See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/244692585

Process industry accidents in Sri Lanka: Analysis and basic lessons learnt

Article in Process Safety and Environmental Protection · November 2008

DOI: 10.1016/j.psep.2008.05.002

CITATIONS READS

14 2,054

2 authors, including:

Ajith de Alwis
University of Moratuwa
84 PUBLICATIONS 1,496 CITATIONS

SEE PROFILE

Some of the authors of this publication are also working on these related projects:

Municipal Solid Waste Management in Local Authorities in Sri Lanka View project

Analysis of Toxic Elements present in Plastic Toys available in Sri Lankan Market using X-Ray Fluorescence Spectrometer View project

All content following this page was uploaded by Ajith de Alwis on 21 July 2020.

The user has requested enhancement of the downloaded file.

 process safety and environment protection 8 6 ( 2 0 0 8 ) 421–426

Contents lists available at ScienceDirect

Process Safety and Environment Protection

journal homepage: www.elsevier.com/locate/psep

Process industry accidents in Sri Lanka: Analysis and basic

lessons learnt

M.Y. Gunasekera ∗ , A.A.P. de Alwis

Department of Chemical & Process Engineering, University of Moratuwa, Moratuwa, Sri Lanka

a b s t r a c t

Accidents in the process industry could be prevented or reduced by having good safety management measures. Such
preventive measures could be further improved through the experiences and lessons learnt from past accidents.
Therefore, analysis results of past accidents are valuable sources of information for determining root causes and as
case material to prevent and reduce the adverse consequences of accidents in the process industry.
This paper looks at accidents in the process industry that have occurred in the past 10 years from 1997 to 2006
in Sri Lanka to gain an understanding of the nature and consequences of accidents. Lessons and main areas of
concern to improve safety in the Sri Lanka process industry are discussed. The analysis is done for different event
types based on specific operating process stage during which the accident occurred such as processing, loading and
unloading, repair and maintenance and storage, the immediate effect types such as fire, explosion, chemical releases
and emissions and the consequences of each accident. Fire incidents were observed in 38 accidents analyzed. The
results show that the highest number of accidents has occurred during processing operations followed by accidents
during maintenance and repair work. The cause analysis shows that many accidents have occurred due to technical
and human failures.
The accidents are then classified according to the severity of the consequences in order to compare the nature of
accidents experienced in Sri Lanka with respect to accidents in other countries in the world.
© 2008 The Institution of Chemical Engineers. Published by Elsevier B.V. All rights reserved.

Keywords: Accident; Sri Lanka; Process industry; Database

1. Introduction is the methyl isocyanate gas leak in Bhopal, 1984. About 3000
people died because of this highly toxic gas release (Gupta,
Past experiences of process industrial accidents have made 2002). The accident occurred in the Flixborough, UK where
the public more aware and concerned of the negative conse- Caprolactum was manufactured resulted in 28 fatalities and
quences involved in them (Roed-Larsen et al., 2004; Christen a huge explosion and a fire which destroyed the whole plant
et al., 1994). Today the public is highly sensitised in this regard. (Marshall, 1987). The Seveso accident happened in 1976 in a
Hazards involved in process plants very often bring public pesticides and herbicides manufacturing plant and released a
protest. Therefore, it is important to adopt safety measures to dense vapour cloud containing tetrachlorodibenzoparadioxin
prevent, mitigate or control such hazards. Measures taken by (TCDD) (Marshall, 1987; Lees, 1996) to the environment. This
way of safety management lead to the prevention of accidents substance was a carcinogenic by product from an uncontrolled
and thus the losses involved. Further, sound safety control exothermic reaction that occurred in a chemical reactor. About
measures installed require continual maintenance through- 600 people had to be evacuated from the contaminated site
out the life of the plant and continuous training to educate or and many were treated for dioxin poisoning. This accident
to remind the staff of necessary safety techniques and policies led to the development of Seveso Directives in the European
in the process plant. Union.
Many accidents happened in the past have resulted in dev- With the experience of these severe accidents, the
astating consequences. One such major accident experienced industrial organizations and nations have adopted many

