Safety, Reliability and Risk Analysis: Beyond the Horizon – Steenbergen et al.

(Eds)
© 2014 Taylor & Francis Group, London, ISBN 978-1-138-00123-7

Probabilistic reliability assessment of steel structures exposed to fatigue

M. Krejsa
VSB, Technical University Ostrava, Czech Republic

ABSTRACT:  The paper describes methods used for probabilistic assessment of reliability of steel struc-
tures and bridges that are exposed to cyclic loads. Propagation of fatigue cracks from surface and edges is
taken into account, the maximum permitted dimension being of particular attention. The model is based
on a linear fracture mechanics. Conditional probability is the basis when designing a regular system of
inspections for the structure. A new method, which is still under development, has been used for probabi-
listic modeling of fatigue damage. Direct Optimized Probabilistic Calculation-DOProC-appears to be a
very efficient for the computation of probabilities. DOProC provides the solution with only a numerical
error and an error resulting from input and output quantities discretizing.

1  INTRODUCING INTO PROPAGATION critical size. The critical size is the final recorded
OF FATIGUE CRACKS size before a brittle fracture appears and the struc-
ture fails. As this type of failure has not occurred
Reliability of the load-carrying structure has been in construction sites for several decades, it would
significantly influenced by degradation result- be advisable to use another method to specify
ing, in particular, from fatigue of basic ­materials. the acceptable final size. Building structures and
­Wöhler’s curves are used when designing such bridges are dimensioned for extreme loads. Fatigue
structures. The service life can be limited and the loads are investigated only in details that are liable
structure can be used until a failure occurs. The to fatigue cracks induced by variable service loads
failure is, however, very difficult to determine. For (Gocal et al., 2010). If the entire load-­carrying ele-
purposes of the modeling, the oscillation ampli- ment is designed exactly for effects of the extreme
tude is considered to be constant, and a certain load with a reasonable safety margin, the crack
number of load cycles is taken into account. The failure will negatively influence the designed
method has been developed which describes real situation.
conditions, all this making the work of design engi- The fatigue crack damage depends on a number
neers easier (Gottvald et al., 2012). of stress range cycles. This is a time factor, which
There are methods under development that influences reliability for the entire designed ­service
would be able to reveal potential defects and dam- life. It is assumed that the failure rate increases in
age resulting from initial cracks that accelerate con- the course of time (this occurs evidently ­during
siderably the propagation of fatigue cracks. Linear corrosion, for instance) and the service life is
fracture mechanics is among alternative methods. reduced (e.g. Holicky et al., 2010). If the propaga-
Experts in mechanics have been dealing with such tion of the fatigue crack is included into the failure
issues for many years. Results have been gradually rate, it is necessary to investigate the fatigue crack
taken over and implemented into designs of the and define the maximum acceptable weakening.
loading structures in buildings (Anderson, 2005). The weakening depends on the acceptable crack
Three sizes are important for the characteris- size, which comprises safety margins for the critical
tics of the propagation of fatigue cracks. The first crack size that may occur in consequence of a brit-
size is the initial size of the crack that corresponds tle fracture and, which is more often in steel struc-
to a random failure in an element subject to ran- tures, in consequence of a ductile fracture. The
dom loads. The random characters of occurrence reason for this type of degradation of a structural
and, in particular, the initial size has been an open element in the course of time is the random exist-
issue all over the world and does not affect build- ence of the initial crack and propagation of the
ing structures only. Existence of the initial cracks crack in the consequence of variable load effects.
during the propagation should be revealed, along The result is the weakening of the element that has
the detectable size of the crack, during inspec- been designed for extreme load effects. The crack
tions. The third important size is referred to as a propagates in a stable way until it reaches the

