Ashgar I Zadeh
Ashgar I Zadeh
Ashgar I Zadeh
www.elsevier.com/locate/compstruc
Abstract
Due to intensive investments in the construction of civil infrastructures in the 1960s and 1970s, the number of
deteriorating structures has increased considerably in the last decade. In many cases these structures require unavailable
nancial resources for inspection, maintenance, repair, rehabilitation, and replacement. As a consequence, there is an
urgent need for developing cost-eective maintenance strategies for deteriorating structures. The existing structures
management models evaluate the need for maintenance using measures of the strength deterioration of the structure.
Strength deterioration is not a consistent measure of safety and for this reason, in this paper, the reliability index is used
as a measure of structural performance. The time-dependent reliability index and the eect of maintenance actions are
described using a model based on that proposed by the second author. In spite of the importance of the cost of a
maintenance action and of its eects on the reliability index, there is very limited information on the relation between
the cost and the eect of maintenance actions. In this paper, a model considering the interaction between maintenance
cost and its eect on the reliability index is proposed. This model is used to compare the cost-eectiveness of several
maintenance strategies for a deteriorating structure. The eect of the parameters associated with the cost model on the
optimal maintenance scenario is also analyzed.
2004 Elsevier Ltd. All rights reserved.
Keywords: Deteriorating structures; Cost; Maintenance; Reliability deterioration; Reliability improvement; Structural reliability;
Optimum maintenance scenario
0045-7949/$ - see front matter 2004 Elsevier Ltd. All rights reserved.
doi:10.1016/j.compstruc.2004.03.012
1078 L.C. Neves et al. / Computers and Structures 82 (2004) 10771089
both the state of the structure and the eect of mainte- deterioration process of the reliability index during a
nance action on the structural performance. However, period of time; and (c) reduction of the deterioration
the cost of maintenance depends not only on the type of rate of the reliability index during a period of time.
maintenance action, but also on the state of the structure In this paper, as previously indicated, time-dependent
before and after its application. As an example, the cost reliability proles are described based on the mainte-
of repairing a corroded steel girder depends on the de- nance model proposed by Frangopol [1]. The eect of
gree of corrosion and the extension of the repair. In this each maintenance action is modeled through several
paper, a model considering the interaction between cost variables (see Fig. 1), such as: (a) increase in reliability
and its eect on the reliability index is proposed. This index immediately after application (i.e., the dierence
model is able to incorporate a wide range of mainte- between the reliability indices after and before applica-
nance actions. The time dependent reliability prole tion of maintenance), c; (b) time period during which the
under no maintenance and the eect of maintenance deterioration process of reliability is delayed, td ; (c) time
actions on the reliability index are simulated using the period during which the deterioration rate of reliability
model proposed by Frangopol [1]. The proposed cost index is delayed or reduced, tpd ; and (d) reliability index
model is used to identify the most cost-eective main- deterioration rate reduction, d. The time of application
tenance scenario for a deteriorating structure over a of the maintenance actions is dened by two variables:
prescribed time horizon. time of rst application, tpi , and the interval between
subsequent applications, tp .
t pd
ti
td td
0
1
RELIABILITY INDEX,
-
1
-
tpi tp tp
1
profile under profile under
no maintenance maintenance
TIME, YEARS
Fig. 1. Reliability index proles under no maintenance and under maintenance and their associated variables.
L.C. Neves et al. / Computers and Structures 82 (2004) 10771089 1079
200
C = C1 + C2 q
175 0
C2 =100- C1
150
50
TOTAL COST, C
125 100 75
100
75 100
50
25 q = 0.5
50
C1 = 0
25
0
0 1 2 3
(a) RELIABILITY INDEX IMPROVEMENT,
200
C = C1+ C2 q 50
175 0
C2 = 100 - C1
150
TOTAL COST, C
125
100 75
100
100
75
50
q = 1.5
50
25
25
C1 = 0
0
0 1 2 3
(b) RELIABILITY INDEX IMPROVEMENT,
200
0
C = C1+ C2 q
175 50
C2 = 100 - C1
150
TOTAL COST, C
100
75
q = 2.5
50
50 25
25 C1 = 0
0
0 1 2 3
(c) RELIABILITY INDEX IMPROVEMENT,
Fig. 2. Total cost of a maintenance action vs reliability index improvement for a discount rate m 0% and (a) q 0:5; (b) q 1:5; and
(c) q 2:5.
