Journal of Building Engineering 26 (2019) 100843

Contents lists available at ScienceDirect

Journal of Building Engineering

journal homepage: [Link]/locate/jobe

Quantification of uncertainty in compressive strength of fly ash brick T

masonry
Santosini Sahua, Pradip Sarkara,∗, Robin Davisb
a
Department of Civil Engineering, National Institute of Technology Rourkela, Odisha, 769 008, India
b
Department of Civil Engineering, National Institute of Technology Calicut, Kerala, 673 601, India

A R T I C LE I N FO A B S T R A C T

Keywords: The compressive strength of brick masonry, an important property for the structural assessment and design, can
Fly ash brick masonry show a significant variation because of several influencing factors like source and proportion of constituent
Stack bonded masonry prism materials, workmanship, curing condition, etc. Quantification of uncertainty in masonry compressive strength is
Compressive strength essential for reliability based limit state design of masonry structures. The present study investigates this aspect
Goodness of fit test
of fly ash brick masonry and recommends the most suitable statistical distribution to describe the variability in
Lognormal distribution
compressive strength. Experimentally obtained compressive strength data are analysed using various probability
distribution functions capable of modelling strength properties of brittle materials and three goodness of fit tests,
i.e. Kolmogorov-Smirnov, Kolmogorov-Smirnov-Lilliefors and Anderson-Darling tests. The analysis concludes
that lognormal distribution is best suited among other distributions to describe the variability associated with
the compressive strength of fly ash brick masonry.

1. Introduction codes use limit state structural design approach to account all the
possible uncertainties. The load components are multiplied by the load
Unreinforced masonry is the oldest construction practice utilized factor and resistance is multiplied by the resistance factor. The basic
around the world, and it includes a significant part of the present form of the limit state function is
building structures. In India, masonry houses comprise 84.7% of the ϕR ≥ γQ (1)
overall housing units [1]. In addition, masonry is used as infill material
in reinforced concrete or steel framed buildings. Traditionally fired clay Where γ is the load factor applicable to load component Q; and ϕ is the
brick has been used for the building of masonry structures. In order to resistance factor applicable to the resistance capacity, R. Probabilistic
utilise a vast amount of fly ash deposition and to preserve the natural models of loads and resistances are required for precise application of
resources, Govt. of India has enforced the usage of the fly ash brick in the formula mentioned above [4]. Strength and load parameters exhibit
areas close to thermal power plants. Besides that, the use of fossil fuels statistical regularity, which calls for appropriate probabilistic models of
for the burning of clay bricks can be avoided as fly ash bricks are hy- load and resistance [5]. The concepts of probabilistic design and as-
draulic-pressed and air-cured, thus conserving natural resources and sessment of existing structures are generally accepted and presented
reducing greenhouse emissions. through several design codes [6,7].
As a critical parameter compressive strength governs the resistance With regard to the development of probabilistic material models,
capacity of both load bearing and infilled masonry structures. Masonry different levels of implementation are observed for different materials.
is weak in tension as it is comprised of two materials (brick and mortar) Probabilistic design with concrete, mortar, wood, and steel is ahead of
of different properties. Similarly, under lateral load also it cannot per- masonry. Many previous literature have reported the variability in the
form elastically even in the small range of deformations [2]. Therefore, compressive strength of cement mortar and concrete [8–10]. The Eur-
masonry is usually expected to resist just the compressive force in the opean probabilistic design code covers material models for steel and
structure. The existing design methodology also considers that the concrete, while masonry is not included so far [11]. Numerious past
masonry structures are subjected to compressive stresses alone [3]. studies [2,12–19]; etc.) have reported that the variability in strength
The characteristic values of material strength, dispersion, and its parameters of clay brick masonry is because of the inherent variation in
probability distribution affect the design of any structure [4]. Modern the structural and chemical properties of constituent materials.

∗
Corresponding author. Department of Civil Engineering, National Institute of Technology Rourkela, Odisha, 769 008, India.
E-mail addresses: santosinisahu62@[Link] (S. Sahu), sarkarp@[Link] (P. Sarkar), robin@[Link] (R. Davis).

