ArticlePDF Available

A Comparative Analysis of Accident Risks in Fossil, Hydro, and Nuclear Energy Chains

Authors:

Abstract and Figures

This study presents a comparative assessment of severe accident risks in the energy sector, based on the historical experience of fossil (coal, oil, natural gas, and LPG [Liquefied Petroleum Gas]) and hydro chains contained in the comprehensive Energy-related Severe Accident Database (ENSAD), as well as Probabilistic Safety Assessment (PSA) for the nuclear chain. Full energy chains were considered because accidents can take place at every stage of the chain. Comparative analyses for the years 1969–2000 included a total of 1870 severe (≥ 5 fatalities) accidents, amounting to 81,258 fatalities. Although 79.1% of all accidents and 88.9% of associated fatalities occurred in less developed, non-OECD countries, industrialized OECD countries dominated insured losses (78.0%), reflecting their substantially higher insurance density and stricter safety regulations. Aggregated indicators and frequency-consequence (F-N) curves showed that energy-related accident risks in non-OECD countries are distinctly higher than in OECD countries. Hydropower in non-OECD countries and upstream stages within fossil energy chains are most accident-prone. Expected fatality rates are lowest for Western hydropower and nuclear power plants; however, the maximum credible consequences can be very large. Total economic damages due to severe accidents are substantial, but small when compared with natural disasters. Similarly, external costs associated with severe accidents are generally much smaller than monetized damages caused by air pollution.
Content may be subject to copyright.
Human and Ecological Risk Assessment, 14: 947–973, 2008
Copyright
C
Taylor & Francis Group, LLC
ISSN: 1080-7039 print / 1549-7680 online
DOI: 10.1080/10807030802387556
A Comparative Analysis of Accident Risks in Fossil,
Hydro, and Nuclear Energy Chains
Peter Burgherr and Stefan Hirschberg
Paul Scherrer Institut (PSI), Laboratory for Energy Systems Analysis, Villigen PSI,
Switzerland
ABSTRACT
This study presents a comparative assessment of severe accident risks in the en-
ergy sector, based on the historical experience of fossil (coal, oil, natural gas, and
LPG [Liquefied Petroleum Gas]) and hydro chains contained in the comprehensive
En
ergy-related Severe Accident Database (ENSAD), as well as Probabilistic Safety As-
sessment (PSA) for the nuclear chain. Full energy chains were considered because
accidents can take place at every stage of the chain. Comparative analyses for the
years 1969–2000 included a total of 1870 severe (5 fatalities) accidents, amount-
ing to 81,258 fatalities. Although 79.1% of all accidents and 88.9% of associated
fatalities occurred in less developed, non-OECD countries, industrialized OECD
countries dominated insured losses (78.0%), reflecting their substantially higher in-
surance density and stricter safety regulations. Aggregated indicators and frequency-
consequence (F-N) curves showed that energy-related accident risks in non-OECD
countries are distinctly higher than in OECD countries. Hydropower in non-OECD
countries and upstream stages within fossil energy chains are most accident-prone.
Expected fatality rates are lowest for Western hydropower and nuclear power plants;
however, the maximum credible consequences can be very large. Total economic
damages due to severe accidents are substantial, but small when compared with nat-
ural disasters. Similarly, external costs associated with severe accidents are generally
much smaller than monetized damages caused by air pollution.
Key Words: ENSAD database, severe accident, energy sector, comparative risk as-
sessment, external costs.
INTRODUCTION
People have always been exposed to a multitude of accident and catastrophe
hazards that can lead to disastrous consequences. Natural disasters such as floods,
windstorms, and earthquakes claim thousands of victims every year worldwide. The
consequences of man-made accidents (e .g ., transportation, major explosions and
Received 25 April 2007; revised manuscript accepted 7 January 2008.
Address correspondence to Peter Burgherr, Paul Scherrer Institut (PSI), Laboratory for En-
ergy Systems Analysis, CH-5232 Villigen PSI, Switzerland. E-mail: peter.burgherr@psi.ch
947
P. Burgherr and S. Hirschberg
fires, collapses of buildings and other infrastructures, releases of chemical and/or
toxic substances, and mining) also regularly disrupt society, and have become in-
creasingly important with the steady growth of industrialization, urbanization, and
the interdependencies of complex infrastructures.
It appears that the number and related damage of natural catastrophes and man-
made accidents have increased since the 1970s. This trend has been recognized by
different stakeholders including researchers (Hirschberg et al. 1998), international
organizations (EM-DAT 2006), and the worldwide operating reinsurance companies
(Munich Re 2005; Swiss Re 2004). This general rise together with specific large loss
events have in recent years led to increased international recognition of risk assess-
ment as an effective, science-based set of tools for tackling the concern by the public
and private sectors in safety, health, and environmental problems (e .g ., Ferguson
and Kasamas 1999; Flynn et al. 2001). This enhanced awareness has significantly
increased the demand for comprehensive data on disasters and accidents by a vari-
ety of stakeholders such as industry and the service sector, government authorities,
development planners, as well as public interest and citizen groups.
A number of initiatives and detailed studies concerning risks emerging from nat-
ural hazards have been undertaken by international organizations and the reinsur-
ance sector to assess risk levels and to delineate vulnerable areas at different spatial
scales. Recent examples include studies such as Natural Disaster Hotspots—A Global Risk
Analysis (Dilley et al. 2005), Reducing Disaster Risk: A Challenge for Development (UNDP
2004), and the Web-based Natural Hazards Assessment Network NATHAN based on Mu-
nich Re’s NatCatSERVICE (Munich Re 2003). Reporting of industrial accidents is
often regulated by national and supra-national frameworks. For example, companies
are obliged to report accidental events from industrial activities falling under the
SEVESO II Directive
1
of the European Union (EU), allowing in-depth analysis of
accident frequencies and consequences (Nivolianitou et al. 2006; Papadakis 2000).
Although accidents in the energy sector have been shown to form the second largest
group of man-made accidents (after transportation), their level of coverage and
completeness was not satisfactory because they were commonly not surveyed and
analyzed separately, but just as a part of technological accidents (Hirschberg et al.
1998).
In the 1990s the Paul Scherrer Institut (PSI) started a long-term research activ-
ity to close this gap and to enable a factual and appropriate treatment of accident
risks in the energy sector. The analytical approach encompassed fossil, hydro, and
nuclear energy chains because all of them entail some significant forms of health,
environmental, or sociopolitical risks. Consideration of complete energy chains is
essential because accidents not only occur during actual energy production, but in
every stage of an energy chain from exploration to extraction, refining, storage,
distribution, and finally waste disposal. Severe accidents are most controversial in
public perception and energy politics. Therefore they are the main focus of investi-
gations, even when the total sum and associated impact of the many small accidents
1
The ”Seveso II” Directive (96/82/EC), developed by the European Commission (EC), pro-
vides guidelines on a “Major Accident Prevention Policy and Safety Management System,”
that is, aiming at the prevention of major accidents involving dangerous substances, and the
limitation of their consequences.
948 Hum. Ecol. Risk Assess. Vol. 14, No. 5, 2008
Comparison of Severe Accident Risks in the Energy Sector
with minor consequences is substantial. For a detailed discussion comparing severe
and smaller accidents we refer to Burgherr et al. (2004).
A comprehensive and undistorted comparative assessment requires the objective
expression of accidents and risks on the basis of extensive data collection and eval-
uation. Considerable difference in the magnitude, timing, and nature of associated
risks can be expected among the various energy chains. It is this difference that
allows a degree of choice in the decision-making process, with regard to selecting
energy alternatives, decisions on energy policies and achieving safety goals. Custom-
tailored information on energy-related accident risks can be useful to a variety of
stakeholders ranging from industry and the services sector to national governments
and national or international organizations and authorities that are engaged in
emergency response, disaster relief and safety, or law enforcement.
The En
ergy-related Severe Accident Database (ENSAD) was used as a basis for the
present study on severe accident risks in the energy sector. It has been developed
and established at PSI, and since its first release in 1998 (Hirschberg et al. 1998)
the ENSAD database has been continuously updated to keep up with the growing
historical experience. The analytic scope has also been substantially extended to
provide solutions to upcoming problems and to meet the specific needs of new
users. Specific advancements were achieved in the course of recent projects: The
China Energy Technology Program (CETP) permitted access to previously restricted
information on accidents in China, particularly in coal mining (Hirschberg et al.
2003a, 2003b). Within the EU 5th Framework Programme Project New Elements for the
Assessment of External Costs from Energy Technologies (NewExt), the analytical framework
was extended to include calculations of external costs of severe accidents in non-
nuclear chains (Burgherr et al. 2004). A study of natural gas accident risks for the Swiss
Gas and Water Industry Association (SVGW) enabled further improvements based
on detailed evaluations of natural gas accident statistics of Switzerland and Germany
(Burgherr 2005a,b; Burgherr and Hirschberg 2005a). Most recently, the database is
being reviewed within the Integrated Project New Energy Externalities Developments for
Sustainability (NEEDS) of the EU 6th Framework Programme.
The principal objective of the present article is to provide a comprehensive, state-
of-the-art overview of the comparative assessment of severe accident risks in fossil
(i.e., coal, oil, natural gas, and Liquefied Petroleum Gas [LPG]), hydro, and nuclear
energy chains. Specifically, the following evaluations were performed: (1) overview
of the available historical experience as collected in ENSAD; (2) comparison of
energy chains using aggregated indicators and frequency-consequence curves; (3)
identification of risk-dominant energy chains; and (4) economic valuation of severe
accidents, that is, calculation of external costs.
APPROACH AND METHODS
Severe Accident Database ENSAD
In the initial phase of ENSAD it was decided that the development of a severe
accident database from scratch would be neither feasible nor efficient. Therefore,
ENSAD uses a multitude of primary information sources whose contents are ver-
ified, harmonized, and merged within the ENSAD framework. The advantages of
Hum. Ecol. Risk Assess. Vol. 14, No. 5, 2008 949
P. Burgherr and S. Hirschberg
Table 1. Overview of sources for the energy-related accidents contained in
ENSAD for the period 1969–2000. Note that the total number of
accidents sums up to more than the 6227 accidents documented in
ENSAD because a specific accident may be reported by several
sources.
