8 risk assessment

8.1 Introduction

8.1.1 Background

The background to the EIA Study and the description of the Project are provided in Sections 1 to 3 of this report. This section presents the methodology, findings and recommendations of the Hazard to Life assessment with regard to the operation of the biodiesel plant.

8.1.2 Legislation Requirement and Evaluation Criteria

The requirement for a Quantitative Risk Assessment (QRA), for projects where risk to life is a key issue with respect to Hong Kong Government Risk Guidelines (HKRG), is specified in Section 12 of the Technical Memorandum on Environmental Impact Assessment Process (EIAO-TM). Annex 4 of the EIAO-TM specifies the Individual Risk and Societal Risk Guidelines.

Individual Risk

Individual risk is the predicted increase in the chance of fatality per year to a hypothetical individual who remains 100% of the time at a given stationary point.

The individual risk guidelines require that the maximum level of off-site individual risk associated with a hazardous installation should not exceed 1×10^-5 per year.

Societal Risk

Societal risk expresses the risks to the whole population. The HKRG is presented graphically in Figure 8.1a. It is expressed in terms of lines plotting the frequency (F) of N or more deaths in the population from incidents at the installation. Two FN risk lines are used in the HKRG to demark “acceptable” and “unacceptable” societal risks. The intermediate region indicates the acceptability of societal risk is borderline and should be reduced to a level which is “as low as reasonably practicable” (ALARP). It seeks to ensure that all practicable and cost-effective measures which can reduce risks will be considered.

Figure 8.1a Hong Kong Government Risk Guidelines

8.1.3 Study Objectives

The objective of this risk study is to assess the risk to life of the general public, including the workers of nearby plants, from the proposed facility during its operational phase. The results of the QRA are compared with the HKRG.

The detailed requirements of the study (see Section 3.4.1.1 of the EIA study brief) are as follows:

· To identify hazardous scenarios associated with the Project operation and then determine a set of relevant scenarios to be included in a Quantitative Risk Assessment (QRA);

· To execute a QRA of the set of hazardous scenarios identified, expressing population risks in both individual and societal terms;

· To compare the individual and societal risks with the criteria for evaluating hazard to life stipulated in Annex 4 of the EIAO-TM; and

· To identify and assess practicable and cost-effective risk mitigation measures.

As required in the EIA Study Brief, the methodology used in this study is consistent with previous studies having similar issues, in particular the PAFF EIA ([1]).

8.2 Process Description

This section summarises those aspects of the process that are relevant to the risk assessment.

8.2.1 Plant Layout

The layout of the site is shown in Figure 8.2a. The process and administration buildings are labelled as items 1A/1B/1C and are located in the centre and towards the south of the site. The tank farm is labelled as items 2A-2F and is situated on the north side. Storage tanks are provided with impoundment bunds to contain any leaks from the tanks. The waste water treatment plant (item 3) is situated on the west side of the site, as is the jetty (item 4A) for loading/unloading of barges. The whole site will be surrounded by a perimeter wall about 2m in height. Access to the site will be through an entrance on Chun Wang Street.

The process operations are carried out inside a building, about 13m in height and 46m by 30m in area which is constructed of concrete walls on two sides and steel structures with panels on the other two sides. The building is provided with continuous ventilation as well as emergency ventilation to prevent flammable atmosphere. All equipment and piping inside the building are sealed and there are no continuous emissions inside the building. The processing equipment is located inside a building to enable to control the ambient temperature as well as for limiting the separation distances imposed by the electrical area classification requirements.

According to the Hong Kong Code of Practice for Oil Storage Installations,[2] minimum separations are recommended between adjacent tanks in a tank farm, and between tanks and buildings. In this Code, combustible liquids are classified according to their flash point. Separation requirements depend on the classification of the tank contents. In the proposed plant, most tank contents would fall into Class 3 (lowest risk), for which no separation requirements are specified. Only one tank, the methanol storage tank T10, falls into Class 1 (highest risk), and this tank complies with the layout requirements. It may be noted that although this Code applies to petroleum products, reference to this Code was made based on the flash point and boiling point of materials being handled in this project to determine the adequacy of inter tank separation distances and bund capacities.

A process flow diagram and flow chart are shown in

Figure 8.2b and Figure 8.2c. A summary of the process will be described below.

Figure 8.2a Site Layout

Figure 8.2b Process Flow Diagram

Figure 8.2c Process Flow Chart

8.2.2 Transport Activities

Feedstock Reception and Handling

Grease trap waste (GTW), waste cooking oil (WCO), gas oil, glycerine and other feedstock will be delivered by sealed 10 m³ road tankers or 10 tonne trucks.

Materials will be unloaded at the designated stations as shown in Figure 8.2a. Four unloading bays will be provided. The GTW and WCO will be unloaded via flexible hoses directly to the receiving tanks under a closed system arrangement.

Typical fire safety measures including spill containment, drainage of spills to a safe location, fire sprinkler systems, fire detection systems, provision of means of firefighting including hydrants and extinguishers, and adequate access for emergency services will be provided. All tanker loading and unloading operations will be supervised by trained personnel.

Jetty Operations

Palm Fatty Acid Distillate (PFAD) will be delivered to site by barge and pumped from the barge to the storage tank. Similar procedures will apply to the delivery of methanol and shipping out of biodiesel. 1,000 tonne barges will be used for all marine-based transport. It is estimated that about 2 barges per week will be required to transport biodiesel out of the plant.

When marine transport is not possible, e.g. due to adverse weather conditions, biodiesel will be shipped out in 20 m³ road tankers (type D vehicles for conveyance of Category 5 Dangerous Goods similar to those used for transport of petroleum diesel). 10 trucks per day will be required to transport biodiesel out of the plant.

The transportation of feedstock and products to and from the biodiesel plant was tabulated in Table 3.2b and is repeated here in Table 8.2a for convenience.

Table 8.2a Estimated Number of Material Delivery to and from Biodiesel Plant

Material	Vehicle / Barge	Frequency
Land-based Delivery
Grease Trap Waste	10m^3 Sealed Road Tanker	60^(a) per day
Waste cooking oil	Trucks with 20ft containers	5 per day
Animal fat	10m^3 Sealed Road Tanker	4 per day
Gas Oil	10m^3 Sealed Road Tanker	1 per day
Glycerine	10m^3 Sealed Road Tanker	2 per day
Fertilizer	10 tonne truck	1 per day
Nitrogen	10m^3 Sealed Road Tanker	1 per week
Other supplies and deliveries	10 tonne Truck/Tanker	2 to 3 per day
Biodiesel ^(b)	20 m^3 Road Tanker	10 per day
Methanol^(b)	10m^3 Sealed Road Tanker	2 per day
Total		76 to 89
Marine-based Delivery
Biodiesel	1,000 tonne barge	2 per week
Palm oil fatty acid distillate	1,000 tonne barge	1 per 10 days
Methanol	1,000 tonne barge or ISO-tanker barge	1 per week
Total		4 per week
Notes: (a) GTW will be delivered to the site on 24-hour basis. (b) Only when marine transportation is not possible (eg during inclement weather).

8.2.3 Oil/Fat Preparation

Crude oil/fat (grease trap waste, waste cooking oil, animal fats or palm oil fatty acid distillate) which is part of the feedstock is cleaned from impurities in a washing step by adding heated water and steam. The separation step is performed by a series of decanters. Solids are collected in waste containers and the aqueous phase is sent to the waste water treatment plant. The clean oil can either be sent to oil drying columns for further treatment or pumped to the storage tank in the tank farm area directly.

8.2.4 Esterification (with catalyst)

The esterification is a one step batch reaction under atmosphere pressure. This reaction is used to convert free fatty acids (FFA) to methyl ester, which is the main ingredient of biodiesel. In this process, FFA is esterified with methanol to methyl ester and water under acidic conditions. The reaction is catalysed by sulphuric acid and is operated at methanol’s boiling point with reflux condensation. The esterification reactor is first filled with oil, then methanol and sulphuric acid are added according to the recipe. The reactor is heated by an internal heating coil. After the batch is completed, the agitator is stopped and the water phase is allowed to settle before sending to the waste water treatment. The oil phase is then cooled and pumped to the transesterification reactor.

8.2.5 High Pressure Esterification (without catalyst)

Feedstock with a high concentration of Free Fatty Acids (FFA) is esterified with methanol under high pressure and temperature rather than using the catalyst process. This high pressure esterification reaction is a continuous process in a tubular reactor jacketed by steam. The operating pressure of the reactor is 100 barg and the temperature is between 180(C and 240(C. The reaction mixture consists mainly of Fatty Acid Methyl Ester (FME), Glycerine, Methanol, Water, Mono-/Di-/Triglyceride and FFA. Unused methanol and water are separated from the oil phase in a demethanolisation dewatering column. Methanol and water are further separated in a methanol water column and the methanol recovered for reuse.

8.2.6 MEK Preparation

MEK is a mixture of methanol and potassium hydroxide and acts as a catalyst for the transesterification reaction. Methanol and potassium hydroxide combine to form potassium methanolate (CH₃OK).

8.2.7 Transesterification, FME-Purification

The transesterification reaction is used to convert triglycerides to methyl ester and glycerine. The reaction is done in a two stage catalytic reaction with MEK. Oil from esterification reactors (both with catalyst and without catalyst) is pumped to the transesterification reactor. Fresh methanol, recycle methanol and MEK (catalyst) are then added according to the recipe. The reaction mixture is agitated for some time before the heavy glycerine phase (GLP) is allowed to be settled at the bottom. The GLP is discharged to a buffer vessel for further processing. Additional fresh methanol and MEK are added to the remaining mixture and the second stage of the transesterification takes place. After draining the GLP, the remaining content goes through a 3 stage washing sequence.

Water is added during the first washing step which helps separate the soap and glycerine from the methyl ester. The aqueous phase is steeled and drained to the GLP collection tank. Phosphoric acid is added to the transesterification tank during step 2 of the washing sequence. This is mainly to convert potassium soap back to FFA. The heavier phase is then partially discharged to the GLP collection tank. At the third washing step, water is dosed to the vessel again to further remove any remaining acid in the oil phase and improve the separation between the lighter oil phase and the heavier phase. Finally the purified oil phase (Methyl ester; Biodiesel) is discharged to the FME buffer tank.

8.2.8 FME Distillation (including Vacuum System)

The purified FME after the 3 washing steps still contains small amounts of methanol and water. The FME is first heated to 200(C and then flashed to remove most of the remaining methanol and water. The flash drum operates under vacuum condition. The FME then enters 2 distillation columns which are both under vacuum condition to allow moderate distillation temperatures (the two columns operate at 230(C and 250(C). The final FEM contains 96.5% or higher methyl ester. The distillate which contains other reaction by-products is used as heating oil to fuel the column reboilers. The FEM is sent to the quality tank where samples are taken to ascertain product quality. Provision is made to route any off-spec product back to the feed for reprocessing.

8.2.9 Acidulation, Phase-Separation

The glycerine phase from the transesterification reactor is collected in the GLP collection tank. A continuous stream from the collection tank is pumped to the acidulation tank where it is mixed with the acidic water from the esterification reactor, which contains sulphuric acid. Inside the acidulation tank, potassium soap will react with acid and form potassium sulphate (solid phase) and FFA. The reaction also produces 2 liquid phases (GLP and FFA). Decanting is used to separate the 3 phases. The solid phase is discharged to containers and sold as fertilizer, FFA is collected and recycled while the GLP phase will be sent to the neutralization tank for further processing.

8.2.10 Neutralization

The acidic glycerine phase (GLP) is collected in the neutralization tank where the pH is adjusted to 7 by dosing with MEK. The solution is then filtered and enters the demethanolization column.

8.2.11 Methanol and Water Recovery

The solution from the neutralization process, which contains glycerine, methanol and water, is sent to the demethanolization column. Glycerine with small amounts of water exit from the column bottom and are sent to the GLP storage tank. Methanol and water from the top of the column are further separated in the MET recovery column. Liquid methanol from the top of the MET recovery column is collected in the recycle methanol tank and water from the bottom of the column is sent to recycle water buffer tank. Both streams are reused in the process.

8.2.12 Wastewater Treatment Plant

Used process water from the Oil/Fat preparation unit and the process areas are sent to the wastewater treatment plant for treatment before routing to the public sewer. The key components of the wastewater treatment plant will include an oil-water separator, a Dissolved Air Flotation (DAF) system, an Internal Circulation (IC) Reactor (an anaerobic treatment utilising up flow anaerobic sludge blanket (UASB) technology), an aerobic treatment system and a secondary clarifier. The IC Reactor is an anaerobic treatment technology that can effectively reduce the organic loading of the wastewater, especially for wastewater with high organic matter content.

The biogas generated from the IC Reactor has a high energy value and will be used as an energy source for on-site facilities, namely as fuel for the steam boiler. The biogas will be temporarily stored in a biogas buffer tank of 30 m³ capacity, under a pressure of up to 5.5 kPa (0.055 barg). Under normal conditions, all biogas will be consumed by the steam boiler. When the steam boiler is under maintenance, the biogas will be sent to flare.

8.2.13 On-site Storage and Ancillary Facilities

The steam boiler system will make use of towngas, biogas, bioheating oil and biodiesel as energy sources for heating. It is estimated that fuel consumption equivalent to about 8.4 tpd of biodiesel will be required for the boiler system.

24 storage tanks are planned for the storage of feedstock and products. The capacities of the tanks for various materials are presented in Table 8.2b.

Table 8.2b Capacities of Storage Tanks for the Biodiesel Plant

Tank Number	Description of Storage Tank	No.	Capacity (m^3)	Capacity (Days)
1 & 2	Raw GTW Tank	2	1,500 each	4.6 (total)
3	Cleaned Trap Grease Tank	1	1,000	10.3
4 & 5	Dewatered GTW (Lipofit)	2	150 each	3.4 (total)
6	Cleaned WCO Tank	1	1,000	11.3
7	PFAD Tank	1	1,500	16.1
8	Raw Animal Fat Tank	1	500	11.2
9	Cleaned Animal Fat Tank	1	500	11.2
10	Methanol Tank	1	500	14.3
11	Sulphuric Acid Tank	1	50	12.5
12	Phosphoric Acid Tank	1	25	83.3
14	Additive Storage Tank	1	50	15
15 & 16	Biodiesel Quality Tank	2	500 each	3.2 (total)
17	Biodiesel Storage Tank A	1	2,500	14.2
18	Biodiesel Storage Tank B	1	1,200	9.2
19	Glycerine (80%) Tank	1	500	30.2
20	Fertiliser Container	1	20	2.6
21	Bioheating Oil Tank	1	200	7.5
22	Gas Oil Tank (as back up fuel)	1	100	8.3
23	Nitrogen Tank	1	25	16.5
24	Crude WCO Tank	1	1,200	-

8.2.14 Safety Features

All vessels/tanks and other equipment for the biodiesel plant will be designed to meet the applicable safety standards and to comply with mechanical, technical and safety standards for chemical plant design and local regulations. The entire production process will be program-controlled. The process visualisation allows monitoring of the process and intervention if required. The process equipment for the biodiesel production line (such as vessels, machines, pipelines, instruments etc.) will be made of stainless steel or other resistant materials fulfilling the respective mechanical, technical and safety standards. The vessels and pipelines will be insulated by aluminium plate. All vessels will be equipped with agitators and a manhole. Pumps for methanol will be equipped with magnetic coupling to eliminate the problems of leaking seals. All pumps will be monitored by a fully automatic process control system (PCS) to prevent dry running.

