TRANSPORT RESEARCH LABORATORY TITLE Accident reduction research in Papua New Guinea by B L Hills, N M Thompson and K Kila IA Overseas Centre Transport Research Laboratory Crowthorne Berkshire United Kingdom HILLS, B L, N M1 THOMPSON and K KILA, 1991. Accidenc reduction research in Papual New Guinea. In: PTRC. Nineteenth Transport Highwa s and Planning Summer Annual Meetins'. Proceedines of Seminar C. University of Sussex. 9-13 September 1991. London: PTRC Education and Research Services Ltd. ACCII)ENT REDUCTION RESEARCH IN PAPUA NEW GUINEA B L Hills Transport and Road Research Laboratory, UK N M Thompson Department of Transport, PNG K xila Department of Transport, PNG 1. INTRODUCTION 1.1 Background Road safety became an issue of concern for the Government of Papua New Guinea in the late 1970's. From the beginning, its Department of Transport was aware of the need for reliable data on which to base future road safety programmes and in conjunction with the Police,' introduced a new traffic accident report form in 1982. Over the 10 years that have passed since that time, Papua New Guinea has developed one of the best accident data recording and analysis systems in any developing country. Since 1986, the TRRL Overseas Unit has been working with the Department of Transport in a programme of. collaborative road safety research. This research covers both: (1) ACCIDENT PREVENTION, which is concerned with standards of design and planning of new road schemes and related devel- opment (2) ACCIDENT REDUCTION, which is concerned with the application of remedial measures to problems identified in the existing road network. This paper describes the second aspect of this work, the Accident Reduction research. 1.2 Papua New Guinea's Road System Papua New Guinea ha's a. total land area of about --460, 000 square kilometres; this comprises a large mountainous and very heavily forested mainland and some 600 smaller offshore islands. Its population is only about 3.5 million. The country is divided into 20 Provinces. The major areas of population have traditionally been in the Highland valleys and basins, and around one third of the population is still drawn from the five Highland Provinces. The two major urban population centres (and ports) are Port Moresby, the capital city with a population of about 200,000, and Lae with over 50,000 people. a Accident Reduction in PNGHlsTopnKa The transport system ref lects the geographic and demographic characteristics of the country, and its improvement has generally been limited to the areas of high population and those with good potential for development. In its early development, transport was largely limited to sea and air, but over the past 20 years a system of ten National Highways has been established, and is being progressively upgraded. The most important of 'these -is the Highlands Highway, which is about 500 Kms long linking Lae to the Highlands region. The present network comprises some 23,480 Kms of road, of which 2130 Kms (9%) are paved. Arterial and trunk roads (4.5 to 7 metres wide) make up 5000 Kms, the remainder being collector and feeder roads. The National Government is responsible for about 20% of the roads, known as National roads. The number of registered vehicles in Papua New Guinea in 1988 was about.50,000, and-this has not increased -significantly since about 1980. Road accidents have been of concern for over ten years and the current fatality rate of 69 deaths per 10,000 vehicles is relatively high by developing country standards. 1.3 TRRL Overseas Unit's Road Safety Research The Overseas Unit of TRRL, began its research into the road safety problems of developing countries in 1972. Early research concentrated on accident rates, trends and costs, and road user behaviour and knowledge (eg Jacobs and sayer, 1982; Downing and Sayer, 1982). In recent years, the emphasis has shifted towards techniques of accident data collection and analysis with a view to establishing the cost-effectiveness of road safety countermeasures in developing countries, for which there is, at present, very little data indeed (Hills and Jacobs, 1980; Hills and Yerrell, 1988). Over the past 15 years, Local Authorities in the UKX have developed techniques of Accident Blackspot Investigation that have proved to be highly cost-effective. The Unit believes these techniques offer considerable potential for reducing accidents in developing countries, particularly in view of the emphasis placed on low-cost engineering countermeasures. It has therefore included in its research programme the development of a number of tools and techniques (accident report forms, microcomputer software, manuals etc) designed to assist the establishment of accident blackspot investigation in developing countries. To test the suitability of the procedures involved, the Overseas Unit is engaged in collaborative research programmes with a number of developing countries, including Egypt, Ghana, Indonesia, Malaysia, Pakistan, and Papua New Guinea. The work in Egypt, Ghana and Pakistan is also described in this seminar (Sayer, Baguley and Downing, 1991). 