Safety philosophy for existing structures

and
partial factors for traffic loads on bridges
R.D.J.M. Steenbergen
TNO, Built Environment and Geosciences, Delft, the Netherlands

A.C.W.M. Vrouwenvelder
Delft University of Technology, and TNO, Built Environment and Geosciences, Delft, the
Netherlands

Evaluating and upgrading existing structures becomes more and more important. For a large
part of the existing infrastructure and buildings the design life has been reached or will be
reached in the near future. These structures need to be reassessed in order to find out whether
the safety requirements are met. Not only for new structures but also for the existing stock the
Eurocodes are starting point for the assessment of the safety. However, it would be
uneconomical to require all existing buildings and civil engineering works like bridges to
comply fully with these new codes and corresponding safety levels. The assessment of
existing structures therefore differs from the design situation. This paper describes the main
differences with respect to the relevant reliability requirements and develops a set of partial
factors to be used in reassessment of existing bridges under traffic load.

Keywords: Reliability, existing structures, bridges, partial factors

1 Introduction

For a large part of the existing buildings and infrastructure the design life has been reached
or will be reached in the near future. This is because a huge part of the existing stock has
been built in the sixties of the previous century. These structures need to be reassessed in
order to investigate whether the safety requirements are met. The assessment of existing
structures is becoming more and more important for social and economical reasons, while
most codes deal explicitly only with design situations of new structures. The assessment of
an existing structure may, however, differ much from the design of a new one. Due to
deterioration and damage it is general practice to inspect existing structures and if

HERON Vol. 55 (2010) No. 2 123

 necessary to repair and/or strengthen them. In general, the safety assessment of an
existing structure differs from that of a new one in a number of aspects [1,2]. The main
differences are:

1. Increasing safety levels usually involves more costs for an existing structure than
for structures that are still in the design phase. The safety provisions embodied in
safety standards have also to be set off against the cost of providing them, and on
this basis improvements are more difficult to justify for existing structures. For
this reason and under certain circumstances, a lower safety level is acceptable.
2. The remaining lifetime of an existing building is often less than the standard
reference period of 50 or 100 years that applies to new structures. The reduction
of the reference period may lead to reductions in the values of representative
loads as for instance indicated in the Eurocode for Actions [11].
3. For an existing building or bridge structure actual measurements with respect to
geometry, material properties and behaviour under normal or design
circumstances (e.g. settlements, cracks, corrosion, survival of certain loads, etc.)
may be made in order to reduce uncertainty.

In the following sections, the safety philosophy for existing structures is discussed. First,
briefly the reliability levels in terms of the β values for new structures are given; then for
existing structures the required β values are presented with motivation. Based on this, for
existing bridges under traffic load, the partial safety factors are derived using a full
probabilistic approach.

2 Reliability levels for new structures

Modern safety standards express the safety target for new structures in probabilistic terms
[3-5]. Here, Eurocode EN 1990 [12] is followed. The safety level of a structure or part of a
structure is in principle expressed as the probability of failure for a relevant period of time.
Instead of the probability of failure, however, use is made of the reliability index β. The
reliability index β has a direct relation to the probability of failure P (see Table 1).

In practice the method to establish the desired safety level is by a proper choice of the
following parameters:

124
 - the consequence class for the structures
- the characteristic loads
- the load factors γ f and the combination factors Ψ
- the design rules and material properties
- the material factor γ m

The load and material factors are chosen in such way that a safety level (expressed by β)
belonging to the vigouring consequence class is obtained. Eurocode EN 1990 gives three
consequence classes CC1, CC2 and CC3 [12]. In Table 2, for new structures, the β values
are provided for these consequence classes. For new structures, the subscript n is used for
the β values.

In general, the codified partial factors for loads as well as resistance may be considered as
being in line with these starting points. As Table 2 shows, an exception is made for the
wind loading. Given the standard variable load factor γQ = 1.5, a wind load dominated

building has a lower reliability level then the target value. According to [6, 7], the values
shown in the last column of Table 2 seem to be appropriate. For bridges, important

Table 1: reliability index β and probability of failure P

reliability index β probability of failure P
1.0 0.16
2.0 0.023
3.0 0.0013
4.0 0.000032

Table 2: Reliability index for new structures [12] and [6, 7]

Consequence Consequences of failure Wind load not Wind load
class Loss of human Economic dominating dominating
life damage
1 Small Small βn = 3.3 βn = 2.3

2 Considerable Considerable βn = 3.8 βn = 2.8

3 Very large Very large βn = 4.3 βn = 3.3

125
 buildings and large civil engineering structures, the target value is β = 4.3. The reliability
index is intended to be used in correspondence with a reference period equal to the design
working life of the structures, usually 50 years for buildings and 100 years for bridges.