1080 L.C. Neves et al. / Computers and Structures 82 (2004) 10771089
maintenance cost and delay and reduction of reliability where C0 is the present value of the cost, Ct is the cost at
deterioration, respectively, and time t and m is the discount rate. The discount rate is very
dicult to predict, since it depends on the economical
c a td dtpd td 5 conditions during the life-time of the structure. In the
United Kingdom, for bridge investments, m 6% [12].
where c is the measure of the eect on the reliability
index prole of the delay in reliability deterioration rate
and the reduction of the reliability deterioration rate. 5. Applications
Assuming maintenance actions with negligible eect
on the reliability deterioration delay and on the reduc- In order to illustrate the importance of the relation
tion of deterioration of reliability (i.e., tpd 0), Eq. (4) between maintenance eect on reliability and mainte-
becomes nance cost, several examples are herein presented.
7 c = 8.33 years
d = 6.25 years
4. Discount rate
RELIABILITY INDEX,
5 SCENARIO C
In addition to the eect of the increase in the reli-
4
ability index, c, the delay in reliability deterioration, td ,
and the reduction of the reliability deterioration rate, the 3
eect of time of application of each maintenance inter- d d d d
2 target
vention must also be considered when the lifetime cost c c c
1
of an existing structure is analyzed. In fact, the same
amount of money spent at two dierent instants in time 0
0 10 20 30 40 50
has dierent present values. As a result, costs can only
TIME, YEARS
be compared if converted to the same instant as follows
Fig. 3. Reliability proles for maintenance scenarios C and D;
Ct lifetime improvement in reliability index due to any mainte-
C0 7
1 mt nance scenario is c50 4.
Table 1
Descriptors of maintenance scenarios considering improvement in reliability index
Maintenance Number of Time of rst Interval between Time of last Reliability index
scenario maintenances n maintenance successive maintenance increase due to
application (years) maintenances (years) application (years) each maintenance c
A 1 25.00 25.00 4.00
B 2 25.00 12.50 37.50 2.00
C 3 25.00 8.33 41.67 1.33
D 4 25.00 6.25 43.75 1.00
E 5 25.00 5.00 45.00 0.80
F 6 25.00 4.17 45.83 0.67
L.C. Neves et al. / Computers and Structures 82 (2004) 10771089 1081
the increase in reliability index due to each nominal The increase of the reliability index, number of
identical maintenance action is set as follows maintenance interventions, and the time of application of
the six maintenance scenarios A, B, C, D, E, and F,
c50 4
c 8 considered are presented in Table 1. In Fig. 3, the reli-
n n
ability proles associated with scenarios C and D dened
where n is the number of applications of maintenance in Table 1 are presented. In order to compare the costs
actions.
C t = C1 + C2 q Ct = C1+ C2 q
800 400
C2 =100 - C1 C2 =100 - C1 q = 0.5
q = 0.5 50= 4
=4 = 2%
50 = 0% Optimal Scenario
TOTAL COST, Ct
Optimal Scenario 300
TOTAL COST, Ct
600
F
F E
E D
D 200 C
400
C B
B A
A
100 A
A A
200 A A
A
A
A
0
0 0 25 50 75 100
0 25 50 75 100
(a) COST C1
(a) COST C1
600
1000 Ct = C 1+ C 2 q
Ct = C1+ C2 q C2 = 100 - C1 q = 1.5
A
C2 = 100 - C1 500 A
50= 4 = 2%
800 =4
50
q = 1.5 Optimal Scenario
TOTAL COST, Ct
600 B
B
300
C
C
D E C
400 E D F C
F 200 E
A F B
200 100
A A
0 0
0 25 50 75 100 0 25 50 75 100
2500
3000 A Ct = C1 + C 2 q
C2 = 100 - C1 A Ct = C1+ C2 q
50= 4 q = 2.5 2000 C2 = 100 - C1
2500
= 0% =4
Optimal Scenario 50
TOTAL COST, Ct
TOTAL COST, Ct
1000
B
C
500
500 D E C C
E F
F D E C
A EF F
A
0 0
0 25 50 75 100 0 25 50 75 100
00
Fig. 4. Total cost of maintenance scenarios considering Fig. 5. Total cost of maintenance scenarios considering
improvement in reliability index for a discount rate of 0% (a) improvement in reliability index for a discount rate of 2% (a)
q 0:5; (b) q 1:5; and (c) q 2:5. q 0:5; (b) q 1:5; and (c) q 2:5.