2 Hills, Thompson, Kila Accident Reduction in PNGHilTopnia 2. COLLABORATION BETWEEN TRRL AND PNG DEPARTMENT OF' TRANSPORT Overseas Unit staff took part in the 1986 Papua New Guinea Road Saf ety Study, which was commissioned by the PNG Department of Transport and funded by the Asian Development Bank. As part of this, the Unit's Microcomputer Accident Analysis Package (MAAP) , which has been specially designed for developing countries -(Hills and Elliott, 1986), was installed on a trial basis. In January 1987, this Package became the national system of PNG, running on microcomputer systems both in the Department of Transport and in the Royal Papua New Guinea Constabulary (RPNGC) headquarters. The RPNGC had previously established a good accident reporting system but had experienced difficulties with the mainframe computing facilities. In addition, the Department of Transport found it needed more in-depth analyses of accidents than this mainframe system could provide. The police accident report form was modified both to improve its content and to enable direct entry of data from the form into the microcomputer. PNG thus became the first country to adopt the TRR.L Package on a national basis. The mainframe accident data from 1982-1986, totalling over 25,000 accidents, were transcribed to the micro system. over 40,000 accident records are now available for analysis. The introduction of the microcomputer accident analysis system combined with the existing accident data base at the current stage of development of the PNG road system offered an excellent opportunity for research that could be of considerable value not only to PNG but to many other developing countries. For these reasons, a programme of cooperative research was agreed between the Department of Transport and the Overseas Unit in 1986. The programme of road safety research now covers the following areas of work: (1) the evaluation of. road safety countermeasures, especially low-cost engineering measures at accident blackspots (2) an investigation of the relationships between the various elements of geometric design and accident rates (3) alcohol: a survey of drinking and driving in Port Moresby (4) the safety of passengers riding in the rear of Pick-ups This paper is particularly concerned with research into accidents at blackspots but in order to put those accidents into perspec- tive, the national characteristics of road accidents in Papua New Guinea are described in the next section. 3 Hills, Thompson, Kila Accident Reduction in PNGQHls hopoKl 3. 'ROAD ACCIDENTS PAPUA NEW GUINEA' Reports The entry of accidents for 1987 into the micro system was completed at the end of 1988 and following a period of analysis of these data, a report entitled 'Road Accidents Papua New Guinea 1987' was published in May, 1989. This was one of the first significant outputs from the collaborative research programme. The 1988 edition was published in October 1990 and the 1989 edition is near completion (July 1991). These are intended to be the first of a series of annual reports presenting the road accident situation in Papua New Guinea and now form the basis of the Department'`s road saf ety programme. Some of the main f indings presented in the 1988 report included: (1) 15% of all vehicles on the road in Papua New Guinea are involved in a reported accident each year. (2) 46% of all accidents in Papua New Guinea involve only one vehicle, and these accidents cause 73% of all casualties, and 76% of all hospitalised and fatal casualties (see Figure 1). These accidents involving only one vehicle are almost entirely made up of: (a) pedestrian accidents (b) rollover accidents (c) "struck object off the road" collisions. (3) rollover (or overturn) accidents are the biggest single cause of casualties in Papua New Guinea. Although they only represent 14% of accidents, they give rise to 33% of all casualties and 27% of all fatalities (see Figure 2). (4) 45% of all casualties are the occupants of Pick-up type vehicles, whereas these vehicles make up 33% of registered vehicles (see Figure 3). (5) 17% of all accidents in Papua New Guinea are reported as 'Alcohol Suspected'. On Saturdays, 26% of drivers involved in accidents are classed as 'Alcohol Suspected'. (6) The National Capital District with 30% of all accidents has only 13% of fatal accidents, whereas the Highlands Region with 25% of all accidents has 38% of fatal accidents. Thus, accidents in the Highlands Provinces tend to result in more serious injuries than accidents in the National Capital District. (7) there are, on average, 0.07 fatalities per reported accident, but for all the National Highways this figure is 0.15, with 0.33 and 0.22 for the Hiritano and Ramu Highways 4 Hills, Thompson, Kila Accident Reduction in PNGHilTopnia respectively. The Highlands Highway has by far the largest total of injuries. (8) although there has been an overall reduction in the total number of recorded accidents from 1981 to 1988, fatal accidents have shown a steady increase from 1982. This almost certainly indicates an increase in under-recording of minor injury and damage only accidents, rather than a genuine reduction in total number of accidents. (9) almost as many pedestrians in Papua New Guinea are killed or hospitalised from accidents when walking along or beside the road (148) as crossing the road (163). In addition to analyses of the accident data at a national level, lists of worst junctions and links in the National Capital District (Port Moresby) and Lae are included in these annual reports and these have formed the basis of a list of priority sites for treatment (see Figures 4 and 5). A histogram of accidents along the major rural highway of PNG, the Highlands Highway, is also presented (see Figure 6). 4. ACCIDENT BLACKSPOT INVESTIGATION USING MAAP This section includes a general outline of the various procedures involved in Accident Investigation, and notes how particular aspects were handled in Papua New Guinea. More detailed accounts are available, such as the UK Accident Investigation Manual (Department of Transport, 1986). 4.1 The Process Accident Investigation involves a continuous programme of identification of accident blackspot locations, analysis of the factors involved, site studies, implementation of countermeasures and evaluation of their effectiveness. The overall process is summarised in Figure 7. In the schematic diagram, the process is divided into activities at the accident SITE and those in the Accident Investigation OFFICE. 