3 Safety in combination with a shorter design life time

It can be a matter of debate what changes are allowed to the design values of the loads on
structures in the case of a shorter design life time. Usually the shortening of the reference
period for the variable loads (wind, snow, etc) results in a decrease of the representative
values. The Eurocodes give formulas to calculate these reductions. On the other hand,
establishing the safety factors for a shorter design life time, both economic arguments and
limits for human safety play a role. The latter play a role because of the maximum
allowable annual probability of failure, calculated for small probabilities through
Pfailure = Φ( −β)/T with T the design life time in years. This annual probability of failure
may not exceed the limits for human safety. This means that too short life times may lead
to unacceptable large probabilities of failure. In the following sections it turns out that a
minimum design life time of 15 year is therefore to be required in structural design. The
economic arguments and limits for human safety are discussed successively in Section 3.1
and 3.2 and brought together in Section 3.3.

3.1 Economics
If only economic optimization is considered and the failure probability increases
approximately linear in time, it makes sense to use the same target failure probability or
reliability index regardless the design life time. As a result the partial factors do not change
when the design life is changed.

For example the reliability index β = 3.8 corresponds to a probability of failure of about
10 −4 . This probability is for the whole reference period, regardless of the length of it. A
structure designed with β = 3.8 for a period of 1 year has a probability of failure Pf = 10−4 .
A structure designed with β = 3.8 for 50 years has in each arbitrary year a probability of
failure of (1 − 10−4 ) = (1 − Pf )50 ⇒ Pf = 2 ⋅ 10−6 (≈ 10 −4 50 , for small probabilities), which is
much smaller. The partial factors are the same for both periods, however, the
representative loads are larger for the long period. This indeed makes sense as it is more

126
economical to invest in safety measures if one can profit from it for a longer period of time.
Therefore, a shorter design live does not provide an argument for a reduction of β.

3.2 Human safety

If human safety is the governing factor in the design, one generally wants to have a
constant annual failure probability. The probability to die as a result of an accident (traffic,
falling from the stairs) is about 10 −4 per year in the Netherlands. For other countries this is
in the same order of magnitude. It is certainly not accepted in society that the probability to
become the victim of structural failure is larger than the normal probability to die as a
result of an accident. For that reason it is agreed to establish the maximum probability to
become the victim of structural failing as order of magnitude 10−5 per year.

In EN 1990 [12], the reliability level is described in qualitative terms with respect to the
danger of life and the economical damage. Here, for each consequence class, this
qualitative description is translated in to the following conditional probabilities P1 for loss

of human life.

CC1: small consequence for loss of human life P1 = 10 −3

CC2: considerable consequence for loss of human life P1 = 3 ⋅ 10 −2
CC3: high consequence for loss of human life P1 = 3 ⋅ 10 −1

These probabilities are conditional, probabilities given the fact that a structural component
fails. The probabilities relate to individual persons that are in the building on a regular
basis. The corresponding probability of failure Pg of a structural part can now be calculated
for one year:

Pg ⋅ P1 < 10 −5 (1)

CC1: Pg ≤ 10 −2 ⇒ β ≥ 2.3
−4
CC2: Pg ≤ 3 ⋅ 10 ⇒ β ≥ 3.4
−5
CC3: Pg ≤ 3 ⋅ 10 ⇒ β ≥ 4.0

The effect of a longer reference period (t) can be incorporated in approximation by

reducing the β values with C log t :

127
 CCl: Pg ≤ t 10 −2 ⇒ β ≥ Φ −1 { t 10 −2 } ≈ 2.3 − 1.10 log t
CC2: Pg ≤ t 3 ⋅ 10 −4 ⇒ β ≥ Φ −1 {t 3 ⋅ 10−4 } ≈ 3.4 − 0.75 log t
CC3: Pg ≤ t 3 ⋅ 10 −5 ⇒ β ≥ Φ −1 {t 3 ⋅ 10−5 } ≈ 4.0 − 0.60 log t

Note that the value of 3 ⋅ 10 −4 per year for CC2 corresponds well with the recommendation
in [7, Chapter 7.2] for moderate failure consequences and relative large costs of safety
measures.

