Collapse from the Inside-out the Impact of the 2003 Bam, Iran Earthquake on the Earthen Architecture

The Arg-e Bam is a remarkable example of earthen architecture and construction that was heavily damaged in the Bam, Iran earthquake of 26 December 2003. This paper presents the hypothesis that the collapse of the walls was caused largely by a combination of the effects of (1) the additive changes made to the walls, particularly in recent restorations resulting in variations in the density and response to vibrations of different layers of unfired earth construction in the walls, and (2) extensive damage from termites and loss of the cohesion of the clay from degradation and excessive drying out, all of which interacted with the earthquake vibrations of unusually high frequency in such a way that many walls effectively burst from the subsidence of their clay internal cores. Concern is raised about the possibility of similar previously unrecognized risks to other earthen monumental structures from future earthquakes. INTRODUCTION: THE CITADEL AND WALLED CITY OF BAM The Arg-e Bam has been identified as the world’s largest complex entirely of earthen construction. Settlement on the site is said to date back as much as 2000 years, with many of the structures and archeological remains that are currently visible dating from 1300 to 1800. Despite its history as a fortified site, all of the battlements, turrets and buildings in the Arg were composed of unfired earth. Unlike many earthen monuments that are clad with brick or stone, the structures in the Arg were entirely composed of unfired earthen construction. The numerous arches, vaults and domes were constructed of sun-dried bricks, often using techniques of construction that avoided the need for structural centering. This construction was of two distinct types – unfired “adobe” masonry, known in Farsi as “Khesht,” and built up earth or “cob” construction, known as “chineh.” 1 (Figure 2) Both types of construction could be found in many of the structures, sometimes in layer s where the later work, including 20 century restoration work, would be in Khesht, while the original work would be chineh. 1 Independent Consultant in Building Technology, Oakland, California, USA. WEB: www.conservationtech.com Collapse From The Inside-Out 2 Figure 2: Chineh wall inside the Arg-e Bam that was only slightly damaged in the earthquake. As an archeological site, many of the structures in the Arg were already in ruins prior to the time of the earthquake. The walled town was gradually abandoned in the nineteenth century as people migrated out to houses located in the date palm orchards nearby. Gradually, the houses and public buildings in the Arg fell into ruin through a slow process of erosion of the earthen walls and domes. Only the structures on the rock outcropping continued to be used and maintained as a military base ,until vacated under orders from Reza Shah following the demise of the Qajar Dynasty in 1925 (Figure 3). Figure 3: Late Qajar Period (19 th Century) view showing soldiers in the inner citadel. Beginning in 1953, the site became recognized as a nationally significant historic site and a gradual process of conservation and restoration began. Most of the restoration work has been carried out over the past 25 years. Some of the ruins in the shadow of the military citadel were restored back into complete buildings. The final step in this restoration process was to plaster the exterior surfaces with a layer of mud plaster reinforced with straw. Most of this modern-day restoration work appears to have been done with square sun-dried bricks, rather than in chineh. Collapse From The Inside-Out 3 THE BAM EARTHQUAKE OF DECEMBER 26, 2003 The earthquake that visited upon Bam in the early morning of December 26, 2003, struck with a suddenness and ferocity that bore little relation to its comparatively modest magnitude of 6.5. There have been few past earthquakes to prepare one for the extent of the destruction seen both in the Arg and in the modern town adjacent to it. There was hardly a single building type, ancient or modern, that did not suffer total destruction. Even many of the steel frame buildings constructed over the last decade ended up with their steel frames wrapped into shapes like pretzels on top of heaps of crumpled infill masonry walls and floors. In the case of the ancient Arg, little remained inside its walls of what had been whole buildings that had been restored over the previous halfcentury. A sea of rubble extended out as far as the eye could see (figure 4 & 13). Even the Governor’s House and Tower astride the hill that formed the central symbolic image for the site disappeared, leaving behind ruins that resembled a natural rock outcropping, untouched by human hands (Figure 1). (See Figure 25 for location of sites) Figure 4: View of the ruins of the Arg-e Bam. Many engineers and seismologists have pointed to the intensity of the Bam earthquake itself as sufficient to explain much of the damage to earthen structures. The seismograph records show that the vertical component of the vibrations near the site of the Arg was greater than the horizontal component, reaching a level of almost 1g. With such intense vertical vibration, the loads on the earthen walls were rapidly cycled from a loss of their overburden weight to having to sustain double that weight. Because, in general, building s are designed to carry well more than their own weight, the vertical earthquake forces generally have not been considered to be as dangerous as the lateral forces. However, for earthen construction, when the overburden weight