Compressive Strength and Modulus of Elasticity

of Masonry Prisms

by

Felix Ip, B. Eng., P. Eng.

A thesis submitted
to
the Faculty of Graduate Studies and Research
in partial fulfillment of
the requirement for the degree of
Master of Engineering

Department of Civil and Environmental Engineering

Carleton University
Ottawa, Ontario
August 18,1999
@ copyright
1999, Felix Ip
 Bibliothque nationale
1+m National Library
ofCanada du Canada
Acquisitions and Acquisitions et
8ibliographic Services services bibliographiques
395 Weilingtori Street 395. rue Wellington
OnawaON K1A ON4 OliawaON K 1 A W
CaMda Canada

The author has granted a non- L'auteur a accord une licence non
exclusive licence allowing the exclusive permettant la
National Library of Canada to Bibliothque nationale d u Canada de
reproduce, loan, dismbute or se1 reproduire, prter, distribuer ou
copies of this thesis in microfom, vendre des copies de cette thse sous
paper or electronic formats. la fome de microfiche/nlm, de
reproduction sur papier ou sur f o m t
lectronique.

The author retains ownership of the L'auteur conserve la proprit du

copyight in this thesis. Neither the droit d'auteur qui protge cette thse.
thesis nor substantial extracts fiom it Ni la thse ni des extraits substantiels
may be printed or otherwise de celle-ci ne doivent tre imprims
reproduced without the author's ou autrement reproduits sans son
permission. autorisation.
 Abstract

Two important elements in structural engineering are the strength and the

stiffness of the structure. In masonry, the primary components are, first of all, strong

and stiff masonry units, and secondly, the relative weak and "soft" mortar.

Masonry units typicaHy have compressive strengths ranging rom 30 MPa to

over 100 MPa in North America, and the standard compressive strength of the mortar

is in the range of 10 to 25 MPa. Their respective moduli of elasticity when tested

independently are significantly different as well, although it is generally accepted that

the units are much stiffer than the mortars.

When these two components are put together to form structural eiements, the

overall strength and modulus are two pieces of important data that are needed by

engineers to design for the strength and deformation of the structure properly.

The research program for this thesis is composed of computer simulation

studies and laboratory testing. The computer simulation was designed to study the

behaviour of the masonry model with respect to the strength and modulus of elasticity

using data obtained from the Phase 1 testing program. Parametric analysis of the

mode1 behaviour was also conducted to examine the effects of different input

parameters for the simulation model.

 The second part of the research program consisted of laboratory testing of

combinations of different strengths masonry units and mortar mixes. The testing

program was designed to study high and low strength masonry units in combination

with two strength extremes of mortar mixes. 'his thesis studied the envelope of this

range of high strength unithigh strength mortar, high strength unitllow strength

rnortar? low strength unithigh strength mortar and low strength unit/low strength

mortar combinations.

The test results were compared with other recent published test results and

Canadian Standards.

The findings of this testing program clearly demonstrate the need for a better

understanding of the mortadmasonry unit interface, and a better measuring technique

for the deformation of mortar joints within a masonry prism assemblage. The

properties of the mortar are changed due to the moisture absorption by the unit, and

this change alone is significant enougn to alter al1 the properties of the mortar when

compared with the non-absorptive specimens prepared for standard mortar testing.

Measuring rnethods using strain gauges across the mortar joints are unreliable due to

the physical size of the strain gauges in comparison to the joint itself.
 Acknowledgements

The author wishes to express his deepest gratitude and appreciation to his

supervisor Professor G. T. Suter for his guidance and teaching throughout the entire

research and reporting program- The author also wishes to extend his gratitude to

Professor S. Kennedy for his guidance and cooperation in reviewing and helping to

select an appropriate finite element program for the research.

The author also wishes to express his sincere appreciation towards Mr. K.

McMartin and his staff for their assistance in the laboratory testing phase of the

program.

The author further expresses his appreciation to the Department of Civil and

Environmental Engineering of Carleton University for the crucial firiancial support

for the duration of this Master's Program, and to the National Research Council for

the generous support at the early stage of the laboratory testing program.

The author especially wishes to express his heartfelt appreciation to his loving

wife, and his children for their unconditional love, encouragement, and support

throughout the years.

Table of Content
1 INTRODUCTION..................................................................... 1

1.1 The Basic Challenge ............................................................................................ 1

l .2 Research Test Program ........................................................................................ 4

1.3 Preliminary Test Program .................................................................................... 4

1.4 Computer Simulation Program .................................................................... 6

1.5 FinalTestPro,gam ............................................................................................... 7

2 RECENT RESEARCH RESULTS .........................................9

3 PRELIMINARY TEST PROGRAM .................................... 15
3.1 General ............................................................................................................... 15

3 -2 Materials ......................................... .................................................................. 16

3.3 Mortar Mixes ........................... ........................................................................ 18

3.4 Compressive Test Specimens.......................................................................... 20

4 COMPUTER SIMULATION ................................................2 1

4.1 General ............................................................................................................... 21

4.2 Basic Simulation Approaches ............................................................................ 21

4.3 Material Properties............................................................................................. 26

4.4 Computer Simulation Modeling Procedure ....................................................... 4 4

5 FINITE ELEMENT ANALYSIS THEORY ........................52

6 COMPUTER SIMULATION RESULTS ................... 56

6.1 Genenl Observation .......................................................................................... 56

6.2 Parametric Analysis .......................

...
....................................................... 63

7 FINAL TEST PROGRAM .

PHASE 1................................. 73
 7.1 Prism Construction and Testing ............................ .
....................................... 75
7.2 Test Procedures of Mortars ................................................................................ 77

7.3 Test Procedures for Ohio Stone ..................... .

.............................................. 80

8 PHASE 1TEST RESULTS ...................................................81

8.1 Properties of Mortars ......................................................................................... 81

8.2 Properties of Ohio Stone................................................................................ 86

8.3 Prisrn Test Result .................................. .

.......................................................... 87

9 FINAL TEST PROGRAM .PHASE II ...............................95

9.1 Mortar Mixes and Masonry Units ................................................................ 95

9.2 Propenies of Mortar Mixes ................................................................................ 96

9.3 Construction of Prisms.......................................................................................

96

9.4 Test Procedure .................................... ............................................................. 98

10 PHASE KI TEST ESULTS ................................................. 108

1O .1 Properties of Masonry Units .......................

... .......................................... 108

10.2 Properties of Strong Mortar ........................ .................................... ...... 111

10.3 Propenies of Weak Mortar............................................................................... III

10.4 Compressive Test Results of Prisms ................................................................ 114

11 DISCUSSION OF RESULTS .e............................................124

1 I .1 Phase 1 Testing ................................................................................................. 124

1 1.2 Computer Simulation Results ......................................................................... 126

1 1.3 Phase II Testing ............................................................................................... 128

12 CONCLUSIONS ................................................................... 137

APPENDIX A .................................................................................... 142
Materials used in final test program .Phase I ........................................................... 142

-
Materials used in final test p r o g m Phase II .......................................................... 143

APPENDIX B .................................................................................... 144

Proposed Estimating Formula ................................................................................... 144

Cornparison of moduli of elasticity between Brook's proposed estimation

formula and Phase i1 fina1 test results ............................................................. 145

REFERENCES .................................................................................146
List of tables

Description Page No.

Table 1 Average compressive strength of prisms ...................................... 116

Table 2 Average modulus of elasticity of prisms..................................... 1 16

Table 3 Range of moduli of elasticity of prisms ...................... .

.....-...-...116
Table 4 Compressive strength comparison of CSA estimations
and Phase II test results ................................................. . ............ 13 1
Table 5 Moduli of elasticity comparison of CSA estimations
and Phase II test results ............................................................... 135

Table Al Mortar mix proportions for Phase I test program.......................... 142

Table B 1 Mortar mix proportions for Phase II test program ...................... 143
List of figures

Fig. 1 Schematic diagram of stress-strain curves of masonry unit. mortar and

pnsm assemblage ........................................................................................... 2

Fig. 2 Typical arrangement of test prism ............................................................... 24

Fig. 3 Basic cornputer mode1 for simulation ................................................... 25

Fig. 4 Typical arrangement for modulus of rupture test of masonry unit

on simply supported beam ........................................................................... 29

Fig. 5 Schematic diagram for split-tensile test of a Stone unit ...............................33

Fig. 6 Cornputer mode1 for split-tensile test .......................................................... 34

Fig. 7 Contour of X-direction stress intensity ........................................................ 35

Fig. 8 Principal stress directions and magnitude ................................................... 36

Fig. 9 Principal stress plot to identify the directions of the stresses only ..............37

Fig. 10 Effective stress of masonry unit and mortar under vertical

compression of pnsm ................................................................................... 39

Fig. 11 3-D mode1 used in computer simulation ...................................................... 45

Fig . 12 Computer mode1 showing the sizes of finite elements ................................ 46

Fig. 13 Stress-strain relationship of elements representing mortar ..........................47

Fig . 14 Stress-strain relationship of elements representing masonry unit ................ 48

Fig . 15 Diagram showing the loading pattern and planes of symmetry...................49

Fig. 16 Typical displacement vector of edges of the elements
under compression ............ .......................................................................... 51

Fig. 17 Schematic diagram of Newton-Raphson iteration process ..........................53

Fig. 18 Iteration process for Newton-Raphson solution.......................................... 55

Fig. 19 Computer simulation showing the direction of displacernent vectors .........58

Fig . 20 Enlarged portion showing the deformation at monar joint interface ...........59

Fig. 2 1 Principal stresses of elements showing the tensile stress of the masonry
unit near the joint interface ......................................................................... 60

F i g 22 Principal stresses of elements showing the magnitude of stresses with

proportional length of arrows .................................................................6 1

Fig. 23 X-axis stress contour with Poisson's ratio of masonry unit being 0.20 .......65

Fig. 24 X-axis stress contour with Poisson's ratio of masonry unit being 0.10 .......66

Fig. 25 X - a i s stress contour with Poisson's ratio of mortar being 0.10 ..................68

Fig. 26 X-axis stress contour with Poisson's ratio of mortar being 0.20 ..................69

Fig. 27 X-mis stress contour with modulus of elasticity of masonry unit

set at 54 500 MPa ........................................................................................ 71

Fig. 28 X-axis stress contour with modulus of elasticity of masonry unit

set at 100 000 MPa .......................................................................................72

Fig. 29 Typical prism layout in Phase 1 test program .............................................. 74

Fig. 30 Typical strain gauge locations on mortar specimen.................................... 78

Fig. 3 1 Stress-strain relationship of 11218 mortar for 50 mm cube specimen .......... 82

 vii

Fig. 32 Stress-strain relationship of 11218 mortar for 100 mm cube specimen ........83

Fig. 33 Stress-strain relationship of 1/0.25/3 rnortar for 50 mm cube specimen .....84

Fig. 34 Stress-strain relationship of 110.2513 mortar for 100 mm cube specimen ...82

Fig. 35 Stress-strain curves of Ohio stone on top and bottom fibre of the simply
supported beam specimen ............................................................................ 86

Fig. 36 Stress-strain relationships of Ohio Stone prsm ~ 4 t h1/2/8 mortar ..............88

Fig. 37 Stress-strain relationships of Ohio stone pnsm with 110.23/3 mortar .........89

Fig. 38 Average stress-strain relatiorships of Ohio stone and 112/8 rnortar in

vertical and horizontal directions ..........................
................................... 92
Fig. 39 Average stress-strain relationships of Ohio stone and 1/0.25/3 mortar in
vertical and horizontal directions .......................... 93

Fig. 40 Typical prism arrangement in Phase II test program .............................. 100

Fig. 41 Schematic arrangement of LVDTs on the test prism for

strain measurement..................................................................................... 104

Fig. 42 Stress-strain relationship of restoration brick ........................................1 0 9

Fig. 43 Stress-strain relationship of flagstone unit ........................................

......1I O

Fig. 44 Stress-strain relationship of strong (1 /0.5/4) mortar ..................................1 12

Fig. 45 Stress-strain relationship of weak (113110) mortar ..................................... 1 12

Fig.46 Stress-strain curves for flagstone prisms with strong mortar ..................... 1 17

Fig. 47 Stress-strain curves for flagstone prisrns with weak mortar ...................... 117
 viii

Fig. 48 Stress-strain curves of brick prisms with strong mortar ............................ 1 18.

Fig. 49 Stress-strain curves of brick prisms with weak rnortar ....................-......... 118
List of Photographs

Phcto 1 Typical mounting of strain gauge using epoxy glue on the unit ..............27

Photo 2 Arrangement of strain gauges on trial prism ........................................... 28

Photo 3 Beam specimen for simply supported bearn testing .............................. 30

Photo 4 Location of strain gauge on simply supported beam specirnen ...............31

Photo 5 Typical failure plane on the specimen as expected in split-tensile test ...32

Photo 6 Split specimen afier failure .................................................................... 33

Photo 7 50 mm cube rnortar specimen being prepared in a non-absorptive

plastic mould ........................................................................................... 41

Photo 8 Strain gauge mounted on cube specimen to determine the stress-strain

relationship during compressive testing................................................... 42

Photo 9 LVDT mounting instrumentation to determine the modulus of

elasticity of mortar on a cylindrical specimen ......................................... 43

Photo 10 Different sizes of rnortar spccimens with strain gauges ......................... 78

Photo 1 1 Prism assemblages with ail strain gauges comected to

data-recording computer ......................................................................... -79

Photo 12 Photograph showing the mortar mix was dry-rnixed thoroughly before
water was added ....................................................................................... 99

Photo 13 Typical construction progress of prism with flagstone .......................... 101

Photo 14 Typical construction progress of prism with restoration bricks.............102

Photo 15 Close-up view of mortar joints showing the moisture
absorption process .................................................................................. 103

Photo 16 Typical arrangement of LVDTs mounted on the special frames ........... 105

Photo 17 Fixed length timber spacers are used to ensure constant

LVDT spacing ........................................................................................ 106

Photo 18 Prism arrangement in loading machine ............................................... 1 0 7

Photo 19 LVDT mounted on 75 mm diameter x 150 mm mortar specimen.........113

Photo 20 Typical brick prism failure mode ..................................................... 1 2 0

Photo 2 1 Typical brick prism exploding at failure ........................................... 121

Photo 22 Typical flagstone prism failure mode ................................................... 122

Photo 23 Typical Stone prism explosion at failure ................................................ 123

List of symbols

Vector of applied loads.

Coefficient matrix.

Tangent manix.

Vector of restoring loads corresponding to the element intemal.

Vector of unknown DOF (degree of freedom) values.

Subscript representing the current equilibrium iteration.

