FIRE and MUD Contents

Building Damage Caused by the Mount Pinatubo Eruption of June 15, 1991

By Robin J.S. Spence,¹ Antonios Pomonis,² Peter J. Baxter,³ Andrew W. Coburn,² Mark White,⁴ Manuel Dayrit,⁴ and Field Epidemiology Training Program Team⁴

¹Department of Architecture, University of Cambridge.

²Cambridge Architectural Research Ltd., The Eden Centre, 47 City Road, Cambridge CB1 1DP, U.K.

³Department of Community Medicine, University of Cambridge, Institute of Public Health, Forvie Site, Robinson Way, Cambridge CB2 2SR, U.K.

⁴Department of Health, Manila, Philippines.

ABSTRACT

This paper presents the results of a survey of the building damage caused by the ash fall from the cataclysmic eruption of Mount Pinatubo. The survey is a damage assessment based on a scale of damage derived from earthquake-damage survey techniques; the damage was related to aspects of the building form and construction technology. Fifty-one houses in the town of Castillejos were surveyed. Of these, one-third either collapsed or were seriously damaged. It was concluded that the roofs that failed did so because the ash load was greater than their vertical load-carrying capacity and that the nature of the roof supporting structure was the principal factor influencing the level of damage sustained; other significant factors included the overall construction type and the roof pitch. The paper also discusses implications for future research on building damage caused by volcanic eruptions. It is suggested that in situ methods for measuring ash loads need to be developed and that a building classification system based on ash-fall vulnerability should be devised for use by disaster planners.

Note to readers: Figures open in separate windows. To return to the text, close the figure's window or bring the text window to the front.

INTRODUCTION

Ash fall is one of the most destructive phenomena associated with massive plinian events like the cataclysmic eruption of Mount Pinatubo on June 15, 1991. Huge quantities of tephra may be transported downwind for tens, or even hundreds, of kilometers and fall in deposits deep enough to cause severe disruption, including damage to, and collapse of, buildings. Although major loss of life from pyroclastic flows was avoided by a successful and timely evacuation of about 60,000 people living within 30 km of the volcano, as of July 17, 1991, the official numbers of casualties were 320 dead and 279 injured, and most of these casualties were believed to have been due to the collapse of roofs under the weight of wet ash on June 15 (Pinatubo Volcano Observatory Team, 1991). Of all the major eruptions in recent times the Mount Pinatubo event provided the first opportunity to study the destructiveness of tephra fallout in a densely populated area, though the impact would have been considerably less without the passage of Typhoon Yunya. We present our findings of a photographic survey of buildings in a representative area of heavy ash fall to draw the attention of emergency planners to the importance of this volcanic hazard and to the need for specific preventive measures to be devised in all areas of active explosive volcanism.

THE SURVEY

Soon after the June 15 eruption, the Field Epidemiology Training Program Team of the Philippine Department of Health carried out photographic surveys of some of the damage. Figure 1 shows typical examples from Olongapo City and the town of Subic. The most detailed survey was in Castillejos, a town with a population of less than 50,000, situated 27 km southwest of Mount Pinatubo, north of the Subic Bay area in one of the worst affected areas. The thickness of the ash fall at this point was about 20 cm (fig. 2). The field survey consisted of choosing a sector of the damaged settlement and recording the state of each building in it photographically, whether damaged or not. Between one and four photographs were taken of each building. Other details were recorded on a specially prepared record sheet. The survey was carried out on June 29th, about 2 weeks after the main eruption; at that stage little work to remove fallen roofs or repair the damage had taken place. Altogether, the Castillejos survey comprised 51 buildings.

Figure 1. Examples of ash-fall damage in Olongapo City and the town of Subic at the end of June 1991. A, Homes and commercial buildings, Olongapo. B, Roadside stores, Olongapo. C, Church with complete roof collapse, Olongapo. D, Collapsed canopy of gasoline station, Olongapo. E, Collapsed roof and upper story of a mixed building with a reinforced concrete lower story and a wooden upper story, plus concrete firewall. F, Damage to roof and eaves of school. G, Collapsed roof of Olongapo health center.
Figure 1 A-E
Figure 1 F-G

Figure 2. Isopach map of layer C, plinian tephra fall from June 15, 1991. Most of the tephra that caused roof collapse was in this event and layer. An additional tenth (roughly) of these amounts fell on June 12-14, in layers A and B. Thicknesses are in centimeters. Modified from Paladio-Melosantos and others (this volume, fig. 7).