Methanol will be stored in a carbon steel storage tank with a double bottom layer and will be maintained at atmospheric pressure. All process tanks and machines will be designed to be gas tight and equipped with a gas displacement system. The whole system will have nitrogen blanketing under positive pressure to prevent air ingress that may otherwise lead to the formation of explosive gas mixtures. The methanol in the exhaust gas will be removed in an air scrubber. A gas warning system will be installed to monitor the methanol concentration inside the process room. The plant will shut down automatically and the emergency ventilation system activated, if the monitoring system detects a methanol concentration of 0.6% v/v inside the room.

The outdoor storage tanks will be built in a bunded area where any spills can be contained. In most cases (some unlikely exceptions are discussed in Section 8.3.2) the impacts of fire caused by loss of containment to tanks would therefore be confined to the bund area and minimise the damage to the surrounding facilities. Bunds for acid storage tanks will be constructed with acid resistant materials.

Explosion Protection

The entire plant is accomplished with equipment according to the required explosion proof class. Open flames and smoking are not permitted. For maintenance and repair works, non-sparking tools will be used.

Each component in which the concentration of methanol is high enough to form an explosive vapour is connected to an inertisation system (ventilation system). Nitrogen is fed to this system to reduce the oxygen content to an amount that no explosive vapour mixtures are formed. Excessive vapour from this system is sent first to a cooling trap in which the methanol is condensed, and then to an exhaust gas washing column. The purified gas is ventilated through a vent above the roof of the process plant.

Furthermore, rotating equipment in which methanol vapour can be present is purged with nitrogen to avoid explosions due to sparks in case of a possible malfunction of the equipment internals.

Within the plant and near all possible methanol emission sources (unloading station, methanol storage, etc.) gas detection instruments will be installed. If gas is detected in the process room, the emergency ventilation is activated automatically and an alarm is displayed in the process control room. Each item of equipment is grounded by proper connections to prevent electrostatic discharges.

Alarms & Shutdown

Every deviation from normal operation condition is reported by the PCS by an alarm. In case of an emergency the process can be stopped by one of the following shut down procedures.

· Loss of utilities - In case of a loss of electrical supply, all electric equipment stops. As the PCS is equipped with an uninterruptible power system, final adjustments for safe shut down and preparation for easy recovery can be made.

There are two redundant cooling pumps installed to maintain cooling. If deviations in temperature occur, the units are shut down automatically by the PCS. In case of complete loss of cooling water, the process is shut down. The cooling capacity in the system allows a controlled shutdown without major evaporation of methanol.

Loss of instrument air or nitrogen automatically activates the protective shutdown procedure.

· Safety pressure relief – Vessels and equipment are fitted with safety pressure relief valves or rupture discs to protect against possible over pressurisation.

· Other measures - The plant will be protected by elaborate fire protection and fire fighting systems.

8.2.15 Plant Personnel

Based on similar existing biodiesel plants, the staffing requirements for the operation of the proposed biodiesel plant will be about 20 in daytime and at least 8 at night time. If necessary, external personnel will be hired for maintenance and repair works.

8.2.16 Plant Documentation

Since the planning of the plant is at a relatively early stage, some plant documents such as safety management system, emergency plan and maintenance system have not yet been finalized. In this assessment, it is assumed that they will be developed later in line with chemical process industry best practices.

8.3 Site Description

The proposed biodiesel plant will be situated in the industrial estate of Tseung Kwan O, along the coast of Junk Bay (see Figure 8.3a).

Figure 8.3a Project Site and its Surroundings

8.3.1 Population Data

The vicinity of the biodiesel plant is generally industrial, with the daytime population significantly exceeding the night time occupancy. A Gammon warehouse and technology park lie to the south, Hong Kong Oxygen about 400m to the north, and the Trade Development Council to the east. Sites labelled as A, B, C and D are currently undeveloped. The nearest high-rise residential buildings are those of the Dream City development, about 800m to the north.

The population within the vicinity of the site was estimated based on a combination of site visits, data provided by the Hong Kong Science and Technology Parks and company websites. The maximum consequence distance from accidents at the facility was calculated at about 300m and so all population within about 500m was considered in the survey. A summary of the estimated population is given in Table 8.3a.

Table 8.3a Current Population in the Vicinity of the Project Site

Site	Day Time Population		Night Time Population
	Outdoor	Indoor	Outdoor	Indoor
Building Population
Gammon Warehouse (North) ^(a)	5	45	1	9
Gammon Technology Park (South) ^(a)	20	180	4	36
Hong Kong Oxygen ^(b)	23	207	5	41
TDC Warehouse ^(a)	30	270	6	54
Asia Netcom Landing Site 1 ^(a)	2	18	1	3
Asia Netcom Landing Site 2 ^(a)	2	18	1	3
HAESL ^(c)	65	585	13	117
Wellcome Warehouse ^(a)	25	225	5	45
Mei Ah ^(d)	21	189	4	38
HAECO ^(e)	37	333	7	67
Sub Total	230	2070	47	413
Road Population
Chun Wang Street (550m) ^(f)	1.5	0	0.3	0
Chun Yat Street (900m) ^(f)	23	0	4.7	0
Chun Kwong Street (370m) ^(g)	1	0	0.2	0
Bus Terminal	10	0	2	0
Sub Total	35.5	0	7.2	0
Marine Population
Water Edge ^(f)	4	0	0.8	0
Junk Bay ^(f)	4	0	0.8	0
Sub Total	8	0	1.6	0
Total	274	2070	56	413
Notes: (a) Populations are estimated based on a total population of 2300 people within 500m of the biodiesel plant. The judgement is based on a site visit and functionality of the building. (b) Hoovers, http://www.hoovers.com/Hong+Kong+Oxygen+&+Acetylene+Company+Limited/--HD__jxjfstyxx,src__global--/free-co-dnb_factsheet.xhtml (c) Hong Kong Aero Engine Services Ltd, http://www.haesl.com/en_frame_facilites.html (d) Hoovers, http://www.hoovers.com/Mei-Ah-Laser-Disc-Co-Ltd/--HD__jjyrcyxky,src__global--/free-co-dnb_factsheet.xhtml, http://www.hoovers.com/Mei-Ah-Video-Production-Company-Limited/--HD__jxjshyxht,src__global--/free-co-dnb_factsheet.xhtml (e) Hong Kong Aircraft Engineering Company Limited Environmental Report 2005, http://www.haeco.com/company_update/HX%20Env%20report%202005.pdf (f) Estimated based on site visit carried out in September 2008. Eight barges were observed anchored within Junk Bay, but no activity was observed on any of the barges. As a conservative assumption, each barge was assumed to have a population of 5 persons indoors giving a total population of 40. Given that the vessels will offer some protection to their occupants, an exposure factor of 0.1 was used in the analysis to give an effective outdoor population of 4. (g) The traffic density for Chun Kwong Street is assumed to be the same as Chun Wang Street which is 2.75 person/km

The night time worker population has been assumed to be 20% of the daytime population. It is also assumed that 90% of the workers would reside indoors, with the remaining 10% being outdoors. A distinction between populations indoors and outdoors is made because the buildings may offer some protection to their occupants from accident scenarios such as fires. Population in vehicles are assumed to be all outdoors.

A distinction is also made between the daytime and night time populations, since significant differences are to be expected. Daytime is defined as 8am to 6pm for 6 days a week and night time from 6pm to 8am. Night time population is also assumed on Sunday. The quoted population estimates represent the average over these time periods.

For marine population, the population of 4 people is distributed evenly over the Junk Bay area of 4km² to derive a population density of 1 person/km². The water edge population of 4 is distributed evenly along the coast of the industrial estate to give a line population.

The traffic populations on the Chun Yat Street and Chun Wang Street were measured during a site visit. One hour of data was collected in the morning between 9am and 10am, and an hour of data was collected in the afternoon between 2pm and 3pm. The daytime traffic population was then calculated by assuming 2 hours of morning traffic and 10 hours of afternoon traffic. Population estimates were obtained by counting the number of vehicles of various types travelling in each direction (Table 8.3b) and multiplying by average occupancy estimates obtained from the Transport Department Annual Traffic Census 2007([3]). These vehicle occupancy estimates are based on cross harbour tunnel traffic and are likely to be conservative for vehicles within the industrial estate.

Assuming an average speed of 20 km hr^-1, the population density on the roads may be calculated from:

Chun Yat Street daytime population =

= 26 persons/km

Similar calculations were performed for Chun Wang Street to give the road population figures presented in Table 8.3b. The site visit indicated that the bus terminal is quiet with few people and so a population of 10 people present continuously was conservatively assumed.

Night time road population is assumed to be 20% of the day time population.

The current population within 500m from the biodiesel plant was estimated by the Hong Kong Science and Technology Parks at 2300. The population data summarised in Table 8.3c was determined so as to be in agreement with this estimate. The Tseung Kwan O Industrial Estate is expected to undergo intensive development in the coming years. Once fully developed, the ultimate worker population within 500m from the project site is estimated by the Hong Kong Science and Technology Parks at 5300. For the purpose of this assessment, these additional 3000 people are assumed to be evenly spread (on a per unit area basis) in the empty lots labelled as A, B, C and D in Figure 8.3a. The road and bus terminal population are also increased proportionally. The future population resulting from this analysis is summarised in Table 8.3c.

Following the above discussion, this QRA study considers two population cases, corresponding to the current and future population estimates. Results are presented for both cases in Section 8.8.

Table 8.3b Traffic Counts near the Project Site

Table 8.3c Future Population Estimates in the Vicinity of the Project Site

Site	Day Time Population		Night Time Population
	Outdoor	Indoor	Outdoor	Indoor
Building Population
Gammon Warehouse (North)	5	45	1	9
Gammon Technology Park (South)	20	180	4	36
Hong Kong Oxygen	23	207	5	41
TDC Warehouse	30	270	6	54
Asia Netcom Landing Site 1	2	18	1	3
Asia Netcom Landing Site 2	2	18	1	3
HAESL	65	585	13	117
Wellcome Warehouse	25	225	5	45
Mei Ah	21	189	4	38
HAECO	37	333	7	67
A	41	369	8	74
B	106	954	21	191
C	53	477	11	985
D	100	900	20	180
Sub Total	530	4770	107	1843
Road Population
Chun Wang Street (550m)	3.5	0	0.7	0
Chun Yat Street (900m)	53	0	10.6	0
Chun Kwong Street (370m)	2.3	0	0.5	0
Bus Terminal	23	0	4.6	0
Sub Total	82	0	16	0
Marine Population
Water Edge	4	0	0.8	0
Junk Bay	4	0	0.8	0
Sub Total	8	0	1.6	0
Total	620	4770	125	1843

8.3.2 Meteorological Conditions

The consequences of accident scenarios, such as the dispersion of flammable gases, depend on meteorological conditions of wind speed, wind direction and atmospheric stability class. Hourly data were obtained from the Tseung Kwan O weather station for the most recent 5 years from 2003 to 2007. These weather data were then rationalised into different combinations of wind direction, speed and atmospheric stability class and the probability of occurrence for each combination determined (see Table 8.3d).

Table 8.3d Tseung Kwan O Meteorological Data (2003-2007)

Direction	Percentage of Occurrence of each Wind Speed (m/s)/Stability Category
	Daytime (9am to 6pm)				Night time (6pm to 9am)				Total
	1.5F	3B	3D	6D	1.5F	3B	3D	6D
N	0.61	2.3	1.78	0.29	12.13	0	3.79	0.8	21.69
NE	0.61	6.11	3.24	0.62	7.02	0	4.29	0.94	22.84
E	0.66	6.01	2.24	0.32	6.47	0	2.9	0.49	19.1
SE	0.23	1.5	0.51	0.08	3.48	0	0.96	0.19	6.95
S	0.19	5.63	0.96	0.12	2.68	0	1.29	0.2	11.06
SW	0.24	1.34	0.55	0.06	5.57	0	1.51	0.13	9.41
W	0.13	0.39	0.11	0.01	2.23	0	0.34	0.03	3.23
NW	0.15	0.32	0.19	0	4.64	0	0.41	0.01	5.73
Total	2.81	23.61	9.58	1.5	44.23	0	15.48	2.8	100
Note: (a) Weather condition 1.5F denotes wind speed of 1.5 m/s and atmospheric stability class F. Similar notation applies to 3B, 3D and 6D.

Note on Atmospheric Stability Class

The Pasquill-Gifford atmospheric stability classes are defined as follows:

A: Turbulent;

B: Very unstable;

C: Unstable;

D: Neutral;

E: Stable; and

F: Very stable.

Atmospheric turbulence is a function of the vertical temperature profile in the atmosphere; the greater the rate of decrease in temperature with height, the greater the level of turbulence. The vertical temperature profile generally depends on conditions of wind speed and cloud cover.

Category A typically occurs in conditions of light wind with strong solar insulation. This leads to rising air pockets, strong vertical mixing and good dispersion characteristics. Stable atmospheric conditions generally occur during light wind conditions, at night time with clear skies. Radiative cooling of the ground leads to a reduced rate of decrease of temperature with height, or even a temperature inversion. This creates a stable atmosphere which inhibits vertical mixing and leads to poor dispersion characteristics.

Category D is neutral and neither enhances nor suppresses atmospheric turbulence. Conditions near class D usually occur during stronger winds and/or overcast conditions.

To represent the range of meteorological conditions possible at the Tseung Kwan O site, 4 weather conditions are considered in the current study: 1.5F, 3B, 3D and 6D.

The annual average temperature and relative humidity were taken to be 25(C and 70% respectively.

8.4 QRA Study Approach

The methodology adopted for the risk assessment comprises the following major elements which are discussed in detail in the following sections:

· Hazard Identification;

· Consequence Analysis;

· Frequency Estimation;

· Risk Summation and Evaluation; and

· Risk Mitigation (if necessary).

The elements of a QRA are shown schematically in

Figure 8.4a. The study focuses on those hazardous scenarios that have a potential to affect the off-site population.

Figure 8.4a Schematic Diagram of QRA Process

8.5 Hazard Identification

8.5.1 Materials Handled on Site

Material Safety Data Sheets (MSDS) were reviewed for all materials handled on site, including feedstock, intermediate products, products and by-products, so as to understand the potential hazards arising from these substances. A summary of the relevant properties of these substances is provided in Table 8.5a.

Methanol

Methanol is used as a reactant throughout the biodiesel process. Methanol (CH₃OH) is a highly flammable liquid which burns with an invisible flame. Release can cause an immediate risk of fire and explosion. Methanol is a volatile, clear, colourless liquid at ambient conditions with weak alcohol odour.

Loss of containment of methanol may lead to a bund/pool fire if ignited, or a flash fire if the dispersing vapour cloud encounters an ignition source. If methanol vapour accumulates in a congested/confined area, a vapour cloud explosion (VCE) may also occur. Nevertheless, unlike most petroleum fires, methanol fires can be extinguished with water.

Methanol is also mildly toxic. Acute exposure by inhalation to high concentrations of methanol vapour can cause irritation to mucous membranes, headaches, confusion, loss of consciousness and even death.

Main Hazard: Highly flammable. Considered extremely flammable when stored at elevated temperature above its boiling point of 64.5 (C.