Modulus of elasticity of prism.

Compressive strength of brick.

Compressive strength of mortar.

Compressive strength of prism.

Ratio of rnoduli of elasticity between brick and mortar.

Water absorption percentage (%).

MegaPascal (N/mm2).

GigaPascal(1O' ~ / m r n ~ ) .
1 Introduction

Two important pieces of data in any structural element are the strength and the

modulus of elasticity of the material. In Engineered Masonry, these two cornponents

are the compressive strength, f ,' and modulus of elasticity, E,, of the element. The

compressive strength is important because it determines the bearing capacity of the

element; the modulus of elasticity is important because it provides the estimate of

deformation of the element under loading.

Full size testing can be done to obtain these data, however, full size testing

programs are typically too expensive to be conducted, and in some cases, impractical

because of other limitations. Pnsms testing in the p s t has been proven to be cost-

effective and to provide some measure of correlation with full size testing. The

Canadian Standards ~ssociation'also recognises this test method to reflect the

compressive strength and the modulus of eiasticity of the structural element.

1.1 The Basic Challenge

Typical compressive strengths of masonry units produced in Canada are

relatively high. They are usually of the order of 60 MPa and up, but the strength of

the mortar, on the other hand, is very low. Typical compressive strengths of Type "S"

or Type "N" mortars are about 12.5 MPa and 9.5 MPa respectively according to

Canadian standards. The modulus of elasticity of masonry units is also relatively high
when compared with the rnodulus of elasticity of typical mortar that is used in the

structure. When a typical masonry element is built, the masons combine the masonry

units with mortar to form that structural element. When these two different materials

with vastly different properties are combined to form the structural element, the

properties of the combined element are not readily identifiable.

Fig. 1 shows a schematic diagrarn of the relationships o f the stress-strain

curves of masonry unit, mortar and the pt-isrn assemblage made of these hvo

components as expected from past experience. A prism assemblage is a tenn used to

describe the small element constructed with several masonry units and mortar in stack

Fig. 1 Schematic diagram of stress-strain curves of masonry unit, mortar and

prism assemblage.
bond so that testing c m be done on a smaller scale. The initial tangent of each stress-

strain curve is defined as the initial modulus OP elasticity of that particular material or

prism. When these two materials are combined together, one wvould nomally expect

the stress-strain curve of the pnsm to lie somewhere between the two b o u n d q

curves. Drysdale explained t his relationship in his book Masonry Structures-

Behuviotcr and ~ e s i g din detail. The question is where would the stress-strain curve

of the prism assemblage lie for any given set of masonry unit and mortar combination.

To add to the cornplexity, while the properties of the masonry unit c m be

easily measured and tested, the properties of the mortar within the structural element

are quite different from the results obtained by testing the standard mortar cube

specimens.

This testing prograrn is an attempt to correlate the relationship of the

compressive strength and the modulus of elasticity of the prism assemblages using

different combinations of high and low strength of masonry units and relatively high

and Iow strength mortars.

In actual structures, many more aspects are just as important for the durability

performance of the structure. Considerations for rainwater penetration, freeze-thaw

cycles in the Canadian climate, bond strength of the mortars, creep and so on are also

critical to the overall performance of the structure. However. they are beyond the

scope of this testing prograrn.

1.2 Research Test Program

Due to a limited amount of research results available, (Chapter 2 of this thesis

outlines some of the recent published results by other researchers), there are many

variables that need to be pre-deterrnined before actual testing could take place.

Therefore the research program was split into the foflowing tluee different stages:

1. Preliminary test program to determine the testing criteria;

2. Computer simulation of the prisrns using a finite element analysis program;

3. Final laboratory testing to determine the strength and the modulus of elasticity

of masonry prisms.

1.3 Preliminary Test Program

Unlike ready-mixed concrete which is plant mixed and therefore usually quite

reliable and consistent in terms of its slump, air content, compressive strength, and

other properties throughout a project, mortar for masonry work is usually site rnixed,

and in many smali projects even hand-mixed in a wheelbarrow; its properties are

therefore highly variable.

The waterkement ratio, or more often in the masonry industry the

watedbinder ratio when both Portland cernent and hydrated lime are used as the

binding materiais, is only one of the variables that would affect the strength and

modulus of elasticity of the mortar. The watedbinder ratio is defined as the ratio

between the weight of water to the total weight of cernent and lime. The labourer
hand-mixes the mortar and adds variable amounts of water to the mortar mix based on

his experience and the amount of moisture present in the sand. The mason adds

additional water to the mortar as required by the construction conditions and

reternpers it as n e c e s s q ; this creates another variable- The contractor uses different

brands of cernent or lime on the same job, or different batches of bulk sand with

different moisture content; al1 this contributes to the variability of the mortar strength

and its modulus of elasticity. Even storing the bulk sand on site without proper

protection against the weather elements would seriously affect the moisture content of

the sand, and therefore, ultimately affect the waterhinder ratio.

In general, however, it can be concluded that the compressive strength of

mortar increases with an increase in cernent content and decreases with an increase of

sand, water or air content. Retempering is usually associated with a decrease in

mortar compressive strength. The amount of the reduction in strength depends on the

amount of water added and the time between mixing and retempering. It is this

variability of mortar that may affect the overall performance of the masonry wall or

structural elements. Although this is widely accepted as the norm, there are very few

quantitative research reports available for reference.

Therefore, a first preliminary test program was carried out to prepare different

mortar mixes and test their strengths to determine which mixes would be better suited

for the final test prograrn.

 With CO-operationfrom the National Research Council of Canada (NRC), the

early stage of testing which consisted of mixing different batches, storage of

specimens? and the subsequent compression testing of those mortar specimens, were

al1 carried out at the Montreal Road campus of the National Research Council in

Ottawa, Ontario.

1.4 Computer Simulation Program

A computer finite element analysis program was used to simulate different

scenarios or properties of the materials, and examine the changes in strength and

modulus of the prism. With the advance of computer technologies, it is both viable

and cost-effective to simulate different scenarios to check against laboratory testing.

Unfortunately, there are so many commercial "finite element analysis"

prograrns available with vastly different capacities and b c t i o n s , that the task of

selecting a suitable program was not easy. Some programs can only handle elastic

analysis; some programs can only handle two-dimensional simulations; some limit the

material properties to metals; others have a tedious user interface; and others are just

not capable of computing large triangular matrices-

Afier evaluating different programs in the computer laboratory of the

Department of Civil and Environmental Engineering of Carleton University and

discussions with faculty members, the final decision was to use the program designed

by ANSY S. This computer program can handle three-dimensional problems, non-

Iinear analysis, and most importantly, has a "concrete" element type that can simulate

cracking and cnishing of the element.

The simulation, first of all, consisted of the simple circular area and square

area to confirm the stress flow and displacement pattern of split-tensile testing of a

circular mortar cylinder and a cubic specimen of masonry unit. This was done to

confirm that the simulation inputs are representative for preparation of the final

simulation process.

The final process \vas to simulate the behaviour of the masonry prism

assemblages with different masonry unit and mortar properties.

1.5 Final Test Program

The final test program was to test suitable prism assemblages with different

combinations of masonry units and mortars. The task included the selection of

suitable low and high strength masonry uni& to match with the appropriate sofi and

strong mortars. These different combinations would give good upper and lower

boundaries of the possible spectrum of strengths and moduli of elasticity of the prism

assemblages.

The final testing program itself was further divided into two stages. Stage 1

included testing different prisms using very high and low strength mortars with the

same masonry units, and mounting sufficient strain gauges for strain measurement

during compressive testing. The properties of the masonry units and mortars were
also separately tested. The data recorded were used to determine which mortar and

masoruy units were to be used in the second stage of the final test program.

Stage II of the final test program included building and testing five prisms for

each combination of different masonry units and mortars. These combinations were

high strength mortar with high strength unit, high strength mortar with low strength

unit, low strength mortar with high strength unit, and finally low strength mortar with

low strength unit. Different masonry units were considered to arrive at the final

seiection of hvo units. Over 20 different mortar mixes were tested to finally choose

two mixes for the final program.

 With shrinking financial resources and ever-advancing computing power.

cornputer simulation is becoming more and more usehl to researchen. J.G. ~ o t s - '

and many others started to use "finite element analysis" prograrns to simulate the

interaction of mortar/masonry unit interface of prisms. The approach by Rots was to

compare the effect on masonry prisms using the modulus of elasticity of mortar and a

very low modulus neoprene material in-between the masonry units. The author

proposed an estimating formula for the relationship between the vertical stress of the

prism to the horizontal stress in the masonry unit. Unfortunately the author appeared

to have used the basic compressive strength of the mortar as obtained from traditional

non-absorptive mould specimens without taking into account the change in properties

of mortar after moisture was absorbed by the masonry units in the prisms. Although

the author concluded "that the edge effecrs md non-hornogeneozts deformation are

important factors that cannot be ignored in rnasonry materials cliaracterization, " the

author did not quantify such edge effect and switched to reporting other aspects of the

masonry structures.

S. Stocki, H. Bierwirth and H. ~upfer' fint reported in 1994 at the Tenth

International Brick and h4asonry Association Conference (IBMAC) that the testing
of the same mortar using different testing methods could yield drastically different

results. They utilised a unique steel loading bnish to virtually eliminate the frictional

effect of regular ngid loading platens. They reported that the compressive strength of

mortars obtained from the traditional testing method must be used carefully as the

non-absorptive specimens and the aspect ratio of the specimens always

underestimated the actual compressive strength of the rnortar joints. In certain

conditions the strength of the mortar joints could be as much as four times the

strength obtained from typical traditional cube specirnens tested according to the

German testing standards. This result was obtained by testing rnortar joint specimens

cut out from prism assemblages utilizing the steel wire loading platens to eliminate

the frictional effect of the traditional loading platens. They were one of the very few

who actually reported quantitatively the increase in mortar joint strength compared

with the traditional cube specimen strengths.

S. Stockl, K. Beckhaus and TH. FritscheSwere again arnong the first to report

the characteristics of actual bed joints in test prisms. They concluded that the

modulus of elasticity of the same mortar could Vary as much as ten times depending

on the test method used. The type of test specimen, the production method of the test

specimen, the size of test specimen, and the measurement system were just several

factors affecting the measured results of the rnodulus of elasticity of the same mortar.

They further concl uded that ''anyune onolyzing masonry behuviour for example by

means offnite elemenfs" should take the differences into account to avoid "getting
n>
zm-ealisticcnlczrlation resrdts using wrong material laws.
 Australian researchers A. W. Page, and D. S. Brooks6 expressed a similar

common philosophy of underestirnating the strength of masonry structures when

prism testing results were compared against the Australian code. They revealed that

when using the Australian code method to estimate the masonry structure's strength

based on the correlation between the compressive strength of masonry unit and mortar

type would typically result in a much lower strength than actual field tested results.

Since the code method of estimation does not reflect the workmanship, site conditions

and other factors, it generally tends to be much more conservative. These researchers

recommended the use of prism testing to accurately reflect the actual workmanship

and the actual materials used at a particular project site. They did not propose,

however, any estimating formula that could be used to estimate the strength or

modulus of elasticity in the event that prism testing is not feasible for that particular

project.

G. T. Suter and E. M. F. ~ a g u i b suggested

' Poisson's ratio of brick masonry is

between 0.16 to 0.20 between bedding planes. They used computer 3-D modelling

together with reviewing other previous research results in reaching their conclusion.

They concluded the modulus of elasticity parallel to its bed planes was srnaller than

that perpendicular to its bed planes. This anisotropic nature of masonry structures

makes any estimating process much more complex than a simple empirical formula

could cover.
 A. W. Hendry8 also confirmed the seemingly straightfonvard compression test

on a masonry unit was not simple at all. Depending on the rate of loading, capping

material used, and aspect ratio of the testing specimen, the test results wouId Vary. He

suggested a loading rate between 7 MPaImin to 40 MPa/min would be appropriate for

consistent "a statistically sign~@cantbut not Imge variation in apparent strength. "

The capping material and its thickness also atiected the strength of the masonry units.

Grinding the testing specimen flat was discussed as an option, however not

recommended due to the inherent nature of the workmanship difficulties and machine

limitations.

Hendry also noted the tensile strength of a rnasonry unit was best based on a

split-tensile test because of its uniform tensile stress distribution throughout the entire

depth of specimen. He went on to suggest other researchers use the split-tensile test

result to calculate the compressive strength of the same unit.

He further pointed out that the worhanship factor was the most important

factor among any others in masonry structures. Incomplete filling of joints would

inevitably give rise to inaccurate results. To ensure proper workmanship,

professional masons were employed to construct the prisrn assemblages at different

phases of the final test prograrn of this thesis. A purpose-made timber jig was

empfoyed in the final test program to ensure a uniform joint thickness of prisms was

achieved.
 The testing method could also significantly affect the test results. R. G.

Drysdale & H. E. won$ suggested that the test prisms should be at least four units

high to avoid the frictional influence of loading platen restraint. The final test prisms

of this thesis were al1 eight units high to allow for the Linear Variable Displacement

Transducers (LVDT) to be mounted on specially made mounting frames.

L. Binda et. al." concluded that "the use of strain gauges for displacement

detecrion in the case of s m d specimens and sofi materids with high porosiy muy

give unreliable results due to the intience of the glue st%fess. " This is particularly

true for mortar joints regardless of the strength of the mortar. She also concurred with

Stock1 that loading platens consisting of steel bnishes would be the most successful

device to avoid platen confinement in compressive testing.

J.J. Brooks and B. H. Abu ~ a k e r "recently published their findings afier

reviewing 184 sets of modulus/strength data. They proposed an empirical formula

relating the modulus of brick pnsms to the strength of their components, Le. the

mortar and unit. In their proposed formula, they included the water absorption of the

unit as one of the variables, together with the strengths of the mortar and unit. With

that many factors that could influence the modulus of the prisms, they claimed that

with their proposed formula they could predict the modulus with better accuracy.

They reported an error coefficient of about 26% when compared with the British

Standards and the corresponding Euro-Code. This was, according to the authors, a

drastic improvement from any other currently used estimation methods. Appendix B
highlights the fomulae proposed in their report, and presents the corresponding

estimation of the modulus of elasticity for the materials used in Phase II o f the final

testing program.
3 Preliminary Test Program

3.1 General

In order to understand the behaviour of mortar, different batches of mortar

were prepared and tested in this prelirninary testing phase of the program. This

preliminary test program was critical in selecting suitable mixes for the final test

program; mortars with different proportions of ingredients were made and tested at

seven days after the specimens were prepared.

