Paper 184 Seismic Damage Model Fractal Dimension Cracking
Paper 184 Seismic Damage Model Fractal Dimension Cracking
Paper 184 Seismic Damage Model Fractal Dimension Cracking
J. Carrillo
Department of Civil Engineering, Nueva Granada Military University, Bogotá, Colombia.
1 INTRODUCTION
Crack width is one of the main indicators of damage severity experimented by reinforced concrete, RC,
structural components during an earthquake. Moreover, the cracking pattern on wall surface is
associated to the experimented seismic demands, that is, to the variation of shear stresses in relation to
high of the wall. Therefore, several studies has focused efforts to estimate the seismic performance level
in terms of damage. For instance, Carrillo and Alcocer (2012a) have applied a damage index based on
the relation between the damaged area (area of cracks) and the area of the façade of the RC wall. Carrillo
(2015a) have proposed a damage index based on the stiffness degradation of walls. Such index depends
on the story-drift ratio and the number of cycles experimented by the wall during a particular seismic
event. In addition, Adhikari et al. (2013) proposes an integral model based on digital images processing
for numerical quantification of cracks, pattern variations and neurological connection between cracks.
Fractals are geometric figures than cannot be described in terms of classical geometry. Such geometric
figures are characterized by having copies of themselves at different scales randomly, i.e., they have a
fine structure. The fractal dimension is a mathematical parameter that measures the geometric
complexity level of a pattern further than evaluates the filling property of a particular geometric plane
or space. The field of structural engineering has applied the approach of fractal theory for proposing
alternative and innovative methodologies of damage evaluation. For instance, Chiaia et al. (1998)
carried out fracture tests of concrete for assessing the failure modes using the fractal dimension of
cracking patterns. Structural health monitoring has also been a field of application of the fractal analysis.
Moustafa et al. (2013) monitored the propagation path of corrosion of post-tensioned structures by
means of the fractal dimension of ultrasonic waves measured in steel tendons. Tzu-Kang et al. (2012)
proposed a novel bridge health monitoring system and a safety index based on the fractal dimension of
the correlation between the scour level and the fundamental period of the bridge superstructure. In a
similar way, Hadjileontiadis and Douka (2007) proposed a cracking detection model based on the fractal
dimension of the vibration mode shapes and the fundamental period of plate elements. Li et al. (1993)
used the fractal theory for describing numerically the shape and distribution of aggregates for concrete.
Miao et al. (2014) studied fractal and multifractal cgaracteristics of 3D asphalt pavement macro-
structures in terms of its depth and friction coeeficient. Werner et al. (2013) used a fractal-based
approach for assessing numerically the parameters related to fractured surface of concrete using laser
scanning techniques.
The approach of fractal analysis for studying the cracking of concrete walls was initially used by
Farhidzadeh et al. (2013). They implemented a theoretical model based on a damage index that depends
on the fractal dimension of the cracking propagation recorded in concrete walls. A damage index
removes the subjectivity and the variability associated with damage assessment based on visual
FDi FDmin
DI (1)
2 FDmin
where FDi is the fractal dimension of the current status of visible cracks (e.g., in the ith inspection); and
FDmin is the fractal dimension computed during the first inspection (e.g., once the cracks become visible
for the first time). The constant “2” in the denominator is the maximum value of the fractal dimension
2
for surface cracks, FDmax. DI varies between 0 and 1 and describes the difference between the current
status of crack patterns and the baseline FDmin. When cracks cover the whole area of the concrete,
Farhidzadeh et al. (2013) proposed that FDmax = 2. However, cracks patterns related to loss of lateral
resistance limit state of thin RC wall specimens have demonstrated that area of cracks on façade is lower
than 30% of the total area of façade (Carrillo and Alcocer, 2012a). For instance, for walls with web
shear reinforcement made of welded-wire mesh and showing a diagonal tension failure, one or some
inclined cracks are merely observed. For walls with web shear reinforcement made of deformed bars
and showing a diagonal compression failure, significant cracks on façade are recorded. Although such
crack patterns can be related to several damage stages, the total area of cracks is certainly not equivalent
to the whole area of the concrete as supposed by Farhidzadeh et al. (2013). Therefore, the final value of
the fractal dimension for surface cracks, FDu, should be used instead of FDmax (FDuu). In addition,
previous studies (Carrillo, 2015b) have demonstrated that fractal dimension of final crack patterns
depend on the geometrical and reinforcement characteristics of thin RC walls. In this study, initial
(FDmin) and final (FDu) values of the fractal dimension of cracks patterns observed in thin RC walls in
terms of the particular characteristics of low-rise housing. Such two values of fractal dimension delimit
the threshold of damage expected in thin RC walls for low-rise housing.