Crude Oil

Crude oil is the main feedstock for producing biodiesel. The main types of oil used are waste cooking oil (WCO), grease trap waste (GTW), palm oil fatty acid distillate (PFAD) and animal fats. The compositions of these oils are highly variable but consist mainly of triglycerides and free fatty acids. They are viscous liquids or even solids at ambient conditions. They have low vapour pressures, high flash points and high boiling points. This means they are difficult to ignite although they are combustible.

Main Hazard: Combustible

Sulphuric Acid

Sulphuric Acid (H₂SO₄) is a strong mineral acid and is highly corrosive. Pure sulphuric acid is an odourless, clear, colourless, oily liquid. Sulphuric acid reacts violently with water and the reaction is highly exothermic.

Sulphuric acid is not considered toxic. Main occupational risks are skin contact leading to burns and the inhalation of aerosols. Exposure to aerosols at high concentration leads to immediate and severe irritation of the eyes, respiratory tract and mucous membranes and may be fatal.

The reported lethal concentration LC₅₀ for sulphuric acid through inhalation is 510mg/m³ for 2 hours exposure in rats. LC₅₀ for humans is estimated to be 625mg/m³ for 10 min exposure using Lee’s method ([4]). The vapour pressure of sulphuric acid at room temperature (25(C) is less than 0.13 Pa which is equivalent to a saturated concentration of 5mg/m³. This is much lower than the LC₅₀. This suggests that a leak from the sulphuric acid storage tank or other equipment near ambient temperature will not pose any risk to personnel due to inhalation of vapours.

Process equipment containing sulphuric acid at the highest operating temperature is the transesterification vessel at 72 (C. However, the acid is diluted to about 12% in this vessel and so the vapour pressure will be correspondingly lower. For comparison, the vapour pressure of pure acid at 50 (C is 0.4 Pa, corresponding to a concentration of 15 mg/m³, still significantly lower than the LC₅₀. In conclusion, the vapour pressure of sulphuric acid is insufficient to cause dangerous concentrations of vapours and hence sulphuric acid is not considered hazardous to offsite population.

Main Hazard: No significant hazard offsite

Phosphoric Acid

Phosphoric Acid (H₃PO₄) is a strong mineral acid and is a white powder under normal conditions. Phosphoric acid solution is corrosive and may cause severe respiration tract, digestive tract, eye and skin irritation with possible burns. Phosphoric acid is non-toxic and non-combustible.

Phosphoric acid has similar properties as sulphuric acid. The reported lethal limit LC₅₀ through inhalation is 850mg/m³ for 1 hour exposure in rats. LC₅₀ for humans is estimated to be 520mg/m³ for 10 min using Lee’s method. The vapour pressure of phosphoric acid is 0.044 Pa at 25 (C, rising to 1.3 Pa at 80 (C. The maximum vapour concentration at process temperatures of 72(C was estimated at 26 mg/m³, much lower than the LC₅₀. In conclusion, the vapour concentration in air is too low to present any hazards to people offsite.

Main Hazard: No significant hazard offsite

Sodium Hydroxide

Sodium hydroxide (NaOH) is a white solid and forms a strong alkaline solution when dissolved in water with liberation of heat. Sodium hydroxide is corrosive and can cause eye and skin burns. Potential severe respiratory tract, digestive tract irritation with possible burns and damage to mucous membranes. Irritation may lead to chemical pneumonitis and pulmonary edema. Sodium hydroxide is non-toxic and non-combustible.

Although sodium hydroxide has a lethal limit LC₅₀ of 2300 mg/m³/2H (rats), sodium hydroxide is extremely non-volatile. The vapour pressure of sodium hydroxide is 1 mmHg (132 Pa) at 739 (C at which is still well below the lethal limit.

Main Hazard: No significant hazard offsite

Potassium Hydroxide

Potassium hydroxide (KOH) is a white solid and forms a strong alkaline solution when dissolved in water with liberation of heat. It has similar properties as sodium hydroxide. Potassium hydroxide is corrosive and can cause eye and skin burns. Potential severe respiratory tract, digestive tract irritation with possible burns and damage to mucous membranes. Irritation may lead to chemical pneumonitis and pulmonary edema. Potassium hydroxide is non-toxic and non-combustible.

Main Hazard: No significant hazard offsite

Additive (Infineum R408)

Infineum R408 is being added to the biodiesel to enhance its combustion properties. Infineum includes the following hazardous ingredients: solvent naphtha, distillates (hydrotreated light), kerosene, alkylhydroxybenzoate formaldehyde condensate, vinyl acetate, mesitylene, 1,2,4-trimethylbenzene, naphthalene. Inhalation of vapours from the heated product can cause irritation of the respiratory tract and the eyes. It has a flash point of 62(C.

Main Hazard: Flammable

Monoglycerides and Diglycerides

Monoglycerides and diglycerides are the side products generated during the esterification and transesterification process. They have similar properties as biodiesel and are combustible under normal conditions. They pose a minor health hazard including skin/eye/respiratory tract irritation on contract.

Main Hazard: Combustible

Triglyceride

Triglyceride (more properly know as triacylglycerol, TAG or triacylglyceride) is a glyceride in which the glycerol is esterified with three fatty acids. It is common in both vegetable oil and animal fats. The melting point of triglyceride is heavily depending on the length of the fatty acid molecule. Triglyceride with a carbon chain longer than 10 carbons atoms would most likely be a solid at room temperature.

Main Hazard: Combustible

Glycerine

Glycerine is generated during the transesterification reaction. It is a colourless, clear liquid without odour. Glycerine poses a minor health hazard including skin/eye/respiratory tract irritation on contract. It has a rather high flash point, giving it a classification of ‘combustible’.

Main Hazard: Combustible

Potassium Phosphate Monobasic

Potassium phosphate monobasic is an intermediate product. Pure potassium phosphate monobasic is a white crystalline solid. Inhalation or ingestion may cause respiratory and digestive tract irritation.

In the biodiesel production process, this material only appears in a few streams with a maximum concentration of 5%wt. No significant hazards have been identified.

Main Hazard: No significant hazard offsite

Biodiesel

Biodiesel is a non-toxic chemical. The main composition of the biodiesel is methyl ester (over 96%). Biodiesel has a very high flash point of over 125 °C and is not volatile. It is therefore considered as combustible rather than flammable. Biodiesel poses a minor health hazard including skin/eye/respiratory tract irritation on contract.

Main Hazard: Combustible

Biogas

Biogas is generated from the IC reactor in the water treatment plant. Biogas is temporarily stored in the biogas buffer tank of 30 m³. Biogas consists mostly of methane and its properties are very similar to Natural Gas (NG). While it is non-toxic, in high concentrations it could lead to asphyxiation. A loss of containment can lead to jet fire (if stored/transferred under sufficient pressure) or to an explosion if the gas accumulates in a confined space.

Main Hazard: Extremely Flammable

Potassium Sulphate (fertiliser)

Potassium Sulphate is a by-product from neutralizing sulphuric acid with potassium hydroxide during the side product treatment step. No specific hazards are identified for potassium sulphate. It is non-toxic, non-flammable, and non-combustible ([5]).

Main Hazard: None

Gas oil and Bioheating oil

Gas oil and bioheating oil are used as supplementary fuel in the biodiesel plant to operate various process equipment such as boilers. It contains medium sized hydrocarbons (C9-C20) and has similar fire properties to biodiesel. Gas oil is, however, bunded separately since unlike biodiesel, gas oil is not biodegradable.

Main Hazard: Combustible

Other Chemicals

Other chemicals involved in the biodiesel process includes Sodium Sulphate and Nitrogen and are considered to pose negligible risk to the offsite population and only a minimal risk to the on-site work-force.

Based on the list of materials on site, potentially hazardous materials identified include Methanol, Crude oil, Infineum R408 (additive), Mono-/di-/tri- Glycerides, Glycerine, Biodiesel, Gas oil/Bioheating oil and Biogas. These are considered further in the analysis.

Table 8.5a Key Properties of Chemicals

Chemical	CAS #	Normal State	Molecular Formula	MW	Vapour Pressure (kPa)	Vapour Density (Air =1)	Melting Point ((C)	Boiling Point ((C)	Flash point ((C)	Auto-ignition Temperature ((C)	Flammability Limit (%)		LC 50	Main Hazard
											UFL	LFL
Feedstock
Methanol	67-56-1	Liquid	CH3OH	32.04	12.8	1.11	-98	64.5	11	455	36.5	5.5	64000ppm /4 hrs (rat)	Highly Flammable^[1]
Crude Palm Oil Fatty Acid (PFAD)	-	Liquid	-	-	<1	-	-	> 200	> 200	> 250	-	-	-	Combustible
Free Fatty Acid	67254-79-9	Liquid	-	-	-	-	-	> 200	315	400	-	-	-	Combustible
Animal Fat (mainly triglycerides)	-	Solid	-	-	-	-	35	-	274	-	-	-	-	Combustible
Sulphuric Acid (solution)	7664-93-9	Liquid	H2SO4	98.08	0.00013	3.4	-15	310	-	-	-	-	510 mg/m^3 /2 hrs (rat)	None
Phosphoric Acid (solution)	7664-38-2	Liquid	H3PO4	98	0.0038	-	-20	158	-	-	-	-	850 mg/m^3 /1 hr (rat)	None
Sodium Hydroxide	1310-73-2	Solid	NaOH	40	-	-	318	1390	-	-	-	-	2300 mg/m^3 /2 hr (rat)	None
Potassium Hydroxide	1310-58-3	Solid	KOH	56.1	-	-	380	1384	-	-	-	-	-	None
Infineum R408 (additive)	-	Liquid	~20% Naphtha, ~5% Petroleum Distillates, ~5% Kerosene, ~5% Alkylhydroxybenzoate, formaldehyde condensate	-	-	-	-	-	62	-	-	-	-	Flammable
Intermediate Products
Mono/Di Glycerides (Glyceryl Mono – Dicaprylate)	26402-26-6	Solid	-	-	<0.27	-	34	155	180	-	-	-	-	Combustible
Monoglyceride (distilled)	97593-29-8	Solid	-	-	-	-	-	250	100	-	-	-	-	Combustible
Triglycerides	85665-33-4	Solid	-	-	-	-	34	-	200	-	-	-	-	Combustible
Glycerine	56-81-5	Liquid	CH2CHOHCH2OH	92.1	< 0.01	3.1	-	171	199	370	-	-	-	Combustible
Potassium Phosphate Monobasic	7778-77-0	Solid	KH2PO4	136.08	-	-	252.6	-	-	-	-	-	-	None
Products/By-products
Methyl Ester (Biodiesel) ^[2]	67784-80-9 73891-99-3 61788-71-2	Liquid	-	-	<0.27	> 1	-	> 200	130	-	-	-	-	Combustible
Biogas ^[3]	8006-14-2	Gas	CH4	-	-	0.59 to 0.72	-182.5	-161.4	-188	580	5	15	-	Extremely Flammable
Potassium Sulphate (fertilizer)	7778-80-5	Solid	K2SO4	14.26	-	-	1067	1689	-	-	-	-	-	None

Notes:

All data are measured at standard state of 20(C and 101kPa.

Flammability classification is according to COMAH guideline (1999 No. 743); Combustible classification is according to OSHA guideline:

· Flammable: Any substance having a flash point higher than 20 (C and lower than 55(C

· Highly Flammable: Substances having a flash point lower than 21(C which are not extremely flammable, or substances which have a flash point lower than 55(C and which remain liquid under pressure, where particular processing conditions such as high pressure or high temperature may create major accident hazards.

· Extremely Flammable: Any substance having a flash point lower than 0(C and boiling point less than 35 (C or flammable substance maintained above their boiling point or gaseous substances that are flammable at ambient temperature and pressure.

· Combustible: Any substance having a flash point above 100(C.

[1] Methanol is highly flammable for storage at ambient temperature. For handling at elevated temperatures in the process areas, it will be classed as extremely flammable if the temperature exceeds the boiling point.

[2] Methyl ester is a group of similar chemicals. Depending on the raw material, different methyl ester will be produced. The three CAS numbers given are associated with the typical biodiesel produced from a combination of animal fats and vegetable oil.

[3] The properties of biogas are very similar to those of Natural Gas (NG), therefore the data for NG is presented.

8.5.2 Review of Previous Incidents

To investigate further the possible hazards from the biodiesel plant, a review of past incidents at similar facilities worldwide was conducted. This involved the review of accident databases such as MHIDAS and the IChemE Accident Database, as well as internet searches.

Incidents at Same Technology Plants

The proposed biodiesel plant in Tseung Kwan O uses the Biodiesel International (BDI) Technology. There are currently 28 biodiesel plants in Europe using the same technology with a combined plant experience of about 280 plant-years. To date, no noteworthy incidents have occurred at any of the plants, demonstrating an excellent safety record. However, the sample size is not large enough to derive a statistically significant incident frequency for comparison with the EIAO-TM risk guidelines.

Incidents at Other Biodiesel Plants

A review of incidents at other biodiesel plants is provided in Table 8.5b. It can be seen that most incidents are related to ignition of methanol vapours although some fires also occurred involving biodiesel and vegetable oil. Further fire incidents relating to methanol are listed in Table 8.5c. A detailed consideration of previous incidents and their relevance to this study is given below, in later paragraphs.

Biodiesel production in both Europe and the US is growing rapidly. In Europe, for example, one estimate[6] puts the number of plants in operation in 2006 at 65, while another report[7] shows European output growing at more than 50% per year. The European biodiesel industry has an excellent safety record, with no major incidents reported in recent years.

We have also considered the US biodiesel industry, which has experienced a number of plant incidents, unlike in Europe, as shown in the following table. According to some estimates,[8] there could be as many as 200 biodiesel plants in the USA at present, although many of these are likely to be small scale and/or idling (as discussed later in this report). Based on this, we could assume approximately 600 plant years of experience in the US.

As shown in the incident reports below, there have been 3 incidents leading to fatalities onsite in the USA in recent years (although no offsite fatalities). Once again, the sample size is not large enough to determine a statistically significant accident frequency but the data may be used to derive an upper bound. There is evidence to suggest a significant difference in the operating philosophy of biodiesel plants in Europe and the US. Whereas in Europe, plants are managed by professionals experienced in chemical processing, US plants often grow organically from ‘backyard’ or entrepreneur start-up operations. This results in substantial differences in the way the plants are operated and managed. To reflect this, we have investigated the recent incidents in more detail, with the aim of examining their relevance to the proposed plant. This has enabled us to obtain a rough quantitative estimate of the incident frequency in the proposed plant, for comparison with the QRA results, as described in the following section.

Diesel Tanker Related Incidents

A search for incidents relating to diesel road tankers produced the results shown in Table 8.5d. Although several incidents resulted in fuel release, in none of the case was the fuel ignited. This demonstrates that the probability of ignition of diesel fuel is rather low due to its high flash point. The properties of biodiesel are similar to petroleum diesel.

Following this review of past incidents, the main hazard is associated with explosions and fires from methanol due to its low flash point. This QRA study therefore looks at possible methanol incidents in detail, and also the possibility of fires from other flammable/combustible materials.

Tank Failures

Table 8.5e provides a list of storage tank failures. The causes are mostly mechanical failure and corrosion. One case was caused by earthquake. It is interesting to note that in at least 2 cases, the bund failed to contain the whole inventory.

Tank failure and the possibility of bund overtopping is considered in the current assessment.