The mortars were prepared with assistance from the National Research

Council (NRC); the mixes were prepared in the mechanical mixer at NRC using white

Type 10 Portland cernent, Type 'S' hydrated lime, and aggregate sand from a local

quarry. While proprietary masonry cements are popular among masons in the

construction industry because of their good workability, they contain different filler

materials that are not listed in the manufacturer's published material data. It is

because of their proprietary secrecy that the true composition o f masonry cements

cannot be identified, and hence this type of cernent was judged unsuitable for the

research purposes in this thesis.

3.2 Materials

In order to control the variables for the entire testing program, the materials

used were carefully selected and maintained. Throughout the testing program. Type

10 white Portland cernent from the same manufacturer was used.

Since mortar is usually exposed to weather, in Canada freezing of moisture

within the mortar is inevitable during winter months. For the durability of the mortar

it is important to have a minimum of 8 to 10% air content within the hardened mortar

to allow the moisture to have the space to freeze within the mortar without

prematurely failing the mortar. Initially, air-entrainment agent originally developed

for reinforced concrete was added to the trial mixes in an attempt to produce the

desired air content.

Different mixes using different proportions of Portland cement, hydrated lime

and sand aggregate with air-entrainment agent were tested for air content, among

other plastic properties. The measurement of the air content of the rnortar was done

using the portable laboratory air content measuring instrument available at NRC. The

instrument is based on the theory that air volume expands with a reduction of pressure;

the measurement of volume expansion was indirectly measured by the linear

potentiometer at different pressure settings. Unfortunately, the measurements were

not reliable, as different readings were recorded for mortars with exactly the same

proportion of ingredients. The inconsistency of air content in the mortar was first

attributed to the fact that the air-entrainment agent was originally developed for
reinforced concrete rnistures, and the dosage needed for a mortar batch was very

small and therefore very difficult to administer accurately. The second factor that led

to the inconsistent measurement was that the equipment was later found to be

malfunctioning by NRC laboratory staff.

After testing different batches of mortar mixes in the laboratory at NRC and at

one of the repointing projects in Ottawa, the final air-entrainment agent chosen for

this research program was Type 'SA' hydrated lime manufactured by Bondcrete.

This product designation was to signify that the hydrated lime satisfies the Canadian

Standards Association's requirement for Type 'S' hydrated lime with air entrainment

aiready pre-mixed during the manufacturing process of the hydrated lime. This type

'SA' hydrated lime was consistently producing an air content of between 10-1 1 %

throughout the repointing project. The field rneasurement was done with the more

reliable air content measuring instrument made of steel construction for checking air

content in concrete.

Recently, A. H. P. Maurenbrecher and K. Trischuk" conducted an extensive

study on the properties of pointing mortar. In their study, various air entrainment

agents were used. They reported that using different brands of air entrainment agents

in the mortar mixes or Type 'SA' lime were al1 successful in achieving over 10% air

content consistently when the mixes were fluid enough to have a flow table reading of'

145 to 150%. Although this report was only available after the current testing

program for the thesis had been completed, it did confirm that the use of Type 'SA'
lime was an etrective method to ensure proper air content in ail the mortar mixes used

in the final test program.

The aggregate sand used throughout the entire testing program was river sand

ihat was wvell graded with a gradation within the Canadian Standards Association's

upper and lower gradation limits.

3.3 Mortar Mixes

The initial trial mixes of mortar ranged from a pure cementlsand mix to a

cement/lime/sand mix. Pure cementhand mixes ranged from a ratio of 1:3 to 1 :7.5 to

obtain the high and low limits of compressive strength, whereas the cementllime/sand

mixes ranged frorn 1:0.25:3 to 1:3:10 to obtain high and low limits. Watedbinder

ratios were recorded after the mixes achieved bedding mortar consistency; they

ranged from 0.9 for high cernent content mixes for the 1:3 cementkand mix or

1:0.25:3 cementAimelsand mix, to 1.25 for high sand content mixes on the other end

of the spectrum. Since the air content is important in regular site work in the

Canadian clirnate, the pure cement/sand mixes were elirninated for the final testing

program due to the inconsistency of air content measurement whm using an air-

entrainment agent. Type 'SA' lime was used throughout the final test program in

order to achieve a consistent air content in the rnortar. Therefore, al1 mortar types

used in the final test program were Portland cement/lime/sand (PCIWS) mixes.
 Plastic properties of the mortar mixes were also recorded, although they were

just recorded to ensure that the laboratory mixes werc similar to the actual field mixes

by masons. Each workable mix had a 135% to 160% flow table reading. Air contents

were initially measured with the portable air content measuring instrument at M C ,

but were later concluded to be not reliable as the instrument \vas not functioning

properly at the time of testing. An air meter designed for measuring air-content for

concrete was finally used to measure the air content; it consistently recorded 10 to

1 1% for al1 mortar mixes using Type 'SA' hydrated lime.

A Vicat cone penetration test was also performed for each of the preliminary

trial mixes. The test is a measure of workability of the mortar by measuring the

penetration depth of a conically shaped metal head that was released from the top

surface of the wet mortar in a fixed size container. This test gave some indication of

the workability of the mortar; any mortar mixes that had a cone penetration test result

of less than 20 mm were considered as having poor workability and were therefore

rejected. Mortar specimens obtained from these low workability mixes were also

likely to have visible horizontal separation cracks at half height of the specimen

showing the distinct different lifts during the moulding of the specimens. This clearly

suggested that the standard tempering procedure during moulding was not adequate

when the mortar was too dry due to low workability.

3.4 Compressive Test Specimens

Six 50 mm cube specimens were prepared for each rnortar rnix using non-

absorptive moulds. Al1 specimens were moist cured for six days using polyethylene

wrapping which effectiveiy achieved 100% humidity inside the wrapping; this was

followed by one day of laboratory curing at about 50% relative humidity and 20" to

22" C before testing. Al1 compressive testing of mortar cube specimens was done

directly on the platens of a universal testing machine without capping materials

placed on the specimen, and with al1 specimens rotated 990" to ensure the rough top

surface of the specirnens was not in contact with the loading platens.

The compressive strengths of the mortar ranged from 1.8 MPa for the

1/3/10(PC/L/S) mix to 18.0 MPa for the 1/0.25/3(PC/LIS) mix. The mixes selected

for the final test program were a IB/lO(PC/L/S) rnix for the low strength mortar and a

1/0.5/4(PC/L/S) mix for the high strength mortar.

 4 Computer Simulation
1

4.1 General

With the advance of computer technologies, the finite element analysis

technologies are becoming increasingly more powerful and readily available. Many

of the commercially available finite element programs can simulate the behaviour of

many civil engineering structures so that a better understanding of the structures can

be achieved without physically testing the full size, or even the reduced model, of the

structures. However, a basic understanding of material properties must fint be

achieved prior to the use of computer programs in order to perform a realistic

simulation.

4.2 Basic Simulation Approaches

There are many different approaches to simulate the behaviour of masonry

structures. The most common appr~aches'~~".

-' are:

1. Equivalent homogeneous material or macro-simulation approach - this

approach assumes an equivalent homogeneous material of the masonry

structure despite the well documented fact that both m a s o q units and mortar

behave quite anisotropically;

 Smear line elenent approach - this approach utilizes line elements

representing horizontal and vertical joints of the masonry structure; and

Detailed definition or micro-simulation approach - this approach defines

mortar and masonry units with clear boundaries and their individual properties.

These different simulation methods offer different degrees of accuracy and

therefore should be used according to the requirements of individual situations. The

first approach offers a general behaviour simulation of the structure and is better

suited for studying large size structures. The second approach offers a better accuracy

of the behaviour of a masonry structure and is suitable for simulation of an individual

element of the structure such as a wall with openings to study the concentration of

stress. The last approach studies the detailed interaction between mortar and masonry

units and is rnost suitable for the current study as it provides the most detailed

accuracy during simulation.

Fig. 2 shows the typical prism used in the final test program. It consists of

eight masonry units, together with a compressive fibre material on top and bottom of

the prism to reduce the frictional effect of the platen during loading. No strain gauges

were mounted on the prisms during the final Phase II test program as it was clear

from the Phase 1 testing results that:

a) The strain gauges mounted on the masonry units of the prisms yielded the

same results as those mounted on the masonry units during individual

 compressive testing. Therefore, the modulus of elasticity of the masonry units

was measured with the strain gauges mounted on the individual units during

compressive testing of the units only.

b) The srnailest practical size of strain gauge used was 5 mm long with a plastic

rnounting base plate of 10 mm length. Therefore, the 10 mm thick standard

mortar joints used in the final Phase II test program were just too small for

strain gauges to be mounted on them without interference between the edges

of the gauges and the masonry units. Smaller gauges of 2 mm length were

available but not suitable for a porous materials like mortar that might have

significant localized voids.

For cornputer simulation work, it was not necessary to deal with the entire

prism. Due to symmetry, the simulation only needed to model one-quarter of the

masonry unit and one-eighth of the mortar joint. Fig. 3 shows the cornputer model

used in the cornputer simulation.

 Applied Load

Location of measuring
point using LVDT's
-

Location of measuring
point using LVDT's

Reaction Table

Fig. 2 Typical arrangement of test prism.

 / USED IN COMPUTER
SIMULATION

Fig. 3 Basic computer mode1 for simulation.

4.3 Material Properties

Using the ANSYS finite element program, the basic properties of the masomy

unit and the mortar were entered into the program. The basic properties for the

rnasonry unit for the purpose of computer simulation were defined as compressive

strength, modulus of elasticity, modulus of rupture, and Poisson's ratio. Several other

properties might affect the actual site condition, but were not essential in the

cornputer simulation; some of these are the rate of absorption, the moisture content,

and size of pores within the units. Although these factors may affect the properties of

the prism assemblages at actual job sites, they were not critical in the computer

simulation simply because the computer program did not have the capability to

actually simulate the process of mortar hardening and water absorption.

The compressive strength of rnasonry units was determined by the normal

compressive testing of 50 mm cubes in the case of a Stone unit, and half bricks of

different units in the case of brick units. The moduli of elasticity of the units were

deterrnined by strain gauges mounted on the specirnens during compressive testing,

and also on some masonry prisms to confirm the results from the compressive testing

in the Phase 1 of the final testing. A third set of strain gauges was mounted on the

small beam specimen for the modulus of rupture test in Phase 1 of the final testing

program to confirm the modulus of the masonry unit. Photo 1 shows a typical strain

gauge mounting with epoxy compound on the surface of the beam specimen. Photo 2

shows the typical strain gauge arrangement on the prism assemblages.

Photo 1 Typical mounting of strain gauge using epoxy glue on the unit.
Photo 2 Arrangement of strain gaugcs on trial prisrn.
The modulus of rupture of masonry units was determined by using a combination of

two different methods. The first method was using a two-point loading on simply

supported beams of dimensions 50 mm x 50 mm x 300 mm, and the second method

was split-tensile testing using a line load across the top of the specimen of a 50 mm

cube specimen. Fig. 4 s h o w a schematic test arrangement of a simply supported

beam specimen and Photo 3 shows the actual testing arrangement of the beam

specimen on the testing machine. Two-point loading ensured a uniform bending

moment between the two loading points.

Fig. 4 Typical arrangement for modulus of rupture test of masonry unit on

simply supported beam.
Photo 4 shows the top view of the beam specimen indicating the location of the strain

gauge mounred at the top fibre of the specirnen that lies between the two loading

points. Photo 5 shows the failure plane of the specimen after failure; a clean failure

plane was to be expected for ail t e n d e stress failures of the beam specimen.

Photo 3 Beam specimen for simply supported beam testing.

Photo 4 Location of strain gauge on sirnply supported beam specimen.
 Fig. 5 shows the schematic test arrangement for the split-tensile test of a Stone

masonry unit and Photo 6 shows the actual failed specimen after testing. In order to

verify the computer modeling procedures, a simple model of the spiit-tensile test was

used in the finite element analysis program to check against the known theoretical

concept. Fig. 6 is the finite element computer simulation of the split-tensile test

model. The loading was in the Y-direction with a negative value as indicated by the

downward arrow, and the model was restrained at the centre of the area at the bottom

as indicated by the chevrons. The stress contour was plotted in Fig. 7 as the intensity

of the stress in the X-direction; Fig. 8 plots the principal stresses of each

Photo 5 Typical failure plane on the specirnen as expected in split-tensile test.

 Applied load
I

Fig. 5 Schematic diagram for split-tensile test o f a Stone un

Photo 6 Split specimen after failure.

Fig. 6 Computer mode1 for split-tensile test.
 ANSYS 5 . 3
AUG 17 1998
13:08:59
NODAL SOLUTIO
STEO=1
SUB =1
TIME=l
SX ( AVG
RSYS=O
DMX =. 849E-03
SMN =-. 87684
SFNB=-2.205
SMX =.146073
SMXB=. 917596

Fig. 7 Contour of X-direction stress intensity.

 ANSYS 5.3
AUG 17 19
13:05:07
VECTOR
STEP=l
S U B =1
TIME=1
S

Fig. 8 Principal stresses directions and magnitude.

L ANSYS 5 . 3
-_ -- --- - - __ - AUG 17 1 9 9 t
t
13:06:03
VECTOR
1 STEP=1
SUB =i

Fig. 9 Principal stress plot to identify the directions of the stresses only.
 element centroid. This plot shows the magnitude of the stresses in proportion to the

intensity of the stress, that is the longer the arrow, the higher the stresses. The

direction of the arrow pain also indicates whether the element is in compression or

tension. The arrows pointing towards the centroid are the compression stresses and

those pointing away from the centroid are the tensile stresses experienced by that

element. This plot demonstrates the tensile stresses experienced in the X-direction

when the downward loading is applied in the middle of the area. For clarity, the same

principal stress is re-plotted in Fig. 9 using the same length of arrows to show the

directions of the stresses only.

The property that was not actually obtained by tests was the Poisson's ratio; it

was initially entered as O. 15 based on previously documented results".

The basic properties of mortar are defined as compressive strength, modulus of

elasticity and Poisson's ratio. The modulus of npture of the mortar was not

considered essential as the failure rnechanism of the prisms was always initiated by

the splitting of the masonry units, whereas the mortar joints were still under a tri-axial

compression state' just prior to the failure of the prism. Fig. 10 shows the schematic

relationship of this phenomenon of the equivalent tri-axial compression state of the

mortar. As the prism was subjected to uni-axial vertical loading, the mortar would

attempt to expand laterally because of the Poisson's effect, however, the masonry unit

did not expand at the same rate due to its much higher modulus of elasticity.

Therefore, if good bonding was achieved at the bed joints between the masonry units
and mortar, the masonry units were restricting the lateral expansion of the mortar.