3 EXPERIMENTAL PROGRAM
Earthquake-induced damage of mid- and high-rise buildings has been widely assessed using tests of
prototypes having various wall layouts. Damage will be symbolized by the pattern and distribution of
cracks observed in 39 isolated thin RC walls tested under quasi-static cyclic load and under shake table
excitations (Carrillo and Alcocer, 2012b). Variables studied are described in Table 1. The typical
geometry and reinforcement layout of some of the full-scale wall specimens is shown in Figure 1.
Behavior of walls was governed by shear deformation so that they exhibited a relatively fragile failure
modes. For evaluating the observed wall behavior, three failure modes were identified: diagonal tension
(DT) failure, diagonal compression (DC) failure, and a mixed failure mode (DT-DC).
3
8#6 8#6 4#5 4#5 4#5 4#5 4#5
4#3
S#2
@200 1#4 1#4
4#5, S#2@150
4#5, S#2@150
#3@250
4#5, S#2@150
E#2@150
4#5, S#2@150
4#5, S#2@150
mesh
6x6-6/6
8#6
#3@250
2#4
2#3
3#3
3#3
Figure 1. Typical geometry and reinforcement layout of wall specimens: (a) hw/lw = 1.0, 100% of min and using
deformed bars (wall MCN100C); (b) wall with openings, 50% of min and using welded-wire mesh (wall
MVN50mC) (Carrillo and Alcocer, 2012a).
Drawings of cracking patterns of walls were achieved by means of marking cracks and recording
damage (notes and pictures) during tests of walls. Cracking drawings of the 39 walls were then converted
to digital format. Shear force, V, and story-drift ratio, R, were also recorded for each characteristic
loading stage. Story-drift ratio was expressed in percentage and was obtained by dividing the relative
displacement measured at mid-thickness of the top slab by the height at which such displacement was
measured. Drift ratios were related to four limit states namely diagonal cracking, maximum shear
strength, loss of lateral resistance and failure of the wall or end of test (Rcr, Rmax, Ru and Ruu) and to three
performance levels namely immediate occupancy, life safety, and collapse prevention (RIO, RLS and RCP).
Limit states for PBSD of low-rise housing having thin RC walls were proposed by Carrillo and Alcocer
(2012b). Diagonal cracking limit state is attained when inclined web cracking is observed. Maximum
shear strength limit state corresponds to peak shear strength. Loss of lateral resistance limit state is
associated to any of the two following scenarios: when a 20% drop to the peak shear strength is reached
or when web shear reinforcement made of welded-wire meshes fractures. Performance levels are those
recommended by Vision 2000 (SEAOC, 1995). Based on technical and economic facts, the IO, LS, and
CP performance levels for low-rise concrete housing are related to initial inclined web cracking, to
extension of web inclined cracks to wall edges without penetration into boundary elements, and to wall
peak shear strength, respectively.
Carrillo (2015b) have correlated results of fractal dimension with story-drift ratio of walls. Based on
such correlations, values of fractal dimension of cracking related to measured drift ratios at defined limit
states (FDcr, FDmax, FDu and FDuu) and performance levels (FDIO, FDLS and FDCP) were proposed. A
statistical analysis was then carried out for proposing values of fractal dimension in terms of the
variables studied.
FDi FDini
DI (2)
FDu FDini
where FDi and FDini are defined similarly to the index proposed by Farhidzadeh et al. (2013), that is,
FDi is the fractal dimension of the current status of visible cracks (e.g., in the ith inspection), and FDini
4
is the fractal dimension computed once the cracks become visible for the first time (“initial stage”).
However, FDu is the value of the fractal dimension for surface cracks related to loss of lateral resistance
limit state of thin RC walls (“final stage”). Values of FDini and FDu are proposed by Carrillo, 2015b.
Similarly to index proposed by Farhidzadeh et al. (2013), the proposed damage index describes the
difference between the current status of crack patterns and the baseline FDmin. As conventional, DI varies
between 0 and 1. Some damage parameters based on cracking have been proposed; however, their
expression in the form of a damage index (0DI1) requires the definition of the limiting value of
cracking after which failure occurs. In this study, a damage index equal to zero indicates no damage or
elastic behavior in the wall, and equal to 1.0 when the wall losses the lateral resistance.
As discussed earlier, in the damage index proposed by Farhidzadeh et al. (2013), the maximum value
of the fractal dimension for surface cracks, FDmax, is 2 because the model assumes that cracks cover the
whole area (100%) of the concrete surface. It can be inferred that the actual cracking stage is not
evaluated when using FDmax = 2. However, such cracking condition is significantly high when compared
with condition of cracks patterns related to loss of lateral resistance limit state of thin RC wall specimens
(30% of the concrete surface). Therefore, the final value of the fractal for surface cracks, FDu, should
be used instead of FDmax. In addition, based on trends of experimental results, Carrillo (2015b) has
proposed FDmax (FDuu) values significantly lower than that proposed by Farhidzadeh et al. (2013); for
instance, FDuu values equal to 1.530 and 1.252 for walls with web shear reinforcement made of deformed
bars and welded-wire mesh, respectively. Such noteworthy different values suggest that the damage
index should be settled based on a particular value of FDu in terms of the characteristics of the wall
instead of a unique value of FDmax (FDuu).
