Comparison of Bridge Collapse Frequencies With Failure Probabilities

You might also like

Download as pdf or txt
Download as pdf or txt
You are on page 1of 10

See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/323868242

Comparison of Bridge Collapse Frequencies with Failure Probabilities

Conference Paper · September 2017

CITATIONS READS

17 417

1 author:

Dirk Proske
Bern University of Applied Sciences
100 PUBLICATIONS   588 CITATIONS   

SEE PROFILE

All content following this page was uploaded by Dirk Proske on 20 March 2018.

The user has requested enhancement of the downloaded file.


15th International Probabilistic Workshop, Dresden 2017

Comparison of Bridge Collapse Frequencies


with Failure Probabilities
Dirk Proske

Axpo Power AG,


dirk.proske@axpo.com

Abstract: In this paper the collapse frequencies of bridges are estimated based on various
publications related to different regions and different times. The various publications are
considered in terms of a meta-analysis to provide a robust estimation of the collapse fre-
quencies and to identify trends of the collapse frequencies. The values are then compared
to probabilities of failure which are the target values.

1 Introduction

1.1 General introduction

Recently the author has compared the computed probabilities and observed frequencies of
core damages in Nuclear Power Plants [1]. This comparison was accompanied by a prelim-
inary comparison of the probabilities of failure and the collapse frequencies of structures.
In this paper the comparison of the probabilities of failure and the collapse frequencies
focuses on bridges only and the database for the collapse frequencies is significantly ex-
tended to provide more robust conclusions.

Bridges form a substantial part of the infrastructure systems in almost all countries world-
wide. The global stock is estimated between five and six million bridges (see for example
VOGEL ET AL. [7]). In industrialised countries approximately one bridge per 500 inhabitants
and in developing countries about one bridge per 2000 inhabitants is counted.

Bridges are designed to function safe over an extreme long period, usually 100 years.
However many cases are known in which bridges function more than hundred years, in
some cases more than thousand years.

Whereas early design concepts of bridges were purely based on empirical rules, modern
safety concepts of bridges are based on certain reliability or risk measures. One of the most
important reliability measures is the computed probability of failure pf (POF) or the safety
index ȕ, which is a substitute of the POF. For new as well as existing bridges target values
of the POF are given in certain codes.

15
15th International Probabilistic Workshop, Dresden 2017

Besides the computed POF values, the observed frequency of bridge collapse (FOC) can be
computed based on the number of bridge collapses and the overall bridge stock for a cer-
tain period and region. Whereas the POF is a bottom-up approach, the FOC is a top-down
approach. Finally both parameters should theoretically yield to the same result.

1.2 Terms and Definitions

In this section terms and definitions are introduced.

1.2.1 Definition of the term “Bridge”

A South-African definition of a bridge states: A bridge is a structure erected with a deck


for carrying traffic over or under an obstruction and with a clear span of six metres or
more. Where the clear span is less than six metres, reference is to a culvert. [2] A German
definition considers bridges as all transitions of a traffic route over another traffic route,
over a body of water or over a lower ground, if the clearance between the abutments is
2.00 meters or more [3].

1.2.2 Definition of the term “Collapse”

A bridge collapse occurs when an entire bridge or a substantial part comes down, at which
point the structure loses its ability to perform its function [14]. Another definition consid-
ers collapse as the development of a kinematic chain, which yields to total or partial de-
struction of the bridge. A local cross section failure or the exceedance of certain limit state
values does not necessarily yield to a collapse since the bridge may remain in a condition
in which it can be repaired [4].

1.2.3 Definition of the term “Frequency of Bridge Collapse”

The frequency of bridge collapse (FOC) is the ratio of the number of collapsed bridges
compared to the bridge stock number related to a certain time scale (usually one year) and
to a certain region or country.

