Main rupture and adjacent belt of right-lateral distortion detected by viaduct at Kaynaşlı, Turkey
12 November 1999 Düzce Earthquake
by
Arvid M. Johnson
Kaj M. Johnson
Joe Durdella
Mete Sozen
Turel Gur
Earthquake Engineering Group
School of Civil Engineering
and
Harry Fielding Reid Earthquake Rupture Laboratory
Department of Earth and Atmospheric Sciences
Purdue University
3 April 2001
Contents
Abstract 3
Introduction 3
Background of this Study 8
Rebound Theory vs. Observation 8
Right-lateral Rupture Zones 10
Composite Idealized Model 11
The Düzce Earthquake 12
The Surveys 14
Deformation of Quadrilaterals of Resurveyed Piers 17
Belt of Right-lateral Shearing at Kaynaşlı 21
Explanation of Right-lateral Process Zone on Either Side of Right-lateral Fault 22
Process Zone 22
Assumptions and Observations 23
Strains Above Strike-slip Fault Propagating in Mode III 25
2 km Depth 25
0.5 km Depth 25
0.1 km Depth 25
Results of Simulations 26
Implications of the Research 27
Acknowledgements 29
Appendix 30
References Cited 33
Abstract
The fault trace of the 12 November 1999 earthquake in the Düzce-Bolu region in Anatolia crossed the alignment of a two-kilometre viaduct in Kaynaşlı that had been carefully surveyed. The builders of the viaduct, the ASTALDI-BAYINDIR Co., resurveyed the viaduct after the earthquake. We repeated the survey for approximately one kilometre of the eastern end of the viaduct and obtained essentially identical results. Though it was unfortunate that the earthquake damaged the new structure, the piers did produce an unusual record of ground deformation of an earthquake. In effect, the viaduct was a giant strain gage that yielded reliable data about ground movement and distortion near a fault. This paper describes the survey data and their evaluation leading to convincing evidence that (a) the fault trace must be considered, not as a fault line or plane, but as a fault zone with a finite width and that (b) the structural damage within the zone was caused, not primarily by ground acceleration, but by ground distortion. Along the right-lateral fault at Kaynaşlı, the fault zone consists of right-lateral movement at the main trace, a zone of right-lateral distortion near the trace, bounded by left-lateral distortion. The 12 November 1999 event in Turkey, like the ground deformation and fracturing at Landers, California (Johnson et al., 1994, 1996), thus affirmed the apparently forgotten conclusions from the studies by Lawson (1908), Gilbert and Reid (1910) of the 1906 San Francisco earthquake that earthquake ruptures typically occur throughout zones or belts, rather than along linear traces or planes.
Introduction
Kaynaşlı Viaduct I (Figure 1) is a double-ribbon bridge, about 2 km long, with piers about 45 m tall. It is part of the E-W Trans-European highway under construction between Istanbul and Ankara. The viaduct trends roughly east-west, lifted above the Asarsu Valley by two sets of piers, one for eastbound and the other for westbound lanes. There are 57 pairs of piers plus an extra one under the eastbound lanes. A typical pier is number 45, shown in Figure 2. The base of the pier is a thick slab of concrete 16 by 18.7 m in plan. The slab rests on piles that extend to bedrock beneath the gravel of the Asarsu Valley. Standing on the slab is the shaft of the pier, 45.6 m long, 8 m wide and 3.7 m broad. The cap on the pier, 5 m by 17.25 m in plan, supports segments of the roadway.
The points in Figure 2 are the survey points of ASTALDI-BAYINDIR Co. The points provide networks of points that were surveyed before and after construction. During the Düzce earthquake the elements of the roadway and, presumably the piers themselves, moved in various directions. Their positions were accurately resurveyed so that the job of repairing and completing the roadway could be planned. The Purdue Task Force was provided with complete sets of the survey data through the good offices of Dr. Çetin Yılmaz of Middle East Technical University. We have used these networks in order to compute strains and deformations in the ground beneath the piers. Our own resurvey was of points a couple of metres below points corresponding to 7 and 19 in Figure 2 on piers under the eastern end of the viaduct. Thus our survey provides a network of points for part of the viaduct.
