SLOPE FAILURES ALONG BRIL ROADS : QUANTITATIVE RISK ASSESSMENT AND RANKING GEO REPORT No. 81 ERM-Hong Kong, Ltd GEOTECHNICAL ENGINEERING OFFICE CIVIL ENGINEERING DEPARTMENT nJv 624.15136 S66 THE GOVERNMENT OF THE HONG KONG SPECIAL ADMINISTRATIVE REGION SLOPE FAILURES ALONG BRIL ROADS : QUANTITATIVE RISK ASSESSMENT AND RANKING GEO REPORT No. 81 ERM-Hong Kong, Ltd BOOKS REGISTRATION ORDINANCE Chapter 142 lJ'J 4293 H No.: HK This report was originally produced in November 1997 under Consultancy Agreement No. GEO 6/97 0 9 JUL 1383 i © The Government of the Hong Kong Special Administrative Region First published, January 1999 Prepared by: Geotechnical Engineering Office, Civil Engineering Department, Civil Engineering Building, 101 Princess Margaret Road, Homantin, Kowloon, Hong Kong. This publication is available from: Government Publications Centre, Ground Floor, Low Block, Queensway Government Offices, 66 Queensway, Hong Kong. Overseas orders should be placed with: Publications Sales Office, Information Services Department, 4th Floor, Murray Building, Garden Road, Central, Hong Kong. Price in Hong Kong: HK$76 Price overseas: US$15 (including surface postage) An additional bank charge of HK$50 or US$6.50 is required per cheque made in other than Hong Kong dollars. Cheques, bank drafts or money orders must be made payable to The Government of the Hong Kong Special Administrative Region PREFACE In keeping with our policy of releasing information which may be of general interest to the geotechnical profession and the public, we make available selected internal reports in a series of publications termed the GEO Report series. A charge is made to cover the cost of printing. The Geotechnical Engineering Office also publishes guidance documents as GEO Publications. These publications and the GEO Reports may be obtained from the Government's Information Services Department. Information on how to purchase these documents is given on the last page of this report. J.B. Massey Ag. Principal Government Geotechnical Engineer January 1999 FOREWORD In 1993, the Geotechnical Engineering Office (GliO) embarked on a programme of research and development studies on landslide risk management under the R & D theme on Quantitative Risk Assessment (QRA). This study was carried out by Mr Venkatcsh Sourirajan of ERM-Hong Kong Ltd under the Consultancy Agreement No. GEO 6/97. The main objective of the study was to formulate a methodology for the ranking of selected roads with a history of landslips (referred to as BRDL roads) based on quantitative risk assessment of slope failures along such roads with due regard given to risk to life and economic losses. Mr K.K.S. Ho and Dr D.O.K. Lo of the Special Projects Division administered the consultancy and reviewed the Report. In this Report, the results of the study are expressed in terms of risk to life (viz. potential loss of life per year) and economic loss per year due to road closure caused by landslips. The 41 BREL road sections considered are ranked based on potential loss of life per year and economic loss per year (due to traffic disruption caused by landslips). The roads are also ranked in terms of the total economic loss considering both the cost of a statistical life and economic loss due to road closure. P.L.R. Pang Chief Geotechnical Engineer/Special Projects EXECUTIVE SUMMARY The technique of Quantitative Risk Assessment (QRA) has been adopted by the Geotechnical Engineering Office (GEO) to assist in the formulation of landslide risk management strategy. This study has been commissioned to apply QRA to evaluate the risk posed by slope failures along selected roads (called BRIL Roads), which have a high incidence of landslides and also have high traffic density. Slope failures along busy roads not only have the potential to cause fatalities and injuries but also cause serious disruption to traffic. In order to evaluate the potential for disruption to traffic, the risk assessment technique has been applied in this study to evaluate the risk of economic loss caused by traffic disruption in addition to risk to life. The main purpose of this study is to provide information on the risks associated with slope failures to help prioritize and co-ordinate the slope and road improvement programmes. The methodology developed in this study includes the determination of frequency of landslide incidents, assessment of the consequences of landslides in terms of fatalities and impact on road traffic and finally calculation of the risk levels in terms of Potential Loss of Life (PLL) and risk of economic loss. The approach for frequency estimation, as described in Section 3, is summarised below: • the frequency of landslides is estimated from historical data for slope failures along each BRIL road section for the period 1984 to 1996. Incidents reported as landslides, washouts, rockfalls and retaining wall failures are considered together as 'landslides' while those reported as boulder falls are considered separately; • the proportion of upslope and downslope failures reported along each road section have been considered to model the consequences of failure as that from cut slope and fill slope respectively; • the slope height-failure volume distribution is estimated for each road section based on the slope height-failure volume data reported in previous incidents for the given road section and the slope height-failure volume distribution derived from the Special Administrative Region (SAR)-wide study on slope failures from pre-GCO man-made slopes. Slope height information from the 1977/78 Slope Catalogue has been used. A modified volume distribution is evolved which attempts to best represent the road specific historical data while at the same time ensuring that certain high volume incidents for the corresponding slope heights, although may not have occurred, are nevertheless represented with a low probability; • for upslope failures, the slope height-volume distribution is estimated for three slope height ranges, <10m, 10 to 