Skattørsundet

Knowing

Many European bridges are more than fifty years old and do not meet current codes, yet they remain an essential part of the road network. In northern Norway the issue is plain: according to a 2017 report by Statens vegvesen, about half of the country's 16,791 bridges lacked adequate inspection and about 1,087 showed serious or critical damage.

The Skattørsundet bridge links the islands of Kågen and Skjervøya, in Troms og Finnmark county, and is a strategic connection: together with the Maursund tunnel it joins Skjervøy to the mainland. It is a reinforced concrete slab bridge opened in 1971, 804 m long, with a longest span of 32 m, divided into four parts by three expansion joints.

The survey of 10 July 2024 revealed heavily degraded and probably locked bearings, marked section loss at the base of several piers, cracks in the abutment walls and corroded external post-tensioning bars.

Measuring

The operational modal analysis campaign took place between June and July 2024, at night only and without closing the bridge. It used 16 high-sensitivity piezoelectric accelerometers (PCB 393B12, 10 V/g), two synchronised LEAP DAQ units and an acquisition server: signals recorded at 1000 Hz in 30-minute acquisitions were resampled to 20 Hz for identification.

With 39 piers and two abutments the whole bridge could not be instrumented at once, so the measurements were organised in five campaigns, with multirun layouts overlapping at a few points.

Modelling

A finite element model with thick beam elements (LUSAS) was built from the original drawings, without the post-tensioning tendons for lack of data. The first fourteen computed frequencies lie between 0.91 and 1.93 Hz, with non-uniform mode shapes involving only parts of the bridge: so the model was used first of all to decide where to place the sensors, and it was checked against the spot tests of 19 June.

The first frequencies identified with the 13 July multirun are close to those of the uncalibrated model, showing that the lowest transverse modes depend mainly on the pier stiffness. For higher modes the measured frequencies are lower than the computed ones, as if the deck concrete were less stiff than assumed or degraded.

On the part containing the main span the model was then updated automatically (MuDI software), changing the elastic modulus of 40 portions of deck and piers to reduce the gap with the measurements: the final result has a MAC of 0.937.

Outcome

The measurements show where to look: locked bearings and a softer zone around the main span.

The first part has no longitudinal modes: it is rigidly connected to the abutment, as the survey suggested. The second part vibrates longitudinally at 1.224 Hz, the third and fourth at about 1.915 Hz: the second part is therefore more flexible, because of taller piers or stiffness reduced by damage or deterioration.

On the 32 m main span the maximum modal displacement does not fall at midspan despite the symmetric geometry, and the mode shapes show an irregularity near pier 12. Model updating concentrates the stiffness reduction in the third and fourth spans and in the second pier of that part: the places to target in the next investigations.