On Mortarless Masonry Walling - Welcome to the … (EWP IIB-4) On Mortarless Masonry R11.docx 23/09/2016 1 Energy & Low-income Tropical Housing On Mortarless Masonry Walling Terry

W01(EWPIIB-4)OnMortarlessMasonryR11.docx 23/09/2016

1

Energy&Low-incomeTropicalHousing

OnMortarlessMasonryWallingTerryThomas–WarwickUniversity2016(originatingasELITHPaperEWPIIB-4)

"This document is an output from a project co-funded by UK aid from the UK Department for International Development (DFID), the Engineering & Physical Science Research Council (EPSRC) and the Department for Energy & Climate Change (DECC), for the benefit of developing countries. The views expressed are not necessarily those of DFID, EPSRC or DECC."

AbstractMortarlessmasonrywallingispotentiallycheaperandlessenergy-intensivethanmortaredwalling and is in use for low-income tropical housing. The saving is becausemortar costsmore,andentailsmoreCO2emissions,perlitrethandomasonryunits(blocksorbricks)andits absence reduces the labour cost of laying. These savings, when using (interlocked)stabilised-soil blocks, lie in the range 10-30%. However mortarless walling has inferiorperformancewithrespecttostraightness,crushingstrength,resistancetolateralloadsandlateral stiffness. Moreover to provide a fully sealed building envelope, a mortarless wallneedsplasteringonatleastoneside.Toachievesatisfactorystraightnessovera2.5meterheight, block contact surfaces should be parallel within 0.5 degrees, be cleaned and begrooved to prevent contact close to their centre line. Unless blocks are very precise, theinterlockpatternshouldpermitblock-reversalduringconstruction.Crushingstrengthisless(typicallybyafactorof5)andlateralstiffnessismuchlessthanforamortaredwall:inbothcases performance is much improved by the same measures as improve straightness,because thesealso increase theeffectiveblock-to-blockcontactarea fromtypically5% to25% of block plan area.Mortarless walling is quite widely used in Africa, but even withaccurate motorised presses, blocks cannot be stackedmore than 3 meters high withoutinsertionofmortaredcoursesorringbeams.

Contents1. Purposeofthisarticle2. Masonryandmortar3. Advantagesofmortarlessmasonry4. Thecurrentpracticeofmortarlessmasonry5. Interlocksandresistancetopunch-through6. Achievinggoodverticalalignment7. Achievingverticalstrength8. Achievinglateralstiffnessandstrength9. Plasteringandpartialmortaring10. Thegeometricaccuracyofmasonryelements11. High-techversuslow-techversionsofmortarlessmasonry.12. Conclusions:Thelikelyfutureofmortarlessconstruction

AppendixA Relationshipbetweenout-of-plumb(overhang)xofamasonrycolumnandtheroll-wedgeangleθofitsconstituentblocks

AppendixB Behaviourofanunmortaredcolumnunderlateralforce


2

1. Purposeofthisarticle

Toidentifytheadvantagesandproblemsassociatedwithmortarlessbrickorblockwallingintropicalcountriesandtoidentifyitsscopeforreducingthecarbonfootprintandcostoflow-incometropicalhousing.

2. Masonryandmortar

‘Masonry’-buildingwithsmall,easilyhandledandinterchangeableunits–suchasbricksorblocks–has severaladvantages.No large liftinggear,nor complex shutteringor formwork, isneeded.Theindividualunitscaneasilybegiventreatmentspriortotheirassembly,suchasfiringorcompression,toenhancetheirstrengthordurability, treatmentshardlypossiblewithawall formed insitu fromraw materials. Interesting patterns can be created using varied bricks styles. Masonry is usuallyassembledwiththeaidofmortar.

Thereishowever,inmanydevelopingcountries,thepracticeofusingmortarlessmasonry-drivenbythebeliefthatomittingmortarinwallingreducesbothmaterialandlabourcosts.Theomissionofmortarhoweverreducesperformanceinvariousways.Thepurposeofthisarticle is to explore how real are the supposed savings and how easily that reduction inperformancecanbe toleratedorcanbeamelioratedbyother techniques thanmortaring.Theadvantagesofmasonryitself–i.e.ofassemblingfairlysmallunitsratherthanforminglargeunitsonsiteorinfactory–willhowevernotbeaddressed.

Amortarmix – of suitable performance – is generallymore expensive per litre than themasonry units it connects,whether the units bemud blocks, fired bricks, hollow cementblocksorstone.(Throughtherestofthisarticletheword‘block’willbenormallyusedforamasonryunitregardlessofthatunit’ssizeorcomposition:thus ‘block’ includesbothbrickandcut stone.) So thematerials fordry (=mortarless)assemblywillusuallycost lessperunitofwallvolumethanthematerialsofamortaredwall.Thecheapertheblocks,themoremarked the savings, so cement-mortar is least used in mud building or for field wallsassembled from field-gathered uncut stones. In both cases the cheapness of the blocksallows the use of big wall thicknesses andmaybe a marked wall taper. Block-laying is ahighly developed skill in some cultures andmortaring is done at considerable speed. Butthere are scenarios such as self-built and community-built housingwheremortaring skillsand speeds will be low: here there may be substantial time savings from employingmortarlessconstruction.

Table1liststhefunctionsthatmortarnormallyperformsinmasonry.

Table1 Theprimaryfunctionsofmortarinmasonrywalling

i Providingadhesioninshear,sopreventingdeliberateoraccidental‘punch-through’ofindividualunits

ii Allowing adjustment during assembly so that walls can be built plumb, and withstraightcourses,despiteanyirregularityintheblocks

iii Cushioning vertical loads, spreading inter-block contact over a greater area and


3

thereby achieve higher crushing strength, Euler buckling strength and lateralstiffness

iv Sealing the building envelope against penetration of light, sound, dust, wind orwater

v Combattingverminandplantgrowth

vi Improvingtheappearanceofwalling

These then are the performance features thatmortarlesswallingmust achieve by ‘othermeans’orwhichmustbesacrificed.

Some alternatives tomortaring, several of which are discussed in the following sections,include:

• Interlockingtorestrictrelativemovementofunitsperpendiculartothewallface(andinsomecasesmovementalongthecourse)

• Usingbricksofsuchaccuracythattheairgapsbetweenthemareverynarrowandtheir ‘dry’assemblyproduceswallsofadequateverticality.

• Injectingaveryfinemortarsubstituteafterassemblingthebrick/blocks.

• Usingathinsurfacerenderand/orinternalplastertoholdthewalltogetherandwind-sealanygaps.

• Designing the brick so its connection to the bricks below and above is so restricted that‘rocking’isprevented

• Having the bricklayer place each brick/block speculatively firstly direct and then reversed,thenchoosingtheorientationgivingthestraighterwall.

• Periodicallyinsertingamortarlayer,lintelorwall-beamto‘reset’thewallstraightness.

3. Advantagesofmortarlessmasonry

Twoprincipalsavingsfromomittingmortarwerelistedabove–cheapermaterialsandfasterassembly.Therearealsosomeminorpossiblesavings–forexampletheomissionofmortarin theperpendsof engineeringbricks allows their use fordamp-courses.Where lime-richmortarsareerodedbysnailsoratmosphericacids,theiromissionmaysignificantlyreducelong-termmaintenancecosts.

Consider twowalls of identical thickness and face area,made of blockswhose individualfaces measure L x H, one wall without mortar and the other with either 20mm ofcementitiousmortar(“thick”)or8mm(“thin”).8to25mmrepresentsthetypicalrangeforAfricanhousing, inEuropetheupperbound isonlyabout16mm.Thevolumetricunit-costratio of mortar material to blockmaterial we can call ‘μ’. μ=2 is representative of mortarcombinedwith hollow cement blocks; μ=5 ofmortar combinedwith stabilised soil blocks orwithlow-quality fired ‘country’bricks; μ=10ofmortaredbutunstabilisedpressed-earthblocks.Table2belowshowsarangeoftypicalcostsavingsfor300mmblocksand200mmbricks.Inthetable,thefractionofthewallthatismortarrangesfrom12%to33%.(Inthefield,fractionsashighas


4

50%maybeobservedine.g.Uganda.) Ineverycasethesaving issignificantenoughtobeworthpursuingandinafewcasesthecostofthecement inthemortartotallydominatesoverallwallingcosts.Costreductionisalsocorrelatedwithreductionin‘embodiedcarbon’-a label for greenhouse-gas emissions before and during construction – because theproductionofcementisemissions-intensive.