[Link]
Received 26 March 2019; Received in revised form 18 June 2019; Accepted 19 June 2019
Available online 22 June 2019
2352-7102/ © 2019 Elsevier Ltd. All rights reserved.
S. Sahu, et al. Journal of Building Engineering 26 (2019) 100843

However, these uncertainties need to be transformed into a statistical factors like constituent materials, manufacturing process, environ-
distribution for limit state/reliability-based design of masonry struc- mental agents, etc. Consideration of all the randomness associated in
tures. Many studies are also reported [20–25]; and [19] on the aspect of the problem is quite complicated. Therefore, the compressive strength
modelling of variability in the different mechanical properties (such as of the fly ash brick masonry is considered as a representative parameter
shear strength and compressive strength) of clay brick masonry and to express the probability distribution of R. The growing usage of fly ash
other concrete block masonry. Sherafati and Sohrabi [19] determined brick in masonry work demands a detailed study of the variability in its
Johnson SB distribution as the best fit for the shear bond strength of compressive strength.
clay brick masonry. Schueremans and Van Gemert [24] and Sykora and
Holicky [25] have suggested lognormal distribution for the compressive
strength modelling of the clay brick unit. Montazerolghaem [23] has 3. Preparation of masonry prism specimens
studied the variability of shear strength and flexural strength of clay
brick, autoclaved aerated concrete block and calcium silicate block and An extensive survey revealed that the properties of fly ash change
concluded that the lognormal distribution might not be the best-fitted from source to source due to differences in coal properties, the method
distribution in the most cases. Many previous studies have reported the of production and loss of ignition. The variation of these factors causes
reliability analysis of clay brick masonry [13,14,26–30]. These studies fluctuation in the fly ash brick strength. It will be a challenging job to
have assumed the standard normal or truncated normal distribution for study the geographic variability of bricks from all possible sources.
the analysis of strength properties of clay brick masonry. Lawrence and Therefore, brick units from one representative source around Rourkela
Stewart [31] calibrated the capacity reduction factor for the design of (eastern part of India) are considered for this study assuming the nature
clay brick masonry assuming normal distribution. Pellegrino et al. [32] of variability in bricks from other sources will be similar. Testing the
expressed the variation of brick and stone masonry compressive masonry wall for determination of its strength properties is quite
strength by standard normal distribution through in situ and laboratory challenging and expensive [33]. Hence, it is a common practice to test
investigations on existing bridges. Barbieri [50] has conducted on flat the masonry prisms in the laboratory to study the masonry compressive
jack test on masonry bridges and reported that compressive strength of strength. Prism is a better illustration of the real masonry structure as it
solid brick and lime mortar assemblages follows lognormal dispersion. incorporates the effect of constituents material properties along with
It is clear from the past studies that, although some studies reported workmanship quality of masonry. A different course of prisms are used
the variability in different properties of fired clay brick masonry, the to obtain the masonry compressive strength in previous studies
same has not gained adequate attention for fly ash brick masonry in [2,34,35] and [36]. The American standard ASTM C1314-16 [48]
spite of its widespread use. However, the properties of fly ash brick proposes a height-to-thickness (h/t) proportion between 1.3 and 5 for
masonry may follow a probability distribution different to that of clay the prism specimens. Accordingly, 3-course stack bonded fly ash brick
brick masonry as the raw material and manufacturing process used are prisms with a height-to-thickness (h/t) ratio of 2.25 is considered in this
different in case of clay and fly ash brick production. Hence, this in- study. Fly ash bricks used in this study are manufactured by mixing fly
vestigation attempts to fill the gap in the existing research by focusing ash, sand, and cement at a proportion of raw materials of 60:30:10 by
on the statistical evaluation of the compressive strength of fly ash brick weight. The unit bricks are of size 230 mm × 110 mm × 75mm. The
masonry and propose a suitable probability distribution model. mean compressive strength of the brick unit is 5.45 MPa with a coef-
ficient of variation (CV) of 0.28. The masonry assemblages are con-
structed using cement mortar of sand to cement ratio 1:6 and water-
2. Research significance binder ratio 0.52 based on existing construction practice and previous
literature [2,37]. The mean compressive strength of the selected mortar
The performance of masonry structures in a probabilistic framework cube is determined as 6.96 MPa with a CV of 0.15. The prepared ma-
presented here, using a limit state function ‘g’ as sonry prism dimension is 230 mm × 110 mm × 250 mm.
g (R, Q) = R − Q ≥ 0 (2) Prior to construction, the bricks are pre-wetted for a suitable period
of time so that the hydration process in mortar is not disturbed by the
Where R and Q, resistance capacity and load component respectively, absorption of water by bricks. Brick units are bonded with cement
are continuous random variables with associated probability-density mortar layer of thickness around 8–10 mm. Typical masonry prism
function. Safety margin evaluation of any structure is dependent on the specimens and a schematic diagram of a specimen constructed for the
probability distribution of R that depends on various contributing compressive test are shown in Fig. 1. Immediately after the prism