4.2 Defining accident locations Being able to indentify the accident's location from the computer record is a basic necessity of any accident investigation system operating at the Local Authority level. MAAP uses three systems for locating accidents: (1) rural highways - kilometre posts (2) urban areas - node/link/cell system (3) any area - grid co-ordinate system 5 Hills, Thompson, Kila Accident Reduction in PNGHilTopnKa In Papua New Guinea, to maintain compatibility with the earlier mainframe system, a fourth location system is used that involves giving each street name in urban areas a three-character code. Junctions are defined by the codes of the two roads involved, and links by the code of the road on which the accident occurred together with the codes of the two intersecting roads either side of the accident location. Ideally, it should be possible to identify the accident location to within lO0m or better. Whilst this should be achievable in a town or city, it is more difficult on inter-city roads where distances are usually defined using kilometre posts. In this case, locating accidents to within 1 Km may be a more achievable goal; in some circumstances, however, 1 Km can be a long distance in which to pinpoint a problem in the road during a site study. Unfortunately, in Papua New Guinea, kilometre posts only exist on sections of the Highlands Highway. However, there are frequent bridges on most sections of the National highways, and the police use these extensively to help identify the accident location. To minimise the problems of identifying accident location, the 'Accident Location Sketch' (Figure 8) has been introduced in revised police report forms being tested in a number of develop- ing countries, including Papua New Guinea. One of the first uses of the 'accident location sketch' was in Bahrain (Ross, 1984) 4.3 Identifying hazardous locations The TRRL Microcomputer Package MAA.P has a number of options for identifying locations with high accident rates: (1) plotting accidents along a (Nodal) Route in a city (2) listing the worst Links or Nodes (Junctions) in a city (3) plotting a co-ordinate accident map for a city (4) listing the worst accident sites by map co-ordinate (5) plotting accidents along a rural highway using kilometre posts (6) listing the worst accident sites by kilometre. In assessing the worst accidents sites, the severity of the accidents is taken into account in some countries. This can be done in at least two ways (1) by allocating points (eg the Malaysia Highway Planning Unit allocates 6 for a Fatal, 3 for a Serious, 0. 8 f or a Slight, 0. 2 f or a Damage Only) ; or (2) by using the average estimated costs of Fatal, Severe, Slight, and 6 Hills,.Thompson, Kila Accident Reduction in PNGHilTopnia Damage Only accidents to total the costs of the accidents at each site. In Papua New Guinea, these costs were estimated in 198 6 to be 1(37,225 for a Fatal accident, K(9,680 for a serious injury accident, K3,020 for a Slight injury accident and 1(1,800 for a damage only accident. The usual practice in the UK is not to weight accidents by their severity (only personal injury accidents are recorded); the underlying argument often given is that it is a matter of chance whether the injury was Fatal, Severe or Slight. In assessing the relative accident records of sections of road between junctions, it is important to remember that the length of a link varies considerably and this must be taken into account in deciding priorities. 4.4 Further analysis of the worst sites The next stage of the Accident Investigation process, as Figure 7 indicates, is for the worst of the accident sites to be selected for more detailed analysis in the Accident Investigation Office. MAAP offers a number of possible analyses: (1) Cross-tabulations can be produced (using program options 6 and 7 in MAA.P) with limitations defined to restrict the analyses to just those accidents occurring at an accident site under investigation. These can be useful in establish ing the overall nature of accidents at the site. (2) a Stick Diagram Analysis can be carried out (using option 8 in MAAP) to. try and establish patterns in the accidents. (3) the serial numbers of the accidents occurring at the site can be listed from the computer (using option 5 in MAAP) and the original police reports then retrieved from the police files. It is then valuable to read the statements of those involved in the accidents and the statements of any witnesses. These sometimes reveal new factors that should be included in the earlier Stick Diagram eg. a high percentage of those involved may have been strangers to the area. Another valuable source of information in the police report is the policeman's Collision Sketch. For a junction site, it is often useful to plot the location of each accident on a plan of the junction in an Accident Plot; more useful still is to represent diagrammatically the movements of all the vehicles in each accident in a Collision Diagram for the junction. Figure 9 shows an example of such a Collision Diagram prepared by the PNG Department of Transport. Figure 5 showed that the worst Link in the National Capital District is Link 157/159. The collision diagram in Figure 9 is based on an analysis of three years data 7 Hills, Thompson, Kila Accident Reduction in'PNGHilTopnia for the site. It shows that there are two main problems: (1) the problem of rear-end shunts associated with the pedes- trian crossing (2) the problem of vehicles turning across the road into premises with direct frontal access to the highway. (It is proposed to construct a central median and put a high skid resistance material down on the approaches to the pedestrian crossing). 