Source # accidents (%)
MHIDAS 2065 19.7
FACTS 1460 14.0
China Coal Industry Yearbook 825 7.9
ETC Tanker Spills Database 726 6.9
HSELINE/LLP 495 4.7
Swiss Re 493 4.7
Vieites et al. 387 3.7
WOAD 266 2.5
EM-DAT 262 2.5
c4tx 256 2.4
MSHA 210 2.0
9 sources with contributions of 1 to <2% 1202 11.5
10 sources with contributions of 0.5 to <1% 709 6.8
153 sources with contributions <0.5% 1104 10.6
such an approach are: (1) The substantial variation among individual databases in
availability, scope, development status, and quality can be balanced; (2) commercial
databases were also included to gain access to proprietary data that are not fully
contained in publicly available information sources; and (3) the combined infor-
mation available from a variety of sources results in a much broader coverage of se-
vere accidents than any single database (Burgherr and Hirschberg 2005b; Burgherr
et al. 2004; Hirschberg et al. 1998). The information sources surveyed to document
energy-related accidents included in ENSAD for the years 1969–2000 are shown in
Table 1. Four sources with shares higher than 5% contributed 48.5%, followed by
seven sources with shares between 2–5% with a cumulative contribution of 22.7%,
and the remaining more than 170 other sources sum up to about 30%. However,
many of the sources with small shares were of critical importance because they cov-
ered specific energy chains and/or countries, were useful to resolve contradicting
statements, and provided supplementary information that would otherwise not be
available. Guha-Sapir and Below (2002) also showed that data sets from three individ-
ual databases were often complementary to each other, even if they were conceived
for different purposes and for different stakeholders. This is in accordance with our
own experience and further supports the approach chosen for the establishment
and development of ENSAD.
The actual process of database building and implementation can be assigned to
the following phases:
1. Selection and survey of relevant information sources.
2. Raw information is collected, merged, harmonized, checked for inconsistencies,
and verified.
950 Hum. Ecol. Risk Assess. Vol. 14, No. 5, 2008
Comparison of Severe Accident Risks in the Energy Sector
3. For each energy-related accident general information (e .g., date, location, and
classification), and specific information on energy chain and stage (e .g., damages,
causes, and technological characteristics) is entered into ENSAD.
4. The database content of ENSAD is once more cross-checked to keep data errors
at the lowest feasible level.
5. Comparative evaluations are then carried out, based on customized ENSAD-
queries.
The complete process has been described in detail earlier and is thus not repeated
here (Burgherr et al. 2004; Hirschberg et al. 1998).
ENSAD provides a comprehensive coverage of severe, energy-related accidents
and their technical aspects, whereas other man-made accidents and natural disasters
are accounted for somewhat less extensively. The historical experience compiled in
ENSAD allows the users to make coherent analyses tailored to their specific needs.
Severe Accident Definition
In the literature no commonly accepted definition can be found of what consti-
tutes a severe accident. This can be illustrated by the following examples:
r
The “World-wide Offshore Accident Database” (WOAD) of the Det Norske
Veritas (DNV 1999) considers an accident as severe or major if more than one
fatality occurred or if the damaged unit (e .g., oil platform, drill ship, or drill
barge) experienced total loss.
r
Glickman and Terry (1994) define an accident from a technological hazard as
significant, if it resulted in at least five fatalities or if it involved the release of a
chemical, petroleum product, hazardous waste, or other hazardous material.
r
The SIGMA publication series of the Swiss Re Company (Swiss Re 2001) and
Rowe (1977) do not use the term “severe accidents.” However, they do investi-
gate and collect data on catastrophic events. The criteria are arbitrary (Rowe
1977), not standardized, and can be adjusted with time (Swiss Re 2001).
r
The Emergency Disasters Database (EM-DAT) by the WHO Collaborating Cen-
tre for Research on the Epidemiology of Disasters (CRED) considers several
reasons for taking a disaster into account (EM-DAT 2006). For example, 10 or
more people killed, 100 or more people affected/injured/homeless, declara-
tion of state of emergency, or when a disaster affected several countries/regions.
It becomes clear from these few examples that accident definitions vary largely with
respect to the actual damage types considered (e.g., fatalities, injured persons, evac-
uees, or economic costs), the use of imprecise categories such as “people affected,”
and differences in damage thresholds used to distinguish severe from smaller acci-
dents.
The database ENSAD uses seven criteria to distinguish between severe and smaller
accidents. Whenever an accident is characterized by one or more of the following
consequences, it is considered to be severe (Burgherr et al. 2004; Hirschberg et al.
1998):
1. at least 5 fatalities or
2. at least 10 injured or
Hum. Ecol. Risk Assess. Vol. 14, No. 5, 2008 951
P. Burgherr and S. Hirschberg
3. at least 200 evacuees or
4. extensive ban on consumption of food or
5. releases of hydrocarbons exceeding 10,000 (metric) tons or
6. enforced clean-up of land and water over an area of at least 25 km
2
or
7. economic loss of at least 5 million USD (2000)
2
Fatalities generally comprise the most reliable indicator with regard to the complete-
ness and accuracy of the data. Fatality information is superior to injured or evacuated
persons information because often the severity of an injury or the duration of an
evacuation is not reported. A typical problem in the case of economic damages is
that sources outside the insurance sector tend to mix the various types of economic
damages (e .g., insured versus total damage), and these categories are not established
in a consistent manner or give no specification of what type of damage is reported
(Munich Re 2001). The other consequence indicators are either relevant only for
specific energy chains or ENSAD contains very few entries with adequate details of
information (Burgherr et al. 2004; Hirschberg et al. 1998). Therefore, results pre-
sented in this study focus on number of fatalities, although in suitable cases results
are also presented for injured persons and evacuees.
Historical Experience and Evaluation Period
The ENSAD database allows comprehensive analyses of accident risks that are not
limited to power plants but cover full energy chains. Accounting for contributions
from all stages of energy chains is essential because fossil chain accidents at power
plants play a minor role compared to the other chain stages, that is, analyses based
on power plants only would strongly underestimate the real impact of accidents
(Hirschberg et al. 1998). Furthermore, identification of weak links in an energy
chain, potential improvements, and effective measures on the technical or regulatory
levels require deep knowledge of events, their possible causes, dimensions, and
relationships (Burgherr et al. 2004; Hirschberg et al. 1998). In addition, results of
Probabilistic Safety Assessment (PSA) are employed to address hypothetical severe
nuclear accidents because past experience is not representative (Hirschberg et al.
2004a).
Severe accidents in the energy sector were analyzed for the years 1969–2000. The
starting year was chosen to reflect the distinct increase in the number of energy-
related accidents at the end of the 1960s, which is primarily due to the increase
in the volume of activities (Hirschberg et al. 1998). Furthermore, consideration of
accidents further back in time may lead to results that lack relevance for the present
situation because they are not comparable due to changes over time (i.e., techno-
logical advancements, more strict safety regulations, etc.) and/or improvements in
reporting and documentation compared to pre-1969 data (Hirschberg et al. 2004a;
Hirschberg et al. 1998). Data from 2001 onward were not included in the current
analysis because it is a known fact that there is a time delay of up to several years for
2
Different currencies were all converted to USD values. To take account of inflation, specific
amounts were extrapolated using the U.S. Consumer Price Index (CPI) to obtain year 2000
values.
952 Hum. Ecol. Risk Assess. Vol. 14, No. 5, 2008
Comparison of Severe Accident Risks in the Energy Sector
certain accidents until consolidated information and final reports become available,
which can then be designated as a final record in ENSAD (Burgherr et al. 2004).
Statistical Analyses
The temporal trends in annual numbers of accidents and fatalities for natural
disasters and man-made accidents were analyzed for the period 1969–2000. During
the past decades, numerous parametric and non-parametric techniques for the de-
tection of trends in time series have been developed (Hirsch et al. 1991; Sirois 1998).
The main objective of trend analysis is to know if there is a significant change in the
time series. The advantage of the non-parametric statistical tests over the parametric
tests is that the non-parametric tests are more suitable for non-normally distributed,
missing data and extreme values (Yue and Pilon 2002). First, the non-parametric
Mann–Kendall test (Mann 1945) was used to detect the presence of a monotonic
increasing or decreasing trend. This test is suitable when a monotonic trend can be
assumed and no seasonal or other cycle is present in the data. Second, the slope
of an existing trend (expressed as change per year) was estimated using Sen’s non-
parametric method (Gilbert 1987; Sen 1968), which is applicable in cases where the
trend can be assumed to be linear. Sen’s method is also little affected by errors within
the data values and it is robust because insensitive to the “extreme” and missing val-
ues. Calculations were performed with the MS Excel application MAKESENS (Salmi
et al. 2002). Results were considered statistically significant when the probability level
(p -value) was smaller than the chosen significance level of α = 0.05.
Comparisons of the various energy chains were based on data normalized to
the unit of electricity production. For fossil energy chains the thermal energy was
converted to an equivalent electrical output using a generic efficiency factor of 0.35.
For nuclear and hydro power the normalization is straightforward because in both
cases the generated product is electrical energy. The Gigawatt-electric-year (GW
e
yr)
was chosen because large individual plants have capacities in the neighborhood of
1GWofelectrical output (GW
e
). This makes the GW
e
yr a natural unit to use in
discussions of total electricity production.
Results are provided separately for OECD
3
and non-OECD countries because of
large differences in levels of technological development and safety performance.
This distinction is also meaningful because of the substantial differences in manage-
ment, regulatory frameworks, and general safety culture between these two groups
of countries (e .g., Burgherr and Hirschberg 2005a; Hirschberg et al. 1998). Finally,
results were complemented by individual calculations for the EU 25
4
and the Chinese
coal chain. In the case of China, coal chain data were only analyzed for the years
3
The Organisation for Economic Co-operation and Development (OECD) was established
in 1961 and currently consists of 30 member countries: Australia, Austria, Belgium, Canada,
Czech Republic, Denmark, Finland, France, Germany, Greece, Hungary, Iceland, Ireland,
Italy, Japan, Korea, Luxembourg, Mexico, The Netherlands, New Zealand, Norway, Poland,
Portugal, Slovak Republic, Spain, Sweden, Switzerland, Turkey, United Kingdom, United
States.
4
The European Union currently comprises 27 member states. Results in this article are given
for EU 25, which includes the former EU 15 countries: Belgium, Germany, France, Italy,
Hum. Ecol. Risk Assess. Vol. 14, No. 5, 2008 953
P. Burgherr and S. Hirschberg
1994–1999 when data from the China Coal Industry Yearbook were available, indicat-
ing that previous years were subject to substantial underreporting (Hirschberg et al.
2003a,b). In other non-OECD no underreporting at a comparable level to China was
found; for example, coal accidents in Russia and Ukraine or oil accidents in Nigeria
receive an adequate coverage in several commercial databases (e.g., MHIDAS, HSE-
LINE [OSH-ROM 2007]). A more in-depth coverage of this issue may be found in
Hirschberg et al. (1998) and Burgherr et al. (2004).
Concerning the analysis of hydropower accidents a conservative approach was
chosen because total fatalities were allocated to electricity generation in those cases
when dams also served other purposes. Furthermore, it is sometimes more difficult
for dams in non-OECD countries to reliably determine secondary uses. Finally, safety
investments in dams used for electricity generation are generally higher because the
loss potential of a dam failure is also higher,
5
which makes the chosen approach less
absolute.