The bases of the piers moved due to offset of the main break of the November rupture. Figure 3 shows a photograph of the main break beneath the viaduct, taken shortly after the earthquake. The rupture, perhaps a metre wide, is trending eastward toward pier 47 R, under the westbound lane. The tension gashes, oriented perhaps 45 degrees clockwise from the trace of the main rupture indicate right-lateral sharing. According to Barka and Altunel (2000), the offset of piers was right-lateral and on the order of 1.15 m. Figure 3 shows the rupture zone approaching pier no. 47L, under the eastbound lane, as viewed eastward. The main rupture steps left about a metre in the central view, and thrusting occurred at the step, again reflecting right-lateral offset across the main rupture. The rupture threaded its way between piers 46 R and 46 L. The caps of both piers 45 R and 47 L were visibly misaligned relative to the roadway, as though they had rotated a few degrees in a clockwise sense (as viewed from the ground), which is consistent with right-lateral, simple shearing.
Figure 2. Survey stations on pier 38
The ground deformation, though, was not restricted to the ruptures shown in Figures 3 and 4. We shall show that the spacing of piers changed many metres on either side of piers 45 and 47. The earthquake rupture intersected Viaduct I at a low angle so a long stretch of the viaduct was disturbed by the main rupture and deformation in ground on either side of the main rupture.
Figure 1. View west-northwest of Kaynaşlı Viaduct I. Arrows to left mark the main rupture. Arrow to right is approximate location of main rupture at viaduct.
Figure 3. View east southeast of main rupture and some of piers of Kaynaşlı Viaduct I. Main rupture is headed toward Pier 47 L at left (photo courtesy of P. Gulkan). Offset here was 1.15 m right-lateral.
Figure 4. View southeast along piers beneath Kaynaşlı Viaduct I. Dirt roadway on left, south side of viaduct. Rupture zone in foreground cuts across road and runs into pier 45R, under the eastbound lane. Left step in main trace in middle distance produces thrust along rupture. (Photo Courtesy of P. Gulkan)
Figure 6. View west of Kaynaşlı Viaduct I. Figure 9 taken from point about 100 m behind photographer of this photo. Multiple fractures characterize the main rupture here, on south side of viaduct. Yellow house in background was destroyed by rupture. (Photo Courtesy of P. Gulkan)
Background of this Study
Rebound Theory vs. Observation
To explain why one would be interested in such measurements, we need to point out some inconsistencies between the elementary notions of elastic deformation during rebound during an earthquake on the one hand and the field observations made along actual earthquake ruptures on the other. We go back to the ideas of the elastic rebound theory of earthquakes and the field observations along the 1906 San Francisco earthquake rupture discussed by G. K. Gilbert (1907) and Harry Reid (1910). In these early papers, based partly on study of the 1906 San Francisco earthquake, an earthquake
Figure 7. Idealized ground deformation along earthquake ruptures. A. Passive marker placed across fault just prior to rupture. B. Offset of passive marker along fault and deformation in ground according to classic elastic-rebound theory. C. Belt of permanent, right-lateral deformation containing main rupture. Elastic, left-lateral deformation (elastic rebound) on either side. D. Right-lateral belt of shearing on one side of main rupture. Left-lateral deformation on either side.
is considered a result of sudden slip on a fault in ground that is under high enough stress for the fault surface to fail. Thus, the earthquake is considered a result of a stress drop at a fault. Reid (1910) explained that markers at the ground surface would be offset in different ways, as a result of strike-slip earthquake rupture that broke through to the ground surface, depending on when the markers were placed. The only case of interest here is the marker placed shortly before earthquake rupture. Thus, let the dashed line in Figure 7A represent the map view of the trace of a strike slip fault and line ab represent a passive marker placed across the fault before earthquake rupture (Figure 7A). The ground is highly stressed. If the fault slips suddenly (traction on fault suddenly drops), the ground will shake due to radiated energy, and the ground on either side of the fault will deform elastically. The passive marker will appear approximately as in Figure 7B, as two deformed line segments, ac and db, broken by the fault. Note that the offset on the fault is right-lateral, but the deformed line indicates that the rock on either side of the fault was distorted in a left-lateral sense. This is, of course, due to the elastic rebound according to the notion of earthquake generation. The left-lateral distortion is indicated via the exaggerated deformation of squares into parallelograms on either side of the fault (Figure 7B).
This is the heart of the elastic rebound theory of earthquake generation.
