Design aspectsThe three most important variableswere found to be: the type of wood, thedirection of the wood grain, and thetreatment against water.

volkensii), which is rougher and notas hard as olive. He charged only$0.25, but again the poor holes inthe bearings meant they could notbe used.

o In Nairobi, 228km away, the mosthigh-tech solution was found: apump rehabilitation workshop witha professional lathe and precisionmeasuring instruments had suitablewood. They could make dimension-ally perfect bearings, but the pricewas still being negotiated.Care was taken that the producers

used the wood they normally used intheir business, so the choice of woodand the producer were linked.

The woodcarvers were not able tomake a bearing of sufficient precision,and the carpenter would need sometools and training, so if local producersare used, someone would have to beresponsible for control of quality andproduction. This would mean addi-tional management responsibilities forthe project.

Even though local production ofbearings means they are more easilyaccessible to local consumers, the bulkof other spare parts come either fromNairobi or overseas. Therefore localproduction of bearings appears torequire the community to establish newsystems for supply. From a productionpoint of view, the most suitable choicefor a producer would be the workshopin Nairobi which is able to make awhole range of spare parts for handpumps.

last for about nine months, and fit intothe SWN 80 handle. (Figure I showssuch a handle with bearings.)

The research investigated both thedesign and the production process ofthe bearing, beginning with the former.Various local woodcarvers and furni-ture manufacturers were given samplebearings and asked to make copies inwood. In addition a fully equippedworkshop in Nairobi also made andtested prototype bearings from wood.o The village woodcarver used itula

(Commiphora baluensis), which isa softwood, and with his chisel andsandpaper made bearings that wereready to be installed for $0.80 each.

o The woodcarvers' factory 33kmaway used African blackwood (Dal-hergia melanoxylon) and olive(Olea africana), both of which arehardwoods. The woodcarvers couldnot get the bearings quite orthogo-nal, and had particular problemsmaking the holes round and cen-tring them, so they could not beused in a pump. They were also themost expensive option, as it took alot of time to make them. Black-wood bearings cost $1.80 and olive$1.40.

o A carpenter 32km away had asimple electric lathe which he usedto make chair legs. He was shownhow to make the bearings using ameasuring calliper, but he was notable to make good holes with theequipment he had. The hardestwood he used was mukau (Melia

Figure 1. The handle of an SWN 80 hand pump with wooden hearings.

ONE PART OF a handpump that oftencauses problems is the bearing. AnAMREF project in Kibwezi, Kenya,which was using the SWN 80handpump, carried out research intothe possibilities of using wood as abearing material. It should be cheaperthan imported fibre plastic bearings,and able to be produced locally. Thefield study at Kibwezi proved thatthere were problems with using wood,but there were also possibilities.

Wood: A local option for handpump bearingsby Nienke Danhof and Mathieu GielenExperiments using wood to make bearings forpumps has shown that there is some potentialfor replacing costly imported bearings withbearings that have been produced locally.

History and problemsSince the SWN 80 handpump wasdesigned in 1980, various bearingshave been designed for and used withit. The bearing housing is designed forball-bearings, which can last for manyyears but which are very expensive forrural communities: they cost US$12.40each and four are needed for a com-plete set.

Plastic bearings, called LFX andAmite, were made as a cheaper alterna-tive (but they are no longer available).These make a journal bearing, with theplastic turning on a stainless steel shaft(Figure 1). The communities inKibwezi preferred these to the ballbearings because they were cheaper,only $5. More durable bearings werelater introduced, called Tufnol. Theyare made of a high-tech fibre plasticand cost $4.50 each, but field observa-tions showed that they wore out thepump shafts. The pump was rede-signed (and renamed the SWN 90) toinclude major improvements to thebearings, but these cannot be used withSWN 80 handpumps. The bearingsmade of Tufnol and other syntheticplastics cannot be produced withinKenya, so they have to be imported atconsiderable cost.

Eucalyptus wood has already beenused successfully in other parts ofKenya, but it is not available inKibwezi, and our research was to findout if local artisans could make bear-ings with local woods. The eucalyptusbearings are also not very good forpumps with water coming in thepump-head: they are soft and swelleasily. The minimum requirementswere that the bearings be producedlocally, be cheaper than imported ones,

WATERLINES VOL.12 NO.2 OCTOBER 1993 23

Figure 2. Expansion of wood in water (R = radial, T = tangential, L =longitudinal).

Figure 3. The redesigned bearing for the SWN 90 handpump.

them in oil did not lead to lessexpansion. The wood expanded orshrank least in the direction of thegrain. Blackwood expanded least, buteven it expanded by 0.4 to 1.6 per cent.

Many of the woods changed theirproperties after being boiled in oil.Blackwood formed powder on itssurface, indicating disintegration,while a series of small radial cracksappeared in the olive.

These results were confirmed byfield trials. The bearings that wereinstalled in pumps on shallow wells,where water was able to reach thebearings, got stuck because of swel-ling. The bearings did not expandwhen installed in pumps where thehousing remained dry (see box).

Field trials also showed that oil canseriously hamper the functioning of thebearing. Debris from corrosion canstick in the oil, forming a crust on theworking surface of the bearing. Thelubricating effect of oil is also verylimited when there is no full rotationand the speed is low.