Table 8.5b Previous Incidents Involving Biodiesel Plants and Related Materials

Date of incident	Location of incident	Material Name	Incident Type	Injuries/ Fatalities	Description	Reference
15/08/2008	DECATURVILLE, TENNESSEE, USA		Fire; Explosion	0 Injured 0 Killed	An explosion took place in a standby biodiesel plant awaiting conversion to glycerine production. The explosion and fire destroyed all the existing stocks of biodiesel, sodium hydroxide, methanol and glycerine. No injury is reported	1
18/05/2008	PRINCESS ANNE, MD, USA	METHANOL/ Biodiesel	Fire; Explosion	1 Injured 1 Killed	2 worker were installing a new methane line at the facility when a massive explosion occurred which blew the walls of the building out and bubbled the roof out. One worker died and the other one injured	2
15/04/2008	CALGARY, CANADA	METHANOL VAPOR	Fire; Explosion	0 Injured 1 Killed	A worker died in an explosion at a biodiesel plant while carrying out welding operation on top of a 30-ft biodiesel settling tank. Fumes and methanol accumulated inside the tank ignited by the welding operation caused the explosion. Fire continued to burn in the tank for several hours until all the biodiesel is consumed	3
04/01/2008	DEFIANCE, OHIO, USA	METHANOL VAPOR	Fire; Explosion	3 Injured 0 Killed	Operator in the biodiesel plant left a manhole cover open on a storage tank holding glycerin, and a spark from an electric motor ignited the methanol vapors that escaped. Three workers inside the biodiesel plant were hurt.	4
07/07/2006	NEW PLYMOUTH, IDAHO, USA	METHANOL VAPOR/ BIODIESEL VAPOR	Explosion	2 Injured 1 Killed	Two explosions happened while one worker was working on top of a tank used to store soy oil. One worker was killed, one suffered 2^nd degree burn and another suffered from smoke inhalation. Nearby highway was shut down for 4 hours.	5
17/02/2006	BAKERSFIELD, CA, USA	METHANOL	Fire	0 Injured 0 Killed	The accident happened outside of the plant building when, during a transfer of methanol, a small spill occurred that ignited (ignition source unknown, probable cause static electricity). The plant was in full production mode when the outside fire spread into the building. The operators followed their training and safety procedures and quickly shut down operations. Then, when they could not contain the fire with on-site extinguishers, they left the premises and quickly notified the Fire Department and Hondo personnel in the other buildings located on the property. No other buildings were affected because they were not in close proximity. Unfortunately, ABF suffered a total loss of the building and equipment. The plant burned violently for several hours and the non-hazardous plumes of smoke could be seen for miles. As a result, although the entire plant was destroyed: five biodiesel tanks (containing approx. 30,000 gallons) and 6,000 gallons of methanol were saved; and, approx. 90,000 gallons of corn oil stored in railroad cars were moved back safely with only some minor fire damage to the exterior of the cars.	6
03/11/1997	ISRAEL	DIESEL	Fire; Explosion	0 Injured 1 Killed	A fire occurred in a diesel storage tank following explosion and caused a fatality. Worker had gone for a test sample when the explosion occurred	IChemE
11/01/1995	USA; TENNESSEE; CHATTANOOGA	VEGETABLE OIL	Fire	Not reported	Derailment of twenty cars of freight train. Box car containing half gallon containers of vegetable oil was first to catch fire.	MHIDAS
03/06/1988	Izmir, TURKEY	VEGETABLE OIL	Ship to shore impact	0 Injured 0 Killed	A marine transportation incident. A marine tanker hit berth at a terminal while manoeuvring in to discharge vegetable oil. Bow heavily damaged. Jetty destroyed.	IChemE
10/11/1979	NORWAY; SKREIA	VEG OIL + SODA	Runaway-reaction	0 Injured 0 Killed	Chemical reaction between vegetable oil and soda. Violent explosion destroyed 2-Storey building of area 25,000m^2.	MHIDAS
10/02/1970	FRANCE; MARSEILLES	VEGETABLE OIL	Fire; Explosion	12 Injured 2 Killed	Explosion in vegetable oil refinery of a soap factory. Refinery gutted by fire.	MHIDAS

1: http://www.biofuels-news.com/news/tennessee_explosion.html

2: http://wjz.com/local/bio.diesel.plant.2.726871.html

3: http://www.cbc.ca/canada/calgary/story/2008/04/15/biodiesel-explode.html

4: http://www.indianasnewscenter.com/news/local/13062367.html

5: http://www.idahobusiness.net/archive.htm/2006/07/17/What-Now-Explosion-at-New-Plymouth-biodiesel-plant-raises-questions-for-highlytouted-market

6: http://www.greaseworks.org/modules.php?op=modload&name=News&file=article&sid=274&mode=thread&order=0&thold=0

Table 8.5c Previous Incident Involving Methanol

Date of incident	Location of incident	Material Name	Incident Type	Injuries/ Fatalities	Description	Reference
20/02/2001	USA; ALASKA; PRUDHOE BAY	METHANOL	Continuous-release	0 Injured 0 Killed	Water frozen in oil pipeline creating an ice plug. Pipeline flushed with warm crude oil and methanol, and pressure raised to help melt the ice. The oil and methanal leaked from a crack on the top of the pipeline. Size of spill unclear.	MHIDAS
06/12/2000	USA; NEW MEXICO; JAL	METHANOL	Fire	Not reported	A natural gas pipeline ruptured and exploded underneath two storage tanks containing methanol and glycol. The tanks burned out and the plant was forced to close.	MHIDAS
05/11/2000	USA; TEXAS; SONORA	METHANOL	Fire	0 Injured 0 Killed	Lightning struck a transformer, igniting a chemical warehouse. All residents within half a mile were evacuated. Fire was extinguished within three hours. Warehouse contained methanol, cleaning solvents and other hazardous chemicals.	MHIDAS
04/09/2000	UK; ESSEX; PURFLEET	METHANOL	Continuous-release	0 Injured 0 Killed	19,000 kg of methanol spilled out onto the dock side after a container loaded with the solvent toppled over. The area was cordoned off by police to allow specialists to remove the chemical. The wharf was cleared of all shipping as a precaution.	MHIDAS
14/03/2000	UK; NORFOLK; NORWICH	METHANOL	Release	0 Injured 0 Killed	Workers were evacuation after a single container of chemical leaked in a warehouse. It is unclear whether the chemical was sulphuric acid or methanol. Fire fighters cleaned up and made safe the leaking container.	MHIDAS
19/08/1999	UK; WEST MIDLANDS; WEST BROMWICH	METHANOL	Gas-cloud	0 Injured 2 Killed	2 night watch men found dead when day shift arrived. They were overcome by fumes believed to be form mixture of methanol, potassium powder & hydrochloric acid, used to strip wheels. High levels of methylene chloride traced in building.	MHIDAS
30/03/1998	UK; CHESHIRE; ELLESMERE PORT	METHANOL	Fireball	Not reported	Tanker carrying 20 tonnes methanol overturned on a roundabout and caught fire. Methanol burns invisibly and fireball was recorded on a thermal imaging camera in a police aircraft which helped direct evacuation and firefighting. Motorway closed for 6 hours.	MHIDAS

Table 8.5d Previous Incident Involving Diesel Tankers

Date of incident	Location of incident	Material Name	Incident Type	Injuries/ Fatalities	Description	Reference
29/10/1998	UK; MID GLAMORGAN; PENCOED	DIESEL FUEL	Continuous-release; Liquid pool formed	0 Injured 0 Killed	A Diesel tanker and a gas tanker collided on the M4 spilling hundreds of gallons of diesel. The motorway was blocked for more than 6 hours. Fire fighters undertook a huge mopping up operation. Both drivers were uninjured.	MHIDAS
09/07/1998	UK; DEVON; CULLOMPTON	DIESEL FUEL	Continuous-release	0 Injured 0 Killed	A Diesel tanker and a BMW collided on the M5 between junctions 28 and 29. No one was injured but the collision and fire crews were called in to deal with the spill. Sand was used to soak up fuel.	MHIDAS
19/02/1996	UK; CLEVELAND; THORNABY	DIESEL	Continuous-release	3 Injured 0 Killed	Tanker with nitrobenzene crossed central reservation & crashed head on with a diesel tanker. The diesel tanker overturned & one compartment of the nitrobenzene tanker ruptured allowing chemical to spill to roadside drain & subsequently to river tees.	MHIDAS
08/09/1995	UK; LINCOLNSHIRE; NORTH THORESBY	GAS OIL	Continuous-release	Not reported	Lorry and diesel tanker collided head on when one of the vehicles skidded in wet and locked brakes. Collision forces part of articulation of tanker back into tank, which then buckled. 400L gas oil from fully laden tanker leaked onto road.	MHIDAS
12/10/1990	UK; ESSEX; BISHOPS STORTFORD	DIESEL	Continuous-release	Not reported	Diesel tanker toppled onto oncoming car spilling 3000galls	MHIDAS

Table 8.5e Catastrophic Tank Failure (100% Instantaneous Rupture) of Petroleum Product Tanks Since 1970

Date of incident	Location of incident	Material Name	Failure Cause	Spill Contained by Bund	Reference
29/07/1993	El Segundo, CA, USA	Fuel oil	Not known	No - about 2% lost	MHIDAS
11/05/1993	Fawley, UK	Bunker oil	Mechanical	Unknown if bund used	MHIDAS
10/1989	Richmond, CA, USA	Gasoline	Earthquake	Yes	MHIDAS
6/02/1989	New Haven, CT, USA	Heating oil	Mechanical	Yes	MHIDAS
11/07/1988	Brisbane, Australia	Gasoline	Corrosion	Not known	MHIDAS
2/01/1988	Floreffe, PA, USA	Diesel oil	Mechanical	No - 40 to 71% lost	MHIDAS
28/12/1980	El Dorado, KS, USA	Solvents	Mechanical	Not known	MHIDAS

Relevance of Serious Incidents at Other Biodiesel Plants

15/08/2008 – Decaturville, Tennessee, USA. This incident relates to an explosion within a biodiesel plant building which was idled. The cause of this incident appears to be unknown, however based on the incident reports, the following may be deduced. The plant covered an area less than 800 m² and producing 35,000 T/year, corresponding to a backyard facility typical to entrepreneur start-up plants in US. In such plants, safety is most likely to be compromised as entrepreneurs are trying to get into the industry at minimal costs often compromising on safety ([9]). Analysing accident reports further, it is noted that an explosion occurred, followed by the building fire. An explosion would only have been possible following a loss of containment of flammable substances. The substance that caused the explosion is not known. It could have been methanol or any other flammable substance stored at the facility. Since the plant had been shut down for several months, the initiating event was likely caused by inadequate procedures for operation (inventory not empty/ inappropriate purging, etc) and maintenance of the facility (e.g. no check for leaks, etc) which would not occur if the facility is under regular checks and monitoring. The safety features are not specified in accident records, but based on typical design of such facilities, if fire mitigation systems were in place and inert gas blanketing was applied the explosion would not have occurred. Therefore, if the operator followed standard operation and maintenance procedures for this facility this accident would not have occurred.

18/05/2008 – Princess Anne, Maryland, USA. This incident relates to an explosion within a biodiesel plant building which was idled. The explosion was triggered by welding work being undertaken by external contractors. In addition, plant modification was carried out without positive isolation. The incident does not, therefore, relate to normal plant operation, and can only arise from lapses in proper maintenance procedures and permit-to-work controls. Therefore, if the operator followed standard operation and maintenance procedures for this facility this accident would not have occurred. Safety management systems will be in place in the proposed plant to minimize the risks of any similar incidents.

15/04/2008 – Calgary, Canada. This incident relates to tank failure outside the process building. This, again, was triggered by welding by contractors. See comments on Princess Anne incident.

04/01/2008 – Defiance, Ohio, USA. This incident relates to an explosion within a biodiesel plant building. This incident appears to have been related to accumulation of methanol vapours inside a glycerine tank, and their subsequent release through an open manhole. The vapour was believed to have ignited by operation of a motorized door. The injured persons were reported to have been located in a plant office. Such an incident is less likely in the proposed plant because:

· methanol is distilled from the crude glycerine before it is sent for storage and in addition any accumulated methanol vapour in the building would be vented;

· the glycerine tank is in the open air, not inside a building, reducing the likelihood of vapour accumulation; and

· All electric motors in chemical-handling areas are designed to be flameproof (Ex-rated).

07/07/2006 – New Plymouth, Idaho, USA. This incident relates to tank failure outside the process building. The plant was under conversion from a non-chemical warehouse to a biodiesel facility. Again, this incident was related to hot work on a tank that had not been properly drained or purged. The tank contained a mixture of methanol and glycerine.[10]

It may be noted that four of the five incidents above occurred during shutdown periods and related to hazards common to any type of facility handling flammable chemicals. Only one fatality (the Defiance, Ohio incident) is known to have arisen from a cause specific to biodiesel plants. Also, it is found that no fatality has been reported beyond the site boundary as the result of any incident at a biodiesel plant.

17/02/2006 – Bakersfield, California, USA. This incident relates to methanol spill during transfer outside the process building. The fire arose from a methanol spill during transfer from a movable bulk container. In the proposed plant, such a spill is unlikely because all methanol transfers (apart from transfer from delivery vehicles) will use fixed piping only, not hoses. Although leaks can occur from flanges, these are rare, especially in low pressure, clean service systems. Leaks can also occur from bleeders if left open and uncapped, but this is poor operational practice and should be controlled by a permit-to-work system.

It was reported that the spill was ‘small’ and yet the fire spread from outside the process building to inside. This suggests that the spill was actually quite substantial, and there may have been no physical barrier between outside and inside (cf. the concrete containment wall around the process building for the proposed plant).

Other Incidents

22/11/2006 – Danvers, Massachusetts, USA. A serious incident occurred in a paint and ink manufacturing facility which, although not directly related to biodiesel, is considered here because the causes and findings are highly relevant. In this incident, a major vapour cloud explosion occurred overnight inside a process building. Although the exact cause has not been established, it appears due be due to leakage of heptane and alcohol vapours from a vessel left heating unattended. The vapours were released due to overheating, and built up because the building ventilation had been turned off at night to prevent noise complaints. The entire plant was destroyed and several neighbouring buildings severely damaged.

The official investigation ([11]) found that the underlying causes were:

· inadequate ventilation in the process area

· absence of automated safeguards, controls or alarms

· no proper management or procedures in place for handling hazardous substances

Each of these issues is addressed in the proposed plant. Adequate full-time ventilation will be provided, with automatic emergency ventilation in the event of a buildup of flammable vapours; automated safeguards, controls and alarms will be provided to control process-related risks; and adequate procedures and management controls will be instituted for all hazardous operations onsite.

28/08/92 – Castle Peak Power Plant, Hong Kong. This was one of the most significant incidents to occur in Hong Kong. An explosion occurred inside a high pressure hydrogen receiver, resulting in damage to the power plant, roof, storeroom and concrete blast walls. Two people were killed and 19 injured. There was however, no injury or fatality involving the public. The incident arose from inadvertent injection of air over a period of 20 hours prior to changeover from a temporary receiver to the permanent one following maintenance.

A design error in the plant allowed the injection of air into the receiver at high pressure. To eliminate such a possibility at the proposed biodiesel plant, a full HAZOP study will be carried out on the proposed plant during the detail design in order to confirm the adequacy of design safety measures.

This incident again highlights the importance of proper maintenance, start-up and shutdown procedures, since a large fraction of incidents occur during these times (see discussion below).