Fig. 1. Fly ash brick prism samples.

2
S. Sahu, et al. Journal of Building Engineering 26 (2019) 100843

Fig. 4. PDF plot with lognormal distribution.

Table 1
Statistics and percentile values of the compressive strength data (in MPa) of
brick samples.
Statistic Value Percentile (%) Value

Sample Size 50 Minimum 1.58

Range 2.77 5 1.78
Mean 3.00 10 2.01
Variance 0.60 25 (quarter 1) 2.37
Standard Deviation 0.77 50 (median) 2.81
Coefficient of Variation 0.26 75 (quarter 3) 3.78
Standard Error 0.11 90 4.13
Fig. 2. Testing of the fly ash brick prism samples. Skewness 0.24 95 4.35
Excess kurtosis −1.00 Maximum 4.35

construction, the specimens are sealed with the moisture-tight bag.

After the initial 48 h, the prisms are cured by covering with wet burlap
Table 1 presents the statistics and percentile values of the compressive
at a temperature of 24 ± 8 °C. Two days before the testing, wet bags
strength data of prism samples. This table depicts that the compressive
are removed, and prisms are stored in dry place. Visible moisture
strength values vary from 1.58 MPa to 4.35 MPa. The mean compres-
should not be present on the surface of the prism during the test.
sive strength of masonry prisms is found to be 3 MPa with a CV of 0.60.
Skewness in the compressive strength values of prism specimens is
4. Compressive strength of fly ash brick masonry +0.24 (positively skewed) which indicates that compressive strength
values higher than the mean value are more widely dispersed than the
Fifty numbers of masonry assemblages prepared following the dispersion of the values below the mean strength. Kurtosis for the
procedure discussed above are tested after 28 days in the compression compressive strength data set is observed to be −1. The negative value
testing machine in line with ASTM C1314-16. The experimental test set- of kurtosis implies that the distribution is platykurtic with heavy tails
up is displayed in Fig. 2. The individual compressive strength values of and broad shoulders.
the fifty specimens are presented graphically in Fig. 3.
Fig. 4 illustrates the histogram plot of prism compressive strength.

Fig. 3. Compressive strength values for all masonry specimens.