4.5 Site studies The next phase of the procedure is to make a site study, armed with the analyses made at the office. The first task is to record the key physical characteristics of the site eg. road width, shoulder width, existence of street lighting etc. A checklist is available in the UK Accident Investigation Manual; such checklists may not always be relevant to the types of site being investigated, in which case the development of a locally applicable checklist should be considered. Photographs also provide a valuable record of the site layout. It is frequently useful to make preliminary surveys of traffic and pedestrian flows. A short survey of traffic approach speeds on different arms of a junction can sometimes be revealing; one simple method to do this is to use one person with a stopwatch stationed 50m from the junction and a second stationed 150m from the junction flagging when a vehicles passes him. In Papua New Guinea, the research made extensive use of pneumatic tube counters that could classify vehicles or speeds. It was found necessary to employ local villagers to guard the meters and tubes. With the latest models, one meter can classify speeds in two directions on "Low Flow" roads. Extensive use was also made of video recording sites before and after treatment. Further site studies should generally be related to the accident problem identified in the office analysis. For example, if wet weather or skidding accidents appear to be a problem, then measurements of skid resistance would be made. If there is a clustering of accidents at a particular time of day or a particular period of the week, then the site study should be made at these times. Again, the UK Accident Investigation Manual offers extensive checklists of possible factors to look for. When the factors leading to the accidents are not clear, then a study of Vehicle Conf licts eg. ranked on a scale of 1 to 5 as illustrated in Figure 10, can often reveal hidden problems at the site. Observations of drivers' head turning movements or 8 Hills, Thompson, Kila Accident Reduction in PNGHilTopnKa observing that drivers have to stretch -to see approaching vehicles can sometimes lead to identifying a visibility problem. Any other problems of visibility should be recorded, eg. shallow hilicrests or vegetation blocking the view of approaching vehicles. Talking to local residents can also be helpf ul, although opinions about causes should always be treated with a good deal of caution. 4.6 Implementing the countermeasure Having identif ied the probable causes of the accidents, the possible countermeasures are listed and ranked according to cost. Normally, the lowest cost solution would be adopted unless there is good reason to believe that it would be unsuccessful. In implementing the countermeasure, due consideration should be given to whether or not advanced warning to the public is necessary, either in pre-publicity or in advance of the site with signs such as 'Experimental Junction Layout Ahead' (white-on- red) . The first days of operation of a scheme should be carefully watched in case an unexpected problem has been incorporated into the design. 4.7 Evaluation After a suitable elapsed period of time, the long term benefits of the scheme should be evaluated with 'Before and After' studies. For details of this, manuals such as the UK's Accident Investigation Manual (Department of Transport, 1986) should be referred to. It is most important to use 'control data' in such an analysis, comparing the 'Before and After' accident levels at the test site, either with those at similar but unchanged sites in the locality or with some regional data. Comparing the 'Before and After' accidents at the test site with those at the control sites eliminates 'global' factors that may occur in the after period, such as a period of particularly good or bad weather, a change in oil price affecting vehicle speeds or a change in traffic law. If this evaluation indicates that the scheme has been unsuccessful then further analyses and site studies may be necessary before an alternative scheme is tried. The Chi-squared statistical test with Yates' correction is the most common statistical test used in evaluating 'Before and After' studies. The 'Regression-to-mean' effect should also be taken into account (Wright et al, 1988). 4.8 Special problems of accident investigation in developing countries It is often difficult to establish the above Accident Investiga- tion process in full in developing countries because of such factors as: a lack of trained manpower, limited financial and physical resources, and problems of co-operation and co-ordina- tion between departments; these are frequently combined with a lack of commitment by departments, partly attributable to the 9 Hills, Thompson, Kila Accident Reduction in PNGHilTopnia fact that Accident Investigation has yet to prove itself in the countries concerned. In Papua New Guinea, the Department of Transport has made a commiitment to establishing an Accident Investigation group with the eventual aim of establishing small teams in each Province. The Department's unit is still in its infancy but already it has made an impact both (1) in getting engineers and planners to think about and design for safety; and (2) in introducing a number of low cost countermeasure schemes, including roundabouts, lane dividers, chevron boards, footpaths alongside rural highways and speed reduction devices. 5. THE EVALUATION OF ENGINEERING COUNTERMEASURES IN PAPUA NEW GUINEA The team identified a number of projects and Figure 11 lists the various engineering countermeasures, both installed and pro- jected, that are under evaluation. The preliminary results presented in this section will be the subject of full statistical and cost-benefit analyses in a later report. The list of schemes is a combination of projects initiated by the team itself and those undertaken by local authorities. The primary reason for some of the measures was to improve traffic management or was for maintenance purposes rather than safety; however, the inclusion of safety benefits could considerably enhance cost-benef it values obtained and might suggest further investment in the measures. 