1.2.4 Definition of the term “Probability of Failure”

The probability of failure pf as a measure of safety can be referred to one year or the life
time of the structures in n years. The measure is based on limit state function g(X) and in-
put random variables x:

pf ³ ... ³
g (X) d 0
f X ( x )dx (1)

p f (n) 1  (1  p f ) n (2)

16
15th International Probabilistic Workshop, Dresden 2017

Since the probability of failure is extremely low due to the high safety requirements, a sub-
stitute measure, the safety index ȕ, has been heavily applied instead of the probability. The
safety index ȕ is defined as:
pf ) 1 ( E ) (3)

ĭ-1 represents the inverse standard normal distribution.

In general, the probability of failure pf is not intended to be a true predicator of the collapse
frequency of bridges. Many possible influence factors, such as human errors or correlations
between different limit states seem neither to be considered at all nor considered to an ade-
quate extent (see Eurocode 0, Table 2.3, Note 2). However, if the POF is considered as an
efficient, effective and robust safety measure, it must prove its worth in reality.

2 Method

This study is carried out as a meta-analysis. A meta-analysis uses the results of former
studies and combines them to achieve further information. This study mixes data from dif-
ferent countries (Germany, UK, US and Japan) and different time scales. Therefore care
has to be taken to draw conclusions for other regions.

3 Comparison of observed FOC and target POF values

3.1 Observed frequency of bridge collapse

Figure 1 combines the FOC given by various authors for different time periods and differ-
ent regions [5] - [16]. First, it becomes clearly visible that there is a decreasing trend over
the last century. FOC of bridges have been decreased by nearly two orders of magnitude.

In Fig. 1 the trend is visualised by two lines, a linear decreasing trend given by COOK [10]
and a nonlinear trend introduced by the author.

Additionally some values are surprisingly low compared to the majority of studies, such as
the values by IMHOF [6] and MCLINN [16]. These numbers are probably based on an in-
complete population. However, several publications and their related numbers form a clus-
ter with highly comparable numbers (rectangle in the diagram). This cluster has a lower
bound by VOGEL [7] and an upper bound by COOK [10]. The data in the cluster is related to
industrialized countries and the data is based on the last decades.

17
15th International Probabilistic Workshop, Dresden 2017

Figure 1: Time dependent Bridge Collapse Frequency based on various references

Still, the difference in the cluster is in the range of one order of magnitude. It would be
interesting to identify the cause of this difference. Therefore the cause of collapses for the
references with high FOC is compared to the references with low FOC. Figure 2 illustrates
the results as histogram of the cause of bridge collapses. COOK [10] and HARIK [11] show
the highest FOC. In COOK collapse is dominated by flooding (scour and flood), whereas in
HARIK [11] impact and overload are dominating. Since for VOGEL ET AL. [7] no detailed
information are available, SCHEER [17] data has been used, which should be comparable
due to the neighbourhood regions (Switzerland and Germany). Based on SCHEER [17] most
collapses in Germany are related to the construction time and to impact.

Based on the diagram it can be concluded that no systematic difference is found, although
the data from COOK [10] and HARIK [11] is mainly related to accidental loads. It is well
known that accidental loads significantly influence the FOC since the collapse of many
bridges in one year due to one major event can heavily shift the FOC. For example the
earthquake and tsunami 2011 in Japan destroyed more than 300 bridges [21].

18
15th International Probabilistic Workshop, Dresden 2017

Figure 2: Cause of Bridge Collapses based on various references

3.2 Target Probabilities of Failure

3.2.1 Introduction

The FOC shall be compared to the target POF. However the target POFs are usually not
related to the entire structure, but rather to the individual limit states. Therefore the target
values have to be converted to the structure level. Additionally the POF intentionally does
not consider several influence parameters as mentioned before.

3.2.2 Target values on limit state level

Several codes give target values for the POF and the safety index such as the Eurocode 0
[24], the DIN 1055-100 [22], the ISO 8930 [25], JCSS Model Code [23] and others. The
target values are often based on optimisation requirements or have an historical back-
ground and are usually in the range of 10-6 per year for ultimate limit states (Fig. 1).

3.2.3 Conversion to structure level

The combination of the individual POF on the limit state level to the structural level can be
carried out to a limited extend by probabilistic computations such as Monte Carlo Simula-
tions using multi surfaces and multi limit states respectively. Additionally bounds are giv-
en for system failure probabilities. A third approach is Fault and Event Tree techniques
including new methods for the consideration of correlations and common cause respective-
ly (e.g. balancing method).