Buncefield, UK, 10 December 2005

One of the tanks at this storage tank farm containing petrol overflowed while being filled from a pipeline. The liquid overflow led to a rapid formation of a vapour cloud which extended offsite. Upon ignition, this resulted in an explosion that led to a fire engulfing over 20 tanks. There were no offsite fatalities since the event occurred in the early morning hours. The explosion caused severe damage to adjoining buildings.

This accident is relevant to some extent to this study since methanol has a low flash point and hence there is potential for a vapour cloud formation upon spillage which can potentially lead to an explosion upon ignition. However, the amount of methanol stored in this plant is less than 500m³ which is very low in comparison with large petroleum storage tank farms.

Relevance of the Six Biodiesel Incidents

In order to estimate the frequency of explosions and fire at the proposed plant based on historical incident data, the relevance of previous incidents at biodiesel plants has been assessed, in terms of the scope for any incident of this type at the proposed plant compared with the affected plants in the USA. These are assigned on the basis that a physical safety system (e.g. inerting), proper maintenance and safety management procedures, and intrinsically safe design (e.g. hard-piping for flammable liquid) will each reduce the likelihood by an order of magnitude. Draining of tanks prior to maintenance is assigned a lower reduction factor of 0.2 instead of 0.1 for other cases because, under exceptional circumstances, it may not be possible. It is also to be noted that these incidents did not cause any offsite fatality and hence the likelihood of an incident that could lead to an offsite fatality may be assumed as 10%. This additional factor has, however, not been considered since the reduction factors assumed for better systems and practices have already accounted for the lower potential for a serious incident.

Table 8.5f Relevance factors for related incidents

Incident	Probable cause and underlying factors	Relationship to proposed plant at TKO	Relevance factor assigned
Incidents relating to explosion within the process building
Decaturville	No emptying of vessels not in operation and inadequate purging Inadequate maintenance of equipment not in operation	Standard operating procedure to empty and purge vessels on shutdown. Safety systems such as gas leak detectors and maintenance checks/rounds.	0.01
Princess Anne*	Damage to gas pipe during hot work No positive isolation before maintenance	Hot work strictly controlled by management system. Positive isolation required in hazardous areas	0.02
Defiance	Accumulation of methanol vapours indoors. Failure to close manway on tank. Non-flameproof electrical equipment in process area	Gas detection and automatic emergency ventilation provided. Flameproof electrical equipment. No storage tanks in indoor areas.	0.001
TOTAL			0.031
Incidents relating to loss of containment events outside process building
Calgary*	Failure to inert, purge and/or empty tank for hot work during shutdown	Hot work strictly controlled by management system. Draining and purging of tank required	0.02
New Plymouth*	Failure to inert, purge and/or empty tank for hot work during shutdown	Hot work strictly controlled by management system. Draining and purging of tank required	0.02
Bakersfield	Fire following methanol spill	Any methanol transfer is a supervised/attended operation with emergency fire fighting systems ready to operate. Methanol container will be iso-container and not subject to catastrophic failure	0.01
TOTAL			0.05

Incidents resulting in an onsite fatality are marked *

Estimating average Potential Loss of Life due to Biodiesel Related Incidents

From the evaluation above, we may estimate the frequency of an incident at the proposed plant as follows. There have been 3 fatalities in the US biodiesel industry (all onsite) in around 600 plant years, with a total relevance factor of 0.06, giving an average PLL (for onsite population) of 1 x 10^-4 per biodiesel plant per year. Based on the review of historical accidents, even if those incidents did not cause any offsite fatalities, it is judged that only one reported vapour cloud explosion incident (Princess Anne incident with relevance factor of 0.02) within the process building may have the potential to cause offsite fatalities, which would lead to a PLL (for offsite population) of 3.3 x 10^-5 per biodiesel plant per year.

Estimating Frequency of Explosions from Related Incidents

Based on the vapour cloud explosion incidents within the process buildings, only one (Princess Anne) was reported to damage the process building directly. Although the building was damaged, the damage was not comparable to explosions which would cause fatalities offsite. Additionally, this facility may not have been provided with emergency ventilation system (typical configuration in US is without emergency ventilation).

It should be noted that there have been no reported serious incidents at biodiesel plants in Europe. The proposed plant will use technology licensed from a European company, already successfully implemented in around 20 other plants across Europe, and will be operated to European safety standards. Thus, the fatality frequency estimated from American incidents can be regarded as a conservative upper limit.

Conservatively accounting that all of the three of the historical biodiesel accidents would be relevant, a total relevance factor of 0.031 may be derived (referring to Table 8.5f). The overall frequency of explosions is therefore estimated at 0.031 x 1/600y, ie 5.2 x 10^-5 per year. This frequency can therefore be considered to represent the overall explosion frequency including minor and major explosions (although no major explosion was reported with offsite fatalities). This may be compared with the derived frequency for explosion in this plant, which is described in later sections.

Hazards during Maintenance and Shutdown

The discussion above indicates that maintenance activities, especially during shutdown periods, have contributed to a substantial proportion of serious previous incidents in biodiesel plants. This is in line with data from the chemical industry generally: for example, a typical refinery operation is shut down for only 2% of the time, but 20% of accidents occur during this period ([12]). In the case of the biodiesel industry, there are indications that, in the USA, a substantial fraction of plants are idle at any given time ([13]). This would explain the preponderance of downtime-related incidents.

In a review of around 500 previous incidents in the oil and chemical industries, Duguid ([14]) notes that about half of the incidents occurred during shutdown (15%), start-up (14%), maintenance (11%) and abnormal operations (13%). Abnormal operations include mainly actions to avoid shutdowns due to faulty equipment.

Also, he reports that around 22% of incidents are related to tank usage. He comments that “one can get the impression that because storage is an ancillary service it may not get the same attention in safety matters as process plant.” This is certainly reflected in the pattern of previous incidents reviewed above, and emphasizes the importance of proper attention to safety management with regard to tank farms.

The scenario of a similar maintenance-related incident occurring at the proposed plant, is discussed in a later section.

8.5.3 Hazards Associated with Marine Transport

Hazards associated with the final approach and manoeuvring of barges within 100m of the jetty are included in this assessment. These hazards include leaks and fires arising from collisions and groundings.

Since methanol is completely miscible with water, this rules out the possibility of a pool fire occurring on the sea surface due to a methanol leak. Small to medium leaks would have little effect since the methanol would simply pour into the sea and disperse. If a large rupture were to ignite, the remaining contents within the tank may ignite leading to a tank fire. This scenario was included in the assessment.

Pool fires on the sea surface for other flammable/combustible materials are included in the analysis.

8.5.4 Hazards Associated with the Tank Farm

Brittle Failure and Material/Weld Defects

The main hazard associated with a storage tank is a loss of containment leading to release of a hazardous substance.

There are a number of old recorded incidents involving a tank suddenly splitting from top to bottom releasing the whole inventory, also known as unzipping. These cases involved a crack propagated suddenly from an initial defect when the tank material was brittle at the prevailing temperature.

Design standards and knowledge of materials have improved since these early recorded incidents. One of the improvements includes welding the plates together in a staggered array such that a single crack cannot propagate vertically over an extended length. Cracks propagating horizontally are less likely since the stresses in this direction are lower.

Brittle failure involves very low ambient temperature, which is not applicable to the proposed biodiesel plant in Hong Kong. All storage tanks will store materials at or near ambient temperature, hence the steel of the tank will behave in a ductile way, in which cracks will arrest or propagate slowly ([15]).

Defects in welds or materials can also lead to tank failures.

Tank failure scenarios are considered in this study, modelled using generic failure frequencies (Section 8.6). It is assumed that the biodiesel storage tanks will be designed and constructed to the appropriate standards so that generic failure frequencies are appropriate.

Internal Explosion

A number of past incidents have occurred involving an explosion in the vapour space of a storage tank. This can result in the failure of the shell to floor seam and instantaneous release of the tank inventory.

Modern tanks are constructed with a weak shell to roof connection which is designed to fail first during an internal overpressure scenario. In this case, the content of the tank will be retained and the worst consequence would be a fire on top of the tank.

Internal explosion is possible only if the vapour generated is able to form a flammable mixture. In the tank farm of the proposed biodiesel plant, all materials handled have insufficient vapour pressure at ambient conditions to form a flammable vapour cloud. The exception is the methanol storage tank. Internal explosion of the methanol tank leading to tank fire was therefore considered further but the consequence modelling demonstrated that the radiation effects at ground level would be negligible. The methanol tank is also a cone roof type with nitrogen blanketing.

Bund Overtopping

Although all the storage tanks are located in bunds, impoundment bunds may not be 100% effective in containing spills, especially large spills from catastrophic failure of a tank. A fraction of the tank contents may overtop the bund wall. This is considered in the analysis.

Tank Fires

Tank fires may occur due to ignition of vents from the tank or due to roof failure or ignition of the vapour space in the tank. This may lead to a fire on tank top. This scenario is considered more likely for the methanol tank (there is only one tank with an inventory of less than 500m³) which has a low flash point. All other tanks hold material with a very high flash point and hence ignition leading to a fire is very low. In any case, the radiation effects from a tank fire is limited to the plant boundary and hence the risk of fatality offsite due to a tank fire is negligible.

8.5.5 Process Related Hazards

Biogas Hazards

The biogas generated from the IC Reactor (average flow about 80 m³ hr^-1) will be temporarily stored in the biogas buffer tank of 30 m³ capacity, under the working pressure of 2.5 to 3.8 kPa and a maximum pressure of up to 5.5 kPa (0.055 barg). A preliminary quantitative assessment using the PHAST model of the potential consequences of a biogas leak or loss of containment demonstrated that due to the low working/maximum pressures and low gas inventory, the potential consequence distances would be small (less than 10 m) and would not affect the off-site population. Nevertheless, for completeness, this hazard is included in the assessment.

Hazards from Loading/unloading Areas

Hazards may also come from the loading and unloading of materials from/onto road tankers and sea barges. Failure of the unloading hoses and piping during the transport of methanol and other flammable materials from barges and road tankers is considered in this study to investigate the effect of a leakage assuming typical unloading rates.

A spontaneous failure of a road tanker may result in a pool fire and has also been considered.

Methanol road tankers would be used only sporadically, when the barge transport is not possible (e.g. during typhoons). The frequency of their failure and resulting pool or flash fires, adjusted by the presence factor, are below 1 x 10^-9 per year, a frequency threshold below which hazardous events are normally not considered in QRAs. Specifically, a road tanker rupture frequency of 2(10^-6 per tanker-year is used. With 2 deliveries (each lasting 1 hour) per day for 10 days during typhoons, the presence factor is (2/24) x (10/365) = 0.0023. With an ignition probability of 0.08 (Section 8.6), the frequency of a fire becomes 3.6(10^-10 per year. The leak frequency is a little higher at 2(10^-6 per tanker-year, but the ignition probability is lower and the fire frequency still below 10^-9 per year. Thus, the risk due to failures of methanol road tankers is considered to be negligible.

Process Building

There are a number of vessels and equipment in the process building that can fail, the main hazard being a release of flammable materials. A release of most of the materials would simply lead to a pool fire. The process area is fully enclosed in a building with a curbed area to contain spills. A fire within the process building would not affect anyone offsite. Even if the fire protection systems were to fail and a pool fire escalated to affect the building itself, the escalation would be gradual. The normal response time[16] of the Hong Kong fire services (FSD) to an emergency call is likely to be around 10 minutes for this location, and so there would be sufficient time for persons offsite to evacuate and avoid danger. Pool fires within the process building are therefore considered to pose negligible risk.

Some equipment items, such as the high pressure reactor, have sufficient pressure to produce a jet fire in case of release. However, similar arguments apply and the jet fire will be contained within the building and will not impact anyone offsite. Jet fires are therefore considered to pose negligible risk.

Releases of methanol, however, pose greater risks. Methanol has a higher vapour pressure and may lead to the accumulation of flammable vapours, particularly when released at elevated temperatures such as from the esterification reactor. Upon ignition, a vapour cloud explosion could occur within the process building due to the high level of confinement and congestion. This scenario is considered in the analysis.

The effects of methanol being released through the ventilation exhaust vent on the roof are also considered.

8.5.6 External Hazards

The concern here is with those hazards that are outside the control of the operating personnel but could still pose a threat to the installation. Such hazards are termed as ‘external hazards’ because they are independent of the operations on-site but can lead to major hazard scenarios. These external hazard events include the following:

· Earthquake;

· Aircraft crash;

· Helicopter crash;

· Typhoons (Storm Surge and Flooding);

· Tsunami;

· Landslide;

· Subsidence;

· Vandalism, sabotage, terrorist attack, acts of war;

· Lightning strike;

· Accidents at neighbouring facilities; and

· Vehicle impact.

Since these events are independent of the operations protection against them needs to be incorporated into the system’s design. Design criteria for the protection of hazardous chemical plants against such events are well established, and are generally dependent on the geographical location of the plant. Additional criteria could also be imposed (for instance development of hazardous chemical plants is subject to stricter planning regulations in areas of high seismic activity).

However, if a major accident occurs despite the fulfilment of the relevant design standards and regulatory criteria, the consequences of this event could be catastrophic. In examining the consequences of a major accident from this source, the examination should be made in the context of the overall situation. For example, if an earthquake, winds or floods of sufficient intensity occurs at a chemical plant complying with relevant design and regulatory criteria, what will be the direct consequences of that occurrence (fatalities due to building collapse, induced landslides, etc.) are typically greater than the indirect consequences as a result of failure of the plant. Each of these hazards is further discussed in the following subsections.

Earthquake

An earthquake has the potential to cause damage to storage tanks, pipework and process vessels. Damage to pipework could be due to ground movement/vibration, with guillotine failure of pipes possibly resulting. Loss of containment from a process vessel could result due to:

· Vessel structural support failure;

· Building roof/structure collapse; and

· Vibration induced stress failure.

Vibration and ground movement can also lead to the failure of storage tanks.

Studies by the Geotechnical Engineering Office([17]) and Civil Engineering Services Department ([18]) conducted in the last decades indicate that Hong Kong SAR is a region of low seismicity. The seismicity in Hong Kong is considered similar to that of areas of Central Europe and the Eastern areas of the USA. As Hong Kong is a region of low seismicity, an earthquake is an unlikely event. The generic failure frequencies adopted in this study are based on historical incidents that include earthquakes in their cause of failure. Since Hong Kong is not at disproportionate risk from earthquakes compared to similar facilities worldwide, it is deemed appropriate to use these generic frequencies without adjustment. There is no need to address earthquakes separately as they are already included in the generic failure rates.

Aircraft Crash

The proposed biodiesel facility is more than 30km away from the Hong Kong’s international airport (see Figure 8.5a). The frequency of aircraft crash was estimated using the methodology of the HSE ([19]). The model takes into account specific factors such as the target area of the proposed hazard site and its longitudinal (x) and perpendicular (y) distances from the runway threshold (see Figure 8.5b).

The crash frequency per unit ground area (per km²) is calculated as:

(1)

where N is the number of runway movements per year and R is the probability of an accident per movement (landing or take-off). F(x,y) gives the spatial distribution of crashes and is given by:

Figure 8.5a Flight Paths at Hong Kong International Airport

Landings

(2)

for km

Take-off

(3)

for km

Equations 2 and 3 are valid only for the specified range of x values. If x lies outside this range, the impact probability is zero.

