Professional Documents
Culture Documents
Tubular Structures X
Jaurrieta, M.A., Alonso, A., Chica, J.A. (Eds.)
Ulrike Kuhlmann
Hans-Peter Günther
Reiner Saul
Marc-Ulrich Häderle
ABSTRACT: The use of circular hollow section members in bridge design is a relatively new concept. The
application of such constructions is strongly influenced by the design and manufacturing of the joints. In gen-
eral, there are two possibilities: either to use cast steel joints or welded joints.
This contribution tries to give an overview about the advantages and disadvantages of both possibilities con-
cerning the aspects of resistance, fatigue, manufacturing and economy in order to help practical engineers in
their decisions and to allow for a further application of circular hollow sections in bridge design.
As a crucial question, special considerations are given to the fatigue assessment of welded circular hollow
section joints. For an example of a recently completed bridge with a typical spatial CHS truss and cast steel
joints it has been shown that also welded connections would have been a possible alternative. This conclusion
is drawn from numerical studies based on FE calculations applying the hot-spot stress approach for the fatigue
assessment of the welded joints.
3.1.3 Fatigue
The smooth shape of the cast joint according to
the flow of internal forces results in only small stress
concentrations, thus making cast steel joints in par-
ticular advantageous for structures subjected to re-
peated loading. The critical part in terms of fatigue
Figure 2. Typical multiplanar KK-joint with notations are the welds between the cast steel joint and the tu-
bular steel members. Concerning the fatigue behav-
iour of this type of connection, there are only few in-
3 JOINT CONSTRUCTION vestigations documented, some associated with the
“Humbolthafen” bridge in Berlin, see Seifried et. al.
3.1 Cast Steel Joints (1999). Based on experimental investigations on
3.1.1 Material Properties small and large scale test specimens, the fatigue re-
sistance could be classified to detail category 71, ac-
Through the ongoing development within the cording to Eurocode 3 Part 1.9 (2002), similar to a
manufacturing of cast steel products it is possible butt welded end-to-end connection of CHS mem-
nowadays to gain almost the same mechanical and bers.
chemical qualities in terms of strength, toughness,
weldability and corrosion resistance as for ordinary 3.1.4 Manufacturing and Quality Assurance
rolled steel products, see e.g. Schober (2001) or
Mang & Herion (2001). The manufacturing of cast joints is quite expen-
sive due to the high costs preparing the cast form-
Figure 3. Definition and extrapolation region of the hot-spot Figure 4. Spatial CHS framework of the bridge Korntal-
stress σhs Münchingen
5.2 Finite Element Analysis
5.2.1 General
For the given CHS joint geometry it is not possi-
ble to verify the fatigue resistance according to the
hot-spot approach as recommended in CIDECT Se-
rial No. 8 (2000). The reasons are as follows:
- the parameter range of the given SCF formulas and
charts is limited to 12 ≤ γ ≤ 24 and 0,3 ≤ β ≤ cos(θ).
These conditions are not fulfilled for almost every
CHS bridge according to Table 1,
- the SCF formulas are restricted to CHS KK-joints
with gap but without any eccentricity and
- the Shs-N curves are limited to a wall thickness
smaller than t = 50 mm.
To overcome these restrictions a detailed FE analysis
has been performed, determining the critical hot-spot Figure 6. Applied loading and boundary conditions
stresses at the joint intersections, Stuba (2002). The
hot-spot stresses respectively SCF values were de- 5.2.3 Stress Concentration Factors (SCF)
termined in accordance to CIDECT Serial No. 8 The stress concentrations factors (SCF) were de-
(2000) and Niemi (1992) using the FE package AN- termined using the quadratic extrapolations method.
SYS. Figure 5 shows the three-dimensional FE Studies by Romeijn (1994) have shown that for CHS
model of the KK-joint. Geometrical symmetries joints this method gives satisfying results.
have not been taken into account. 20-node solid In circumferential direction the tubular sections
elements with an 3x3x3 integration scheme were have been divided into 36 elements/nodes resulting
used. Various mesh densities were investigated and in altogether 36 x 4 x 2 = 288 SCF values for one
compared in order to ensure enough convergence of joint and one loading condition.
the stresses in the vicinity of the weld toe. Since no experimental results have been avail-
able, the numerical FE results of test calculations
were compared to the values given in CIDECT Se-
rial No. 8 (2000), leading to an acceptable difference
less than 5%.
