AMMONIUM NITRATE IN PORTS STORAGE AND TRANSPORTATION

Transcription

1 AMMONIUM NITRATE IN PORTS STORAGE AND TRANSPORTATION Robert Hutchison and Philip Skinner Lloyd s Register; robert.hutchison@lr.org and philip.skinner@lr.org Large quantities of ammonium nitrate are manufactured and transported by ship around Australia supplying the main raw material for explosives used by the mining industry. Thus there is the potential for accidental s involving hundreds or thousands of tonnes of ammonium nitrate in stockpiles near ports awaiting loading to ships or on board a ship. The ammonium nitrate used as a raw material for explosives has a lower density and a higher porosity than the ammonium nitrate used in the fertiliser industry, which makes it more sensitive to propagation. There have been numerous accidents involving ammonium nitrate around the world over the past century involving various grades and mixtures of ammonium nitrate. These include the major accidents at Oppau, Texas City and Toulouse. More recently, there have been terrorist attacks that have used ammonium nitrate based explosives such as the Oklahoma City bombing. Risk assessments of the transportation of ammonium nitrate must take account of all these factors, which requires addressing the following questions:. How should the consequences of an of ammonium nitrate be modelled? What is the TNT equivalence of pure and contaminated ammonium nitrate? Will the ammonium nitrate detonate or deflagrate?. How should the consequences of a fire involving ammonium nitrate be modelled? Can the grade of ammonium nitrate support combustion? What are the products of combustion?. How should the likelihood of an accidental involving ammonium nitrate be estimated? Is history a good predictor of future s?. How should the likelihood of terrorist activities be estimated?. What fraction of a load or stockpile of ammonium nitrate could explode? What fraction of a load or stockpile of ammonium nitrate is likely to explode? Risk assessments prepared for Australian ports that handle significant quantities of ammonium nitrate are used to provide guidance on the above questions. This will assist the preparation of future risk assessments to accurately assess the risk associated with transportation of large quantities of ammonium nitrate by ship. KEYWORDS: ammonium nitrate, fire,, transportation, ports, shipping INTRODUCTION Within Australia, there is a large market for explosives to support the mining industry. The main explosive used in Australian mining has ammonium nitrate as a precursor. Ammonium nitrate is both manufactured in Australia and imported from overseas. Due to the size of the market in ammonium nitrate, large quantities are transported both internationally and within Australia. Ships are used for the international transportation of ammonium nitrate and potentially for movements between the east and west coasts of Australia. The local manufacturers and the importers of ammonium nitrate want to minimise their costs and so large shipment sizes have occurred and are proposed for the future within Australia. Where the load is sufficiently large that a ship can be chartered for exclusive use, the manufacturer or importer has increased control over the condition of the ship and other specific aspects of the transportation. This can improve the delivered quality of the material through reduced potential for contamination and less damage to the bags. This also reduces the transport cost per tonne. However, larger shipments have potentially larger accident consequences. The larger a shipment of ammonium nitrate, the larger is the maximum possible, the distance travelled by smoke from a fire and the pollution potential. This paper examines the changes in risk and costs that occurs with changes in the shipment size of ammonium nitrate. This paper focuses on the accidental risks associated with transportation of ammonium nitrate and does not consider smoke from fires and the risks associated with terrorism. Previous Quantitative Risk Assessments (QRAs) that have been undertaken in Australia were used to provide the scenarios and likelihoods. Indicative costing estimates have been provided by people involved in the transportation of ammonium nitrate. 1

2 This paper focuses on the differences in cost and risk associated with an annual trade of 1 tonnes (te) of ammonium nitrate through an Australian port. This is a large but credible annual trade volume for an Australian port. The shipment sizes considered are 1 te, 2 te, 5 te, 1 te, 2 te, 5 te, 1 te and 2 te. SCENARIO IDENTIFICATION Many of the scenarios that are considered in QRAs are not affected by the shipment size. For example the risks associated with of a truck carrying ammonium nitrate to or from the port are only affected by the truck load size and the annual trade, not the quantity that is on the ship transporting the ammonium nitrate. The scenarios that have been considered in this paper are those considered in previous QRAs in Australia: 1. A small of 2 te of ammonium nitrate. This is considered a small only by comparison to the quantities that can be carried on a ship. A fire on a ship could conceivably directly affect 2 te of ammonium nitrate, causing it to be heat affected or contaminated by fuel or other organic material. The fire could then cause the 2 te of contaminated heat affected AN to explode. 