Subsequently, the photographs and other information were sent for analysis to Cambridge, where both building typologies and damage levels for all buildings were assessed. The building type analysis included identification of:

• principal constructional materials used
• number of stories
• roof structure, shape, and pitch
• building usage (residential or nonresidential)

For damage assessment, a scale of damage categorization appropriate to ash-fall damage was needed. The MSK earthquake intensity scale (Karnik and others, 1984), which defines six levels of damage, has proven easy and reliable for postearthquake damage assessment, and it was decided to adapt this scale for use in the present survey. The six degrees of damage were defined as:

D0: No damage.

D1: Light roof damage.--Gutter damage; few tiles dislocated.

D2: Moderate roof damage.--Bending or excessive deflection of roof sheeting or purlins; no damage to principal roofing supports.

D3: Severe roof damage and some damage to vertical structure.--Severe damage or partial collapse of roof overhangs or verandahs; severe deformation of main roof sheeting; some damage to roof supporting structure, columns, trusses.

D4: Partial roof collapse and moderate damage to rest of building.--Collapse of sheeting but not truss; partial collapse of sheeting and some truss failure; failure of supporting structure; moderate damage to other parts of building resulting from roof collapse.

D5: Complete roof collapse and severe damage to the rest of the building.--Collapse of roof and supporting structure over more than 50 percent of roof area; partition walls destroyed; external walls destabilized.

CONSTRUCTION TYPOLOGIES AND USAGE

Table 1 shows the distribution of the overall construction types identified. The confined masonry buildings had concrete block walls with reinforced concrete columns in corners or near openings, but there were no confining beams. Some buildings had ground floors of masonry or reinforced concrete and an upper floor of timber frame. In the analysis of damage, these were classified as timber buildings, since it was the upper floor that was exposed to the ash fall.

Table 2 shows the analysis of roof types. In virtually all cases the roof covering material was corrugated galvanized steel sheet, of indeterminate gauge, but the supporting structure varied. In most cases roofs were of domestic scale, with short spans: roof sheets rested on purlins that were supported on closely spaced rafters running down the slope of the roof and resting on external walls or ridge beams. In a smaller number of cases, roofs had much larger spans; roof sheets were supported on more substantial purlins that were in turn supported on more widely spaced trusses spanning between walls or columns a distance of 5 m or more apart. Table 2 shows the observed distribution of the buildings between the different principal roof types. The pitch of the roof also varied; in most cases it was below 20°, but in a smaller number of cases it was between 20° and 30° (table 3).

The distribution of buildings by usage category is shown in table 4. Over half were residential buildings, and the rest a mixture.

Table 1. Distribution of buildings by construction type.

Table 2. Distribution of buildings by roof support.

Table 3. Distribution of buildings by roof pitch.

Table 4. Distribution of the buildings by usage.

DAMAGE DISTRIBUTION

The damage distribution over the whole sample is shown in table 5. The variation in the level of damage is very considerable: about one-third of the sample suffered no damage, while another one-third suffered either total or partial collapse.

It is important to try to identify the significance of the various attributes of the construction classification in contributing to damage potential. Because numbers are small, in subsequent analysis, damage has been divided into just two classes: buildings without significant damage (D0+D1) and buildings with partial or complete roof collapse (D4+D5).

Table 6 shows the distribution of damage with aspects of the building construction and usage. It shows that timber frame buildings suffered significantly more damage than reinforced concrete frame buildings, that buildings with steeper roof pitches suffered worse than those with shallower pitches, and that nonresidential buildings were much more vulnerable than residential ones. The most significant indicator of damage, however, was the roof supporting structure. Only 16 percent of the short-span roofs suffered major damage, with 43 percent having no significant damage; while of the long-span roofs, 75 percent were severely damaged, and only 17 percent were without significant damage. Long-span roofs were, therefore, nearly 5 times more likely to suffer major damage than short-span roofs.

Table 5. Damage distribution of the whole sample.

[D0, no damage; D1, light roof damage; D2, moderate roof damage; D3, severe roof damage and some damage to vertical structure; D4, partial roof collapse and moderate damage to rest of building; D5, complete roof collapse and severe damage to the rest of the building]

Table 6. Damage distribution by characteristics of construction and usage.