Neglecting labour-time savings, the fractional cost savings due to omittingmortar from brick (orblock)wallingequals

Sc=(costperunitofwallvolumewithmortar–costwithoutmortar)/costwithmortar

=λm(μ-1)/(1+λm(μ-1))

whereμisdefinedaboveandfractionλmofthewallfaceareaismortar.λmdependsuponthebrick-mortargeometry.Itcanbeexpressedas

λm=Am/(Am+Ab)whereAmandAbarewall-faceareasofmortarandbrickrespectively

Am=(L+h+t).tandAb=L.h;wheretismortarthickness,Landharethefacelengthandheightofthebrickface.

Table2 MaterialscostsavingsbyomittingmortarThesesavingsareindependentofwallthickness,whichwouldtypicallybe150mmforawallofblocks,100mmforasinglestretcher-bond,fired-brickwalland200mmforheader-bondordoublestretcher-bondbrick.

Blocklength

Blockheight

Mortarthickness

Mortarfractionof

Assumedvolumetriccostratio

Wallcostratio

Wallcostsavingif

L H t wallvol’m mortar:block mortared: nomortar

mm Mm mm λ μ unmortared %

300 100 thick(25) 0.26 2 1.26 21

300 100 thin(10) 0.12 2 1.12 11

200 75 thick 0.33 2 1.33 25

200 75 thin 0.16 2 1.16 14

300 100 thick 0.26 5 2.05 51

300 100 thin 0.12 5 1.48 32

200 75 thick 0.33 5 2.33 57

200 75 thin 0.16 5 1.64 39

300 100 thick 0.26 10 3.35 70

300 100 thin 0.12 10 2.08 52

200 75 thick 0.33 10 4.00 75

200 75 thin 0.16 10 2.44 59


5

The table immediately suggests that mortar should be kept thin. In practice there areseveral reasons for not doing so: these include the irregularity of cheap bricks, problemswiththepoorworkabilityofthinmortarandlackofbricklayingskills.

Anextremeexampleofpossiblesavings isstandardUgandanheader-bond200mm-thickwallingofclamp-fired ‘country’brick forwhichmortar thickness is typically30mm,and theunit cost ratioμ(mortar:brick) is about4.Omittingmortarwould saveabout57%ofmaterials cost (or52% in thecaseofstretcherbond).Thesehugesavingsarehoweverunattainableastheirregularityof‘countrybricks’rulesoutinterlockedmortarlessconstruction.Morefeasibleisthe20%savingobtainablebyreducingmortarthicknessfrom30mmto10mm,whichmayinturnrequirehigherqualitycontrolsin clamp-firing the bricks. In this article wewill henceforth focus onmasonry that uses not firedbricksbutstabilised-soilpressedblocks.

Usingpressedstabilised-soil300mmblocks (sand:cementratio=12)and15mmofmortarcosting,perlitre,3timeshigherthantheblocks(i.e.μ=3),thematerialscostsavingwouldbeabout21%.Thissubstantialfiguremakesmortarlesswallingworthinvestigating.TheinterlocksusuallymouldedintoSSblocks intendedformortarless layingdonotaffectthematerialscost-savingandhavelittleeffectonmouldingtimes.

The monetary value of time saving by omitting mortar varies greatly from country tocountry.Table3comparesUKwithTanzania.

Table3 Timeandcostsavingsbyomittingmortarinwallconstruction

Scenario Mortarused

Unitsperm2

Layingspeed

Wagerate

Wage Savinginwages

m2/hour £+/hour £+/m2 %

UKbricklayer Yes 56 1.1 20* 18.1

“ No 66 6.0 13 2.2 88

Tanzanianself-build(blocks) Yes 29 2 0.4** 0.20

““ No 33 5 0.2 0.04 80

Tanzanianfundi(blocks) Yes 29 4 0.6* 0.15

““ No 33 12 0.4 0.033 80

*Includingmatetomixandfeed. **Forself-buildthisisanopportunitycost.+BasedonTZS3000=£1.

Buildingwithcompressedbutunstabilisedearthblocks isaspecialcase.The‘mortar’maybecement-basedormaynot.Inthelattercase(‘mudmortar’)thematerialscostdifferencebetween ‘mortared’ and ‘unmortared’ construction is likely to be negligible and the onlyextra cost of mortaring would be for labour. Under these circumstances there is littlebenefitinomittingmortarandincurringallthedisadvantagesimpliedinTable1.

4. Thecurrentpracticeofmortarlessmasonry

Stone masonry uses rocks either as found, or (ashlar) as found/quarried and then cut.Sometimesnomortarisusedintheassemblyofashlarwallsbutthisincreasestheaccuracy


6

ofcuttingrequired–anditscost.Mortarlessashlarisgenerallytoocostlyforhousingexceptforspecial‘feature’sections.

Figure1Drystoneandashlarwalling.

Whererocksarefromlayeredstrataandthereforenotrounded,theymaybeheapedfairlystablywithoutmortarintowallshavingatrapezoidalcross-section.Thistechniqueisusedtoconstruct‘drystone’boundarywallstofieldsinmountainousdistricts.Suchwallsarethickandsoconsumemuch rockwhich ishoweveroften readilyavailable from field-clearance.They are easily dislodged, require considerable maintenance and are for many reasonsunsuitedtohouseconstruction.

Mouldedearthblockshavetraditionallyconstitutedmuchwallingroundtheworld.Inorderto get reasonable stability and resistance to surface erosion, the blocks are usually ofcompressedmoistsoil,thecompressionbeingimpulsiveorofupto1MPapressureslowlyapplied. Such blocks have no resistance to standing long in water and therefore earthbuildingsrequireaflood-protectedenvironmentandagooddampcourse.Havinglongbeenregardedasprimitive,earthbuildingsofvarioustypesareslowlyreturningtopopularityandarebeingpromotedbysuchagenciesasCRATerrecraterre@grenoble.archi.frinFranceandproducerassociations in Europe and the Americas. Some ‘welding’ of the surfaces of contiguousblockscanbeengenderedbymoistening them immediatelyprior to laying. Evenpressedblocksmadewithawell-chosensoilmixhavepoorerosion resistance,aredustyandmayharbour vermin (triatomin bugs, cockroaches, mice etc), so mortarless soil-block wallingneeds either a surface seal or periodic repair. The new attraction of all forms of earthbuildingisitsverylowcarbonfootprint.

Themostcommonformofmortarlessmasonryemployspressed,interlocking,stabilised-soilblocks,‘ISSBs’,whicharetypically300mmlong,150mmthickand100mmhigh.Machinestoproduce such blocks are available in many countries. Motorised presses give higher


7

pressures, more accurate blocks and somewhat higher labour productivity than manualpressesbutaretooexpensiveforartisanalbuilders.

,

Figure2aManual(Makiga)andMotorised(Hydraform)block-makingpresses

(ii)Tanzanian(NHBRA)manualblock (ii)HydraformBlock

Figure2bInterlockdesigns

A150mmwallthickness,andhenceblock-thickness,isthatwhichgivesjustadequatelateralstiffnessduringconstruction(afterwhicharingbeamisaddedtostabilisethetopofawall).Thecement(oroccasionallylime)contentoftheproductionmixis low,e.g.under8%andtheblockhollownessupto25%.Thesoilmixmustcontainsomeclay(forcohesionduringproduction) andahighmouldingpressuremustbe supplied (1 to 10MPa). This pressurestrengthensthematerialandpermitsafairlydrymixtobeused,withawater:cementrationclosetotheidealof0.5.(Adrymixislessvulnerabletounevendrying-shrinkagethanawetone.)Formationpressuresofgreaterthan10MPamayhoweverdriveoutsomuchmoistureandcementfinesthattheinteriorofblocksiseffectivelyunstabilisedandthereforeweak.