on the walls is reduced or eliminated, the lateral forces can be far more destructive than if there were only lateral motion. In Collapse From The Inside-Out 4 addition, with vertical forces of almost 1g, the ancient walls were forced to sustain almost double their weight with each cycle. As will be described below, the degraded state of the inner cores of some of these walls may simply have been unable to sustain this momentary additional weight. Even so, the extent of the collapses in the Arg was greater than one would have expected. There was almost no mid dle ground. With the notable exceptions described below, almost every structure suffered partial or total collapse into formless piles of rubble. Of those parts that did survive; some structures came through with remarkably little damage. What occurred to cause all of this damage? Is it explained by the intensity of the shaking alone? In a comprehensive ten-year research project on the seismic behavior and protection of historic adobe building by the Getty Conservation Institute, the researchers concluded that “It is often assumed that an unreinforced masonry structure (such as adobe or brick) is safe only while it is largely undamaged, that is, if it has not sustained substantial cracking. The usual analysis assumes that once cracks have developed the materials have lost strength and continuity – and therefore the building is unsafe. However, a thick-walled adobe building is not unstable after cracks have fully developed, and the building still retains considerable stability characteristics even in that state.” Since it took only a little over 10 seconds for the earthquake not only to crack, but to collapse much of the Arg-e Bam, the Getty project’s important findings on adobe structures clearly were not applicable to this site in this earthquake. Why did the walls and vaults of the Arg prove to be so unstable? Shouldn’t the structures and ramparts, with their thick earthen walls, have remained standing, even if heavily cracked? Were they simply overwhelmed by the unusually large surface shaking for a 6.5 earthquake, or is this now an unsettling exception to the Getty Seismic Adobe Project’s findings that disproves the rule? In either case, does this mean that the rest of Iran’s most celebrated monuments, many of which are largely constructed of unfired earth, will eventually suffer the same fate? In spite of the extent of destruction in the Arg (as the ancient walled -city is called), it is important to correct one miss-impression that has served to call into question the safe continued use of buildings of unfired earthen construction throughout the world . The news accounts that spread around the world gave the impression that tens of thousands of people died in ancient mud buildings. Instead, almost all of the 30,000 who died in the earthquak e were in buildings that were less than thirty years old. For five decades prior to the earthquake, the Arg was an archeological museum. At the time of the earthquake, which occurred at 5:26 am, only three people were sleeping in the Arg complex. The two guards sleeping in the gatehouse were killed, but the chief conservator, who was sleeping in the archeology office in the Arg, was rescued from under the rubble. Had the earthquake happened during the daytime, there undoubtedly would have been more fatalities in the Arg, but it is important to note that the massive loss of life from this earthquake was almost entirely the result of the collapse of modern buildings – buildings constructed not only of unfired earthen masonry, but also of steel and concrete. The destruction of earthen and masonry structures in large earthquakes is often accepted by observers as inevitable. Thus , inquiry into the causes of such destruction often stops with the analysis of the lateral forces measured against the capacity of the unreinforced earthen structures without consideration of other factors such as pre-existing pathologies. Yet one important anomaly in the damage distribution in the Arg-e Bam flies in the face of this common assumption – those structures that had not been recently maintained or restored survived with significantly less damage than did those that had been restored and even strengthened in recent years. (Figures 2, 5, 11, 18, 22) Collapse From The Inside-Out 5 Figure 5: This unrestored ancient earthen structure , the Shahrbast Wall, was only lightly damaged by the earthquake despite the height of its unbuttressed earthen walls. Earthquake damage to this archeological ruin is evident only where there is debris on the ground. Notice the height of the walls compared to the figure in the middle distance. TERMITE DAMAGE IN THE ARG During the four months that followed the December 26, 2003 earthquake that destroyed much of the Iranian desert city of Bam, much has been said in the international press about the damage to the Arg-e Bam. Nowhere in this coverage, however, was there any mention about termites. While attending the International Workshop on Bam sponsored by UNESCO, ICOMOS, and the Iranian Cultural Heritage Organization (ICHO) I noticed evidence of an insect infestation in the broken remains of the city’s walls. The Iranian archeologists working on the site identified the insects as termites, explaining to me that such termites are relatively common in Iran. In contrast to the archeologists working at the site, few conservation architects or engineers who had come to Bam to observe the damage during the preceding four months that followed the earthquake either noticed or were aware of the existence of the extensive termite infestation of the Arg. While