3.3 Economics and human safety in structural design

In this section the above mentioned arguments of economics and human safety are
translated into a practical method that can be used by structural engineers.

In Figure 1, the annual failure probability as a function of the design working life has been
plotted for new structures in consequence class 2. As stated above, considering economic
optimization only, the β values remain the same for different design life times; this means
that the probability of failure per year is obtained as Pfailure = Φ( −β)/T , with T the design

9,0E-04

8,0E-04
human safety
7,0E-04 CC2
probability of failure [1/year]

CC2 wind
6,0E-04

5,0E-04

4,0E-04

3,0E-04

2,0E-04

1,0E-04

0,0E+00
0 10 20 30 40 50 60
design life [year]
Figure 1: Annual failure probability as a function of de design working life, new structures CC2.

128
 1,2E-02

1,0E-02
probability of failure [1/year]

8,0E-03

human safety
6,0E-03 CC1
CC1 wind

4,0E-03

2,0E-03

0,0E+00
0 10 20 30 40 50 60
design life [year]
Figure 2: Annual failure probability as a function of de design working life, new structures CC1.

1,8E-04

1,6E-04
human safety
1,4E-04 CC3
probability of failure [1/year]

CC3 wind
1,2E-04

1,0E-04

8,0E-05

6,0E-05

4,0E-05

2,0E-05

0,0E+00
0 10 20 30 40 50 60
design life [year]
Figure 3: Annual failure probability as a function of de design working life, new structures CC3.

129
 life. This probability is plotted for β = 3.8 and β = 2.8 (wind dominant). Also the lower
boundary for human safety Pg ≤ 3 ⋅ 10 −4 for CC2 is displayed. For consequence class 1 and 3
the same is shown in Figures 2 and 3 respectively. From these figures it can be concluded
that for CC2 and CC3 in the case of wind dominated structures and design lives smaller
than 15 years the probability of failure exceeds the level for human safety. In that case,
instead of raising partial factors for short periods, the Dutch Code simply demands a
minimum design life of 15 years for CC2 and CC3. In Table 3 the lower β limits for human
safety are collected for both a 1 year and a 15 year design life. In Section 4 these values will
be employed.

Table 3: Lower β limits for human safety

Consequence class 1 year design life 15 year design life
1 2.3 1.1
2 3.4 2.5
3 4.0 3.3

4 Safety targets for existing structures

In this section the required β values are derived for two types of decision.
First we have the level below which the structure is unfit for use. If this safety level is not
reached, the authorities have to send immediately a notification that the structure has to be
closed and to be adapted. Secondly, we have the safety level for repair of existing
structures.

Establishing safety targets for existing structures, both economic arguments and limits for
human safety play a role. For existing structures normally a shorter design life is
employed, however, as shown in Section 3.1, this does not provide arguments for a
reduction of β. On the other hand, increasing safety levels usually involves more costs for
an existing structure than for structures that are still in the design phase; this is the reason
that from an economical point of view the required safety level for existing structures is
lower than for new structures. However the limits for human safety may not be exceeded.
Both the economic arguments and the human are discussed in the Sections 4.1 and 4.2.

130
4.1 Economics
As stated earlier, based on economic arguments, the reliability index for existing structures
may be reduced. This is the result of an economic optimization of the total building costs
and the product of damage costs and the probability of failure.

The actual reliability index for new wind dominated structures is, as already mentioned,
about one unit lower then the standard value of 3.8. Higher design wind loads are not
accepted because of economic reasons. From that argument it seems reasonable to propose
a level βu below which existing structures are unfit for use:

βu = βn − Δβ (2)

with Δβ > 1.0. The value Δβ = 1.5 is chosen based on a crude study of economic
optimization for existing structures [6] and corresponds to commonly accepted target
safety levels for existing structures [14,15]. For Eurocode reliability class 2 a reduction by
1.5 means a shift from β = 3.8 to β = 2.3 and for the wind dominated cases from β = 2.8 to
β = 1.3 (life time basis).