∗
Corresponding author.
E-mail address: manisha@cheng.mrt.ac.lk (M.Y. Gunasekera).
Received 12 September 2007; Accepted 13 May 2008
0957-5820/$ – see front matter © 2008 The Institution of Chemical Engineers. Published by Elsevier B.V. All rights reserved.
doi:10.1016/j.psep.2008.05.002
422 process safety and environment protection 8 6 ( 2 0 0 8 ) 421–426

improvements in safety methods and also legislation aimed at of 10 years from 1997 to 2006 that were published in local
prevention and control of such accidents. These approaches newspapers in Sri Lanka. The accident records in AcciHaz con-
try to avoid or eliminate hazards, or reduce their magni- tain data on, the name of the plant, accident location, date
tude, severity or likelihood of occurrence. Such improvements of the accident, chemical(s) involved, description of the inci-
in safety have shown results of reduced number of acci- dent, damages to the people, property, animals, ecosystem
dent and losses (Gardener, 2003). To further develop modern and the environment, costs of losses, cause of the accident
safety devices and methods, lessons from past accidents could and sources of information. Further, data on accident statis-
be used (Uth and Wiese, 2004). Lessons and causes of acci- tics, safety award winners in the past years and information
dents are well understood by the analysis of past accidents. on safety-related regulations in Sri Lanka are also available in
Such analyses are valuable sources of information not only AcciHaz.
to improve existing safety measures but also to prevent sim- In this work data from AcciHaz database were taken for
ilar accidents from occurring again (Roed-Larsen et al., 2004). analysis. Although the information reported by newspaper
Many other facts such as knowledge of the primary causes of reporters may not be very accurate, in the absence of any
accident, information on consequences and impacts and how other accurate source of information this source of informa-
to respond to accidental situations also could be obtained by tion was considered to be reasonably accurate and the most
these sources and are essential data in developing successful acceptable. Newspaper reporters of different newspapers have
safety management systems in the process industry. reported the same incident in a similar manner. Further, some
In this paper, major accidents observed in Sri Lanka during accidents such as those resulting in a death or deaths always
the past 10 years from 1997 to 2006 are analyzed. The current involve a police investigation and a court hearing. Some of the
situation of accidents experiencing in this country, the lessons data in the newspaper reports were based on these court hear-
that can be learnt, causes and measures that could be adopted ings. Therefore, it can be assumed that the information taken
to prevent the harmful consequences are also discussed. from the newspaper articles as reasonably accurate. It is also
assumed that the media attention is usually received for major
accidents. Therefore, the accident records in this database are
2. Process industry accidents in Sri Lanka considered as major accidents in this country. These reports
have been made during the first few days of occurrence of
Sri Lanka is an industrially developing economy where the the accident. Therefore, the information is not available in a
process industry is largely found developed as small- and comprehensive form and thus, these reports are considered
medium-scale industries. The number of employees in these as equivalent to a preliminary report of the accident.
industries is less than 50 and the processing techniques used