Plan

FrontView


8

Table3Wetcompressivestrengthaftercuring(σcinMPa)vmouldingpressure&cementcontent;fromGooding1994.(Figuresinbracketsshowstrengthincreaseratiooverdatum)

Formationpressure→

Cementfraction↓

MPa 1.25(datumpressure)

2.5 5 10

5% σc=0.86 1.02(1.19) 1.22(1.42) 1.60(1.86)

9% σc=1.80 2.20(1.22) 2.73(1.51) 3.40(1.89)Forconstantcementcontent,eachdoublingincompactionpressureproducesabouta23%increase in cured compressive strength (as shown by the ratios in brackets). The sameincrease could alternatively be achieved by increasing absolute cement content by about1%.Thereisthus,foragivenblockstrength,theoptionoftradingextramouldingpressureagainstextracement.

Pressingintoamould,ratherthanextrusion,isinvariablyused.Themouldthereforeinitiallydefines some dimensions of the blocks and these can be made closely repeatable.Unfortunately this definition is least reliable on the critical top and bottom surfaces of ablock.Moreoverduringhandlingandcuring,blocksareoftenlocallydamaged,ordistorteddue to uneven curing and subsequent drying, or acquire small blemisheswhich preventsthemseatingproperlywhenlaidontheblocksbelow.Goodon-sitemanualpresses–usuallybasedon the1950sColombian ‘CinvaRam’andhavingonemovingpiston - cangenerateabout 1.5MPa pressure. Expensivemotorised presses in block-yards or on large housingsitesoftenusetwoopposinghydraulicpistonsandexertamouldingpressureupto15MPa.Block replicability – which is especially important in mortarless construction – can beimprovedbyweight-batchingofthesoilusedforeachblock.Controlofmouldingpressureisachievedbymonitoringleverforce(‘effort’)inthecaseofmanualpressingandbyhydraulicpressureinthecaseofmechanicalpressing.

Someblock-setsarejustlaterallyconstrained:theirinterlockfeaturespreventablockbeingeasilyknockedthroughitswall,asthatwouldentailinterlockprotrusionsbeingshearedoff,but sliding along a course of bricks is not constrained. Such interlocking is commonlyachievedbyhavingthetopandbottomfacesofblockslongitudinallygroovedandridgedtoforma‘tongueandgroove’joint,asinFig2b(ii)above.Theymayinadditionhaveanend-interlocktoclosetheperpendwhentheyarelaid.Thewidely-marketed‘Hydraform’block,madewith a 10MPamechanised press is of this type. Unfortunately such blocks do notformneatcornersorteejoints,norcantheybe‘reversed’duringlaying

Other block sets, e.g. as illustrated by Fig 2a, are both laterally and longitudinallyconstrained. Neat strong corners/quoins may be formed and blocks may be reversed,


9

however the perpends are not fully closed. From inside a building daylight chinks can beseenthroughthewalls.

Withmortaredbrickwork,on-sitecutting isemployed toproduce thequarterbricks,half-bricks,three-quarterbricksandclosersnecessarytocreatecorners,jointsandstraight-sidedopenings.Site-cutbricksoften lookrough,but this roughnesscanbe largelyconcealedbyapplication of mortar.With mortarless construction it is more desirable to avoid on-sitecuttingandinsteadtomanufacturehalfbricksandthree-quarterbricksaswellasfullbricks.Mouldsmay be adapted so that divider plates can be inserted to create half-bricks. It isgeneral with mortarless masonry to dimension the width of wall-sections as integermultiplesofastandardblock-lengthLorofL/2.

Figure4 3-bedroomhouseinTanzania,oneofhundredsmadeofmanuallypressed,unmortared,stabilised-soilblocks

.

5. Interlocksandresistancetopunch-through

‘Punch-through’isthepenetrationofawallbydislodginganindividualblock.Itcanhappenby accident (nailing up a shelf, a lurching cow, a colliding car) or as ameans of criminal‘breakingandentering’.The force required topunch-througha singleblock isusually lessthan that to create a larger hole spanning several blocks, so our primary interest is toidentifyandincreasetheformer.

Mortarbetweentwomasonrycoursesactspartlyasanadhesiveabletoresistshearfailure.Itseffectivenesscanbe increasedbysuchpartial interlock featuresas the ‘frogs’ (mortar-locating indentations) moulded into the top of some fired bricks. Widely-used hollow‘cement’ blocks are commonly laid so that their cavities act as such frogs but on their


10

undersides. There is a tradition in e.g. parts of Latin America and Asia of incorporatingbarbed wire within mortar layers to better resist deliberate punch through. In Tanzaniaillegalentryusingabatteringramhasitsownterm(‘faduma’).Inmostcountrieshowever,forced(criminal)entryisusuallyviawindowsordoorsratherthanthroughmasonry.Withoutmortarorinterlockfeatures,theprimaryresistancetoshearfailureinawallisthefrictioncreatedbytheverticalpressureacrossahorizontaljoint.Thecoefficientoffrictionbetweenblocksorbricksistypicallyabout0.7http://www.supercivilcd.com/FRICTION.htmandtheverticalpressureinasingle-storeywallrangesfromnearzeroatthetopcoursetoabout50kPaatthe lowest course, giving a shear strength in the range 0 to 35 kPa. Relying only on thisfrictionis likelytoresult inpunch-throughwhenlargeout-of-planeforcesareappliedhighup in a wall. Since walling is quite rigid, such large forces can readily be generated viaimpulsive blows. In consequence of this vulnerability, most mortarless walling systemsemployinterlockingblocks,whosetopandbottomsurfacescarrymatchingprotrusionsanddepressions.

Table410desirablefeaturesthatinterlocksshouldallowa Clearancesufficienttoaccommodateproductiontolerances,butkeptlowenoughthat

theinterlockcanbereliedontocorrectly(withinsay1mm)locateablockrelativetotheblocksbelowitinthedirectionperpendiculartothewallface.Some‘double-constraint’interlockdesignsalsolocateblockscorrectlyalongacourse;

b Nointrusionintotheblock-to-blockbearingsurfaces(whichshouldideallybenearthefrontandtothebackofeachblock.(Actually–asdiscussedlater-itisusefuliftheinterlockactuallypreventsblock-to-blockcontactclosetotheblocks’centrelines.);

c Adequateverticaloverlapandverticalcontactareasuchthattopunchablockthroughthewallentailsshearingoffasignificantareaofinterlockmaterial;

d Nolight,ormuchwindorsound,allowedtopassthroughthewall;

e Productionofneat strongcorners,openingsand tee jointswithouton-sitecuttingofblocks(thismayneedtheblock-settoinclude½blocks,¾blocks,corners);

f Some– sayat least10% -blockhollowness, locatedclose to theblockcentreline, tosavematerial;

g Creation of vertical passages through blocks to permit electrical wiring orreinforcementbarstobeplacedwithinwalls;

h Relativerotationofadjacentblocksalongacoursebyat least50andpreferably80toenabletheconstructionofcurvedwalls

i Blockreversal(viasuitablesymmetry)asthismayaidproductionofplumbwalls;

j Productionbyextrusion.

Features(a)to(c)wemayconsideressential,(d)to(j)‘desirable’.Nocurrent,commerciallyavailable, block-set geometries satisfy all these desiderata. Requirement (d) = sealedenvelope can be obtained by other means – for example by plastering the wall.Requirements (d) and (e) are normally in conflict with each other, as are (e) and (j).


11

Incorporationof ‘block-end’ interlocksenhances (d)butconflictswith (i). Indeed is lackoffeature (v) – e.g. ability to formneat strong corners – that is themostobviousdefect ofsomeblock-sets.To date, suitability for extrusion (j), has not been considered necessary as interlockingblockshaveneverbeenextruded,evenwhentheinterlockdesignwouldpermitit.

Todatethemortarlesslayingof215mmfiredbricksisveryrarelypracticed,althoughTable2 indicates that thematerials advantages ofmortarless assemblywould be greatestwiththese smaller units. Clay bricks can be extruded with a tolerance of 0.1mm, but thesubsequent operations of drying and firing cause additional size-variations that result inmanufacturingheight tolerances increasingtoat least1mm.Bricksare intendedforbeingplacedwithonehandandthereforeareweightlimited.InUKastandardbrick(65x102.5x215mm)occupies1.4litres,muchlessthanatypical‘cement’orstabilised-soilblockwhichoccupies 3.5 to 4.5 litres. The number of bricks required per square meter of walling istypically60,increasingto72intheabsenceofmortar;thecorrespondingblockcountis30increasingto33.