For repair a safety level βr is defined:

β n < βr < βu (3)

leading to:

βr = βn − 0.5 (4)

Here in the case of repair, it has to be avoided that too many structures designed according
to old building codes and at that time were considered to be safe enough, have to be
replaced or radically changed. Therefore the safety level βr for repair is chosen so that

structures that have been designed with old building codes can meet these requirements
without any problem. Δβ = 0.5 is on the average the difference between the safety levels for
the new consequence classes of the Eurocode and the safety levels for the consequence
classes of the old national Code.
This leads to the values in Table 4 indicating the economically optimal values for β in the
cases of repair and unfitness for use. In this table, according to the Dutch NEN-8700 [13],
consequence class 1 from EN 1990 has been subdivided into 1A and 1B, these classes are

131
the same except for the fact that in 1A no danger for human life is present. However, these
economic arguments should not lead to concessions in human safety. This is discussed in
the following section.

Table 4: β values for repair and unfitness for use; based on only economic arguments
Consequence class βn new βr repair βu unfit for use

wn wd wn wd wn wd
1A 3.3 2.3 2.8 1.8 1.8 0.8
1B 3.3 2.3 2.8 1.8 1.8 0.8
2 3.8 2.8 3.3 2.3 2.3 1.3
3 4.3 3.3 3.8 3.3 2.8 2.3
wn = wind not dominant
wd = wind dominant

4.2 Economics and human safety

Here, the same limits as presented in Section 3.2 have to be observed. It was explained in
Section 3.3 that, for human safety reasons, the Dutch code requires a minimum design life
of 15 years; this holds for both new and existing structures. Therefore, the representative
loads used in design will be equal to or larger than the 15 years values. Table 3 shows for
human safety the required β values based on the minimum design life of 15 years and
leading to the maximum annual probability of failure as defined in Section 3.1; these
values are also adopted in this section. If the design life for existing structures is larger
than 15 years, the required β values for human safety become some smaller. However, for
existing structures the design life time is mostly about 15 years and for e.g. 20 years the
differences are negligible small. So, here, the 15 year β values are adopted as the human
safety requirement, which is a somewhat conservative approximation for existing
structures with design lives larger than 15 years.
The same graphs as made in Figures 1-3 can be made in order to observe which β values
have to be used. An example is given in Figure 4. For non wind dominated structures CC3,
economic optimization would lead to β = 4.3 - 1.5 = 2.8.

132
 1,0E+00

human safety

1,0E-01 CC3 existing; economy

probability of failure [1/year]

CC3 wind existing; economy

1,0E-02

1,0E-03

1,0E-04

1,0E-05
0 10 20 30 40 50 60
design life [year]
Figure 4: Annual failure probability as a function of the design working life, existing structures
CC3

However we see here that for CC3 the human safety criterion is decisive. The β value
becomes β = 3.3 (see Table 3, third column), this corresponds to the annual probability of

failure of 3 ⋅ 10 −5 in Figure 4. For CC2 a similar analysis can be performed, also here the
human safety criterion is always decisive. The β value becomes β = 2.5 (see Table 3, third
column). For CC1B also the human safety criterion is decisive with β = 2.3. However in
order to obtain uniformity, here we also take a minimum reference period of 15 years; in
that case the value for the representative variable load increases and the reliability index

decreases. It is found β = Φ −1 {15 ⋅ 10 −2 } = 1.04 which is rounded off to 1.1 (see Table 3, third

column) for human safety. For CC1A, no minimum reference period is chosen, because
human safety is not decisive. The final values for β can be determined by taking each time
the maximum of Tables 3 and 4. This has been done in Table 5.

133
 Table 5: Required β – values for the minimum reference period
Consequence Minimum βn new βr repair βu unfit for use
class reference period
wn wd wn wd wn wd
1A 1 year 3.3 2.3 2.8 1.8 1.8 0.8
1B 15 year 3.3 2.3 2.8 1.8 1.8 1.1*
2 15 year 3.8 2.8 3.3 2.5* 2.5* 2.5*
3 15 year 4.3 3.3 3.8 3.3* 3.3* 3.3*
wn = wind not dominant
wd = wind dominant
(*) = in this case the minimum limit for human safety is decisive

5 Probabilistic calculations for deriving partial factors for existing bridges

In this section probabilistic calculations are used in order to establish the partial factors
that are needed to obtain the required reliability for existing bridges with a relatively small
span under traffic load. In The Netherlands, the bridges in highways have to satisfy the
requirements belonging to consequence class 3, so they must have a reliability index of β =
4.3 for new bridges, β = 3.8 for repair and β = 3.3 for unfit for use. Bridges in less important
roads are in consequence class 2 and have to satisfy β = 3.8 for new bridges, β = 3.3 for
repair and β = 2.5 for unfit for use.