19
15th International Probabilistic Workshop, Dresden 2017

However for most structures these techniques are not applied since it is assumed that the
structural elements show a high correlation, in other terms, fails one component, fail all
components.

Based on this conclusion, the theoretical POF should be slightly above 10-6 per year, ap-
proximately in the range of 10-5 per year. Here it is assumed:
pf structure = pf Limit State u p HF / c (4)

pf structure = 10-6 u 1.10 / 0.80 = 1.4 u 10-6

with c as overall correlation value and p HF as human failure probability. However the re-
sult of pf is in the range of the variation of the target values for the single limit states be-
tween different codes.

3.2.4 Time-dependent Target Values

Furthermore time-dependent target values of the POF for existing structures have been
introduced. Figure 3 shows the FOC and the trend of the target POF given SIA 269 [20]. It
clearly shows that the target value has an opposite trend compared to the FOC seen over
the last decades.

Figure 3: Comparison of the observed FOC trend and the function of the POF

20
15th International Probabilistic Workshop, Dresden 2017

3.3 Comparison

Observed FOC of bridges are in the range of 10-5 to 10-4 per year related to the last decades
and to industrialised countries. Target values of POF on the structural level are probably in
the range of 10-6 to 10-4 per year. However, as shown in section 3.1 the FOC values are
often impacted by catastrophic events, e.g. accidental loads affecting many structures such
as floods or earthquakes. This could explain why the FOC values are slightly larger. How-
ever, in general the values comply very well.

4 Possible Future Trends

Whereas the observed FOC slightly decreased in the US and Europe in the last decades, the
target POF values for existing structures suggest an increasing POF. If the POF target
curve is true, then the FOC should only decrease in times, when more old bridges were
replaced by new ones. If the codes were always be the same, the level of FOC should be
constant, however it is slightly decreasing. This decrease can be caused by

x better design (planning),

x better construction and

x better maintenance and usage.

or combinations in comparison to former times. We do not assume that no critical loadings


occurred. It is assumed that:

x bridges are better designed today,

x bridges are not better build today and

x bridges are better maintained and usage is better controlled.

The new design is related only to a small number of bridges per year in relation to the
overall bridge stock of 120 000 in Germany or more than 600 000 in the US. Therefore the
influence of new bridges on the FOC is limited. In contrast, better maintenance strategies
would be related more or less to the entire bridge stock which yields to improvements and
which is probably visible in the trend.

5 Conclusion

Surprisingly, the comparison between the probabilities of failure and the collapse frequen-
cies reveals a fairly good agreement between the target values and the observed values.
This is a strong indicator, that the simplification in the numerical design proofs, e.g. ne-
glecting human error, is reasonable. However, interestingly the trend of the time-dependent

21
15th International Probabilistic Workshop, Dresden 2017

target values and observed values deviates. This may be related to better design and better
maintenance strategies in the last decades.

6 References

[1] Proske, D.: Differences between Probability of Failure and Probability of Core
Damage. 14. International Probabilistic Workshop, Ghent, 2016, Springer: Cham,
pg. 109-122

[2] Wolhuter, K.M.: Geometric Design of Roads Handbook, CRP Press, Taylor &
Francis Group, Boca Raton, 2015

[3] DIN 1076: Ingenieurbauwerke im Zuge von Strassen und Brücken, November 1999

[4] Klingmüller, O.; Bourgund, U.: Sicherheit und Risiko im Konstruktiven Ingenieur-
bau, Vieweg, Wiesbaden, 1992

[5] Zerna, W.: Grundlage der gegenwärtigen Sicherheitspraxis in der Bautechnik, In:
Große technische Gefahrenpotentiale: Risikoanalysen und Sicherheitsfragen, Hrsg.
S. Hartwig, Springer Verlag: Berlin/Heidelberg, 1983, pp 99-109

[6] Imhof, D.: Risk Assessment of Existing Bridge Structures, University of Cambridge,
Dissertation, Kings College, December 2004