Lackofnegativefieldreportssuggestthatcurrent ISSB interlocksareadequatetoprevent‘punch-through’.However little numerical data is available to compare the forces (or theenergyinblows)neededtopunchthrough(a) interlockedunmortaredblocks,(b)smooth-topped unmortared blocks and (c) smooth-topped mortared blocks of the same size.Includinginterlockfeaturesinblockdesignincursonlyaminorproductioncost;howeverifdone as badly as it is in a few commercial cases, the interlock itself may interfere withaccurateassembly.MoreoverthepenultimateofthedesideratalistedinTable4–namely(ix)blockreversibility-whichisnotavailablewhenusingsomewidelyusedblockpresses–is significant for achieving verticality (plumb) of walls as discussed below. The lack ofdesideratum (x) – compatibility with block formation by extrusion – is likely to seriouslydiscouragetheapplicationofmortarlessnesstofired-brickwalling.AnidealblockformTable4listsdesirablefeaturesoftheblock-to-blockinterlockandhenceoftheblockform.Fig2bshowstwoblockforms,eachwithitsownadvantages.Bothvariants(i)and(ii)satisfycriterion(a)inTable4;bothfailthecurvedwallcriterion(h)andextrudabilitycriterion(j).Theridgedformof2b(ii) ,with itsadditionalend interlocksatisfiescriteria (b), (c)and(d)whereastheformof2b(i)satisfiescriteria(e),(f),(g),and(i).Itwillnotbepossibletocombinethemeritsofeachformifwearelimitedtoasingleblockshape,soweshouldbeconsideringblockSETscomprisingmultipleshapes.Takingthe‘tongue-and-groove’form2b(ii)asastartingpointweneedto

• Create½blocksand¾blocksforedgingopeningsandjoiningcrosswalls


12

• Createacornerunitthatwillgiveafullcornerinterlockwithnovoids–forexamplea block that for half its length has a longitudinal tongue-and-groove and for itssecondhalfhasalateraltongue-and-groove.

• Removetheendinterlockorreplaceitbyasymmetricform(suchasan‘S’shapedend) so that only one version of ½ and ¾ and corner blocks are needed and inaddition the reversibility criterion (i) is satisfied at least for full blocks. Notehoweverthatwithoutendinterlockstheperpendsarenolongercompletelysealedagainstlightorwindpenetration.Thismaybeacceptableormayrequireperpendpointing. Ifblocksareveryaccurate, reversalduringassembly isnotrequiredandcriterion(i)nolongerapplies.

• Create two symmetric voids (e.g. with centres at ¼ and ¾ of the block length)withinthetongueandgrooveswatheinordertosatisfythematerialsavingcriteria(f)andthepresenceofverticalpassagesforwiringorreinforcement(g).Thecrosssectionalareaofeachpassageishoweverunlikelytoexceed5%oftheblock’splanareasomaximummaterialsavingis10%.

We now have a block SET that largely satisfies all criteria except rotatability (h) andextrudability(j).

6. Achievinggoodverticalalignment(SeealsoAppendixA)

A column assembled from unmortared and inaccurate blockswill rapidly become out-of-plumbasitrises.Intheworstcasesthetopofcolumnsonly10blockshighwilloverhangbyso much that they topple. This is clearly an unacceptable failing. Ideally the (standarddeviationof the)overhangX of a single-storey (e.g. 2.5meter)wall shouldbeacceptablysmall–thisheightcorrespondsto25coursesof100mm-highblocks.Howeverundersomecircumstancesawall’sstraightnesscanbereseteverysay8coursesby introductionofanintermittentmortarlayeratforexamplewindowsillandlintelheights.Ring-beamsandfloorslabs are commonlyused in conjunctionwithmortarlessmasonrywalls. These effectively‘reset’ a wall’s straightness at each floor, so the inability of such walls to keep withinoverhanglimitsovertwostoreysisgenerallynotaproblem.FortunatelyISSBwallsleanlessthancolumnsdo,andsomearesoflexiblethattheycanbepulledintoverticalitybeforethecastingofaringbeamontopofthem.Nevertheless,restrictingoverhangtoanacceptablevalueisthemostchallengingoftheproblemsfacingthebuilderofamortarlesswall.

Block imperfections includevariation inheight, lateral taper, longitudinal taper, twist andbowing. These are caused by variability in themoulding process and by block distortionduringdrying, curingor firing.Wall imperfections that followare lackof ‘plumb’, rocking,wavycourses,unevenstressesleadingtostressamplificationandlossofcrushingstrength,reduced Euler-load capacity, much reduced stiffness etc. We might distinguish betweensmall-scale‘blemishes’thatreducetheeffectivecontactareabetweenadjacentblocks,andoverall geometric inaccuracy. To completely describe such inaccuracy would require aninfinityofdata,whereasanypracticalclassificationofthegeometric‘quality’oraccuracyof


13

ablockmustemployonlyafewdata.Asthefrontandrearofablockarenotnormally incontactwithotherblocks, their imperfection israrelyof importance– it is thematingtopandbottomsurfacesthatmatter.

Fig6 Simplifieddescriptionofaparallelepipedblock

Plan Endelevation Sideelevation

Gapgbetweenundersideandareferencesurfacee.g.glassplate

Inpracticetheblockheightsandthegapsbetweentheblock’sundersideandareferenceplane will vary slightly at the different plan-points A to F. A perfect prism would haveuniformheight (hA=hB=hC=hD=hF=hG=h0,whereh0 is thenominalheightandheightvariationσh=0),equalgaps(gA=gB=gC=gD=gF=gGandσg=0)andrightanglesbetweenfaces(α=β=γ=δ=900).

Fromthepointofviewofassemblingblocks(onalevelbase)intoaverticalwall,wewantthetopandbottombearingfacesoftheblockstobeparallel(α+β=1800);howeverwehaveonlyaweakrequirementforeitherαorβtobecloseto900(e.g.87o<α<930wouldsuffice)Ourbestestimateofthe‘out-of-parallel’angleofthesebearingfacesis

‘Rolltaperangle’θ=α+β-1800=((hA+hB+hC)-(hD+hF+hG))/3W(inradians)

whosecalculationthereforerequiresthemeasurementof7datapoints.

However even this measure gives an under-estimate of the imperfection in roll becauseneithertheblockbeingconsidered,northatsupposedlymatingwithit,haveanabsolutelyplanetopandbottomsurfaces.Theremaybebowing/hogging(byadistanceb)andtwisting(byanangleφ) thatdonot showupasdifferences in theblockheightshA tohG. For thebottomfacethesemaybeestimatedasrespectively

bowing: b=(gB+gF)/2–(gA+gD+gF+gG)/4

twist: φ=(gA+gD-gC-gG)/2LwhereListheblocklength.

Themain impact of bowing and twist is to increase the instability of thewall units – asmanifestbyrockingduringassembly.

AppendixAcontainsanalysisthatrelatesthevariabilityofacolumn’soverhang(i.e.out-of-plumb)distanceXtothevariabilityoftheblocks’smallbutrandomroll-taperangleθandtocolumnheightH.Howevertheanalysisneglectsbowingandtwist,sothatitunderestimateslikely overhang X. It predicts that the standard deviation σx of X is proportional to the

Heighthbetweenbearingfaces

αγ

βδ

ABCWidthWFED


14

standarddeviationσθoftheblocks’roll-taperangleθ(providedθhasazeromean)andtoH1.5.(SeeEqA2b:σX≈0.58H1.5σθ/h0.5).

Severalexperimentshavebeenundertakentotestthisrelationship.