Figure 8.5b Aircraft Crash Coordinate System

NTSB data ([20]) for fatal accidents in the U.S. involving scheduled airline flights during the period 1986-2005 are given in Table 8.5h. The 10-year moving average suggests a downward trend with recent years showing a rate of about 2(10^-7 per flight. However, only 13.5% of accidents are associated with the approach to landing, 15.8% are associated with take-off and 4.2% are related to the climb phase of the flight ([21]). The accident frequency for the approach to landings hence becomes 2.7(10^-8 per flight and for take-off/climb 4.0(10^-8 per flight. The number of flights at Chek Lap Kok in 2007 was about 300,000 ([22]).

Considering landings on runway 25R for example, the values for x and y according to Figure 8.5b are 32km and 14km respectively. Applying Equation 2 gives km^-2. Substituting this into Equation 1 gives:

The number of plane movements has been divided by 8 to take into account that half of movements are take-offs and only a quarter of landings use runway 25R. This effectively assumes that each runway is used equally and the wind blows in each direction with equal probability.

Table 8.5h U.S Scheduled Airline Accident Rate

The target area is estimated at 20,000m² or 0.02km². This gives a frequency for crashes into the biodiesel plant associated with landings on runway 25R as much less than 10^-9 per year.

The combined frequency of all take-off and landing crashes onto the biodiesel plant from activities on all runways is much less than 10^-9 per year. The risk of aircraft crash is therefore negligible.

Helicopter Crash

A helicopter landing pad is situation to the north of the site, in the unoccupied lot labelled as ‘B’ in Figure 8.3a. The landing pad is used rarely, however, with a frequency of less than once per year. The distance to nearest storage tank is about 130m.

The approach, landing and take-off stages of a flight are associated with the highest risk of helicopter crashes. Historical incidents show that helicopter accidents during take-off and landings are confined to a small area around the helipad ([23]). 93% of accidents occur within 100m of the helipad. The remaining 7% occur between 100 and 200m of the helipad.

Data from offshore helicopter activities ([24]) gives a helipad related helicopter crash frequency of 2.9(10^-6 per flight stage (i.e. per take-off and landing). However, most of these incidents are minor such as heavy landings. For a helicopter incident to damage a facility, it must be a serious, uncontrolled impact. Only accidents involving fatalities were therefore considered in the analysis. 4% of incidents resulted in one or fatalities and so the frequency of uncontrolled crashes was estimated at 2.9(10^-6(0.04 = 1.2(10^-7 per flight stage. For one flight per year using the helipad, the frequency becomes 1.2(10^-7 per year.

The storage tanks are approximately 130m from the helipad. Only 7% of accidents occur within the range of 100 to 200m. The probability of a helicopter crashing into a storage tank may be calculated as:

per year

where a typical storage tank has been estimated to be 6m in radius. The frequency of damage by helicopters is very small compared to the process related failures. For example, the generic failure frequency adopted for catastrophic failure of a tank is 5(10^-6 per year (Section 8.6). The contribution from helicopters to the frequency of tank or equipment failure may therefore be neglected with negligible error.

Typhoons (Storm Surge and Flooding)

Flotation of piping and tanks is possible if equipment becomes submerged in water.

Flooding from heavy rainfall is not possible due to the coastal location of the site. The primary hazard from typhoons is the storm surge and waves, which if combined with a high tide could lead to flooding of the site. Winds, and to a lesser extent pressure, cause a rise in sea level in coastal areas. In general, storm surges are limited to several meters.

The foundation of the storage tanks varies from 4.4m to -0.1m above sea level, however all are surrounded by a bund wall which is 6.25m above sea level. The process buildings are 4.9 meters above sea level and are therefore protected against any risk from storm surge, wave and other causes of flooding.

Tsunami

Similar to storm surges, the main hazard from tsunamis is the rise in sea level and possible floatation of piping and tanks. The highest rise in sea level ever recorded in Hong Kong due to a tsunami was 0.3m high, and occurred as a result of the 1960 earthquake in Chile, the largest earthquake ever recorded in history at magnitude 9.5 on the Richter scale. The effect of a tsunami on the biodiesel plant is therefore considered negligible.

The reason for the low impact of tsunamis on Hong Kong may be explained by the extended continental shelf in the South China Sea which effectively dissipates the energy of a tsunami, and also the presence of the Philippine Islands and Taiwan which act as an effective barrier against seismic activity in the Pacific ([25]). Secondary waves that pass through the Luzon Strait diffract and lose energy as they traverse the South China Sea.

Seismic activity within the South China Sea area may also produce tsunamis. Earthquakes on the west coast of Luzon in the Philippines have produced localised tsunamis but there is no record of any observable effects in Hong Kong.

Damage from tsunamis is therefore considered to pose negligible risk.

Subsidence

For subsidence to result in failure of pipework or vessels, the ground movement must be relatively sudden and severe. Normal subsidence events will occur gradually over a period of months and thus appropriate mitigating action can be taken to prevent failures. In the worst cases, the plant would be shut down and the relevant equipment isolated and emptied.

The biodiesel plant will be built on a reclaimed land but the reclamation was completed more than 10 years ago, so any settling will have already taken place. Also, appropriate foundation work will be undertaken for the site. No undue risk from subsidence is therefore expected and failures due to this are deemed to be included in generic failure frequencies.

Intentional Acts

The risks from intentional acts such as sabotage, vandalism and terrorist attacks are difficult to quantify. Sabotage and arson together are believed to be involved in around 1% of all major incidents at chemical facilities, where the cause is known ([26]). Terrorist attacks on chemical sites are, however, extremely rare in developed countries ([27]). Hong Kong is an area where such activities are very rare. At the proposed site, physical security will meet normal standards for a facility of this type, including access control via manned gate, a physical wall or gate surrounding the entire site, and video surveillance. Moreover, the site is unlikely to be of interest to terrorists or outside saboteurs because (1) it is not located close to a major population centre, (2) the maximum possible scale of any intentional incident is rather small, and (3) the nature of the business is environmentally positive and so it is unlikely to be targeted by eco-activists. Based on these considerations, intentional acts are considered to pose negligible risk.

Lightning strikes

Lightning strikes have led to a number of major accidents world-wide. For example, a contributory cause towards the major fire at the Texaco refinery in the UK in 1994 was thought to be an initial lightning strike on process pipework.

The installation will be protected with lighting conductors to safety earth direct lightning strikes. The grounding will be inspected regularly. The potential for a lighting strike to hit the facility and cause a release event is therefore deemed to be unlikely. Failures due to lightning strikes are taken to be covered by generic failure frequencies.

Risks from Neighbouring Facilities

Hong Kong Oxygen (HKO) has a facility about 400m to the north of the proposed biodiesel plant. HKO provides compressed gases such as acetylene, hydrogen, argon, nitrogen, oxygen, compressed air etc. These are for welding applications as well as medical and special uses. Since these gases are contained in compressed gas cylinders, a fire at the site could lead to explosions with fragments travelling sufficient distance to strike storage tanks and equipment within the biodiesel plant. The main concern is an impact on one of the storage tanks leading to a tank leak.

A past incident in 1992 at the Castle Peak Power Plant involved a hydrogen explosion and produced a fragment that travelled 500m. This demonstrates that it is feasible for incidents at HKO to impact on the biodiesel plant.

Lees ([28]) reviews past incidents involving BLEVE type explosions and notes that missile fragments are generated in about 80% of cases. Missile fragments may be produced by ejection of a single item such as the valve assembly, or by disintegration of the vessel. In reviewing a number of past incidents, Lees observes that the number of fragments produced per explosion can be up to about 30. Lees also assesses the distance travelled by fragments, from which it may be estimated that less than 10% will fall within the range of 400m to 500m.

Combining this information, the probability of a storage tank being hit by an explosion fragment may be estimated:

per explosion

where it has been assumed that the radius of a storage tank is about 6m (one of the larger tanks).

No information is available on the facilities and equipment at HKO, so the frequency of explosions was therefore estimated as follows. The frequency of a gas leak from a compressed gas cylinder is about 10^-5 per year([29]). Assuming a typical fire frequency of 1x10^-4 per year on a conservative basis, and assuming that every fire results in a single cylinder explosion, the number of fragments colliding with one of the storage tanks within the biodiesel plant would be:

per year

If a fragment were to strike a storage tank, it could potentially cause damage to the tank, and result in a leak that would be contained by the bund. The generic failure frequency being adopted in this study for leaks from storage tanks is 10^-4 per year (Section 8.6). The frequency of damage from explosion fragments from HKO is much less than this and therefore the risks from HKO may be considered to be included in the generic frequencies for tank failures.

Vehicle Impact

Only authorised vehicles will be allowed within the site and the speed limit will be strictly enforced. Thus, the possibility of an accident where a vehicle impacts equipment causing failure is considered to be remote. Also, since all the tanks are located in bunded areas and the process equipment is in enclosed process buildings, vehicle impact is considered as highly improbable. Vehicle impacts are therefore considered to pose negligible risk.

Conclusion on External Hazards

All external hazards either make site specific contributions to failure frequencies that have been quantified as negligible, or are shown to be already covered in the generic frequencies.

8.5.7 Potential Hazardous Outcomes

As seen from the incident review and discussions above, the following hazardous outcomes may result from leaks of materials such as biodiesel, gas oil or methanol:

Pool Fire

A pool fire occurs when a flammable liquid is spilt onto the ground and ignited. If the release source is protected by a bund, the maximum size of the pool fire is limited by the bund area; such case is sometimes termed a ‘bund fire’. The possibility of bund overtopping has also been considered for catastrophic tank failures.

Jet Fire

A jet fire may occur in case of immediate ignition of a pressurised release of flammable liquid or gas. In this biodiesel plant, the only item of equipment under significant pressure is the high pressure esterification reactor. A jet fire from this equipment will be contained by the process building and will not affect offsite population. Jet fires are therefore considered to pose negligible risk.

Flash Fire

Following a release of a gas, or liquid with sufficient volatility, a large vapour cloud may form around the release point. If this cloud is not ignited immediately, it will move with the wind and be diluted as a result of air entrainment.

The dispersing vapour cloud may subsequently come in contact with an ignition source and burn rapidly with a sudden flash. If the source of material which created the cloud is still present, then the fire will flash back to the source generating a pool fire. Direct contact with the burning vapours may cause fatalities but the short duration of the flash fire means that thermal radiation effects are not significant outside the cloud and thus no fatalities are expected outside of the flash fire envelope.

Vapour Cloud Explosion

If a flammable vapour is allowed to accumulate in a confined and/or congested area, and is subsequently ignited, an explosion (Vapour Cloud Explosion or VCE) may result. At the biodiesel plant, the methanol storage tank will be located outdoors in an open area. The impact of a release leading to a VCE onsite has been considered. The process building, is confined and heavily congested and so VCEs are considered in this area.

Toxic Release

(a) Ingestion of Methanol

Methanol is toxic by ingestion. In the event of a catastrophic release of methanol from the storage tank, a person nearby may be affected by the surge of methanol overtopping the bund, and involuntarily ingest some of the liquid. The fatal dose is around 100mL ([30]). However, symptoms take several hours to emerge, and during this time, antidotes can be administered. Also, tank failure modelling indicates that no methanol will be released offsite in a catastrophic release (see Table 8.7a). Thus, the overall risk from methanol ingestion is considered to be negligible.

(b) Inhalation of Methanol Vapour

If methanol is released without ignition, offsite population may be exposed to a toxic risk from the vapour, although the inhalation toxicity of methanol is relatively low. Dispersion of unignited methanol clouds is included in the assessment.

8.5.8 Formulation of Representative Hazardous Scenarios

Based on the above discussions, a list of accident scenarios is produced for further analysis (see Table 8.5i). The assumed leak sizes for tanks and process vessels are based on the largest instrument tapping or other nozzle that is potentially vulnerable to damage. Leak losses of 5% and 15% from barges correspond to the figures assumed in the PAFF report ([31]).

Table 8.5i Hazardous Scenarios Considered

Section No.	Section Code	Description	Equipment Tag	Hazardous Material	Physical State	Temp. ((C)	Press. (barg)	Inventory (kg)	Leak Size	Potential Outcomes
Tank farm
1	T1/T2	Crude grease trap waste storage tank	B100300/	Fats & oils	Liquid	Amb.	0	1,270,000	6” leak	Bund fire
			B100400						Catastrophic rupture	Pool fire with bund overtopping
2	T3	Cleaned grease trap waste storage tank	B100500	Fats & oils	Liquid	Amb.	0	863,000	6” leak	Bund fire
									Catastrophic rupture	Pool fire with bund overtopping
3	T4/T5	Crude waste cooking oil storage tank	B100600/	Fats & oils	Liquid	Amb.	0	41,100	6” leak	Bund fire
			B100700						Catastrophic rupture	Pool fire with bund overtopping
4	T6	Cleaned waste cooking oil storage tank	B100800	Fats & oils	Liquid	Amb.	0	863,100	6” leak	Bund fire
									Catastrophic rupture	Pool fire with bund overtopping
5	T7	PFAD storage tank	B100900	Fats & oils	Liquid	Amb.	0	1,270,000	6” leak	Bund fire
									Catastrophic rupture	Pool fire with bund overtopping
6	T8	Crude animal fats storage tank	B100100	Fats & oils	Liquid	Amb.	0	426,000	6” leak	Bund fire
									Catastrophic rupture	Pool fire with bund overtopping
7	T9	Cleaned animal fats storage tank	B100200	Fats & oils	Liquid	Amb.	0	426,000	6” leak	Bund fire
									Catastrophic rupture	Pool fire with bund overtopping
8	T10	Methanol storage tank	B102400	Methanol	Liquid	Amb.	0	309,000	6” leak	Bund fire, flash fire, VCE, toxic release
									Catastrophic rupture	Pool fire with bund overtopping, flash fire, toxic release, VCE, tank fire
9	T14	Infineum (additive) storage tank	B102200	Naphtha,	Liquid	Amb.	0	41,100	6” leak	Bund fire
				distillates					Catastrophic rupture	Pool fire with bund overtopping
10	T15/	Quality biodiesel storage tank	B200500/	Biodiesel	Liquid	Amb.	0	363,000	6” leak	Bund fire
	T16		B200600						Catastrophic rupture	Pool fire with bund overtopping
11	T17	Biodiesel-Europe storage tank	B200700	Biodiesel	Liquid	Amb.	0	1,797,000	6” leak	Bund fire
									Catastrophic rupture	Pool fire with bund overtopping
12	T18	Biodiesel-Hong Kong storage tank	B200800	Biodiesel	Liquid	Amb.	0	870,000	6” leak	Bund fire
									Catastrophic rupture	Pool fire with bund overtopping
13	T19	Glycerine storage tank	B202100	Glycerine,	Liquid	Amb.	0	510,000	6” leak	Bund fire
				FFA					Catastrophic rupture	Pool fire with bund overtopping
14	T21	Bioheating oil storage tank	B202200	Bioheating oil	Liquid	Amb.	0	150,000	6” leak	Bund fire
									Catastrophic rupture	Pool fire with bund overtopping
15	T22	Gas oil storage tank	B102300	Gas oil	Liquid	Amb.	0	74,000	6” leak	Bund fire
									Catastrophic rupture	Pool fire with bund overtopping
16	T24	Crude waste cooking oil	B101000	Fats & oils	Liquid	Amb.	0	1,022,000	6” leak	Bund fire
									Catastrophic rupture	Pool fire with bund overtopping
17	G01	Biogas buffer tank	V601	Biogas	Gas	Amb.	0	20	2” leak	Flash fire
									Catastrophic rupture	Flash fire
Process Area
18	P01	Esterification with catalyst	B10200/	11%wt methanol	Liquid	72	0.05	89,534	1” leak	VCE, toxic release
			B10400						Catastrophic rupture	VCE, toxic release
19	P02	Methanol buffer tank	B111100	95% methanol	Liquid	60	0.05	9060	1” leak	VCE, toxic release
									Catastrophic rupture	VCE, toxic release
20	P03	Pipe reactor	SP112000	40%wt methanol	Liquid	260	100	6600	1” leak	VCE, toxic release
									Catastrophic rupture	VCE, toxic release
21	P04	Methanol recycle tank	B72300	100%wt methanol	Liquid	60	0.05	16308	1” leak	VCE, toxic release
									Catastrophic rupture	VCE, toxic release
22	P05	MEK buffer tank	B20500	84%wt methanol	Liquid	50	0.05	12728	1” leak	VCE, toxic release
									Catastrophic rupture	VCE, toxic release
23	P06	GLP settling tank	B60300	51%wt methanol	Liquid	50	0.05	12949	1” leak	VCE, toxic release
									Catastrophic rupture	VCE, toxic release
24	P07	GLP collection tank	B50100	33%wt methanol	Liquid	65	0.05	103950	1” leak	VCE, toxic release
									Catastrophic rupture	VCE, toxic release
25	P08	Acidulation tank	B50400	38%wt methanol	Liquid	65	0.05	17160	1” leak	VCE, toxic release
									Catastrophic rupture	VCE, toxic release
26	P09	FFA buffer tank	B50900	5%wt methanol	Liquid	65	0.05	15456	1” leak	VCE, toxic release
									Catastrophic rupture	VCE, toxic release
27	P10	Neutralization tank	B60100	51%wt methanol	Liquid	65	0.05	21105	1” leak	VCE, toxic release
									Catastrophic rupture	VCE, toxic release
Marine scenarios
29	M01	Biodiesel/PFAD barge	-	Biodiesel/	Liquid	Amb.	0	1,000,000	5%, 15% leaks	Pool fire on sea surface
				PFAD					Catastrophic rupture	Pool fire on sea surface
30	M02	Methanol barge^[1]	-	Methanol	Liquid	Amb.	0	1,000,000	5%, 15% leaks	No effect
									Catastrophic rupture	Tank fire
31	M03	Methanol ISO-tanker^[1]	-	Methanol	Liquid	Amb.	0	16,000	1” leak	Pool fire, flash fire, toxic
									Catastrophic rupture	Pool fire, flash fire, toxic
Transport scenarios
32	L01	Unloading line from jetty to tanks	-	Biodiesel/	Liquid	Amb.	4	71,000	2” leak	Pool fire
				PFAD					Full bore rupture	Pool fire
33	L02	Unloading line from jetty to tanks		Methanol	Liquid	Amb.	4	65,000	2” leak	Pool fire, flash fire, toxic release
									Full bore rupture	Pool fire, flash fire, toxic release
34	RT1	Road tanker for fats & oils	-	GTW, WCO,	Liquid	Amb.	0	9,000	1” leak	Pool fire
				Gas Oil, animal fats,					Catastrophic rupture	Pool fire
				Glycerine
35	L03	Road tanker unloading line		Fats & oils,	Liquid	Amb.	4	4,300	1” leak	Pool fire
				biodiesel					Full bore rupture	Pool fire