are often less complex.
3. Analysis of accidents
Industrial accidents in Sri Lanka are reported to the
Department of Labour. According to the Factories Ordinance,
Events occurred during the period 1997–2006 were studied.
accidents occur in factories which causes loss of life to
A total number of 96 incidents occurred during this period
a person employed, or disables any such person for more
have been recorded in the database. These include all acci-
than 3 days or make any such person unconscious must
dents in the process industry including mining industry and
be reported in writing to the factory inspecting engineer
household process industries and also the accidents during
(Factories Ordinance of 1942). Further this ordinance requires
transportation and distribution of raw material and finished
notification of dangerous occurrences in factories (Factories
goods related to the process industry. Other accidents such
Regulations, 1965), which involves fire and explosions, col-
as those related to road traffic and civil construction industry
lapses and failures of some machinery and equipment. The
were not considered in this work.
information on accidents that cover above incidents is avail-
able as compiled data with the Labour Department of Sri
Lanka. However, the data reported by the employer on these 3.1. Event types and consequences
accidents is done on a form issued by the regulator and it
does not require the employer to give a detailed description The operating processes at the time of accident and the imme-
of the accident. Therefore, in this study other sources of data, diate effects are used in studying the type of event. The
which provides much more detailed and accurate information operating processes at the time of accident could be classified
on process industrial accidents in Sri Lanka, were referred. as processing, loading or unloading, maintenance and repair,
Major accidents in the country usually grab the attention onsite or off site transportation and storage. Fig. 1 shows the
of the local newspaper reporters and most of these get pub- number of accidents that come under each type of operating
lished in the local press. Therefore, local newspapers were process.
referred for the past 10 years from 1997 to 2006 in order to Some accidents were not classified under any of the above
gather articles on accidents. The newspapers namely, ‘Daily operations because they have occurred during non-operating
Mirror’, ‘Daily News’, ‘The Island’ and ‘Sunday Observer’ were time or the set of operating conditions was not known. There
the main news papers referred. Most of the accidents were are 18 such accidents falling under unknown operating con-
reported in all or at least two of the newspapers and the acci- ditions. Many of the accidents appear to have occurred during
dent information in those records were found to be consistent. the time of manufacturing in the plant. Accidents during
The newspaper collections in the National Library of Sri Lanka maintenance or repair work were the next highest with 11
and the National Archives of Sri Lanka were the main sources accidents and the lowest accident number was observed with
of information for this study. onsite transportation operations.
The information on these articles was then compiled and The immediate effects involved in the accident are con-
a database called ‘AcciHaz’ was developed (Narangoda and sidered as an emission to the atmosphere, fire, explosion or a
Gunasekera, 2004). This database contains accident records release of a chemical substance or substances to the aquatic or
 process safety and environment protection 8 6 ( 2 0 0 8 ) 421–426 423