5.1 Roundabouts Following research by the TRRL, roundabouts have been more widely used in recent years in the United Kingdom and Australia to solve both traffic management and road safety problems at junctions (eg Summersgill, 1991). In many developing countries, on the other hand, particularly in the Middle East and South East Asia, roundabouts are being replaced by traffic lights. The failure of roundabouts in these countries can very frequently be attributed to the failure of drivers to give way on entering the round- abouts. Between 1982 and early 1986, six roundabouts were installed in PNG's National Capital District at uncontrolled major/minor junctions that had both traffic management and road safety problems. As in some African countries, PNG drivers do seem to understand the concept of 'Give Way' and these roundabouts are generally regarded as being a successful innovation. Five roundabouts have recently been constructed at the worst junction blackspots in Lae, the second largest city *in PNG. These were half funded by the Department of Transport's Accident Investiga- tion Unit. 10 Hills, Thompson, Kila Accident Reduction in PNGHil, Top n, K a TABLE 1 (a) Analysis of accidents 1982-1988 at roundabout sites in the National Capital District (Port Moresby). Roundabout 1 installed in 1983/84, the other roundabouts in 1985/86. TABLE 1 (b) Analysis of accidents at all junctions in the Capital District (Port Moresby). National 1 1 Number of accidents per year Injury Accidents Damage Only SITES Before After Before After Waigani Drive / 2.0 0.4 15.0 9.2 Wards Road _____ waigani Drive / 1.0 1.3 9.2 15.0 Cameron Rd Waigani Drive / 1.2 0.3 11.7 8.3 Koura Way _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ Waigani Drive / 1.2 1.0 4.2 4.0 Goro Keaga _ _ _ _ _ _ _ _ _ _ _ Angau Drive/ Lah- 1.2 0 8.5 5.7 ara Ave Bisini Prde/ Kau- 0.2 0.3 2.7 2.0 bebe St _____ Average Rate for 65 1.17 0.55 8.6 7.4 sites Year 82 83 84 85 86 87 88 INJURY ACCIDENTS All coded junctns in 140 122 127 132 133 130 115 NCD All Stop, Give Way and 21 32 36 39 37 39 33 Traffic Signals in NCD Hills, Thompson, Kila Accident Reduction in PNG Hills, Thompson, Kila Table 1(a) shows an analysis of the accident data so far accumulated for the roundabout sites in the NCD. Except for one site, the analysis is based on four years 'before' data and three years 'after' data; the exception is Waigani Drive/Wards Road for which there are two years 'before' data and five years 'after' data. It will be seen that generally the figures look encourag- ing, with on average a halving of injury accidents per year per site. For purposes of comparison with this analysis of round- abouts, the Table 1(b) is an analysis of injury accidents over the same seven year period for (1) all junctions in NCD; and (2) junctions with Stop, Give Way and Traffic Signals as traffic controls. It can be seen that over the seven year period there has been little change in the general injury accident rate at junctions in NCD, suggesting that the introduction of roundabouts at the sites listed in Table 1(a) has indeed reduced injury accidents over that period. Figure 12 shows accident trends at two closely located junctions on a major urban arterial, Waigani Drive. The junctions are separated by less than 200m. They are encountered at the end of a fast dual carriageway, first the Cameron Road junction and then the Wards Road junction. The Wards Road roundabout was con- structed in 1983 and the Cameron Road roundabout in 1986. The Waigani/cameron roundabout clearly increased accidents at that location during its first year (1986) but, at the same time, seems to have reduced accidents at the already existing Wards/Waigani roundabout 200m down the road. The Waigani/Cameron roundabout, which was previously a T-junction, is encountered before the Wards/Waigani junction from the fast dual carriageway section, so it would appear that it now bears the brunt of the fast entry traffic. The accident types need to be analysed to confirm this. Figure 12 shows the apparent crossover effect, with the combined data showing an overall decline in accidents. 5.2 Central reservation During 1986, a 300m section of the Hubert Murray Highway between Taurama and Boroko was upgraded from a single carriageway to a dual carriageway. This involved associated road widening and the closure of one T-junction. The scheme was carried out more for traffic management purposes than for road safety. For the pedestrian, the median helps the-crossing task-but the inevitable increase in vehicle speeds will have made judgements more difficult. ('Before and After' speed measurements are not available as the scheme began before the research programme got under way) . Table 2 shows an analysis of accidents along the stretch of road for the period 1982-1989. It can be seen that Head On, Rear End, 90 Degree and Sideswipe accidents halved after the scheme; these accidents were large in number but their severity was almost exclusively minor injury or damage only. 'Hit object off the road' and Pedestrian accidents show a completely 1 2 2 Accident Reduction in PNGHilTopnKl TABLE 2 Hubert Murray Highway, Boroko, NCD: of constructing 30Gm of central The effects on accidents reservation in 1986. I J ~~~~~~~~~BEFORE* SCHEME _ __AFTERSCHEME 82 8 3 84 85 86 87 88 89 ACCIDENT TYPE Head On 5 1 6 8 310 0 Rear End 2 4 30 22 1 17 8 8 1 2 90 degree 6 6 10 7 12 3 2 Sideswipe 7 .10 11 10 75 3 7 SUB-TOTAL 42 47 49 42 28a.. 16 14 21 Hit Object 0 1 0 20 3 1 3 Off Road ... ____ Pedestrian .1 0 1 2 12 2 3 SUB-TOTAL, 1 1 1 4 15 3 6 (ALL ACCIDENTS 4 3 49 50o 46 3 21 18 28_ TABLE 3 Hubert Murray Highway, Boroko, NCD: The effects on accidents of constructing a pedestrian footbridge in 1985. _________ BEFORESCHEME _ __-AFTER *SCHEME___ ACCIDENT 82 83 84 85 86 87 88 89 TYPE ....... Pedestrian 4 4 2 11 1 2 2 Other Acc- 2 3 21 1 8 26 1 4 1 4 21 1 6 idents_______ Total 27 25 20 .27 1 5 1 5 2 3 1 8 1 3 Hills, Thompson, Kila Accident Reduction in PNGHlsTopnKa opposite picture: they have doubled since the scheme was introduced, and although they are small in number, they are high in severity, with eight out of the 22 accidents being fatal or resulting in hospitalisation (six were pedestrians) . The need to prevent pedestrian accidents on this section of road has thus become a high priority. 