Because so many bearings gotstuck in the field trials, only limited

-Mukau

-0- Olive (boiled in oil)

- - Blackwood

-- Blackwood (boiled in oil)

-"- Fresh olive (boiled in oil)

....... Slingna

-.,- Fresh olive

r-·---------lR = Rlldi.1T = TengemialL = Longitudinal

TT

SWN 80 bearings

SWN 90 bearings

ResultsFigure 2 shows the results of soakingthe wood. All types of wood reactedto the water, and boiling or soaking

......l

..................l

/"--".' "/ '.

......................... ·\ ....•... • ... T

.' \ It_-- ------t"

..-... ..-

o ~).; .. ~- .. t!~"tj+H.II+U.}.'!=l\1'2 /l 'iO,-.-.--..-"I\'~ "--'j- .._ ",\ ..---"./' 19

-0.8 \ '\"y__-=--=.':..:=., --L Time (days). .\ \

-1.6 I. ,,-...... \t I '.'.J • '. \

24 • I ',",- . I I' "'. ...---T

\ it' '<.::- ..-- .. _ ..- .../\ l '.... _,_._R'_,./t "'"'-,._1'-~

~ 2.4c.51[

1.6'"UJ

0.8

o Type of wood The hardness, density,and structure of the wood all influ-ence the usefulness of wood as abearing.

o Direction of the wood grains in thebearing There are two ways to cutthe bearing: the axis can be parallelwith or perpendicular to the grain.The direction of the wood will bedifferent for each option. Someliterature was available on theseaspects, but further field trials werecarried out.

o Treatment against water When abearing is located inside the pump-head there is always a risk that itwill get wet, which will causeswelling in a wooden bearing. Thisrisk is increased when the pumpoutlet is blocked or inadequate forthe flow of water. In this studypieces of plain and oil-treated woodwere soaked in water to observetheir expansion. This enabled theusefulness of oil treatment in pre-venting bearings swelling by waterto be determined. Motor oil wasused because it is cheap, heatresistant, and widely available.

b

1\

Figure 4. The grain in the bearing can be parallel to the axis of the bearing (a), or perpendicular to it (b).

24 WATERLINES VOL.12 NO.2 OCTOBER 1993

Why itula failedBearings made of the softwood itula were installed in a pump with a dryhead and did not swell. The grain in these bearings was perpendicular tothe axis of the bearings. As the wood dried out it lost its flexibility,however, and the load on it then caused cracks along the grain, as is shownin Figure 5.

This shows that the direction of the wood was poorly chosen, and itappears that the same bearing, only installed in the pump slightly rotated,could have made a stronger bearing. There is a big difference in the strengthdepending on to which side of the wood grains it applies. It is howeververy difficult to make sure that in every well the bearings are thuspositioned that the strongest side is loaded.

If the bearing is made with the wood grain going in the other direction,with the axis of the bearing parallel to the wood grains, the wood does notshow so much difference in strength.

lOW TECHNOLOGY,WATER WEll DRilliNG

SPECIALIST

information was obtained about thewear. hula, the softest of all woods,showed considerable wear after onlyfive weeks. It is estimated that thesebearings would only last for threemonths.

Conclusionso All types of wood which were tested

expand when wet and boiling orsoaking in oil does not prevent this.

o Oil has unwanted side-effects, suchas weakening the wood and making

debris stick to the bearing. Oil isnot an effective lubricator in ahandpump bearing.

o The direction of the grain in awooden bearing should be parallelto the axis of the bearing.

o Of all the researched woods black-wood expanded least in water (ra-dial expansion 0.4-1.6 per cent,longitudinal shrinkage 0.9 per cent).The following alternatives should

be considered when wood is used forhandpump bearings and when there isa risk of water reaching the bearings:

* MACHINES!

* MATERIALS!

* TRAINING!

Figure 5./tula bearings, cracked in thedirection of the grain.

o Bearing with a bigger holeIf the hole in the bearing is bigger,the bearing can swell without get-ting stuck to the shaft. If blackwood,the wood which swells least, isused, the hole needs to be 31.4mmin diameter instead of the usual30.2mm. If the wood does not swell,the bearing will be worn out muchearlier! The life span of this largerdiameter bearing still needs to befield-tested.

o Design changes to the pump-standChanges to the pump-stand can bemade to prevent water from reach-ing the bearings. This alternative isonly possible for new pump-stands,and does not solve the problem ofexisting wells. Operators shouldalso ensure that the correct sizecylinder has been chosen.

o Better waterproofing of woodOther methods of protecting wood,for example paint, coating, or im-pregnation, might be better thansoaking in oil. More research isnecessary to find out if other meth-ods protect the wood better andprevent it from soaking up water.If after trying these suggestions the

wooden bearings are still not workingsatisfactorily, synthetic materials willhave to be used.

So the best option for now for awooden bearing is to have it made ina workshop with a professional latheout of blackwood, and with a biggerhole. Field tests must demonstrate itslife span .•

Contact us to request comprehensive brochure.

Eureka UK Ltd., 11 The Quadrant, Hassocks, West Sussex, BNG8BP, EnglandTelephone: 0273 844333/Fax: 0273846332

WATERLINES VOL. 12 NO.2 OCTOBER 1993

Nienke Danhof and Mathieu Gielen are fifth-year students at Delft University of Technology,Faculty of Industrial Design Engineering, il'.J.F.M. Molenbroek. Jaffa/aan 9, 2628 BX Delft.The Netherlands. This research was carried outin co-operation with AMREF-Kenya, G1Z-Kenya. Delft University of Technology. andCICAT.

25