In addition, damage index computed with Eqn. 1 is related with a non-conservative estimation of the
structural damage. For instance, for a wall having hw/lw = 2 and web steel ratio equivalent to 50% of min
(0.125%) and using deformed bars, the fractal dimension of surface cracks related to the last cracking
record at failure of the wall is FDi = 1.220. Men value of FDini for walls having web shear reinforcement
made of deformed bars is 0.927 (Carrillo, 2015b). Damage index computed using FDu = 2 (as proposed
by Farhidzadeh et al., 2013) is 27%. However, when using the mean value of FDu proposed by Carrillo
(2015b) for walls with deformed bars (FDu = 1.530), the damage index is 51%. The damage index
computed using the model proposed in this study (DI = 51%) is significantly higher than that computed
using the model proposed by Farhidzadeh et al. (DI = 27%). Thus, it is demonstrated that the model of
Farhidzadeh et al. can provide a non-conservative estimation of damage stage of walls having the
particular characteristics of low-rise housing.
5
In Eqn. 2, FDi is the fractal dimension of the current status of visible cracks related to a particular limit
state or performance level. Therefore, the proposed damage index is an effective tool for explaining the
seismic behavior of a structure based on the performance of the walls. Table 2 shows the expected
damage index at defined limit states (DIcr, DImax, DIu and DIuu) and performance levels (DIIO, DILS and
DICP). Damage index were arranged in terms of aspect ratio of walls, type of concrete, web steel ratio,
type of web shear reinforcement, and type of testing. For the IO (immediate occupancy) performance
level, walls reinforced with deformed bars and welded-wire mesh have attained 20% of the performance
capacity (DIIO = 0.20). For the LS (life safety) performance level, walls reinforced with deformed bars
and walls with welded-wire mesh have attained 46% (DILS = 0.46) and 56% (DILS = 0.56) of the
performance capacity, respectively. For the CP (collapse prevention) performance level, walls
reinforced with deformed bars and walls with welded-wire mesh have attained 69% (DICP = 0.69) and
99% (DICP = 0.99) of the performance capacity, respectively. Although two types of walls have
comparable shear strength capacities (Carrillo and Alcocer, 20xx), residual capacity at CP performance
level of walls with welded-wire mesh is scarcely 1% (1-0.99) while such capacity of walls with
deformed bars is 31% (1-0.69). These significant differences of residual capacity is directly related with
lower costs of seismic rehabilitation of walls with deformed bars when compared with wall with welded-
wire mesh.
Table 2. Damage index associated to limit states and performance levels in terms of different variables
Limit states Performance levels
Variable DIini
DIcr DImax DIu DIuu DIIO DILS DICP
hw/lw = 2.0 0.13 0.79 1.38 0.09 0.46 0.87
Aspect ratio
hw/lw = 1.0 0.00 0.26 0.75 1.00 1.29 0.21 0.47 0.71
(hw/lw)
hw/lw = 0.5 0.21 0.75 1.31 0.19 0.48 0.79
Normalweight 0.24 0.77 1.34 0.21 0.47 0.78
Type of
Lightweight 0.00 0.24 0.76 1.00 1.22 0.18 0.49 0.73
concrete
Self-consolidating 0.21 0.72 1.35 0.14 0.40 0.63
0% ρmin 0.20 0.82 1.19 0.05 0.20 0.25
Web steel
50% ρmin 0.00 0.22 0.81 1.00 1.28 0.18 0.49 0.85
ratio
100% ρmin 0.26 0.69 1.34 0.22 0.47 0.69
Type of Deformed bars 0.24 0.69 1.40 0.20 0.46 0.69
web reinfor- Welded-wire mesh 0.00 0.23 0.91 1.00 1.09 0.20 0.56 0.99
cement No reinforcement 0.20 0.82 1.19 0.05 0.20 0.25
Quasi-static monot. 0.18 0.67 1.17 0.11 0.29 0.43
Type of
Quasi-static cyclic 0.00 0.25 0.76 1.00 1.31 0.20 0.47 0.76
testing
Shake table 0.23 0.83 1.38 0.27 0.63 0.98
5 CONCLUSIONS
Damage quantification based on visual inspection of cracking pattern is a subjective estimate because
the damage criterion depends on the expertise of the inspector. Although characteristics of cracks
(length, maximum width, residual width) are a key indicator of structural damage, pattern and
distribution of cracks of the damaged structural component should also be considered. The damage
observed in thin reinforced concrete (RC) walls for low-rise housing subjected to seismic demands is
evaluated in this study by means of fractal dimension of the cracking propagation. The study was aimed
at establishing a new evaluation method of seismic damage based on the fractal dimension of cracking.
