EXPERIMENTAL AND ANALYTICAL APPROACH FOR THE ASSESMENT
OF FLEXURAL STRENGTH OF ADOBE MASONRY
Daniel To eal a1, Ma io Solís2, Pa icia San illán1, Gonzalo Mon oya2
1Enginee ing Depa men
Pon i icia Uni e sidad Ca ólica del Pe ú
A da. Uni e si a ia 1801
San Miguel, Lima 32 Pe ú
Tel: 511 6262000
E-mail: d o [email protected]
School o Enginee ing
Uni e sidad de Se illa
Camino de los descub imien os s/n
41092 Se illa Spain
Tel: 0034 954487294
E-mail: [email protected]
Theme 7: Ancien /His o ic and Inno a i e solu ions o na u al disas e s damage p e en ion and
enhancemen o s uc u al pe o mance
Keywo ds: adobe mason y, bending beha io , expe imen al analysis, analy ical model
Abs ac
This pape p esen s an in es iga ion abou he lexu al beha io o adobe mason y. I is ocused
on he de elopmen o cons i u i e models ha can se e as a basis o he es ablishmen o
new design guidelines o adobe cons uc ions, wi h special emphasis on seismic
ein o cemen s. The pape analyzes he lexu al beha iou o geog id ein o ced adobe walls.
The expe imen al seismic es s o geog id ein o ced adobe houses ha e p o en he
e ec i eness o he ein o cemen echnique. Howe e , addi ional esea ch is needed o
de elop cons i u i e models ha can be used o quan i y he ac ual pe o mances o he
ein o ced adobe. Fo his pu pose, bending es s ha e been ca ied ou o ob ain expe imen al
cu a u e-momen ela ionships o ein o ced and non ein o ced adobe walls. Analy ical
models ha e been de eloped o app oach hese expe imen al laws, using equilib ium and
compa ibili y equa ions simila o hose usually applied o he lexu al beha iou o ein o ced
conc e e. The cons i u i e models o he indi idual ma e ials a e p e iously ob ained h ough
expe imen al es s, and simpli ied cons i u i e models a e p oposed o he go e ning equa ions.
The analy ical models show he duc ili y o adobe mason y, and how duc ili y inc eases when
adobe is ein o ced wi h geog ids. The p oposed ma hema ical models and me hodology can be
applied o o he s uc u al elemen s and ein o cemen sys ems. They can se e as a basis o
he de elopmen o new design guidelines o adobe mason y.
1. INTRODUCTION
Rammed ea h exhibi s well known ea u es as a building ma e ial (low cos ,
sus ainabili y, he mal and acous uc isola ion, e c). Howe e , i exhibi s low mechanical
s eng h, and un o una ely i is mos commonly used in a eas whe e he e is a high
seismic isk. The e o e, i is clea ly necessa y o de elop ein o cemen sys ems and
design guidelines o ob ain adequa e le els o esis ance, essen ially o seismic loads,
bo h o building housing and conse ing his o ic si es. This equi es ho ough and in-
dep h esea ch s udies on he beha io o s uc u es buil wi h his ma e ial. The
p esen pape aims o ake a i s s ep owa ds he s udy o he s uc u al beha io o
adobe mason y, ocusing on he bending beha io o geog id ein o ced and non
ein o ced walls.
O e he las ew decades, a ious ein o cemen echniques ha e been s udied o
imp o e he esis ance p ope ies o adobe. Types o ein o cemen s s udied include
na u al ma e ials (s aw, cane, wood, e c.) (O azzi e al, 1988, p. 1123-1128, Ba iola e
al, 1988, p. 1154) as well as indus ial ma e ials such as chicken wi e, elec o-welded
s eel wi e mesh, PVC pipes and o he plas ic ma e ials (Zega a e al, 1997, p. 1,
Zega a e al, 2001, p. 1, Blonde e al, 2005, p. 1-20). Al hough all hese echniques
imp o e he s uc u al beha io o adobe, no ma hema ical models o he s uc u al
beha io o hese echniques ha e been de eloped so a o quan i y he posi i e e ec
o he ein o cemen sys em.