2016experimentsatNHBRA,DaresSalaam,140poorqualityISSBblocksweremeasuredtoestablish theirSD of roll-taper angle asbeingσθ = 0.0102 radians and theirmean roll-taperangleasmθ=0.0013 radians. The latterwas reduced to .0005 radiansby reversingalternateblockswhichoperationhoweverhadnegligibleeffectontheirSD(σθ).1.5m(i.e.15-course) columnswerebuiltwiththeseblocksandtheiroverhangsXweremeasuredtoyieldthecolumnstatistics(SD=σXandmean=mXofoverhang)below:• for20columnsbuiltwithrawblocks(butbrushed):σX=22mm, mX=83mm• for30columnsbuiltwithrawblocks(50%reversed):σXR=19mm,mXR=14mm

andforthebestcase: • 30columnswithblocksreversed&grooved σXG=14mm,mXG=8mm

Sofor50%reversing,σXR/σX=0.86;foralsogrooving,σXG/σX=0.64for50%reversing,mXR/mX=0.17;foralsogrooving,mXG/mX=0.10Thusreversingandgroovingmarkedlyreducethemeanoverhangbutonlymodestlyreducethe SDof overhand.Groovingwas to prevent blocks contacting each other close to theirlongitudinalcentrelines.Reversingisonlypossiblewithsomeinterlockdesigns.

Note that even the best case above (blocks 50% reversed and grooved) was barelysatisfactory,astheoverhangreached50mminsomecolumns.Extrapolatingthe limit (X<9mm for awall ofH = 2m)mentioned in Appendix A, wemight adopt the acceptabilitycriterion: σXshouldbelessthansay6mmforanH=2.5mcolumn.In that case these columns are unacceptable even when made of grooved bricks. Wallswouldhoweverhavea loweroverhangvariability (σX) thancolumns.Somewillarguethatforlow-cost,low-rise,tropicalhousing,wall-accuracystandardscouldberelaxed.

Theory(usingEqA2b)–basedonassumingplanartopandbottomsurfacesand(usingthemeasuredblockdataσθ=0.0102rad,H=1.5m,h=0.1mandassumingmθ=0)predictedahighσXGvalueof34mm.

2009ExperimentsatURDTCampus,Kagadi,UgandabyWarwickUniversityStudents:10-blockcolumns (i.e.1.0mhigh)ofpoorquality ISSBblocks (each300x150x100mmhigh)wereassembledandtheiroverhangsmeasured.Threevariantsofassemblywerecompared.• for40columnsbuiltwithrawblocks, σX=30.5mm,mX=6.9mm• for40columnsbuiltwithgroovedblocks, σXG=25.7mm,mXG=8.5mm

Howeverwhen blockswere ‘bespoke reversed’ so as to best fit a builder’s level test forverticality• σXL=3.6mm, mXL=1.6mmfor10columnsbuilt


15

Soforgroovingalone,σXG/σX=0.84;forbespokereversing&alsogrooving,σXL/σX=0.12forgroovingalone,mXG/mX=1.23;forbespokereversing&alsogrooving,mXL/mX=0.23

ThuswhilstgroovingproducedonlyaminorreductioninoverhangvariabilityσXG,‘bespokereversing’ofindividualblocksproducedamajor(88%)reduction.

A ready, if crude, indicator of the poor precision of these blocks is that the SD of theirheights,measuredatpointsA…FforeveryNHBRAblock,waslargeat1.6mm.Itissuggestedthat a height variation within a block of more than 0.5mm should be taken as it beingunsuitableforuseinunmortaredwallingunlessitistobe‘bespokereversed’bythemason.

Ifwewere toanalyse theoverhangofawallbuiltof randomlyvariableblocks,wewouldexpect it to be less than that of a column by a factor of at least 2: unfortunately suchanalysisisimpracticallycomplex.

Clearlyweshouldbe lookingforwaysofmanufacturingblockswith lowSDvalues(σθ) forlateraltaperandverylowvaluesformeantapermθ.Itisunlikelythatblockswithalateraltaperexceeding0.5ocanbeusedtoconstructamortarlesswallexceeding1meterinheight.Achieving low values for σθ requires accurate press-moulds andwell-maintained presses.Achieving low values for mθ can be assisted by reversing alternate blocks (betweenmanufactureanduseinawall).

Measurestoachievestraighterwalls,alsoexploredinAppendixA,thusinclude:• Brushingblockstoremovesmallblemishesandbumps• ‘Grooving’blocksalongtheircentre-linestoprevent‘rocking’block-to-blockcontact

inthecentralregion.Thisreducesσxby15-30%butatacostofsomewhatreducingthebearingsurfaceareaonthetopofeachblock.

• Bespoke reversing of individual blocks during their assembly to best conform to abuilder’slevel.ThisslowsdownwallorcolumnassemblybutreducesσxmassivelyasshownbytheKagadidataabove.

• Identifyingandrejectingblockswhoseroll-taperangleθexceedssay0.5oorwhichdisplayothermajorblemishes.Thisidentificationwouldrequirea‘go–nogo’gaugeintowhicheachblockisdroppedfortesting–aprocedurecurrentlyunknowninISSBconstructionbutnotunacceptablycostly.

ThemaximumoverhangXmaxacceptableduring single-storeymortaredbrickwork is about10mm(seeAppendixA).Withoutapplyingsomeofthefourmeasuresjustlisted,thisqualityis difficult to achieve with unmortared blockwork. A less exacting standard (e.g. Xmax <25mm)isapparentlyacceptableinthosecountriescurrentlypracticingmortarlessmasonry.Thereisalsoevidencethatmechanisedblock-pressingproducesmoreaccurateblocksthanmanualpressing.

7. Achievingverticalstrength

Column failureunderabalancedvertical loadFcanbebycrushingorbyEuler instability.Respectively:


16

Fcrush=σcAc whereσcisthecompressivestrengthofthewallmaterialandAcistheblock-to-blockcontactarea;foranon-centredverticalloadFcrush<σcAc

FEuler=4EIπ2/H2forawallofheightHmadeofmaterialofYoung’sModulusE

BothFvaluesarereducedbyreducingthewallthicknessorbyreducingAc.

Becauseofimperfectcontact(Ac<LW)the2ndmomentattheblock-to-blockcontactplane,(Ic=approx.AcW2/12),hasalowervaluethanwithinblocks(Ib=LW3/12).Duetoabsenceofcushioning, inmortarlesswallsthe2ndmoment Ivariesthroughthewall’sheight inaverycomplexwaydependentonfluxpaths.

InastraightwallofheightH,subjecttoonlyitsownweight,theverticalcrushingpressureatits base will be pbase = ρgH. However we should apply a loading factor f, where f >1, toaccount for the transmitted weight of suspended floors and roofing. Moreover anyeccentricityofverticalloading,orout-of-plumborapplicationofwindloadingwillresultinuneven pressure (vertical stress) through thewall. Taking as a limiting case the situationthat the minimum pressure is zero on one wall face, then at the base of the wall themaximumpressureneartheoppositefacewillbe pbase,max=2fρgH which,toavoidcrushingfailurewithinblocks,shouldbelessthanthecompressivestrengthσcofthewallingmaterial.

However at the block-to-block interface the only-partial contact (Ac/LW = γ, where γ<1)effectivelyincreasesthepressureatactualcontactpointsevenfurther–bythefactor1/γ.

Thustoavoidcrushingfailurewerequire: plowestinterface,max=2λρgH/γtobelessthanthewallmaterial’scompressivestrengthσc.

Foratwo-storeybrickwalling,typicalvaluesforthevariousparametersare:

H=5m;ρ=1800kgm-3;λ=1.5;γ=(veryapproximately)0.2,giving pinterface,max=1.35MPa

Sotoavoidcrushingfailurethematerialcompressivestrengthσcshouldexceed1.35MPa,or higher, as a safety factor is required. In fact a compressive strength σc of 1.35MPa isrepresentiveofalow-qualityfiredbrick,ablockofdriedpressedearthoraweaklystabilisedsoil-cementblock,sounmortaredblockwallingisbarelysuitablefor2-storeybuildingsorfor5m-high gable-endwalls. The distinctive negative feature ofmortarless walling is its lowcontact-areafactorγ=Ac/LWdueto

• imperfectcontactevenoverthenominalbearingsurfaces• bearingsurfacesbeingreduced(bychamfers,interlocksanddeliberategrooving),to

about60%oftheplanareaLxW

Weneedtoraiseγtoatleast0.5tohaveconfidencein2-storeymortarlesswallingoflow-quality masonry. Fortunately in practice very localised crushing of small bumps on the


17

bearing surfaces has the effect of increasing γ as critical loading is approached, therebyincreasingthecrushingstrengthofacolumn.

