TRANSPORTATION RESEARCH RECORD 1652 103

Accident Characteristics on Low-Volume

Roads in Indonesia
NAWAWI ACHWAN AND DIDIK RUDJITO

Road accident statistics from the Indonesian Traffic Police Headquarters ACCIDENT INFORMATION SYSTEM
for the last decade show that there have been 10,000 fatalities on average
per year. This situation has become a serious national issue. Since 1989 Triple L System
the Overseas Unit of Transport Research Laboratory has been involved
in a collaborative research study on road safety with the Institute of Road
Engineering in Bandung, using the Microcomputer Accident Analysis
Since 1989 the overseas unit of TRL has been involved in a collab-
Package (MAAP). A pilot study to implement the MAAP system was set orative program of road safety research with IRE in Bandung, par-
up in four regions. The characteristics of road accidents on selected low- ticularly by using the Microcomputer Accident Analysis Package
volume roads are described and processes for identifying contributory (MAAP). The MAAP is used in parallel with a Traffic Accident
factors at hazardous locations are introduced. Report Form (TARF) for the purpose of accident data storage,
retrieval, and analysis (5). The MAAP system has been renamed
Lahta Laka Lantas (LLL or Triple L), meaning data processing,
During the last decade, Indonesia has experienced high numbers of
accident, traffic.
road accident fatalities, with figures in the region of 10,000 per year.
The fully computerized version of Triple L was set up as a trial
These statistics, although grievous, are reasonably typical of the
in four areas of Indonesia (6,7 ). These are Bandung City (1989),
impact of the development of a national transportation system, par-
Semarang and Surabaya cities (1991), and the Purwakarta region
ticularly a country’s first period of motorization (1). However, there (1992) for rural areas. During the initial implementation of the pilot
is still considerable scope for road safety improvements in Indonesia. study in Bandung during 1989 and 1990, TARF was designed with
The situation has become a serious national issue, not only because 86 items of data. In early 1991 a new TARF was designed for use in
of the loss of life and the injuries and suffering caused, but also the pilot study areas with 99 data fields.
because of the loss of productivity and the waste of scarce resources
resulting from road accidents. International research (2) indicates that
annual economic losses arising directly from road accidents typically Accident Data
amount to about 1 percent of a country’s gross domestic product. If
this ratio is applied in Indonesia, the total costs would be about U.S. This study examined all the accident data from 1993 to 1995 regis-
$1,000 million per year. tered by the traffic police in the Purwakarta Police Region, which
Thus it is vital that a systematic approach for road safety is devel- includes three districts (Karawang, Purwakarta, and Subang) (8,9).
oped, one that will help reduce both the severity and number of road The data set was corrected to ensure that the number of records on
accidents (3). Development of a road safety approach also is essen- the computer equaled the number in the police registers.
tial for supporting the Directorate General of Highways’s stated
goal, which is improvement of the level of road service in Indone-
Accident Severity and Injury Severity
sia. To meet these targets the Institute of Road Engineering (IRE)
has developed and trialed a road accident database in collaboration
Accidents often are separated according to severity—fatal, serious
with the traffic police, with support from the Transport Research
injury, slight injury, or damage only (10). An accident is classified
Laboratory (TRL) in Great Britain. This first step is very important
according to the most severe injury incurred. For example, an accident
to Indonesia because reliable data are critical both for diagnosis of
in which 2 people are killed and 10 injured is a fatal accident.
the road accident problem and for research and development. Casualties often are divided into killed persons, seriously injured
The database known as Triple L has been implemented as pilot persons, and slightly injured persons. Indonesia adapted the defini-
projects in four regions (three cities and a rural area) (4). This paper tions that a killed person is one who dies within 30 days of the acci-
describes the magnitude of the road accident characteristics on low- dent, and a seriously injured person is one who is admitted to a
volume roads by using data from the rural areas, particularly on kabu- hospital. Slight injuries are minor injuries.
paten (district) collector roads, and introduces the processes for
identifying common contributory factors by using a stick diagram at
hazardous locations. ROAD ACCIDENT CHARACTERISTICS