2671
acceptable size that is a limit for the required reli- make it possible to reach the required reliability if
ability. The acceptable fatigue crack is the size that the acceptable damage method is used. The method
might be achieved, in cross-sections and elements describes probabilistic calculation in modeled prop-
of steel structures and steel bridges dimensioned agation of a crack up to the size, which is the speci-
for the combined extreme loads, as a result of fied acceptable size for real structural details.
gradual degradation when the required reliability
is reached at the end of the designed service life of
2  PROBABILISTIC APPROACH TO
the structure. Only one dimension of the fatigue
PROPAGATION OF FATIGUE CRACKS
cracks is monitored for purposes of the fatigue
crack propagation, while the weakening relates to
Three sizes are important for the characteristics
the entire crack surface. Shapes of the crack sur-
of the propagation of fatigue cracks. These are
faces are different for the cracks originating from
the initial size, detectable size and acceptable size,
the surface or edge initial cracks.
which occur prior to failure caused by a fragile or
Complex and demanding monitoring of crack
ductile crack.
propagation for acceptable sizes have not been
The fatigue crack that deteriorates a certain
developed yet to the extent which would make
area of the structure components is described
it possible to use the method for general/stand-
with one dimension a only when investigating the
ardized applications. The probabilistic methods
­propagation. In order to describe the propagation
should be applied when investigating the propaga-
of the crack, the linear elastic fracture mechanics is
tion rate of the fatigue crack until the acceptable
typically applied. This method uses Paris-Erdogan’s
size is reached because the input variables include
law and defines relation between propagation rate
uncertainties and reliability should be taken into
of the crack da dN and the stress range rate coeffi-
account. The most important inputs are the ini-
cient ∆K in the face of the crack (Paris et al., 1960):
tial crack size and the acceptable crack size. The
definition of the acceptable crack size is a neces-
da
sary, but not the only one, condition because the = C ⋅ ( ∆K )m , (1)
initial crack size is most important for the crack dN
propagation.
Standards introduce two standardized assess- where C, m are material constants, a is the crack
ment methods for evaluation of fatigue crack size and N is the number of loading cycles. The
resistance of steel structures. Locations that tend fatigue crack will propagate in a stable way only
to be liable to fatigue cracks should be pre-treated if the initial crack a0 exists in the place, where the
in a rather complicated way during the fabrication stress is concentrated. This place is located at the
of the structures. If defects are not pre-treated, edge or on the surface of the element.
they may result in initial cracks. The first method The primary assumption is that the primary
is the safe pre-treating service-life method: it does design should take into account effects of the
not expect the cracks to be propagating from the extreme loading and the fatigue resistance should
initial sizes. This, however, does not mean that the be assessed then. This means, the safety margin in
cracks will not appear during the designed service the technical probability method is:
life of the structure. Such cracks must not, how-
ever, exceed the acceptable crack size. The accept- Z( R,S ) = RF = R − S , (2)
able damage method suggests by its name that
acceptable damage should be defined using that where R is the random resistance of the element
method. But conditions are rather general (Kotes and S is the random variable effect of the extreme
et al., 2012). It permits occurrence of the accept- load. If such an element is subject to the service
able crack size during the designed service life of load, following cases can occur (Fig. 1):
the structure. Damage is caused there by a defect
a. safe service life—the fatigue effects do not degrade
that has not been corrected. Such defect becomes
the element by means of the fatigue crack,
an initial crack. The expected crack size or non-
b. acceptable failure rate—the fatigue effects
­existence should be revealed during a special
degrade the element and decrease the load-
system of inspections (e.g. Moan, 2005; Van der
­carrying capacity of the element,
Weide et  al., 2010; Krejsa, 2012). Those inspec-
c. acceptable failure rate—fatigue effects are
tions are considerably more important than stand-
expressed as stress changes.
ard inspections. Both times and required quality
are of essence in the inspections. The calculation model of the fatigue crack
The description below focuses on clarification propagation defines the stress when the maximum
and specification of certain requirements defined acceptable crack results in the constant resistance
in standards. Such specifications and requirements of the structure, R, that corresponds to the stress in

2672
 If the length of the crack a1 equals to the ini-
tial length a0 (this is the assumed size of the ini-
tial crack in the probabilistic approach) and if a2
equals to the final acceptable crack length aac (this
is the acceptable crack size which replaces the criti-
cal crack size acr if the crack results in a brittle frac-
ture), the left-hand side of the equation (5) can be
regarded as the resistance of the structure—R:
aac
da
R( aac ) = ∫ . (6)
( )
m
a0 π ⋅ a ⋅ F( a )

Similarly, it is possible to define the cumulated

load effect that is equal to the right side (randomly
variable effects of the extreme load) (5):
N