2. A partial of 2% of the ammonium nitrate in the shipment. The scenario that is envisaged here is the of 2 te of ammonium nitrate boosting a fraction of the ammonium nitrate carried on the ship. The historical record suggests that only a fraction of stockpiles of ammonium nitrate involved in s have contributed to the overpressure wave. 3. A complete of 1% of the ammonium nitrate in a full shipment. This scenario is considered the worst credible accident and could occur due to an uncontrolled fire in a fully laden ship with all the ammonium nitrate stored in one hold or in close proximity. The uncontrolled fire could cause a fraction of the ammonium nitrate potentially contaminated and heat affected by the fire to detonate and then to propagate through the rest of the shipment. CONSEQUENCES OF SCENARIOS In assessing the consequences of the scenarios, there are a number of important parameters. 1. TNT equivalence. The TNT equivalence of pure ammonium nitrate is considered to be in the range of 3% to 55% based on both theoretical and experimental results. In this study an equivalence of 34.6% is used based on the ratio of the heat of detonation of ammonium nitrate of 378 kcal/kg to the heat of of TNT of 194 kcal/kg (Lawrence Livermore National Laboratory 22). If the ammonium nitrate is contaminated during the accident with a fuel, the TNT equivalence is increased to approximately 1.. For the large shipments of ammonium nitrate being considered in this analysis, it is unlikely for an accident to cause contamination of a significant fraction and a TNT equivalence of 34.6 has been used. Despite the variation in the values of TNT equivalence for ammonium nitrate, this is a parameter with less uncertainty than the others in the analysis and the differences in equivalence do not greatly affect the consequence distances. 2. Fraction of inventory that contributes to the. The fraction of inventory that contributes to the overpressure shockwave produced by the is a critical parameter but is very uncertain. In historical accidents, the fraction of the inventory that has contributed to the overpressure has ranged over two orders of magnitude from less than 1% at Cherokee Nitrogen in 1973 (Freeman 1975), to 1% at Oppau in 1921(Medard 199), to 1% 6% for Toulouse in 21 (Creemers, et al. 22) and close to 1% at Texas City in 1947 (Klintz, et al. 1947). The reasons for the differences in the inventory fractions that contributed to the s include differences in the material or grade of ammonium nitrate, the degree of confinement and the proximity of the initial to the bulk pile. In QRAs within Australia, the uncertainty surrounding the fraction of inventory that will contribute to the shock wave has been accommodated by postulating different likelihoods for various scenarios. In this study, the three different accident scenarios primarily differ in the quantity of ammonium nitrate that explodes and thus are considered with different likelihood estimates. This method implicitly utilises a risk based framework where both the consequence and likelihood are considered. The regulators in Australia have established risk-based criteria for assessing new developments. 3. Explosion modelling. Most recent QRAs in Australia on ammonium nitrate have modelled of an equivalent quantity of TNT to represent the of the ammonium nitrate. The relationship between quantity of TNT and overpressure is well known and documented (Mannan 25). The modelling estimates the overpressure as a function of distance. There are also rudimentary models that estimate the shrapnel distribution from an but they are not considered further in this study. The relationship between overpressure and likelihood of fatality varies whether the affected person is inside or in the open air. In this study, an overpressure of 14 kpa was considered to have negligible fatality risk, 21 kpa to have a 2% fatality risk, 35 kpa to have a One accident that has occurred was where a bunker hatch at the bottom of a hold was not sealed correctly. After loading with ammonium nitrate bags, the ship filled its bunker tanks. The fuel flowed into the base of the hold and over the duration of the voyage soaked into the bags. If this material was exploded a TNT equivalence of 1 would be appropriate. 2

