Viaduct Damage Assessment After the 2023 Earthquake in Turkey

To view the figures and tables associated with this article, please refer to the flipbook above.

Fourteen viaducts are situated along the Tarsus-Adana-Gaziantep (TAG) Highway, which is the main transportation route in Southern Turkey. Three lanes of traffic flow in each direction. Of the 14 viaducts, five were damaged during the 2023 Mw 7.8 earthquake.

Built just before the year 2000, all five viaducts are large and impressive structures set in the low mountains, with interesting designs for earthquake response. Originally designed for a peak ground acceleration (PGA) of 0.4 g (with return period of about 500 years), the damaged bridges were loaded with significantly more ground shaking than they were designed for. Station 2712, the closest free-field strong motion station to all five of these damaged viaducts, recorded a PGA of 0.607 g (see full references with the online article). Other free-field strong motion stations along the fault rupture line showed even larger PGA values (from USGS), but these were not close to the TAG Highway, or any of the damaged bridges (Fig. 1). Hence, based on the available measured data, and how close these five bridges are to each other and to Station 2712, it is reasonable to expect that they were all overloaded by about 50% beyond what they were designed for.

Figure 1 shows how close the USGS-defined fault rupture line for the Mw 7.8 earthquake is to the five damaged bridges, and how the TAG Highway (O-52 in Fig. 1) turns and follows it, in parallel, before turning away again. This proximity to the fault line explains why these five viaducts were damaged, and why the other nine viaducts on the TAG Highway weren’t; for a large earthquake, PGA reduces with normal (perpendicular) distance from the fault rupture line, and not as the distance from the earthquake epicenter. Also, near-field earthquake conditions existed for these damaged bridges since they were all within 225 meters (738 feet) of the fault line that ruptured.

Two of the five major bridges—the Ataturk and Turgut Ozal Viaducts—are very similar in both scale and seismic design features. As such, their new seismic retrofit schemes were similar as well. Part 2 of this series will appear in the October issue of STRUCTURE and focus on the other three viaducts.

Seismic retrofit designs for the Ataturk and Turgut Ozal Viaducts were done by Cenan Ozkaya, as the engineer of record, working within the PONTEM Engineering Co. The typical seismic retrofit design of a bridge structure is in anticipation of a future large event that will probably never happen, and that the bridge was not originally designed for. This project, however, is the seismic retrofit design for a bridge that has already been subjected to the maximum considered earthquake (MCE). Importantly, these two major bridge structures were damaged but did not collapse, and they were saved for future use by the ongoing seismic retrofitting to larger PGA values than considered in the original design. The high quality of both (1) the original design details and (2) the construction, are important features that helped save the viaducts during the February 2023 Mw 7.8 earthquake.

Retrofitting works are being carried out by the SNH Construction Company, and the owner of the viaducts is the Motorway Division of the General Directorate of State Highways, in the Ministry of Transportation and Infrastructure of the Republic of Turkey.

Just after 4 a.m., local time, on February 6, 2023, a Mw 7.8 earthquake struck Southern Turkey followed by a Mw 7.5 earthquake only six hours later. The first one is the largest earthquake to ever hit Turkey and is consistent with an MCE event in California. When converted to the older Richter scale, it has a magnitude of 8.1, which is the same magnitude as the famous 1906 San Francisco earthquake and the future “big one” in California. Furthermore, the right strike-slip fault mechanism that caused the 2023 Turkey earthquake is the same mechanism for both California earthquakes, past and anticipated future, along the San Andreas fault. Since Turkey closely follows Caltrans seismic bridge design specifications and practices, it is of interest in California to see how Turkey’s major bridges and viaducts performed in this very large earthquake, especially since Caltrans’ infrastructure has not yet been tested like this (the many California bridges were built after the 1906 San Francisco earthquake).

The 2023 Mw 7.8 earthquake, in Turkey, occurred due to sudden slip along the Eastern Anatolia Fault Zone (EAFZ), which is the second most active fault system in Turkey and is the most significant for the two major viaducts considered in this article. It is a right strike-slip fault system that is 550 km (342 miles) long and separates the Anatolian and Arabian tectonic plates. These two plates move relative to each other at a rate of about 10 to 12 mm (0.394 to 0.472 inches) per year. Using the maximum measured relative displacement of 40 feet from the 2023 earthquake, and the upper value of the relative rate of plate motion, this translates to a return period for such a large earthquake of 40(12)/(0.472) = 1,017 years. Therefore, a large earthquake is expected along this region about every 1,000 years, which is consistent with the historical record going back just over 2000 years, to 30 BC. A large earthquake happened in 30 BC, then again in 1114 AD, and the most recent earthquake was in 2023. The average time between these three historical records is 1,027 years, which agrees with the approximate time of 1,000 years between big events.