The location of maximum stress concentration
(hot-spot) always appeared in the brace near the sad-
dle. Figure 8 shows the variation of the SCF value
along the circumferential direction of the brace in
case of axial balanced brace loading. Figure 7 re-
flects the corresponding von Mises stress pattern at
the surface. The maximum SCF value appears in the
brace at node number 22.
3.0 3.0
1.5 1.5
0.0 0.0
1 6 11 16 21 26 31 36 1 6 11 16 21 26 31 36
node number node number
Figure 8. Stress concentration factor (SCF) in circumferential foot length tw = 0 foot length tw > 0
direction of the brace, brace loading
Figure 10. Influence of the weld shape on the stress concentra-
5.2.4 Influence of Weld Shape tion factor (SCF)
For critical connections in terms of fatigue and
where the wall thickness is large, full penetration
6 FATIGUE VERIFICATION OF THE KK-JOINT
welds should be used. However, a realistic modeling
of the weld shape in circumferential direction is a 6.1 General
very complex task, because of the ever-changing an-
In order to clarify the feasibility of welded joints
gle Ψ between the surface of the chord and the di-
for the bridge “Korntal-Münchingen”, the fatigue
agonal braces that have to be joined, see figure 9.
verification was performed according to Eurocode 3
Especially in the “heel” zone of the connection with
Part 2 –Steel bridges– (1997) and the hot-spot fa-
a small intersection angle full penetration is difficult
tigue design method recommended for hollow sec-
and leads to a much higher throat thickness com-
tion joints in CIDECT Serial No. 8 (2000).
pared to the “crown” zone. Two different weld
Three different joints of the aforementioned tri-
shapes have been considered:
angular truss girder were selected. These joints differ
- a full penetration weld with a weld foot length
regarding their wall thickness t0 and t1 and loading
tw = 0 (butt weld) and
condition because of their different positions in lon-
- a full penetration butt weld with a constant weld
gitudinal direction of the bridge girder. Table 3
foot length tw, see figure 9.
summarizes the joint dimensions and parameters.
Figure 10 shows the influence of the these different
types of weld shapes on the stress concentration 6.2 Stress Range
factor. The consideration of an additional foot length
Fatigue load model 3 (FLM 3) of Eurocode 3
tw leads to a reduction of the SCF of about 10 to
Part 2 (1997) was used to determine the nominal
15%.
stress ranges in the corresponding tubular members.
FLM 3 consists of a single 4-axle vehicle truck with
axle loads of 120 kN.
According to Eurocode 3 Part 2 (1997) the re-
sulting stress range is transformed into the damage
equivalent stress rang ∆σE,2 related to 2·106 cycles in
order to make it comparable to the fatigue strength
∆σC. This is realized by the so-called damage
equivalent factors λi, see equation (2). These factors
depend on fatigue relevant parameters as e.g. the
traffic volume, the design life of the bridge or the
Table 3. Joint dimensions and parameters
joint 1 joint 2 joint 3
do 457 457 457
t0 65 55 45
d1 267 267 267
t1 45 36 28
g 79.0 65.5 52.8
β = d1/d0 0.58 0.58 0.58
Figure 9. Different weld shapes
γ = d0/2t0 3.52 4.15 5.07
τ = t1/t0 0.69 0.65 0.62
Θ 60 60 60
φ 90 90 90
shape of the influence line. Table 4 summarizes the The recommended minimum SCF value is a rea-
values applied as they are given in Eurocode 3. sonable assumption and should especially be applied
for thin sections used for buildings or crane struc-
∆σ E , 2 = λ 1 ⋅ λ 2 ⋅ λ 3 ⋅ λ 4 ⋅ ∆σ nom (2) tures. However, the authors believe, that for thick
sections and high quality full penetration welds with
good accessibility SCF values in the range between
6.3 Fatigue Verification 1.5 and 2.0 may be acceptable as well.
Using the hot-spot method, the fatigue limit state
can be verified using equation (3): With the above mentioned assumptions and based
on the SCF values given in Table 5, the fatigue limit
γ Ff ⋅ SCF ⋅ ∆σ E , 2 ≤ ∆σ C.hs γ Mf (3) state has been verified for all three joints, thus
clearly indicating, that for the specific type of bridge
Where γFf and γMf are the partial safety factors for also welded CHS joint would have been possible.