there is no question but that termites did not cause the destruction of the historic Arg-e Bam, the evidence of extensive infestation in the ancient earthen monument was unmistakable (Figure 6 ). This raises the following question: Did this infestation contribute to the extraordinarily large amount of earthquake damage? While it took only 12 seconds for the earthquake to shake this majestic monument down into formless piles of rubble, the seeds of its destruction in this earthquake may have been laid over the many centuries of continuous erosion, decay, and rebuilding that have taken place on the site. Collapse From The Inside-Out 6 Figure 6: Evidence of termite damage in a wall of the Arg-e Bam showing extensive deposits of frass with insect tunnels. This question of the importance of the termites as a factor in the earthquake induced collapses was debated during the International Workshop on Bam, without arriving at a single consensus. The observation and discussion was, however, valuable in that awareness of this particular pathology has now been increased throughout Iran. In terms of the analysis of the causes of the collapses in the Arg, the discovery of the termites, did serve to focus my own attention onto the nature of the earthen material itself . No longer did it seem adequate to treat this material as a generic material which, in engineering terms, can be assumed to have a certain strength and structural capacity. The earth in the Arg may have been in a seriously degraded state going into the earthquake – a degradation that was largely hidden from view by the subsequent layers of new material and surface earth plastering from the recent restorations – until the earthquake dramatically brought it to attention. The following observations on the damage to the Arg were made over a brief two-day series of visits to the Arg, during a seven-day period in April 2004 when the International Workshop on Bam was held. The following explanations of the causes of the damage are hypotheses based on this rapid survey. Definitive determinations on all of the causes of the damage could not be done during such a short visit, but it is hoped that these observations can help to define areas for further research. DAMAGE TO THE ARG-E BAM At first view, the damage to the Arg is so extensive as to defy any attempt to classify or interpret it. The structures were pulverized, often leaving only mounds of rubble at the base of a few remai ning standing walls and piers. Few of the walls survived to their pre-earthquake height, and many of those structures that had been fully restored back into buildings were returned to a ruined state with less remaining standing than had existed prior to the last fifty years of restoration work. Collapse From The Inside-Out 7 Figure 7: The same view of the walls and towers of the inner citadel before and after the earthquake. Before photo by Mary Hardy, 2003. After an exploration of the site, some patterns in the damage began to emerge. These included the following: (1) the circular structures, such as the turrets on the ramparts, fared worse than the long straight walls and rectangular structures (Figure 7); (2) the Governor’s House and other structures on the top of the hill were more completely destroyed than were the structures lower down the hill (Figure 1); (3) almost every structure in the Arg that remained standing showed evidence of the onset of damage through the spreading to their walls from the inside-out as evidenced by the preponderance of vertical cracks (Figures 8, 17, 18); (4) most of the earthen masonry domes and vaults in the complex, many of which had been rebuilt in the late 20 th century, collapsed. The largest dome in the complex on the icehouse, a structure that was outside of the walled town that had been converted to an auditorium, collapsed as if punched in. Collapse From The Inside-Out 8 Figure 8: Piers in a partially collapsed section of the Caravansary showing the bursting of the outer layers from internal expansion from the earthquake vibrations. With regard to interesting examples of surviving structures, one could not help but notice the following: (5) a brick reconstruction of a structure with internal vaults over an ancient water cistern in the center of the stables courtyard (figures 9 & 10) survived with no evidence of even so much as a crack from the earthquake. In aerial photographs taken in 1974, the cistern was uncovered and the current structure is a recent reconstruction in modern fired brick masonry. (6) The outer ramparts on the south, east and west sides of the walled city suffered a great deal of damage, with the loss of their projecting turrets and complete destruction of the top crenellations and walkway, yet the north facing ramparts survived in better condition (figure 18). (7) In the structure known as the “Small Caravansary,” the second level of the side that had a series of buttresses along the outside wall collapsed, whereas the opposite side, which had no buttresses, survived almost intact (Figure 19). Figure 9&10: Before (November 2003) and After (April 2004) of the same view of the Stables courtyard showing the superstructure over the cistern that was recently reconstructed in fired bricks. Before photo by James Conlon, 2003. Most intriguing and significant, perhaps, are (8), those structures that had been maintained and repeatedly modified and expanded over time (such as the structures of the inner citadel) and those structures that had been partially or wholly strengthened and restored during the late 20 century (such as the outer ramparts and buildings of the lower town) fared