The last 12 months have seen substantial changes at the Natural Hazards Research Centre. Firstly, we have changed the name to Risk Frontiers – Natural Hazards Research Centre. While this change reflects the potential for extending the range of our research and development activities, we have no immediate intention of working on other than natural perils and their consequences.

Similarly, Risk Frontiers will maintain very close links with the insurance industry. From 1994 to mid-2001, the Centre was sponsored by Benfield Greig, Swiss Re, QBE, and Guy Carpenter. We are most appreciative of their vision in establishing the NHRC and in maintaining funding over the past seven years.

We are even more delighted that the four companies have agreed to continue their involvement for the next three years as Lead Partners – the top level of sponsorship. The “First Four” have been joined by NRMA Insurance and AON Re as Lead Partners. Employers Re, CGU Insurance, Gerling Global Group and Royal Sun Alliance have become Major Partners. We anticipate that other insurers and reinsurers will also become partners over the next few months.

In the short term (1-2 years) Risk Frontiers will continue to focus on modelling natural perils for Probable Maximum Loss estimation, risk rating and portfolio management. Our particular emphasis will be on flood and hail, perils with (frequently) underestimated consequences but where we already have market-leading models. We will also direct resources to investigations of other natural perils in Australia.

Risk Frontiers’ focus will be Australian primarily, but also Australasian – we have begun work on volcanic hazards in Auckland and will continue our interests in perils and consequences in the south Pacific and Asia.

In the longer term, Risk Frontiers will invest considerable sums in innovative technologies using spatial data analysis and remote sensing with the aim of providing new tools for portfolio and natural perils risk management. The initial emphasis of this R&D is Australian, but the broader implications are global.

Interwoven with the short and longer-term aims, we intend to work closely with our partners in providing training solutions, emphasising the ways in which Risk Frontiers’ staff investigate and assess natural perils risk.

Notwithstanding the breadth and depth of our links with the global insurance industry, we intend to strengthen ties with other research groups, partners and clients. Our strengthened financial position provides possibilities for new collaborations and new directions in our research. We would also like to pay increased attention to sustainability issues, primarily as they relate to the insurance industry but also in a broader context – our research, for example, has already identified issues relevant to building codes and land use planning.

At present, Risk Frontiers has five full-time staff, though we also employ part-time assistance and, occasionally, consultants. Our 3˝ postgraduate students, working on a range of flood and volcanic hazard problems, provide an additional strength. With our current R&D interests and new possibilities in mind, Risk Frontiers will advertise, in the near future, for a Deputy Director and an additional staff member at the post-doctoral level. We anticipate that these developments will attract additional postgraduate students.

Finally, and personally, I would like to thank again our partners for their marvelous support and encouragement – the “First Four” for the past seven years, and all ten partners for the next three years!

Russell Blong

Volcanic ashfall from the September 1994 eruption of Tavurvur and Vulcan volcanoes destroyed the southern and central parts of Rabaul town, Papua New Guinea. This report documents the effects of volcanic ashfall on timber-framed, metal-roofed residential buildings in Rabaul town.

Figure 1 illustrates the distribution of ash loads produced by the 1994 eruption. Southern parts of Rabaul town received up to 800 mm of compacted ash, equivalent to a load of as much as 16 kNm-2, while the section of the town at the northern western end of Simpson Harbour received around 50 mm (or 2-5 kNm-2). A few degrees more westerly component in the Tavurvur tephra elipse would have increased destruction in the town. Had the axis of the Tavurvur tephra elipse been a few degrees to the east, or had the eruption styles of Tavurvur and Vulcan been reversed, much of the damage to Rabaul would not have occurred.

Figure 1: Tephra loads in kNm-2,. Rabaul urban area enclosed by black line.

All buildings in Rabaul were destroyed during World War II, so that the greatest age of buildings in the town was less than 50 years. Most buildings appeared to be less than 30 years of age when the eruption occurred but there are no adequate statistics. The sample of buildings considered here is not a random sample, as it is drawn largely from those buildings that had volcano insurance cover. This suggests that the sample is biased toward the better quality buildings. Only 13% of all buildings in the sample had more than one floor. The roof of almost every building was of sheet metal, generally with a pitch of <10 degrees. Most roofs were supported on timber frames. Houses generally had timber frames, with external walls of timber or fibro (fibre/asbestos-cement sheeting). A minority of residential buildings had one or more external walls of concrete block.