Itisdifficulttomeasureγdirectly,butwecaninferitsvaluefromexperimentsthatcomparethe crushing strength, σcolumn, of short mortarless columns with that, σblock, of individualblocks testedwith good load-cushioning.We assume γ = σcolumn/σblock.We do have somelimiteddata(seeEWPIIB-8-5)for2x2ISSBprismscomparedwithsingle ISSBblockswhichindicateavalueforγofca0.54.

Table5Strengthof3-blockunmortaredcolumncomparedwithblockstrength

Data to Follow However we do not yet possess the data for the table below. SomeexperimentsatWarwickin2016-7mayprovideextradata.

Blocks are 300mm x 150 mm, *pressed at ca 1 MPa. Burnt bricks are 200mm x100mm.Botharestrength-testedwithtopandbottomcushioning.

Columnsareunmortared ‘prisms’3blockshighand2blocksdeepbut toppedandbottomedwithmortarcushioning.

Allstrengthvaluesaretheaverageforasampleof5tests.

Unit Pressed*soilblock

‘Countrybrick’

Kiln-firedbrick

Pressed*Stabilisedsoil

Columnstrengthσcolumn MPa

Blockstrengthσblock MPa

γ=σcolumn/σblock Ratio

8. Achievinglateralstiffnessandstrength(seealsoAppendixB)For allwalling, but especially boundarywalling,wall-thickness is chosen to give adequatelateral stiffness and strength under wind-loading, seismic loading and casual impacts.Thicknesscanbereducedifbuttressing,closely-spaced‘returns’orstepped(orwavy)wallplansareemployed.150-200mmisthenormalrangeforthethicknessofdomesticmasonrywalling,although100mmmaysufficeforinternalwalls.

The importance of lateral stiffness and strength is debatable. There is little doubt thatboundary walls do fail during wind-storms or earthquakes due to insufficient lateralstrength.Housewallshoweverarestiffenedbyboththe‘returns’thatoccurateveryroomcornerandbyanyring-beamholdingtogetherthetopedgeofthewall.Theimportanceoftheringbeam(orceilingslab)isillustratedbytheAfricanconstructionpracticeofavoidingcompletingtheupperpartsofmortarlesswallsonwindydays.Highlateralstiffnessisclearlydesirableinseismicareasbecauseitraisesthenaturalfrequencyofwallsabovethatoftheshakingmotion–resonanceistobeclearlyavoided.Stiffnessalsohasasecurityfunctionasa wall of low lateral stiffness can be manually broken down by rocking it and therebyincrementallybuildinguplargemovements.


18

Acolumnorwallofperfectblocksofapre-specifiedheightHwouldhaveaninitialstiffness(toout-of-plane forces) thatdependedon theYoung’smodulusof thematerialEand thesecondmoment of area I of the column’s plan. This stiffness would be the same for anunmortaredasforamortaredwall.Forawallorcolumnofunitwidth,subjecttoalateralforce F spread along its top edge causing a displacement δ, we would expect an elasticstiffnessF/δ=3EI/H3whereI=thickness3/12.HoweverwhenappliedforceF,andhencetheoverturningmoment,reachesasufficient(‘yield’)valueFy,partsofonewallfacewillgointotension. Amortared or plasteredwallmay be able to sustain some tensile stress but anunmortaredwallcannot.ForthelatterthereforeatF=Fyblockseparationwouldbeginonthe faceunder tensionand the incremental stiffnesswould reduce.What followsmaybe‘peeling’,cracking&slidingorcracking&hinging–as isanalysed inAppendixB.Howeversliding–aftertakingupanyslackintheblock-to-blockinterlocks-isusuallythenpreventedbysuchinterlocks.

In the case of actual, i.e. imperfect, mortarless blocks, surface irregularity significantlyreduce theeffective2ndmomentof area I andhence theelastic stiffness. Indeed if inter-block contact is limited to a zone close to the blocks’ centreline, this reduction can bemassive: a displacement versus force graph for such a scenario may include horizontalsectionsofzeroincrementalstiffnessduringwhichoneblockrocksontheblockbelow.Thuspreventing (by grooving) contact occurring in this central zone, already noted above asimprovingwallstraightness,alsoenhancescolumnstiffness.Indeed,ingeneral,measurestoimprovestraightnessalsoimprovelateralstiffness.

Experimentaldataconcerninglateralstiffness(andcomparisonwiththeoryofunmortaredmasonryandofcontinuouslycastcolumns)asrecordedinELITHWorkingPapersEWPIIB-8-1,-2,-3,-4.Indicate:

• Lack of ISSB stiffness compared with theoretical continuous & actual mortaredcolumns.

• Benefitsofplastering

9. Plasteringandpartialmortaring

‘Fullmortaring’isthefillingofallhorizontalandvertical(perpend)gapsbetweenunitswitha workable plastic solid, that subsequently hardens, and whose thickness is adjustable.Therearehoweverseveralformsof‘partialmortaring’thatincludedustingwithsand,useofnon-settingmastics,applicationofpointing,useof thin fluidadhesives,and injectionofagroutafterassemblyetc.For thepurposesof thispaperwecountall these techniquesas‘mortarless’.Noneofthemoffersanymeansofadjustingthestraightnessofawall,indeedthey may worsen it if their thickness is significant (e.g. more than 1mm), but they canimprovethosemasonrypropertiesthatdependuponblock-to-blockcushioning.


19

Thereisalsotheoptionofalternatingmortaredandunmortaredmasonry,forexamplebymortaringevery5thor10thcourse,orusingringbeamsandlintelbeamstoperiodicallyresetthewall’splane.Plastering, or even just pointing a wall, undoubtedly improves its lateral strength andstiffness. The plaster thickens the wall by typically 10% and thereby increases its 2ndmomentofarea Ibyover30%. Italsoprovidesa layerwithsome tensile strengthexactlywhere(i.e.furthestfromthewall’sneutralaxis) itcanmostcontributeto lateralstrength.Plasteringprovides sealingof thewallingenvelope that someblocksetsdon’tprovide, sothatitsomissioncanreduceprivacyandresistancetorainpenetration.Indeeditisusualtouse mortarless masonry only where roofing overhangs walls by 50 cm or more, suchoverhangs(intendedtoimproveshading)reducingtheincidenceofdrivingrain.

10. Thegeometricaccuracyofmasonryelements

In theparagraphs followingFigure6 inSection6above, several formsof inaccuracywerementioned – e.g. unacceptably high values for roll-taper angle θ, twist φ and bowing b.Theseare‘withinblock’defects.Thereisalsoanimportant‘betweenblocks’defect,namelyexcessive variation in the average heights of adjacent blocks in a course. Such variationexceeding about 0.5mm will result in some blocks being ‘unsupported’, i.e. having nocontactwiththeblockbelowforhalfoftheirlengthandhencebeingliabletocrackundertheloadfromabove.Suchheightvariationisalsooftenvisiblebecausewhenlookingalongacoursetheblocklinesareseentobewavy.

Thecausesofinaccuraciesinfiredbricksarewellknown–unevendryingandunevenfiring.Forstabilisedsoilorfor‘cement’blocksmouldedfromafirmmix,thecauseliesmainlyinthe moulding process prior to curing, although post-ejection handling can also causedistortionandaccumulationofmicro-debrisonbearingsurfaces.

ForISSBmasonrywerequirenotjusthighergeometricaccuracythanformortaredmasonry,we also require adequate material quality which is dependent on the mix used, themoulding pressure achieved and the thoroughness of curing. Geometric and materialproperties interact.Dependingonthedesignofablockpresswecaneither tightlycontrolsize or tightly control material strength but not both. Fixed-geometry presses produce anearly-fixedblocksizebutuncertainblockquality.