significantly worse than did those ancient structures – both inside and outside of the Arg – that had not been maintained, modified or restored. Collapse From The Inside-Out 9 Figure 11: LEFT: Unrestored ruins in the Konari neighborhood of the Arg-e Bam that survived the earthquake with comparatively little damage. RIGHT: The Shahrbast Wall from above (see also Figure 5). The unmodified and restored structures included most of those in the north-west section of the walled town known as the “Konari” neighborhood, and also those structures just outside of the Arg to the north-east including the tall “Shahrbast Wall,” (figures 5, 11) located near the icehouse, and the “Khale Dokhtar,” (figure 22) located on the opposite riverbank to the north. Some of these surviving unrestored structures are of considerable size and height, and were undoubtedly subjected to shaking of close to the same characteristics as the rest of the Arg, but they remained standing, except for some smaller parts that broke off. (Even in these few collapsed sections in the Khale Dokhtar and other structures, termites were also in evidence.) The question that presented itself after these observations was: Is there any single condition that can explain all of these phenomena? During the brief study of the site, two unrelated experiences have contributed to my assessment of what may have caused so much damage, in addition to the high frequency vertical earthquake vibrations. One was the discovery of the termite infestations, and the second was the chance experience of the largest aftershock to be felt at the site in many weeks. The aftershock, 3.8 on the Richter Scale, rolled through the site at 7:10 a.m. on the 20 th of April. Fortunately that was a day that a small group of us had visited the site shortly after dawn. Standing in the middle of the Arg, the aftershock was felt as a high frequency vertical vibration. It can be described as being like standing on a platform above an engine that was just starting up, but not firing on all cylinders. It lasted only for about four or five seconds. A small amount of dust rose from the complex, but no further damage was sustained. This vibration was at the opposite end of the spectrum from the kind of earthquake that had, for example, affected Mexico City in 1985 or even San Francisco in 1989. Emanating from directly below the site, rather than from some distance away, the waves caused vertical shaking and vibrated at a high frequency. The earthquake records from the one instrument that was in Bam that was located near the site of the Arg recorded strong vertical vibrations of about 10 hz (cycles per second). Strong high-frequency vertical shaking alone is capable of causing extensive damage to load-bearing earthen and masonry structures, but there had to be a plausible explanation for the counter-intuitive observation that the unrestored parts of the complex did better than those that had been strengthened and restored. That is where the issue of the termites enters into the picture. After first noticing the termite bore holes and frass in a damaged section of one surviving rampart wall in the center of the Arg, I began looking for further evidence of the insects. Selecting walls at random, I looked to see if similar insect evidence could be found on other broken surfaces. In every instance, insect damage was in evidence on each of the newly exposed inner surfaces that had been broken open by the earthquake. This evidence consisted of both tunnels into the still standing portion of the walls, and large amounts of frass on the interface between the fallen and standing portions. The earth itself in these areas was extremely friable. There was evidence that the surfaces Collapse From The Inside-Out 10 between many of the fallen and standing portions of walls had been the interface between earlier and later work. This interface had contained many channels left by the insects that gave access to those tunnels that drove deeper into the (usually) older material that was still standing. (figures 2 & 12) I have learned since that termites were also common in dwellings in the city, even to the extent that people have avoided using wood for kitchen cabinets and other fittings. Since the earthquake, the termites have been found to have consumed the wood of the front doors and woodwork of the citadel that became buried when the front towers collapsed during the earthquake. (Figure 12) Figure 12: The Arg-e Bam archeologists, conservators and technicians examining the termite damage on the doors to the Arg recently exhumed from the collapsed tower. Photo courtesy of Mehrdad Hejazi. Termites live in earth and feed on organic material – that is, the same kind of cellulose that is frequently used to reinforce adobe bricks and the earth stucco used in earthen construction. Thus, the concentration of termite passageways in the interface between newer and older construction appeared to have weakened and separated the different layers of construction. If further research does prove that the termites were concentrated in the interface between zones of construction of different periods, it can explain why the later construction tended to fall off of the older cores of the walls. In addition, once they have perforated the matrix of the earthen wall, the termite tunnels may have contributed to the further drying out of the earth itself, with a commensurate loss of cohesion that comes from an excessive drying out of the clay. Collapse From The Inside-Out 11 COLLAPSE FROM THE INSIDE-OUT The termites are only a part of the larger problem of the internal degradation of the walls, but