[1] Methanol is anticipated to be delivered either by barges or by ISO-tanker barges. Both are therefore considered in the analysis.

8.6 Frequency Estimation

8.6.1 Release Frequencies

Frequency estimation involves estimating the likelihood of occurrence of each of the representative release events identified in Table 8.5i. The Consultants have collated a large in-house database of internationally recognised failure data for process plant and this was used for the study and adapted as necessary to take account of local factors. The frequency of each identified hazardous release event was estimated using this generic failure data. Event tree analysis is used where a release can lead to different outcomes such as pool fires or flash fires.

8.6.2 Ignition Probability

Equipment in the process areas will be fire and explosion rated to minimise sources of ignition. The highest temperature equipment in the process area is 260(C, below the auto-ignition temperature of all materials handled in the area. Therefore, there are expected to be few ignition sources within the process building. In outside areas, the main ignition source is likely to be trucks delivering feedstock.

This study adopts generic ignition probabilities adapted from Cox, Lees and Ang([32]) , which are summarised in Table 8.6a.The ignition probabilities for liquids are understood to apply to liquids above their flash point. These are therefore appropriate for methanol which has a flash point of 11(C.

Other materials have a higher flash point, significantly higher in most cases. A lower ignition probability is appropriate for these materials. A factor of 10 reduction is assumed, to give the ignition probabilities summarised in Table 8.6b.

Table 8.6a Ignition Probabilities from Cox, Lees and Ang Model

Leak Size		Ignition Probability
	Gas Release		Liquid Release
Minor (< 1 kg s^-1)	0.01		0.01
Major (1 – 50 kg s^-1)	0.07		0.03
Massive (> 50 kg s^-1)	0.30		0.08

The ignition probabilities are further split between immediate ignition and delayed ignition, in equal proportions. Immediate ignition will give rise to a pool fire, while delayed ignition of a volatile substance may cause a flash fire or vapour cloud explosion.

Table 8.6b Ignition Probabilities used in this Study

Leak Size		Ignition Probability
	Biogas		Methanol	Other flammable/combustible substances
Minor (< 1 kg s^-1)	0.01		0.01	0.001
Major (1 – 50 kg s^-1)	0.07		0.03	0.003
Massive (> 50 kg s^-1)	0.30		0.08	0.008

For ignition of releases within buildings, some adjustments have been made to the ignition probabilities above to take account of the mitigation provided by ventilation system.

8.6.3 Transport

Barges

1,000 tonnes barges will be used for transporting both feedstock and final products to and from the plant. The frequency for loss of containment is estimated based on the frequency of collisions and groundings.

The collision frequency adopted in PAFF ([33]) is 3.5×10^-5 per encounter. The encounter frequency is given as 0.69 per km. The same is applied to the current study. This effectively assumes that the marine traffic near the proposed biodiesel site is similar to that near the Permanent Aviation Fuel Facility, which is considered a conservative assumption. The interaction distance is 0.1 km as only marine transport within 100 m of the jetty is considered for this study. Therefore the collision frequency is given as 3.5×10^-5 × 0.69 × 0.1 = 2.4×10^-6 per visit. In addition, a spill probability of 0.015 is included to take into account that only 1.5% of all collisions lead to a loss of containment event. Therefore the frequency for loss of containment due to collision becomes 3.6×10^-8 per visit.

Similar calculations are applied to groundings. While the water depth is sufficient for a barge to dock at the jetty without grounding, the barge may collide with the sloping sea wall if it goes off course or loses control. The possibility of grounding is therefore included in the analysis. The adopted frequency for grounding in PAFF is 4.3(10^-6 per km travelled. Applying a distance of 100m for the approach, and a spill probability of 0.03, gives a base frequency for loss of containment due to grounding as 1.3(10^-8 per visit.

There are on average 4 shipments per week for the biodiesel plant, 2 for biodiesel, 1 for PFAD and 1 for methanol, all using 1,000 tonne barges. The maximum loading and unloading rate is specified as 500 m³/hr, however 12 hours is assumed to be the average time spent at the jetty for each barge.

The frequency data adopted in the PAFF study is for double hull tankers, while the barges used by the biodiesel plant will be single-walled liquid tankers. The frequency of collision for both types of tankers will be the same; however the probability of spillage following a collision for a single hull barge will be higher than that of the double hull tanker. A correction factor is needed to allow for the lower integrity of a single hull barge. In comparing data published in the “Purple Book”, the frequency for large leaks from a single hull tanker is 16.7 times higher than for a double hull. This factor of 16.7 was therefore applied to the data to give a failure frequency of 6.0(10^-7 per visit due to collisions and 2.2(10^-7 per visit due to groundings. The combined frequency becomes 8.2(10^-7 per visit.

Three different leak sizes are modelled as indicated in Table 8.6c.

Table 8.6c Barge Leak Size Distribution ([34])

Leak Size	Description	Probability
Small	5% of inventory released in 20 min	20%
Large	15% of inventory released in 20 min	20%
Rupture	100% of inventory released instantaneously	60%

An event tree for biodiesel/Palm Oil Fatty Acid Distillate (PFAD) barges is shown in Figure 8.6a. With 3 barges per week for these materials, the initiating frequency becomes 1.28(10^-4 per year. Since both immediate and delayed ignition result in a pool fire, no distinction is made between immediate and delayed ignition.

Figure 8.6a Event Tree for Biodiesel/PFAD Barge

Similarly, an event tree for a methanol barge is shown in Figure 8.6b. Since methanol is miscible with water, if there is a leak, methanol will simply pour into the water with no effect. A large rupture may lead to a tank fire on the barge if ignited and so this scenario was considered in the analysis.

Figure 8.6b Event Tree for Methanol Barge

ISO-tanker barges may be used rather than normal liquid tanker type barges. ISO-tankers would then be unloaded at the jetty using a reachstacker and loaded onto a flatbed truck for delivery to the tank farm. Since this delivery method for methanol would be used only occasionally, a factor of 1/52 (one week per year) is applied in the risk calculation. The ISO-tankers are considered similar to road tankers, which have a rupture frequency of 2(10^-6 per tanker-year and a leak frequency of 5(10^-6 per tanker-year ([35]), based on failure frequencies given in the Purple Book. These are increased by a factor of 10 to reflect the higher risks associated with the additional handling.

One barge per week is anticipated for delivering methanol to the 500m³ storage tank. Assuming that a complete filling is required, and the capacity of each ISO-tanker is 20m², 25 ISO-tankers per week would be required. It is further assumed that 1 hour of handling is required to unload an ISO-tanker. The leak frequency may then be calculated as 5(10^-6 ( 10 (handling factor) ( 25/168 (hours per week) ( 1/52 (one week per year) = 1.43(10^-7 per year, and rupture frequency as 5.71(10^-8 per year.

An event tree for ISO-tanker leaks is shown in Figure 8.6c. Immediate ignition, delayed ignition and unignited releases are considered. An ignition probability of 0.08 is used for methanol, based on figures for a large liquid release from Lees ([36]). This is distributed equally between immediate and delayed ignition to give 0.04 each.

Operation with either 1,000 tonne methanol barges or ISO-tanker barges will be used. Since these events are not expected to contribute significantly to the overall risks, both are included in the analysis. The risks are essentially double counted and hence the approach is conservative.

Figure 8.6c Event Tree for ISO-Tanker Failure

Road Tankers

A road tanker rupture frequency is assigned as 2(10^-6 per tanker-year and a 1” leak frequency of 5(10^-6 per tanker-year ([37]). This frequency is modified to take into account presence factors, assuming that each tanker will be present within the plant area one hour per visit.

The event tree for crude oil (GTW, WCO, gas oil, animal fats, glycerine) is shown in Figure 8.6d. With 72 road tankers per day, each spending 1 hour on site, this is equivalent to 3 tankers present on a continuous basis. The initiating event frequency then becomes (2(10^-6 + 5(10^-6) x 3 = 2.1(10^-5 per year. The relative leak size probability is (5(10^-6)/(2(10^-6 + 5(10^-6) = 0.714 for a 1” leak and (2(10^-6)/(2(10^-6 + 5(10^-6) = 0.286 for rupture.

Biodiesel shipping out by 20 m³ road tankers is anticipated during adverse weather when barges cannot be used. Biodiesel transport by road tanker is assumed to take place 10 days per year, with 10 tankers a day present at the premises for 1 hour each. With such a low presence factor and low ignition probability of 0.008, the frequency of biodiesel fires due to tanker leaks is below 10^-9 per year and is therefore considered to pose negligible risk. Similarly, methanol may be delivered by road tankers when marine operations are not possible but the frequency is again below 10^-9 per year.

Figure 8.6d Event Tree for Fats/ Oils Road Tanker

Unloading/Loading Lines and Hoses

Hoses, piping and connection errors are considered for the loading/unloading from road tankers and barges. Failure frequencies are obtained from:

· The failure frequencies of hoses, 9(10^-8 per hour, is taken from Blything ([38]);

· Hose disconnection error and failure to rectify is taken from LPG experience in Hong Kong with 1 incident in 31,718 operations per day and 5 years of data: 1.7(10^-8 per operation;

· Hose misconnection error and failure to rectify is based on experience in Hong Kong with LPG with 2 incidents in 31,718 operations per day and 5 years of data: 3.5(10^-8 per operation; and

· Piping failure frequencies are taken from the “Purple Book”, assuming a length of 150m from the jetty to the tank farm. Piping lengths from road tanker bays to storage tanks are estimated from plot layouts.

Hoses and piping are combined into a single scenario, with frequencies summed, since the consequences of leaks are essentially identical. 90% of leaks are assumed to be small leaks, modelled as 2”. The remaining 10% of leaks are assumed to be full bore ruptures. Hose failure frequencies are quoted on a per hour basis. These are multiplied by the hours of operation for each type of barge and road tanker. Piping frequencies are per year, and hence are also corrected to take into account the fractional period of time in operation. It is assumed that unloading from barges is performed at a rate of about 100m³/hr so that unloading to the 500 m³ methanol storage tank would require 5 hours, while loading/unloading of PFAD and biodiesel barges would take 10 hours. Road tankers are assumed to be on site for about 1 hour, however, actual pumping time is estimated at 20 minutes. Thus, operating fractions are calculated as (total transfer time per year) / (number of hours per year). For example, the methanol barge operating fraction is obtained by (52 operations/year x 5 h/operation) / (24 x 365) = 0.030.

A summary of the resulting leak frequencies is shown in Table 8.6d.

Table 8.6d Failure Frequency for Unloading Operations

	Pipe Length (m)	Operating fraction	Small Leak Freq. (yr^-1)	Rupture Freq. (yr^-1)
Jetty Operations
Methanol	150	0.030	2.57(10^-5	3.06(10^-6
PFAD	150	0.059	4.90(10^-5	5.84(10^-6
Biodiesel	150	0.12	9.80(10^-5	1.17(10^-5
Road Tanker Operations
Gas oil	50	0.014	2.83(10^-5	3.20(10^-6
Trap grease, crude	75	0.83	1.74(10^-3	1.98(10^-4
Animal fat, crude	40	0.056	1.12(10^-4	1.26(10^-5
WCO, crude	40	0.069	1.40(10^-4	1.58(10^-5
Bioheating oil	50	0.014	2.83(10^-5	3.20(10^-6
Glycerine	75	0.028	5.80(10^-5	6.61(10^-6

An example event tree for fats and oils is shown in Figure 8.6e. This is similar to previous event trees except that unloading hoses and line can be isolated and operations are manned so it is assumed that 90% of leaks can be isolated quickly before any hazardous outcomes occur. The event tree for methanol unloading (see Figure 8.6f) is similar.

Figure 8.6e Generic Event Tree for Fats & Oils Unloading Hose/Lines

Figure 8.6f Event Tree for Methanol Unloading Hose/Line

8.6.4 Tank Farm

Two release scenarios are considered for each tank: a leak through a 6” diameter hole and a catastrophic rupture (releasing 100% of the tank contents instantaneously). A 6” release is chosen to correspond to the size of tank connections. All leaks are taken to be confined to the bund and so this choice of 6”, although larger than normally considered for leaks, is not expected to have any impact on the results.