6
Damage rates in the form of aggregated indicators provide a direct comparison
of severe accident consequences (e.g., fatalities) per unit of electricity produced
(e.g.,GW
e
yr) between energy chains and country groups. A second set of values is
also given, accounting for the fact that a large number of severe accidents occur
in non-OECD countries at stages in the energy chain relevant for export to OECD
countries. This can be incorporated in the calculations by adding the appropriate
share of the consequences of accidents that occurred at such fuel cycle stages in
non-OECD countries to the damages that physically occurred in OECD countries,
that is, OECD countries import fatalities with their fuel. The net amounts of energy
carriers imported to OECD countries from non-OECD countries form the basis
for this allocation procedure, which has been described in detail in Hirschberg
et al. (1998). Aggregated indicators with allocation are particularly useful within a
sustainable development perspective because they assume that the industrial OECD
countries should bear a certain share of these damages.
In a second step, the comparison of results was expanded beyond aggregated
chain-specific values by combining results of the frequency and consequence analy-
ses to generate risk estimates. Frequency-consequence (F-N) curves are a common
way to express collective or societal risks in quantitative risk assessment, that is, the
F-N curves provide an estimate of the risk of accidents that affect a large num-
ber of people. F-N curves show the relationship of the cumulative frequency (F)
of events having N or more fatalities, which is usually presented in a diagram with
two logarithmic axes. In this way, they comprise more information than aggregated
Luxembourg, The Netherlands, Denmark, United Kingdom, Ireland, Greece, Portugal, Spain,
Austria, Finland, Sweden as well as Czech Republic, Estonia, Cyprus, Latvia, Lithuania, Hun-
gary, Malta, Poland, Slovenia, Slovak Republic that joined recently. Bulgaria and Romania
that became member states as of 2007 are not yet considered here.
5
A failure of a hydropower dam could cause a large number of fatalities in densely populated
industrial areas downstream where power is delivered, or a failure may lead to large damages
to the costly installations for electricity production (e.g., generators, turbines, transformers).
6
Note that a recalculation of aggregated indicators for hydropower (i.e., fatalities distributed
evenly in the case of multi-purpose dams) would result in about 0.002 and 5.015 fatalities per
GW
e
yr for OECD and non-OECD countries, respectively, but not change the energy chain
ranking (compare Figure 4).
954 Hum. Ecol. Risk Assess. Vol. 14, No. 5, 2008
Comparison of Severe Accident Risks in the Energy Sector
indicators because they show the probability of accidents with varying degrees of
severity of consequences, including chain-specific maximum damages.
7
It is impor-
tant to stress that the FN curve shows the cumulative number of events. The method-
ology for generating an F-N curve from accident frequency and consequence data
has been extensively described in literature (e.g., CCPS 1989; IChemE 1985). Finally,
the 5% and 95% confidence intervals can be calculated for the point estimates of
the F-N curve, based on the Chi-square distribution (H¨artler 1983).
The economic losses of severe accidents (and natural disasters) are often ex-
pressed as total economic damage or insured damage; the latter only reflecting the
part covered by insurance premiums. However, insurance does not lower total dam-
age per se, it only spreads the losses inflicted by a disaster; that is, it provides financial
protection to the insured party against a large loss event with a small probability of
occurrence. Generally, the ratio of economic losses to insured losses (L/I) is high
in developing countries with a limited insurance market, whereas the highly devel-
oped OECD countries have a much lower L/I-ratio (Kunreuther and Michel-Kerjan
2006). In contrast, fatalities are often not considered in economic loss assessments, al-
though their monetization may become important in the less-developed non-OECD
countries where insurance coverage is limited. Accordingly, internalization of exter-
nalities prevails at substantially lower levels in these countries.
Fuel chain external costs or externalities denote those costs imposed on society
and the environment that are not accounted for by the producers and consumers of
energy. In the case of severe accidents, the assessment and valuation of external costs
has been limited to the nuclear chain in the past. Therefore, the present study pro-
vides a consistent approach to calculate external costs of non-nuclear energy chains.
External costs of severe accidents in different energy chains were estimated, based
on unit cost values for the various types of consequences (Burgherr et al. 2004).
8
Unit
values for fuel cycle accident endpoints were derived, expressing the welfare impacts
of accidents in monetary terms in order to enable calculation of the external costs of
accidents. Endpoints were established for different damage types such as premature
death, physical injury or evacuation, among others. The central estimates derived
for the value of a statistical life (VSL) are considered representative, with minimum
and maximum ranges reflecting the uncertainty that remains in the derivation of
these values (Markandya et al. 2004). For injuries it is often difficult to determine
the degree of severity. Therefore, unit values for a “typical” injury correspond to
the average of “severe” and “minor” injury categories, following the differentiation
proposed by Lindberg (1999). Evacuation costs consist of two components, namely
fixed and daily costs per household, the latter often difficult to estimate because
the duration of evacuation is often not precisely stated for specific accidents. In this
7
The chain-specific maximum damage or maximum consequences denote the accident with
most severe consequences (e.g., fatalities) of a specific energy chain, based on the historical
data contained in ENSAD or derived from a Probabilistic Safety Assessment (PSA) for the
nuclear chain.
8
For further reading we suggest references cited in the reports by Burgherr et al. (2004) and
Markandya et al. (2004) that were produced within the NewExt project (see introduction). Ad-
ditionally, a number of other references may be of particular interest (European Commission
1995, 1999, 2003; Jones-Lee and Loomes 1995; Viscusi and Aldy 2003, 2007).
Hum. Ecol. Risk Assess. Vol. 14, No. 5, 2008 955
P. Burgherr and S. Hirschberg
Table 2. Summary of unit values for fuel cycle accident endpoints (in EUR at
price level of year 2002) for occupational and public fatalities, provided
for various levels of internalization (expressed in parentheses). The
values were derived by University of Bath (see Burgherr et al. 2004;
Markandya et al. 2004).
Central Minimum Maximum
Value of a Statistical Life 1,000,000 400,000 3,310,000
Occupational fatalities OECD
Central (80%) 200,000 80,000 662,000
Lower internalization (70%) 300,000 120,000 993,000
Upper internalization (100%) 0 0 0
Occupational fatalities non-OECD
Central (50%) 500,000 200,000 1,655,000
Lower internalization (0%) 1,000,000 400,000 3,310,000
Upper internalization (100%) 0 0 0
Public fatalities OECD
Central (50%) 500,000 200,000 1,655,000
Lower internalization (30 %) 700,000 280,000 2,317,000
Upper internalization (70%) 300,000 120,000 993,000
Public fatalities non-OECD
Central (20%) 800,000 320,000 2,648,000
Lower internalization (0%) 1,000,000 400,000 3,310,000
Upper internalization (50%) 500,000 200,000 1,655,000
study only external costs of severe accidents with at least five immediate fatalities
were evaluated.
An important prerequisite for economic analysis is the separation between public
and occupational fatalities in single accidents because they differ in their degree of
internalization, which directly affects calculation of external costs. Severe accidents
in the energy sector are often work-related, but can also affect the general public
(Burgherr et al. 2004). For example, consequences of coal mine accidents are gener-
ally restricted to the workers present, although rescue parties may be at risk as well.
In contrast, hydro dam failures may have large effects on downstream residents. In
many cases, however, accident fatalities may not be exclusively allocated to one or the
other category. The monetary valuation results obtained for the fatality endpoints
in OECD and non-OECD countries are summarized in Table 2 in internalization for
occupational and public taking into account the differences victims.
The actual calculation of external costs from fatalities of severe accidents consists
of a simple multiplication, that is, fatalities × VSL × degree of internalization (OECD
or non-OECD, occupational or public fatalities), which is then normalized to the
unit of electricity production (kWh). For a detailed description of the methodology
and findings for the various categories compare Burgherr et al. (2004).
If not denoted differently, all statistical analyses were performed using the freely
available language and environment R Software for Statistical Computing (R Develop-
ment Core Team 2006).
956 Hum. Ecol. Risk Assess. Vol. 14, No. 5, 2008
Comparison of Severe Accident Risks in the Energy Sector
Figure 1. Number of fatalities for severe (5 fatalities) accidents that occurred
due to natural disasters and man-made accidents in the period 1969 to
2000 (n = 32). Time series trends analyzed by the Mann–Kendall test
and Sen’s slope estimator are shown as an insert, where Z = test statistic,
p = p -value, Q = Sen’s slope, ns = not significant.
RESULTS AND DISCUSSION
Current Status and Properties of ENSAD
The ENSAD database currently contains 18706 accident records, of which 88.4%
occurred in the years 1969–2000, that is, the evaluation period chosen in this study.
Within this period, 6995 accidents resulted in five or more fatalities, of which 39.5%
were natural disasters and the other 4233 were man-made accidents. The latter can
be further divided into energy-related accidents (1870, or 44.2%) and other man-
made accidents (2363, or 55.8%).
In Figure 1 fatalities in all categories of severe (5 fatalities) man-made accidents
and natural disasters from 1969 to 2000 are shown, amounting to about 3.4 million
fatalities. Of these victims, more than 90% were due to natural catastrophes and
about 10% due to severe man-made accidents; 37% of the latter were killed in
energy-related accidents.
The largest natural disasters
9
were a storm and flood catastrophe in Bangladesh
in 1970 (300,000 fatalities), the Tangshan earthquake in China in 1976 (290,000),
and a drought and civil strife in Sudan in 1983 (250,000). In contrast, the largest
9
The December 2004 Indian Ocean seaquake (magnitude of 9.1 to 9.3 on the Richter scale)
triggered a series of devastating tsunamis. Initial estimates put the death toll in the range of
300,000, but recent analyses reduced it to about 220,000 (e.g., Swiss Re 2006). Nevertheless,
this catastrophe is one of the deadliest disasters in modern history.
Hum. Ecol. Risk Assess. Vol. 14, No. 5, 2008 957
P. Burgherr and S. Hirschberg
man-made accidents resulted in fatalities one to two orders of magnitude lower. The
top-ranked energy-related accidents include the Banqiao/Shimantan dam failure in
China in 1975 (26,000 fatalities), the collision of the tanker Victor with the Ferry Dona
Paz off the Philippines in 1987 (4386), and a tank truck collision with another vehicle
in the Salang tunnel in Afghanistan’s Parvan province in 1982 (2700). Large non-
energy-related severe accidents include the accident at a pesticide plant in Bhopal in
India in 1984 (5000 fatalities), the sinking of the ferry Neptune near the coast of Haiti
in 1993 (1800), and the failure of the Gouhou dam (primary purpose: irrigation and
water supply) in China in 1993 (1250).