5.3 Pedestrian Footbridge During 1985 and early 1986, a -pedestrian footbridge was con- structed over the Hubert Murray Highway in Boroko, NCD, at one end of the section of road described in 5.2. Table 3 shows an analysis of the accidents within about lO0m of the bridge. The numbers are small, but suggest the bridge has reduced pedestrian accidents from an average of three per year to 1.5 per year. 5.4 Rural Footpaths - Goroka A major area of neglect throughout the developing world is the provision of pedestrian footpaths alongside major rural highways. On some of these highways, large numbers of pedestrians can be observed but it is clear that highway designers have not taken their needs into consideration. Indeed, the authors know of no surveys that have ever been carried out to establish potential pedestrian needs when a major rural highway has been constructed or upgraded in a developing country. The Highlands Highway in PNG is one such highway. It is the major rural highway of the country with very heavy pedestrian usage in the Highlands regions. Analyses of the accident data confirmed that rural pedestrian casualties were a major problem on the highway. One of the major blackspots was on the approaches to the Eastern Highland's Provincial Capital, Goroka. Pedestrians were making their own tracks in many places to get away from the traffic - in one extreme case, they were prepared to walk on slopes of 45 degrees to achieve this. In another case, pedes- trians used embankment benching as a footpath. (This could be adopted as a formal design in many such situations). As an initial trial, the Department of Transport is half-funding 13 kilometres of footpath, one metre wide, to be hand-dug alongside the drainage ditch from Goroka towards Asaro. Approxi- mately 5 Km has so far been, constructed. A-survey---indicated that the footpath is being used by about two-thirds of the pedestrians (68%), but there remains a significant proportion who continue to walk along the highway itself. More studies of its use are required, particularly to establish the reasons for pedestrians not using it, and whether its use is increasing or decreasing. To improve pedestrian usage, it is already planned to widen and generally upgrade the standard of the footpath so far con- structed; and-estimates for sealing it are being obtained. 1 4 Hills, Thompson, Kila Accident Reduction in PNGHisTopnia 5.5 Chevron Boards Three accident-prone sections of rural highways containing difficult curves were identified. These were: (1) from Goroka to Kamaliki Creek on the Highlands Highway, one of the worst blackspots along the Highlands Highway; (2) on the 'Buruni Coast Road in the National Capital District; and (3) Gusap Bridge on Ramu Highway (see Section 5.5 below). They were found to have single vehicle night-time accidents as a predominant type. Chevron boards have been erected at each of the sites and are under evaluation. This was the first use of chevron boards other than at roundabouts in Papua New Guinea. Analyses of their effectiveness will have to wait for the 1990 accident data to be completed. 5.6 Accident Blackspot at Gusap Bridge, Ramu Highway From December 1988 to April 1989 there were three fatal accidents and several injury accidents at the Gusap and Bora River bridges, which are a short distance apart on the Ramu Highway some three kms from the Ramu Sugar factory. The road carries only about 200 vehicles/day. The bridges are on a recently sealed section of the Highway. This section of the road is very straight, and therefore fast, from Wataris to the Gusap Bridge, where the road then does a greater than 90 degree bend to cross the Gusap River - a classic accident producing situation. Before the beginning of the bend, there was a 40 kph speed limit sign at approximately 150m and a 'dangerous curves' symbolic warning sign at approximately 50m. There were no other advance warnings of the bends nor any chevron boards at the bends. When first sealed, there were apparently more signs but a number were stolen. There is no background, just open sky, to the Gusap River bend, giving no other visual clues. Before the road was sealed, there were only a few accidents at the bridges; it has been suggested that this was due to the large potholes on the approaches to the bridges. To improve safety, bar patterns (originally developed at TRRL) covering some 200m were painted on the road on the approaches to the bridges and four reflectorised chevron boards were also erected on the 90 degree bend. Approach speeds were measured 'Before and After' the installation of the chevron boards 'and bar patterns. These showed, as has been found in various studies (eg Jarvis, 1989), that the bar pattern had little effect on the entry speeds to the bend. However, research in the UK (Helliar-Symons, 1981) has demonstrated that speed-related accidents at rural roundabouts are dramatically reduced by the bar pattern, suggesting that the pattern acts as an alerting device rather than a slowing device. So far, no fatal accidents have occurred since the installation, but accident monitoring is continuing. Eventually, it is planned to re-align the road in the vicinity of Gusap Bridge. 