In ecen yea s, p oposals ha e been made o ein o ce adobe walls wi h geog ids, a
polyme ic ma e ial (Blonde e al, 2005, p. 1-20, Blonde e al, 2006, p. 1-8, To eal a
e al, 2008, p. 1-6). This ein o cemen imp o es he esis ance o adobe walls,
essen ially inc easing i s capaci y o wi hs and ensile s ess. An addi ional ad an age
o he s uc u al bene i s o his ein o cemen echnique is ha i can be applied o
exis ing s uc u es wi hou changing hei ex e nal appea ance. Mo eo e , i has good
du abili y and esis ance o co osion and is economically a o dable.
The p ocedu e used o build houses wi h geog id- ein o ced adobe is explained in
se e al bookle s p oduced o dissemina e he cons uc ion echnique based on he
econs uc ion o he a eas damaged in he Pisco ea hquake in 2007 (Va gas e al,
2007, p. 22-29, GTZ, 2007, p. 37-54).
The basic idea o he geog id ein o cemen echnique is o w ap he ein o cemen
a ound he walls, wo king join ly wi h hem. To do so, he geog id is ied o he wall wi h
s ings h eaded h ough he walls du ing hei cons uc ion. When building he walls,
s ands o ope o nylon should be placed be ween he b icks. Once he walls a e buil ,
he geog id should be placed on bo h sides and ied wi h he s ings. This a achmen
me hod is comple ed by co e ing he geog id wi h mud mo a . Fo he geog id o w ap
he whole adobe s uc u e e ec i ely, he di e en pieces o geog id should be placed
wi h su icien o e lapping and a ached o one ano he wi h pieces o polyme s ing.
The geog id should be ancho ed o he s em wall a he bo om and w apped a ound
he ing beam a he op.
When using he ein o cemen echnique on exis ing cons uc ions, such as his o ic
buildings, o example, he cons uc ion echnique is essen ially simila o he p ocess
desc ibed abo e. In hese cases, howe e , he placemen o he s ings o a ach he
geog id o he wall equi es d illing holes in he wall.
The p esen pape a emp s o s udy he beha io o his compound ma e ial o med by
adobe and geog id ein o cemen . Fo he i s ime, he pape p oposes analy ical
beha io models o ea hen buildings so ha in he u u e design p ocedu es a e no
only based on expe ience and es ic ed o quali a i e c i e ia o geome ical
p opo ions.
The pape is o ganized as ollows. Fi s ly, he expe imen al cons i u i e law o each
indi idual ma e ial is analyzed. Then, he pape shows he expe imen al esul s o he
bending es s o he ein o ced and non- ein o ced adobe walls. Finally, he pape
p esen s an analy ical model o he s uc u al beha io o he walls o app oach he
expe imen al cu a u e-momen ela ionship. The las sec ion includes conclusions and
p oposals o u u e esea ch.
2. MATERIAL PROPERTIES
Adobe was cha ac e ized by pe o ming comp ession es s on indi idual adobe b icks
and piles o 5 b icks. The dimensions o he adobe b icks we e 120x210x100 mm.
These es s showed an a e age esis ance o 1.0 MPa o he b icks. The
expe imen al cons i u i e law ob ained o he adobe mason y ( he b ick piles) show
ha adobe mason y ha e a maximum comp ession s ess a ound 1.1MPa when he
comp essi e s ain eaches 0.4%.
To cha ac e ize he beha io o he geog id, ensile es s we e pe o med ollowing he
speci ica ions o he ASTM D6631-01 s anda d. The geog id exhibi s a e y duc ile
beha io . I showed an expe imen al a e age ensile s eng h o 25 kN/m o a
maximum s ain o 13%.
The expe imen al comp ession cons i u i e law o adobe and he ensile cons i u i e
law o he geog id was app oached by piecewise linea unc ions. These linea ized
cons i u i e laws we e used o he analy ical models.
3. EXPERIMENTAL FLEXURAL BEHAVIOR
To cha ac e ize he bending beha io o walls, bending es s we e pe o med a 3
poin s o e ical walls, as shown in Fig. 1. The walls we e 1.60m high, 0.80m wide and
0.22m hick. O he 3 walls es ed, one did no ha e any kind o ein o cemen and he
o he wo we e ein o ced wi h geog id.