Low-Volume Roads
N. Achwan, Directorate General of Highways, Ministry of Public Works, Jl.
Pattimura n. 20, Jakarta 12110, Indonesia. D. Rudjito, Institute of Road
The Integrated Road Management System (IRMS) of the Direc-
Engineering, Ministry of Public Works, Jl. Raya Timur no. 264, Bandung torate General of Highways has classified rural roads according to
40294, Indonesia. their function. The IRMS distinguishes among three functions of
104 TRANSPORTATION RESEARCH RECORD 1652

rural roads: national roads, province roads, and local or kabupaten Where?
roads. According to this approach, low-volume roads in Indonesia
are kabupaten roads that typically connect small towns and villages. Most accidents (83 percent) occurred where the road surface was
Low-volume roads in the Purwakarta region situated in the center of in good condition (asphalt pavement). However, it must not be
west Java Island constitute about 85 percent of the road network of assumed that road surface condition has no effect on accidents as the
2323 km, generally are 5 m wide, and commonly carry between 200 police description was based on a simple manual check and not on
and 300 vehicles per day. Terrain of the area through which the road engineering judgment or measurements.
passes is relatively flat. A road condition problem was identified by the police in which
the level of the shoulder was lower than the road surface. This situ-
ation was reported in 66 percent of the accidents. The high incidence
Road Accident Data, 1993–1995 of this factor is probably due to loss of control when the vehicle
wheel drops onto the shoulder. It is clear that this poor design or
Just over 6 percent of the accidents and 4 percent of the fatalities in maintenance condition of low-volume roads could contribute to the
the Purwakarta region occurred on kabupaten roads. From Table 1 high accident figures.
it can be seen that the number of accidents on kabupaten roads rose
from 1993 to 1994 and then dropped for the next year. Indeed, there
is considerable variation in the number of accidents and casualties When?
from one year to the next and, as this is true of fatalities, it seems
unlikely that this is due to any variation in the underreporting of There is an even distribution of accidents throughout the day from
accidents. 6:00 a.m. to 6:00 p.m. and a drop during the night from 6:00 p.m. to
morning. This pattern reflects the periods of most activity on the vil-
lage roads. The number of accidents from Monday to Saturday is
Who? relatively constant, with a slight rise on the weekend.
According to the police, 65 percent of all accidents on low-volume
In the Purwakarta’s kabupaten roads, the largest single group of roads occurred in the hours of daylight, and 19 percent occurred at
casualties are motorcyclists, who accounted for 25 percent of the night on roads without street lighting. Given the lower levels of night-
fatalities and 15 percent of the casualties. Occupants of trucks time usage there is a significant nighttime accident risk and measures
were the next most vulnerable group, with 14 percent of the fatal- such as improved delineation and improved street lighting could have
ities and 1 percent of the casualties. Pedestrians were also a prob- considerable potential for accident reduction.
lem group, representing 9 percent of both the fatalities and the Most accidents occurred in fine weather (76 percent). This is some-
casualties. what surprising and the problem of wet road accidents (6 percent)
The most vulnerable age group for driver casualties was 21– needs to be investigated further. However, drivers tend to slow down
30 years old (44 percent of all casualties, 20 percent of all fatalities, considerably on wet roads and this may be why there are few wet road
24 percent of serious and slight casualties). The reason for young driv- accidents.
ers being injured may be insufficient driving training. The worst age
group for passenger casualties was the 16- to 20-year-olds, with 4 per-
cent of the fatalities and 52 percent of the casualties. Young children How?
(0–10 years) also were frequently involved in accidents and they
accounted for 4 percent of both fatalities and casualties. It seems very The most common collision types on low-volume roads are head-
likely that the lack of safe pedestrian facilities and the poor knowl- on (37 percent), pedestrian hit (26 percent), and rollover (22 per-
edge and behavior of young children coupled with the high vehicle cent). These predominant characteristics are extremely important
speeds on low-volume roads have resulted in this increased risk for and have significant implications for the solutions required. Head-
young children. on collisions are the most dangerous and measures preventing dan-

