror to the start of the endure te It detis perfrmIance test results for the pump. s pelmina report gives manuIactur an oppornty to comment on he tesing to date, and to qustion any of theslts. CRL hopes this approach wi encomrge a diaogue with ma c 10.2 Contas wil be made with the cient as reqired when significant problems are encountered in endurance teing. 10.3 The final Technical Report includes ful details of the pumps, test procedures, results, rleva drawings, and photraphs. 150 URDI-W.d Bak Waw and Swiade himm REFRECES Arlosoroff, S., et al. 1987. Commuiy Water Supply: The Hadpsmp Op6tn Washington, D.C.: The World Bark. Bsselink et al. 1990. 'Bebaviour of Deepwell Handpumps with PVC Rising Mains.' IAD Handpump Project Comume Resech Labotory. 1983. "Acetl Bearings." Harpnden, U.K.: Consumer Research Laboratory, Report A91060. 1985. "Dry Bearings.' Harpenden, U.K.: Consumer Research Laboratory. Report A9161. -. 1986a 'ncdia-Mali Handpump.' Harpenden, U.K.: Conumer Research Laborty. Report A9123. 1986b. Plastics in Below-ground Components of Handpumps. Harpenden, U.K.: Coumer Reserch Labotory. Report A91538. ---= 1986c. Besobell Micro Handpump.0 Harpenden, U.K.: Consmer Research Laboratory. Report A9170. -. 1986d. 'Wavin Handpump.' Harpaden, U.K: Consumer Research Laboratory. Report A9178. -. 1986e. ODevelopment Project on Plastics in Below-grund Equipment.' Harpenden, U.K.: Consumer Research Laboratoy. Report A9944 -. 1986f. wDynamisch Dichtencer Kolben.0 Harpenden, U.K.: Consumer Research Laboratory. -. 1986g. OStroemungsbilder im Labyrindhspalt.* Harpene, U.K.: Consumer Reseah Labortory. -. 1987a. 'Pumpenboese Handpump.' Harpede, U.K.: Cnumer Research Laboatoy. Report A9101. -. 1987b. OAbi-ASM Handpump." Harpenden, U.K.: Coumr Reearch Labotry. Report A9127. Amguwu Rupef Sofae 151 -. 1988. "Aquamont Hndpp." Harpeaden, U.K.: Consumer Rsearch Labory. Report A9108. -. 19886. "Dry Bearings at Elovatod Temperatues and Humidity." Harpenden, U.K.: Consumer Research Labotory. Report A9109 -. 1988c. "Adas Copoo 122 and 111 andp. H aden, U.K.: Consumer Resarch Laboratory. Report A9136. -. 198&. *Sealless Pions and Diodic Valves." Harpenden, U.K.: Consumer Reearch Labortory. Report A9202 -. 198&. Lightweght Pump Rods." Harpenden, U.K.: Consumer Research Laboratory. Report A9189. -. 1988f. 'Rising Main and Pump Rod Brealages." Harpenden, U.K.: Consm Rsarch Labotory. Report A9141. =--= 1989a. "Knebel Handpump." Harpenden, U.K.: Consmer Resach Laboratozy. Report A9204. 1989b. "Afridev Handpump." Hapn, U.K.: Consumer Research Laboy. Report A9218. -. 1989c. "Aquadev Handpump. HHapeden, U.K.: Cos Research Laboratory. Report A9242. -. 1990. "uPVC Pipe of Vadable Quality." Hapenden, U.K.: Consume Resarch Labortory. Report A9260 -. 1991a. Rising Main Conectors - Intrim Harpendn, U.K.: Consumer Research Labotoy. Repot" Report A9239. -. 1991b. "Pumprod Connectos - Interim Harpenden, U.K.: Consmer Research Laboatory. Report." Report A92729 Hoehere Technikche Lebranstalt 1985. aKobnumpe " lCjeeup, B. et al. 1989. "Te Taa Handpump - The Birth of a Star." UNDP-World Bank Water and Sataio Program Disui Paper No. 1. Washigton, D.C.: The World BankL 152 CWP.NWed BeA Wowand Sania ion h,Vm Lewis. 1991. 'Investigation into the Quality of uPVC Rising Mains for Handpumps in the Developing World." The Open University CRL reports are available from: Consumer Research Laboratory Hapenden Rise Hapenden Herts ALS 3BJ England Copies of the IAD Handpump Project report are available frm: APP bv Hofsean 14 6955 BA Ellecom The Nethedands Progmm Rewo Seies 153 Ba*. k \Wte and Sanitation, U'NDP-World Bank WAater and Sanitation Program The Wdorld Bank 1818 H Street, NW' Wkashington. DC 2(M33 * USA U~nited Nations Development Programme One Unrited Nat ions Plaza New N'ork. NN 100 17 U'SA Regional Water and Sanitation Groups Regional Water and Sanitation Network Eastern and Southern Africa Central America c/o The W.orld Bank - co UNICEF P.O. Box.30577 P.O. Box 525-01901 Nairohi, Kenva Ciudad de Guatemala' West Africa c/o The W~orld Bank 01 B.P. 1850 Abidjan 01, C6te divoire East Asia and the Pacific lo The )World Bank P.O. Box 324/JKT Jakarta. Indonesia South Asia do The Norld Bank P.O. Box 416 New Delhi 11000.3. India l SA l rnited Nations Dvelopment Pr_gramm