∫ C ⋅ (∆σ )
m
S= dN = C ⋅ ( ∆σ )m ⋅ ( N − N0 ), (7)
N0

where N is the total number of stress range peak

Figure 1.  Safety margin Z related to designed and oper-
ating load effects. ∆σ when the crack size increases from a0 to aac, and
N0 represents the number of load cycles in the time
of the fatigue crack initial (it is typically equal to
zero).
the yield point fy. The approach c) is more demon- The reliability function RF can be defined as
strative and has been preferred to the approach b) follows:
because it describes the non-linear growth of the
both stresses in the element under degradation. RF( Χ ) = R( aac ) − S( N ) , (8)
When using (1), the condition for the acceptable
crack length, aac, is: where X is a vector of random physical proper-
aac
ties such as mechanical properties, geometry of
1 da the structure, load effects and dimensions of the
N=
C ∫ ∆K m
> Ntot , (3) fatigue crack.
a0
The failure probability pf equals to:
where N is the number of cycles needed to increase
the crack from the initial size a0 to the acceptable ( ) (
p f = P RF( Χ ) < 0 = P R( aac ) < S( N ) . ) (9)
crack size aac, and Ntot is the number of cycles
throughout the service life.
The equation for the propagation of the crack 3  USING THE CONDITIONED
size (1) needs to be modified for this purpose. The PROBABILITY TO DETERMINE TIMES
stress range rate coefficient, ∆K with the constant TO INSPECT THE CONSTRUCTION
stress range ∆σ is:
Because it is not certain in the probabilistic cal-
∆K = ∆σ ⋅ π ⋅ a ⋅ F(a) , (4) culation whether the initial crack exists and what
the initial crack size is and because other inac-
where F(a) is the calibration function which repre- curacies influence the calculation of the crack
sents the course of propagation of the crack. After propagation, a specialized inspection is neces-
the change of the number of cycles from N1 to N2, sary to check the size of the measureable crack
the crack will propagate from the length a1 to a2. in a specific period of time. The acceptable crack
Having modified (1) and using (4), the following size influences the time of the inspection. If no
formula will be achieved: fatigue cracks are found, the analyses of inspec-
tion results give conditional probability during
a2
da
N2 occurrence.
∫ = ∫ C ⋅ ( ∆σ )
m
dN . (5) While the fatigue crack is propagating, it is pos-
( )
m
a1 π ⋅ a ⋅ F( a ) N1 sible to define three random phenomena that are

2673
related to the growth of the fatigue crack and may Those three phenomena cover the complete
occur in any time, “t”, during the service life of the spectrum of phenomena that might occur in the
structure (Krejsa et al., 2011). Then: t-time. This means:
U(t) phenomenon: No fatigue crack failure has been P (U (t ) ) + P ( D(t ) ) + P ( F(t ) ) = 1. (16)
revealed within the t—time and the fatigue crack
size a(t) has not reached the detectable crack size, The probable course of the growth of the fatigue
ad. This means: crack is shown in Figure 2.
a(t ) < ad , (10) The probabilities in eqs. (13) through (15) can be
determined in any period of time, t, using any of the
probabilistic methods. The probabilistic calculation
D(t) phenomenon: A fatigue crack failure has been is carried out in time steps where one step typically
revealed within the t—time and the fatigue crack equals to one year of the service life of the construc-
size a(t) is still below the acceptable crack size aac. tion. When the probability of failure P(F(t)) reaches
This means: the designed failure probability pd, an inspection
should be carried out in order to find out fatigue
ad ≤ a(t ) < aac , (11) cracks, if any, in the construction element. The
inspection provides information about conditions
F(t) phenomenon: A failure has been revealed within of the construction. Such conditions can be taken
the t—time and the fatigue crack size a(t) has into account when carrying out further probabil-
reached the acceptable crack size aac. This means: istic calculations. The inspection in the t-time may
result in any of the three mentioned phenomena.
a(t ) ≥ aac . (12) Using the inspection results for the t-time, it is
possible to define the probability of the mentioned
phenomena in times T > tI. For that purpose, the
If the crack is not revealed within the t-time,
conditional probability should be taken into con-
this may mean that there is not any fatigue crack
sideration (probability of A if B has occurred):
in the construction element. This might be also
an initiative phase of nucleation of the fatigue P (A ∩ B)
crack (when a crack appears in the material) and P (A B ) = , kde P ( B ) ≠ 0. (17)
this phenomenon is not taken into account in the P (B)
fracture mechanics (Routil et al., 2012; Seitl et al.,
2012). Even if the fatigue crack is not revealed it is The probability that the U phenomenon occurs
likely that it exists there but the fatigue crack size in tI is:
is so small that it cannot be detected under existing
P (U (T ) ∩ U (tI ) ) P (U (T ) ) 
conditions.
Using the phenomena above, it is possible to
(
P U (T ) U (tI ) =) P (U (tI ) )
= 
P (U (tI ) ) 

) ( P(U )
define following probabilities: P (T ) ∩ U (tI )
D 
– The probability that the failure is not detected (
P D(T ) U (tI ) =
(tI ) )

 ∑ = 1.

within the t-time, this means the probability that
P ( (T ) U (tI ) )
F ∩ 
the fatigue crack size a(t) is below the measurable
crack size ad: (
P F(T ) U (tI ) = ) P (U (tI ) )



P (U (t ) ) = P (a(t ) < ad ), (13)  (18)

– The probability that the failure detected

within the t-time has the crack size a(t) that is less
than the acceptable size aac:

P ( D(t ) ) = P (ad ≤ a(t ) < aac ), (14)

– The probability that the failure occurs within

the t-time, this means the probability that the
fatigue crack size a(t) reaches the acceptable
size aac:
Figure  2.  Probabilistic growth of the fatigue crack in
P ( F(t ) ) = P (aac ≤ a(t ) ). (15) the course of time.