Fig. 1. Pic u es o he bending es s
These es s eco ded applied o ce, which de e mines he bending momen in he
middle sec ion o he wall. A he same ime, s ains we e measu ed a wo poin s o he
ensile side (ε+) and wo poin s o he comp essi e side o he wall (ε-) in i s middle
a ea, using LVDT senso s eco ding he ela i e displacemen be ween wo poin s
40cm apa . This yields wo measu emen s o he cu a u e (χ) in he middle sec ion,
ob ained om he ollowing exp ession:
h
 


(Eq. 1)
whe e h is he hickness o he wall (h=0.22m).
Fig. 2 shows he combined esul s o he momen -cu a u e law o he es ed walls.
This law was ob ained om wo poin s o each wall, p oducing 6 cu es (2 o wall 1,
wi hou geog id, and he emaining 4 o walls 2 and 3, wi h geog id).
Du ing he ials wi h he ein o ced walls, load and unload cycles we e pe o med o
s udy he duc ile beha io o he wall and i s abili y o eco e om he load and s ain
le el a e each cycle. The beha io obse ed was simila o ha o a ein o ced
conc e e beam.
As shown in he Fig. 2, all he expe imen al cu es ob ained we e e y simila . Hence,
i can be concluded ha he geog id ein o ced adobe walls show a ul ima e s eng h
(4.4kNm) h ee imes highe han a non- ein o ced wall (1.4kNm). In addi ion, he
duc ili y o he ein o ced walls is much highe . The maximum cu a u e is also h ee
imes highe han he non- ein o ced one. The beha io a e he elas ic i s s age is
also di e en . The non- ein o ced wall show a so ening p ocess (bending momen
dec eases o highe de o ma ion) whe eas he ein o ced wall s ill inc easing i s
bea ing load o inc easing de o ma ion.
Fig.2. Expe imen al momen -cu a u e ela ionships
The cu a u e-momen ela ionships and he load and unload cycles show ha he
geog id ein o cemen signi ican ly inc eases he amoun o ene gy ha can be
dissipa ed du ing a shaking exci a ion. This is a majo enhancemen o he seismic
esponse o he wall.
4. ANALYTICAL APPROACH
This pape p oposes he use o simila equa ions o hose used o he ein o ced
conc e e. The go e ning equa ions o he bending p oblem o he c oss-sec ion a e
used o p edic analy ically a momen -cu a u e ela ionship.
The bending p oblem is go e ned by compa ibili y and equilib ium equa ions in he
c oss-sec ion o he wall. The p esen esea ch assumes ha he wall beha es like an
Eule -Be noulli beam, wi h small displacemen s, small s ains and la sec ions
emaining la , wi h negligible shea s ain.
Thus, a linea dis ibu ion o s ains was de ined along he c oss-sec ion o he wall
( hickness h=0.22 m o he es ed walls). Fig. 3 shows he s ains along he c oss-
sec ion, as a unc ion o a y coo dina e de ined along he sec ion. The ex eme
comp ession ibe is loca ed a he y=0 coo dina e, which co esponds o he geog id’s
s ain unde comp ession εgc, whe eas he ex eme ensile ibe is loca ed a y=h, which
co esponds o he geog id’s s ain unde ension εg . The neu al axis is loca ed a he
y=x coo dina e, and he angle o med by his s ain p o ile wi h he unde o med
posi ion is he cu a u e o he wall’s ans e se de lec ion (χ).