3 5% fatality risk and 7 kpa to have a 1% fatality risk (DIPNR, 199). 4. People in the vicinity of the. The modelling must consider the people who may be in the danger zone when an occurs. In addition, the property that may be damaged also should be considered, particularly to assess the potential for domino or knock-on accidents. In this study, consequences have been restricted to fatalities to people. The port considered in this study is fictitious but is based on the numbers of people at distances from a number of the ports that handle ammonium nitrate in Australia. In the immediate vicinity of the ship (,5 m from the centre of the accidental ), there are likely to be 2 people including ship personnel, stevedores and inspectors. In the area between 5 m and 1 m from the centre of the accidental, there is likely to be 4 people, primarily associated with security, port administration and truck flow management. Between 1 m and 5 m there are likely to be other ships and local storages and buildings on the port. 5 people are assumed to be present in this region. The area between 5 m and 1 m is likely to include administration buildings, as well as general port facilities, including warehouses, storages, container loading onto trucks or rail cars. In this large area 1 people are assumed to be present. Beyond 1 m is likely to be commercial areas and residential areas. The population density in this area is assumed to be 2/ha (which is the average residential density in Sydney). Figure 1 illustrates this data and shows the concentration of people at the ship during the loading/unloading operation. However the density of population generally increases with distance from the ship, as does the numbers of people. In most of the scenarios, a fire precedes the. This period can be used to evacuate people from the vicinity of the ammonium nitrate, which can significantly reduce the numbers of people who may be killed or injured by an. This factor has not been considered in this study. People in Rings Surrounding Ship Local Population Distance (m) People per Hectare Population Population Density Figure 1. Populations surrounding shipment of ammonium nitrate LIKELIHOOD The likelihood of the scenarios is the area of greatest uncertainty. Since the three ship s in 1947 and the one in 1953, I am unaware of any ship s involving ammonium nitrate. Following those s (over 5 years ago), significant changes were made to the composition of ammonium nitrate and the emergency response actions that would occur in the event of a fire. However, the historical record cannot show that a ship is impossible. Thus, fault trees have been developed for numerous QRAs, which identify the causal sequences that are required for an to occur and suggest likelihood or frequency values for the scenarios. The likelihood estimates of the scenarios considered in this study are based on recent QRAs prepared in Australia. These likelihood estimates are:. 2 te AN on a ship in a port for loading or unloading of ammonium nitrate: per cargo.. Partial of shipment (2% of full load): p.a.. Complete of shipment (1% of full load): p.a. RISK OF ACCIDENT SCENARIOS The risk of the s is estimated taking into account the physical consequences and the likelihood. In Australia, the New South Wales Department of Planning criteria are used by many state governments to assess the risk of proposed developments. The criteria are based on location specific individual fatality risk contours and the risk of defined overpressure and heat radiation levels. The criteria apply to various land uses e.g. a new development should not expose any residential land to fatality risks above p.a. Another criterion that is often considered is societal risk expressed as an FN curve. There are no official FN criteria in Australia but there are various criteria that have been applied elsewhere in the world. However, in a comparative risk assessment, such as the subject of this study, a single measure of risk was desired. The Potential Loss of Life (PLL) is the summation of the individual fatality risk levels at all the locations where a person is assumed to be located. This estimate of risk does not take account of the potential for people to be injured but not killed, neither does it include the potential for shrapnel to strike people, for people to be injured by smoke from a fire or the potential for property to be damaged. COST OF SHIPPING The cost of shipping ammonium nitrate comprises two components: 1. A fixed administration cost per shipment which is independent of the size of the shipment. This is assumed to be $5. 3

4 Cost per Tonne $2 $15 $1 $5 Average cost per tonne $ Shipment Size (te) Figure 2. Average cost per tonne of ammonium nitrate shipments 2. The cost of the shipping. For loads smaller than a full ship, the cost is a fixed price per tonne (assumed to be $12 per tonne). For loads of 5 te and above, a full ship can be chartered. The cost for loads of 5 te to 2 te is assumed to be $12 5 ¼ $6. The costing is shown in Figure 2. It is significantly less costly to transport larger shipments, particularly if a ship can be chartered for a full load. RESULTS The overpressure produced by s of shipments of ammonium nitrate is shown in Figure 3. The distances to fatal overpressures of 35 kpa range from less than 2 m for 1 te of AN to approximately 1 m for 2 te of AN. The distances to overpressure that will not cause fatality and only has a low likelihood of injury (3.5 kpa) range from 1 km for 1 te AN to 5.6 km for 2 te AN. The fatality risk to people located in the vicinity of a port undertaking 1 te shipments of ammonium nitrate is shown in Figure 4. The fatality risk to people located close to the ship (,15 m) is dominated by the small ship because the likelihood is higher. The risk to people located between 15 m and 5 m is dominated by the partial ship and for distances between 5 m and 14 m the risk is dominated by