SUMMARY OF DAMAGE

1. Although many roofs had been cleared by the time of the survey, there was evidence from the uncollapsed roofs that the wet ash was able to accumulate to depths of at least 15 cm on metal sheet roofs of pitch up to 25°, without slippage.
2. Out of the total sample of 51 buildings in Castillejos, 17 suffered partial or complete roof damage, while 18 suffered no damage or only light damage.
3. Buildings tended to suffer worse damage if they were (a) constructed with long-span roofs (greater than 5 m clear span), rather than with short-span domestic scale construction, (b) of timber frame rather than reinforced concrete frame construction, (c) of higher rather than lower roof pitch, or (d) non-residential rather than residential.
4. Some other factors that seem to have contributed to damage, though statistical evidence is inadequate to demonstrate their significance are (a) unbraced supporting walls or columns, and (b) large unsupported roof overhangs.

The principal cause of the damage was that the load of ash on the roof exceeded the strength either of the roof sheets or of the roof supporting structure, or both. This is to be expected, because the load of 15-20 cm of water-saturated ash on the roof would probably have exceeded 2.0 kN/m² (approximately 200 kg/m²), whereas typical design loads for pitched roofs in tropical cyclone areas would rarely approach this level even for engineered structures, and they would be designed principally to resist wind suction forces. Some buildings, by virtue of their form of construction, had a better inherent resistance than others, and this explains in part the large variation in performance, although some roofs may also have been cleared during the first 2 days of the eruption. In particular, it appears that the spacing and span of the principal roof-supporting members was critical. Domestic-scale roofs supported on closely spaced rafters seemed to have a much greater reserve of strength than the longer span roofs. This is perhaps because the longer span roofs would have been designed more precisely for the expected wind and maintenance load. It is not clear why roofs with steeper pitch were more severely damaged, but it is worth noting that none had a slope great enough to overcome the frictional resistance of the roof sheeting material.

The principal design recommendation arising from this study is that, to protect lives, roofs of buildings exposed to possible ash fall should be designed for a superimposed load related to the probable level of ash fall, in a manner analogous to design for snow loading in cold climates. In the case of Castillejos, a design load of 2.0 kN/m² would almost certainly have been sufficient to avoid collapse during the 1991 eruption.

To our knowledge, this study is the first published attempt to investigate the impact of heavy ash fall on buildings, and its deficiencies highlight the need for more comprehensive scientific evaluations to be undertaken in future eruptions to enable disaster reduction measures to be developed as one of the goals of the International Decade for Natural Disaster Reduction (IDNDR). The results of surveys of building damage need to be linked to epidemiological surveys designed to define the size of the population at risk in the buildings at the time of the ash fall, the numbers of fatalities, and the numbers and types of injuries encountered. Autopsy studies will be needed to clarify the causes of death, such as severe trauma, asphyxia from inhaling ash, or a combination of the two. Unfortunately, these data were not collected at the time of our survey. Conventional advice to protect property against ash falls is to shovel or sweep by hand as a way of removing ash from roofs, as was done in the eruption of Heimaey, Iceland in 1973 (UNDRO, 1985). Such advice is difficult and risky to follow during a Pinatubo-type eruption. With the heavy fallout of tephra combined with rain and lightning, most people were too afraid to venture onto their roofs, and those who did might have found the ash difficult to shovel and impossible to sweep. Also, experience at Cerro Negro, Nicaragua, and elsewhere has suggested that the risks of falls, death, and injury from working on ash-laden roofs might exceed the risk from spontaneous collapse, so the conventional wisdom about sweeping must be tempered by judgment in any specific circumstances. Some residents reported to one of us how, while sheltering in their houses during the night of June 15, they listened, terrified, to the sound around them of roofs breaking and the accumulated ash sliding inside. Recognition that most of the lightning was intracloud lightning might have eased fears, but probably not by much.

IMPLICATIONS FOR FUTURE RESEARCH

One of the weaknesses of the present survey is that no measurement of the loading actually imposed by the ash on the roofs was possible. It is known that the specific weight of dry ash can vary from 400 to 700 kg/m³, and that rainwater can increase this by 50-100 percent if the ash becomes saturated (Blong, 1981). Although there was heavy rainfall during the eruption of June 15, the extent to which the ash was saturated, and whether this caused compaction, which would reduce the thickness of the layer, is not recorded. Ash layers of at least 15 cm are certainly observable on some of the roofs in Castillejos that survived the ash fall. In future building surveys, it will be essential to measure not just the thickness, but the insitu specific weight of the ash. A small portable device for this purpose needs to be designed.