This restriction results in a lateral bi-axial compression confinement pressure thus

subjecting the mortar to tri-axial compression. Similar to reinforced concrete, the

compressive strength would increase significantly when lateral-confining pressure

was applied.

The reverse was true. however, for the masonry units; the lateral expansion of

the mortar acted as the tensile forces on the masonry units, thus the masonry unit was

undergoing a uni-axial compressive and bi-axial tensile stress state. The net result

was the pnsm would fail by the tensile splitting of the masonry units once the units

exceeded their modulus of rupture. This applied loading would then be considered as

VERT. LOADING

BI-LATERAL TENSION
& UNI-AXIAL COMP.

- - - . - a - . - . -
2
0 -
0-
0

I '1 '1 '1 '1 '1 '

VERT. LOADING

Fig. 10 Effective stress of masonry unit and mortar under vertical

compression of prism.
the ultimate strength of the prism. This phenomenon is also referred to as the indirect

tensile testing of prism assemblages by some researchers.

Similarly, the compressive strength of the mortar was determined by the

compressive strength of standard 50 mm cube specimens from non-absorptive plastic

rnoulds as shown in Photo 7. The modulus of elasticity of the mortar was also

determined by strain gauges mounted both on cube specimens and on joints of prisms

as shown in Photo 8, and also by a Linear Variable Displacement Transducer (LVDT)

mounted on a 3" diameter x 6" high cylindrical specimen as shown in Photo 9.

Poisson's ratio was assurned as 0.15 fom previous research data-

The compressive strengths of masonry uni& and mortars are two of the most

tested properties for typical projects simply because the specimens are reiatively easy

and inexpensive to prepare when compared with testing for other properties. Most

irnportantly, they are reasonably reliable strength indicators for commercial or

research laboratories.

Other properties, however, are seldom tested, because they are either very

espensive to carry out or often very unreliable. The preparation work for testing

using strain gauges is both expensive and, at times, erroneous if the gauges are not

prepared properly. Before the strain gauge can be mounted on the specimen, the

surface of the base material must be sanded, thoroughly cleaned, and treated with

epoxy resin to fifi the voids on the surface. The strain gauge must then be carefully

applied on the surface of the epoxy resin under specific manufacture's instructions.
The data transrnitting wiring is then soldered ont0 the conducting terminals before the

strain gauge readings c m be recorded with another remote recording machine or

cornputer. The strain gauge will simply give erroneous result if any one, or any

combination, of the preparation procedures is not followed correctly. Another

possible area of concern is that the epoxy resin fills the voids of the base material and

may have altered the properties of the base material somewhat.

Photo 7 50 mm cube rnortar specimen being prepared in a non-absorptive

plastic mould.
Photo 8 Strain gauge mounted on cube specimen to determine the stress-
strain relationship during compressive testing.
Photo 9 LVDT mounting instrumentation to determine the modulus of
elasticity of mortar on a cylindrical specirnen.
4.4 Cornputer Simulation Modeling Procedure

Only a quarter of the masoxuy unit and one-eighth of the mortar joint were

needed for simulation of a prism assemblage because of symmetry. Thecefore, the

CO-ordinatesof the corners for the masonry unit and the mortar were entered into the

computer to form two comecting solid volumes. The volume representing the mortar

was fitted with elements that had the properties of rnortar. The volume representing

the masonry unit was hirther divided into two regions. Al1 these two regions were

fitted with elements that had the same properties of the masonry unit. The volume

closer to the mortar, however, was fitted with smaller elements, and the volume

further away from the mortar was fitted with larger elements. This would provide

better computation accuracy near the mortadunit interface plane. Fig. 11 indicates the

three regions of the computer model represented by the three different volumes. Fig.

12 shows the element configuration in the model.

Each type of elements was defined with its own properties in terms of its

respective uni-axial compressive strength, cracking tensile stress or modulus of

rupture, initial modulus of elasticity and stress-strain relationship. The stress-strain

relationships were al1 multi-linear approximations. Fig. 13 shows one of the multi-

linear stress-strain curves approximating the non-linear characteristic of masonry unit,

and Fig. 14 shows a typical stress-strain relationship of mortar.

 The three principal X-Y, X-2,and Y-Z planes were assigned as planes of

symmetry for the simulation. The appiied Ioading was on the top surface of the

model, in the negative Y-direction representing the compressive loading. Fig. 1 5

shows the direction of loading pressure indicated by the downward arrows and the

planes of symrnetry indicated by the chevrons.

Fig. 1 1 3-D model used in cornputer simulation.

Fig. 12 Cornputer mode1 showing the sizes of finite elements.
 MIS0 Table For Material 2 ANSYS 5.3
AUG 6 199r
15:20:30
Table Data

Fig. 13 Stress-strain relationship of elements representing mortar.

 -4PR 2 5 1 9 9
10:11:33
NODES
TYPE NUM
- 7

Fig. 15 Diagram showing the loading pattern and planes of symmetry.

 The simulation was carried out in sub-steps of 1 MPa increments until the

mode1 failed due to non-convergence of differential equations. The final loading was

then funher sub-divided into 0.1 MPa increments to obtain the failure pressure to the

nearest 0.1 MPa. This compressive loading would then be defined as the ultimate

strength of the prism in the computer model. Since the elements defined in the

simulation contained no steel reinforcement, as soon as they reached either the

cracking stress (modulus of rupture) or the crushing stress (ultimate compressive

stress), the matrix equations would fail to converge and this signaled the program to

terminate fiirther calculations.

Fig. 16 shows typical displacement vectors of the elements in the mode1 under

compressive loading from the Y-direction. The results are discussed in later chapters.

In order to study the effect on the failure strength of the model due to changes

in properties of different individual components, the input variables were changed and

the failure pressures recorded. These changes were carried out systematically one

variable at a time.
 ANSYS 5 . 3
AUG 12 19:
O9:36: 1 3
VECTOR
STEP=1
S U B =40
TIME=l
U
NODE=1320
MIN=O
MAX=. 0 2 7 1 ~
O
. ~ b 336
~

d
~ - 3 ) -

.G C 0 7 3 i
- v3:='7
.2;3543

-927205

Fig. 16 Typical displacement vector o f edges of the elements under compression.

5 Finite Element Analysis Theory
I
The Finite Element Analysis uses a "discretization" process wiih the following

set of simultaneous equations to equate the interna1 forces with the extemal applied

loads:

where [KI = coefficient matrix

{u} =vector of unknown DOF (degree of freedom) values

{F. } = vector of applied load

This set of equations would be correct if the matrixes are al1 linear. However,

if the simulation were to be non-linear, this simple matrix solution would give rise to

incorrect accumulation of computational errors.

Fig. 17(a) shows the relationship of the assumed true response and the

calculated responses. An iteration process is therefore required to overcome this

shortcoming of straight simulation and Fig. 17(b) shows the general schematic of the

iteration process.
 calculated
response

true response
-- --.. -
. .- ..-. .- . - . - - ..

(a) Pure incremental solution

- - -,
u
(b) Newton-Raphson Iterative Solution

Fig. 17 Schematic diagram of Newton-Raphson iteration process.

 When the coefficient matrix [KI is itself a function of unknown DOF values,

then the original matrix equations become non-linear equations. The Newton-

Raphson method is an iterative process of solving the non-linear equations and can be

written as:

where [KT]= Tangent matrix

i = subscript representing the current equilibrium iteration

{F;"'}= vector of restoring loads corresponding to the element interna1

loads.

The rnatrixes [ ~ r ] a n d ma are evaluated based on the values given by (ui),

and the term IF')- {F,"} represents the residual vector or sornetirnes referred to as the

out-of-balance load vector. This is the amount of out-of-equilibrium from the system.

Fig. 18 shows the graphical relationship of a single solution iteration of a single

Degree of Freedom (DOF) model.

1
In this case, [KT]represents the tangent of the stiffness matrix, { is the

restoring force matrix calculated frorn the element matrix, and { u i ) is the

displacement matrix of the system. As seen from the graph, the tangent solution does

not represent the actual response, and therefore an iteration process is necessary for
the solution to converge. The solution obtained at the end o f each iteration process

represents the load level { F U } .The final converged solution would be in equilibrium,

and in such case, the restoring load vector {F;"'}computed frorn the current stress state

would equal the applied load vector {F"} within the acceptable toIerance.

The ANSYS program implements this Newton-Raphson procedure and the

program would go through al1 the intermediate steps until convergence is reached for

each load step. Since rnasonry prisms typically behave in a non-linear manner, this

method is necessary for accurate simulations. The convergence criterion was set at

1o - ~a, relative small value in order to ensure the computation error did not exceed the

simulation errors.

response

Fig. 18 Iteration process for Newton-Raphson Solution.

6 Cornputer Simulation Results

6.1 General Obsewation

The computer model was to simulate the behaviour of the rnasonry prism

under vertical compressive loading until faiture. Fig. 19 shows the displacement

nodes of the elements in the model as simulated using the ANSYS finite element

analysis program. It shows the displacement of the nodes of the elements is changing

gradually from the vertical negative Y-direction near the top to the lateral direction

n e z the horizontal X-Z plane. Fig. 20 shows the enlarged portion of the deformation

of the mode1 at the joint interface. The negative Y-direction (downward) pressure

causes the vertical dimension to decrease? and this deformation is directly related to

the strain measurernent. At the same time, the Poisson's effect causes the X- and the

2-direction dimensions to increase. However, the rate of increase depends on the

modulus of elasticity of the elements; that is, the elements with a lower modulus of

elasticity would expand laterally more than the those with a higher modulus of

eIasticity.

Fig. 21 shows the directions of the principal stresses of each element at its

centroid. Each element has three pairs of arrows representing its three principal

stresses. The arrows pointing away from the centroid of the element represent the

tensile stresses, whereas those arrows pointing towards the centroid of the element
would indicate compressive stresses. Fig. 22 shows the same plot of principal

stresses except the magnitude of the stresses are scaled, in other words, the higher the

stresses, the longer the line benveen the arrows, and vice versa.

From these plots it is clear that those elements near the masonry unit/mortar

interface but inside the mortar have al1 three principal stresses pointing to their

respective centroid of the elements. indicating compressive stresses in al1 three

principal directions. This matches exactly the theory of the equivalent tri-axial

compression stress of the mortar.

The elements just above the unit/mortar interface representing the masonry

units al1 show compressive stresses in the Y-direction parallel to the loading.

However, for the ones near the symmetry planes of the masonry unit, arrows are

pointing away from the centroid of the elements indicating tensile stresses are present

in the hvo principal stresses perpendicular to the loading pressure. When the tensile

stress in these elements exceeds the input criterion of the modulus of rupture, the

elements in the simulation would "crack" and the simulation would stop due to non-

convergence of the triangular matrix. This clearly explains the laboratory observation

that the failure mode of prisms is always the splitting of the masonry units and not the

compressive failure of the mortar.

...... .- ..-.... ..-. ................. - -.-.-.- *

.. - t / s . t;.
,
..
 ANSYS 5 - 3
SEP 9 1998
08:45;32
NODAL SOLUTIC
CiTEP=1
SUB =40
TmE=1
u m
RSYS=O
D m =,O27165
SLZX =,O27165

Fig. 20 Enlarged portion showing the deformation at mortar joint interface.

 AUG 11 199
17:10:57
VECTOR
STEP=l
S U B =40
TIME=1
s

Fig. 21 Principal stresses of elements showing the tensile stress of the

masonry unit nea. the joint interface.
 ; ANSYS 5 . 3
1 SEP 9 1998
08:S2:51
VECTOR
STEP=L
SB = 4 0
TIHE=l
s
PRIN1

PRIN3

Fig. 22 Principal stresses of elements showing the magnitude of stresses with

proportional length of arrows.
 The failure strength of the prism in the simulation is dependent on any of the

following three conditions: it depends on the modulus of rupture of the masonry unit,

the compressive strength of the masonry unit and of the mortar. The computer mode1

would indicate prism failure when the integration points on the elements do not

converge, and this non-convergence will occur when:

a) the compressive stress in the elements exceeds its respective pre-defined

maximum strength of either the mortar or masonry unit, or,

b) the tensile stress of the elements exceeds the input value for the modulus of

rupture of the masonry unit. This usually occurred just above the

unithortar interface.

Using the basic compressive strength of the standard 50 mm cube non-

absorbent specimens, the simulation result will produce a non-convergent result once

the compressive stress of the mortar reaches the input compressive strength data.

This occurs because the computer program interprets this compressive stress as the

"cmshing" stress of the element, thereby generating non-controllable displacements

of the nodes. The program will indicate locaI failure of the mortar, and will halt the

simulation. This will occur once the loading pressure exceeds the compressive

strength of the mortar, even though laboratory testing clearly reports that a much

higher strength can be achieved. As previously stated, earlier research work has

observed that, the failure of a prism is initiated by the splitting of masonry units when

they reach their modulus of rupture stress. The crushing strength of the mortar can be
much higher than that recorded by testing the non-absorptive specimens t'ur two

reasons. First, the monar is experiencing an effective tri-axial compressive pressure;

second, the moisture absorption during setting would increase the mortar strength

considerably. Therefore, in order to allow the simulation to continue until the

phenornenon of splitting of masonry unit could be achieved, with no other data being

available for the effective lateral confining pressure of the mortar, the crushing stress

of the mortar was eventualiy omitted in the simulation.

When the monar tries to expand laterally under vertical loading, and with

good bonding behveen mortar and unit, the net effect is a tensile stress being

developed in the masonry unit at or very close to the joint interface. The computer

simulation ciearly demonstrates this phenornenon in agreement with actual laboratory

testing and the previously reported failure mechanism when the crushing strength of

the mortar is omitted. The result is that the simulation will stop only when the tensile

stress of the elements representing the masonry unit exceeds its pre-determined

modulus of rupture.

6.2 Parametric Analysis

In order to gain a better understanding of the different factors influencing the

ultimate strengths of the prisms that are not norrnally tested in laboratory, different

simulations using different constants were conducted and the results were anaiyzed.

The possible parameters that can be changed are Poisson's ratio of masonry unit,
Poisson's ratio of mortar, the modulus of elasticity of masonry unit, and the modulus

of elasticity of mortar.

6.2.1 Change in Poisson's Ratio

Poisson's ratio of the masonry unit and the mortar is a piece of data that is not

normally tested for and reported in traditional compressive testing, but is important in

computer simulation as it influences the lateral expansion of the masonry unit and the

mortar under compressive loading. Based on previous research results, the initial

Poisson's ratios were chosen as 0.15'' for both masonry units and mortar.