seeing how pervasive the insect tunnels were throughout the ruins did alert me to consider the possibility that the many of the collapses in the Arg may have initiated from failures deep inside the thick walls. As I explored the ruins of the still impressive earthen complex, it was, of course, difficult to come up with a single theory that could explain the nature and extent of the damage. I had expected to experience the kind of damage described by the Getty Seismic Adobe Project with the classic signatures of structural weaknesses inscribed on them: shear “X” cracks, cracks propagating out from the tops of windows and doors, collapsed corners, overturned walls, etc. However, in the Arg, even the usually common diagonal or “X” cracks were relatively rare. It appeared as if the structures had exploded from the inside and crumbled straight to the ground in a scatter of small pieces. Rubble was everywhere. It formed a mat of broken material that in some places was higher than the still-standing remains of the walls and piers. The previously completely restored Grand Mosque, for example, was completely unrecognizable after the earthquake. In the rubble pile, there was even barely enough left intact to discern the outline of what had been its large courtyard (Figure 13). It was only after having taken in all of the evidence seen in four short visits to the site over a six -day period that a pattern began to emerge. First it became apparent that walls did not crack into a series of larger sections that could rock back and forth as the Getty project had predicted based on the adobe buildings they had studied. Instead, the Arg buildings appeared to have responded to the high-frequency vibrations like unconsolidated earth fill. The study of the situation thus seemed to require a change of discipline – from structural engineering to soil dynamics. Figure 13: Ruins of the formerly impressive four-sided courtyard of the Main M osque, which has completely disappeared from view, leaving only the ritual washing place as the only thing still identifiable. Collapse From The Inside-Out 12 The more we examined the site, the more compelling this explanation became. In a number of locations there was evidence of lateral spreading of the kind that one would expect to find around the shore of a lake – but in this case it was located on dry ground where historically the surface had been built up to create the level terraces on which the buildings were constructed on the lower hillside. One section of the terrace that supported a building above the stables collapsed altogether, carrying away the front half of the rooms constructed on it (Figure 14). The round turrets appeared to have failed at the bottom, instead of splitting apart at the top as one might have expected. Their seemingly strong walls were simply pushed out at the base, with sections of the upper walls having slid down the rubble with the upper ornamentation remaining as the only recognizable pieces left. As the earthen core in a wall loses its cohesion, then the loose sand and clay material can settle. The spreading forces of this settling earthen material are sufficient to blow out the containing exterior surfaces, leading to the collapse of the wall. Figure 14: Buildings above the Stables courtyard that collapsed from the lateral spreading failure of the retaining wall and fill beneath. The aftershock at 7:10AM on the 20 of April provided a palpable sense of what had happened during the main shock. As the records did show, the main earthquake on December 26 th was recorded as having a high frequency vibration, particularly in the vertical direction. From the sensation felt during the aftershock, it did feel like the kind of vibration that could cause soil subsidence – much the same way that a vibrator causes freshly-placed wet concrete to flow . The experience then began to explain each of the seemingly disparate and sometimes counter-intuitive phenomena described in the list above. Collapse From The Inside-Out 13 For example, in the case of (1) the first observation, the particular vulnerability of the circular turrets may be explained by the fact that they had contained large amounts of unconsolidated fill in their bases. One of the few that survived is to the right of the 2 gate (Figure 15), and it has a timber floor diaphragm with timbers penetrating the walls located beneath the upper windows, with a room, rather than solid fill, below. The absence of the fill, combined with the effectiveness of the floor diaphragm may have been instrumental in holding it together. (Figure 15) Figure 15: The one of the few surviving turret on the rampart wall of the inner citadel visible on the right has a room in its base. The second floor diaphragm timbers extend through the walls just below the window. By contrast, the turret on the left that has collapsed was filled solid below the level of the room that is visible at the upper level. In the case of (2) the collapse of the structures at the top of the hill (Figure 1), this seemed to be caused in part because the failure of the retaining walls and fill that had been constructed up from the lower hillside to widen the platform at the top of what had originally been a narrow rock outcropping. (A similar failure can be seen in Figure 14.) The failure of (3) the walls of many of the buildings and ramparts was also consistent with the lateral spreading of the material in the cores of their walls, with the exterior adobe bricks being forced out as manifested by the greater frequency of vertical cracks as compared to diagonal cracks (Figure 8, 16 & 17). In the case of (4), the collapse of the domes throughout the complex, many simply may have followed their bursting supporting walls to the ground. Others which collapsed inward, like that of the Icehouse, suffered from the effect of the intense vertical vibrations on the soft adobe brick masonry. The momentary doubling of the weight of the domed structures was probably more than they could handle. In the case of the Icehouse, the 1974 aerial photographs provide evidence that the part of the dome that did collapse had been reconstructed after that date, as it was missing in those photographs. 