3
S. Sahu, et al. Journal of Building Engineering 26 (2019) 100843

Table 2 and rank of various distributions (having a different number of cali-

Distribution parameters for the various probability distribution functions. brated parameters) obtained from K-S, K-S-L and A-D tests. It is seen
# Distribution Parameters from the table that two-parameter lognormal distribution ranks first in
case of all three considered GOF tests (K-S, K-S-L, and A-D). Based on
1 Gamma (2P) α = 15.041 β = 0.199 the analysis performed, it can be proposed that lognormal (2P) dis-
2 Gamma (3P) α = 11.009 β = 0.234 γ = 20.416
tribution is best suited for compressive strength of fly ash brick ma-
3 Gumbel Max (2P) σ = 0.602 μ = 2.649
4 Gumbel Min (2P) σ = 0.602 μ = 3.344
sonry.
5 Johnson SB (4P) α = 0.261 δ = 0.755 λ = 3.112 ξ = 1.641 The observed compressive strength data is positively skewed with a
6 Lognormal (2P) σ = 0.261 μ = 1.064 longer right tail, and it is a platykurtic distribution, which might be the
7 Lognormal (3P) σ = 0.182 μ = 1.425 γ = −1.233 reason why lognormal distribution fits the data better than other se-
8 Normal (2P) σ = 0.772 μ = 2.997
lected distributions. However, it should be noted here that the differ-
9 Weibull (2P) α = 4.363 β = 3.259
10 Weibull (3P) α = 2.366 β = 1.906 γ = 1.309 ence between the statistic values of some of the highly ranked dis-
tributions is very small. Also, there might be a minor change in ranks if
α, δ are continuous shape parameters; β, σ, λ are continuous scale parameters; the data pool expands or changes. Considering all such cases the other
γ, μ, ξ, are continuous location parameters. higher ranked distributions may also be acceptable to represent the
compressive strength model.
5. Variability in compressive strength Suggested lognormal is a bounded distribution with two parameters.
The probability density function for lognormal (2P) distribution is de-
Statistical techniques are used to find the regularities of the ob- fined as:
served compressive strength data and propose the probability dis-
ln x − μ 2
tributions that can model the variability. The compressive strength data
obtained are statistically analysed for ten selected probability dis- f (x ) =
( (
exp − 2
1
σ ))
xσ 2π (3)
tribution functions based on the past literature [22,23]; and [10] using
the computer program EasyFit 5.6 Professional [49]. The details of the The corresponding cumulative distribution function is given as fol-
selected distributions are available in the published literature [38]. The lows:
parameters obtained for the compressive strength of fly ash brick ma-
ln x − μ ⎞
sonry for each of the selected distributions are calculated and presented F (x ) = φ ⎛
in Table 2. ⎝ σ ⎠ (4)
Three Goodness of Fit (GOF) tests, namely K-S, K-S-L, and A-D are where σ is the mean, μ is the standard deviation and ϕ is the Laplace
carried out for the data set to obtain the best-fit distribution model for integral. The performance of lognormal distribution to represent the
the compressive strength of fly ash brick masonry. It should be noted probability distribution of the compressive strength can be shown in
here that all of the considered GOF tests are non-parametric and are various ways. In general, the variability is described with the prob-
applicable for continuous distributions. ability density function (PDF) plot and Cumulative density function
The basic principle of K-S [39] and K-S-L [40] test is to measure the (CDF) plot. The PDF and CDF plot of compressive strength with log-
“distance or difference” between the observed cumulative distribution normal distribution is shown in Figs. 4 and 5 respectively. These two
function and the hypothesized cumulative distribution function (called figures indicate how well lognormal distribution fits the observed data.
the test statistic) and compare that distance/difference with some A probability-probability (P-P) plot is used to represent the observed
threshold value (critical value). The critical value relies upon the values of the cumulative distribution function (CDF) against the fitted
sample size and the chosen significance level. The lower the statistic of CDF values. Similarly, quantile-quantile (Q-Q) plot represents the ob-
a function, the higher is its rank. The A–D test is more sensitive to served data values plotted against the fitted distribution quantiles. Both
deviations in the tails of the distribution than the K-S test and an im- of these plots are used to analyse the match between observed data and
proved version of the K-S test [41]. Thus A-D test is additionally con- the modelled distribution. P-P plot emphasises more at the centre of the
sidered as a check on the outcomes. This test (A-D) is used in several distributions, but less at the tails. On the other hand, Q-Q plot focuses
previous literature [42,43]; etc.) for a similar purpose. more at the tails, but less in the centre. P-P and Q-Q plots of the ob-
The logic of applying various GOF tests is the same for all cases. served data with the lognormal distribution presented in Figs. 6 and 7
However, they differ in what way the test critical values and test sta- are almost linear along the 45-degree line which indicates that both
tistics are calculated. Further details about the selected GOF test can be centre and the tail of lognormal distribution predicts the variability of
available elsewhere [44,45]. The test statistic values for the chosen compressive strength of fly ash brick masonry prisms with significant
probability distribution function are obtained individually, and the
functions are ranked accordingly. Table 3 presents the statistic value

Table 3
GOF Statistic for compressive strength of fly ash brick masonry.
# Distribution K-S/K-S-L A-D