Temporal trend analyses using the Mann–Kendall test and Sen’s slope method
are given as an insert in Figure 1. The annual numbers of accidents and fatalities in
severe man-made accidents increased from 1969 to 2000. For natural disasters such
a significant trend was only found in the number of accidents, but not for fatalities.
This is primarily due to the few extremely large (100,000 fatalities) catastrophes,
of which more occurred in the 1970s than in the 1980s and 1990s. If they were
excluded, cumulated fatalities for the 1990s become substantially larger than for the
two decades before.
Overview of Man-Made Energy-Related Accidents
The ENSAD database at PSI includes 1870 severe accidents for the various energy
chains in the period 1969–2000, amounting to 81,258 fatalities (Table 3). The coal
chain accounted for 65.3% of all accidents, with oil a distant 2nd at 21.2%. Con-
tributions by the natural gas (7.2%) and LPG (5.6%) chains were much smaller,
while both hydro and nuclear account for less than 1% each. This dominance of
coal-chain accidents is fully attributable to the release of detailed accident statistics
by China’s coal industry, data that were not previously publicly available (Hirschberg
et al. 2003a,b). Altogether, 819 of the 1044 accidents collected for the Chinese coal
chain occurred in the years 1994–1999, implying substantial under-reporting prior
to the release of the annual editions of the China Coal Industry Yearbook.
Fatalities were clearly dominated by the Banqiao/Shimantan dam failures, which
together resulted in 26,000 deaths. As a consequence, the hydro chain accounts for
36.8% of all fatalities. Among the fossil chains, coal accounted for the most fatalities,
followed by oil, LPG and natural gas.
In Figure 2 percent contributions of energy-related severe (5 fatalities) acci-
dents, fatalities, and insured losses (5 million USD (2000) per continent are shown
for OECD and non-OECD countries in the years 1969–2000. The number of acci-
dents and fatalities is substantially higher in non-OECD countries, with Asia being
the dominant contributor. Concerning the number of accidents, the Chinese coal
chain accounts for 55.7% of the total, whereas the Chinese hydro and coal chains
contribute 22.1% and 31.9% to total fatalities. In contrast, insured losses are distinc-
tively higher in OECD countries, with a combined share of 69.0% for Europe and
North America. In these two continents 63.0% of insured losses are attributable to
the oil and gas chains. The high insured losses and low death toll in OECD countries
and vice-versa in non-OECD countries can be explained by the significantly higher in-
surance density in OECD on the one hand, and the exposure characteristics of many
population centers as well as the less strict safety regulations and lack of engineered
958 Hum. Ecol. Risk Assess. Vol. 14, No. 5, 2008
Comparison of Severe Accident Risks in the Energy Sector
Table 3. Summary of severe accidents with at least 5 immediate fatalities that
occurred in fossil, hydro, and nuclear energy chains in the period
1969–2000. Accident statistics are given for the categories OECD
(incl. EU 25), EU 25 alone, and non-OECD. For the coal chain,
non-OECD without China and China alone are given separately. Acc
= number of accidents, Fat = number of fatalities.
OECD EU 25 non-OECD World total
Energy chain Acc Fat Acc Fat Acc Fat Acc Fat
Coal 75 2259 32 697 102 4831 1221 25107
1044 18,017
(819)
a
(11,334)
a
Oil 165 3713 60 1185 232 16,505 397 20218
Natural gas 90 1043 28 276 45 1000 135 2043
LPG 59 1905 20 564 46 2016 105 3921
Hydro 1 14 0 0 10 29,924
b
11 29938
Nuclear 0000 1 31
c
131
Total 390 8934 140 2722 1480 72324 1870 81258
a
First line: Coal non-OECD without China; second and third line: Coal China
1969–2000, and in parentheses 1994–1999. Note that only data for 1994–1999 are
representative because of substantial underreporting in earlier years (Hirschberg et al.
2003a, 2003b).
b
Banqiao/Shimantan dam failures together caused 26,000 fatalities.
c
Only immediate
fatalities. In the case of Chernobyl estimates for latent fatalities range from about 9000
for Ukraine, Russia, and Belarus to about 33,000 for the whole northern hemisphere in
the next 70 years (Hirschberg et al. 1998). According to a recent study (Chernobyl
Forum 2005) by numerous United Nations organizations (IAEA, WHO, UNDP, FAO,
UNEP, UN-OCHA, and UNSCEAR) up to 4000 persons could die due to radiation
exposure in the most contaminated areas. This estimate is substantially lower than the
upper limit of the PSI interval, which, however, was not restricted to the most
contaminated areas.
safety features in many highly exposed areas in non-OECD on the other hand. In
accordance with our results Swiss Re (2006) reports that (1) industrialized countries
dominate insured losses of man-made accidents; (2) the most costly of these acci-
dents are fires and explosions in industrial operations and in the oil and gas industry;
and (3) insured damages of man-made accidents increased about 1.5 times between
the 1970s and 1990s.
Weak-Point Analysis of Individual Energy Chains
The majority of accidents in fossil energy chains do not occur in power plants, but
rather in other stages in the energy chains (Figure 3). More than 95% of the victims in
the coal chain lose their lives in mines, primarily due to gas explosions. With oil, the
transportation to the refinery and regional distribution are the most accident prone
stages; most frequent are tanker accidents at sea and street accidents involving tank
trucks. Transportation is also the weak stage in the gas chain, which is dominated by
pipeline accidents in transmission (long-distance) and distribution (regional/local)
Hum. Ecol. Risk Assess. Vol. 14, No. 5, 2008 959
P. Burgherr and S. Hirschberg
Figure 2. Percent shares of energy-related severe (5 fatalities) accidents, fatali-
ties, and insured losses (5 million USD (2000)) per continent for OECD
and non-OECD countries in the years 1969–2000.
networks. In the LPG chain transportation accidents were most prominent too, par-
ticularly in regional and local distribution. In contrast, hydropower and nuclear
power accidents (not shown in Figure 3) only occur near the area of the storage
dam or reservoir, and the plant site, respectively. While coal chain victims are al-
most exclusively work-related, gas and oil accidents involve a significant number of
innocent bystanders as victims. If a storage dam breaks then the general populace is
almost exclusively affected, with the exception of the dam operators. Nuclear plant
accidents may also lead to immediate fatalities, but here the deaths are dominated
by latent (see footnotes 7 and 10) fatalities due to eventual cases of cancer.
Comparative Evaluations of Severe Energy-Related Accidents
Aggregated indicators
Numerical comparisons of severe accident risks associated with the different en-
ergy chains are presented for three types of damage indicators, namely immediate
fatalities, injured persons and evacuees. Damage rates expressed as aggregated in-
dicators are given in terms of affected people per GW
e
yr, differentiating between
OECD and non-OECD countries. It should be noted that the statistical basis for the
indicators for individual energy chains may differ substantive. For example, there
are 1221 severe accidents with at least 5 fatalities in the coal chain and only 1 in the
nuclear chain (Chernobyl).
Although different damage indicators were considered and analyzed, the con-
clusions stated in this paper are primarily based on fatality rates. The reasons for
this are threefold: (1) historical data on fatalities is most complete (see the section
on Severe Accident Definition); (2) fatalities associated with large accidents are the
960 Hum. Ecol. Risk Assess. Vol. 14, No. 5, 2008
Comparison of Severe Accident Risks in the Energy Sector
Figure 3. Relative shares (logarithmic scale) of accidental fatalities in the stages of
various energy chains.
major indicator driving public awareness and the reaction of society; and (3) overall
patterns for other indicators are in some cases quite similar to those for fatality rates,
whereas differences may be explained by chain-specific characteristics (Hirschberg
et al. 2004a; Hirschberg et al. 1998).
For example, in coal mine accidents evacuations often affect fewer than 200 peo-
ple, so that an accident is not considered severe according to this criterion. Par-
ticularly in some non-OECD countries escape routes in subsurface mines may also
become blocked in an emergency due to insufficient safety management of the
mines, and thus people may be killed instead of being evacuated or rescued. In the
natural gas and LPG chains the absolute numbers of evacuees in OECD countries
are two orders of magnitude higher than in non-OECD countries due to efficient
safety measures and emergency planning, resulting in much higher rates of evacuees
per GW
e
yr, that is, saved lives. Finally, the number of evacuees after a dam failure is
strongly dependent on the warning time and the population size downstream of the
dam.
That significant differences exist between the aggregated damage rates assessed
for the various energy chains is shown in Figure 4. However, one should keep in mind
Hum. Ecol. Risk Assess. Vol. 14, No. 5, 2008 961
P. Burgherr and S. Hirschberg
962 Hum. Ecol. Risk Assess. Vol. 14, No. 5, 2008
Comparison of Severe Accident Risks in the Energy Sector
that from the absolute point of view the damage rates for fatalities from fossil sources
are much smaller than the corresponding rates associated with health impacts of
normal operation, which can be demonstrated by monetization of damages (e.g.,
Burgherr et al. 2004; EC 2003) and calculation of external costs (see section on
“external costs of severe accidents”). For this reason the evaluation focuses here on
the relative differences between the various energy carriers.
Generally, OECD countries exhibit significantly lower fatality rates than those for
non-OECD countries. Among the fossil chains LPG is most accident-prone, followed
by oil and coal, whereas natural gas performs best. Although the EU 25 showed
lower values compared to OECD, differences were minimal in some cases. Before
the extension from EU 15 to EU 25, that is, the integration of accession countries, this
difference was more pronounced because several of the new EU member countries
already had OECD status. Therefore, values for the OECD can also be considered
representative for the EU 25, and they also have a broader statistical basis.
Western style nuclear and hydropower plants have the lowest fatality rates. The
recent experience with hydro in OECD countries points to very low fatality rates, com-
parable to the representative PSA-based results obtained for nuclear power plants in
Switzerland and in the USA, whereas in non-OECD countries dam failures can claim
large numbers of victims. The figure also shows that the Chinese coal chain should
be treated separately as its accident fatality rates are about ten times higher than in
other non-OECD countries and about forty times higher than in OECD countries.
The previous discussion was restricted to immediate fatalities; however, in case of
the nuclear chain latent fatalities dominate total fatalities. When one reviews these
latent fatalities for the only severe (5 fatalities) nuclear accident with an impact on
human health (Chernobyl), then estimates of latent deaths would range from 13.9 to
51.2 deaths per GW
e
yr (for non-OECD countries). However, extending these risks for
nuclear energy to current OECD countries is not appropriate, because OECD plants
use other, safer technologies. This is also predominantly true for the current situation
in non-OECD countries. In the OECD, PSA-based, latent fatalities are therefore gen-
erally significantly lower, around 0.02 fatalities per GW
e
yr (Hirschberg et al. 1998).