7 Collapse From The Inside-Out 14 Figure 16: Collapsed round turret on the ramparts at the main entrance to the Arg shows evidence of having burst apart from collapse of internal layers that are composed of construction of different periods. By contrast to the bursting walls (5), the masonry structure over the cistern in the center of the stables courtyard performed much better despite being of unreinforced masonry. (Figure 9 & 10) Most likely in this case, the walls were of a uniformly solid and bonded masonry construction without a rubble core. The fact that it was constructed of fired brick would have contributed to its strength, but what may even have been more importa nt was the fact that the walls were of a fully cohesive material of uniform density without voids or vertical gaps. Figure 17: Another view of a collapsed section of at the main entrance where the surface of earlier construct ion has been exposed underneath. Note the multitude of vertical cracks – evidence of spreading forces on the exterior layers from the compression of the internal solid clay fill. Collapse From The Inside-Out 15 As for the better behavior of the north-facing ramparts compared to the other city walls (6), the subsurface soil conditions along the riverbank where the wall is located may account for some of this difference because the alluvial soils may have served to damp out some of the vibrations, rather than intensify them, as is often the case with earthquakes where the epicenter is farther away. Further research is needed to determine whether this may be one explanation. Also , like the nearby Konari neighborhood, these walls had not been altered and restored along their tops as much as had the other walls around the Arg. It was the newer restored upper level battlements, walkways and crenellations that consistently suffered the most, possibly because of the different densities and vibration response of the new and old material and the additional weight. (Figure 18) Figure 18: LEFT: Ramparts on the West side of the Arg showing the spreading failure of the turret and the collapse of the crumbling of the walls. RIGHT: The North-facing ramparts, which were significantly less damaged in the earthquake than the other city walls. Notice that the crenellations are still intact on this one section, the only section where that was observed to be the case. The second-to-last item (7) is the Caravansary , where the rooms on the second level of the buttressed west side of the complex collapsed, leaving the east side that lacked buttresses largely intact. The buttresses themselves were also damaged, with one collapsing from the crushing of its base. (Figure 19) The story of this complex became even more interesting when I learned from the old photos that the side that collapsed had been almost completely reconstructed only a few years earlier, whereas the still standing side had mostly survived from antiquity. In aerial photographs of the caravansary taken in 1974, the domes on the east side were almost completely intact, whereas on the west side they had collapsed. At that time, the buttresses on the west side only extended up to the level of the first floor. As late as 1996 the condition of both sides was similar to 1974, except that the small holes in the east side domes had been fully repaired. At the time of the earthquake, photographs show that the restoration of the west side of the Caravansary had been completed. The domes had been reconstructed and the west wall and buttresses had been extended up to the roof level. Ironically, in the earthquake it was this newly constructed and fully buttressed side that fell. This was simply one more example of the finding that the areas with the greatest amount of strengthening, reconstruction, or even of continued maintenance were the most heavily damaged. Collapse From The Inside-Out 16 Figure 19: The ruins of the Small Caravansary show that the domed rooms on along the buttressed wall (UPPER LEFT) have collapsed, while the ones along the non-buttressed wall (UPPER RIGHT) survived. Some of the buttresses themselves are in a state of partial collapse. All of this evidence taken together seems to point to a phenomenon where those earthen walls that are composed of material of different densities and construction characteristics resulting from their different phases of construction, repair and reconstruction, proved to be more vulnerable to the earthquake vibrations. As long as one perceives of the earthen construction as having a uniform composition, it is difficult to understand why the strengthened and restored walls would fare worse than the unrestored and naturally eroded walls. However, the succeed ing phases of construction in the Arg over the centuries had produced walls of a very different composition than that of a newly constructed earthen building. No longer did many of these walls consist of horizontal layers of earth or sun-dried bricks, and those that did consistently appeared to be less damaged. Instead, through generations of erosion, repair and