Conventionally a block is formed within a rigid 5-sided steel mould by moving a pistonformingthe6thside(usuallytheblock’sbottomface)underahighforce.Later,duringblockejection themechanismhas to be altered. Thepiston’smovement is fixed by the press’sgeometry,howeverthisrestraintcanbecompromisedbywearinbearingsandbyfailureofthepistonfacetoremainexactlyperpendiculartothesidewallsofthemould.Researchhasalso shown that such ‘single-sided’ pressing results in considerable variation in materialformationpressurefromonepartofablocktoanother.Theblockfacenexttothemovingpistonismorecompressed(andthereforeismoredenseandstrong)thantheoppositeface.Blockmaterialqualitywillvarywiththeappliedpressure (which in turn isaffectedbythe


20

soil-cementmixused), thewatercontentofthemixandmostcriticallybythequantityofthemixputintothemould.Thusmaterialqualitycontrolrequiresaccuratebatching.

Fixed-force presses allow the operator to set the pressure applied during moulding butthereby lose control over block size. As howevermaintaining geometric accuracy ismoreimportant (andmucheasier togauge) thanmaintainingmaterial strength, fixed-geometrypressesaretobepreferredtofixed-forceones.

Acompromisebetweenthetwopressdesignsisonethathastwoopposingmovingpistons,one geometrically limited and the other pressure limited. This can give more uniformpressurethanaone-pistonpress,butofcoursecompromisedgeometricalaccuracy.

Presseshavepartsmovinginaveryabrasiveenvironmentandarethereforeliabletowearthat leads toblockvariability.Greasingofbearings isessentialbutoftenneglected.Somedesignsofmanualpress canonlydeliverabout10,000blocks (i.e. enough tobuildonly2houses)beforetheirwearresultsinunacceptableblock-geometryvariation.

11. High-techversuslow-techversionsofmortarlessmasonry.

There are two forms of mortarless masonry in common use, employing respectivelymechanisedandmanualblockpresses.Asmechanisedpresses typicallycostover$10,000butdeliverhigherpressuresandmoretightlycontrolledmix-batching:wecanregardthemasthe‘hightech’option.Artisanalbuildersinthetropics,whogenerallydonothaveaccessto so much capital, employ manual presses, the low-tech option, costing around $1000.Productionrateswithmotorisedpressescanexceed800blocksperdaywithawork-gangof5, withmanual presses only 400 blocks per day can bemadewith awork gang of 3, somechanisation does increase labour productivity as well as block quality. In practice,mechanisationsuitsproduction inapermanentblock-yard, (especiallywhere there isalsosome mechanisation of block-handling and mixing) or in a temporary yard servingconstructionofanestateofnewhousing.Manualproductionbettersuitstheconstructionofasingledwelling(saywith2000blocks)andwhereslopinglandpermitsmuchofthesoiltobestabilisedtobedugfromthesiteofthedwelling.

12. Conclusionsandthelikelyfutureofmortarlessconstruction

Thematerial,energyandcostsavingsfromusingmortarlessconstructionaresignificant,butthequalityofthatconstructionisoftenpoor.Qualitycontrol,especiallyinblockproduction,requiresmore attention than it is currently receiving. Improvement requires better pressdesign,regularpressmaintenance,measurementofblockdimensionsandblock-hardness,care in curing and batching. Blocks from mechanised presses set a standard (e.g. ofcrispness)thatmanualpressingshouldbeabletomatchbutdonotmatchatpresent.

Interlockdesignsthatallowneatcornerswithouton-sitecuttingarehighlydesirable,asaredesignsthatpermitblockreversalbeforeorduringconstruction.


21

Mortarlessmasonryissuitableforinternalwalls,somewhatsuitableforexternalwallsandlargely unsuitable for boundary walling. Unplastered mortarless masonry should not beusedinactivelyseismicareas.

Plasteringofatleastonesurfaceisdesirableforsealingandforwallstiffness.Itmaybethatblockthicknesscouldbereducedfrom150mmto120mmifcompensatedbyuseof15mmof plaster. Outdoor faces, if not rendered, need protection, by overhanging roofs, fromwater streaming down them; otherwisewaterwill penetrate thewall. Some tongue-and-groove interlock designs do provide some internal sealing. Unfortunately this is at theexpense of preventing block-reversal during construction and so requires a high level ofgeometricaccuracy.

Acceptably straight (i.e. vertical) single-storeywalls canonly bebuiltwith close-toleranceblocks–forexamplethosethat,afterremovalofthemostinaccurateblocks,havewithin-blockandbetween-blockheight variationsofunder0.5mm. Less accurateblocks requireoneofthefollowingprocedures:

• Thatthewallberesettoverticalusingamortarjointeverysay8courses• That‘bespokereversing’andabuilder’slevelisusedduringconstruction• Straightnessstandardsarerelaxedforsingle-storeytropicalhousing.

Blocksshouldbechamfered(tofacilitateadhesionofthinplaster)and‘centre-grooved’topreventblocktoblockcontactclosetotheblocks’centreline.Groovingimprovesbothwallstraightness andwall stiffness. Both processes however reduce the fraction of block-planare that is availableas load-bearing surface. If this fraction fallsbelowsay50% there is apossibilityofblockcrushingatthebottomofwallsover5mhigh.Mortarlessmasonryseemsunsuitedtomulti-storeyconstructionevenwherestraightnessisresetateachfloorbythepresenceofa(concrete)floorslab.

Mortarlessmasonryhasalowstiffness(under20%ofthatofmortaredbrickwork)toout-of-plane forces such as wind pressure, seismic accelerations or roof out-thrust. Stiffness ismuch enhanced by plastering. Lateral strength is also very low and in seismic zonesmortarless walls should be plastered, close-buttressed or crenelated to enhance theirstrength.

Theneedtopress,ratherthanextrude,interlockingblocksgivesanopportunitytoincludethrough-holesclosetotheblock’scentreline.Thesehavelittleeffectoncrushingstrengthorlateral strength but save up to 15% ofmaterial and facilitate the placement of electricalwiringinwallingwithouthavingtochasesurfacegrooves.

Experimentsshouldbepursuedto investigate the feasibilityof injectinga thin (e.g.1mm)layerofmortarbetweenblocksafterconstruction.


22

AppendixA Relationship between out-of-plumb (overhang) X of amasonrycolumnandtheroll-wedgeangleθofitsconstituentblocks

An ideal brick is a rectangularparallelepiped of uniform heightandwith parallel top and bottombearing surfaces. These twosurfaces are indeed the criticalones and especially so formortarless construction. Howeverno artefact has perfectdimensions,allwilldeviatewithinspecified tolerance limits fromsomeideal.

The simplest of models is a twodimensionaloneinwhichwetreatthe contact surfaces (i.e. thosepartsofthetopandbottomfacesthat touchotherbricks)asplaneswith some taper bothlongitudinally and laterally. Thelatter, which we might measureby a ‘roll-taper’ angle θ is themore critical as it affects thewallbeing‘out-of-plumb’.

Uniformblocks:Thesimplestcaseis that θ is the same (= θ0) forevery brick, for which the offset(out-of-plumb)distancexn for the

topofncoursesmaybeverylarge,asthewallcentredescribesacirculararcofradiusR.

xn=R(1–cos(nθ0))=approxRn2θ02/2=n2hθ0

/2. (A1)

Forexampleifθ0=0.01rad(=0.60)andh=0.1mandwallheight=2.5m(thusn=25courses)thentheoverhangwouldbeaquiteunacceptable0.31m.

Howeverthissimpleformofbrickdistortioncanbeeasilycompensatedbyreversingeverysecondbricktoremoveanybiasduetopress-mouldinaccuracy.

Variable blocks:More generally the angle θ is a random variablewhose average θm andstandarddeviationσθmaybemeasured.Byrandomlyreversingbrickswecanmakeθmclosetozero,butthishasnoeffectonσθ.Inthiscasexnwillalsobearandomvariable.