2674
 The probability that the D phenomenon occurs to time the inspections well, the equation (21)
in tI is: is most important. It defines the failure prob-
ability in T  >  tI provided that no fatigue cracks
P (U (T ) ∩ D(tI )) P ({0}) 
(
P U (T ) D(tI ) = ) P (D(tI ))
= = 0
P (D(tI )) 
have been revealed during the last inspection. It
is clear from the equation that the results of the
failure probability are influenced by mutual rela-
P (D(T ) ∩ D(tI )) 
(
P D(T ) D(tI ) ) =
P (D(tI )) 


 ∑ = 1.

tions between the three crack sizes—the initial
crack size, measurable crack size and acceptable
∑ = 1 crack size.
P (F(T ) ∩ D(tI ))  
(
P F(T ) D(tI ) ) =
P (D(tI )) 


The probabilities in the equation (21) can be cal-
culated in any T  >  tI time using any probabilistic

method. When the failure probability (21) reaches
 (19) the designed failure probability pfd, an inspection
should be carried out in order to reveal fatigue
The probability that the F phenomenon occurs cracks, if any, in the construction component. The
in tI is: inspection may result in one of the three ­mentioned
phenomena with corresponding probabilities. The

) ( P( ()F ) ) = PP(F({0})) = 0 
PU ∩F 
( T (tI ) entire calculation can be repeated in order to ensure
P U (T ) F(tI ) = well-timed inspections in the future.
(tI ) (tI )

(
P D(T ) ∩ F(tI ) P ({0}))= 
(
P D(T ) F(tI ) ) =
P (F(tI )) P (F(tI ))
= 0 ∑ = 1.

4  APPLYING THEORETICAL APPROACH
TO PROPAGATION OF FATIGUE

P (F(T ) ∩ F(tI )) P (F(tI ))
CRACKS
(
P F(T) F(tI ) ) =
P (F(tI ))
=
P (F(tI ))
=1 
 Fatigue cracks appear most frequently in decks

of railway or road bridges. The fatigue cracks
 (20) may appear easily because each force induced by
vehicle represents one loading cycle. The load-
In order to specify the time for the next inspec- ing effects are more evident if the construction
tion, it is necessary to determine the conditional element is located close to the point of loading
probabilities, which can be expressed using the full application.
probability rule: An important factor influencing occurrence of
the fatigue crack is the weld itself because inter-
(
P F(T ) U (tI ) ) nal tension and initial cracks may appear because

( ) ( ) ( ),
of poor workmanship. Where the cross-section
=
( )
P F(T ) − P F(tI ) − P D(tI ) ⋅ P F(T ) D(tI ) changes suddenly, the fatigue damage can be influ-
P (U ) (tI )
enced by differences during the real tension (where
peaks appear in the weld) and designed tension (an
 (21) even tension in the flange).
Depending on the position of the initial crack, it
and is possible to monitor the crack propagating from
the edge or surface (Krejsa et  al., 2011). Regard-
(
P F(T ) D(tI ) ) ing the frequency, relevance and concentration of
stresses, those locations rank among those with

=
(
P ( F(T ) ) − P ( F(tI ) ) − P (U (tI ) ) ⋅ P F(T ) U (tI ) ). the major hazard of fatigue cracks appearing in
the steel structures and bridges. The cases are dif-
P ( D(tI ) ) ferent in calibration functions F(a) and in weakened
 (22) surfaces which are appearing during the crack
propagation.
If re-distribution of stress from a point that is The acceptable crack size aac can be described
weakened by the crack is not taken into account, for the fatigue crack from the edge by a formula
the crack propagation is in the range of detectable resulting from the reduced weakening of the cross-
values usually rather high in practical cases. If a section area of the flange:
fatigue crack is found during the inspection, it is
necessary to monitor the safe growth of the crack  σ 
or to take actions that will slow down or stop fur- aac = b ⋅ 1 − max  . (23)
ther propagation of the fatigue crack. In order  fy 