1082 L.C. Neves et al. / Computers and Structures 82 (2004) 10771089
associated with dierent maintenance scenarios a cali- ferent values of C1 and q are also indicated considering
bration point (c 1; C 100) is dened. Therefore, that costs are not discounted (m 0%).
C2 100 C1 . The eect of the xed cost C1 on the total From Fig. 4(a) it is clear that for a value of the
cost of each of the six maintenance scenarios considered parameter q lower than 1 the optimal maintenance sce-
is shown in Fig. 4. In this gure the optimal (i.e., mini- nario is A for all values of C1 . In Fig. 4(b) the costs of
mum cost) maintenance scenarios corresponding to dif- the six maintenance scenarios considering q 1:5 are
140 Ct = C1+ C2
q Ct = C1+ C2 q
C2= 100 - C1 600 C2 = 100 - C1
120
50 = 4 50= 4
q = 0.5 = 0% q = 0.5
Optimal Scenario = 6% q = 1.0
TOTAL COST, Ct
100
q = 1.5
Total Cost, C
Any E DE q = 2.0
400 C C
80 F B q = 2.5
F F A F
E
D F A
60 C
B A
A F
A A
40 A 200 A A
A A
A A
AA AA A
20
0 0
0 25 50 75 100 0 25 50 75 100
Optimal Scenario = 6%
TOTAL COST, Ct
q = 2.0
150 F F CDE C q = 2.5
200 B C
F
F F B
A
F
B A
100 F A A
C 100 A
A
D F D AA A A A
E B
50 F
A
0
0 25 50 75 100
0
0 25 50 75 100 (b) Cost C1
(b) COST C1 100
Ct = C1+ C2 q
C2 = 100 - C1
80 50= 4
800
A Ct = C1+ C2 q q = 0.5
= 6%
C2= 100 - C1 q = 1.0
q = 1.5
Total Cost, C
50 = 4 60 F
C
F
DE
E C
D
q = 2.0
TOTAL COST, Ct
600 B B q = 2.5
Optimal Scenario A F
q = 2.5 F
A
F
A
= 6% 40 F
A
400 F A
AA A A A
20
B
200
C
DE F E C 0
F A
0 25 50 75 100
0
0 25 50 75 100 (c) Cost C1
(c) COST C1
Fig. 7. Total cost of optimal maintenance scenarios considering
Fig. 6. Total cost of maintenance scenarios considering improvement in reliability only for dierent values of q (i.e., 0.5;
improvement in reliability index for a discount rate of 6% (a) 1.0; 1.5; 2.0; 2.5) and C1 (i.e., 0; 25; 50; 75; 100) and dierent
q 0:5; (b) q 1:5; and (c) q 2:5. discount rates: (a) m 0%; (b) m 2%; and (c) m 6%.