−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−
Figure 4. Comparison of aggregated, normalized, energy-related damage rates for
(a) fatalities, (b) injured, and (c) evacuees, based on historical experi-
ence of severe accidents that occurred in OECD (EU 25 in rectangular
brackets for fatalities only) and non-OECD countries in the period 1969–
2000, except for coal China where complete data from the China Coal
Industry Yearbook were only available for the years 1994–1999. Note that
only immediate fatalities were considered; latent fatalities, of particular rel-
evance for the nuclear chain, are commented on separately in the text.
Results for OECD and non-OECD countries are given with and without
allocation of damages. The bars in dark grey show domestic fatalities,
and the light grey bars show the “imported” or “exported” fatalities. For
the OECD the dark and light grey bars should be added, and for the
non-OECD the light bar should be subtracted from the total bar to ob-
tain allocated damage rates. The exact values for each bar are shown in
the figure, with the allocated values in brackets.
Hum. Ecol. Risk Assess. Vol. 14, No. 5, 2008 963
P. Burgherr and S. Hirschberg
Finally, Figure 4 also displays a full allocation of damages for fossil energy chains
on the basis of imports and exports (see section on Statistical Analyses) because
a large number of severe accidents in non-OECD countries are related to energy
exports to the OECD. Severe fatality rates for the oil and LPG chains exhibited the
most distinct increase for OECD countries and decrease for non-OECD countries in
comparison to the rates without allocation, whereas differences for coal and natural
gas chains were distinctly smaller. Within the framework of sustainable development,
it could be argued that the highly industrialized OECD countries should assume a
certain share of these damages.
Frequency-consequence curves
Frequency-consequence (F-N) curves retain the ranking of energy chains de-
rived from aggregated indicators, but provide additional insight on chain-specific
maximum damages and on the probability of an accident exceeding a specified
consequence threshold (e.g., number of fatalities).
Shown in Figure 5 are F-N curves for severe accidents with at least 5 fatalities
in the OECD, EU 25, and non-OECD countries (Figure 5a–c) and for individual
chains including confidence intervals (Figure 5d–i). For OECD countries, fossil en-
ergy chains clearly exhibited higher frequencies of severe accidents than hydro and
nuclear (Figure 5a). Among fossil chains, LPG exhibits the worst performance and
natural gas the best, whereas coal and oil chains are ranked in between. When com-
paring maximum consequences, the impact of hydro is marginal, followed distantly
by natural gas (109 fatalities), and other fossil chains having 2.5 to 4.5 times greater
values compared to natural gas. The ranking of F-N curves for the EU 25 is similar
to OECD countries; the however maximum consequences of fossil chains are about
2to4times lower (Figure 5b and 5d–g).
For non-OECD countries (Figure 4c) the ranking of F-N curves was comparable
to the OECD, except for the Chinese coal chain that showed a significantly worse
performance than other non-OECD countries. Furthermore, frequencies at corre-
sponding numbers of fatalities were generally higher for non-OECD compared to
OECD, and for LPG and Coal China (1994–99) chain frequencies at lower death tolls
were even greater than 10
1
(Figure 5d–i). Regarding chain-specific maxima, non-
OECD values of coal (without China), oil and LPG were substantially higher than
the corresponding OECD values. Additionally, the range in observed maximum fa-
talities among individual chains was larger in non-OECD, particularly because the
oil chain can reach maximum numbers up to one order of magnitude higher than
other fossil chains.
For nuclear energy immediate fatalities play a minor role, whereas latent fatal-
ities
10
clearly dominate. Accident frequencies associated with the nuclear chain
(Chernobyl) are relatively low, but the maximum credible consequences may
be very large due to the dominance of latent fatalities, that is, comparable to
10
In the case of the nuclear chain fatalities could continue to occur over an extended period,
therefore total fatalities are split in the following categories: early (immediate) fatalities would
occur shortly after exposure, whereas latent fatalities include all those occurring within 70
years from the radioactive release. For details see Burgherr et al. (2004) and Hirschberg et al.
(2003a).
964 Hum. Ecol. Risk Assess. Vol. 14, No. 5, 2008
Comparison of Severe Accident Risks in the Energy Sector
Figure 5. Comparison of frequency-consequence curves for full energy chains,
based on historical experience of severe accidents in (a) OECD, (b) EU
25, and (c) non-OECD countries for the period 1969–2000, except for
China 1994–1999 (compare text). Additionally, panels (d) to (i) show
individual chains, including 5% and 95% confidence intervals.
Hum. Ecol. Risk Assess. Vol. 14, No. 5, 2008 965
P. Burgherr and S. Hirschberg
the Banqiao/Shimantan dam accident that occurred in China in 1975. Results
concerning Chernobyl were published in Hirschberg et al. (1998). Studies by
EC/IAEA/WHO and UNSCEAR
11
formed the main basis for the numerical esti-
mates of total latent fatalities associated with Chernobyl, supported by numerous
sources including the Russian ones. Estimated latent fatalities due to delayed can-
cers range from 9000 (Ukraine, the Russian Federation, and Belarus) to 33,000
(entire northern hemisphere) over the next 70 years (Hirschberg et al. 1998), indi-
cating that the upper range in PSI’s estimate is conservative (as intended) because it
was not limited to the most contaminated areas. In September 2005 a new report on
the consequences of the Chernobyl accident was released by a Forum consisting of
a number of professional organizations of the United Nations (IAEA, WHO, UNDP,
FAO, UNEP, UN-OCHA, and UNSCEAR
12
)aswell as the World Bank and the govern-
ments of the Russian Federation, Belarus, and Ukraine (Chernobyl Forum 2005).
This report reflects the findings of a large team of natural scientists, economists and
health specialists. One of the conclusions of the report is that in the areas with high
contamination up to 4000 people could eventually die due to radiation doses from
the Chernobyl accident, most of them among the so called “liquidators.”
13
This is
significantly lower compared to the previously mentioned PSI values because of the
more limited area considered.
Finally, the large differences between Chernobyl-based estimates and probabilistic
plant-specific estimates for a Swiss nuclear power plant (Figure 5a,c,i) illustrate the
limitations in applying past accident data to cases that are radically different in
terms of technology and operating environment. To obtain realistic calculations for
Western plants, results of full scope PSA should constitute the relevant reference.
Risk dominant energy chains
The overview of the risk dominant energy chains based on historical accidents
for OECD and non-OECD countries in the period 1969–2000 is shown in Table 4.
Only accidents with at least 5 fatalities or 10 injured persons or 200 evacuees were
considered. The following evaluation categories are used in the table:
I. Largest number of accidents with consequences exceeding the aforementioned
threshold values.
II. Largest aggregated number of fatalities, injured, or evacuees.
III. Largest number of fatalities, injured, or evacuees in a single accident.
IV. Largest aggregated number of fatalities, injured, or evacuees, averaged per ac-
cident.
V. Largest aggregated number of fatalities, injured, or evacuees, per GW
e
yr.
11
EC: European Commission, IAEA: International Atomic Energy Agency, WHO: World
Health Organization, UNSCEAR: United Nations Scientific Committee on the Effects of
Atomic Radiation.
12
UNDP: United Nations Development Programme, FAO: Food and Agriculture Organization
of the United Nations, UNEP: United Nations Environment Programme, UN-OCHA: United
Nations Office for the Coordination of Humanitarian Affairs; for others see footnote 11.
13
Emergency workers and persons who were involved in the clean-up operations after the
accident.
966 Hum. Ecol. Risk Assess. Vol. 14, No. 5, 2008
Comparison of Severe Accident Risks in the Energy Sector
Table 4. Risk-dominant energy chains based on historical experience of severe
accidents in OECD (first line) and non-OECD (second line) countries
for the years 1969–2000. Evaluation categories are described in more
detail in the text.
Immediate fatalities Injured Evacuees
Evaluation category (fat)
a
(inj) (eva)
I: largest number of
accidents with fat,
inj, eva
Oil
Coal/Oil
Oil
Oil
Natural Gas
Oil
II: largest aggregated
number of fat, inj,
eva
Oil
Hydro/Oil [Coal]
b
LPG
Oil
LPG
Oil
III: largest number of
fat, inj, eva in a
single accident
Oil
Hydro/Oil [Oil]
b
LPG
Oil
LPG
Hydro, LPG
c
IV: largest aggregated
number of fat, inj,
eva; averaged per
accident
LPG
Hydro/Hydro
Hydro
Nuclear
d
Nuclear
Nuclear
d
V: largest aggregated
number of fat, inj,
eva; normalized per
GW
e
yr
LPG
LPG/LPG
LPG
LPG
LPG
Nuclear
d
a
In non-OECD countries, the ranking for immediate fatalities is given for total non-OECD
and non-OECD without China because of the dominant influence of the Chinese coal
chain; as indicated by the separating slash.
b
The ranking changes if the Chinese dam failures of Banqiao/Shimantan with a total of
26,000 fatalities are excluded from the analysis; as indicated by squared brackets.
c
150,000 evacuees were reported for both energy chains.
d
Chernobyl accident.
Generally, fossil energy chains appeared to be most accident-prone in most cat-
egories. However, natural gas outperformed the other fossil chains because it is
represented just once in Table 4 in evaluation category I for evacuees. In non-OECD
countries, the ranking for immediate fatalities is sensitive to whether China is treated
separately or not. The presence of nuclear in these tables is primarily due to the Cher-
nobyl accident in non-OECD, and a contribution from the Three Mile Island (TMI)
accident to the category evacuation in OECD.
External costs of severe accidents
The external costs for immediate fatalities associated with the different energy
chains were calculated on the basis of historical experience with severe accidents
in OECD and non-OECD countries, separated for the power plant stage and rest of
chain (Figure 6). Generally, total external costs for non-OECD countries were clearly
higher than for OECD countries; however, it is interesting to look at their specific
composition within different chains and regions. In the coal chain occupational
fatalities attributable to rest of chain (i.e., extraction and exploration) dominate
Hum. Ecol. Risk Assess. Vol. 14, No. 5, 2008 967
P. Burgherr and S. Hirschberg
Figure 6. Summary of full chain external costs ( -Cent
(2002)
/kWh
e
)ofsevere ac-
cidents with at least five immediate fatalities in OECD and non-OECD
countries, separated for the power plant stage and rest of chain. Values
on the right of the bars indicate total external costs per energy chain
and region. The reference coal, oil, natural gas, and LPG electricity gen-
erating plants have been assigned efficiencies of 40, 31, 53, and 36%,
respectively. See Table 2 for central value of a Statistical Life and corre-
sponding degrees of internalization. OECD results for nuclear are based
on PSA for a Swiss plant and commented on in the text, whereas reported
numbers for non-OECD refer to the Chernobyl accident. Definition: ng
= negligible.
because accidents occur in the mines with very few exceptions. In contrast, total
external costs in the other fossil chains were primarily driven by public fatalities that
occurred in the different transportation stages. For the oil and natural gas chains the
relative contributions from rest of chain were higher in OECD countries, whereas
for the LPG chain the relative share was higher in non-OECD countries. Concerning
the hydro chain, external costs arise almost exclusively from public fatalities related
to the power plant because floods resulting from failures of large hydro dams are
primarily affecting downstream settlements, that is, the general public. In this case
occupational fatalities are mostly negligible because of the limited number of staff
needed for operation of a hydro dam. However, accidents during the construction
phase of a hydro dam can lead to many occupational fatalities as in the case of
the Mattmark dam (Switzerland) in 1965 when an ice-avalanche catastrophe caused
the death of 88 workers, or the Guavio dam (Colombia) in 1983 when torrential
rains lead to mudslides burying and killing 160 workers changing shifts at the dam
construction site (Burgherr et al. 2004; Hirschberg et al. 1998).