2675
 The calculation of acceptable crack size aac for The fatigue reliability of the structure with cracks
the fatigue crack from the surface is based on propagating from the surface or edge is calculated
relation: using the Direct Optimised Probabilistic Calcula-
tion (“DOProC”) (Krejsa, 2012), which has been
bf t f under development since 2002. The calculation
σ max . ≤ fy . procedure for a certain task in DOProC is clearly
1 0.303 2 
bf t f − π a  . a + 1.020.a + 0.007.t f determined by its algorithm, while Monte Carlo
2  tf  simulation methods generate calculation data for
 (24) simulation on a random basis (Krejsa et al., 1999;
Konecny et al., 2009; Vorechovsky et al., 2009).
It is difficult to describe the crack size a directly In the name of the method, there is the word
explicitly. In order to calculate the acceptable “optimized”. Reasons are following: The number
crack size aac, it is necessary to use a numerical of random quantities, which are used in the calcu-
iteration approach where restrictions resulting lation of the probability of failure, pf, is limited by
from (24) should be taken as a basis (for details see the ability to manage the task numerically. If there
Krejsa et al., 2011; Krejsa, 2012). are too many random quantities, the tasks require
too much time even if advanced computational sys-
tems are used. Therefore, efforts have been made
5  PROBABILISTIC ASSESSMENT to optimize calculations in order to reduce the
OF FATIGUE DAMAGE TO THE number of operations, keeping, at the same time,
STRUCTURE reliable calculation results. Currently, the DOProC
along with the optimizing steps are a good tool
Many calculation methods exist now for design- for rather many probabilistic tasks ­(Krejsa et  al.,
ing and reliability assessments of load-carrying 2013). Detailed theoretical background has been
structures and elements with the specified reli- published (Janas et  al., 2010; Janas et  al., 2012;
ability. Those methods are based on the probabil- Krejsa et al., 2012).
ity theory and mathematic statistics (Bris, 2010; FCProbCalc has been developed using the
Vavrusova et al., 2012) and risk analyses (Praks et al., ­aforementioned techniques. By means of FCProb-
2010). Such methods have been becoming more and Calc (which stands for Fatigue Crack Probability
more popular (Kralik, 2012). The methods which Calculation), it is possible to carry out the proba-
are referred to as probabilistic or stochastic make bilistic modeling of propagation of fatigue cracks
it possible to analyse a safety margin defined by a in a user friendly environment and to propose a
calculation model where at least some input quan- system of regular inspections which should reveal
tities are of a random nature (Tesar et  al., 2008; damage to the structure, using the computational
Krivy et  al., 2011; Kala, 2012). Another category procedures described in Chapters  2, 3 and 4
of advanced computation methods is based on ­(Krejsa, 2012).
artificial neuron networks (Sadovsky et  al., 2010; The input quantities can be determined deter-
Yilmaz et  al., 2012). The calculation procedures ministically or stochastically using non-parametric
contribute to a qualitatively higher level of the reli- (empirical) and parametric probability distribu-
ability assessment and, in turn, higher safety of tions (see Tables 1 and 2). A sample probabilistic
those who use the buildings and facilities. calculation has been carried out for fatigue cracks

Table 1.  Input quantities expressed in a histogram with parametric distribution of probabilities.

Parametric distribution

Parameters

Quantity Type Mean value Standard deviation

Oscillation of stress peaks Δσ [MPa] Normal 30 3

Total number of oscillation of stress Normal 106 105
peaks per year N [–]
Yield stress fy [MPa] Lognormal 280 28
Nominal stress in the flange plate σ [MPa] Normal 200 20
Initial size of the crack a0 [mm] Lognormal 0,2 0,05
Smallest detectable size of the crack ad [mm] Normal 10 0,6

2676
Table  2.  Input quantities expressed in a deterministic cyclic loads. A particular attention is paid to the
way. maximum acceptable crack size. The theoretical
model of fatigue crack progression is based on a
Quantity Value linear fracture mechanics.
Material constant m 3 Propagation of the fatigue cracks and possible
Material constant C 2.2 ⋅ 10-13 forecast of such propagation in time since the start
Width of the flange plate bf [mm] 400 of variable loading effects is the case when proba-
Thickness of the flange plate tf [mm] 25 bilistic methods must be used because too many
Nominal probability of failure pd 0.02277 uncertainties influence the determination of the
input values. The uncertainties include both load-
ing effects and construction resistance (for instance,
the stochastic response to effects of the variable
operation form by oscillation of stress in locations
which are susceptible to fatigue damage). In the
global context, it is the size of the expected initial
crack, which is managed with most difficulties.
The calculation uses the newly developed Direct
Optimized Probabilistic Calculation (“DOProC”),
which is suitable for several probabilistic calcula-
tions. Examples of the probabilistic methods used
in calculations have been proving that the method
is suitable not only for the reliability assessment,
but also for other probabilistic calculations, includ-
ing the propagation of the fatigue cracks. DOProC
appears to be a very efficient tool that results in
Figure 3.  FCProbCalc desktop with results of the prob- the solution affected by a numerical error and by
abilistic modeling of propagation of a fatigue crack from an error resulting from the discretizing of the input
the edge. The result is that the bridge should be inspected and output quantities only.
for the first time after 49 years of operation. The probability of propagation of fatigue cracks
from the edge/surface was calculated in FCProbCalc.
Using this software, it was possible to make proba-
propagating from the surface and edge. The reli- bilistic assessment of the structural reliability on the
ability index for the structure is β = 2, which corre- basis of the exact definition of the acceptable size
sponds to the designed probability Pd = 0.02277. of the fatigue crack. The probabilities were obtained
If a period of time is specified and the step is for three basic phenomena, which are related to
1 year, it is possible to determine load effects, S, propagation of the fatigue cracks. On the basis of
according (7), resistance of the construction R(ad) those data, the probability of failure can be calcu-
and R(aac) according (5) and (6), and the probability lated for each year of operation of the construction.
of elemental phenomena, U, D and F, pursuant to When determining the required degree of reliability,
(10) through (12) which are the basis for specifica- it is possible to specify the time of the first inspec-
tion of inspection times. tion of the structure, which will focus on the fatigue
Figure 3 shows results of the probabilistic mod- damage. Using a conditional probability, times for
eling of a fatigue crack from the edge. The curves subsequent inspections can be determined.
describe dependence of the probability of failure, The methods and application can considerably
pf, on time of operation of the bridge construction. improve estimation of maintenance costs for the
When the probability of failure exceeds the specified structures and bridges subject to cyclic loads.
designed probability, pd, the inspection should be If this methodology is developed further, the
performed. For the task in Figure 3, it was decided goal of investigations seems to be, in particular,
that the first inspection of the bridge should take application of Bayesian networks (Bensi et  al.,
place after 49 years of operation. This inspection will 2013) in the computational model, which describes
focus on growth of the fatigue crack on the edge. propagation of fatigue cracks.