Design recommendations such as those given above may help to avoid future life loss in buildings designed in accordance with them. But some assessment of the probable losses of existing settlements at risk, in the Philippines and elsewhere, is also needed to alert civil protection authorities to the scale of evacuation and other protection measures that may be needed. For this purpose, it will be necessary to develop some understanding of the relation between ash-fall loading and the extent of damage, in a manner analogous to the vulnerability functions that have been developed for assessing earthquake damage (Spence and others, 1992). Ideally, such relationships would be built on the basis of extensive field damage data, covering a range of different levels of ash-fall loading and a large range of building types. There is, however, no existing body of such data, and the study reported here provides just one data point for the relation between ash fall and damage level for the two principal building types used locally, so building such relations on the basis of damage data alone will be a lengthy task. However, where enough is known about the structural form and materials of a building at risk, an assessment of its response to any level of loading can be made by calculation of its resistance. It may be possible to extend this approach beyond that of a single building to estimate the damage to a whole population of buildings, if enough is known about the conventional construction technologies in use. But such a predictive approach will depend on the ability to predict future ash-fall loading intensity, not just thickness.

This paper has concentrated on the damage caused by ash fall because it is clear that roof overload was the principal cause of the failures observed in Castillejos. Some earthquake activity also took place during the period up to June 29th, and this activity could in some cases have increased the ash-fall damage, but the ground shaking recorded was not sufficient, in itself, to have caused the damage seen. However, if a predictive approach to damage assessment of buildings is to be developed, it will be important not to ignore the potential building damage from all the hazards associated with possible future eruptions, including earthquakes, tephra bombs, pyroclastic flows, and lahars. as well as ash fall. The distributions of these hazards will differ from that of ash fall, as will the characteristics of buildings that are vulnerable to each hazard. Thus, composite hazard and risk maps will need to be developed. There is a need, however, to develop simple and rapid building and settlement survey techniques to identify buildings at risk and to classify them according to their vulnerability to each of the separate hazards.

Note from the editors: Densities of Pinatubo tephra that were collected at Clark Air Base and weighed later in a laboratory ranged from 1.2 to 1.6 g/cm³ (dry) and 1.5 to 2.0 g/cm³ (wet) (G. Heiken and D. Riker, written commun., 1994). If the density of wet tephra in Castillejos, Olongapo, and Subic was approximately 1.5 g/cm³, a 20-cm thickness of that tephra would have exerted a force of 3 kN/m².

REFERENCES CITED

Blong, R.J., 1981, Some effects of tephra falls on buildings, in Self, S., and Sparks, R.S.J., eds., Tephra studies: Dordrecht, D. Reidel Publishing Co., p. 405-420.

Karnik, V., Schenkova, Z., and Schenk, V., 1984, Vulnerability and the MSK scale: Engineering Geology, v. 20, p. 161-168.

Pinatubo Volcano Observatory Team, 1991, Lessons from a major eruption: Mount Pinatubo, Philippines: Eos, Transactions, American Geophysical Union, v. 72, p. 545, 552-553, 555.

Office of the United Nations Disaster Relief Coordinator (UNDRO), 1985, Volcanic Emergency Management: New York, United Nations, 86 p.

Paladio-Melosantos, M.L., Solidum, R.U., Scott, W.E., Quiambao, R.B., Umbal, J.V., Rodolfo, K.S., Tubianosa, B.S., Delos Reyes, P.J., Alonso, R.A., and Ruelo, H.R., this volume, Tephra falls of the 1991 eruptions of Mount Pinatubo.

Spence, R.J.S., Coburn, A.W., Pomonis, A., and Sakai, S., 1992, Correlation of building damage with strong ground motion, in World Conference of Earthquake Engineering, 10th, Madrid, Spain, Proceedings, v. 1: p. 551-557.

FIRE and MUD Contents

PHIVOLCS | University of Washington Press | U.S.Geological Survey

This page is <https://pubs.usgs.gov/pinatubo/spence/>
Contact: Chris Newhall
Last updated 06.11.99