1 5 Hills, Thompson, Kila Accident Reduction in PNG.HilTopn, ia This accident blackspot emphasises the need to anticipate potential new problems when a road is upgraded, and the need to "over-design" warning devices to alert drivers of the impending problem. It is planned. to paint the TRRL bar pattern on the approaches to all single lane bridges on the Highlands Highway. An analysis of accidents has identified the bridges with the worst accident records and a detailed analysis of the accidents is being carried out. It is planned to treat all single lane bridges in one Province on the Highway at the same time; those with high accident rates will get the full TRP.L bar pattern, those with low accident rates a half-length pattern. 6. CONCLUSIONS This project has shown how important a sound accident reporting and analysis system is to a developing country. It has enabled Papua New Guinea to investigate its accident problems in depth, and determine priorities and introduce improvements on a rational basis. Also, it has provided a much improved evaluation capabil- ity. The wide range of schemes that have been or are being implemented has been listed in Figure 11. These schemes, combined with the availability of sound historical accident data, provide an invaluable opportunity for research. Many of the accident remedial measures have only recently been implemented and others are yet to come. It is therefore too early to present general conclusions relating to the effectiveness or otherwise of the countermeasures. However, in the case of roundabouts, preliminary evidence is presented which suggests that they have been effective in reducing accidents. Special studies planned for the future include the introduction of parking bays on the Highlands Highway and trials with high skid resistance surfaces in the National Capital District. 7. ACKNOWLEDGEMENTS The work described in this paper forms part of the programme of the Transport and Road Research Laboratory, United Kingdom, and the paper is published by permission of the Director.It also forms part of the research programme of the Department of Transport, Papua New Guinea and is published by permission of the Secretary. 8. REFERENCES' DEPARTMENT OF TRANSPORT, 1986. Accident Investigation Manual. Distributed by the Royal Society for the Prevention of Accidents, 1 6 Hills, Thompson, Kila Accident Reduction in PNGHilTopnia Birmingham, UK. DOWNING, A J, and I A SAYER, 1982. A preliminary study of childrens' road crossing knowledge in Three Developing Countries. TRRL, Supplementary Report 771. Crowthorne: Transport and Road Research Laboratory. HELLIAR-SYMONS, R D, 1981. Yellow bar experimental carriageway markings - accident study. TRRL LaboratorV Report 1010. Crowthor- ne: Transport and Road Research Laboratory. HILLS, B L and G J ELLIOTT, 1986. A microcomputer accident analysis package and its use in Developing Countries. Srinagar: Indian Road Cong~ress Road Safety Seminar. HILLS,'B L and G D JACOBS, 1980. The application of road safety countermeasures in Developing Countries. Traffic Engineering & Control, 22 (8/9) , 464-8. HILLS, B L and J S YERRELL, 1988. Accident black-spot investi- gation in Developing Countries. Berlin: International Road and Traffic Conference. JACOBS, G D and I A SAYER, 1983. Road accidents in Developing Countries. Accid Anal & Prey 15(5) pp. 337-353. JARVIS, J R 1989. The effect of yellow bar markings on driver behaviour. Research Report ARR No 173. Australian Road Research Board. SAYER, I C BAGULEY, and A DOWNING, 1991 Low-cost engineering measures in Egypt, Ghana and Pakistan. PTRC Summer Annual Pleeting, University of Sussex. 9-13 September 1991, London:Proc of Session C, Towards Safer Roads in developing countries. PTRC Education and Research Services. SUMhMERSKILL, I 1991. What determines accident risk? Safety 91: Proceedings and programme. 1-2 May 1991. Crowthorne: Transport and Road Research Laboratory. WRIGHT, C C et al, 1988. Estimating the regression-to-mean effect associated with road accident blackspot treatment: towards a more realistic approach. Acc. Anal. Prey, 20(3), 199-214. Crown Copyright 1991. The views expressed in this paper are not necessarily those of the Department of Transport, United Kingdom or the Department of Transport, Papua New Guinea. Extracts from the text may be reproduced, except for commercial purposes, provided the source is acknowledged. 1 7 Hills, Thompson, Kila Accident Reduction in PNGHilTosnil FIGURE 1 Casualties by Number of Vehicles involved in the accident - Figure from 'Road Accidents Papua New Guinea 1988' 1 8 1988 ACCIDENTS - Casualty Statistics RPNGC / Dept of Transprt 3000' 0 . ..... .......... ..................................................... E z 5 0 0 .......................................................... 1 2 3 4 5 6 7 8 9 10 Number of Vehicles in Accident M Fatal M Hospftal " Not-Hospital INJURY SEVERITY by COLLISION TYPE 1988 Fatal Hospital Not-Hosp TOTAL Head On 38 180 18,4 402. Rear End 7 27 53 87 90degree 7 47 100 1 54 Sideswipe 23 98 10,4 225 Overturn 90 534 506 11 30 Object On Rd 8 27 22 57 Object OffiRd 35 35 .267 607 Parked Vehicle 0 6 1 3 1 9 Pedestrian 88 242 180 510 Other 32 108 76 216 TOTAL 328 1574 1505 3407 [FIGURE 2 Table from 'Road Accidents Papua New Guineal1988' Hills, Thompson, Kila Accident Reduction in PNGHilTosnia 1988 ACCIDENTS - Casualty Statistics RPNGC / Dept of Transport 800 700 ................. :D 3 0 0 .............. j 2001---- ~~ -- . -.............. Pedestrian HGV Bus PickUp Other Car Tractor M/Cycle Cycle " Not Hospital 1 SFatal M Hospital FIGURE 3 Casualties by Class of Road User -Figure from 'Road Accidents Papua New Guinea 1988' 1 9 Hills, Thompson, Kila Accident Reduction in PNG Hills, Thompson, Kila NO 77,D Ju cto Accidents...1988. Jitn~~n Alli ~~Aum~~~~~er fl~~~~~~L N~~~~m~~ .... ..... . it ~~~~~& urray/W ai~~~~~~~~~~arii. .. ...... ..... 1. 250 1-lubertWPA drayTurmaWrdd3 359 aigail~png Grden15 441 Lapwing/Wo~~~~~~~~~~~~~~~dc~~~ock..... 14....1 251 Hubert P~urrayl~ahara 13r 25~ Hbert urrayBorok 13U 363 Waigani/Camemn 1~~~~~~~~ imo 3~4 wu anJhnuse12 058 E~~a Beac~~i....... 10 2.... 365 Waiga~~~/K~~ura 10 1. 159 Hubert6 K rryKooeb 254 Hubert~~. .. rr .. ....... 