Fig. 23 shows the maximum tensile stress of the rnasonry elements near the

interface was 1.9 MPa as indicated in the red zones of the plot, with Poisson's ratio of

masonry unit being set at 0.20. Fig. 24 shows the maximum tensile stress of the

masonry elements was 2.7 MPa, with every other parameter being constant but

Poisson's ratio of the masonry unit changed to O. 10. These simulations indicated that

when Poisson's ratio of the masonry unit was reduced, the tensile stress of the

elements in the masonry unit, in the horizontal directions, increased. When the prism

is subjected to vertical loading, the masonry units and the mortar will expand laterally

at different rates due to their different moduli of elasticity even if their Poisson's

ratios are the same. Therefore, if Poisson's ratio of the masonry unit is reduced, the

difference in the rate of lateral expansion is further increased causing an increase in

tensile stress of the masonry unit.

Fig. 23 X-axis stress contour with Poisson's ratio of masonry unit being 0.20.
 ANSYS 5.3
MAY 8 1938
14: 59: 16
ELEHEEJT S O L U T
STEP=l
SUB =40
TIHE=I
Sl I NOA'
DMX =. 03161
SMN = - 3 - 826
SMX = Z - 676

Fig. 24 X-axis stress contour with Poisson's ratio of masonry unit being 0.10.
This ultimately reduces the compressive strength of the pnsm as the tensile stresses in

the masonry would reach the modulus of rupture at a much faster rate.

Fig. 25 shows the tensile stress near the mortadrnasonry unit interface with

Poisson's ratio of the mortar being 0.10. Fig. 26 shows the same interface with

Poisson's ratio of the rnortar increased to 0.20. The X-mis stress plots fiom the two

different simulations revealed the tensile stress in the masonry unit increased from 1.0

M'Pa to 3.8 MPa just by changing the value in Poisson's ratio of mortar. This is just

the reverse of the previous parameter. As Poisson's ratio of the mortar increases, so

does the difference in the lateral expansions of the mortar and the masonry units.

This increase of the difference causes the increase of the tensile stress in the masonry

unit.

It can be concluded that the higher the difference of Poisson's ratios between

the masonry unit and the mortar, the faster the prism would fail. This result is

important as this relationship would explain one aspect that contributes to the

variability of masonry structures simply because traditional strength testing of the

masonry unit or mortar simply does not reflect variations of Poisson's ratio of the unit.

Without the proper material data, the commonly obtained data of the compressive

strengths of masonry unit and rnortar are insufficient to understand the complete

behaviour of the prisms.

,LL*ZG-- 1-1
SOI-T-
T 8 b 'T- rn
 BFJSYS 5.3
EAY 15 1938
09:29:56
NODAL SOLUTION
STEP=l
S m =40
m = 1
SX (-%WC)
RSYS=O
DHX =-O30835
SLIN =-6-283
SHX =3-84

Fig. 26 X-axis stress contour with Poisson's ratio of mortar being 0.20.
6.2.2 Change in Modulus of Elasticity

The effect of changes in Young's modulus of the masonry unit was exarnined

next. The result showed that the higher the Young's modulus of unit, the higher the

tensile stress in the unit along the interface. Fig. 27 shows the plot of the nodal stress

in the X-a.xis for rnodulus of unit being set at 54 500 MPa, and Fig. 28 shows the

corresponding plot when the modulus of unit was set at LOO 000 MPa. With the lower

modulus of elasticity, the tensile stress was only 0.17 MPa, while the higher modulus

elasticity caused the tensile stress to increase to 2.4 MPa. The reverse is tme for

modulus of mortar. The simulation revealed the higher the Young's modulus of

mortar, the lower the tensile stress in the Stone along the interface. It appears that the

relative difference between the masonry unit and the mortar caused the increase or

decrease in the tensile stress in the masonry unit. The higher the relative difference,

the higher the tensile stress.

Fig. 27 X-axis stress contour with modulus of elasticity of masonry unit
set at 54 500 MPa.
 ANSYS 5.3
AUG 11 1998
09: 39: 31
NODAL S O L m I i
STEP=1
SVB 140
m = 1
SX (AV(
RSYS=O
D m = - 016473
SIM =-3-448
SEX =2-182

Fig. 28 X-axis stress contour with moduius of elasticity of masonry unit

set at 100 000 MPa.
7 Final Test Program - Phase I
The results of the computer simulation of a masonry prism suggested that the

interface of the mortar and the unit was the key investigation area. The principal

stresses of the elements c h a n p d from tri-axial compression in the mortar, to bi-axial

tension in the masonry just above the joint interface. A test prograrn to investigate the

relationship of the modulus of elasticity of the prism and across the mortar section

was therefore developed.

Since there was an other research program being canied out with Ohio stone at

the National Research Council W C ) , it was decided to use the same Ohio stone to

compare with their findings. Ohio stone is a type of sandstone used quite frequently

in the Ottawa area. This sandstone exhibits a fairly homogeneous behaviour within

the stone units compared with limestone which is more layered and therefore behalves

quite differently in orthogonal directions. The size of Ohio stone unit was chosen

such that both strain gauges and DEMEC points could be mounted o n the stone

surface to obtain Young's modulus of the stone. The recorded value can then be used

to compare with values obtained from other different measuring methods, as well as

with NRC's value. Fig. 29 shows the arrangement of the prisms for the phase 1 of the

final test prograrn.

1 L 1 0 mm fibre

ctr.

gauge in
transverse direction

Fig. 29 Typical prism layout in Phase 1 test program.

7.1 Prism Construction and Testing

The size of the Stone unit was 100 mm x 100 mm x 80 mm; IO-mm length

strain gauges were mounted on opposite faces of the unit. Furthemore, two DEMEC

discs were mounted at 50 mm spacing, also on opposite faces of the same unit, for

direct strain rneasurements using a 50 mm DEMEC gauge. Strain gauge data were

recorded during compressive testing along with its associated loading, with the on-

line data recording computer connected to the compression testing machine. At the

sarne time, the 50 mm DEMEC gauge was used to obtain strains at varies stress levels.

These two sets of strain measurements were compared against each other, as well as

against the results supplied by NRC.

The thickness of mortar joints was chosen as 20 mm to allow for the mounting

of strain gauges across the mortar joints. This relatively thick joint was needed for

the installation of suitable size strain gauges directly on the mortar joints without

interference of the masonry units; 5-mm length gauges with 10 mm base plastic

mounting plates were used throughout. This allowed the strain gauges to have a

minimum 2 to 3 mm spacing between the ends of the strain gauge bases and the

unitlmortar joint interfaces of the prisms. Another pair of strain gauges were mounted

in the transverse direction at the centre of the mortar joint to record the strain in the

orthogonal direction of the loading.

Two different strengths of mortar, one with extremely high strength of 25.5

MPa and the other with relatively low strength o f 5.3 MPa, were selected in this
Phase 1 test program. The mortar mix proportions were obtained from earlier test

results: the high strength mortar of 25.5 MPa was achieved with the mix of 1/0.25/3

(PC/L/S), representing 1 part o f white Portland cernent, 0.25 part of Typs 'S' hydrated

lime and 3 parts of sand, while the low strength mortar of 5.3 MPa was achieved with

a 1/2/8 (PC/L/S) mixture. The mortar mixes and the prisms were prepared and

constructed by an experienced mason to ensure good workrnanship. The prisms were

then wrapped with plastic for moist curing for 7 days, and laboratory condition curing

for another 21 days before the fint compressive testing sequence. The laboratory

conditions were at about 20 to 22O C and 35 to 50 % relative humidity.

Instead of loading the prisms to its ultimate strength and failure, these prisms

were loaded to approximately 25 percent of the estimated ultimate capacity based on

the preliminary data supplied by NRC. The prisms were unloaded prior to repeating

another two cycles of loading and unloading.

The testing was carried out on a Tinius-Olsen 400-kip universal testing

machine in the Structural Laboratory of Carleton University. The loading was

applied at a rate of 0.05 k N per second but halted at every 1.5 kN interval for the 50

mm DEMEC measurements across the mortar joints and on the Stone to be taken. A

200 mm DEMEC measurement for the overall strain of the prisms was also recorded

at the same loading levels.

 Fig. 30 Typical strain gauge locations on mortar specimen.

Photo 13 Different sizes o f mortar specimen with strain gauges.

As mentioned earlier, 5 mm strain gauges were mounted on the mortar joints in

vertical and horizontal directions on the prisms to obtain the stress-strain relationships,

and the approximate value of Poisson's ratio of the mortar. Photo L 1 shows the prism

assemblage with al1 the strain gauges connected into the data-recording computer.

Photo 1 1 Prism assemblages with al1 strain gauges connected to data-

recording computer.
7.3 Test Procedures for Ohio Stone

The properties of Ohio stone were measured using different testing methods

and specimen types. The first test involved compressive tests on 50 mm cube

specimens with two 10-mm strain gauges mounted o n opposite sides to obtain the

stress-strain relationship and the compressive strength of the stone. The second

involved 50 mm x 50 m m x 200 mm sirnply supported beam specimens loaded at two

points to obtain the modulus of rupture of the stone; 10 mm strain gauges were also

mounted on the top compression fibre and bottom tension fibre to compare the stress-

strain relationship of the stone obtained by other methods. The third test involved

split-tensile tests on 50 mm cube specimens to compare the modulus of rupture

obtained from the beam specimens. The fourth involved the mounting 10 mm strain

gauges on the stone units of the prisms to measure the stress-strain relationship of the

stone unit during the compression tests of the prisms.

8 Phase I Test Results

8.1 Propertiesof Mortars

The stress-strain relationships of the mortars are plotted in Figs. 3 1 to 34. The

plots are the average value of the two strain gauges mounted on opposite sides of each

specimen. Fig. 3 1 shows the stress-strain relationship of the 1/2/8 mortar on a 50 mm

cube specimen and Fig. 32 is the relationship on a 100 mm cube specimen of the

same mortar mix; Fig. 33 shows the stress-strain relationship of the 1/0.25/3 mortar

on a 50 mm cube specimen and Fig. 34 shows the relationship on a 100 mm cube

specimen.

For the 1/2/8 mortar, the maximum strength of the 50 mm cube specimen was

5.3 MPa and of the 100 mm cube was 4.6 MPa, representing 86% of the 50 mm cube.

This is consistent with previous findings that the size of cube specimens affects the

ultimate strength of the sarne mortar mix.

 StressStnin Cuwe for 112/8 Mortar for 50 mm Cube

O 500 1O00 1500 2000 2500 3000

Stnin (x IO* mrnlmm)

Fig. 3 1 Stress-strain relationship of 11218 mortar for 50 mm cube specimen.

Fig. 32 Stress-strain relationship of l/2/8 mortar for 100 m m cube specimen.
 Curve for 110.2513 Mortar for 50 mm cube specimen

Strain (x l o J mmlmm)

Fig. 33 Stress-strain relationship of 1/0.25/3 mortar for 50 mm cube specimen.

Fig. 34 Stress-strain relationship of 1/0.25/3 mortar for 100 mm cube specimen.
 For the 110.2513 mortar, the maximum strength of the 50 mm cube specirnen

was 25.5 MPa and of the 100 mm cube specimen 19.7 Mpa, representing 77% of the

50 mm cube strength-

8.2 Properties of Ohio Stone

The compressive strength of the 5 0 mm cube Ohio stone specirnen was

detennined as 58.0 MPa. The modulus of rupture of the Ohio stone was calculated as

2.8 MPa for the two-point simply supported beam testing. This value was in good

agreement with the results from the average value of 2.9 MPa obtained from the split-

tensile tests for three 50 mm cube specimens. Fig. 35 shows the plot of the suess-

Stress-Strain Relationslzip of Ohio stone in sirnpIy supported beam test

Comp. Strain - - - - - - Tens. Strain

-600 -400 -200 O 200 400 600

Strain (x 10-6mmlmm)

Fig. 35 Stress-strain curves of Ohio stone on top andbottom fibre of the simply
supported bearn specimen.
strain relationship of the stone, recorded by the strain gauges that were mounted on

the beam specimen.

The initial modulus of elasticity for the Ohio stone was also obtained as 8.0

GPa. The modulus of elasticity was well within other measuring methods used in this

thesis, as well as within the reported result from M C .

8.3 Prism Test Result

Phase 1 of the test program recorded several different stress-strain relationships

of Ohio stone and rnortars, in addition to the overall relationship for the prisms. Fig.

36 shows the different stress-strain relationships of the prism constructed with the

weak rnortar and Fig. 37 shows the matching stress-strain relationships of the prism

made with strong rnortar. Each figure has three different sets of data representing

DEMEC gauges at different locations of the prisms.

The first set, consisting of 50 mm DEMEC data, was used to determine the

modulus of elasticity of the Ohio stone only; six measuring points were averaged to

plot each data point on the curve. The second set, consisting of 50 mm DEMEC data,

was used to determine the modulus of elasticity of the section across the rnortar joints.

Four data points were recorded for the average value used for each point of the plot.

The third set of data was plotted using 200 mm DEMEC data to measure the overall

strain deformation of the prisms; only two measuring points were available for the

average value used to plot the graph.

Stress-strain curve for weak mortar with 200m Demec & 50mm Demec gauges

n
-C- MOmm Dcmdc for tbc pnsm
AV for 50mm Drn*r for pnwn
+5 i h m k ~ -
on sconc

Fig. 36 Stress-strain relationships of Ohio Stone prism with 1/2/8 mortar.

Stress-strain curve for strong mortar with 200mrn Demec & 50rnm Demec gauges

- - -

Fig. 37 Stress-strain relationships of Ohio Stone pnsm with 110.2513 mortar.

 The forth set of data points were the weighted average of al1 50 mm DEMEC

points as shown on the plots in Fig. 36 and Fig. 37, used to compare the modulus of

elasticity of the prisms obtained from the 200 mm DEMEC points. Both curves

confirmed a good match of the initial modulus of elasticity of the prisms between two

different sets of DEMEC data points. It was decided, therefore, in the phase II of the

final test program, only the measurement across the 200 mm length would be needed

to obtain the overall modulus of elasticity for the pnsms.

These two plots also showed the initial modulus of elasticity of both prisms

were approximately the same regardless of the mortar strengths. Both test results

showed the initial moduli of elasticity of the prisms were in the range of 7 800 MPa.

When compared with the basic modulus of elasticity of the Ohio stone, the initial

modulus of elasticity of the prisms did not appear to be affected by the strength, nor .

the modulus of elasticity, of the mortar.