Statistic Rank Statistic Rank

1 Gamma (2P) 0.128 5 0.913 4

2 Gamma (3P) 0.124 3 0.884 3
3 Gumbel Max (2P) 0.131 8 1.150 7
4 Gumbel Min (2P) 0.180 10 3.130 9
5 Johnson SB (4P) 0.128 4 8.164 10
6 Lognormal (2P) 0.119 1 0.601 1
7 Lognormal (3P) 0.129 6 0.923 5
8 Normal (2P) 0.140 9 1.133 6
9 Weibull (2P) 0.130 7 1.227 8
10 Weibull (3P) 0.123 2 0.882 2
Fig. 5. CDF plot with lognormal distribution.

4
S. Sahu, et al. Journal of Building Engineering 26 (2019) 100843

Fig. 6. P-P plot with lognormal distribution. Fig. 9. Probability Diff. plot with lognormal distribution.

Fig. 7. Q-Q plot with lognormal distribution. Fig. 10. HF plot with lognormal distribution.

Fig. 8. SF plot with lognormal distribution. Fig. 11. CHF plot with lognormal distribution.

accuracy. demand of 3.1 MPa is 0.39. The probability difference plot shows the
A survival function plot signifies the probability of exceeding the distinction between the observed CDF and the fitted CDF. The prob-
compressive strength of a particular value. It is frequently used in re- ability difference plot of the observed data with the lognormal function
liability and related fields to find out the likelihood of the variant to is shown in Fig. 9. The plot depicts that the maximum value of the
take on a higher value than the specified value. If ‘X’ is a continuous probability difference is less than 0.12 for lognormal function. Both the
random variable with probability density function f(x) and cumulative survival function plot and probability difference plot shows that log-
distribution function F (x ) = P (X ≤ x ) , mathematically, the survival normal distribution fits the observed data quite well and supports the
function, S(x), can be expressed as: conclusions obtained from P-P and Q-Q plots.
Hazard function and cumulative hazard function are two para-
S (x ) = P (X > x ) = 1 − F (x ) (5) meters often used in the reliability analysis. Hazard function, h(x) for
the distribution of ‘X’ is defined as the instantaneous rate of occurrence
Fig. 8 presents the survival function graph of the experimental data of an event as follows:
with the lognormal function. This figure illustrates the chances of sur-
vival of the selected fly ash brick masonry as a function of the com- f (x ) f (x )
h (x ) = =
pressive strength demand. For example, the probability of surviving for S (x ) 1 − F (x ) (6)
the chosen fly ash brick masonry beyond a compressive strength

5
S. Sahu, et al. Journal of Building Engineering 26 (2019) 100843

Fig. 12. Crushing failure of the brick.

Fig. 13. Face shell Separation.

Where f(x) is probability density function and F(x) is the cumulative This figure indicates that the failure rate of selected fly ash brick ma-
distribution function. The greater the hazard at instantaneous com- sonry increases with the increase of compressive strength demand. It is
pressive strength, the greater is the risk of exceedance. Cumulative to be noted that the shape of the hazard function strongly relies upon
hazard function for the distribution of ‘X’ is defined as: the data and the selected distribution. The shape of the hazard function
presented in Fig. 10 considering lognormal distribution appears to be
x
representing the compressive strength of fly ash brick masonry appro-
H (x ) = ∫ h (X ) dX priately. The lognormal hazard rate, h(x), is generally an increasing
−∞ (7)
followed by a decreasing function, and can be shown to approach zero
The hazard function plot can be used to relate between the physical for large value of x (Sweet, 1990). However, this trend is not visible in
knowledge and mathematical structure of the distributions that model Fig. 10 as this figure is plotted for a range of x (compressive strength)
the phenomenon [46,47]. Fig. 10 illustrates the hazard function plot within the limit of observed data.
and Fig. 11 shows the cumulative hazard function plot considering
lognormal distribution for the compressive strength of fly ash brick 6. Failure pattern of prism specimens
masonry. These plots might be used to assess the safety of fly ash brick
masonry structures. The hazard function depicts the likelihood of ASTM C1314-16 identifies seven types of possible failure modes in
failure as a function of force demand (in terms of compressive strength brick masonry under compressive strength test as follows: conical
demand) in this study. The hazard plot presented in Fig. 10 shows the break, cone and shear, cone and split, tension break, semi-conical
trend in the failure rate as a function of compressive strength demand. break, shear break, and face shell separation. None of the above seven