5.1 Set up of the probabilistic calculation for existing bridges

Considered is a period of 15 years with loading by dead weight and traffic load only. Other
loads play a minor role in the design and assessment of traffic bridges. The limit state
function is as follows:

Z = R - mG G - mT T (5)

where R is the resistance of a structural element, mG G is the effect of the dead weight and
mT T is the effect of the traffic load, where mG and mT represent model uncertainties. These
model uncertainties have been estimated as normally distributed with a mean value of 1.0
and a coefficient of variation of 0.07 for self weight and 0.10 for traffic load. The model
uncertainty both covers the schematisation of the load and the determination of the load
effect by means of a structural calculation. The weight G has a mean value that is equal to

134
the characteristic value and a coefficient of variation of 7 %. For the traffic load T the
maximum weight is used of a truck combination that passes the bridge in a period of 15
years. We can make this assumption because in the case of a relatively small span, a single
truck will determine the decisive load on the bridge [8]. The weight of a single truck has
the following statistical distribution (units kN):

FV (T ) = 0.996 × Φ(200, 117) + 0.004 × Φ(600, 224) (6)

This distribution has been derived from weigh in motion measurements in 2004 on the
Dutch highway RW16 near the Moerdijk bridge. In [8, 9, 10] it was shown that this
distribution provides values for the traffic load that are comparable to the ones from EN
1991-2, Traffic loads on bridges. In the expression above, Φ(μ, σ) is the distribution
function of the normal distribution with mean μ and standard deviation σ.

The distribution for the maximum truck weight in a period of 15 years follows from:

F15 year (T ) = FV (T )n (7)

For n is used: n = 15 × 2 × 106 , the number of trucks passing in 15 year.

The ratio between the self weight of the bridge and the traffic load will be varied in order
to study the influence of that ratio on the result. The ratio Trep / Grep is called χ. For the
strength R a lognormal distribution is taken with a coefficient of variation VR = 0.10. This
is a common value primarily for the bending resistance of steel of concrete structures. The
mean value of R results from the design value via:

Rm = Rd exp(αβVR ) (8)

where α is the sensitivity coefficient and β the required reliability index. The design value
of the strength is found with help of the basic design formula:

Rd = Gd + Td (9)

Here, the maximum of formulas 6.10a and 6.10b from EN 1990 [12] has to be taken.
Format 6.10a (weight dominant):

135
 Gd = γ G Grep
(10)
Td = Ψ o γT Trep , with Ψ o = 0.8

Format 6.10 b (traffic dominant):

Gd = ξ γ G Grep
(11)
Td = γT Trep

The representative value of the traffic load T follows from:

1 − FV (Trep ) = 1/ n (12)

With the above mentioned value for n and the distribution for FV a value Trep = 1560 kN
results. This value is comparable to the representative value for a bridge with a span of
about 20 m (the type of spans under discussion here); see [8]. The representative value for
Grep is taken equal to Trep /χ, where χ is varied from 0.25 tot 2.0.

5.2 Results for the partial factors

For each set of partial factors γ G and γT (see Tables 6 and 7) the reliability index belonging
to the probability of failure P(Z<0) is calculated and plotted as a function of χ =
Trep / Grep . Both the reliability indices for format 6.10a and format 6.10b are plotted.
An example is shown in Figure 5 where for the partial factors belonging to CC3, repair (see
Table 6) the reliability level is plotted. In Figure 5, the maximum of 6.10a and 6.10b has to
be taken as stated above. It can be observed that the target value in this case of β = 3.8 is
not entirely obtained. However in the old Dutch Building Code NEN 6702 a value of β =
3.6 was prescribed for new structures, so this value is accepted for the time being.
The same exercise is performed for CC3, the level at which the structure is unfit for use and
CC2, the levels repair and unfit for use. The corresponding partial factors γ G and γT are
determined in such a way that a sufficient safety level results from the probabilistic
calculations. Finally the partial safety factors that are given in Table 6 and 7 are obtained
for use in structural calculations. These values will be incorporated into the Dutch National
Annex of EN-1990-Annex A2, Bridges.