Six strain gauges were mounted on each of the prisms to record the modulus of

elasticity of Ohio stone. The average modulus of the stone of these six sets of data

was 8.0 GPa. This was in good agreement with the 2-point load simply supported

beam test, and was also in good agreement with the separate testing carried out by the

National Research Council of Canadal2. This good agreement of results led to the

conclusion that strain gauges could be omitted in phase II of the test program.

Therefore, in phase II of the final test program, the measurement of the modulus of

elasticity for the masonry units would be done with the strain gauges rnounted on the
compressive testing specimens only and no strain gauges would be used on the

masonry units of the prisms.

Fig. 38 shows the plots of th: average values of stress-strain relationship of

Ohio stone as well as the vertical and the transverse directions of rnortar, using the

strain gauge data recorded during the compression testing of the pnsm made with

1/2/8 mortar. The modulus of elasticity of the mortar is interpreted as about 5.5 GPa,

and together with the transverse direction strain data, the value of Poisson's ratio is

calculated as about 0.14; this value is also in good agreement with the value of O. 15

used in the Finite Element Analysis simulations.

Fig. 39 shows similar plots of the strain gauge data of the Ohio stone prism

made with the U0.2513 mortar. The rnodulus of elasticity of Ohio stone was again

consistent with other tests conducted. However, the modulus of elasticity of the

mortar for this strong mortar was interpreted as 14.0 GPa with Poisson ratio recorded

as approximately 0.14.

The moduli of elasticity of these two different mortars are very different.

Furtherrnore, the values also differ quite significantly from the values obtained from

the compressive testing of their respective non-absorptive cube specimens.

 Stress-straia relationship of Ohio Stone and 1/2/8 mortar

- - - Vertical strain gauges on rnortar

---
--- --
Horizontal strain gauges on monar
. . --
- - -- - --- - .- -- - - - - .. - .. - - . ...- -

200 400 600 800 1000 1200 1400 1600

Strain (x 10" mmlmm)

Fig. 38 Average stress-strain relationships of Ohio Stone and 1/2/8 mortar in

vertical and horizontal directions.
L

Stress-strain relationship of Ohio stone and 1/0.25/3 mortar

-- - - - - - - - - -
Vertical strain gauges on Ohio Stone
- - - Vertical ~traingauges on rnortar
--- Horizontal ~traingauges on mortar
- - - - . --

r
c3

w
Vi
rn
2
G

-500 O 500 1O00 1500

Strain (x 1o - m
~ rnhm)

Fig. 39 Average stress-strain relationships of Ohio stone and 1/0.25/3 morta. in

vertical and horizontal directions.
 This is to be expected as it is well documented that the mortar strengths of

absorptive specirnens, and in the joints, are always higher than those of the non-

absorptive specimens. This response is further explained in detail in the recent

published papers authored by Stockl, Biewirth, and Kupfer4, and Stockl, Beckhaus

and ~ritsche'. They reported that not only the compressive strength measured in the

actual mortar joints can be four times that of the strength obtained from regular non-

absorptive specimens, the modulus of elasticity of the same mortar can differ up to

ten times depending on the testing method chosen.

They have suggested multiplying the standard test results obtained in

accordance to the German Standard D M 18555 by four to estimate the compressive

strength of the mortar joint; in this they have accounted for the effect of the water

absorption process of' the brick unit, and of the friction on the ngid loading platen.

However, they still have not proposed any formula for estirnating the effective lateral

confinement pressure experienced by the mortar.

9 Final Test Program - Phase II

9.1 Mortar Mixes and Masonry Units

After the results of the phase 1 test data had been reviewed, it was decided that

the range of the mortar strengths should be 2.0 MPa for the fow-end, and about 15

MPa for the high-end strength. These values are closely related to the Type "O"

rnortar for the low strength and Type "MWmonar for the high strength mortar. Type

"K" mortar was not being considered as it would be a purely "lime/sand" mortar mix
and not likely to be representative for modem masoruy work.

Since the low-strength of the masonry unit was to have a low compressive

strength and most sandstones have compressive strengths of 65 MPa and higher, the

decision was made to select a solid restoration style clay brick. Afier searching '

several sources, it was decided to choose an American restoration brick with an

average compressive strength of about 30 MPa. This type of brick was newly

manufactured to the sizes and shapes of older bricks for the restoration of heritage

buildings, hence the name of restoration brick. As a high strength masonry unit of

about 100 MPa, a sandstone flagstone from a nearby quarry from the Province of

Quebec was selected. These two strength levels provided good lower and upper

strength limits for the masonry units.

9.2 Properties of Mortar Mixes

Further trial mixes of mortar with different compositions were prepared and

tested prior to the final composition being decided. The final mix for the low strength

mortar was 1/3/10 (PC/L/S), with a waterhinder ratio of 1.2. This mix produced an

average of 2.1 MPa compressive strength at 28 days and would be referred as the low

strength mortar. The high strength mix was 1/0.5/4 (PC/L/S), with a waterhinder

ratio of 0.9. This rnix produced an average of 14.4 MPa compressive strength at 28

days and would be referred to as the high strength mortar. As it was not the

controlling factor of failure for the prisms, the modulus of rupture for the mortar was

no longer tested in the phase II program. The mortars were mixed in a wheelbarrow

by a professional mason, who measured the portions of each component into the

wheelbarrow, dry mixed the mixture, and then added the pre-determined amount of

water slowiy while rnixing the mortar.

9.3 Construction of Prisms

For compati bi lity reasons, the dimensions of the flagstone units were chosen

as 90 mm x 90 mm x 30 mm to closely match the sizes of a quater of a brick. To

ensure good quality of cutting, both flagstone and restoration bricks were cut by a

local commercial cutting facility in Ottawa. The final dimensions of the restoration

bricks were 90 mm x 90 mm x 27 mm thick due to the thickness of the cutting blade.

The mason mixed one batch of mortar at a tirne, using the wheelbarrow

available at the laboratory as shown in Photo 12. Both the fiagstone prisms and the
restoration brick prisms were constnicted using the same batch of mortar. The

procedure was then repeiited for the other mortar. Six prisms for each combination

were constructed al1 in the same day.

The prisms were eight units high, with standard 10 mm joint thickness. In

order to ensure good quality of workmanship, the same professional mason who

prepared the mortar was retained to constmct these prisms. A jig marked with the

height of each unit was used throughout the construction of prisms to ensure

uniformity of the joints. Fig. 40 shows the typical arrangement of prisrns while

Photos 13 and 14 show typical prism construction of flagstone and restoration bricks.

Photo 15 is a close-up view of several joints of the flagstone prism to show the

moisture being drawn fiom the mortar into the flagstone. This moisture absorption

process enhances bonding benveen the mortar and the masonry units. However, this

same process also alters the properties of the mortar in the prism assemblages.

The prisms were then stored in the laboratory wrapped in plastic for 7 days of

moist curing, and then under laboratory conditions of about 20" C and 30 to 50

percent relative humidity for another 20 days.

One prism from each combination was trial tested on the Wh day after

construction to record the approsirnate failure loading of each combination. This

information was used to predict the final failure load of each combination of prisms.

The final testing of prisms was carried out on the 28" or 29'hday afier construction.

Since the properties o f the masonry units do not change, but the mortar strength will
increase over time, the prisms with the same mortar mix were tested on the same day.

That is, al1 ive stone prisms with weak mortar and al1 five brick prisms with the same

mortar were tested on the 28" day, and al1 ten prisms with strong monar were tested

on the 2gthday.

9.4 Test Procedure

The prisms were tested in the 400-kip universal testing machine at the

Structural Laboratory of Carleton University with two sets of LVDTs mounted on

frames. These metal frames were custom-made by a local tool and die Company to

ensure precise a l i m e n t of anchoring points and therefore accurate LVDT recording.

Fig. 41 shows the schematic arrangement of prisms and the location of LVDTs for the

recording of the stress-strain relationship of the prism, and Photo 16 shows the

mounting frame with the LVDT already in place. The mounting frames were spaced

consistently at 200 mm by inserting a set of fixed length timber spacers behveen

frames for each prism as shown in Photo 17. Each pnsm was then tested with 12 mm

fibreboard behveen end platens of the testing machine as shown in Photo 18. Each

prism was subjected to the same loading rate of 0.05 kN per second to ensure a

consistent loading rate for each prism. The loading and LVDT displacements were

recorded at every second interval during the testing period of each prism, with the

cornputer program for LVDT recordings pre-programmed to have 200 step

increments per inch accuracy.

Photo 12 Photograph showing the mortar mix was dry-mixed thoroughly
before water was added.
 Applied Load

Location of measuring
point using LVDT's

Location of measuring ,
point using LVDT's

Reaction Table

Fig. 40 Typical prism arrangement in Phase II test program.

Photo 13 Typical construction progress of prism with flagstone.
Photo 14 Typical construction progress o f prism with restoration bricks.
Photo 15 Close-up view of mortar joints showing the moisture absorption process.
 onar

Point

l
-Fibre board
Reaction Table

Fig. 41 Schematic arrangement of LVDTs on the test prism for strain measurement.
Photo 16 Typical arrangement of LVDTs mounted on the special frames.
Photo 17 Fixed length timber spacers are used to ensure constant LVDT spacing.
Photo 18 Prism arrangement in loading machine.
10 Phase II Test Results

10.1 Properties of Masonry Units

The primary properties of the masonry units used in the final testing program

are presented in the next sections.

10.1.1 Restoration Brick

The compression tests were carried out with the Tinius-Olsen 400-kip

universal testing machine in the Structural Laboratory of Carleton University. The

compressive strength of the restoration brick was 33.0 MPa obtained by averaging the

compressive test results of three 5Omm cube specirnens. The modulus of elasticity of

the brick was 14.0 GPa calculated from the stress-strain data recorded with the strain

gauges mounted on the brick specimens during compressive stren-g$h testing. Fig. 42

shows a typical plot of the stress-strain relationship of the brick. The tensile strength

of the brick was 4.0 MPa calculated from split-tensile tests on two 57 mm x 90 mm x

90 mm specimens. This represents 12% of the compressive strength of the brick.

Fig. 42 Stress-strain relationship of restoration brick.
 Flagstone

The ultimate compressive strength of this particular flagstone was averaged at

103.0 MPa from compressive testing of three 50 mm cube specimens. The modulus

of elasticity was interpreted as 36.3 GPa obtained from strain gauge data on one

specimen only. Fig. 43 shows a typical stress-strain relationship of flagstone used in

the final testing program. The modulus of rupture, on the other hmd, was calculated

from split tensile testing of three other 50 mm cube specimens, and the average result

was 14.7 MPa, representing 14.3% of the compressive strength.

Stress strain relationship of flagstone

2000 3000

Strain (x 1'0 mm/mm)

Fig. 43 Stress-strain relationship of flagstone unit.

10.2 Properties of Strong Mortar

The 28-day compression test result of the strong mortar (U0.94)revealed that

the average compressive strength of six 50 mm cube specimens was 14.4 Mpa; this

result was confirmed by compressive testing of 75 mm diameter by 150 mm high

cylinders. The stress-strain relationship of the cylinder testing is shown in Fig. 44 and

the ultimate strength was 13.7 MPa. The modulus of elasticity of the mortar was

interpreted using this stress-strain relationship recorded with the LVDT mounted on

the cylinder specimen during the compression test. Photo 19 shows the testing set up

of the LVDT on a cylinder specimen; the average modulus of elasticity was

deterrnined as 10.8 GPa.

10.3 Properties of Weak Mortar

Fig. 45 shows the stress-strain relationship of the weak mortar; based on 75

mm diameter by 150 mm high cylinders, the compressive strength of this mortar was

1.9 MPa while its Young's modulus was interpreted as 8.0 GPa. The average

compressive strength of six 50 mm cube specimens was, however, averaged as 2.1

MPa.
 Stress-strain curve for 1 /OS14 mortar

Strain (x 1o4 mm/mm)

-

Fig. 44 Stress-strain relationship of strong (1/O.5/4) mortar.

Stress-strain curve of l/S/I O mortar

4000 6000

Strain (x 1 0" m m h m )
-

Fig. 45 Stress-strain relationship of weak (l/Y 10 ) mortar.

Photo 19 LVDT mounted on 75 mm diameter x 150 mm mortar specimen.
10.4 Compressive Test Results of Prisms

The final test resutts of the prisms are plotted in Figs. 46 to 49. The computer

prograrn used in the collection of data had fixed increment steps for the LVDT

displacements. The accuracy of the LVDT recording by the cornputer was limited to

200 divisions per inch, Le. a continuous interpretation of the displacement of the

LVDT was not possible. This resulted in data points which were tmncated down to

the lower increment or rounded up to the next higher increment, and caused slight

zigzagging of data points along the curves.

Fig. 46 shows the plots o f the stress-strain relationships of the prisms for the

flagstone with strong mortar. The average compressive strength of the prisms was

4 1.O MPa and the average modulus of elasticity was 11.3 GPa. Al1 five stress-strain

curves were extremely similar and so were their ultimate strengths, indicating very

low variability among these prisms.

Fig. 47 shows the stress-strain relationships of the prisms for the flagstone and

weak mortar. The compressive strength of the prisms \vas averaged as 34.0 MPa, and

the associated initial moduli of elasticity were interpreted as between 4.8 GPa and

10.2 GPa with a mean value of about 7.4 GPa. Although al1 five stress-strain curves

were also very similar with very low variability among these prisms, the

interpretation of the modulus of elasticity could be very subjective depending on the

individual engineer. The stress-strain curves of these prisms did not exhibit a typical
ascending and slow descending before failure; they followed a fairly consistent

smaller modulus of elasticity once the stresses passed beyond 6.0 MPa. FinalIy,

sudden rupture took place at fairly similar ultimate failure stresses.

The average ultimate strength of the brick prisms with strong rnortar was 13.8

MPao and the moduli of elasticity could be interpreted between 4.3 and 7.1 GPa with

an average of 5.7 GPa as shown in Fig. 48. Prism 1 was discarded in the averaging

due to a pre-cracked condition in one of the bed joints. The bottom sections of the

stress-strain curves were discarded in the determination of moduli.

The average compressive strength of the brick prisms with weak mortar was

1 1.O MPa, and the moduli of elasticity were between 2.8 GPa and 5.7 GPa depending

on the interpretation of the initial tangent of the curves as shown in Fig. 49. This set

of data was the least consistent in terms of their ultimate strengths and their moduli o f

ehsticity, although the average was interpreted at 4.3 GPa.

 Table 1 summarizes the average compressive strengths of prisms, Table 2

summarizes the average moduli of elasticity of prisms, and Table 3 presents the range

of moduli of elasticity of prisms interpreted.