L.C. Neves et al. / Computers and Structures 82 (2004) 10771089 1083
presented. In this case, the optimal maintenance sce- If costs are discounted there is a signicant decrease
nario strongly depends on the xed cost. If the xed in the lifetime cost, as shown in Figs. 5 and 6 for dis-
cost, C1 , is not considered, the cost of maintenance is count rates m 2% and m 6%, respectively.
only dependent on the eect of the maintenance action In Figs. 7(a), (b), and (c) and Tables 24 the optimal
and, as a result, a scenario based on maintenance actions maintenance scenario for each combination of q and C1
without signicant eect on reliability improvement but is presented for a discount rate of 0%, 2%, and 6%,
applied more frequently, is the most cost-eective (i.e., respectively. From these gures and tables it is clear
scenario F). On the other hand, if the xed cost, C1 , is that, as q increases the optimal maintenance scenario
equal to 100, the cost is independent on the eect of the changes from the application of maintenance actions
maintenance scenario and, as a result, the most cost- with signicant eect on reliability to actions with
eective maintenance scenario is that associated with smaller eects applied more frequently. Conversely, as
fewer applications (scenario A). If the parameter q is C1 increases scenarios corresponding to a smaller num-
taken as 2.5 (see Fig. 4(c)) the optimal maintenance ber of applications become more cost eective. This
scenario is that associated with a smaller eect on reli- tendency is unchanged by the discount rate. The optimal
ability (i.e., scenario F) for the lowest value of the xed maintenance scenario for each combination of the xed
cost, and with fewer applications (i.e., scenario A) for cost C1 and coecient q, for all values of the discount
the highest value of C1 . rate considered is presented in Table 5.
Table 2
Total cost of optimal maintenance scenarios considering 0:5 6 q 6 2:5; 0 6 C1 6 100; and m 0%
Cost C1 q
0.5 1.0 1.5 2.0 2.5
0 200.0 (A) 400.0 (X) 326.6 (F) 266.7 (F) 217.7 (F)
25 175.0 (A) 325.0 (A) 393.3 (E) 350.0 (F) 313.3 (F)
50 150.0 (A) 250.0 (A) 380.9 (C) 400.0 (D) 393.1 (E)
75 125.0 (A) 175.0 (A) 275.0 (A) 350.0 (B) 379.0 (C)
100 100.0 (A) 100.0 (A) 100.0 (A) 100.0 (A) 100.0 (A)
The maintenance scenario associated with the minimum total cost is indicated in parentheses. X denotes any maintenance scenario (A,
B, C, D, E or F).
Table 3
Total cost of optimal maintenance scenarios considering 0:5 6 q 6 2:5; 0 6 C1 6 100; and m 2%
Cost C1 q
0.5 1.0 1.5 2.0 2.5
0 121.9 (A) 200.3 (F) 163.6 (F) 133.6 (F) 109.1 (F)
25 106.7 (A) 189.9 (B) 197.8 (F) 175.3 (F) 156.9 (F)
50 91.4 (A) 152.4 (A) 198.7 (C) 204.4 (D) 198.5 (E)
75 76.2 (A) 106.7 (A) 158.2 (B) 186.9 (C) 197.6 (C)
100 61.0 (A) 61.0 (A) 61.0 (A) 61.0 (A) 61.0 (A)
The maintenance scenario associated with the minimum total cost is indicated in parentheses.
Table 4
Total cost of optimal maintenance scenarios considering 0:5 6 q 6 2:5; 0 6 C1 6 100; and m 6%
Cost C1 q
0.5 1.0 1.5 2.0 2.5
0 46.6 (A) 55.3 (F) 45.1 (F) 36.8 (F) 30.1 (F)
25 40.8 (A) 58.1 (C) 54.6 (F) 48.4 (F) 43.3 (F)
50 34.9 (A) 51.8 (B) 58.5 (D) 58.0 (E) 55.6 (E)
75 29.1 (A) 40.8 (A) 50.3 (B) 55.5 (C) 58.5 (D)
100 23.3 (A) 23.3 (A) 23.3 (A) 23.3 (A) 23.3 (A)
The maintenance scenario associated with the minimum total cost is indicated in parentheses.