remodeling, many of the walls had evolved into a series of vertical layers of earth, standing together like books on a shelf without bookends. Each of the different layers was of a different density and cohesion resulting from the different ages, construction characteristics, and degradation. For example, modern Khesht (adobe masonry) frequently encased older chineh (cob) construction (Figure 17), and the organic material used for reinforcement had rotted or been consumed by insects, leaving cavities and friable earth (Figure 12). Collapse From The Inside-Out 17 Figure 20: View of a surviving damaged section of wall showing the cracks and separation between different types and vintages of construction that make up the same wall. Further research is needed on this subject, but it was only after I began to interpret what I saw at the site as the behavior of vertically disconnected unconsolidated earth, rather than as the uniform horizontally bedded earthen construction of the sort analyzed for the Getty project, that a possible explanation for the nature and extent of the earthquake damage began to emerge. Based on the Getty research, the thick walls of the ramparts and main citadel in the Arg of Khesht or chineh construction would normally be expected to be the most resistant, rather than the most vulnerable as they turned out to be. If the hidden interior parts of the walls are composed of a series of vertical segments, and especially if the inner segments had large voids, crevices, and dried -out unconsolidated fill that lacked connections to the outer layers, the high frequency vibrations of this earthquake could cause the inner older and more degraded portions of the walls to settle. This then could exert a horizontal force from the inside-out onto the outer layers at the base of the structures, causing the walls to collapse, not by tipping over, but by crumbling in place. In addition, variations in the density and cohesion in the earthen layers in a wall, particularly resulting from different periods of erosion and reconstruction and changes from chineh to Khesht quite possibly can cause the earthquake vibrations – particularly vibrations in the high frequency range experienced in Bam – to ricochet off of the layers of different densities, causing the onset of damage from the local intensification of the vibrations. Evidence of such behavior can be seen in (figure 20) where the wall began to break out at the interface between two vintages of Khesht construction, both of which appear to be pre-20 th century. This is a subject in need of further research specific to earthen construction in order to establish if this phenomenon can be explained in this way, but such research may have a significant bearing on the protection of other earthen monuments that have been altered over the centuries. The North and South American adobe structures that were the primary subjects for the Getty research are less likely to be subject to this Collapse From The Inside-Out 18 phenomenon because they are not of such great age, and thus their walls are more uniformly of horizontally bedded adobe bricks without the many stratifications of different construction in the walls. This is why the observation of the widespread infestation by termites may turn out to be important. Not only did it appear that the ancient construction in the Arg was perforated by the insects, but that the insects had also succeeded in separating the different vertical stratif ications in the walls, and reducing the cohesion of the inner core of earthen material. In a trip to Isfahan after the mission to Bam, during meetings with the professional restorers of several of the historic monuments in that splendid city (figure 21), I learned that termites were also found in the walls during the recent restorations. These professional conservators explained that the damage caused by the termites had to be addressed in the restorations by consolidating the earthen cores of some of the walls. Figure 21: The Imam Mosque in Isfahan, April, 2004. The inner cores of many of these walls and the walls of other great monuments in Isfahan are constructed of unfired clay. In contrast to this strengthening work in Isfahan, some of the 20 th century restoration work seen in Bam may have aggravated the problem. Clay stucco added before the earthquake was reinforced with copious amounts of straw – a material that appeared in many areas to have been consumed by termites. By contrast, the older reinforcement of shredded date palm tree bark appeared to have been more resistant. Perhaps the termite population inadvertently has been increased in modern times, simply because of this “banquet” of non-resistant straw-reinforced stucco. 11 THE RISK TO EARTHEN MONUMENTS What many people do not realize is that new buildings in any country constructed of modern materials to code are not designed to withstand major earthquakes without damage. For earthen structures, elastic analysis procedures provide little guidance on how such buildings will behave in the post-elastic range. Quoting again from the Getty Seismic Adobe Project report: “The sole use of an elastic approach can be justified only when there is a known relationship between the level at which yielding first Collapse From The Inside-Out 19 occurs and the level at which the structure collapses. In the case of thick -walled adobe construction, there is no clear relationship between these two events. ... While a strength-based analysis can accurately predict when cracks will occur, it cannot provide insight into the post-elastic performance of adobe buildings.” This report makes a