Damage data has been derived from the author’s own observations, the observations of loss assessors, and the reports of structural engineers, quantity surveyors and other building professionals. Tephra thicknesses were taken from the Rabaul Volcano Observatory maps or from the author’s own observations and a Damage Index (V1 to V5) for tephra fall was developed (Table 1). A range of similar Damage Indices and Central Damage Values, developed for a variety of purposes, can be perused at

Table 1: Volcanic Damage Index

Figure 2 illustrates the relationship between total tephra load and the Damage Index for 10 residential building classes. A total of 78 buildings are included in this sample. Virtually all of the single-storey houses had similar timber roof framing structures and a corrugated iron or ribbed metal roof. However, these houses varied considerably in age, quality of maintenance, and the span of the roof from wall to wall or across internal partitions. In addition, some buildings sat on a concrete slab, while others were raised on short concrete or steel columns.

Figure 2: Total tephra load versus Damage Index for the sample of residential buildings. In the key H=house, U=apartment or unit, B=concrete block construction, HS=highset houses, and numerals indicate the number of storeys >1.

Figure 2 suggests the tephra load required to initiate damage at each level of the index increases progressively, though the difference only becomes dramatic between V4 and V5. At V1 damage to buildings is largely cosmetic with cleanup of the interior required and replacement of some guttering and/or water tanks. It is surprising that some residential buildings sustained only this level of damage despite carrying loads >4 kNm-2. Many buildings subjected to similar loads suffered much more severe damage (V3 and V4), with failure of the roof structure and damage to some external walls.

Houses that collapsed completely (V5) all experienced tephra loads greater than about 7.5 kNm-2 (equivalent to a tephra thickness ~460mm). Some collapsed houses experienced tephra loads around 15 kNm-3, but they may have collapsed long before the total load was imposed.

All two-storied residential buildings (H2, HB2, HS2) in Figure 2 are clustered around the V 5 level indicating that most of these structures were damaged beyond repair. However, it is not possible to conclude that these buildings were less resistant to tephra loads than single-storey houses. Most of the two-storey houses experienced tephra loads in the top-half of the range.

The other notable feature in Figure 2 is that residential buildings with one or more external walls of concrete block construction survived tephra loads with less damage than houses with timber framed walls. Almost all buildings with concrete block walls (HB, HB2, UB and UB3) sustained less damage. This reflects the fact that concrete block walls generally remained upright and reasonably intact, though there are a few examples where the top course of blocks was damaged or walls were deformed, no doubt as a result of roof collapse.


Most metal roofs on timber frames are of relatively lightweight construction. As Rabaul is outside the area affected by tropical cyclones, roofs are not designed for significant uplift forces, nor are roofs, walls and floors tied together in the ways recommended in modern cyclone/hurricane loading codes. It is not surprising, given this context, external walls were pulled down when roofs collapsed under a tephra load.

As Figure 2 shows, the minimum tephra loads causing damage to the roof structure was about 2 kNm-2, while loads >7.5 kNm-2 were apparently required to cause total collapse of buildings. Roof access loads are likely to be in the range 0.75-1.0 kNm-2. Thus, the loads experienced in Rabaul leading to partial collapse of the roof structure were at least twice the design load of the roof itself, while loads ~7 times the design load frequently produced failure of the whole structure.

One Canadian study of conventional timber roof frames built for the same design load and the same building code showed failure loads varying from 0.9 – 6.0 kNm-2. The wide range of failure loads was attributed to the size of rafters used, the heel joint details, the type of end support used and the quality of workmanship. As the Canadian roofs were designed to carry snow loads, it is not surprising roofs in Rabaul failed under tephra loads 2 to >7.5 kNm-2. Similarly, comparisons with roof failures around Pinatubo (Philippines) in 1991 suggests that most buildings performed reasonably well in Rabaul with tephra loads around 2 kNm-2, despite the absence of a tropical cyclone loading code.

It seems likely that relatively simple measures can be designed into building codes to protect structures from ash falls of less than, say, 200 mm or loads of <2.5 kNm-2. Larger loads are much more problematic for timber-framed buildings. Even if roof-framing systems could be strengthened, there is ample evidence to indicate that external walls would also require strengthening, presumably by abandoning timber-framing in favour of concrete block construction.


Colin Taylor of National Insurance of New Zealand assisted the author’s first visit to Rabaul after the eruption. Chris McKee of Rabaul Volcano Observatory provided invaluable field companionship and advice while Ian Graves of QBE Insurance and Peter Greenlees of AIG assembled valuable information on building damage.

For further information please contact Russell Blong
Tel. +61-2-9850 9683. Fax: +61-2-9850 9394

Risk Frontiers-NHRC
Macquarie University NSW 2109 Australia
Telephone: +61-2-9850 9683. Facsimile: +61-2-9850 9394