You are on page 1of 22

Image size: 7,94 cm x 25,4 cm

Workshop on Eurocode 4-2

Composite Bridges Swedish experiences from Eurocode
Robert Hllmark
Bridge Designer, Rambll Sverige AB

Table of contents
Eurocode from a Swedish bridge designers point of view Eurocode vs. the old Swedish Bridge Code A more national approach than previous speakers

Problems using the codes

Improvement proposals

Reference bridges

Three bridges designed according to Eurocode will be the reference objects for this presentation
- One single span bridge - One multi span bridge - One multi span bridge I-girders I-girders Box-girder

Bridge 1: Skulns Bridge over E4

Single span bridge in northern Sweden Span length 32 m, two I-girders. Steel girder height 1,5 m. Local road that crosses a highway AADT 50-200

Bridge 1: Skulns Bridge over E4

The first composite bridge that was designed by Rambll Sweden, according to Eurocode. On a pre-design stage, the steel weight was estimated by a calculation according to the old Swedish codes.
A possibility to compare the results from the different codes.

Bridge 2: Forsj-bridge, Katrineholm RV55

Five span bridge, two I-girders Span length 45-58 m Steel girder height of 2,5 m AADT ~5000-6000

Bridge 3: NN (under tender evaluation)

2 cable stayed main-spans 2 approach bridges > 500 m, composite cross-section

Span length 64-90 m Box-girder height > 4,0 m

Main objectives in this presentation

Global analysis Design in ULS Fatigue design Design in SLS

Global analysis
One of the largest differences in Sweden, is how we are dealing with the cracked concrete, at internal supports.
Swe. code ~20-25% of the span length on both side of an In midspan we (always) have cross-section class 1, and a internal support is treated as very cracked-concrete. The possible failure is expected to be ductile. cracked concrete is modelled with a 60% stiffness It is unexpected that the simplified method seems to decrease compared to uncracked concrete. the support moments and lower the safety margin at internal supports, where we have possible failure modes that can occur The 15%-method has been used so far, due to its more suddenly. (lateral torsional buckling of bottom flanges, stiffeners etc.) simplicity (only one global analysis is necessary). Why was exactly 15% chosen?


(measurements? comparison to the tensional stiffening model?)

EC Bro2004 Sw.code Eurocode


Result from Bridge 2

M [MNm]


The support moments decreases by ~10 %. The field moments are increased by ~10 %.

-5 0 0 x [m] If tension stiffening is used, the result might be different. 50 100 150 200 250




Global analysis
Another difference is how we are dealing with concrete of varying age.
Previously, as soon as composite action is achieved, the concrete has been given its long term stiffness (one third of the short term stiffness).
-This stiffness has been used for all long term loading (concrete dead load, shrinkage, nonstructural bridge equipments etc.)

Today, the bridge designer must estimate the construction procedure on an early stage, since the concrete long term stiffness is treated as time dependent.
-The construction procedure and time schedule is often not known on an early stage. Therefore the bridge designer has to estimate how long time it takes to cast each segment. -For concrete dead loads, one mean value of the concrete age is used to model the concrete stiffness.

A small sensitivity analysis gives

A bad guess will not effect the design of the bridge significantly.

ULS Ultimate Limit State

Before: Equivalent load type 1 governing the design regarding shear forces, midspan moments and often also internal support moments.
Alane1 = 250 kN Alane2 = 170 kN plane1 = 4 kN/m2 plane2 = 3 kN/m2 plane3 = 2 kN/m2


LM1 is often not governing the design regarding shear forces or moments

Special load model in Sweden
A = 160 kN B = 260 kN q = 5 kN/m fyd = fyk/1,2


A = 260 kN B = 300 kN q = 5 kN/m

Bnew = Bold*1,2 Bnew = 312 kN qnew = 6 kN

Today, the EG-vehicles is often the governing load regarding shear forces and moments.

Still the relative load effect is less than before

From our point of view, the design in ULS according to Eurocode (compared to Bro2004) results in lower load effects lower safety level in ULS

FLS Fatigue Limit State

The simplified method ( -method) for road bridges up to 80 m is still the only method we have used.
FLM 3 is used to calculate the stress amplitude (4 x 120 kN)

Old code
- Generally, fatigue was not governing the design of the structural elements in road bridges. - The number of shear studs was usually the only detail that was limited by the fatigue limit state. - In road bridges, the flanges were often made in S420/S460, since fatigue was not a problem.