968 Hum. Ecol. Risk Assess. Vol. 14, No. 5, 2008
Comparison of Severe Accident Risks in the Energy Sector
Because the costs in Figure 6 only cover immediate fatalities it is of interest to
relate them to the accident damage costs based on PSA for a Swiss nuclear power
plant, which are dominated by the costs of latent fatalities. The mean value has
been assessed at 1.2·10
3
USD-cent/kWh
e
, with 5% and 95% percentiles at 1.0·10
4
and 3.8·10
3
USD-cent/kWh
e
, respectively. It should be noted that these results also
include damage costs of non-health effects (Hirschberg et al. 1998).
The corresponding average external costs of injured and evacuated persons (not
shown in Figure 6) were 1.0·10
3
-Cent
(2002)
/kWh
e
and 3.9·10
5
-Cent
(2002)
/kWh
e
,
which is roughly one to three orders of magnitude lower than for fatalities (2.5·10
2
-Cent
(2002)
/kWh
e
). However, one should note that estimates for injured and evac-
uees have higher uncertainties because the severity of an injury or the duration of
an evacuation are often not specified for individual accidents.
Although external costs of severe accidents in the energy sector are an impor-
tant component in comparative risk assessment, the absolute values are scarcely
significant when compared to the quantifiable external costs (such as global warm-
ing, public health, occupational health, material damage) of electricity generation,
which range from less than 1
-Cent/kWh
e
for nuclear to 1–4 -Cent/kWh
e
for
natural gas and up to 15
-Cent/kWh
e
for coal (EC 2003). External costs resulting
from severe energy-related accidents are also substantially smaller compared to nat-
ural disasters. This is primarily due to the fact that annual numbers of fatalities from
natural disasters are one to three orders of magnitude higher, and most fatalities
occur in low developed non-OECD countries (also compare Figure 1) with marginal
insurance coverage and thus very low degrees of internalization.
CONCLUSIONS
The ENSAD database developed and maintained at PSI provides a comprehensive
and detailed compilation of severe energy-related accidents. The basis for compar-
ative assessment of severe accident risks associated with the different energy chains
has been significantly extended compared to previous evaluations (e.g., Hirschberg
et al. 2004a; Hirschberg et al. 1998). Recent advancements and developments in-
clude in particular the continuous extension of the period of observation, improved
completeness of historical records, upgrades in quality and consistency of the in-
formation, and better coverage of various types of damages. Furthermore, new unit
cost values for fuel cycle accident endpoints were generated, which in combina-
tion with the energy chain specific accident indicators made it possible to estimate
the corresponding external costs. These estimates are the first of their kind for the
non-nuclear fuel cycles.
This study demonstrates that the comprehensive historical experience of energy-
related severe accidents available in ENSAD can be used as a sound basis for quan-
tifying the corresponding damages and external costs. However, analyses should be
complemented by a PSA approach when full chain risks are dominated by the power
plant stage or when availability and applicability of historical experience is strongly
limited, as it is the case for most Western hydro and nuclear power plants.
Up-to-date evaluations assembled in this publication show quite large numerical
differences between the different energy chains as well as country groups. Hydro
Hum. Ecol. Risk Assess. Vol. 14, No. 5, 2008 969
P. Burgherr and S. Hirschberg
power in non-OECD countries and upstream stages within fossil energy chains are
most accident-prone, whereas the natural gas chain exhibits the lowest risks among
the fossil chains. When comparing country groups, energy-related accident risks
are distinctly lower in the OECD and EU 25 countries than in non-OECD coun-
tries. Overall accident risk values for the EU 25 alone are lower than for OECD
countries, but differences are mostly quite small. Thus the more statistically robust
estimates obtained for OECD countries can also be considered representative for the
EU 25.
Consideration of regional differences is particularly important for the nuclear and
hydro chains, where expected values for fatality rates and corresponding external
costs due to severe accidents are lowest for Western power plants. At the same time
the consequences of hypothetical extreme accidents are largest in the case of hydro
and nuclear. However, valuation of this aspect is subject to stakeholder preferences
that can be addressed in the context of a sustainability assessment using multi-criteria
decision analysis (Hirschberg et al. 2000; Hirschberg et al. 2004b).
The damages caused by severe accidents in the energy sector are significant,
although still small in comparison to natural disasters. More important, external
cost estimates of energy-related accidents are rather insignificant when compared to
the quantifiable external costs (such as global warming, public health, occupational
health, material damage) of electricity generation, which pose the most serious
problem. Nevertheless, future research should aim to improve the still incomplete
information for some of the unit values for fuel cycle accident endpoints that would
enable a more robust monetization of the various damage categories for accidents
in the energy sector.
Since the establishment of ENSAD, many customers representing a variety of
stakeholder groups have ordered studies to be carried out, based on tailored ENSAD
queries. This demonstrates that the use of ENSAD is not restricted to purely scien-
tific purposes, but can contribute to decision-making processes for energy policies,
the realization of safety goals, and improved technology transfer to other countries
(Burgherr et al. 2005).
Acknowledgments
The authors thank Dr. Warren Schenler and three anonymous reviewers for their
valuable comments and critiques on an earlier version of this article. This study was
partially performed within the Integrated Project NEEDS (New Energy External-
ities Development for Sustainability, Contract No. 502687) of the 6th Framework
Programme of European Community.
REFERENCES
Burgherr P. 2005a. Unfallrisiken im Energiesektor. Gas Wasser Abwasser 85:683–93
Burgherr P. 2005b. Unfallrisiken von Erdgas: eine umfassende Analyse. Gas Erdgas GWF
146:574–81
Burgherr P and Hirschberg S. 2005a. Comparative Assessment of Natural Gas Accident Risks.
PSI Report No. 05-01. Paul Scherrer Institut, Villigen PSI, Switzerland
Burgherr P and Hirschberg S. 2005b. Severe accident database for comprehensive risk assess-
ment in the energy sector. In: Mock R and Birchmeier J (eds), Risk and Safety: Impacts
970 Hum. Ecol. Risk Assess. Vol. 14, No. 5, 2008
Comparison of Severe Accident Risks in the Energy Sector
on Research and Education. Step-Commerce AG / MAXIMA, Zurich, Switzerland, pp. 83–
93
Burgherr P, Hirschberg S, Hunt A, et al. 2004. Severe Accidents in the Energy Sector. Final
Report to the European Commission of the EU 5th Framework Programme ”New Elements
for the Assessment of External Costs from Energy Technologies” (NewExt). DG Research,
Technological Development and Demonstration (RTD), Brussels, Belgium
Burgherr P, Hirschberg S, and Hardegger P. 2005. The Severe Accident Database ENSAD.
PSI Scientific Report 2005, vol 3, Technology Transfer, pp 78–79. Paul Scherrer Institut,
Villigen PSI, Switzerland
CCPS (Center for Chemical Process Safety). 1989. Guidelines for Chemical Process
Quantitative Risk Analysis. American Institute of Chemical Engineers, New York, NY,
USA
Chernobyl Forum (IAEA, WHO, UNDP, FAO, UNEP, UN-OCHA, UNSCEAR, World Bank,
Governments of Belarus, the Russian Federation and Ukraine). 2005. Chernobyl’s Legacy:
Health, Environmental and Socio-Economic Impacts and Recommendations to the Gov-
ernments of Belarus, The Russian Federation and Ukraine. The Chernobyl Forum: 2003–
2005. Second revised version. IAEA, Vienna, Austria
Dilley M, Chen RS, Deichmann U, et al. 2005. Natural Disaster Hotspots. A Global Risk Analysis.
Disaster Risk Management, Series No 5. The World Bank, Hazard Management Unit,
Washington DC, USA
DNV (Det Norske Veritas). 1999. Worldwide Offshore Accident Databank (WOAD). WOAD
Statistical Report 1998. AS, Hovik, Norway
EC (European Commission). 1995. ExternE Externalities of Energy. Vol 2—Methodology.
A Report produced for the EC—DG XII, Luxembourg. Office of Publications for the
European Communities, Brussels, Belgium
EC. 1999. ExternE Externalities of Energy. Vol 7—Methodology Update. A Report produced
for the EC—DG XII, Luxembourg. Office of Publications for the European Communities,
Brussels, Belgium
EC. 2003. External Costs—Research Results on Socio-Environmental Damages Due to Elec-
tricity and Transport. European Communities, Brussels, Belgium
EM-DAT. 2006. The OFDA/CRED International Disaster Database. Available at
http://www.em-dat.net
Ferguson CC and Kasamas H. 1999. General introduction. In: Ferguson CC and Kasamas H
(eds), Risk Assessment for Contaminated Sites in Europe, vol. 2 Policy Framework, pp 1–6.
LQM Press, Nottingham, UK
Flynn J, Slovic P, and Kunreuther H (eds). 2001. Risk Media and Stigma: Understanding Public
Challenges to Modern Science and Technology. Earthscan, London, UK
Gilbert RO. 1987. Statistical Methods for Environmental Monitoring. Pacific Northwest Lab-
oratory, Van Nostrand Reinhold, New York, NY, USA
Glickman T and Terry K. 1994. Using the News to Develop a World-Wide Database of Haz-
ardous Events. Center for Risk Management, Resources for the Future, Washington, DC,
USA
Guha-Sapir D and Below R. 2002. Quality and Accuracy of Disaster Data: A Comparative
Analysis of 3 Global Data Sets. Working Paper Prepared for the Disaster Management
Facility, The World Bank. CRED, Brussels, Belgium
artler G. 1983. Statistische Methoden f¨ur die Zuverl¨assigkeitsanalyse. VEB Verlag Technik,
Berlin, Germany
Hirsch RM, Alexander RB, and Smith RA. 1991. Selection of methods for the detection and
estimation of trends in water quality. Water Resources Res 27:803–13
Hirschberg S, Spiekerman G, and Dones R. 1998. Severe Accidents in the Energy Sector—First
Edition. PSI Report No. 98-16. Paul Scherrer Institut, Villigen PSI, Switzerland
Hum. Ecol. Risk Assess. Vol. 14, No. 5, 2008 971
P. Burgherr and S. Hirschberg
Hirschberg S, Dones R, and Gantner U. 2000. Use of external cost assessment and multi-
criteria decision analysis for comparative evaluation of options for electricity supply. In:
Proceedings of the ”5th International Conference on Probabilistic Safety Assessment and
Management PSAM 5,” 27 Nov–1 Dec 2000, Osaka, Japan. Universal Academy Press, Inc.,
Tokyo, Japan
Hirschberg S, Burgherr P, Spiekerman G, et al. 2003a. Assessment of severe accident risks.