1084 L.C. Neves et al. / Computers and Structures 82 (2004) 10771089
Table 5
Optimal maintenance scenarios considering 0:5 6 q 6 2:5; 0 6 C1 6 100; and 0% 6 m 6 6%
Cost C1 q
0.5 1.0 1.5 2.0 2.5
0 AAAA XFFF FFFF FFFF FFFF
25 AAAA ABCC EFFF FFFF FFFF
50 AAAA AABB CCCD DDDE EEEE
75 AAAA AAAA ABBB BCCC CCCD
100 AAAA AAAA AAAA AAAA AAAA
First, second, third, and fourth letter denotes the optimal maintenance scenario associated with a discount rate of 0%, 2%, 4%, and 6%,
respectively.
X denotes any maintenance scenario (A, B, C, D, E, or F).
RELIABILITY INDEX,
The second example analyzes maintenance actions SCENARIO C1
6
producing both an increase in the reliability index and a
delay in reliability index deterioration rate. The reli- td = 3 years
ability index prole under no maintenance considered is
4
the same as that described in the previous example. The
denition of each of the ve scenarios A1, B1, C1, D1,
and E1, considered is presented in Table 6. The sce- = 0.96
2 target 9 years 9 years
narios are dened so that, within the time horizon of 50
years, the reliability index would never down-cross the
target value btarget 3:0. For all the ve scenarios a 0
0 10 20 30 40 50
delay in reliability index deterioration of three years is
TIME, YEARS
assumed. In Fig. 8 the time-dependent reliability prole
under maintenance scenario C1 is shown. For all sce- Fig. 8. Reliability prole for maintenance scenario C1; increase
narios in Table 6, shown in Figs. 9(a) and (b), the in reliability index c 0:96 and delay in deterioration td 3
parameters in Eq. (4) are as follows: C1 C2 C3 50; years.
q1 q2 1:5. The last two columns of Table 6 indicate
the present value of the total cost of each maintenance
scenario, considering discount rates of 0% and 6%. The results show that the optimal maintenance scenario, as
discount rate plays a signicant role in choosing the well as the cost of each maintenance scenario, is strongly
optimal scenario. In fact, for a discount rate of 0% dependent on the xed cost. For smaller values of C1 the
the optimal scenario is B1, and for a discount rate of 6% optimal maintenance scenario is that corresponding to a
the optimal scenario is C1. more frequent application of maintenance actions with
In Fig. 10 the costs associated with the ve mainte- smaller impact on the reliability of the structure. As the
nance scenarios considered are presented for dierent xed cost increases, the optimal maintenance scenario
values of the xed cost C1 and for q1 q2 1:5. The tends to be associated with maintenance actions with a
Table 6
Descriptors of maintenance scenarios and associated cumulative total costs considering improvement in reliability and delay in
deterioration
Maintenance Time of rst Interval be- Time of last Reliability Delay in reli- Present value of total cost
scenario maintenance tween succes- maintenance index increase ability deteri- m 0% (7) m 6%
application sive mainte- application due to each oration td
(years) nances (years) (years) maintenance c (years)
A1 25.00 25 3.52 3 396.8 92.5
B1 25.00 13 38 1.60 3 335.6 57.4
C1 25.00 9 43 0.96 3 341.0 51.4
D1 25.00 7 46 0.64 3 368.9 51.6
E1 25.00 5 45 0.32 3 378.4 53.5
Note: Bold characters indicate total costs of optimal maintenance scenarios.
L.C. Neves et al. / Computers and Structures 82 (2004) 10771089 1085
10
RELIABILITY INDEX,
A1, B1, D1
6
D1
0
0 10 20 30 40 50
(a) TIME, Years
10
6 A1, C1, E1
2 C1
E1
0
0 10 20 30 40 50
(b) TIME, Years
Fig. 9. Reliability proles for maintenance scenarios (a) A1, B1, and D1; and (b) A1, C1 and E1.
500 = 0%
B1 In the third example, a deteriorating structure is
C1
400
C1 analyzed considering maintenance actions whose eects
D1 B1
E1 are similar to those of maintenance actions on existing
300 E1 A1 bridges. The minimum admissible value for the reli-
ability index is btarget 4:0. The prole under no main-
200
A1 tenance is dened by an initial reliability index b0 7:0,
100 a time of initiation of deterioration of reliability ti 3
years, and a deterioration rate of reliability index under
0
0 20 40 60 80 100 no maintenance a 0:16/year.