the fatigue limit state, ∆σC.hs is the characteristic
value of the fatigue strength against hot-spot stresses
for 2·106 number of cycles and SCF is the stress 7 CONCLUSIONS
concentration factor.
The safety factors were chosen to γFf = 1.0 and This contribution covers the application of circu-
γMf = 1.15. lar hollow sections in bridge design and tries to give
Values for the fatigue resistance ∆σC.hs for CHS an overview about the advantages and disadvantages
section joints are given in CIDECT Serial No. 8 of either cast steel or welded joints in order to help
(2000) by the following formula: practical engineers in their decisions and to allow for
∆σ C.hs = 1 3 ⋅ (12.476 − log( N f ) ) a further application of circular hollow sections in
(4) bridge design.
+ 0.06 ⋅ log( N f ) ⋅ log(16 / t ) Concerning a cost effective and robust design of
CHS joints for bridge structures the following con-
Where Nf is the number of cycles to failure and t
clusions are drawn:
the wall thickness. Although the given formula is
- cast steel joints should be used if there are several
limited to a maximum thickness of t ≤ 50 mm it has
members to be connected at one joint resulting in a
been applied in this case.
complex joint geometry. For such cases, the casting
process allows an optimal design of the joint ac-
Based on the aforementioned FE calculations, the
cording to the flow of internal forces, increasing
relevant stress concentration factors (SCF) for the
their static and fatigue resistance compared to
observed three joints are given in Table 5. Herein the
welded joints.
first lower index describes the member (ch = chord,
- for standard joint types such as e.g. K- or KK-
b = brace) and the second one the loading condition
joints that are typically used for triangular truss
(ch = chord loading, ax = axial balanced brace load-
girders, welded joints are the more economic solu-
ing).
tion and, providing a high manufacturing standard
In contrast to the given values in Table 5,
especially for the welds, a possible alternative.
CIDECT Serial No. 8 (2000) recommends a mini-
mum SCF = 2.0 unless it is negligible. The follow-
Through detailed investigations connected to a
ing reasons are given:
recently completed CHS truss bridge near Stuttgart,
- a possible and uncontrollable crack initiation from
it has been shown, that for this bridge instead of cast
the weld root, that is not covered within the SCF
steel also welded joints would have been a possible
value determined for the weld toe,
alternative. This conclusion could be drawn from
- possible deviations of the principle stress direction
numerical studies based on FE calculations applying
from the direction perpendicular to the weld toe
the hot-spot stress approach for the fatigue assess-
and
ment of the welded joints.
- difficulties in FE modeling.
Table 4. Assumed damage equivalent factors λi Table 5. Stress concentrations factors (SCF)
NObs = 0,5·106 lorries per year joint 1 1.99 1.68 1.40 0.30
λ3 design life of the bridge; N = 100 years 1.0 joint 2 2.06 1.78 1.39 0.36
λ4 number of lanes with heavy traffic; k = 1 1.0 joint 3 2.08 1.85 1.37 0.41
λ = λ1⋅λ2⋅λ3⋅λ4 1.41
Furthermore, these studies clearly indicated that 9 ACKNOWLEDGMENT
for a wide application of cost effective and robust The authors would like to express their gratitude
welded CHS joints, the currently existing design to Mr. G. Stuba who did the FE-Analysis within his
guide CIDECT Serial No. 8 (2000) should be ad- diploma-thesis.
justed to the specific situation of bridge structures,
e.g. by extending the parameter range of the SCF
values or providing values for the fatigue resistance
also for a large wall thickness.