6
S. Sahu, et al. Journal of Building Engineering 26 (2019) 100843

Fig. 14. Tension separation.

failure modes is explicitly observed among the fifty prism specimens Appendix A. Supplementary data
tested as part of this study. In the present study, it is found that the
crack is not forming uniformly around the specimen. In most of the Supplementary data to this article can be found online at https://
specimens, the cracks started from one edge of the brick and stopped at [Link]/10.1016/[Link].2019.100843.
mortar joint without further propagation. This may be attributed to the
fact that the brick strength is lesser than the adjoining mortar strength. References
Figs. 12–14 summarized different failure pattern observed in masonry
specimens throughout the compression test carried out in this study. [1] J.K. Banthia, Census of India, 2001, Controller of Publications; Office of the
Crushing of edge brick as shown in Fig. 12 is the most common type of Registrar General & Census Commissioner, 2001 New Delhi, India.
[2] H. Kaushik, D. Rai, S. Jain, Stress-Strain characteristics of clay brick masonry under
failure pattern observed. Fig. 13 shows a typical failure pattern illus- uniaxial compression, J. Mater. Civ. Eng. 19 (No. 9) (2007) 728–739.
trating vertical tensile splitting cracks along the direction of loading [3] J. Thomas, Concrete block reinforced masonry wall panels subjected to out-of-plane
similar to the face shell separation mentioned in ASTM C1314-16. monotonic lateral loading, Proceedings of National Conference on Recent Advances
in Structural Engineering, Hyderabad, India, February, 2006, pp. 123–129.
In a few cases, cracks initiated inside the mortar joint and after that, [4] A.S. Nowak, K.R. Collins, Reliability of Structures, McGraw-Hill, New York, 2000.
it propagated to the brick units as illustrated in Fig. 14. This kind of [5] B. Ellingwood, T.V. Galambos, J. MacGregor, C.A. Cornell, Development of a
failure pattern is termed ‘tension separation’ in ASTM C1314-16. Probability Based Criterion for American National Standard A58, NBS Special
Publication 577 National Bureau of Standards, Washington, DC, 1980.
Thirty-six out of the fifty tested specimens are failed through crushing
[6] EN 1990, Eurocode – Basis of Structural Design, European Committee for
of edge brick while nine specimens are failed through tension separa- Standardization, Brussels, 2002.
tion failure mode. Face shell separation was observed for remaining five [7] JCSS probabilistic model. "Joint committee on structural safety." URL: [Link].
ethz.
specimens.
[8] T. Chmielewski, E. Konopka, Statistical evaluations of field concrete strength, Mag.
Concr. Res. V. 51 (No. 1) (1999) 45–52.
[9] B. Graybeal, M. Davis, Cylinder or cube: strength testing of 80 to 200 MPa (11.6 to
7. Conclusions 29 ksi) Ultra-high-performance fiber-reinforced Concrete, ACI Mater. J. 105 (No. 6)
(2008) 603–609.
[10] L. Sorrentino, P. Infantino, D. Liberatore, Statistical tests for the goodness of fit of
This study aims to investigate the statistical variations of the com- mortar compressive strength distributions, 16th International Brick and Block
pressive strength of stack bonded fly ash brick masonry prisms. Ten Masonry Conference (IBMAC), 26-30 June, Taylor & Francis group, Rome, 2016,
pp. 1921–1928.
distributions are selected to find out the best-fit probabilistic model that [11] A.C.M.V. Vrouwenvelder, Development toward full probabilistic design code,
closely describes the dispersion in the compressive strength of the Struct. Saf. 24 (2002) 417–432.
masonry prisms. The most appropriate distinct statistical distribution [12] X. Chen, S. Wu, J. Zhou, Variability of compressive strength of concrete cores, J.
Perform. Constr. Facil. V. 28 (No. 4) (2014) 06014001.
for masonry compressive strength is found out using the three good- [13] B. Ellingwood, Analysis of reliability for masonry structures, J. Struct. Div. V. 107
ness-of-fit tests on the available experimental data set. This study (No. 5) (1981) 757–773.
concludes that the variation in the compressive strengths of fly ash [14] B. Ellingwood, A. Tallin, Limit states criteria for masonry construction, J. Struct.
Eng. V. 111 (No.1) (1985) 108–122.
brick masonry can be described using the “lognormal (2P)” probability [15] J. Li, Masia, M.J. Stewart, S.J. Lawrence, Spatial variability and stochastic strength
distribution. The results of this study will form the basis for limit state prediction of unreinforced masonry walls in vertical bending, Eng. Struct. 59 (2014)
design of fly ash brick masonry structures. Additionally, the findings of 787–797.
[16] T.C. Ling, K.H. Mo, Q. Lie, J. Yang, G. Lei, Mechanical strength and durability
this study will support the probability-based analysis and design of fly
performance of autoclaved lime-saline soil brick, Constr. Build. Mater. 146 (2017)
ash brick masonry structures where several compressive strength data 403–409.
are required to be generated to consider the uncertainty associated with [17] M.N. Rahmat, N. Ismail, J.M. Kinuthia, Strength and environmental evaluation of
the compressive strength of fly ash brick masonry in the structural stabilised Clay-PFA eco-friendly bricks, Constr. Build. Mater. 125 (2016) 964–973.
[18] G. Sarangapani, B.V. Reddy, K. Jagadish, Brick-Mortar bond and masonry com-
model. The parameters of the proposed distributions can be evaluated pressive strength, J. Mater. Civ. Eng. 17 (No. 2) (2005) 229–237.
from the available data utilizing any of the conventional design of ex- [19] M.A. Sherafati, M.R. Sohrabi, Probabilistic model for bed-joint shear- sliding
periment (DOE) technique. strength of clay-brick walls based on field test data, J. Risk Uncertainty Eng. Syst.,
Part A: Civ. Eng. 2 (No. 4) (2016) 04016007.
[20] L.R. Baker, G.L. Franken, Variability aspects of the flexural strength of brickwork,