very important distinction between what is described as a “strength-based” approach and a “stability-based” approach to seismic upgrading. The report identifies that the current “conventional engineering approach to seismic retrofitting” is a strength-based approach, which is based on increasing the elastic strength of a building’s structural system. They go on to explain that for adobe buildings, a “stability-based approach” is more suitable. The objective in stability-based design is to ensure that the structure remains standing long after the elastic range of its structural system has been exceeded. Since the elastic capacity of adobe masonry is low, seismic hazard mitigation of adobe structures depends on maintaining its stability long after it begins to crack. The Getty Report then goes on to describe the potential that adobe buildings have for “structural ductility” even though they lack “material ductility.” The structural ductility can come from the inherent stability that even the cracked adobe walls can have so long as the cracked wall sections remain bearing one on another. This is an important finding that can be used as effective basis for design for many adobe structures that otherwise would be condemned. However, in retrospect, what the Bam earthquake has proved is that, in spite of the thickness of the ancient earthen walls in the Arg, stability was quickly lost after the elastic range was exceeded. The earthquake on 26 th December 2003 was recorded to have lasted only 12 seconds, so the collapse of the Arg was almost instantaneous. Structural ductility thus was not to be found in the structures of the Arg. Who could have ever known there was such a risk? The interiors of the walls with their voids and degraded materials were hidden. An engineer doing a structural analysis would normally have based the analysis on measurements of the thick walls without anticipating the effect of the fractured and weakened conditions of their internal cores. Research is now needed to find ways to be able to evaluate such walls as non-destructively as possible, and then to find ways to address the kinds of problems that may be discovered that would lead to their rapid loss of structural cohesion when subjected to earthquake vibrations. WHY DID THE ARG -E BAM PROVE TO BE SO VULNERABLE? If the changes to the walls of the Arg over the centuries, and during recent restorations can be proven to be a major part of the cause of its wide-spread collapse, then there are two remaining questions that need to be asked. The first is: Since the Arg is located in a known earthquake region, why was its construction not more responsive to the threat? Before answering this question, one must ask whether or not the original construction prior to its alteration and degradation over time was in fact designed to be more resistant. These are both particularly difficult questions because the subject of the inquiry is archaic construction using a limited palate of materials – namely unfired earth with a small number of undressed timber logs – not something that provides much opportunity for modification to resist large earthquake forces. Earthquake resistance is not an absolute. One must alter one’s expectations to reflect what could be achieved in a culture where only unfired earth and a limited amount of timber were available. In the present, expectations of earthquake safety have been shaped by the existence of steel, either as the structural building material itself, or imbedded into concrete, or even when used to reinforce the connections between timbers. The frequent catastrophic failures of modern buildings of steel and concrete, as we have seen in Bam, in the case of steel, and other recent earthquakes in the case of reinforced concrete, notwithstanding, the existence of steel as a building material has raised expectations, making it harder now to recognize pre-modern mitigation efforts. Collapse From The Inside-Out 20 Traditional construction – particularly in a desert environment – did not have the luxury of an abundance of timber, much less access to modern steel. Much of what could be done was the product of trial and error leading to the evolution of building practices over many centuries. Thus, out of all of the influences on the evolution of construction practices, it is not easy to discern those specifically that are a response to earthquakes. Earthquakes themselves are not always the same. The December 26, 2003 earthquake in Bam was only 6.5, but it was a shallow earthquake with its epicenter located almost directly beneath the Arg-e Bam. The likelihood of being directly above the epicenter of an earthquake is significantly less than being nearby. Thus, it is entirely possible that the Arg had never been subjected to such a high frequency vertical vibration over the prior 2000 years of its history. The earthquake risk most often analyzed by computing the static equivalent forces on structures, but in this case, the frequency of the vibrations may have been particularly significant. Even a small difference in the location of the epicenter, would have shifted the ground shaking under the Arg away from such high frequency vertical vibrations to a low er frequency vibration with a smaller vertical component. How this may have affected the spectrum of damage in the Arg is difficult to determine, but important to know because it can shed light on the degree to which other monuments are at risk. If high frequency is a large part of the cause of the damage, the odds of a recurrence under other monuments is less than if a broad range of frequencies will cause similar damage.