Fig. 3. Dis ibu ion o s ains and s esses in he c oss-sec ion o he wall
The s ains o each poin can be w i en as a unc ion o he y coo dina e, o he dep h
o neu al ibe x and o cu a u e χ , acco ding o he ollowing exp essions:
)(),,( xyxy
a

(Eq.2)
xx
gc ),(

(Eq.3)
)(),( xhx
g 

(Eq.4)
P o ided ha he cons i u i e laws o he ma e ial a e known, he dis ibu ion o s ains
can be used o assess he esul ing s ess sus ained by each ma e ial (Fig. 3), knowing
ha equilib ium mus be eached in he sec ion. In hese equilib ium equa ions, he
esul ing o ce sus ained by he adobe is gi en by he in eg al o he s ess dis ibu ion
σa(εa) along he sec ion ( he wid h o he wall is gi en by pa ame e b=0.8 m). These
s esses can be comp essi e o ensile and a e de ined by he cons i u i e law o he
ma e ial. Such s esses a e ze o o s ains g ea e han hose accep able unde
ension and comp ession. The esul ing o ces o he geog id unde ension and
comp ession a e de ined by i s s ess pe uni o leng h (Sg and Sgc espec i ely). The
au ho s conside ed ha he p esence o mo a a ound he comp ession geog id allows
i o wo k unde comp ession, since his seemed logical. Howe e , u u e ials will be
necessa y o e i y whe he his hypo hesis is co ec o no .
The o ce and momen equilib ium equa ions can he e o e be w i en as
ollows:
WbxSbxSdybxy g g gcgc
h
aa 
)),(()),((),,((
0

(Eq. 5)
MhWhbxSdyhybxy gcgc
h
aa 
2/)),(()(),,((
0

(Eq. 6)
In he o ce equilib ium equa ion (Eq. 5), he alue o W is ha o he co esponding
dead load o he sec ion unde analysis. Gi en he dimensions o he walls es ed and
he weigh o he conc e e elemen a he op used o hei handling, an app oxima e
alue o W=-4kN was conside ed.
In he momen equilib ium equa ion (Eq. 6), he o igin o momen s chosen was he
ex eme ibe co esponding o y=h, whe e he ensile geog id was loca ed.

By in oducing he cons i u i e law o each indi idual ma e ial in he equilib ium
equa ions, one can ob ain he p o ile o s ains in he sec ion ha co esponds o e e y
alue o bending momen . Hence, he analy ical cu a u e-momen ela ionship is
ob ained.
5. ANALYTICAL RESULTS
The i s analy ical app oach o bending beha io is based on in oducing ma e ial
cons i u i e laws based on hose ob ained in he indi idual es s o he ma e ials in o
he equa ions o he p oblem. As i was s a ed abo e, hese expe imen al laws we e
app oached by means o a piecewise linea law o simpli y nume ical calcula ions.
In o de o analyze and unde s and he eal beha io o he walls obse ed
expe imen ally, a ious cons i u i e models o adobe unde ension we e conside ed.
This pape shows he alues esul ing om conside ing a ew cha ac e is ic models,
called A, B, C and D. The laws o hese models unde ension a e shown in Fig. 4.
Fig. 4. Tensile cons i u i e laws o adobe conside ed o he analy ical models
Law A conside s ha adobe does no o e any esis ance o ension. Law B supposes
an equal ini ial ensile and comp essi e s i ness (752.5 Mpa), which is equal o he
ini ial expe imen al comp essi e s i ness. I assumes a maximum ensile s ess o
0.196 MPa. This alue co esponds o he ensile c ack s ess ob ained by assuming a
c acking momen o 1.4kN.m, which is he maximum momen ob ained expe imen ally
o he wall wi hou geog id, and sub ac ing he s ess caused by dead load. This
yields an app oxima e alue o ensile s eng h in adobe.
Law C supposes a lowe ini ial ensile s i ness (369 MPa), eaching a maximum
ension o 0.075 MPa. I also o e s g ea e duc ili y, by means o p og essi e so ening
o each a maximum s ain o 1.5%.
Finally, model D di e s om C in i s so ening and has e en highe duc ili y. So ening
occu s i s , eaching a s ess o 0.035 MPa, and i is ollowed by a pe ec elas o-
plas ic beha io , eaching a s ain o 5%.
Fig. 5 shows he cu a u e-momen ela ionships ob ained o each model o he non
ein o ced wall. Resul s o model A shows ha i ensions in adobe a e no conside ed,
he wall can only wi hs and he momen hanks o he p e-comp ession caused by he
dead load, and i s highe alue (0.4 kNm) is much lowe han he eal alue. This
illus a es he ac ha , in non- ein o ced adobe cons uc ions, he ma e ial’s ensile
esis ance mus be conside ed o es ima e i s bending esis ance, excep in hose
cases in which he dead load bo ne by he wall is much highe han i s ensile
esis ance. This is no he case in u al adobe houses bu may occu in his o ic
buildings, whe e walls a e hicke and he weigh o ceilings o oo s can be g ea e .