Fig. 2 – Consequences of accidents.

Fig. 1 – Accident event type.
noted that these results are based on newspaper reports and
the terrestrial environment or to both. These effects could be the priorities placed by newspapers on the types of events they
seen as a separate effect or a combination of several. In this report also influence the results of this analysis.
study a total of 60 accidents have been observed which are The accidents involving above impacts are further ana-
associated with the above effects and 36 accidents without lyzed into accidents having various impacts in combination
any of the above immediate effects. Some of these accidents (Table 2).
involve liquids or gaseous chemical substances but some do
not involve any such substance. The distribution of the num-
Table 2 – Accidents having various impacts in
ber of accidents based on their immediate effects (Table 1) combination
indicates that the fire type incidents involving only a fire
Accidents with fatalities 49
effect are the highest with 34 incidents. A total of fire effect
Only fatalities 30
is observed in 38 incidents including incidents involving only Fatalities/human injuries 14
fire effect, fire effect with explosion effect and fire effect with Fatalities/material losses 1
chemical release. An accident which involves separate emis- Fatalities/human injuries/material losses 3
sion, explosion or release effect varies between 10% and 11.6% Fatalities/human injuries/material 1
of the total 60 accidents with above effects. However, the losses/damaged ecosystem
Damaged ecosystem 7
effects in combination are observed in small numbers. The
Only damages to ecosystem 2
number of accidents involved in each effect category for the
Human injuries/damaged ecosystem 1
60 accidents, which had shown the above effects, is given in Human injuries/material losses/damaged 1
Table 1. Here the analysis of accidents based on immediate ecosystem
effects is exclusive to one category. Damaged ecosystem/material loss 2
The analysis of accidents based on the consequences was Fatalities/human injuries/material 1
done distributing them as fatalities, injuries to human, dis- losses/damaged ecosystem
Community disruption 2
ruption to the community in a way of evacuation of people,
Only Community disruption 1
ecological harm which includes impacts on animals, vegeta- Material losses/community disruption 1
tion, soil and water and other elements in the environment.
Accidents with no impacts reported 7
Losses due to destruction of material or equipment are also
considered in the analysis of consequences. Of the accidents Human injuries 37
studied in this work, the impact that is observed in highest Only human injuries 10
Fatalities/human injuries 14
number of accidents is the human fatalities and next high-
Human injuries/material losses 7
est is injuries to humans. Material losses are also seen in a
Human injuries/damaged ecosystem 1
considerable number of accidents. Some of the accidents are Human injuries/material losses/damaged 1
in more than one impact category because several impacts ecosystem
have resulted from those accidents. These results of analyz- Fatalities/human injuries/material losses 3
ing impacts due to accidents are shown in Fig. 2. It should be Fatalities/human injuries/material 1
losses/damaged ecosystem
Material losses 31
Table 1 – Types of accidents based on immediate effects Only material losses 15
Material losses/fatalities 1
Effect category Number of accidents %
Human injuries/material losses 7
Fire 34 56.7 Damaged ecosystem/material loss 2
Explosion 6 10.0 Human injuries/material losses/damaged 1
Emission 7 11.6 ecosystem
Release 6 10.0 Fatalities/human injuries/material losses 3
Fire/explosion 3 5.0 Material losses/community disruption 1
Emission/explosion 2 3.3 Fatalities/human injuries/material 1
Emission/release 1 1.7 losses/damaged ecosystem
Fire/release 1 1.7 Accidents with no impacts reported 7
424 process safety and environment protection 8 6 ( 2 0 0 8 ) 421–426