7
S. Sahu, et al. Journal of Building Engineering 26 (2019) 100843

4th Int. Brick Masonry Conference, April 1976, pp. 4–11 (Brussels). under axial compression, J. Struct. Eng. V. 118 (No. 7) (1992) 1751–1769.
[21] L.M. Heffler, Variability of Unit Flexural Bond Strength and its Effect on Strength in [36] A.J. Francis, C.B. Horman, L.E. Jerrems, The effect of joint thickness and other
Clay Brick Unreinforced Masonry Walls Subject to Vertical Bending, Master’s thesis factors on compressive strength of brickwork, Proceedings of 2nd International
University of Newcastle, 2009. Brick Masonry Conference, Stoke-on-Trent, 2017, pp. 31–37.
[22] S.J. Lawrence, Random variations in brickwork properties. 7th Int. Brick Masonry [37] N.N. Thaickavil, J. Thomas, Behaviour and strength assessment of masonry prisms,
Conference, February 17-20, 1985, pp. 537–547 Melbourne. Case Studies Construct. Mater. 8 (2018) 23–38.
[23] M. Montazerolghaem, Analysis of Unreinforced Masonry Structures with Uncertain [38] C. Forbes, M. Evans, N. Hastings, B. Peacock, Statistical Distributions, ” fourth ed.,
Data, P.h.D Thesis Technische Universität, Dresden, Germany, 2015. John Wiley, New York, 2011.
[24] L. Schueremans, D.V. Gemert, Probability density functions for masonry material [39] J.R. Benjamin, C.A. Cornell, Probability, Statistics, and Decision for Civil Engineers,
parameters - a way to go? 5th Int. Conference on Structural Analysis of Historical McGraw-Hill, New York, 1970.
Constructions, 6–8 November, 2006, pp. 921–928 New Delhi. [40] H.W. Lilliefors, On the Kolmogorov-Smirnov test for normality with mean and
[25] M. Sykora, M. Holicky, Probabilistic model for masonry strength of existing struc- variance unknown, J. Am. Stat. Assoc. 62 (No. 318) (1967) 399–402.
tures, Eng. Mech. 17 (No. 1) (2010) 61–70. [41] A.H.S. Ang, W.H. Tang, Probability Concepts in Engineering: Emphasis on
[26] M.A. Maes, L. Schueremans, K.V. Balen, Reliability based assessment of existing Applications to Civil and Environmental Engineering, second ed., Wiley, New York,
masonry structures, December, “8th Int. Conf. On Applications of Statistics and 2006.
Probability (ICASP8), 1999, pp. 689–696 Sydney. [42] A.K. Chandrappa, K.P. Biligiri, Flexural-fatigue characteristics of pervious concrete:
[27] P. Müller, L. Miccoli, P. Fontana, C. Ziegert, Calibration of partial safety factors for statistical distributions and model development, Constr. Build. Mater. V. 153
Earth Block Masonry under compression loading, 16th International Brick and (2017) 1–15.