[7] Vogel, T.; Zwicky, D.; Joray, D.; Diggelmann, M.; Hoj, N.P.: Tragsicherheit der
bestehenden Kunstbauten, Sicherheit des Verkehrssystems Strasse und dessen Kun-
stbauten, Bundesamt für Strassen, Dezember 2009, Bern

[8] Taricska, M.R.: An Analysis of Recent Bridge failures in the United States (2000-
2012), MSc. Thesis, The Ohio State University, 2014

[9] Lee, G.C.; Mohan, S.B.; Huang, C.; Fard, B.N.: A Study of U.S. Bridge Failures
(1980-2012), Technical Report MCEER-13-0008, Jun 15, 2013, University at Buf-
falo, State University of New York

[10] Cook, W.: Bridge Failure Rates, Consequences, and Predictive Trends. Disserta-
tion, Utah State University Logan, 2014

[11] Harik, I.E., Shaaban, A.M., Gesund, H., Valli, Y.S., Wang, S.T. United States
bridge failures, 1951-1988, Journal of Performance of Constructed Facilities
(ASCE), 1990, Volume 4, Ausgabe 4 pp 272-277

[12] Sharma, S.; Mohan, S.: Status of Bridge Failures in the United States (1800-2009).
TRB 90th Annual Meeting: Transportation, Livability, and Economic Development
in a Changing World, Washington D.C., 2011

[13] Hersi, M.: Analysis of Bridge Failure in United States (2000-2008), Ohio State
University, Master of Science Thesis, 2009

22
15th International Probabilistic Workshop, Dresden 2017

[14] Wardhana, K.; Hadipriono, F.C.: Analysis of Recent Bridge Failures in the United
States, Journal of Performance of Constructed Facilities, ASCE, August 2003, pp
144-150

[15] Christian, G.A: Bridge Failures – Lessons learned, Bridge Engineering Course,
University at Buffalo, March 29, 2010, http://mceer.buffalo.edu/education/
bridge_speaker_series/2009-2010/presentations/P1%20Lessons%20learned%20
from%20Bridge%20Failures_FINAL.pdf

[16] McLinn, J.: Major Bridge Collapses in the US and Around the World, IEEE Relia-
bility Society 2009 Annual Technology Report. IEEE Transactions on Reliability,
Volume 59, Heft 3, September 2010, 5 pp

[17] Scheer, J.: Versagen von Bauwerken: Band 1: Brücken. Verlag Ernst und Sohn:
Berlin, 2000

[18] Bast (2016) Brücken an Bundesfernstrassen, Brückenstatistik, http://www.bast.de/


DE/Statistik/Bruecken/Brueckenstatistik.pdf?__blob=publicationFile&v=6

[19] Bundesministerium für Verkehr, Bau und Stadtentwicklung: Bericht Strategie zur
Ertüchtigung der Strassenbrücken im Bestand der Bundesfernstrassen, 22. Mai
2013, Berlin

[20] SIA 269 2007: Schweizerischer Ingenieur- und Architektenverein: SIA 269:
Grundlagen der Erhaltung von Tragwerken; Schweizerischer Ingenieur- und Archi-
tektenverein, Zürich Entwurf 03/2007.

[21] Maruyama, K., Tanaka, Y., and Hosoda, A. (2012). Damage of Bridges Structures
by Huge Tsunami and Evaluation of Tsunami Force on Bridges. The 8th Interna-
tional Symposium on Social Management Systems, SSMS2012- Disaster Prevention
and Reconstruction Management. 2-4 May 2012, Kaohsiung, Taiwan

[22] DIN 1055-100: Einwirkungen auf Tragwerke, Teil 100: Grundlagen der Tragwerks-
planung, Sicherheitskonzept und Bemessungsregeln

[23] JCSS - Joint Committee of Structural Safety: Probabilistic Model Code,


http://www.jcss.byg.dtu.dk, 2015

[24] Eurocode 0: EN 1990 Basis of Structural Design, 2nd Edition, Draft 30 April 2017

[25] ISO 8930: General principles on reliability of structures, 1987, Edition 1

23
View publication stats

You might also like