6  CONCLUSION 7  Appendix

This paper describes methods dealing with propa- For a lite version of FCProbCalc and other soft-
gation of fatigue cracks from the edge/surface in ware products based on DOProC method, please,
steel structures and bridges, which are subject to visit http://www.fast.vsb.cz/popv.

2677
Acknowledgement Janas, P., Krejsa, M. & Krejsa, V. 2012. Statistical
dependence of input variables in DOProC method.
The publication has been completed thanks to Transactions of the VŠB—Technical University of
the financial support provided to VSB-Technical Ostrava: Construction Series [online]. Warsaw, Poland:
Versita, vol. 12, issue 2, pp. 1–11 (11 p). ISSN 1804-
University of Ostrava by the Czech Ministry of 4824 (Online); ISSN 1213–1962 (Print). doi: 10.2478/
Education, Youth and Sports from the budget for v10160-012-0017-3.
conceptual development of science, research and Kala, Z. 2012. Geometrically Non-Linear Finite Ele-
innovations for the year 2013. ment Reliability Analysis of Steel Plane Frames
with Initial Imperfections. Journal of Civil Engi-
neering and Management, vol.  18, issue 1, pp.
REFERENCES 81–90 (10 p). doi: 10.3846/13923730.2012.655306,
WOS: 000300478300008.
Anderson, T.L. 2005. Fracture mechanics: ­fundamentals Konecny, P. & Brozovsky, J. & Krivy, V. 2009. Simulation
and applications, Third edition, CRC Press, ­Taylor & Based Reliability Assessment Method using Parallel
Francis Group, Boca Raton, Florida, ISBN 0-8493- Computing. In Proceedings of 1st International Con-
1656-1. ference on Parallel, Distributed and Grid Computing for
Bensi, M., Kiureghian, A.D. & Straub, D. 2013. Efficient Engineering, Civil Comp Proceedings, issue 90, pp. 542–
Bayesian network modeling of systems. Reliability 549 (8 p), ISSN: 1759-3433. WOS: 000271452700038.
Engineering & System Safety, vol. 112, pp. 200–213 (14 Kotes, P. & Vican, J. 2012. Reliability levels for exist-
p). ISSN 0951-8320. doi: 10.1016/j.ress.2012.11.017. ing bridges evaluation according to Eurocodes. In
WOS: 000315251300019. Proceedings of 23rd Czech and Slovak International
Bris, R. 2010. The effective iterative algorithm to Conference Steel Structures and Bridges, vol. 40,
solve a maintenance optimization problem of a pp. 211–216 (6 p), ISSN 1877-7058. doi: 10.1016/j.
highly reliable system. In: Proceedings of European proeng.2012.07.082. WOS: 000314665300037.
Safety and Reliability Conference (ESREL 2009), Kralik, J. 2010. A RSM method for probabilistic non-
Reliability, Risk and Safety: Theory and Applica- linear analysis of reinforced concrete bubbler tower
tions vols 1–3. London: Taylor & Francis Group, structure integrity. In: Proceedings of European Safety
pp. 635–641 (7 p). ISBN 978-0-415-55509-8. WOS: and Reliability Conference (ESREL 2009), Reliability,
000281188500087. Risk and Safety: Theory and Applications vols. 1–3.
Gocal, J., Vican, J., Hlinka, R. & Jost, J. 2010. Labo- London: Taylor & Francis Group, pp. 1369–1372 (4 p.).
ratory tests of a typical fatigue prone riveted steel ISBN 978-0-415-55509-8. WOS: 000281188500188.
railway bridge structural detail. In Proceedings of Krejsa, M. & Tomica, V. 2011. Determination of Inspec-