Block Masonry Conference, Taylor & Francis group, Rome, 2016, pp. 857–864 26- [43] K. Kilinc, A.O. Celik, M. Tuncan, A. Tuncan, G. Arslan, O. Arioz, Statistical dis-
30 June. tributions of in-situ micro core concrete strength, Constr. Build. Mater. 26 (2012)
[28] F. Parisi, N. Augenti, Seismic capacity of irregular unreinforced masonry walls with 393–403.
openings, Earthq. Eng. Struct. Dyn. V. 42 (No. 1) (2013) 101–121. [44] T.B. Arnold, J.W. Emerson, Nonparametric goodness-of-fit tests for discrete null
[29] F. Parisi, N. Augenti, Uncertainty in seismic capacity of masonry buildings, distributions, The R J. V. 3 (No. 2) (2011) 34–39.
Buildings 2 (No. 4) (2012) 218–230. [45] G.W. Corder, D.I. Foreman, Nonparametric Statistics: A Step-by-step Approach,
[30] A. Taliercio, L. Binda, On the reliability of linear elastic analyses of historical ma- John Wiley & Sons, 2014, pp. ISBN978–1118840313.
sonry buildings: a case study, J. Build. Apprais. V. 2 (No. 4) (2007) 301–312. [46] E. Garavaglia, A. Anzani, L. Binda, Probabilistic model for the assessment of historic
[31] S.J. Lawrence, M.G. Stewart, Reliability-based calibration of the capacity reduction buildings under permanent loading, J. Mater. Civ. Eng. 18 (No. 6) (2006) 858–867.
factor for design of masonry in compression to AS3700, Aust. J. Struct. Eng 9 (No. [47] E. Garavaglia, A. Anzani, L. Binda, G. Cardani, Fragility curve probabilistic model
2) (2009) 97–110. applied to durability and long term mechanical damages of masonry, Mater.
[32] C. Pellegrino, M.A. Zanini, P. Zampieri, C. Modena, Contribution of in situ and Struct./Matériaux et Constructions 41 (No. 4) (2008) 733–749.
laboratory investigations for assessing seismic vulnerability of existing bridges, [48] ASTM, Standard Test Method for Compressive Strength of Masonry Prisms
Struct. Infrastruct. Eng. 11 (No. 9) (2015) 1147–1162. C1314–16 ASTM, West Conshohocken, PA, 2016.
[33] K.S. Jagdish, B.V.V. Reddy, K.S.N. Rao, Alternative Building Materials and [49] EasyFit version 5.6 [Computer software], MathWave technologies, dnepropetrovsk,
Technologies, New Age International, New Delhi, India, 2008. Ukraine. Mathwave, Available at: [Link]
[34] R.G. Drysdale, A.A. Hamid, Behavior of concrete block masonry under axial com- [50] D.M. Barbieri, Two methodological approaches to assess the seismic vulnerability of
pression, ACI J. Proc. V. 76 (No. 6) (1979) 707–722. masonry bridges, J. Traffic Transp. Eng. 6 (No. 1) (2019) 49–64.
[35] T.P. Ganesan, K. Ramamurthy, Behavior of concrete hollow-block masonry prisms