In: Eliasson B and Lee YY, (eds), Integrated Assessment of Sustainable Energy Systems in
China. The China Energy Technology Program—A Framework for Decision Support in
the Electric Sector of Shandong Province. Alliance for Global Sustainability Series, vol 4,
pp 587–660. Kluwer Academic Publishers, Amsterdam, The Netherlands
Hirschberg S, Burgherr P, Spiekerman G, et al. 2003b. Comparative Assessment of Severe
Accidents in the Chinese Energy Sector. PSI Report No. 03-04. Paul Scherrer Institut,
Villigen PSI, Switzerland
Hirschberg S, Burgherr P, Spiekerman G, et al. 2004a. Severe accidents in the energy sector:
Comparative perspective. J Hazard Mater 111:57–65
Hirschberg S, Dones R, Heck T, et al. 2004b. Sustainability of Electricity Supply Technolo-
gies Under German Conditions: A Comparative Evaluation. PSI-Report No. 04-15. Paul
Scherrer Institut, Villigen PSI, Switzerland
IChemE (Institution of Chemical Engineers) 1985. Risk Analysis in the Process Industries.
IChemE, Rugby, UK
Jones-Lee M and Loomes G. 1995. Scale and context effects in the valuation of transport
safety. J Risk and Uncertainty 11:183–203
Kunreuther HC and Michel-Kerjan EO. 2006. Climate Change, Insurability of Large-Scale
Disasters and the Emerging Liability Challenge. Risk Management and Decision Processes
Center, The Wharton School, University of Pennsylvania, Philadelphia, PA, USA
Lindberg G. 1999. Values of Statistical Life (Risk Value). Prepared for EC UNITE Project,
Brussels, Belgium
Mann HB. 1945. Nonparametric test against trend. Econometrica 13:245–59
Markandya A, Hunt A, Ortiz RA, et al. 2004. Monetary Valuation of Increased Mortality from Air
Pollution. Final Report to the European Commission of the EU 5th Framework Programme
”New Elements for the Assessment of External Costs from Energy Technologies” (NewExt).
DG Research, Technological Development and Demonstration (RTD), Brussels, Belgium
Munich Re. 2001. Topics—Annual Review: Natural Catastrophes 2000. Munich Re Group,
Munich, Germany
Munich Re. 2003. NatCatSERVICE—A Guide to the Munich Re Database for Natural Catas-
trophes. Available at http://www.munichre.com/nathan, Munich Re, Munich, Germany
Munich Re. 2005. Topics Geo—Annual Review: Natural Catastrophes 2004. Munich Re Group,
Munich, Germany
Nivolianitou Z, Konstandinidou M, and Michalis C. 2006. Statistical analysis of major acci-
dents in petrochemical industry notified to the major accident reporting system (MARS).
J Hazard Mater 137:1–7
OSH-ROM. 2007. Bibliographic Databases Covering Critical International Occupational
Health and Safety Information: Major Hazard Incident Data Service (MHIDAS), Library
and Information Services of the Health and Safety Executive (HSELINE). Complete to
11/2007. Ovid Technologies Inc. Copyright Search and Retrieval Software, 1990 to present,
SilverPlatter Information Limited, London, UK
Papadakis GA. 2000. Assessment of requirements on safety management systems in EU regu-
lations for the control of major hazard pipelines. J Hazard Mater 78:63–89
R Development Core Team. 2006. R: A Language and Environment for Statistical Computing.
R Foundation for Statistical Computing. Available at http://www.R-project.org, Vienna,
Austria
972 Hum. Ecol. Risk Assess. Vol. 14, No. 5, 2008
Comparison of Severe Accident Risks in the Energy Sector
Rowe W. 1977. An Anatomy of Risk. John Wiley & Sons, New York, NY, USA
Salmi T, M¨att¨aA,Anttila P, et al. 2002. Detecting Trends of Annual Values of Atmospheric
Pollutants by the Mann-Kendall Test and Sen’s Slope Estimates—The Excel Template Ap-
plication MAKESENS. Publications on Air Quality No. 31. Finnish Meteorological Institute,
Helsinki, Finland
Sen PK. 1968. Estimates of the regression coefficient based on Kendall’s tau. J Am Stat Assoc
63:1379–89
Sirois A. 1998. A Brief and Biased Overview of Time Series Analysis or How to Find That
Evasive Trend. WMO Report No. 133: WMO/EMEP Workshop on Advanced Statistical
Methods and Their Application to Air Quality Data Sets, Helsinki, Finland
Swiss Re. 2001. Natural Catastrophes and Man-Made Disasters in 2000. Sigma No. 2/2001.
Swiss Reinsurance Company, Zurich, Switzerland
Swiss Re. 2004. Natural Catastrophes and Man-Made Disasters in 2003. Sigma No. 1/2004.
Swiss Reinsurance Company, Zurich, Switzerland
Swiss Re. 2006. Natural Catastrophes and Man-Made Disasters in 2005. Sigma No. 2/2006.
Swiss Reinsurance Company, Zurich, Switzerland
UNDP Bureau for Crisis Prevention and Recovery. 2004. Reducing Disaster Risk. A Challenge
for Development. Available at http://www.undp.org/bcpr/disred/rdr.htm
Viscusi WK and Aldy JE. 2003. The value of a statistical life: A critical review of market estimates
throughout the world. J Risk and Uncertainty 27:5–76
Viscusi WK and Aldy JE. 2007. Labor market estimates of the senior discount for the value of
statistical life. J Environ Economics Manage 53:377–92
YueSand Pilon P. 2002. Power of the Mann-Kendall and Spearman’s rho tests for detecting
monotonic trends in hydrological series. J Hydrology 259:254–71
Hum. Ecol. Risk Assess. Vol. 14, No. 5, 2008 973
... A distinction is also made between the member countries of the Organization for Economic Cooperation and Development (OECD) and nonmember countries (non-OECD). This distinction has proven to be meaningful because of the substantial differences in management, regulatory frameworks, and general safety culture between these clusters (Burgherr & Hirschberg, 2008;Hirschberg, Burgherr, Spiekerman, & Dones, 2004). The second objective of this study is to propose a method to evaluate changes in accident frequencies of dams serving different purposes over the considered 1911-2016 time period. ...
... The primary focus of ENSAD is on severe accidents, that is, accidents fulfilling at least one of seven consequence thresholds (e.g. five or more fatalities, 10 or more injured persons, etc.) (Burgherr & Hirschberg, 2008). ...
Article
This study presents probabilistic analysis of dam accidents worldwide in the period 1911–2016. The accidents are classified by the dam purpose and by the country cluster, where they occurred, distinguishing between the countries of the Organization for Economic Cooperation and Development (OECD) and nonmember countries (non‐OECD without China). A Bayesian hierarchical approach is used to model distributions of frequency and severity for accidents. This approach treats accident data as a multilevel system with subsets sharing specific characteristics. To model accident probabilities for a particular dam characteristic, this approach samples data from the entire data set, borrowing the strength across data set and enabling to model distributions even for subsets with scarce data. The modelled frequencies and severities are combined in frequency‐consequence curves, showing that accidents for all dam purposes are more frequent in non‐OECD (without China) and their maximum consequences are larger than in OECD countries. Multipurpose dams also have higher frequencies and maximum consequences than single‐purpose dams. In addition, the developed methodology explicitly models time dependence to identify trends in accident frequencies over the analyzed period. Downward trends are found for almost all dam purposes confirming that technological development and implementation of safety measures are likely to have a positive impact on dam safety. The results of the analysis provide insights for dam risk management and decision‐making processes by identifying key risk factors related to country groups and dam purposes as well as changes over time.
... We use the ENergy-related Severe Accident Database as sole input to our study [11,13]. Severe accidents are defined in ENSAD as events leading to five or more fatalities, ten or more injuries, at least 200 evacuees, release of hydrocarbons exceeding 10,000 tons, enforced clean-up of land and water over areas of 25 km 2 or greater, or an economic loss equal or greater to $5 million USD (inflation adjusted) [36]. Chains-of-events are extracted for the CIs part of the oil & gas sector. ...
Article
Full-text available
Organized into a global network of critical infrastructures, the oil & gas industry remains to this day the main energy contributor to the world's economy. Severe accidents occasionally occur resulting in fatalities and disruption. We build an oil & gas accident graph based on more than a thousand severe accidents for the period 1970-2016 recorded for refineries, tankers, and gas networks in the authoritative ENergy-related Severe Accident Database (ENSAD). We explore the distribution of potential chains-of-events leading to severe accidents by combining graph theory, Markov analysis and catastrophe dynamics. Using cen-trality measures, we first verify that human error is consistently the main source of accidents and that explosion, fire, toxic release, and element rupture are the principal sinks, but also the main catalysts for accident amplification. Second, we quantify the space of possible chains-of-events using the concept of fundamental matrix and rank them by defining a likelihood based importance measure γ. We find that chains of up to five events can play a significant role in severe accidents, consisting of feedback loops of the aforementioned events but also of secondary events not directly identifiable from graph topology and yet participating in the most likely chains-of-events.
... What, If; Checklist; Fishbone diagram; Case studies; Event Tree; FMECA; etc. [16 in dependence of data quality and amount. They were also considered get-at-able results of other authors [6], [17], [18], [19], [20], [21], [22. The results of these methods were critically assessed and separated into classes according similarity of causes and created the basis for Decision Support System enabling to multicriterial assessment of possible technical facility risks [6]. ...
Article
Full-text available
The aim of risk management of socio-cyber-physical systems at operation is the integral safety which ensures their co-existence with their vicinity throughout their life cycles. On the basis of present knowledge and experience, part of risks that threaten socio-cyber-physical systems is coped by preven-tive measures during their designing and manufacturing. Due to dynamic changes of the world, the con-ditions of socio-cyber-physical systems at operations change. If changes exceed the socio-cyber-physical systems´ safety limits which were inserted into their designs, the accidents or socio-cyber-physical systems´ failures occur. The paper contains the Decision Support System for determination of risk rate of socio-cyber-physical systems. Its regular application shows present-day risk rate and allows to reveal danger situations and in time to apply mitigation measures. The methodology can find several applications in Education, especially in Engineering Education.