Table 1 Average compressive strengths of prisms

Compressive Strong % of Unit Weak % of Unit
strengths of prisms rnortar strength mortar strength
(Fm) (MW (%) (MW
Flagstone 41.0 39.0 34.0
Restoration brick 13.5 40.9 11.5

Table 2 Average moduli of elasticity of prisms

Strong % of Unit Weak % of Unit
Moduli of elasticity monar mortar modulus
modulus
of prisms (E,)
(GW (%) (GPa) ("A) 1

Flagstone 11.3 31.1 7.4 20.4

Restoration brick 5.7 40.7 4.3 30.7

Table 3 Range of moduli of elasticity of prisms

Strong mortar Weak mortar
Moduli of elasticity
Low High Low High
of prisms (E,)
(GPa) (GW (GPa) (GW
Flagstone 11.3 4.8 10.2
Restoration brick 4.3 7.1 2.8 5.7
Fig. 46 Stress-strain curves for flagsstone pnsms with strong mortar.

Stress-strain curves of stone pnsms with 1/3/10 mortar

0.004 0.006

Strain (mrnhm)
- - --

Fig. 47 Stress-~traincurves for flagstone prisms with weak mortar.

Fig. 48 Stress-strain curves of brick prisms with strong mortar.

Stress-strain curves of brick prisrns with 1/3/10 rnortar

0.004 0.006

Strain (mm/mm)

Fig. 49 Stress-strain curves of brick prisms with weak mortar.

 Photos 20 and 21 show two typical failed brick prisms, and Photos 22 and 23

show some typical failed stone prisms. In al1 cases, the splitting of the masonry units

is very clear, agreeing with the reported failure mechanism of masonry prisms. Photo

23 frther captured the instant of the failure of a prism. The sudden release of energy

caused the near explosion of the stone units.

Photo 20 Typical brick pnsm faiiure mode.
Photo 2 1 Typical brick prism exploding at failure.
Photo 22 Typicai flagstone prism ilure mode.
Photo 23 Typical Stone prism explosion at failure.
II Discussion of Results

11.lPhase I Testing

Since the phase 1 test program was only designed to help arrive at the

parameters for the phase II test program, there were an insufficient number of prisms

built to have any statistical reliability. However, the knowledge gained was still

valuable in designing the phase II test program.

First of all, the modulus of elasticity o f the masonry units was consistent

behveen different test methods. Therefore, it was not necessary to use strain gauges

on the masonry units on the prisms in phase II testing to obtain the modulus of

eiasticity of the units. The modulus of elasticity of masonry units could be

determined from simple compressive testing of units, with strain gauges mounted on

the test specimens.

Secondly, the determination of the modulus of elasticity of mortars proved to

be much more complicated than this testing program could measure. Therefore, the

attempt to measure the modulus of elasticity of the mortar in the prisms was omitted

due to lack of suitable instrumentation and equipment available at the time of testing.

This phenomenon was later confirmed by the German researcher ~ t o c k l ' . ~

after the

testing program for the thesis was completed. Stock1 et al. concluded that there

appear to be three different zones across the vertical plane of the mortar joint in a
prism. The top zone, the middle zone and the bottom zone each had its own distinct

properties in tenns of their compressive strengths and moduli of elasticity. They

compared the results by testing the bed joint mortar taken from the actual prism and

loaded with steel brush platen to eliminate the frictional effect of the platens and

effects from non-absorptive specimens.

It was due to unique "clipsn utilised in their testing program they were able to

obtain the modulus of elasticity of the mortar across the bed joints in the prisms.

They concluded that the modulus of elasticity of the sample taken out from bed joint

mortar was 5 to 15 times higher than the results obtained from specimens prepared

using regular non-absorptive moulds for the sarne mortar mix. This explains the

unexpected result of the phase 1 test prism with high strength mortar which actually

had a higher modulus of elasticity in the mortar joint than that of the overall prism.

The test prism with low strength mortar was so low in compressive strength, the

modulus of elasticity of the mortar in the bed joint actually exceeded the overall value

of the prism.

Thirdly, the use of 200 mm DEMEC gauges for the measurement of the

modulus of elasticity of prisms was confirmed by the weighted average of the 50 mm

DEMEC gauges. The results obtained using these two different DEMEC gauges were

exactly the same; i.e., the 200 mm DEMEC gauges proved to be just as effective as

the weighted average of 50 mm DEMEC gauge measurernents in measuring the

overall modulus of elasticity of prisms. Therefore, LVDT's with proper mounting

frames with anchoring points spaced at 200 mm could be used to obtain the modulus

of elasticity of the prisms.

Fourthly, the modulus of rupture value of the masonry units was essential for

the computer simulation, but neither the compressive strength nor the modulus of

rupture of the mortar played a role. The modulus of rupture of masonry units was a

piece of important data needed in the computer simulation to detennine the failure

load of the prisms, but the ultimate compressive strength of mortar derived from the

non-absorptive moulds was not needed.

11.2 Cornputer Simulation Results

The computer simulation provided a valuable tool in understanding the

behaviour of the masonry pnsms, especially when different parameten could be

entered to simulate different situations. Unfortunately, however, there are just not

enough research data available to fully utilize the finite element analysis program to

understand the behaviour of the mortar and unit interface in the prisms. In their

separate respective tests, Beckhaus and Fritsche2' measured the modulus of elasticity

with the same mortar type in the bed joints of prisms and found startling results when

compared with the current German code. Beckhaus found the rnodulus was averaging

4500 MPa while Fritsche found the rnodulus was between 700 to 1200 MPa

depending on whether the voids in the bricks were taken into account or not.

Compared to the estimated value of 11 000 MPa that the German code allowed for

that combination of brick and mortar, Beckhaus and Fritsche's results were extremely
low. They representing only approximately 35% to 10% of the code provision, thus

making the computer simulation in this thesis very difficult as there are simply too

rnany unknown variables. Until a better understanding of the properties of mortar in

the actual bed joint is achieved in ternis of its strength and modulus, simulating the

true behaviour of prisms in cornputer programs is virtually impossible.

The A N S Y S finite element analysis program has the capability of simulating

crushing failure of the elements, and the program checks the stress of every element

to find if any reaches the crushing stress that was input. When the element reaches the

crushing stress, the program would interpret that element has cracked. Since the

computer program does not have the capability to account for the effective lateral

compressive pressure provided by the masonry units, the increase in crushing strength

of the mortar is not being accounted for in the simulation. Thus when the elements in

the mortar reach the specified crushing stress, the program would interpret that the

mortar has failed, thus prematurely halting the simulation process. The crushing

strength of the mortar \ a s eventually deleted altogether from the input data in order

to achieve more realistic prism strengths.

The moduli of elasticity for masonry units and mortars were needed to

simulate the deformation of the model. The modulus of rupture of the rnasonry unit

was needed to simulate the splitting (cracking) failure of the model. Since the

elements did not have any reinforcing steel, any "cracking" of an element would

cause non-convergence failure of the program. Therefore, as far as the simulation

using finite element analysis program is conceme, the modulus of rupture of the unit

is the most important piece of data required.

However, it was clear from the simulation results that it was the combination

of the properties of masonry unit and mortar that determines the strength and the

modulus of elasticity of the prism. Isolating the properties of individual elements in

the pnsm simply did not yield rneaningful results. This emphasizes the need to devise

testing prograrns to give proper input data for the cornputer simulation process to

work properly in the future.

1f .3 Phase Il Testing

The compressive strength o f the prisms in the phase 11 test program depended

mainly on the tensile strength of the masonry unit, but not so much on the strength of

the mortar. The prism always failed by tensile splitting of the masonry units in

laboratory testing; however, the tensile strength of the masonry unit can be correlated

with the compressive strength of the units. The flagstone prisms had average

compressive strengths between 34.0 to 41.0 MPa with the associated low and high

mortar strengths of 2.1 M'Pa and 14.4 MPa. The ratio of the high strength to low

strength mortar is 685%, but the increase of the strength of the prisms is only 20.6%.

The low strength restoration brick prisms exhibited a similar relationship. The

average compressive strengths o f the brick prisms were avenging 1 1.5 MPa with the

low strength mortar and 13.5 MPa with the high strength mortar. An increase of only
14.7% in prism compressive strength was obtained while the mortar strength

increased 685% as noted before.

The strengths of the prisms represent 33.0% of the flagstone unit strength

when low strength mortar was used and 39.8% when high strength mortar kvas used;

similarly, the strengths of the prisms represent 35.5% of the strength of the restoration

bricks with the low strength mortar and 40.9% with the high strength mortar.

This relative minor influence of mortar strength on pnsm strength is in

agreement with the statement made by S. I. Lawrence, and G. T. caoI6 that "there is

n general misconception among practitioners that increasing the quantity of cernent

will reszrit in higher strength. " There is the same general misconception that a

stronger mortar will result in a stronger structure. In order to increase the strength of

the mortar, more cement is usually added to the mortar. This will not only reduce the

workability of the mortar, it will also increase the shrinkage of the mortar thus

increasing the risk of separation cracking behveen the mortar and the units at both bed

joints and head joints. This separation cracking, if developed, would be harmfl to

the structure as water may penetrate into the joints. Freeze-thaw problems and other

water related problems could occur. Therefore, the compatibility between mortar and

masonry unit is much more important than mortar compressive strength alone.

Compatibility such as water absorption rate of masonry units and workability of

mortar to ensure good bonding is much more important than the increase of the

strength of the mortar.

 Furthemore, the current Canadian standard" for unit strengths of over 90

MPa and Type S rnortar, provides for a strength of 25 MPa as specified in Table 3 of

CSA S304.1-M94, the prisms built with flagstone and the strong mortar achieved 41

MPa which is well in excess of the CSA standard's 25 MPa requirement. Even with

the low strength mortar, the stone prism strength was averaging at 34 MPa. Note

however, that CSA requirernent is intended for solid bricks. The Standard recognizes

that stone units may not behave exactly the same as solid bricks and requires the

designer to carry out prism testing to find the appropriate compressive streagth. The

same holds true for the prisms built with restoration brick; the interpolation of Table 3

from CSA S304.1-M94 leads to a specified compressive strength of prisms of 11.5

MPa for 33 MPa bricks and Type S rnortar. The test results of the brick prisms made

with low strength mortar, which had a compressive strength of less than 2.0 MPa,

aIready achieved this strength. This shows the current Canadian Standard is

conservative in ternis of assessing the compressive strength of the prisms. This

consemative assessment of the ultimate strength of the structure is necessary for the

safety of the structure when no prism testing program is conducted for a particular

project. Table 4 summarizes the measured results in this thesis and the CSA

provisions.
 I

Table 4 Compression strength cornparison of CSA estimations and Phase II test

results

P r i m Combination
1 CSA estimates / results
test 1 Difference 1
(MW (MPa) (%)
Strong mortar 25.0 4 1.O -39.0
F1agstone ~ e a rnortar
k NIA 34.0 NIA
1 I 1

Restoration Brick
Strong mortar 11.3 13.5 - 16.3
rnortar N/A 11.5 N/A

By testing prisms with either DEMEC gauges or LVDTs, the overall

deformational behaviour of prisms can be studied, but this information still does not

yield a proper understanding of the behaviour of the mortar at the interface. In the

phase 1 testing, the modulus of elasticity recorded by the strain gauges mounted in the

mortar joints of the 1/0.25/3 mortar actually showed a higher modulus than the Ohio

stone, and therefore, the prisms. This phenornenon can only be understood with the

aid of Stockl's recently published paper. Further detailed studies of the bed joint

rnortar must be developed in order to gain a better understanding of the mortar joint

behaviour.

Unlike the strength of the prisms, which were between 30-40%of the strength

of the masonry unit, the interpretation of the moduli of elasticity of the prisms was

much more subjective. In many cases, the interpretation could be made within a wide

range and a single value of modulus for that particular combination of masonry unit

and mortar could not be assigned without knowledge of the variability. The

variability increased as the compressive strengths of the individual components

decreased. The least variable set of data for the modulus was found in the high

strength unitktrong mortar combination, and the greatest variable set was found in the

low strength unitheak mortar combination. However, the average initial modulus of

elasticity appears to be the in the same range for a given masonry unit regardless of

the mortar strengths.

The wide range of modulus of elasticity of prisms clearly indicates the

difficulties researchers face in their attempt to determine the prism modulus

accurately. As discussed earlier, the strengths of mortars increased quite significantly

in the prisms compared with the traditional cube strengths from non-absorptive

specimens. An increased strength of mortars would inevitably increase the modulus

of elasticity of the mortar.

Stock1 et. al. concluded that different specimen types or testing methods would

produce different moduli of elasticity for the sarne mortar. The rneasured values of

the modulus of elasticity of mortar made with non-absorptive moulds were very much

different from the results obtained from testing the actual mortar joints. The -

difference was attributed to the moisture absorption process of the monar in the joints

of the prism, and it was this difference which caused the basic properties of the mortar

obtained from the non-absorptive moulds to be virtually useless.

Although water absorption appears to be a simple process, it depends on many

different factors. Firstly, it depends on the water content of the mortar. Obviously if

the mortar is too dry, there is an insufficient amount of water that c m be absorbed by
the units. On the other hand, if the mortar is too wet, the mortar wiii be too sofi even

after some of the water is absorbed by the units. Water absorption also depends on

the rate of absorption of the masonry units, the environmental conditions such as

temperature and relative humidity at the job site, and moisture lost by evaporation. C.

~ e a l l "explains these factors in detail in her book "Masonry Design and Deraifing".

The extra strength gained by adding too much Portland cernent to the rnortar

may even have an adverse affect on the overall performance of the prism. An

excessive amount of cement usually reduces the workability of the mortar and thereby

increases the chance of poor workmanship in both head and bed joints. Furthemore,

even if the joints were forrned properly, the shrinkage due to an excessive cernent

content might cause separation cracking in the joints. A11 these would contribute to a

worse overall performance of the structure than with relative low strength mortar but

good bonding between the mortar and the masonry units.

The modulus of mortar itself was extremely difticult to measure accurately. L.

Binda et al." also stated that the measurement of the mortar modulus was difficult

using conventional testing methods and hence developed a totally different method to

measure the modulus of elasticity of masonry prisms. Their use of electronic speckle

pattern interferometry (ESPI) offered good agreement with the strain gauge

measurement on mortar specimens, therefore this technique may become a useful tool

in measuring the modulus of elasticity of mortar in view of the fact that direct

measurements were so difficult.