Knowing that the last large earthquake in this region was in 1114 AD, one could have predicted this latest large earthquake to occur where it did in about 2114, with perhaps a 150-year window in either direction to allow for natural variations in soil/rock strength. So, it would have been expected anytime between 1964 and 2264, for example. Now that the slip has occurred along this fault, the strain energy that had built up for 1,000 years is released, and it will take another 1,000 years or so before enough strain energy builds up to suddenly fracture the rocks at the tectonic plate interface and produce a large earthquake again.

At a height of over 130 m (427 feet), the Ataturk Viaduct was the tallest viaduct in Turkey when it was built, and is the tallest bridge anywhere in the world to be subjected to an earthquake of this magnitude (to be hit by its MCE). As shown in Figures 2-5, the bridge is so tall that clouds and mist often form below the superstructure.

The viaduct is supported by hollow rectangular reinforced concrete columns with 9 m x 6 m (29.5 feet x 19.7 feet) main core (without ears/protrusions) cross-section dimensions at the column top and a wall thickness of 0.60 m (1.97 feet). Column flares linearly increase these cross-section dimensions from the top to the bottom of the column, with primary vertical rebar following these flared regions to increase the internal moment arm and, hence, moment capacity of the column toward its base, where the moment demand is largest. As with the three damaged bridges in the companion article, the columns have primary vertical rebar cutoffs but with no adverse effects for this viaduct. The largest reinforced concrete foundations have dimensions of 22 m x 34 m in plan and are 6 m thick (72.2 feet x 112 feet x 19.7 feet). This eight-span viaduct has a total length of 802 m (2,630 feet; about half a mile), with span lengths that range from 70.7 m (232 feet) to 110 m (361 feet) and is on a 1,200 m (3,937 feet) horizontal curve, with 3% longitudinal slope. Column heights range from 130 m (427 feet) to 9.62 m (31.6 feet). Expansion joints are provided at the ends of the bridge.

Side-by-side superstructures consist of weathering steel U girders, with width of 9 m (29.5 feet) and depth of 4.9 m (16.1 feet), are 17.5 m (57.4 feet) wide, and have a 4.5 m (14.8 feet) clear space (gap) between them. Girders were placed by incremental launching and have a cast-in-place (CIP), reinforced concrete topping slab that is 39.5 cm (15.6 inches) thick, as well as an asphalt concrete overlay for the driving surface. The steel girders are supported by a prestressed concrete transverse beam and reinforced concrete single-column-bents (Fig. 5). Hence, the reinforced concrete columns act as cantilevers in both the longitudinal and transverse directions, with large reinforced concrete footings at their bases, and micropiles below the footings to prevent overturning; as well as to eliminate problems arising from karstic cavities present along all the viaducts in the Nurdag district. The tallest columns at Bents 5, 6 and 7 have caisson foundations.

The original seismic design included multiple shock-absorbing bumpers added to the expansion joints at both ends of the bridge, and at Bents 2, 3, 4, 5 and 8, which takes the longitudinal force, with steel beams and slider pot bearings not allowing longitudinal movement at the other bents. These unique, elastic, high-force-capacity bumpers act in tension and compression and were intended to provide a self-centering behavior to the viaduct after an earthquake. Tall columns at Bents 6 and 7 are longitudinally guided by using a custom steel connection detail by the contractor. However, since the seismic ground shaking was significantly larger than the viaduct was designed for, the bumpers and expansion joints failed, with the bumpers no longer functioning after the earthquake (Fig. 6). Sliding surfaces of the bearings were also damaged and are no longer functional. In the transverse direction, the bridge was restrained from movement by steel seismic braces.