... Initially, the indicator is quantified for each generation technology and then aggregated according to the production mix of each country. The fatalities do not only include events from the actual power production, but all stages of the production, transmission and transfer chain ( Burgherr & Hirschberg, 2008 ). A high value indicates that severe accidents are frequent, resulting in a higher fatality rate, occurring in countries with lower safety standards or inadequate technological expertise. ...
Article
Full-text available
The resilience of an energy system and especially the national electricity supply is a complex and multidimensional concept, which is receiving growing attention at a European level. In particular, the increasing risks of extended electricity supply disruptions and/or severe electricity price fluctuations are stressing the need for an assessment of the European countries’ electricity supply resilience. This paper proposes an elaborative multicriteria decision support methodological framework, based on three major resilience dimensions, namely: “resist”, “restabilise” and “recover”. In total, 35 European countries are evaluated and ranked according to their performance on 17 interacting evaluation criteria. The overall evaluation of the countries is established based on a synergy of two complementary multicriteria decision aid (MCDA) methods. Specifically, the Choquet Integral is approached and assessed as an importance indicator to properly accommodate and incorporate the interacting criteria into the value system, negating therefore their arbitrary effects on the final benchmarking. Simultaneously, a procedure based on the Simos method supports the robust elicitation of the Choquet capacities. All preferential information required for the implementation of the MCDA framework has been elicited from an expert, highly knowledgeable of the European energy system. This research work aims to support energy policymakers in Europe and provide guidelines and areas for improvement at a country level.
... What, If; Checklist; Fishbone diagram; Case studies; Event Tree; FMECA; etc. [16] in dependence of data quality and amount. They were also considered get-at-able results of other authors [6,[17][18][19][20][21][22]. ...
Article
The human’s lives are strongly dependent on operation of socio-cyber-physical systems that create vitally infrastructure of human society system. These systems are complex systems, which have form “systems of systems”. Due to complexity, these systems are threatening not only by risks influencing their elements, but also with risks connected with uncomfortable links and flows among elements, which occur at critical conditions due to occurrence of unexpected interdependences. The aim of risk management of socio-cyber-physical systems at operation is the integral safety which ensures their co-existence with their vicinity throughout their life cycles. On the basis of present knowledge and experience, part of risks that threaten socio-cyber-physical systems is coped by preventive measures during their designing and manufacturing. Due to dynamic changes of the world, the conditions of socio-cyber-physical systems at operations change. If changes exceed the socio-cyber-physical systems´ safety limits which were inserted into their designs, the accidents or socio-physical -cyber-physical systems´ failures occur. The presented risk management plan is tool which ensures the correct response to such unaccepted situations and fast ensuring the safety.
... Such crises' ramifications extend to threaten the social fabric, economic propensity, and political structure (Rosenthal and Kouzmin, 1997). For instance, the Chernobyl incident is known as a worse nuclear disaster due to the abrupt surge in power during the reactor systems that caused an explosion, emission of radiation across the Western Union and Europe, and displaced more than 220,000 people (Burgherr and Hirschberg, 2008). In terms of a man-made crisis, the infrastructure failure incidents and structure fire contributed in formulating the governmental response based on the crisis concepts. ...
Article
Crisis management is a relatively new field. It was founded in the mid-1980s following several natural and man-made crises stemming from modernisation, according to the pioneers of the field. This paper’s aim is to discuss the theoretical models and concepts required to manage a crisis. Moreover, it addresses public administration challenges with dealing with a crisis. The research strategy is built upon Cooper’s taxonomy of a literature review to determine the appropriate methodology for achieving an inclusive and critical review for crisis management. A theoretical literature review is conducted to focus on theoretical concepts so as to foster an understanding of the management mechanisms of crisis and the main policy landmarks to mitigate the repercussions of the crisis. This paper addresses the concept of risk society, the typologies of crisis, the interaction risk dimensions, and the role of public administration in responding to the crisis. It considers international crisis management due to the spillover effect of the crisis. It also offers some discussion about the challenges of managing the crisis, including the impossibility of circumscribing the effect of the crisis, the effect of politics in responding to the crisis, and the small events that may cause greater consequences
... Initially, the indicator is quantified for each generation technology and then aggregated according to the production mix of each country. The fatalities do not only include events from the actual power production, but all stages of the production chain ( Burgherr and Hirschberg, 2008;Hirschberg et al., 2004). The specific normalized fatality rates per technology and geographical region are given in Table S4. ...
Article
Full-text available
Indicator-based approaches are suitable to assess multi-dimensional problems. In order to compare a set of alternatives, one strategy is to normalize individual indicators to a common scale and aggregate them into a comprehensive score. This study proposes the Electricity Supply Resilience Index (ESRI), which is a measure of a nation's electricity supply resilience. Starting from an initial set of individual indicators derived through a structured selection process, the ESRI is calculated for 140 countries worldwide. To account for robustness of the resulting resilience index, 38 combinations of eight normalization methods and six aggregation functions were considered. Results show a clear country ranking trend, with robust top-and low-performing countries across all combinations. However, the ranking disparity becomes large for average performing countries, especially if their indicators show high variability. Furthermore, the differences of the rankings are quantified through the Rank Difference Measure (RDM), which identifies the categorical scales and the minimum aggregator as the most different ones. Finally, the effects of different compensation levels of the aggregation functions are discussed. The findings of the present study aim to provide recommendations for policymakers on how composite indexes results depend on assumptions and chosen approaches.
Article
Full-text available
The aim of risk management of socio-cyber-physical systems at operation is the integral safety which ensures their co-existence with their vicinity throughout their life cycles. On the basis of present knowledge and experience, part of risks that threaten socio-cyber-physical systems is coped by preven-tive measures during their designing and manufacturing. Due to dynamic changes of the world, the con-ditions of socio-cyber-physical systems at operations change. If changes exceed the socio-cyber-physical systems´ safety limits which were inserted into their designs, the accidents or socio-cyber-physical systems´ failures occur. The paper contains the Decision Support System for determination of risk rate of socio-cyber-physical systems. Its regular application shows present-day risk rate and allows to reveal danger situations and in time to apply mitigation measures
Article
Although energy systems are well known to lead to positive or negative externalities, one less explored attribute has been a rigorous historical and future looking assessment of energy accidents. In this study, we analyze an extensive dataset of 4,450 energy accidents from 1800 to 2018 across eleven energy systems. Our analysis reveals that these collective energy systems resulted in more than 278,000 human fatalities and approximately $421.3 billion in economic damages. Historically, coal accidents are the most frequent, accounting for almost half of all accidents. In terms of severity, accidents at hydroelectric dams were the most fatal, accounting for 67 percent. In terms of cost, nuclear power accidents are by far the most expensive, accounting for 62 percent of damages. Coupling our data and an econometric model with future projections of energy demand underscores the magnitude of the trends identified: 986,000 to 1.72 million potential energy accident deaths in 2040, as well as almost $1 trillion in damages. This leads to compelling policy implications, especially concerning the need for safety improvements in energy systems such as bioenergy and nuclear power, as well as the need for the IEA and IRENA, among others, to begin to better track and account for energy accident trends. We find that across all accidents, fuel extraction and processing, transmission and distribution, and transportation have the most fatalities, and yet conversion and operation, transmission and distribution, and transportation have the most damages. Moreover, achieving strong climate goals leads to an unacceptably higher risk of accidents and human health and economic consequences. Finally, as its economic development propels increases in energy consumption, Africa will become the future center for energy accident fatalities.
Article
Full-text available
The purpose of this paper is to present a knowledge management approach to complex crises, having as a reference the multiple crises generated by COVID-19, crises we are living with, and for which we cannot anticipate a specific end due to the knowledge absence concerning this complexity. A complex crisis is a system crisis that induces new crises in connected systems. The knowledge management approach shows that understanding complex crises and managing them needs an excellent understanding of these crises' knowledge gaps and the knowledge dynamics phenomena able to describe their manifestation and evolution. There is no a priori knowledge or model to give us the key to coping with these crises and designing the best solutions for all the problems generated by them. The present research is based on qualitative research of the significant literature by using the Grounded Theory. The findings show that a knowledge management approach offers new insights into the complexity of these phenomena, which can be symbolized by the Black Swan metaphor defined by Nassim Taleb in his seminal book The Black Swan. The impact of the highly improbable.
Article
Full-text available
One result of increased scientific and public interest in water quality over the past few decades has been the gradual accumulation of reliable long-term quality data records and an interest in examining these data for long-term trends. This paper summarizes and examines some of the major issues and choices involved in detecting and estimating the magnitude of temporal trends in measures of stream water quality. The first issue is the type of trend hypothesis to examine: step trends versus monotonic trend. The second relates to the general category of statistical methods to employ: parametric versus nonparametric. The third issue relates to the kind of data to analyze: concentration data versus flux data. The fourth relates to issues of data manipulation to achieve the best results from the trend analysis. These issues include the use of mathematical transformations of the data and the removal of natural sources of variability in water quality due to seasonal and stream discharge variations. The final issue relates to the choice of a trend technique for the analysis of data records with censored or less than values. The authors' experiences during the past decade with the development of several trend detection techniques and application of these techniques to a large number of water quality records provide insight into the issue related to a choice of a statistical test for trend in water quality.
Article
Natural gas shows both the lowest accident fatality rates and smallest maximum credible consequences of all fossil energy carriers. Western hydropower and nuclear power have even lower rates, but maximum consequences may be very large. Within the natural gas chain fatality rates are lowest for "Central Europe" followed by EU15 and OECD, whereas non-OECD is substantially more accident-prone. Finally, it could be demonstrated by means of the detailed DVGW-statistics for natural gas accidents in Germany that normalized accident rates for fatalities and injured decreased over the last two decades.
Article
This chapter deals with the Risk Assessment Task within CETP. This task addressed the comparative assessment of accidents risks associated with the various electricity supply options. The full account of the work carried out within this task is provided in the technical report (Hirschberg et al., 2003a).
Article
The least squares estimator of a regression coefficient β is vulnerable to gross errors and the associated confidence interval is, in addition, sensitive to non-normality of the parent distribution. In this paper, a simple and robust (point as well as interval) estimator of β based on Kendall's [6] rank correlation tau is studied. The point estimator is the median of the set of slopes (Yj - Yi)/(tj-ti) joining pairs of points with ti ≠ ti, and is unbiased. The confidence interval is also determined by two order statistics of this set of slopes. Various properties of these estimators are studied and compared with those of the least squares and some other nonparametric estimators.