TABLE 1 Accident Data, 1993–1995

Achwan and Rudjito 105

gerous overtaking such as road markings, road widening, and Preliminary Results
sealed shoulder, plus providing lay-byes, bus stopping areas, and
off-road facilities for stalls, could lead to considerable reductions A summary of the links with three or more accidents is shown in
in overtaking accidents. However, these measures would need to Table 2 and in Figures 2, 3, and 4.
be backed up by information campaigns and enforcement. Sealed The level of accidents over the 3 years is low (a maximum of 5) but
shoulders and safer crossing facilities together with improved road the real number of road accidents is likely to be substantially higher
safety education and awareness could help reduce pedestrian acci- because many of the less-serious accidents would have gone un-
dents. Improving the shoulder should also prevent many of the reported. On the basis of accident frequency alone there appears to
rollover accidents. be no great need to treat any low-volume road link as a special case
According to police reports most vehicles were traveling for remedial measures. However, it is also important to determine
straight ahead before the accident (68 percent). Overtaking was the whether any routes have a typical pattern of accidents as this could be
next most frequent vehicle maneuver, resulting in 16 percent of the due to below-normal road design standards or conditions. These pat-
accidents. tern have been examined for the worst link in each district by using a
Wearing a seat belt is the most important way to prevent injury stick diagram. These diagrams present the key information for each
when traveling by car. In 90 percent of accidents on low-volume accident in a single column or stick and they are useful for identify-
roads occupants were not using seat belts. For road fatalities, 42 per- ing common combinations of dominant characteristics. Table 2 shows
cent of the drivers were not wearing a seat belt. Further investiga- the hazardous locations as identified for each district.
tion is required to gain reliable evidence on the use of seat belts From Figure 5 (Karawang district, road code 79) it can be seen that
because police generally arrive at the scene of accidents well after there were two dominant accident types: three cars hitting pedestri-
the accident occurred. ans and two rollovers. In the Purwakarta district (road code 34; Fig-
ure 6) there were three accidents involving motorcyclists, two of
which were head-on and all of which were fatal. Interestingly, there
Road Accident Costs were no accidents in 1993 or 1995 in this district. In the Subang dis-
trict (road code 7; Figure 7) two of the three accidents involved trucks
Accident costs for low-volume roads in the Purwakarta region from and all three were fatal.
1993 to 1995 were estimated by using the accident cost figures based Although the number of accidents is small, it would appear that
on 1991 prices (fatal, Rp 38,849,000; serious injury, Rp 1,924,000; different road links do have different accident patterns. Through
slight injury, Rp 662,000) determined by a 1992 study (11) and adjust- more research it may be possible to group low-volume roads into
ing them according to cost increases per year of 10 percent for each categories with similar accident problems and then identify common
type of accident. The average accident cost per year is an estimated contributory factors and remedial treatments.
Rp 867 million.

CONCLUSIONS
HAZARDOUS LOCATIONS
• It can be seen from the evidence given that there is a serious
Investigatory Approach road safety problem on low-volume roads in the Purwakarta region.
• These accidents caused significant losses in production, dam-
Research in motorized countries has shown that the systematic iden- age, and suffering, with a total estimated cost per year of Rp 867
tification, investigation, and treatment of hazardous locations can million. The country can ill afford to waste scarce resources. Hence,
produce significant accident reductions and a number of highly cost- all institutions involved in road safety should make every effort to
effective countermeasures have been identified (12,13). It is there- reduce accidents.
• Key vulnerable groups are motorcyclists and pedestrians, with
fore important that the Triple L system enables accident
truck casualties also being a problem on low-volume roads.
investigators to identify high frequency accident locations, analyze
• Rollover is a common collision type and it appears that this is
accident patterns, and determine the contributory factors. To help
caused by poor shoulder conditions in the majority of accidents; it
accident investigators and researchers the following facilities were
is relatively rare on the national highway network.
developed in collaboration with TRL:
• The accident numbers on individual road links over the 3-year
period were low (maximum of 5) but it seems likely that many less-
• Accident location systems for kabupaten roads using the road serious accidents were not reported.
link numbering system from the Directorate General of Highway • Because of these low numbers it is difficult to justify prioritiz-
plus the name of the village (kilometer posts generally are not ing any links for remedial treatment on safety grounds alone. How-
available). ever, clearly more research is needed to determine whether links can
• Triple L software facilities including cross tabulation, route be categorized into groups with similar contributory factors such as
analysis, and stick diagrams (a special technique for determining poor road shoulders and thus to identify potentially cost-effective
accident characteristics and pattern at site). route improvement schemes.
• A systematic methodology for accident investigation. The • The trial of Triple L in the pilot area has proved successful and
methodology was based on the U.K. accident investigation approach. useful. Therefore, it should be adopted on a wider basis, along with
It consists of four stages—identification, diagnosis, selection, a systematic approach to accident investigation, prevention, and
evaluation—and 12 steps (see Figure 1). reduction.
FIGURE 1 Methodology of accident investigation.
 TABLE 2 Hazardous Locations Identified for Each District