COST C1 The rst maintenance scenario considered, A2, is
associated with the repair and corrosion protection
Fig. 10. Total cost of maintenance scenarios considering of most of the deteriorated components of the struc-
improvement in reliability and delay in reliability deterioration ture (Fig. 11(a)). This leads to a signicant improvement
for a discount rate m 0% and q1 q2 1:5. of the reliability of the structure (c 1:4) and to a
1086 L.C. Neves et al. / Computers and Structures 82 (2004) 10771089
8
Scenario A2
7 =0.16/year
RELIABILITY INDEX,
1 =0.025/year
6 -
5
1 =1.40
=1.40
4
target
3
No Maintenance
2
0 10 20 30 40 50
(a) TIME (years)
8
Scenario B2 = 0
= 0.16/year
td = 3 years
7
RELIABILITY INDEX,
5 1
target
3 No Maintenance
2
0 10 20 30 40 50
(b) TIME (years)
Scenario C2
7
= 0.16/year
RELIABILITY INDEX,
1
6 - = 0.025/year
5
= 0.90
1 = 0.90
4
target
3
2
0 10 20 30 40 50
(c) TIME (years)
Fig. 11. Reliability proles associated with (a) scenario A2; (b) scenario B2; (c) scenario C2; (d) scenario D2 and (e) scenario E2.
reduction of the deterioration rate of the structure scenario, B2, models the replacement of the structure
(d 0:025/year) after application of maintenance. It is (Fig. 11(b)). This maintenance action sets the reliability
assumed that the reliability index cannot increase above index, b, to its initial value, b0 , and leads to a delay in
the initial reliability index. The second maintenance deterioration of reliability index of three years corre-
L.C. Neves et al. / Computers and Structures 82 (2004) 10771089 1087
RELIABILITY INDEX,
6 Scenario D2
5 1 = 0.16/year
td = 6 years
4
target
3 No Maintenance
2
0 10 20 30 40 50
(d) TIME (years)
8
= 0.16/year
td = 3 years
7
Scenario E2
RELIABILITY INDEX,
5 1
target
3 No Maintenance
2
0 10 20 30 40 50
(e) TIME (years)
Fig. 11 (continued)
Table 7
Description of maintenance scenarios, their eects on reliability, and the associated cost considering m 0%, q1 q2 2:5, C1 0 and
C2 C3 100.
Eect on reliability index
Maintenance Time of Time of sub- Improve- Delay in Deteriora- Duration of Cost of application of
scenario application sequent ment c deterioration tion rate maintenance each maintenance ac-
of rst applications td (years) reduction d eect tpd tion considering
maintenance tp (years) (years1 ) (years) m 0%,
tpi (years) q1 q2 2:5, C1 0,
C2 C3 100
A2 10 15 1.4 0 0.025 10 135.9, 235.0, 235.0a
b
B2 20 20 3 0.000 3 1236.1
C2 10 10 0.9 0 0.025 10 80.0
D2 2 6 0.0 6 0.000 6 90.3
E2 9 9 0.0 6 0.000 6 90.3
a
135.9, 235.0, and 235.0 correspond to the cost of application of rst, second, and third maintenance actions.
b
Reliability index is set to its initial value.
sponding to the initiation of damage in a new structure. on the reliability of the structure, but also the one with
This is the maintenance action with the highest impact a higher cost per application. The third maintenance
1088 L.C. Neves et al. / Computers and Structures 82 (2004) 10771089
scenario, C2, models the replacement of only a few rate of reliability index (d 0:025/year). The fourth and
deteriorated elements (Fig. 11(c)). It leads to a lower fth maintenance scenarios, D2 and E2, are associated
increase in the improvement in reliability (c 0:9) than with the application of coating or painting protection.