10th International Conference Fatigue 2010, Procedia tions of Structures Subject to Fatigue. Transactions of
Engineering, vol. 2, issue 1, pp. 1761–1766 (6 p), ISSN the VŠB—Technical University of Ostrava: Construc-
1877-7058, doi: 10.1016/j.proeng.2010.03.189. WOS: tion Series [online]. Warsaw, Poland: Versita, vol. 11,
000278762900187. issue 1, pp. 1–9 (9 p). ISSN 1804-4824 (Online); ISSN
Gottvald, J. & Kala, Z. 2012. Sensitivity Analysis of 1213-1962 (Print). doi: 10.2478/v10160-011-0007-x.
Tangential Digging Forces of the Bucket Wheel Exca- Krejsa, M. 2012. The Probabilistic Calculating of Fatigue
vator SchRs 1320 for Different Terraces. Journal of Crack Propagation Using FCProbCalc Program. In:
Civil Engineering and Management, vol.18, issue 5, pp. Proceedings of 18th International Conference Engineer-
609–620 (12 p). doi: 10.3846/13923730.2012.719836, ing Mechanics 2012, Svratka, Czech Republic. Prague:
WOS: 000309445300001. The Institute of Theoretical and Applied Mechanics AS
Holicky, M. & Markova, J. & Sykora, M. 2010. Assessment CR, v.v.i., pp. 745–754 (10 p). ISBN 978-80-86246-39-0.
of deteriorating reinforced concrete road bridges. In: Krejsa, M. 2012. Probabilistic Calculation of Fatigue Crack
Proceedings of European Safety and Reliability Con- Progression Using FCProbCalc Code. Transactions of
ference (ESREL 2009), Reliability, Risk and Safety: the VŠB—Technical University of Ostrava: Construction
Theory and Applications vols 1–3. ­London: Taylor & Series [online]. Warsaw, Poland: Versita, vol. 12, issue 1,
Francis Group, pp. 1377–1383 (7 p). ISBN 978-0-415- pp. 1–11 (11 p). ISSN 1804-4824 (Online); ISSN 1213-
55509-8. WOS: 000281188500190. 1962 (Print). doi: 10.2478/v10160-012-0003-9.
Janas, P. & Krejsa, M. & Krejsa, V. 2012. Structural Reli- Krejsa, M. 2012. Stochastic Modelling of Fatigue Crack
ability Assessment using a Direct Determined Proba- Progression using the DOProC Method. In: Pro-
bilistic Calculation: paper 72. In: Proceedings of the ceedings of the Eleventh International Conference on
Twelfth International Conference on Civil, Structural and Computational Structures Technology: Civil-Comp
Environmental Engineering Computing: CC 2009, Civil- Proceedings: 99, Dubrovnik, Croacia, 2012. Edited by
Comp Press, (20 p), ISBN 978-1-905088-31-7. Elsevier B.H.V. Topping. Stirlingshire, Scotland: Civil-Comp
B.V., ISBN 978-190508830-0. doi: 10.4203/ccp.91.72. Press, (18 p). ISBN 978-1-905088-54-6, ISSN 1759-
Janas, P., Krejsa, M. & Krejsa, V. 2010. Using the Direct 3433. doi: 10.4203/ccp.99.113.
Determined Fully Probabilistic Method (DDFPM) Krejsa, M. & Janas, P. & Krejsa, V. 2012. Direct Optimized
for determination of failure. In: Proceedings of Euro- Probabilistic Calculation. In: Recent Advances in Sys-
pean Safety and Reliability Conference (ESREL tems Science & Mathematical Modelling: Proceedings
2009), Reliability, Risk and Safety: Theory and of the 3rd International Conference on Mathematical
Applications vols 1–3. London: Taylor & Francis Models for Engineering Science (MMES ‘12). Paris,
Group, pp. 1467–1474 (8 p). ISBN 978-0-415-55509-8. France: WSEAS Press, North Atlantic University
WOS: 000281188500203. Union, pp. 216–221 (6 p). ISBN 978-1-61804-141-8.