FIGURE 2 Road network of Karawang district.

FIGURE 3 Road network of Purwakarta district.

FIGURE 4 Road network of Subang district.

Achwan and Rudjito 109

FIGURE 5 Stick diagram for worst link in Karawang district.

FIGURE 6 Stick diagram for worst link in Purwakarta district.

110 TRANSPORTATION RESEARCH RECORD 1652

FIGURE 7 Stick diagram for worst link in Subang district.

ACKNOWLEDGMENT 6. Rudjito, D., S. Umar, and B. Sargeant. Blackspot Investigation in

Indonesia. In Proc., 2nd Conference on Asian Road Safety, Beijing,
China, 1996.
The authors are grateful for the assistance of and valuable discussions
7. Rudjito, D., M. Idris, and A. J. Downing. Accident Investigation in
with Andrew J. Downing of TRL. Three Cities Pilot Project Areas in Indonesia. In Proc., 9th REAAA
Conference, Wellington, New Zealand, 1998.
8. Rudjito, D. Development of Road Safety Technology. Institute of Road
REFERENCES Engineering, Bandung, Indonesia, 1997.
9. Rudjito, D., and Mardiansyah. Accident Situation in the Four Pilot
1. TRL and Ross Silcock Partnership. Towards Safer Roads in Developing Study Areas in Indonesia. In Proc., International Conference on Road
Countries. Crowthorne, Berkshire, U.K., 1991. Safety, Visakhapatnam, India, 1997.
2. Downing, A. J., C. J. Baguley, and B. L. Hills. Road Safety in Devel-
10. Indonesia Highway Capacity Manual. Traffic Safety and Environmental
oping Countries: An Overview. Overseas Centre, Transport Research
Laboratory, Crowthorne, Berkshire, U.K., 1991. Impacts, WP Report A:2.1. Bandung, Indonesia, 1995.
3. Downing, A. J. Engineering Approach to Accident Prevention & Reduc- 11. IRE and TRL. Accident Cost Study. Technical Assistance Research
tion. Road Research Development Project, Institute of Road Engineering, Project Phase II. Bandung, Indonesia, 1992.
Bandung, Indonesia, 1997. 12. IRE and TRL. Interim Manual on Accident Investigation Procedures
4. Thompson, N. M., and D. Rudjito. Traffic Accident Investigation in and the Development of Low Cost Engineering Improvement Schemes,
Indonesia. In Proc., 7th REAAA Conference, Singapore, 1992. Part I and II. Bandung, Indonesia, 1992.
5. Hills, B. L., and G. J. Elliot. A Microcomputer Accident Analysis Pack- 13. Royle, B., J. Bromnfield, D. Lynam, D. Rainbird, and B. Sargeant. High-
age and Its Use in Developing Countries. Transport and Road Research way Safety Accident Reduction and Prevention, 2nd ed. The Institute of
Laboratory, Crowthorne, Berkshire, U.K., 1986. Highways and Transportation, London, U.K., 1986.