 J.J. Brooks and B.H.Abu Baker" proposed a simple estimating formula for

the modulus of elasticity for clay brick prisms. Using their formula, the estimated

modulus of elasticity for the brick/strong mortar prisms would have been 1 1.9 GPa

assuming 10% water absorption by the bricks. The measured range for this

combination in this thesis was between 4.0 to 6.0 GPa; this represents close to a 295%

and 198% over-estimation of the actual measured values. When compared with the

bricWweak mortar prisms, the estimating formula would have given the modulus of

the prisms as 5.2 GPa, which lies within the measured range of between 3.8 GPa to

6.4 GPa. Appendix B shows the calculation of the moduli of elasticity of different

bricWmortar combinations. Stone prisms were not compared as the proposed formula

was developed for brick prisms. However, the large error percentage for the

bricldstrong mortar combination further confirms the need for more systematic

research on the modulus of elasticity of masonry.

The current Canadian standard" allows the moduli of elasticity be estimated

by multiplying the compressive strength of prisms by 850 for clay bricks. Using this

estimating method, the modulus of elasticity of the prisms with the restoration

bricldstrong mortar combination would be 11.5 GPa (850 x 13.5MPa) compared to

the test result of 5.7 GPa, and 9.8 GPa for the bricldweak mortar combination with the

measured result of 4.3 GPa. The CSA does require prism testing be done for

estimating the modulus of elasticity for Stone masonry structures. Table 5 shows the

cornparison of the average moduli of elasticity of prisms from the phase II test results

and the CSA estimations. The Canadian Standard's estimating method is clearly not
 very valid as it does not take into account the modulus of elasticity of either the

mortar or the masonry unit.

TabIe 5 Moduli of elasticity cornparison of CSA estimations and Phase II test

v
resul ts
Phase II test
CSA estimates Di fference
Prism Combination results
(GW (GW (%)
Strong mortar N/A 11.3 NIA
Flagstone
Weak mortar N/A 10.2 N/A
Strong mortar 11.5 5.7 101.8
Restoration Brick
Weak mortar 9.8 4.3 127-9

Maurenbrecher and ~rischuk"conducted many tests on repointing mortar with

Ohio sandstone, and in their findings the average strength of the pnsms was 26.5 MPa,

approximately 40% of the Ohio's average compressive strength of 60.0 MPa. The

averaged modulus of elasticity of the prisms was 5.5 GPa from over 15 different

batches of rnortar. If the Canadian Standards estimation were to be used, the

modulus of elasticity would have been 22.5 GPa (850 x 26.5 MPa), almost four times

higher than the actual measured results.

Thus, more innovative testing methods for the measurement of the modulus of

elasticity of mortars and prisms are needed to improve the estimation accuracy of the

modulus of elasticity.

The effect of the joint thickness has not been studied in this thesis as it has

been documented by A. I. Francis, C . B. Horman and L. E. J e r r e r n ~that

' ~ the thicker
the joints, the lower the strength of the prism; the current Canadian Standard to some

extent recognizes this effect as the standard joint thickness is specified as 10 mm.
12 Conclusions

With the relatively recent development of high strength concrete, it is tempting

for designers to '3ump" to the conclusion that a higher monar compressive strength

will automatically yield a higher overall strength of an element in masonry structures.

With the results of this research, however, it is clear that this may not necessarily be

true. Based on the results of this research, the following conclusions c m be drawn:

1. The compressive strength of prisms is influenced very little by the strength of the

mortar. When the compressive strength of the mortar was increased 685%, the

increase of prism strengths were only 20.7% and 14.7% for stone prisms and brick

prisms respectively. Therefore, the structural engineer should specify the proper

mortar for good compatibility with the masonry units, good workability, suitable

water retention, and suitable air content, without umecessarily over-designing the

strength of the mortar.

2. The influence of the mortar strength on the modulus of elasticity of prisms is

again very small. The brick prisms had virtually the same initial tangent modulus

of elasticity for both high and low strength mortars. Furtherrnore, the initial

tangent modulus of elasticity for the stone prisms were also very similar

regardless of mortar strengths, although modulus for the prisrns with low strength

mortar started to fa11 off quite rapidly at a relatively low stress level. This thesis
 shows that the modulus of elasticity of prisrns is mainly controlled by the

properties of the masonry units.

3. The interpretation of the moduli o f elasticity for the prisms is very subjective.

Until there is a reliable method of measuring the properties of the mortar in the

prism, the interpretation will always be subjected to a high degree of variability

even when al1 the individual components rernain unchanged.

4. The process of water absorption during laying and curing reduced the

waterhinder ratio of the mortar and significantly changed the properties of the

hardened mortar in the prisms. The relatively minor increase in compressive

strength and the similarity of the moduli of elasticity of the prisms suggests that

the properties of the hardened mortars may be very similar despite their difference

in properties when tested using non-absorptive moulds. The true properties of the

hardened mortars remain very difficult to determine.

5 . The computer simulation confirmed the flow of forces and stresses within the

prism and the cracking mechanism of the pnsm in agreement with laboratory

observations. The computer simulation confirmed that the modulus of rupture of

the masonry unit is the most important piece of information required in initiating

the failure of prisms.

6. The computer simulations confirmed that the critical properties of the mortar are

Poisson's ratio and modulus of elasticity, and not the compressive strength of the
 mortar. As observed from laboratory testing, the failure mechanism of the prisms

is always initiated by the splitting of the masonry units. This observation led to

the omission of the mortar's compressive strength in order to simulate the proper

failure mechanism. Therefore, the compressive strength is irrelevant as far as the

computer simulation is concerned, at least until the effective confining pressure of

the mortar can be quantified.

7. Other important factors for the computer simulation are the relative differences of

the Poisson's ratio between masonry units and rnortar, and the relative difference

of the moduli of elasticity between masonry units and mortar. The Iarger the

difference of their relative Poisson's ratios, the lower the compressive strength of

the prism. The same holds m e for modulus of elasticity, i.e. the larger the

difference, the lower the strength of the prism.

8. The lack of measuring techniques of the mortar properties in the bed joints makes

it difficult to assign proper values for a computer simufation. Any attempt to

assign a simple value to the mortar would be wrong. Although the mortar is

understood to be experiencing uniaxial compressive stresses during a simple

compressive loading of a prism, the effective confining pressure of the mortar

cannot be assigned as there is no quantifiable value available at the present time.

9. Finally, the current CSA Standard, similar to other countries' standards and the

Australian Standard reported by Page and Brooks6, underestimates the strength of

the prism assemblage with certain masonry unithortar strength combinations to

 ensure safe design of stmctural elements, yet generally overestimates the modulus

of elasticity of the prisms. While the CSA Standard restricts the use of the simple

estimating formula for the modulus of elasticity of masonry to clay bricks and

concrete unit, the Standard's formula does not take into account either the

modulus of elasticity of the unit or the mortar. The current work has shown that a

better understanding of the mortar joint properties is needed to accurately estimate

the modulus of elasticity of prisms.

Future research is recornrnended in the following areas:

1. Since the properties of mortar in the prisrns are changed significantly by the

moisture absorption process, there is a need for better testing procedures to

measure the properties of mortar properly. This includes the measurement of the

compressive strength as well as the modulus of elasticity of the hardened mortar.

2. Although it is generally accepted that the mortar in the prism assemblage is

subjected to tri-axial compressive pressures when the prism is loaded uni-axially,

there appears to be little if any research undenvay to quantify the equivalent

lateral confining pressure. This information, if avaiiable, will definitely enhance

the understanding of the prism behaviour by either cornputer simulation or

laboratory testing.

3. Poisson's ratio of both the masonry units and mortars are also important to

determine the uItirnate strength and modulus of elasticity of prisms in the

 cornputer simulation. Therefore funire research should include the study of the

variability of Poisson's ratio especially within a hardened mortar joint close to and

away from the unit interface.

4. The recording of strain using LVDT's was carried out with an accuracy of 200

divisions; this resulted in al1 the stress-strain curves not being smooth but rather

"toothed". The cornputer program used for data recording should be refined to

achieve better accuracy of the strain data.

1 Appendix A

Materials used in final test prograrn Phase I -

Masonry Unit:
Ohio stone
S ize: lOOmm x lOOmm x 80 mm

Supplier: Plouffe Park, Public Works Canada Materials Depot

Cernent:
Manufacturer:Federal Cernent Ltd.
Type: Type 10
Colour: White

Lime:
Manufacturer:Bondcrete
Type : Type SA (Type S with Air entrainment)
Sand:

Manufacturer:George Drummond Ltd.

Type: Aggregate sand for construction

Table A l Mortar mix proportions for Phase 1 test program

Mortar Type -- 1 Cernent 1 Lime 1 Sand 1 Waterhinder ratio
1 ~ t r o n mortar
e 1 1 1 0.25 1 3 0.89 1
1
- -
V 1 1 1 1

Weak mortar 1 1 2 8 1 0.90

 -
Materials used in final test program Phase II

Masonry Unit:

FIagstone sandstone

S ize: 90mm x 90mm x 30 m m

Supplier: Merkley Supply Ltd.

Restoration brick
Size: 90mm x 90 mm x 27 m m

Supplier: Merkley Supply Ltd.

Cernent:
Manufacturer:Federal Cernent Ltd.
Type: Type 10
Colour: White
Lime:
Manufacturer: Bondcrete

Type: Type SA (Type S with Air entrainment)

Sand:

Manufacturer:George Drummond Ltd.

Type: Aggregate sand for construction

1
Table A2 Mortar mix proportions for Phase II test program
Mortar Type 1 Cernent Lime 1 Sand 1 Waterlbinder ratio
Strong mortar I 0.5 4 0.96
Weak mortar 1 3 10 1.17
Appendix B

Proposed Estimating Formula

J.J. Brooks and B.H. Abu Baker" proposed an estimating formula for the
rnodulus of elasticity in their paper published in Novernber 1998. Their formula is as
fo1lows:

Where:
E, = Modulus of elasticity of prism.
f, = Compressive svength of brick.

y,= Ratio of moduli of elasticity between brick and mortar.

& = Compressive strength of mortar.
and

,y = (1 - 0.016 W, )/(1 - 0.029 W, ) for masonry cured continuously under

polyethylene.
Where:
W, = Water absorption percentage (%)
Comparison of moduli of elasticity between Brook's proposed

estimation formula and Phase II final test results

The following table summarizes the measured test results of the brick prisms

and the estimated value using the proposed formula. The stone pnsms were not

compared as the water absorption percentage was too low when compared with the

brick, and was considered outside the proposed formula's proposed parameters.

Table B 1 Comparison of moduli of elasticity between Brook's formula and

-
laboratorv measured values.
Brook's estirnated Phase II measured
Difference
Prism combination moduli moduli range
(GW (GPa) (W
BricWStrong Mortar
I 11.9 1 4.3-7.1 1 277-168

B r i c W e a k Mortar 1 5.2 1 2.8 - 5.7 /

- - - -

Within range
 References
4

Masonry S369.1-M90, "Method of Test for Compressive Strength of Masonry

Prisms", Canadian Standard Association.

Drysdale, R.G., Hamid, A. A., Baker, L. R., "Masonry Strtrctrrre-Behaviotrr and

Design ",Chapter 5, Prentice Hall, Inc., 1994.
Rots, J.G., "Cornputer Simulafion of Masonry Fracture: Continzrttm and
Discontinuum Models ", pp. 93-1 03, Proceedings o f the International Symposium
on Computer Methods in Structural Masonry Swansea, UK, Apnl 199 1.

Stockl, S., Bienvirth, H., and Kupfer, H., "The influence of Test Method on the
Resirlts of Compression Tests on Mortar ",Proceedings, 10" IBMAC, University of
Calgary, pp. 1397-1406, 1994.

Stockl, S., Beckhaus, K., and Fritsche, TH., "Influence of Test Method on the
Reszllts of Deformation Menswements on Uniaxially Loaded Mortar Specimens ",
Masonry International Vol 12, No 1, 1998, pp. 32-38.

Page, A. W., and Brooks, D. S., "Load Bearing Masonry - A Review ",Proceedings
of the 7IhInternational Brick Masonry Conference, pp. 8 1-99: i985.

Suter, G. T., and Naguib, E. M. F., "Effect of Brick Stiyness Orrhorrapy on the
Lateral Stress in Stuck-Bonded Brick Masonry Prisms ", Proceedings, Fourth North
American Masonry Conference, paper 18, 1987.
Hendry, A. W., "Testing Methods in Masonry Engineering", Proceedings, Fourth
North American Masonry Conference, paper 49, 1987.

Drysdale, R. G., & Wong, H. E., "lrirerpretation of the Compressive Strcngth of

Masonry Prisms ",Proceedings o f the 7'h International Brick Masonry Conference,
1985.

Binda, L., Facchini, M., MirabelLa Roberti, G. Tiraboschi, C., "Ekctronic speckle
intcrferometryfor the deformation measlirement in masonry testing ",Construction
and Building Materials 12 (1998) pp. 269-28 1, 1998.

Brooks, J.J., and Abu Baker, B. H., "The Modlrlrrs of Elasticity ofMasonryW,The
British Masonrv Societv Journal. Vol. 12. No. 2. November 1998.
" Maurenbrecher, A.H.P. and Trischuk, K., "Durability of Mortars-Assessrnent of
properties and durabilily of pointing rnortar ",pp. 11-12, Drafi Report for Hentage
Conservation Program of Public Works & Government Services Canada, Institute
for Researc h in Construction, National Research Council, 1998.
' Loureno. P. B., "Computational Strategies for iMasonry Strttcfures ", Chapter 2,
Delfi University Press, The Netherlands, 1996.
1-1
Saliba, J. E., Alakkad, R. S., and Sawaya, G . E., "Use of Anisotropic Behaviotrr in
Mosonry ".Proceedings of the 61hCanadian Masonry Symposium, 1992.
1S
Eddington, J.R., & Gambo, A.H., "A Two Stage Analysis Procedure For Uftimate
Strength Analysis of Mmonty Shear Strrtctttres", pp. 84-92, Proceedings of the
International Symposium on Computer Methods in Structural Masonry, Swansea,
UK, April 1991.
16
Lawrence, S. J., and Cao, H. T., "An Experimental Study of the Interface Betrveen
Brick and Mortar ", Proceedings, Fourth North American Masonry Conference,
paper 48, 1987.

l7 Design of Masonry for Buildings (Limit States Design) CSA S304.1-M94;

Canadian Standards Association, 1994.

" Beall, C., "1Maronry Design and Detailing: For Archiectsl Engineers, and
Contractor ", pp. 123-144, McGraw Hill Text, 1997.
19
Francis, A.J., Horman, C.B. and Jerrems, L.E., "The Effecf ofJoint Thickness and
Other Factors on the Compressive Strength of Brickwork ", Proceeding, Second
International Brick Masonry Conference, Stoke-on-Trent, April 1970.