A corollary of the elastic rebound theory of earthquakes has been that there are three causes of earthquake damage, all described in the 1906 earthquake report (Lawson, 1907):
· Shaking of structures in excess of their design capabilities.
· Collapse of ground beneath structures due to landsliding or liquefaction of soil generated by shaking.
· Direct offset along a main fault rupture.
During the 80 years between the 1906 San Francisco earthquake and the 1989 Loma Prieta earthquake in California earthquake damage has been explained in terms of one of these three phenomena.
The far-field, left-lateral distortion that we associate with elastic rebound along a right-lateral fault, of course, has been thoroughly documented in the literature, new and old. This is the deformation that is shown by regional GPS and geodetic data for an entire fault rupture. For example, strains were calculated by Fleming and Johnson (1997) from GPS measurements of displacements in a network of points with spacings of 20 to 70 km reported by Hudnut et al. (1994) for the north end of the Landers earthquake rupture. The principal strains were 7 x 10-5 and –7 x 10-5 on the east side of the Landers rupture, confirming left-lateral distortion in the NW direction that one would expect from elastic rebound along a right-lateral fault.
Right-lateral Rupture Zones
Observations made at the ground surface closer to the main rupture at Landers, though, are at odds with the simple elastic rebound theory (Johnson et al., 1993; 1996a; 1996b). The most obvious feature that deviates from the simple rebound theory is that there may be a complex rupture belt at the ground surface, not merely a fault trace. In the case of right lateral, strike-slip faulting[1] at Landers, there was a zone or belt of right-lateral shift 50 to 500 m wide along a right-lateral earthquake rupture, not a simple fault plane of right-lateral slip (Figures 6C and 6D) (Lawson, 1908; Johnson et al., 1994). Instead of a fault plane, there may be a main rupture, which may be a metre or more wide, along which much of the right-lateral shift occurs, on either side or perhaps within the belt of right-lateral rupturing (Johnson et al., 1994).
It turns out that Johnson et al. (1994) re-discovered belts of rupturing; they had been described by Lawson, Gilbert and Reid after the 1906, San Francisco earthquake, at least along the northern stretches of the rupture (Lawson, 1908; Reid, 1910). The following quote describes a broad shear zone, about 100 m wide, as well as a narrow shear zone with long, echelon fractures at an acute clockwise angle to the walls of the shear zone. The zone of most intense rupture, "the fault–trace or rupture plane" occurred on one side or the other of a shear zone:
“ . . . The surface of the ground was torn and heaved in furrow–like ridges. Where the surface consisted of grass sward, this was usually found to be traversed by a network of rupture lines diagonal in their orientation to the general trend of the fault . . . The width of the zone of surface rupturing varied usually from a [metre] up to [15 m] or more. Not uncommonly there were auxiliary cracks either branching from the main fault–trace obliquely for [30 to 100 m], or lying subparallel to it and not . . .directly connected to it. Where these auxiliary cracks were features of the fault–trace, the zone of surface disturbance, which included them, generally had a width of [about 100 m]. The displacement appears thus not always to have been confined to a single line of rupture, but to have been distributed over a zone of varying width" (Lawson, 1908, p. 53; italics ours).
Thus, it has been established that belts of fracturing and distortion can form along an earthquake rupture, so the picture of simple elastic rebound needs to be emended, at least near the ground surface.
Composite Idealized Model
Although at Landers there was a belt of right-lateral deformation, made visible by ground fracturing, as well as far-field left-lateral deformation, shown with GPS stations spaced 10 to 20 km apart, no information could be obtained about ground deformation in the ground between these quite different types of deformation. At Tortoise Hill, Landers, a double ladder array of control points, spaced 500 m to 1000 m apart, extended from a distance of about 4 km SW of the fault to the fault and a short distance NE of the fault (Fleming and Johnson, 1997). A re-survey of the control points showed that the horizontal strains are about 10-4, which is smaller than the combined surveying errors, about 3 ´ 10-4. The strains in ground outside the right-lateral rupture zone, which was about 500 m wide there, could not be determined, so it remained unclear whether they were left-lateral, rebound strains or right-lateral, permanent strains. Nor was it possible to relate the left-lateral, far-field rebound, the near-field right-lateral rupture zone, and the included main ruptures or faults.