2678
Krejsa, M. 2012. Inspection Based Probabilistic Modeling Sadovsky, Z. & Soares, C.G. 2010. Artificial neural net-
of Fatigue Crack Progression. In: Recent Advances in work in probabilistic assessment of strength of thin
Mechanical Engineering & Automatic Control: Pro- imperfect plates. In: Proceedings of European Safety
ceedings of the 3rd European Conference of Mechani- and Reliability Conference (ESREL 2009), Reliability,
cal Engineering (ECME’ 12). Paris, France: WSEAS Risk and Safety: Theory and Applications vols. 1–3,
Press, North Atlantic University Union, pp. 104–109 London: Taylor & Francis Group, pp. 1373–1376 (4 p).
(6 p). ISBN 978-1-61804-142-5. ISBN 978-0-415-55509-8. WOS: 000281188500189.
Krejsa, M., Janas, P. & Cajka, R. 2013. Using DOProC Seitl, S. & Simonova, H. & Kersner, Z. & Fernandez-
Method in Structural Reliability Assessment. Applied Canteli, A. 2012. Evaluation of concrete fatigue meas-
Mechanics and Materials: Mechatronics and Applied urement using standard and non-linear regression
Mechanics II. Zurich, Switzerland: Trans Tech model. Applied Mechanics and Materials: Frontiers of
Publications, vols. 300–301, pp. 860–869 (10 p). Manufacturing and Design Science II, Pts. 1–6, Trans
ISSN 1662–7482. doi: 10.4028/www.scientific.net/ Tech Publications, vols. 121–126, pp. 2726–2729 (4
AMM.300–301.860. p), ISSN 1660–9336. doi: 10.4028/www.scientific.net/
Krejsa, M. & Marek, P. 1999. Transition from determin- AMM.121–126.2726, WOS: 000307425401200.
istic to probabilistic structural steel reliability assess- Tesar, A. & Melcer, J. 2008. Structural monitoring in
ment with special attention to stability problems. advanced bridge engineering. International Journal for
Stability and Ductility of Steel Structures. pp. 19–26 Numerical Methods in Engineering, vol. 74, issue 11,
(8 p), ISBN 0-08-043016-3, WOS: 000083313700003, pp. 1670–1678 (9  p), ISSN 0029–5981, doi: 10.1002/
doi: 10.1016/B978-008043016-4/50003-9. nme.2224. WOS: 000257353700002.
Krivy, V. & Cajka, R. 2011. Design and Reliability Van der Weide, J.A.M. & Pandez, M.D. 2010. Condition-
Assessment of Roof Structural Elements Using the based maintenance of engineering systems with peri-
New Digital Ground Snow Load Map of the Czech odic inspection. In: Proceedings of European Safety
Republic. In Proceedings of Engineering Mechanics and Reliability Conference (ESREL 2009), Reliability,
2011, pp. 335–338 (4 p), ISBN: 978-80-87012-33-8, Risk and Safety: Theory and Applications vols 1–3.
WOS: 000313492700077. London: Taylor & Francis Group, pp. 535–542 (8 p).
Moan, T. 2005. Reliability-based management of inspec- ISBN 978-0-415-55509-8. WOS: 000281188500074.
tion, maintenance and repair of offshore structures. Vavrusova, K. & Lokaj, A. & Zidek, L. 2012. Reliance
Structure and Infrastructure Engineering, Life-Cycle of embedment strength on dowel diameter in joints
Design and Performance, vol. 1, issue: 1, pp. 33–62 of cement-splinter boards. Applied Mechanics and
(30 p), Taylor & Francis Ltd., ISSN 1573-2479. doi: Materials: Applied Mechanics and Civil Engineering
10.1080/15732470412331289314. II, vol. 188, Trans Tech Publications, pp. 242–246
Paris, P.C. & Erdogan, F. 1960. A Critical Analysis of (5  p), ISSN 1660–9336, doi: 10.4028/www.scientific.
crack propagation laws. Journal of Basic Engineering, net/AMM.188.242, WOS: 000310935900041.
ASME, vol. 85, pp. 528–534 (7 p). Vorechovsky, M. & Novak, D. 2009. Correlation control
Praks, P. & Medonos, S. & Raman, R. 2010. Fire loads for in small-sample Monte Carlo type simulations I: A
realistic risk assessment applications in petrochemical simulated annealing approach. Probabilistic Engineer-
industry. In: Proceedings of European Safety and Reli- ing Mechanics, vol. 24, issue 3, pp. 452–462 (11 p), ISSN
ability Conference (ESREL 2009), Reliability, Risk 0266–8920, doi: 10.1016/j.probengmech.2009.01.004,
and Safety: Theory and Applications vols 1–3, Lon- WOS: 000265818100019.
don: Taylor & Francis Group, pp. 2199–2206 (8 p). Yilmaz, I., Marschalko, M., Bednarik, M., Kaynar, O. &
ISBN 978-0-415-55509-8. WOS: 000281188500302. Fojtova, L. 2012. Neural computing models for pre-
Routil, L. & Vesely, V. & Seitl, S. 2012. Fracture Analy- diction of permeability coefficient of coarse-grained
sis of Cube- and Cylinder-shaped WST Specimens soils. Neural Computing & Applications, vol. 21, issue
Made of Cementitious Composites with Various 5, pp. 957–968 (10 p), ISSN: 0941-0643, doi: 10.1007/
Characteristic Length. Key Engineering Materials, s00521-011-0535-4, WOS: 000307552000015.
Advances in Fracture and Damage Mechanics X,
Trans Tech Publications, vol. 488–489, pp.  533–536
(4 p), ISSN 1013–9826, doi: 10.4028/www.scientific.
net/KEM.488–489.533, WOS: 000306531200130.

2679