that due to the maintenance action associated with sce- The maintenance actions associated with these two sce-
nario A2, but to the same reduction of the deterioration narios produce no improvement in reliability, but only a
delay in the deterioration rate of reliability index. In
1200 maintenance scenario D2 (Fig. 11(d)), actions are ap-
1100 C2 = 100 - C1 plied as soon as deterioration in reliability has started or
q1 = 0.5
1000 C3 = 100 - C1 the eect of the previous action has ended. This leads to
Optimal Scenario q2 = 0.5
900 C2 = 0%
an extremely safe structure during its entire lifetime but
TOTAL COST, Ct
6. Conclusions
0
0 20 40 60 80 100
deterioration, and the reduction of the deterioration rate Strategies for Dierent Bridge Types, Highways agency,
of the reliability index. The main advantage of the London, 1998, November.
proposed cost model is the exibility and expandability [2] Augusti G, Ciampoli M, Frangopol DM. Optimal plan-
obtained from using the same expression for several ning of retrotting interventions on bridges in a highway
network. Eng Struct 1998;20:9339.
maintenance actions. The results presented show that
[3] de Brito J, Branco FA, Thoft-Christensen P, Sorensen JD.
the relation between reliability and cost can signicantly An expert system for concrete bridge management. Eng
aect the optimal maintenance scenario. The proposed Struct 1997;19:51926.
cost model can be used in structure maintenance sys- [4] Estes AC, Frangopol DM. Repair optimization of highway
tems, as it provides a powerful tool for taking into ac- bridges using system reliability approach. J Struct Eng
count the interaction between the eects of maintenance 1999;125:76675.
actions on structural reliability and the costs of main- [5] Frangopol DM, Kong JS, Gharaibeh ES. Reliability-based
tenance actions. Further research is needed to ade- life-cycle management of highway bridges. J Comput Civil
quately quantify the parameters used in Eq. (4) for Eng 2001;15:2747.
dierent structures. Steps in this direction are already in [6] Frangopol DM, Maute K. Life-cycle reliability-based
optimization of civil and aerospace structures. Comput &
progress [13].
Struct 2003;81:397410.
[7] Klatter HE, van Noortwijk JM, van Eck NV. Bridge
management in the Netherlands; Prioritisation based on
Acknowledgements network performance. In: Casas JR, Frangopol DM,
Nowak AS, editors. First International Conference on
The support provided to the second author by the US Bridge Maintenance, Safety and Management (IABMAS).
National Science Foundation through grants CMS- International Center for Numerical Methods in Engineering
(CIMNE). Barcelona, Spain, 14-17 July 2002, CD-ROM.
9912525 and CMS-0217290 is gratefully acknowledged.
[8] Frangopol DM, Gharaibeh ES, Kong JS, Miyake M.
The support provided to the rst author by the Portu- Optimal network-level bridge maintenance planning based
guese Science and Technology Foundation (FCT) is also on minimum expected cost. J Transport Res Board;
gratefully acknowledged. The FCT support provided the Transport Res Rec 2000;1696(2):2633.
necessary means to the rst author to spend three years [9] Onoufriou T, Frangopol DM. Reliability-based inspection
at the University of Colorado, at Boulder, under the optimization of complex structures: a brief retrospective.
supervision of the second author. The opinions and Comput & Struct 2002;80:113344.
conclusions presented in this paper are those of the [10] Stewart MG. Reliability-based assessment of ageing
writers and do not necessarily reect the views of the bridges using risk ranking and life cycle cost decision
sponsoring organizations. analyses. Reliab Eng & Syst Saf 2001;74:26373.
[11] Thoft-Christensen P. Assessment of the reliability proles
for concrete bridges. Eng Struct 1998;20:10049.
[12] Tilly GP. Principles of whole life costing. In: Das PC,
References editor. Safety of bridges. London: Thomas Telford; 1997.
p. 13844.
[1] Frangopol DM. A probabilistic model based on eight [13] Kong JS, Frangopol DM. Cost-reliability interaction in
random variables for preventive maintenance of bridges. life-cycle cost optimization of deteriorating structures.
Presented at the Progress Meeting Optimum Maintenance J Struct Eng; ASCE (in press).