Fig. 5. Analy ical and expe imen al momen -cu a u e laws o a non ein o ced wall.
Conside ing a ensile law like ha o model B, i can be obse ed ha he s i ness
eached in he ini ial s e ch o he momen -cu a u e law is oo high. The maximum
momen , eached when he ex eme ensile ibe eaches i s maximum allowable s ain
(0.0005), is also oo high. F om ha poin , he momen alls sha ply and con e ges o
he cu e o he model wi hou ensile esis ance (model A). This means ha , when
subjec ed o ension, adobe p esumably has a lowe ensile esis ance han ha
es ima ed om he c acking momen , a lowe s i ness o comp ession han ha
ob ained expe imen ally and pa icula ly highe duc ili y.
Models C and D a e aimed a illus a ing his ac . As can be seen, model C adap s o
he expe imen al beha io wi h a high deg ee o app oxima ion and has a sligh ly less
s i and duc ile beha io han he eal beha io . The opposi e is ue o model D, which
is mo e dis an om he expe imen al beha io han model C.
Fo a ein o ced wall (Fig. 6) model A is also unable o simula e he c acking
phenomenon (which happens app oxima ely a a momen o 1.8kNm). I does no make
i possible o es ima e he alue o he maximum momen co ec ly ei he . Ye , once i s
c acking momen is eached and he geog ids s a o wo k signi ican ly, he cha
shows a simila slope o he expe imen al condi ions (simila s i ness) and shows a
ce ain con e gence owa ds he maximum momen ob ained expe imen ally. This is
because in his si ua ion s i ness is de e mined by he geog ids and, o a lesse ex en ,
o adobe unde comp ession, and he ension o adobe plays a negligible ole in a
si ua ion close o b eaking.
Model B shows good beha io un il he c acking momen . Ye , esis ance d ops sha ply
a e his and apidly ends owa ds ype A beha io (wi hou ensions in adobe).
Fig. 6. Analy ical and expe imen al momen -cu a u e laws o a ein o ced wall.
Again, models C and D ep oduce eal beha io qui e closely, al hough model D is
close o eali y. A compa ison be ween his si ua ion and ha o he non- ein o ced
wall sugges s ha he geog id i sel no only inc eases he esis ance o he wall bu
also b ings cohesion o adobe ha con ibu es o imp o ing i s p ope ies, inc easing i s
esis ance and duc ili y.
The esul s ob ained show ha he ela ionship be ween he ensile and comp essi e
esis ance o adobe mason y is ela i ely high i compa ed wi h he usual alues ound
in conc e e. Howe e , conside ing a comp essi e esis ance o 1 MPa in adobe and
applying he app oxima e co ela ions commonly used o in e he ensile s eng h o
conc e e om i s comp ession s eng h (Cala e a, 1992, p. 24 EHE S anda d, 2008, p.
114), he alues o ensile s eng h ob ained would be abou 30% o hose o
comp ession; in conc e e, howe e , he ypical alue o his would be 10-15%. This is
because, as happens wi h conc e e, i can be unde s ood ha a lowe esis ance o
comp ession leads o a highe ela ionship be ween ensile and comp essi e s eng h.
The in luence o his ensile beha io is ob iously negligible when calcula ing he
ul ima e bending momen o he wall. Howe e , esul s show ha i p o ides high
duc ili y and is essen ial o ob ain an app oxima e law o i s beha io be o e b eaking.
Such beha io ul ima ely de ines and cha ac e izes he wall’s abili y o dissipa e ene gy
and de end i sel in he e en o an ea hquake.
6. CONCLUSIONS AND FUTURE RESEARCH
This pape aims o con ibu e o he de elopmen o calcula ion models o ea hen
cons uc ions. Such models a e essen ial o espond o he need o de ine design and
calcula ion c i e ia in his ype o cons uc ions, whe he he pu pose is o build
inexpensi e ea hquake- esis an houses o o conse e his o ic buildings.