Variables of the experimental program were the aspect ratio of the wall, type of concrete, steel ratio and
type of web shear reinforcement, and type of testing. To improve the quantitative analysis of structural
damage under a particular seismic excitation, an empirical damage index was developed in this study.
It was demonstrated that other models proposed in the literature can provide a non-conservative
estimation of damage stage of walls having the particular characteristics of low-rise housing.
ACKNOWLEDGEMENT
The author gratefully acknowledges the financial support from the Research Office (Vicerrectoría de
Investigaciones) at Nueva Granada Military University (UMNG, Colombia), through project number
6
IMP-ING-1574. The views expressed in this paper are solely those of the authors and do not necessarily
reflects the views of the sponsor.
REFERENCES:
ACI Committee 318 2011. Building code requirements for structural concrete (ACI 318-11) and commentary.
American Concrete Institute, Farmington Hills, MI, 503 pp.
Adhikari, R., Moselhi, O. and Bagchi, O. 2013. Image-based retrieval of concrete crack properties for bridge in-
spection. Journal of Automation in Construction, 39, 180-194, doi:10.1016/j.autcon.2013.06.011.
Carrillo J. 2015a. Damage index based on stiffness degradation of low-rise RC walls. Journal of Earthquake En-
gineering and Structural Dynamics, 44, 831-848. doi:10.1002/eqe.2488.
Carrillo J. 2015b. Evaluation of fractal dimension of cracking recorded on concrete walls for housing subjected to
seismic effects. Proceedings of the International Conference on Earthquake Engineering and Seismology,
Kiel, Alemania, Paper 211.
Carrillo J. and Alcocer S. 2012a. Acceptance limits for performance-based seismic design of RC walls for low-
rise housing. Journal of Earthquake Engineering and Structural Dynamics, 41, 2275-2280.
DOI:10.1002/eqe.2186.
Carrillo J., and Alcocer, S. 2012b. Backbone model for performance-based seismic design of RC walls for low-
rise housing. Journal of Earthquake Spectra, 28(3), 943-964. doi:10.1193/1.4000068.
Chiaia B., Van Mier J. and Vervuut, A. 1998. Crack growth mechanisms in four different concretes: microscopic
observations and fractal analysis. Journal of Cement and Concrete Research, 28(1), 103-114.
Farhidzadeh A., Dehghan-Niri E., Moustafa C, Salamone S. and Whittaker A. 2013. Damage assessment of
reinforced concrete structures using fractal analysis of residual crack patterns. Journal of Experimental
Mechanics, 53, 1607-1609. doi:10.1007/s11340-013-9769-7.
FEMA-306 1998. Evaluation of earthquake damaged concrete and masonry wall buildings. Federal Emergency
Managmanet Agency, Washington, 270 pp.
Hadjileontiadis L. and Douka E. 2007. Crack detection in plates using fractal dimension. Journal of Engineering
Structures, 20, 1612-1620. DOI:10.1016/j.engstruct.2006.09.016.
IAEA 2002. Guidebook on non-destructive testing of concrete structures. International Atomic Energy Agency.
Training Corse Series No. 17, Vienna, Austria. 231 pp.
Li L., Chan P., Zollinger D. G. and Lytton R. 1993. Quantitative analysis of aggregate shape based on fractals.
ACI Materials Journal, 90(4), 357-365.
Miao Y., Song P. and Gong X. 2014. Fractal and multifractal caractheristics of 3D asphalt pavement macrotexture.
Journal of Materials in Civil Engineering, 26(8), 1-11.
Moustafa A., Dehghan E., Farhidzade, A. and Salamore S. 2013. Corrosion monitoring of post-tensioned concrete
structures using fractal analysis of guided ultrasonic waves. Journal of Structural Control and Health Moni-
toring, 21, 438-448, DOI:10.1002/stc.1586.
SEAOC 1995. Vision 2000: performance-based seismic engineering of buildings. Report, Structural Engineers
Association of California, Sacramento, California.
Tzu-Kang Lin A., Chih-Shiuan Lin E., Rih-Teng Wu B., Kuo-Chun Chang C. and Yu-Chi Sung D. (2012). Appli-
cation of fractal theory on bridge health monitoring system. 15th World Conference on Earthquake Engineer-
ing, Paper 2338.
Werner S., Neumann I., Thienel K. and Heunecke O. 2013. A fractal-based approach for the determination of
concrete surfaces using laser scanning techniques: a comparison of two different measuring systems. Journal
of Materials and Structures, 46, 245-254. DOI: 10.1617/s11527-012-9898-y.