Forblock1,seeFig.A1,thefrontfaceleansforwardatangleβ1(assume=θ1/2)andthetopofblock1overhangsitsbaseby

x1=hsin(β1)≈hβ1 (asβ1isverysmall)

block1

block2

h Lateral taperangleθ1

Blockoverhangx1

CumulativedeviationX2=x1+x2

SIDEELEVATIONOFWALL(2COURSES)

abc

def

chamfer

rearbearingzone

no-contact zone &interlockfeatures

frontbearingzone

chamfer

L

We

PLANVIEWOFABLOCK


23

Byaccumulation,theleanangleofthetopblockwillbeβn=θ1/2+θ2+…+θn

Thetopofthewallwilloverhangthewall’sbaseby:

Xn=x1+x2+…+xn=h{β1+β2+…..+βn}=h{nθ1/2+(n-1)θ2+…+θn}

We may regard the taper angles θ1…. θn of the individual blocks as being independentrandomvariableswithameanvalue(αm)ofzeroandavarianceofσθ2.WeshouldliketobeabletocalculatethevarianceσXn2ofthedeviationofthetopofthewall.

varianceσXn2=h2timestheexpectedvalueof{nθ1/2+(n-1)θ2+…+θn}2

Fromthepropertiesofsuchvariables,wecansaythat

theexpectedvalueofeachofθ12,θ22,…θn2isσθ2

theexpectedvalueofθiθj,=0fori≠j becausetheerrorsareuncorrelated.

So σXn2=h2{n2/4σθ2+(n-1)2σθ2+….σθ2}=h2σθ2{2n3+3n2+n-4.5n2}/6

=n3h2σθ2(1-0.75n-1+0.5n-2)/3 [A2a]

andforlargevaluesofn(typicallyn>20)wecansimplifythisto

σXn2≈n3h2σθ2/3=nH2σθ2/3 giving σXn≈0.58n1.5hσθ [A2b]

Asexpected,highvariation(σθ)inthelateraltaperofthebricksgiveshighdeflectionsinanywallbuiltwiththem:standarddeviationσXnofthewall-topoverhangisproportionaltothetaperangle’sstandarddeviationσθ.Alsoif,foragivenwallheight,thebricksareindividuallythinner(hissmallerandsonislarger)thenthewalltopdeflectionwillbemorevariable.

Moreinterestingly,theoverhang’sstandarddeviationσXnisproportionaltoH1.5whereH iswall height; in general, for a particular block set, doubling a wall’s height will increasemaximumoverhang(forexampletakenas2σXn,i.e.2xstandarddeviationofX)byafactorofabout3.Thisrelationwasborneoutbylaboratoryexperimentswith30columnsofhalf-size blocks for which it was found that σXn was proportional to H1.46 Kintingu, (2009).Howeverintheseexperiments,formula[A1b]wasfoundtounder-estimateσXnbythelargefactorof3,indicatingthatwarpingorbowingofthecontactzoneswerealsocontributingtothevariations inoverhang.Unfortunatelythesedefectsare lesseasytomeasurethanthelateraltaperangleandmuchmoredifficulttomodelthanequation[A1a]above.

A full formulaic approach to assessing whether blocks are acceptably accurate seemsunlikely.Onlyby fieldexperiments cancriteria forblock-erroracceptabilitybedeveloped.Walling standards BS 5628-3: 2005 TableA-2 andBS 5606: 1990 Table 1 specify that themaximum overhang for a 2m wall shall not exceed 9mm, and up to 14m high wall theoverhang shall not exceed ±14mm (this corresponds to σXn < 20mm). This might beexperimentallytiedtoatestonblockssuchas“Whenablockisplacedonaglassplate,atno point around its perimeter should the height of its top surface deviate bymore than1mmfromitsintendedvalue(ofe.g.100mm)”.Thepracticalobjectiveiseithertobeabletodecide if a set of blocks can be assembled to a specified height without exceeding a


24

specifieddeflection limitor to indicatehowmanycoursescanbebuiltwith thatbrick setbefore‘relevelling’(usingmortar)isrequired.

So foragivenwall (i.e.givennandh)andagivenbricktaper (θ0),wecoulddecide if theoffset from vertical X is acceptable or not. When it is NOT acceptable, then we mustimprovebrickaccuracyorlimitnbyintroducingamortarlayereverysay10courses.

UnfortunatelythemeasuredstandarddeviationofoffsetXisusuallyfoundtosubstantiallyexceed the value predicted by this formula – see Section 7 below - indicating that themodelunderlyingtheformulaistoosimple.Thetopandbottomfacescannotbeassumedtobeplanesbutinsteaddisplaysagging/hoggingandtwisting.

AppendixB Behaviourofanunmortaredcolumnunderlateralforce.

Although themodeof failure of a column subject to a large lateral forceFmaybe quitecomplex, we will consider the most likely – namely initiation of ‘peeling’ (progressiveseparation) at some block-to-block joint. To simplify the analysiswithout disqualifying itsvalidity,we consider a column of unitwidth (e.g. 1meter) of heightH, thicknessW anddensityρ,subjecttoanout-of-planeforceFappliedtoitstopedge.Weassumetheheightofthejointatwhichseparationoccursisataheightu(0<u<H).TheheightofthatpartofthecolumnthatisabovetheassumedfailureplaneisthereforeH–u.Wealsoassumethataseparation(‘crack’)hasalreadyprogresseddistanceλWthroughthewall–followingamortarlessjoint.SotheuncrackedsectionB-CiscarryingtheweightρguWofthewallaboveandintheabsenceofanyappliedmomentswouldbesubjecttocompressivestressσc=ρguW/(1-λ)W=ρgu/(1-λ).

The section B-C is also however subject to both a clockwise coupleM1= F.u (due to thelateral top loading) and an anti-clockwise coupleM2 (because thewall’s weight is acting

distance λW/2 to the left ofthe centre of section B-C). Sothetotalclockwisemomentis

M=M1-M2=Fu-λWρgu/2(1-λ).ThismomentactingalonewouldcreateatensilestressatBofσT=Mymax/IwherebothymaxandIapplytosectionB-C.ForthisB-Csection, ymax=0.5(1-λ)WI=((1-λ)W)3/12,soymax/I=6/((1-λ)W)2.

CrackλW

Uncracked(1-λ)W

weightof wall=ρguWA B

C

Distanceu to topofwall

DistanceH-u tobottomofwall


25

Foramortarlesscolumninwhichtensilestressescannotbeborne,themaximumloadFwillbethatwhichreducesthecompressivestressatpointBtozero.i.e.makesσT=σc.Thus ρgu/(1-λ)=[Fu-λWρgu/2(1-λ)].[6/(1-λ)2W)2] (B1)

As dimension u cancels out, it transpires that the size of force F to progress the crackbeyond fraction λ of the wall’s thickness is the same for all potential crack locations(successivecoursejoints):hingingfailuredoesnotalwaysoccuratthebaseofawall.

F=(1-λ)ρgW2/6+λρgW2/2=(ρgW2/6).(1+2λ) (B2)

Thusasthepeel/crackprogressesfrompointA(whereλ=0)topointC(whereλ=1)theappliedlateralforcerisesthree-foldfrom

Fy=(ρgW2/6)atinitiationofpeeling(‘yielding’)to Ff=(ρgW2/2)athingingfailureaboutpointC..

In practice as peeling progresses and λ tends to 1, i.e. the zoneB-C shrinks, compressivestressesbecomeveryhighandlocalcrushingorslidingmayoccur.ItisthereforeprudenttoconsiderFyasthedesignstrengthofthewallagainstlateralloadingofitstop.

Wind-loadingorquake-loadingisdistributedfairlyevenlyacrossthewholeofawall’sfaceandfailurevaluesfortotalforcewillbeabouttwicethoseapplicablewithtop-edgeloading.For quake-loading the applied forces will be proportional to wall mass, so there is anincentivetomakewallsthinner–yetstilllaterallystrongandstiff.

Upto loadF=Fy theelasticstiffnessof the loadapplicationpointshouldstayconstantatleastduringunloading. (During loading theremayhavebeen instancesof rocking contactandlocalisedcrushingofsurfaceblemishes.BeyondloadforceF=Fy(intheoryuptofailureatF=Ff=3Fy)theincrementalstiffnessiserratic,verylowandhardtoestimate.

Documents

On Mortarless Masonry Walling - Welcome to the … (EWP IIB-4) On Mortarless Masonry R11.docx 23/09/2016 1 Energy & Low-income Tropical Housing On Mortarless Masonry Walling Terry