The force-displacement response of the bumpers is initially linear but with increasing slope as the displacements increase; hence, a stiffening spring. The steel superstructure, columns, footings, and bent cap were not damaged in the earthquake. Both approaches to the viaduct settled, and trees adjacent to the bridge moved downhill in the soil. The concrete slab was locally damaged at the expansion joints due to impact forces, while the asphalt concrete overlay buckled, and many potholes, bumps, and undulations are now on the driving surface, significantly affecting traffic.

Since these elastic bumpers are no longer made, and due to the extreme difficulty of retrofitting the large, hollow rectangular reinforced concrete columns and large reinforced concrete footings, it was decided to reduce any future seismic forces on the substructure by incorporating seismic isolation at the tops of all the columns. This lengthened the longitudinal period of the structure from 2.6 seconds to 3.45 seconds, reducing maximum structural accelerations and, hence, the seismic substructure forces. It also ensured that the shorter, stiffer, columns wouldn’t get more seismic load than the taller and more flexible columns. In the transverse direction, the structure period increased only slightly from 4.06 seconds to 4.20 seconds. Viscous dampers were also added in the bridge longitudinal direction. The final seismic retrofit design was validated using nonlinear time-history analyses (NTHA), showing that the superstructure, bent caps, columns and footings remain linear-elastic from a design maximum considered earthquake event. This viaduct was 75 m (246 feet) from the fault rupture line.

The Turgut Ozal Viaduct (Fig. 7) is very similar to the Ataturk Viaduct, with the same use of elastic, high-force-capacity bumpers at the end expansion joints, as well as at the bents with shorter columns, Bents 2 and 5. It is also a very large structure with two side-by-side steel superstructures of the same dimensions as the Ataturk Viaduct, with reinforced concrete topping slab and asphalt concrete overlay for the driving surface. Likewise, the superstructure of the Turgut Ozal Viaduct is supported by a prestressed concrete bent cap, and hollow rectangular reinforced concrete columns that act as cantilevers, in both the longitudinal and transverse directions. However, unlike the Ataturk Viaduct, there are no column flares and no vertical rebar cutoffs. It is on a horizontal curve with 2,600 m (8,530 feet) radius, and 3% longitudinal slope.
While the reinforced concrete column cross-section details are the same as the Ataturk Viaduct, the tallest column of the Turgut Ozal Viaduct is 76 m (249 feet), which is significantly shorter than the 130 m (427 feet) column height of the Ataturk Viaduct. Also, the total length of 424 m (1,391 feet) for the Turgut Ozal Viaduct is much less than the 802 m (2,630 feet) length of the Ataturk Viaduct. Pot bearings with sliding surfaces were used at interior bents and abutments. These sliding surfaces had a displacement capacity which was consistent with the displacement capacity of the high-force-capacity elastic bumpers, demonstrating a good design. Longitudinal movement of the tall columns, at Bents 3 and 4, was restrained by steel beams. Hence, in the earthquake, all of the bents did not share the longitudinal load equally.

Damage occurred to the elastic impact bumpers and expansion joints, as well as all sliding surfaces, which are no longer functional. As with the Ataturk Viaduct, the seismic retrofit design consisted of providing seismic isolation to the tops of the columns and adding viscous dampers in the bridge longitudinal direction, which lengthened the period of the structure, reducing substructure forces, and resulting in linear-elastic superstructure, bent cap, column and footing responses from a future maximum considered earthquake event. Thus, no retrofit of the substructure was required. This viaduct was 225 m (738 feet) to the fault rupture line. Micropiles of 25 m (82 feet) length were provided under the foundations.

NTHA showed that a combination of seismic isolation at the tops of the columns and added viscous dampers allows the viaduct to survive a future maximum considered earthquake event (about a 2,500-year return period) based on the Turkish National Earthquake Design Code. PGA of 0.983 g was used for the base motions. Vertical accelerations were also included in this NTHA. The program Larsa 4D was used for static and dynamic analyses, with the XTRACT program utilized for section moment-curvature analyses. Although the axial force for any given column is constant along its length from dead load of the bridge superstructure and bent cap, the total axial force in the column varies over its height due to the added weight of the big column, requiring various moment-curvature analyses up the column height. For typical, smaller, bridge structures, the changing axial load along the column length from the weight of the column often can be ignored, allowing a single moment-curvature analysis to represent the whole column height, so long as the cross-section and/or rebar details don’t change over the column length.