Table 3 – Accidents where chemical substances are

involved
Chemical substance Number of accidents

Fuel oil
Petrol 1
Diesel 2
Kerosene 2
Fuel oil type not mentioned 3
Sub total 8
Fig. 3 – Accident percentages with different causes.
Chemical (other than fuel oil)
Acetylene 1 ure in a component due to mechanical damage or corrosion.
Chlorine 1
In this work, highest number of causes are found to be due
Benzene substance 1
to technical failures. There are 34% of the accidents having
Hydrochloric acid 1
Carbide substances 2 unknown causes which is a number similar to the technical
Sub total 6 failures. Human failures account for 29% of the accidents. The
natural causes such as lightening has caused fire in one inci-
Chemical name not mentioned
Poisonous gas 5 dent. This analysis was done based on accident records made
Flammable gas 1 during the first few days of accident occurrence. Therefore,
Chemicals not described 6 the results are mainly based on the preliminary data avail-
Sub total 12 able at that time. A detailed investigation could have revealed
Flammable substances more information on the cause of accidents where the cause
Fireworks industry 6 is unknown. The distribution of cause of accident for all the
Matchstick industry 2 accidents in this study is shown in Fig. 3.
Sub total 8
4. In-depth analysis of the accidents
Total 34

In effect types analysis a large number of fire incidents were

3.2. Accidents involving chemical substances observed. A more in-depth analysis shows that these fire inci-
dents whether resulting individually or with another type or
The chemical substances involved in accidents are mentioned types of effects such as explosion or release are triggered
only in few instances. Therefore, the number of chemical sub- by human failures and technical failures. Six incidents with
stances associated with the accidents is also few. One of the human failure and seven incidents causing technical failure
common substances is fuel oil. This includes diesel, kerosene were observed. However, from a total of 38 fire incidents in
and petrol oils. There are eight incidents involving fuel oil and the database, 24 incidents did not disclose a reason for its
most of these incidents have shown a fire as an immediate cause. The data used in this work were obtained from reports
effect. Six incidents have indicated the chemical substance that have been made immediately after the incidents. A fur-
other than fuel oil involved in the accident. Five poisonous gas ther comprehensive investigation to these incidents must be
incidents and one flammable gas incident have been reported, done in order to determine the causes. One fire incident was
though the gases involved are not mentioned. From the acci- reported under environmental causes which had occurred due
dents involving chemicals, six accidents have not indicated to a lightening strike.
the chemical substance involved. The number of accidents Accidents with fire effect were looked into with respect to
where chemical substances are involved is shown in Table 3. their operation at the time of the accident. Although the oper-
In many incidents with fire as an immediate effect, destruc- ation carried out when the accident had occurred was not
tion of the entire plant was observed. The inventory of known for 14 incidents, the fire accidents had mainly occurred
materials in these plants very often includes chemical sub- during a processing operation within the manufacturing plant.
stances as raw material in manufacturing goods. For example, The number of fire incidents under various operating condi-
plastic goods manufacturing or fiberglass boat manufacturing tions is shown in Table 4.
involves many chemical substances and a fire in such a plant Further analysis shows that human failures and technical
could burn these substances producing noxious gases and failures have occurred mainly during a manufacturing process
hazardous residues. For these accidents, the data on chemical stage. Out of the 33 accidents caused by technical failure, 17
substances involved are not available in the reports obtained have happened during processing. While carrying out main-
from the newspapers. Therefore, it is recognized that for more
complete and accurate analysis, the detailed or full reports
resulting from accident investigations should be referred.
Table 4 – Number of fire effect accidents under different
operating conditions
3.3. Causes of accidents
Operation Number of fire accidents