9. 157 Hubert~.. aurJ cra `d ' i t. ro ab 4,:~.".::.I.~: 303 G cRE 4 Tale fom Roa A.des 8au Ne1une 98 20 Accident Reduction in PNG Hills, Thompson, Kila NCD INK CCIDENT 1.988 Nwnber Road Names Accidents Acc~~~~~~~~~~~~~id---ts 157/159 Hubr ?Irry crathy.oveS .5'5 H..... .an a ur~-a~an 2 250(25 HuejMra:Wad.Iaa1 364(65 algni:JohnGuie-Kura16n 291/31 Anga: NitaBisin 13 0 25~356 Waia* WuetMra-trs1 261/262 Mc~~~rea Tobc~~~ Hubert Murray-Do~ ~~.......1..3 155/156 Hubert~~~~. ..urray: .. i-c . n9 037(058 Ele~each: Durvill...... ... 01~~020 Musgrave:~~~ Eta Bacr.wy9 35235 Wa-cpigGrte-a~rr 0241037 Ela ~~~~~~~~~~~~~eac~~~~~~~: Hunter~~~~~~~~~OurviI~~~~~~~eK. ..: *..~ ~ ~ ~ ~ ~ ~ ~~~~. ..... 4241431 B gotoko Geag-$prng ...... 7: 2511252 H'ubert.. M ray Lahar..a.N..t........ 380(363 Bora~~~~~~~~~~~ga~ino: claEbny-Gnas a7 365(391 Koura WagniGr 6 3 159(160 Hubert Niurray1 U0 ,a: Go'ob- P:: -o6 FIGURE 5 Table from 'Road Accidents Papua New Guinea 1988' 2 1 .1.... ... 1.1 Accident Reduction in PNGHil, ho sn, K a ACCIDENTS 27 *** 5 X 2 xx 2 xx 02 *X 5 xxxx 02 *x 1 X 2 *x 03 *x 5 *~cX= 03 xx 7 *Xx 24 *** 16 *xxxxxxu 21l X 5 * 3 xx 2 xx 3 xx 8 ***xxx is - 24 * 1013 *xxxxx 6 xx3 0 3 xxx 21117*15*162 xx 8 *Xxxoa0x 31* 14 **A K 16 * 0 Total= 422 ~~~~ ------- --------------------- 2 2 KILakrR= 0 -9 10 -19 20 -29 30 -39 40 -49 50 -59 60 -69 70 -79 80 89 90 -99 100 -109 110 -119 120 -129 130 -139 140 -149 150 -159 160 -169 170 -179 180 -189 190 -199 200 -209 210 -219 220 -229 230 - 239 240 -249 250 -259 260 -269 270 -279 280 -289 290 -299 300 -309 310 -319 320 -329 330 -339 340 -349 350 -359 360 -369 370 -379 380 -389 390 -399 400 -409 410 .- 419 420 -429 430 -439 440 -449 450 -459 460 -469 470 -479 480 -489 490 -499 FIGURE 6 Accidents along the Highlands Highway in 1988 - Figure from 'Road Accidents Papua New Guinea 1 988' ------- ---------- ~------- --------------------------------------------------- Hills, Thompson, Kila Accident Reduction in PNGHil, To sn, ia 1FIGURE 7 The process of Accident Blackspot Investigation i 2 3 FIGURE 8 An example of an Accident Sketch and a Site Location Sketch from the records of the Royal Papua New Guinea Constabulary Hills, Thompson, Kila Accident Reduction in PNGHisTosnia Q.5 t.~Cl. Locaflor~i z h I. -.--- =-... ~7. - Collision Diagram prepared by member of DoT staff 2 4 1FIGURE 9 1 Hills, Thompson, Kila =3 S_~ - Accident Reduction in PNG Hills, Thompson, Kila 1.FIGURE 1 0 .Conflict study-rating scale adopted by the TRRL 2 5 CONFLICT REPORT SHEET DATE DIAGRAM OF OCCURRENCE CONFLICT SEVERITY Precautionary conflict (I.e brakiirg for vehicle 6'fl/t~~1t~e~t~ 1 maiting to emerue. precaui~onary lane chrangte. ~~ry GRO- 0,o anricipatoey braking)l Controlled braking or lane change to avoid 2 collision but mittE ample time for the manoeuer Rapid deceleration, lane change or stooding to avod collisionC. rsuting in a near miss dr/ : Emergency braking or imolen1t-sere to avoid coilision resulting in very near miss situation or occurrence 0f a mi-nor collision. C& Cr -Y~ K5~IC.4- &, 5 Evergency action, lollomno by a colision, Accident Reduction in PNGHilTopnia Department of Transport/TRRL-- CO-OPERATIVE ROAD SAFETY RESEARCH PROGRAMME EINGINEERING COUNTERMEASURES UJNDER EVALUATION * Location 0 Date Installed Im~plemnenting Authority Roundabouts -Installed Waigani Drive/Wards Road Waigani Drive/Koura Way Waigani Drive/Goro Keaga Rd Angau Drive/Lahara Avenue Bisini Prde/Kaubebe/Okari Waigani Drive/Cameron Rd Waigani Drive/Spring Garden (Gyratory) Central Ave nue/9th St MilfordHaven Rd/Markham Rd Milford Haven Rd/Air Corps Rd Milford Haven Rd/Malaita St Huon Rd/Burnbu Rd 1983 Aug 85 Sept 85 Dec 85 Dec 85 Feb 86 87 1985 Nov 88 Nov 88 Nov 88 Sept 88 NODC13 NCODIC NCIDIC NCD3IC NCODIC NCODIC NCODIC Lae City M orobeDoW/ DoT M orobeDoW/ DoT MorobeDoW/DoT Lae City Rniinel1qhniit-- -Prnitec,~ted NOD Hubert Murray Hwy/Scratchley Rd NOD Ela Beach Rd/Lawes Rd Lae Milford Haven Rd/Huon Road NCODIC/DoT NCODIC/Dot Lae City * (NCOD Waigani Drive/John Guise Drive -under consideration NCDIC/DoT) Traffic Liahts -Proiected Re-imina * -NOD Hubert Murray Hwy/Taurama Rd/Wards Rd * NOD Hubert Murray Hwy/Boroko Drive (NBC) * NOD Hubert Murray Hwy/Waigani Drive Junction Improvements -implemented * Lae Butibum Rd/Orion St * Lae Central Av/7th St * Lae Huon RdI7th St I nk lmorovements -orrpiecteD 0 0 1985 1985 1 988 NCOD Hubert Murrray Hwy: Scratchley Rd-Gorobe St NOD Waigani Drive: John Guise Drive-Koura Way GorokoHighlands Hwy: Mack St- Brechin St NCD1C4DoT NDCM/DoT NCDIC4DoT Lae City Lac City Lae City NCDI/ DoT NCDI1C/ DoT EHDoWIDoT Pedestrian Footbridge -installed NCOD Hubert Murray Hwy (Turumu St) 0destrian Footpaths -installed * GorokoHighlands Highway * NCOD Hubert Murray Hwy:Taurama Rd-Lahara Av FIGURE 1 1 1988/1989 EH13oW/DoT NCD1C4DoT 2 6 NCOD NCOD NCOD NCOD NCOD NCOD NCOD 0 0 0 0 0 0 0 0 0 0 0 0 Lae Lae Lae Lae Lae 0 0 1 986 NCDIC Hills, Thompson, Kila Accident Reduction-in PNGHil, To sn, ia Chevron Boards -installed 0 GorokoHighlands Hwy (GKO-Kamaliki Ck) NOD Baruni Coastal Rd Gusap Bridge, Ramu Highway 3heed Reduction -projected rumble stip 0 0 Jan 88 Sept 89 Sept 89 NOD Waigani Drive/John Guise Drive NOD Diho Avenue Lane Dividers -implemented * GorokoHighiands Highway Zentral Reservations -implemented Jan 88 NOD Hub Mur-ray Hwy:Taurama Rd-Okari Stl987 NOD Waigani Drive: Tomu Place-lKoani St 1 988 Gflimbng Lane * NOD Gavamani Rd Single Lane Bridge -bar pattern * Umi River, Highlands Highway * Gusap & Boro Rivers, Ramu Highway Bu Bay -implemented * NOD and Lae 1988 Aug 89 1984-1989 EHDoWI DoT NODiC4DoT Morobe DoW/DoT NCDIC4DoT NODIO/DoT EHDoWIDoT NCODIO NODIC NODIC MorobeDoW M orobe DoW! DoT NCDIC/Lae City FIGURE 1 1 (Continued) 2 7 ,Ac._cidents at 2 Roundabouts- 2 00 m apart 1st instala 1983 - 2nd early 1986 35. E15 5 - -----..... FIGURE 1 2 Analysis of accidents at two adjacent roundabouts in Port Moresby Hills, Thompson, Kila