The detailed analysis demonstrates that with this seismic retrofit, the steel superstructure, prestressed concrete bent caps, reinforced concrete columns and foundations remain linear-elastic from a maximum considered earthquake event. Therefore, no strengthening was required for these members. With spherical sliding (curved surface slider-friction pendulum) bearings provided at the tops of all the columns as the seismic isolation scheme, longitudinal forces from a future earthquake will be almost the same at all the bents. Displacement capacity had to be ensured, since the isolation system reduces forces but increases displacements. A total of 14 time-history analyses were conducted. Rotational mass inertia of the superstructure was included in the dynamic analyses, which was shown to be important in a study by Robert K. Dowell for single-column-bent structures.

Material tests indicated the concrete and steel had strengths that were consistent (or higher) with the values given in the original design plans. For both the Ataturk and Turgut Ozal Viaducts, the friction pendulum bearings (used as the seismic isolation system) had a dynamic friction coefficient of 0.07 and radius of 5 m (16.4 feet). For the Ataturk Viaduct, at some bents and at the abutments, supplemental elastomeric bearings were added, working in the longitudinal bridge direction to control the bridge’s natural period. At both viaducts, to save the shorter columns from high earthquake forces, transverse seismic movement is now allowed by removing shear bracing that would, otherwise, restrain these transverse movements.

Five of the 14 viaducts on the TAG Highway in Southern Turkey were damaged in the 2023 Mw 7.8 earthquake, which was the strongest earthquake to ever hit Turkey. Each viaduct was designed for a PGA of 0.4 g, which is much smaller than the closest measured PGA to all five of these bridges of 0.607 g. Importantly, all damaged viaducts were very close to the fault rupture line—225 m (738 ft) or closer—while the remaining, undamaged, viaducts on the TAG Highway were further away from the fault that ruptured.

The Ataturk and Turgut Ozal Viaducts are major bridge structures with similar superstructure, column and footing details, and dimensions. They both have impressive reinforced concrete columns of hollow rectangular cross-section, as well as very advanced and innovative original seismic designs and details, that would probably have worked well if the earthquake had been about 50%, or maybe even 75%, of the size that it was. Hence, the seismic retrofit strategy for both of these viaducts was the same, to provide seismic isolation at the top of each column by use of friction pendulum sliding bearings and added viscous dampers. This lengthened the period of the two bridges and reduced the substructure seismic forces so that the columns and footings did not need any work done, since detailed NTHA showed that they remain linear-elastic in this modified design.

Importantly, these two major viaducts on the TAG Highway didn’t collapse and were saved for future use by the ongoing seismic retrofitting, with no closure to vehicles. It is expected that under a future large earthquake of similar size to the 2023 earthquake, these two retrofitted bridge structures would not be damaged. However, it is unlikely that an earthquake of this magnitude would hit these structures again within their remaining service life. ■

About the Authors

Cenan Ozkaya got his B.S., M.S. and Ph.D degrees from the Civil Engineering Department of Middle East Technical University-Turkey. He is working as Technical Manager in the PONTEM Engineering Company.

Robert K. Dowell received his B.S. degree in Civil Engineering from San Diego State Univeristy (SDSU), and his M.S. and Ph.D degrees in Structural Engineering from the University of California at San Diego (UCSD). He is a licensed Civil Engineer (PE) and a Professor of Structural Engineering at SDSU.

Faruk Yildiz got his B.S. degree from the Civil Engineering Department of Yildiz Technical University-Turkey, and is working at the Motorway Division of the General Directorate of State Highways-Turkey.

References

Dowell, R. K. (2023).  Reconnaissance Report of Observed Structural Bridge Damage: M_w 7.8 Turkiye (Turkey) Earthquake of 2023.  SDSU Structural Engineering Research Project, Report No. SERP – 23/03, March 2023

Dowell, R. K. (2023).  Observations of the Bridge Damage Caused by the M_w 7.8 Türkiye (Turkey) Earthquake of February 6, 2023, STRUCTURE Magazine, October (Yearly Bridge Issue)

Ozkula, G., Dowell, R.K., Baser, T., Lin, J.L, Numanoglu, O.A., Ilhan, O., Olgun, C.G., Huang, C.W., Uludag, T.D. (2023).  Field reconnaissance and observations from the February 6, 2023, Turkey earthquake sequence, Natural Hazards (2023) 119:663-700.