The p esen s udy has de eloped models o he bending beha io o geog id- ein o ced
adobe walls. I p o es ha i is possible o ob ain and apply such models in his building
echnique, in spi e o he ac ha ea h is usually conside ed as a “non-enginee ing”
ma e ial.
The models de eloped p o e ha , con a y o he gene al belie , adobe mason y has
ela i ely high duc ili y, in spi e o i s low ensile s eng h. Mo eo e , he duc ili y o
adobe conside ably imp o es when geog id ein o cemen is used. Thus, he geog id
ein o cemen con ibu es o dissipa e he ene gy ansmi ed by an ea hquake, and
con ains he adobe mason y a oiding he collapse o he s uc u e e en when i is
subjec ed o g ea displacemen s. Mo eo e , om an enginee ing iewpoin , he
geog id ein o cemen p o ides con olled mechanical p ope ies ha can be used o
make sa e and mo e eliable p edic ions abou he s uc u al beha io o ein o ced
adobe mason y.
The pape also p o es he need o conside he ensile beha io o adobe when
explo ing he bending beha io laws o he walls analyzed. Such laws de ine he walls’
capaci y o dissipa e ene gy o he ma e ial in load and unload cycles, as happens in an
ea hquake. Howe e , as can be expec ed, his ensile s eng h is negligible in he
calcula ion o he ul ima e collapse momen unde a s a ic load.
Ye , new expe imen al es s a e needed o alida e, comple e and imp o e he models
p oposed in his pape . To do so, we sugges o s a by pe o ming bending ials a 4
poin s, wi h di e en ein o cemen condi ions in he a ea be ween he poin s whe e he
load is applied (a ea o cons an bending momen ). This would make i possible o es
walls wi h di e ences in his a ea: no union mo a be ween adobe blocks, no mud
co e ing he geog id, o nei he o hem. These es s will shed mo e ligh on he
impo ance o hese elemen s on he global beha io o he compound adobe-geog id
ma e ial wi h g ea e accu acy. This is ele an , i s o all, because hese a e he
componen s whose p ope ies a e subjec o he highes a iabili y depending on he
skills and pe o mance o he people doing he wo k. Secondly, he mud co e may all
o in an ea hquake. These es s would help o quan i y he con ibu ion o each
componen o he compound adobe-geog id ma e ial wi h g ea e accu acy and lead o
mo e eliable models yielding p ac ical calcula ion alues ha a e on he sa e side.
Mo e p og ess is also needed o de elop shea beha io models o his ype o walls o
comple e knowledge abou hei s uc u al beha io . In hese shea es s, as in he
bending es s men ioned abo e, i will be necessa y o s udy walls o di e en hickness
o check he alidi y o he models de eloped.
Once models o bending and shea beha io a e ob ained, i will be possible o de elop
ini e elemen models o model he beha io o cons uc ions made wi h his ma e ial, a
leas app oxima ely. This would be a g ea s ep o wa d in s uc u al in eg i y analysis
and he design o ein o cemen sys ems o hese cons uc ions.
The au ho s will app ecia e any con ibu ions made by o he esea che s in he a eas o
esea ch p oposed he e, us ing ha coope a ion be ween di e en esea ch g oups
and labo a o ies will lead o he g ea es p og ess in his ield.
6. Acknowledgmen s
The au ho s would like o hank he Spanish o eign aid agency (Agencia Española de
Coope ación pa a el Desa ollo – AECID) h ough i s In e uni e si y Coope a ion
P og am o he unding ecei ed.
Re e ences
J. Ba iola, J. Va gas, D. To eal a, G. O azzi (1988). Resis an p o isions o adobe
cons uc ion in Pe u. 9 h Wo ld Con e ence on Ea hquake Enginee ing, Tokyo-Kyo o,
Japan.
M. Blonde , D. To eal a, G. V. Ga cía, F. Ginocchio, I. Madueño (2005). Using
indus ial ma e ials o he cons uc ion o sa e adobe houses in seismic a eas. Ea h
Build 2005, Sydney, Aus alia.