- Fatigue will govern the design of quite a lot of details in a road bridge. - The number of shear studs is nowadays often governed by the ULS and SLS. - It will not longer be economical to use S420/S460 in the same extent as earlier, since we do not manage to fulfil the fatigue requirements in the steel girders at midspan.


Bridge 2, Katrineholm-bridge, is chosen to illustrate the fatigue problems.

- AADT < 6000 vehicles/day - Techical lifetime 80 years - The -factors are calculated Nobs = 125 000 (EN 1991-2 Tab.4.5)

Bending moment, endspan 1 & 5


This factor allows us to have the following stress amplitude in the bottom flange, concerning detail category 80 MPa (web stiffeners welded to bottom flange).

p = 80 / (1,35*1,0*1,00) p = 59,3 MPa

In Bridge 2, we have almost 100% utilization ratio in FLS at several places and details along the bridge. In the same bridge in ULS we have only 85% utilization ratio in midspan and 94% at internal supports. This is a huge difference compared to what we are used to in Sweden.
not be From now on (using this fatigue method), on highway bridges, it will economical to use S420/S460 in the bottom flanges in midspan.


Another detail with fatigue problem, on Bridge 2, was the notches in the assembly joints, c = 71 MPa. In order to increase c small plates were welded in the notches in order to seal them. Is it possible to achieve a higher c in another way?


If Bridge 2 would have an AADT > 10 000 vehicles/year

Nobs = 500 000 heavy

- due to fatigue the total amount of steel is increased with several percentage. - larger bottom flanges would be needed in the midspan (S460 S355) - the utilization ratio in ULS, would be far below 90% along the whole bridge.

Bridge 1 have an AADT < 200

Nobs = 50 000 heavy vehicles/year

- still this bridge has a utilisation ratio that is higher in FLS than in ULS FLS seems to be governing the design of the midspan in all composite bridges, no matter how few cars that crosses the bridge.

SLS Serviceability Limit State

SLS is mainly used to check the allowed deflection and breathing (in the design of the steel girders)
- Breathing Previously, breathing has quite often been governing the web thickness in the area where the bending moment are shifting from positive to negative. (concrete shrinkage compression over the whole web height) Eurocode allows us to use extremely slender web plates without getting problems with breathing. For a road bridge with a span of 50 m, and a web height of 2,0 m, the web plate can be as thin as 9 mm without problems with breathing.

Web breathing will never be a problem in Swedish bridges according to this formula

Conclusions and discussion

It seems as the safety level in ULS has been lowered. At the same time the safety level in FLS has been raised. Is this reflecting the reality?
- Fatigue failure seems to occur in the details that are not checked in the design stage, and often due to other stresses than the nominal stresses in the girders. - Do we have problems today with fatigue in composite road bridges? (in the details that we are checking for fatigue) - Is the national value Qm1 = 300 kN reasonable?
- UK Qm1 = 260 kN

- Other countries?

Problems using Eurocode

Not one code

several codes, regulations, national constitutions

- hard to find what you are looking for - a lot of references followed by another reference. - a huge amount of paper to handle

Lack of knowledge
- both among the bridge designers and the bridge owners

In a lot of formulas it is hard to understand the background to all factors/parameters

- the simplified fatigue calculation model is one example - it is hard to get a genuine feeling, how the formulas is reflecting the reality

Improvement/Research proposals

In the design of Bridge 3 we have struggled a lot with patch loading.

the contractor would like to launch the steel-girders together with the prefabricated concrete deck elements (except in the cantilevering part)

we have suggested to use two different plate thickness in the same web
- the lowest 1,3 m will have a permanent thickness of 28 mm - the upper 3,2 m will have a varying thickness (from plate to plate) between 15 - 28 mm

since some of the Nordic steel producers have a limitation of their plate width to ~3,2 m. It might be economic to use different web-thickness when a longitudinal butt weld is already necessary.

it would be nice with some guidelines how to deal with varying web thickness Maybe this is a possible research topic!