the complete ship. The magnitude of the risks at all locations is 1.E-6 1.E-7 1.E-8 1.E-9 Fatality Risk 1.E Small ship Partial ship Complete ship Total Figure 4. Location specific individual fatality risk surrounding 1 te shipments very low, less than p.a. This value should be juxtaposed with the NSW Department of Planning individual fatality risk criterion for sensitive locations in the vicinity of a proposed development, which is p.a. The societal risk as a function of shipment size is shown in Figure 5. The larger shipments have generally lower likelihoods of s but the consequences are substantially larger. With the smallest shipment size (1 te), the likelihood of a complete ship is p.a. and this could kill 28 people. All these people are working on the port and have some degree of voluntary acceptance of risk. With the largest shipment size, the likelihood of a complete ship is much lower at p.a. but the number of people who could be killed is much higher at 36. Also, the majority of these people would be members of the public with no voluntary acceptance of the risk. The criteria lines shown are the indicative societal risk criteria suggested in NSW but are not mandatory. The societal risk associated with 1 te and 2 te shipments is closer to the lower criteria line. The smaller shipments lie closer to the upper criteria line and the 1 te and 2 te shipments extend beyond the upper criteria line because they could cause in excess of 1 fatalities, which is the limit tolerable using these criteria lines. However, other criteria lines are used in other jurisdictions and the conclusions of the analysis would be different. Peak Overpressure (kpa) Distance (m) Figure 3. Overpressure from s of various sized shipments F (frequency of events causing N or more fatalities) 1.E-4 1.E-5 1.E-6 1.E-7 1.E-8 Societal Risk 1.E N (Fatalities) 1 te 2 te 5 te 1 te 2 te 5 te 1, te 2, te Lower Limit Upper Limit Figure 5. Societal risk as a function of shipment size 4

5 Cost vs Benefit.4 $2 PLL $15 $1 $5 Shipment Cost ($/te) Potential Loss of Life Average Cost of Shipment Ratio $ Shipment Size (te) Figure 6. Costs and benefits as a function of shipment size Figure 6 shows the comparison between the cost of larger shipments to society due to the potential risk and the benefits to the owner of the shipment through lower costs. The overall fatality risk, measured by PLL, decreases as a function of shipment size until the 2 te shipment size is reached. The PLL is larger for 5 te shipments, lower for 1 te shipments and higher for 2 te shipments. These changes in fatality risk are due to the combination of the further extent of larger s, the population distributions and the lower likelihood of larger s. The lowest PLL is at the 1 te shipment size. The ratio between the shipment cost and the PLL is lowest at the 2 te shipment size. Either side of the 2 te shipment size, the changes in PLL are greater than the changes in the shipment cost. However the risks are still very low for all the shipment sizes. CONCLUSIONS The risks of transporting large shipments of ammonium nitrate through Australian ports is low due to the very low likelihood of large s coupled with the significant distances (.1 km) to large populations. The risk is likely to meet current individual fatality risk based criteria. The societal risk associated with shipments of ammonium nitrate varies significantly with larger shipments. The distance from the ship to residential or commercial populations is an important factor in determining the fatality risk. For the population distribution and shipment costs considered in this study, the lowest ratio between PLL and shipment cost is for a shipment size of 2 te. This shipment size also corresponds to the societal risk that is closest to the lower societal risk criterion suggested for use in NSW. When considering the tolerability of the risk from larger shipment sizes, it is important to consider societal risk and not just the individual fatality risks as the magnitude of the consequences can be significant if a large shipment of ammonium nitrate explodes. REFERENCES Creemers, A.F.L., Kersten, R.J.A., van der Steen, A.C., Opschoor, G. 22, The ammonium nitrate in Toulouse, France The incident and its consequences for industrial activities, TNO web site. DIPNR 199, Risk Criteria for Land Use Planning, NSW Department of Infrastructure, Planning and Natural Resources: Hazardous Industry Planning Advisory Paper No. 4 Freeman, R. 1975, The Cherokee Ammonia Plant Explosion, Chemical Engineering Progress 71(11), November 1975 Klintz, G.M., Jones, G.W. and Carpenter, C.B. 1947, Report of Investigations Explosions of Ammonium Nitrate Fertilizer on Board the S.S. Grandcamp and S.S. High Flyer at Texas City, Tex., April 16, 17, Lawrence Livermore National Laboratory 22, Cheetah Model Description, web page current in September 22, Mannan, S. 25, Lees Loss Prevention in the Process Industries, 3 rd Edn. Medard, L.A. 199, Accidental Explosions Volume 2: Types of Explosive Substances, Translator P. Fawcett, Ellis Horwood Limited, chapter 23 (Ammonium nitrate and its thermal decomposition) and chapter 24 (The explosive properties of ammonium nitrate). 5