136
 3,9

3,7

3,5

3,3
beta

3,1 6.10.a
6.10.b
2,9

2,7

2,5
0,0 0,5 1,0 1,5 2,0 2,5

chi=T/G
Figure 5: Reliability index for CC3 repair using the proposed rules of verification

Tabel 6: Partial factors for traffic bridges in highways, CC3

Reference Obtained β Partial factor
period [year]

Weight γ G Weight ξ γ G Traffic γT

6.10a 6.10b
new 100 4.3 1.40 1.25 1.50
repair 15 3.6-3.8 1.30 1.15 1.30
disapproval 15 3.3 1.25 1.10 1.25

Tabel 7: Partial factors for traffic bridges CC2

Reference Obtained β Partial factor
period [year]

Weight γ G Weight ξ γ G Traffic γT

6.10a 6.10b
new 100 3.8 1.30 1.20 1.35
repair 15 3.1 – 3.2 1.25 1.10 1.20
disapproval 15 2.7 – 2.8 1.10 1.10 1.10

137
6 Conclusions

In this article a theoretical background and corresponding results for the safety assessment
of existing structures have been presented. The first choice is to compare the structure with
the code requirements for newly built structures, either the present ones or the ones at the
time of erection. However, there may be many occasions where this level is not reasonable
or economically justifiable. In that case authorities may be willing to relax the
requirements. Such reduced requirements should be based upon the latest codes, but with
reductions in partial factors, design working life and representative load values. The
reductions however should not affect human safety. In this article the reliability levels for
repair and disapproval of existing structures have been established. For economic reasons
a reduction Δβ = 0.5 in the reliability index has been proposed for the repair level and a
reduction Δβ = 1.5 for closing down a structure. In the latter case, however, the limit value
introduced for human safety becomes decisive in most of the cases; here a minimum
design life of 15 years has to be observed. Subsequently, for existing bridges under traffic
load adapted partial factors for weight and traffic load have been established using full
probabilistic calculations. These can be used for reassessment of existing bridges under
traffic load.

138
References

[1] Diamantidis, D. at al., Probabilistic Assessment of Existing Structures, RILEM Report

032, 2001.
[2] Vrouwenvelder, A.C.W.M., Evaluation of Existing Structures, item Codification, IABSE
Congress, Copenhagen, June, 1996.
[3] Ostlund, L. et al, Recommendation for loading- and safety regulations for structural
design, Nordic Committee on Building Regulations, NKB-Report No. 36, 1978
[4] Ellingwood, B., Reliability Basis of Load and Resistance Factors for Reinforced
Concrete Design, National Bureau of Standards, Building Science Series 110, 1978
[5] Vrouwenvelder, A. and A. Siemes, Probabilistic calibration procedure for the
derivation of partial safety factors for the Netherlands building codes, Heron, Vol. 32,
no 4, Delft, the Netherlands, 1987.
[6] Vrouwenvelder, A.C.W.M. and Scholten N.P.M. , Veiligheidsbeoordeling bestaande
bouw, Achtergrondrapport bij NEN 8700, TNO report 2008-D-R0015,2008.
[7] JCSS, Background Documentation Eurocode 1 ENV, Part 1: Basis of Design, ECCS,
Report No 94, 1996.
[8] Vrouwenvelder, A.C.W.M., Waarts, P.H., De Wit, S., Algemene
veiligheidsbeschouwing en modellering van wegverkeersbelasting voor
brugconstructies, TNO report 98-CON-R1813, 1998
[9] Allaix, D.G., Bridge Reliability Analysis with an Up-to-Date Traffic Load Model,
Politecnico di Torino, 2007.
[10] Steenbergen, R.D.J.M., Morales Napoles, O., Determination of design axle loads for
bridges from measurements. In Proceedings of ESREL 2010 – European Safety and
Reliability Conference 2010.
[11] EN 1991 Eurocode 1: Actions on structures; European Committee for Standardization.
[12] EN 1990 Eurocode 0: Basis of structural design; European Committee for
Standardization.
[13] NEN 8700:2009 Grondslagen van de beoordeling van de constructieve veiligheid van
een bestaand bouwwerk, Nederlands Normalisatie instituut.
[14] Diamantidis D., Bazzuro P. Safety acceptance criteria for existing structures. Workshop
on Risk Acceptance and Risk Communication, Stanford University, USA, March 2007.
[15] Vrouwenvelder A.C.W.M., Scholten N.P.M., Assessment Criteria for Existing
Structures, Structural Engineering International 1/2010.

139
140