Causes of the recorded accidents were analyzed based on the Processing 13

factors human failure, technical failure, physical or chemical Loading/unloading 2
reactions, environmental or as unknown. The human failures Maintenance/repair 4
Transportation (onsite/offsite) 1
considered are due to factors such as operating errors and
Storage 4
inadequate organizational procedures. Technical failure cause Unknown 14
is taken as if there was an inadequate design or process, a fail-
 process safety and environment protection 8 6 ( 2 0 0 8 ) 421–426 425

intensifying training on safety, ‘inherent safety’ technology

applications would be more appropriate in reducing fire haz-
ards involved.
Authors believe that the importance of maintenance,
awareness of the inventory involved in the plant and their
chemical, physical characteristics and the safety culture pre-
vailing in the country are key factors that must be looked at
in preventing accidents in the future. Absence of information
collection mechanism to gather comprehensive data on acci-
dents in all types of process industries in Sri Lanka is a major
draw back in achieving a way towards an industrially safer
nation. An accident data collection system that can centrally
store and analyze facts would help in determining the lessons
that could be learnt and help in understanding why similar
accidents happen. Works of other researchers have shown
Fig. 4 – Accidents caused by human and technical failure that data of all types of accident when collected centrally facil-
under different operations. itate effective analysis and evaluation (Uth and Wiese, 2004;
Basso et al., 2004; Kirchsteiger, 1999).
tenance and repair work, human failures have been observed
in more situations than technical failures.
The distribution of total number of accidents, 28 caused 6. Classifying accidents
by human failures and 33 caused by technical failures with
respect to the type of operation during which the accident has The harmful consequences of accidents in the process indus-
occurred are shown in Fig. 4. try are seen as various impacts. The magnitudes of these
impacts vary depending on the seriousness of the effect. The
5. Main areas of concern impacts of some of the world’s disasters such as the Bhopal
and the Flixborough have shown very serious impacts killing
Maintenance of plant equipment at appropriate time is impor- many people, animals and also damaging the ecosystem and
tant in avoiding accidents due to technical failures. In the natural resources (Marshall, 1987). Many methods have been
analysis of many accidents when happened during main- proposed by researchers for classifying such accidents.
tenance, repairing or inspection, it was found that proper The ‘Gravity Scale’ (Kirchsteiger, 1998; Amendola et al.,
isolation of the equipment had not been done. Accidental 1994) approach is used in classifying accidents reported to
switching on of the electricity supply, entering into a vessel the Major Accident Reporting System (MARS) operated by the
for repair without isolation for hazardous substances, carrying Major Accident Hazards Bureau of the European Commission.
out welding work without isolating for flammable substances Similar approach has been taken by Basso et al. (2004) to define
are some examples. types of incidents in their work of reviewing safety man-
When private contractors were hired to carry out main- agement system by incident investigation and performance
tenance or repair work it appears that most of them were indicators. They have classified accidents as major accidents,
not aware of the hazards associated with the work that accidents and near miss events. The criteria used by Basso et
they have undertaken to carryout. It is also seen that during al. (2004) were the same as those contained in the Annex VI of
maintenance operations proper precautions and procedures the Seveso II Directive.
necessary to prevent hazardous situations are not taken. Morrison (2004) has classified accidents based on severity
The incidents studied in this work where the immediate as ‘incidents’, ‘serious incidents’, ‘major incidents’ and ‘criti-
cause of failure was human error, have shown root causes cal incidents’. The type of effects and damage losses have been
such as lack of awareness of proper safety precautions needed used in this classification.
and inadequate or absence of proper maintenance and repair Christen et al. (1994) have classified accidents based on the
procedures. In many incidents the root cause of the problem magnitude of the impacts due to the accident. The impact on
was not known. The workers were not aware of the basic risks man, animals, ecosystem, natural resources and property are
involved in the operation that they were carrying out. These considered and a ‘disaster value’ is defined in order to classify
industries lack expert knowledge and qualification in accident accidents.
prevention. Insufficient knowledge of the operation condition The accidents in the AcciHaz (Narangoda and Gunasekera,
during the incident stresses the need for carrying out training 2004) database were attempted to classify according to the
sessions on a continuous basis. disaster value proposed by Christen et al. (1994). In deter-
There were evidences of recurrence of similar type acci- mining the disaster value impacts resulting from an event is
dents in different industries of same nature as well as in the described by nine indicators. They are namely number of fatal-
same installation. Fire incidents in the fireworks industry have ities, injuries, evacuees, dead animals, alarmed duration, area
been observed in six incidents. Some of these works are unau- damaged ecosystem, area of contaminated soil, area of pol-
thorized installations carrying out fireworks manufacturing luted groundwater and property damaged. Based on the actual
process. The fireworks manufacturing industry uses materi- impacts observed each of these indicators are given an impact
als which are inherently flammable and could be considered value. All nine impact values are then combined and a disaster
as an industry that could experience fire incidents if proper fire value is obtained which indicates how catastrophic the acci-
precautions are not taken. However, according to Sonnemans dent occurred is. The disaster value varies from 0 to 1. A value
and Korvers (2006) accidents still occur in spite of the out- 0–0.1 indicates a ‘normal operation’ condition and a value
standing safety performances. In such industries rather than 0.1–0.3 indicates an ‘incident’ event. Disaster value between
426 process safety and environment protection 8 6 ( 2 0 0 8 ) 421–426