20m and >20m and for five volume ranges, <20m3,20 to 50m3,50 to 500m3,500 to 2000m3 and >2000m3. For downslope failures, the volume distribution is estimated without any reference to slope height for five volume ranges, <20m3/ 20 to 50m3,50 to 200m3,200 to 1000m3 and >1000m3; - 6 - • the slopes along BRIL Roads are mainly pre-GCO slopes which are potentially sub-standard. A small percentage of these slopes have been improved to current geotechnical standards. In order to account for such improvements in the frequency analysis, a reduction factor equivalent to the proportion of total number of slopes along each road section that have been improved to current standards is applied. It is further assumed that the chance of failure of slopes designed to currently applicable standards is negligible in comparison with pre-GCO slopes; The approach for consequence estimation, as described in Section 4, includes estimation of fatalities and estimation of economic loss due to road closure. The approach for estimation of fatalities from landslides and boulder falls is summarised below: • the methodology for consequence assessment (ie, estimation of fatalities caused by landslides) is based on suitable adaptation of the approach evolved in previous studies by GEO on SAR-wide Pre-GCO Man-Made Slope Failures, Landslide Consequence Severity Classification of Roads, etc. • the number of fatalities from landslides is expressed as a function of vulnerability factor, scale of failure and expected number of fatalities for a reference landslide affecting a facility which is road in this case; • the vulnerability factors derived in the SAR-wide study as a function of runout angle and failure volume for cut slopes failures (ie, upslope failures) and as a function of crest distance and failure volume for fill slope failures (ie, downsiope failures) are adopted. The proximity of the affected facility below a slope is expressed in terms of shadow angle, which is estimated considering three slope height ranges and the distance to the centre of each lane from the slope toe to determine the appropriate vulnerability factor. The average of vulnerability factors computed for each lane as the affected facility is then considered. Similar approach, but based on crest distance, is adopted for downsiope failures; • the scale factor which is the ratio of the width of actual landslide to the width of reference landslide is computed for different failure volumes as given in the SAR-wide study; • the expected number of fatalities for a reference landslide (of 50m3 volume) is estimated from the nomograph which correlates fatalities with actual Annual Average Daily Traffic (AADT) and number of lanes for a given road; • the estimation of fatalities from boulder fall is based on a model which evaluates the probability of boulder hitting a vehicle for various scenarios- falling boulder hitting a moving vehicle, moving vehicle hitting a fallen boulder etc. The probability of death of occupants in the road vehicle is considered for each of the scenarios. The approach for computing economic loss due to landslides is summarised below: • economic loss due to landslides is computed by considering the potential loss of life in terms of Value of life1 and the loss due to road closure resulting from landslides; • the value of life, or the implied cost of averting a fatality is assumed as HK$ 24 million; • road closure following landslips may result in increase in travelling time which may cause indirect effects on the economy due to lost time for productive work; • the increase in travelling time due to congestion on the roads caused by lane/road closure is based on the initial suggestions by the Financial Services Bureau. The increase in travelling time is estimated for different class of roads (rural, urban, trunk roads etc) and for different proportion (or extent) of road closure; • the time value for additional hours lost in commuting to work(which is higher than time value for non-working hours) is adopted from studies undertaken for the Transport Department in Hong Kong and in the UK; • the increase in vehicle operating costs due to additional distance travelled as a result of congestion on roads due to road closure is considered; • the duration and extent of lane closure caused by landslides is derived for various failure volumes based on data from past incidents provided by the Highways Department for the period 1994 to 1996. Data from past incidents on duration of closure of more than one lane is limited and therefore suitable assumptions have been made; • the probability of different extent of closure (ie, one lane, two lanes, etc) is derived for various failure volumes from the GEO Annual Report on Rainfall and Landslides. Again, this data is adopted for estimating the probability of closure of the first lane while suitable assumptions have been made for the probability of closure of more than one lane. Results The results from the study are expressed in terms of both risk to life and risk of economic loss due to road closure. Risk to life is expressed in terms of PLL (Potential Loss of Life per year), which is a measure of societal risk. The 41 BRIL Road sections considered under this study are ranked based on PLL per year and economic loss per year (due to traffic disruption caused by landslides; other costs such as cost of slope repair, etc. are not considered). The roads are also ranked based on total economic loss (considering both cost of life saved and economic loss due to lane/road closure caused by landslides), expressed in units of per kilometre per year, to provide a more uniform basis for comparison between roads given that their lengths are different. The Castle Peak Road section ranks first both on PLL and on economic loss per year as a result of lane/ road closure. However, when the total economic loss is expressed in units of per km per year, the Kwun Tong Road section ranks first. The results for PLL, economic loss and ranking are presented in the attached Table.