that the current system must be altered to be able to gather

Table 5 – Accident classification
and collect process industrial accident data centrally. Legisla-
Event classification Number of incidents tion should be in place to collect these data and upkeep for
Normal operation 37 continuous updating.
Incident 13 When compared to accidents observed in the world many
Severe accident 2 of the incidents taken place in Sri Lanka are not severe. How-
Catastrophe 0 ever, a couple of severe accidents also have been experienced.
The severity of these accidents could be better determined
values 0.3 and 0.5 is a ‘severe accident’ and above 0.5 the event using currently available accident classifying methods if more
is considered as a catastrophe. According to Christen et al. data related to the accidents are made available.
(1994) the Bhopal incident in 1984 is a catastrophic accident
with a disaster value 1.0. The disaster happened in Houston
References
in 1976 (Christen et al., 1994) with a road tanker containing
ammonia is considered as a ‘severe accident’.
Amendola, A., Francocci, F. and Chaugny, M., 1994, Gravity scales
The accidents studied in this work did not have data for
for classifying chemical accidents, In Proceedings of the 7th
determining all nine indicators given in the work of Christen ESReDA Seminar on Accident Analysis
et al. (1994). Therefore, only the data on fatalities and injuries Basso, B., Carpegna, C., Dibitonto, C., Gaido, G., Robotto, A. and
were considered in this classification. A total number of 52 Zonato, C., 2004, Reviewing the safety management system
accidents had consequence data on these indicators to deter- by incident investigation and performance indicators.
mine the disaster value. Many of the incidents are classified Journal of Loss Prevention in the Process Industries, 17:
225–231.
as ‘normal operation’ according to this analysis. Two of the
Christen, P., Bohnenblust, H. and Seits, S., 1994, A methodology
accidents were severe accidents and events of catastrophic
for assessing catastrophic damage to the population and
nature were not observed. The number of incidents for each environment: a quantitative multi-attribute approach for risk
event type is shown in Table 5. analysis based on fuzzy set theory. Process Safety Progress,
Although the above classification indicates a large num- 13(4): 234–238.
ber of incidents as ‘normal operation’, authors consider these Factories Ordinance of 1942, Notification of accidents Section 61,
incidents also as accidents with significant impacts though Sri Lanka.
Factories Regulations, 1965, Factories Ordinance of 1942,
they may not have impacts as severe as the impacts observed
Notification of Dangerous Occurrences, Sri Lanka.
in major catastrophes such as the incident in Bhopal, India. Gardener, N.J.L., 2003, Safety improvement through learning from
For some accidents analyzed in this work, the property incidents, HAZARDS XVII, Symposium Series No. 149,
damages and material loss costs data are available as initial IChemE, pp. 205–216.
estimates. The costs of the property damages data available Gupta, J.P., 2002, The Bhopal gas tragedy: could it have happened
are based on the property values in Sri Lanka. Therefore, these in a developed country? Journal of Loss Prevention in the
Process Industries, 15: 1–4.
values were not attempted to use in the disaster value deter-
Kirchsteiger, C., 1998, Absolute and relative ranking approaches
mination because the impact values for property damage used
for comparing and communicating industrial accidents.
in the Christen et al. (1994) work were not based on prop- Journal of Hazardous Materials, 59: 31–54.
erty in values in Sri Lanka. The property damage and other Kirchsteiger, C., 1999, The functioning and status of the EU’s
factors also must be included in a good classification sys- major accident reporting system on industrial accidents.
tem. Therefore, it is recognized that accident classification Journal of Loss Prevention in the Process Industries, 12: 29–42.
methods should be developed to suite more specifically to the Lees, F.P., (1996). Loss Prevention in the Process Industries (2nd
edition). (Butterworth-Heinemann, UK).
country concerned for understanding the nature of accidents
Marshall, V.C., (1987). Major Chemical Hazards. (Ellis Horwood
better. The technique used in this paper is recommended to
Limited, Chichester).
be adopted for Sri Lanka because data available at this stage Morrison, L.M., 2004, Best practices in incident investigation in
is not adequate to use a complex method. the chemical process industries with examples from the
industry sector and specifically from Nova Chemicals. Journal
7. Conclusions of hazardous Materials, 111: 161–166.
Narangoda, N.A.C. and Gunasekera, M.Y., 2004, Development of a
database for the process industrial accidents in Sri Lanka, In
Accidents are caused mainly due to technical failures and
Proceedings of the Institute of Engineers Sri Lanka (IESL), 98th
human failures. Absence of proper safety measures is the Annual Sessions
main cause of these accidents. Lack of awareness of the safety Roed-Larsen, S., Valvisto, T., Harms-Ringdahl, L. and Kirchsteiger,
precautions required and inadequate expert knowledge are C., 2004, Accident investigation practices in Europe—main
very evident in the accident data available. Although there responses from a recent study of accidents in industry and
are few organizations which are highly organized with respect transport. Journal of Hazardous Materials, 111: 7–12.
Sonnemans, P.J.M. and Korvers, P.M.W., 2006, Accidents in the
to safety, most companies are small-scale and medium-scale
chemical industry: are they foreseeable? Journal of Loss
installations with informal safety routines. Prevention in the Process Industries, 19: 1–12.
These accidents analyzed are not all the accidents that Uth, H.J. and Wiese, N., 2004, Central collection and evaluating of
have occurred during the past decade in this country but a major accidents and near miss events in the Federal Republic
small number that were severe enough to bring the attention of Germany—results, experiences, perspectives. Journal of
of the public. In order to learn all the events authors believe Hazardous Materials, 111: 139–145.

View publication stats