Bund Overtopping

An atmospheric storage tank is assigned a catastrophic rupture frequency of 5(10^-6 per year and a leak frequency of 1x10^-4 per year ([39]). The possibility of bund overtopping is considered for tank rupture scenarios. The proportion of tank contents that overtop the bund depends on the geometry, specifically the hydrostatic head of liquid, the height of the bund wall and the distance of the bund from the tank. For the biodiesel site, the likelihood of bund overtopping is therefore estimated based on the orientation of a release in relation to the nearest bund wall. Directions are considered in quadrants (see Figure 8.6g). The probability of a release being directed towards the nearest bund wall is taken to be 0.25, while the probability of a leak being directed towards the inner regions of the impoundment area is taken as 0.75. For tanks located in the corner of the bund area, the probability of overtopping is taken as 0.5. The methanol tank has bund walls in close proximity to the tank on 3 sides; the probability of overtopping is therefore taken as 0.75 for this tank.

Figure 8.6g Bund Overtopping Considerations

A generic event tree for feedstock, biodiesel etc. is shown in Figure 8.6h, and an event tree for the methanol tank is shown in Figure 8.6i.

Bund Overflowing

It would, in theory, be possible to release the contents of more than one tank into a bund at the same time. Bunds are required to have a capacity of at least 110% the capacity of the largest tank. In the biodiesel tank farm, bunds containing multiple tanks have a capacity of at least 150% of the largest tank. To cause overflowing, two large tanks would need to fail and be filled to near full capacity at time of failure. A tank fire leading to subsequent escalation to other tanks is calculated below:

The frequency of bund fire from all causes, including escalation, can be taken as 1.2 x 10^-5 ([40]). The risk at the proposed plant is further reduced because:

· Bunds are equipped with fire fighting measures such as heat detectors connected to the fire alarm, deluge systems (to cool tanks and pipelines threatened by fire), a manually operated AFFF foam system with 100% redundancy backup;

· Bunds can be drained to remove flammable/combustible material. This will limit the duration of a bund fire;

· Tank failure below the liquid level is normally not to be expected during an external fire, because of the cooling effect from the liquid.

The maximum duration of a bund fire incident resulting from tank rupture can be estimated as follows. We consider the worst case of tank T17 rupturing and spilling its entire contents of biodiesel into bund 2B; the biodiesel then ignites and burns without any mitigation. The area of the fire is 526m² and the burning rate 0.04kg s^-1 m^-2. ([41]) The maximum volume of biodiesel in T17 is 2140m³, corresponding to 1,881,000kg. Thus the burning time t_B can be estimated as:

s = 24.8 hours

Thus, a bund fire incident can be expected to last only a day. The frequency of 2 out of the 8 tanks in bund 2A randomly failing within a day may be estimated from:

per year

where 10^-4 per year is the leak frequency of a tank and 0.008 is the ignition probability. The resulting frequency of an overflowing pool fire was hence estimated to be below 10^-9 per year and was therefore considered to pose negligible risk. Tank to tank escalation may occur but this would take many hours allowing ample time for evacuation of the surrounding area (see PAFF EIA ⁽¹⁾ section 10.5.4.2).

Figure 8.6h Event Tree for Feedstock/Biodiesel Storage Tank

Figure 8.6i Event Tree for Methanol Storage Tank

8.6.5 Process Area

Following the discussions given in Section 8.5.5, pool fires within the process building are not expected to create hazards that affect offsite population. Only leaks of methanol leading to vapour cloud explosions within the building and toxic releases from the ventilation system are considered in the analysis. Frequencies for loss of containment scenarios in the process area, for equipment handling methanol, are estimated based on the failure rate of the process vessels/reactor vessels ([42]), except for the pipe reactor for which pipe failure frequencies have been adopted. The leak frequency for process vessels is taken as 10^-4 per year and the rupture frequency is 10^-5 per year. For the pipe reactor, the length of the pipe was calculated to be 350m based on equipment specifications. The failure frequency was taken as 5(10^-7 m^-1 y^-1 for leaks and 10^-7 m^-1 y^-1 for ruptures. Taking into consideration the number of vessels in each part of the plant, the leak frequencies are as summarised in Table 8.6e.

Table 8.6e Failure Frequencies in Process Area

	No. Vessels	Leak Freq. (per year)	Rupture Freq. (per year)
Esterification vessel	2	2(10^-4	2(10^-5
Methanol buffer tank	1	1(10^-4	1(10^-5
Pipe reactor	350m	1.75(10^-4	3.5(10^-5
Methanol recycle tank	1	1(10^-4	1(10^-5
MEK buffer tank	2	2(10^-4	2(10^-5
GLP settling tank	1	1(10^-4	1(10^-5
GLP collection tank	1	1(10^-4	1(10^-5
Acidulation tank	1	1(10^-4	1(10^-5
FFA buffer tank	1	1(10^-4	1(10^-5
Neutralization tank	2	2(10^-4	2(10^-5

An example event tree for methanol release (from Process Vessels) and methanol release (from Reactor) inside the process building is shown in Figure 8.6j and Figure 8.6k respectively. The effect of emergency isolation and ventilation in the event tree analysis are discussed below.

Emergency Isolation

Safety interlocks for emergency shutdown will be provided for systems within process area. The primary purpose of isolation is to minimize inventory available for release by means of tripping pumps or actuating emergency isolation valves.

The probability of actuating an isolation depends on the ability to detect a leak, the integrity of the interlock system and the response time for closing the valve/tripping a pump. Isolation can be actuated by gas detection as well as deviation in process parameters (a leak in a system would usually be accompanied by process upset which can be detected by process parameters like flow, pressure and temperature). Gas detectors will be provided at strategic locations throughout the plant where there is a potential for gas leak. The interlock system is relatively reliable and can be expected to respond positively in 90% to 95% of the demands calculated based on typical failure rates suggested by Lees ([43]).

For the purpose of this study, reliability of 90% is assumed where an isolation system can be actuated within 5min from when a leak occurred. This can significantly reduce the released inventory, thus the strength of explosion and likelihood of vapor cloud ignition. Release duration of 15 min is assumed for isolation failure.

For vessel ruptures, isolation failure has been assumed.

Emergency Ventilation

Emergency ventilation will be provided in the process building, in accordance with the relevant design codes, such IP 15 : Area Classification Code for Installations Handling Flammable Liquids, IEC 60079, Part 10 :2002 Electrical Apparatus for Explosive Gas Atmospheres and NFPA 30 : The Flammable and Combustible Liquids Code. It is assumed in the analysis that this emergency ventilation will be designed such as to prevent the accumulation of flammable concentrations of vapours, ie the ventilation rate will be sufficient to bring down the concentration to 50% of lower explosive limit (LEL). Also, the ventilation system will be designed to avoid any stagnant pockets. Hence, the likelihood of vapour cloud ignition will be significantly reduced. The emergency ventilation is assumed to have a reliability of 90%.

Figure 8.6j Event Tree for Methanol (Process Vessels) Release inside Process Building

Figure 8.6k Event Tree for Methanol (Reactor) Release inside Process Building

Frequency of Maintenance Related Incidents

Based on past incidents in the bio-diesel industry and the general process industry, maintenance related incidents are found to be a significant contributor. Therefore the frequency and consequence of such incidents are assessed separately below.

Possible maintenance-related causes of an explosion in the process area of the proposed biodiesel plant include:

· Failure to purge equipment of flammable substances prior to opening or hot work

· Failure to purge equipment of air prior to introducing flammable substances

· Spillage from pump during maintenance or changeover to/from spare pump

· Spillage or leak from drain valve used for draining or purging

An example event tree for potential air ingress scenario into process equipment is shown below. The general practices followed in the industry for purging of equipment include pressure purging (ie repeated pressurisation and depressurisation with nitrogen to remove any flammables and this process is repeated several times) or flow purging (where the equipment or piping is continuously purged with nitrogen for several hours). This purging step is carried out after isolation of an equipment and prior to opening. Similarly, this step is carried out after completion of maintenance and before flammables are introduced. At the end of the purging step, the equipment atmosphere is tested for flammables or air as applicable, using portable instruments for gas analysis. There are strict procedures followed in chemical process industry for purging and testing and similar procedures will be adopted in this plant. It is also important that the purging operation is independently verified by another operator, to reduce the risk of human error. In some plants, the supervisor will also check plant records (e.g. pressure recording charts) to verify the purging procedure. All of these checks are normally built into the permit-to-work system for maintenance. It is assumed that the operator of the proposed plant will ensure that appropriate cross-checking procedures of this type are developed and followed, and the internal atmosphere of the plant equipment is tested using a gas analyzer before introduction of the process materials.

In order to estimate the likelihood of human error in carrying out the task of purging, generic human error probabilities are adopted [Ref: Lees, F. P., Loss Prevention in the Process Industry, Second Edition, 1996, Human Factors and Human Error, Chapter 14, Table 14.26]. A nominal human unreliability of 0.003 is assumed for the task of purging. A nominal human unreliability of 0.09 is assumed for the task of testing by an independent operator following the purging step. This gives an overall human unreliability of 0.00027.

In order for an internal explosion to be possible, it is also necessary to have a significant volume of flammable vapour/air mixture. Depending on the plant configuration, the incoming process material may simply sweep the air away without significant mixing, leading to little or no possibility of an explosion. This is reflected in the event tree below as ‘Explosive mixture formed’, with an assumed probability of 0.5.

The frequency of maintenance events resulting in air ingress is taken conservatively as 10 times per year. Major maintenance is normally undertaken during shutdown once every 1-2 years, but it may occasionally be necessary to open plant equipment for emergency repairs between shutdowns. The number of such operations assumed as ten may be slightly conservative.

The ignition probability is taken as 0.008. This is based on the values of 0.01 to 0.08 (depending on quantity) adopted for unconfined methanol vapour explosions (see Table 8.6b), reduced to take account of the fact that there is no ignition sources inside the equipment except hot surfaces, and these will be below the autoignition temperature of methanol.

Figure 8.6l Event Tree for Example Maintenance Scenario

Based on the above, the frequency of an internal explosion is derived as 1.08x10^-5 per year. All of the equipment in this plant is operating at atmospheric pressure, except for the reactor which is a pipe. Hence the impact of an internal explosion will be limited as compared to that from a high pressure equipment, since the explosion energy imparted to fragments in a vessel burst is a function of the initial pressure. Referring to the incident in Castle Peak power plant in Hong Kong, where an internal explosion in a high pressure hydrogen equipment led to projection of fragments 500m away (although it did not cause any offsite fatality, there were two onsite fatalities), a similar scenario in this plant could lead to one or more fragments being projected offsite. Assuming a probability of 10% that any of the fragments could hit an individual, the overall frequency of this scenario with potential for fatality is estimated as 1.08x10^-6 per year. The actual frequency is expected to be lower since the fragments will be contained by the building walls/ structure. As a conservative assumption, it is assumed that such an incident could cause two fatalities. This scenario is carried forward as a separate scenario in the derivation of FN curve and PLL to represent the potential for incidents leading to an internal explosion causing fragment hazards to public offsite. This scenario has been considered for future population case. For the current population, there is only one building at the Gammon site, with a much smaller footprint area, and shielded by the building in the project site.

There is some uncertainty in the estimation of the frequency and consequence of such events as it relies largely on the effectiveness of the procedures and the management systems in place. These are not developed yet and has been assumed to be of similar standard as those adopted by well managed companies in the process industry and the best practices adopted by companies in Hong Kong, such as those operating LPG, flammable liquids, and gas.

With regard to other maintenance related scenarios which lead to a spill inside the process building, their frequency of incidence is included in the base case frequency. These types of incidents are not specific to biodiesel facilities and could, in principle, occur in any plant handling flammable liquids. Also, since the vapour or liquid spill volume from such an incident is likely to be small, the explosion itself will cause little damage or harm.

Escalation Caused by Vapour Cloud Explosion in the Process Building

Vapour cloud explosion in the process building can potentially damage the storage tanks in the tank farm. Overpressure of 3 psi can cause uplifts (half tilted) of the tanks ([44]), which might lead to catastrophic rupture. In the worst case, the 3 psi blast wave can affect multiple tanks on site.

Figure 8.6m Maximum area affected by vapour cloud explosion (process building) at an overpressure of 3 psi

The fat preparation room and steam boiler room are located between the main process building and the tank farm, providing protection for most of the storage tanks from the blast wave. However, T10 (methanol storage tank) and T7 (PFAD storage tank) are directly exposed and may be seriously damaged. Considering all the vapour cloud explosion scenarios that can reach these tanks at an overpressure of 3 psi, the frequency of explosions causing rupture of T10 is estimated as 2.59(10^-6 per year, and the frequency causing rupture of T7 is estimated as 1.59(10^-6 per year. These frequencies are obtained by adding up the frequencies of all the relevant scenarios, as indicated in Table 8.6f. Catastrophic loss of containment from T10 and T7 could lead to overtopping of their bunds, followed by potential pool fire in the plant. These scenarios are shown in Table 8.5i and the consequences are detailed in Table 8.7g. Based on the analysis carried for bund overtopping due to the failure of these tanks, no spillage beyond the plant boundary is expected.

Table 8.6f Frequency of tank failure due to escalation from explosion in process building

Equipment involved in initial incident	Rupture/Leak	Isolation Successful	Ventilation Successful	Freq. of Incident	Tanks Affected
Esterification with catalyst	Rupture	N	Y	1.80E-07	T7, T10
Esterification with catalyst	Rupture	N	N	8.00E-08	T7, T10
Esterification with catalyst	Leak	N	N	2.00E-08	T10
Methanol buffer tank	Rupture	N	N	4.00E-08	T7, T10
Methanol buffer tank	Leak	N	N	1.00E-08	T10
Pipe reactor	Rupture	N	Y	1.26E-07	T10
Pipe reactor	Rupture	N	N	3.50E-08	T7, T10
Pipe reactor	Leak	N	Y	6.30E-07	T10
Pipe reactor	Leak	N	N	1.75E-07	T7, T10
Pipe reactor	Rupture	Y	N	1.26E-07	T7, T10
Pipe reactor	Leak	Y	N	6.30E-07	T7, T10
MEK buffer tank	Rupture	N	N	8.00E-08	T7, T10
MEK buffer tank	Leak	N	N	2.00E-08	T10
GLP collection tank	Rupture	N	Y	9.00E-08	T10
GLP collection tank	Rupture	N	N	4.00E-08	T7, T10
GLP collection tank	Leak	N	N	1.00E-08	T10
Acidulation tank	Rupture	N	N	4.00E-08	T7, T10
Acidulation tank	Leak	N	N	1.00E-08	T10
FFA buffer tank	Rupture	N	N	4.00E-08	T7, T10
FFA buffer tank	Leak	N	N	1.00E-08	T10
Neutralization tank	Rupture	N	N	8.00E-08	T7, T10
Neutralization tank	Leak	N	N	2.00E-08	T10
Methanol recycle tank	Rupture	N	N	4.00E-08	T7, T10
Methanol recycle tank	Leak	N	N	1.00E-08	T10
GLP settling tank	Rupture	N	N	4.00E-08	T10
Total frequency for rupture of T10 due to escalation				2.59E-06
Total frequency for rupture of T7 due to escalation				1.59E-06

8.6.6 Summary of Scenario Outcome Frequencies

Combining the initiating event frequency with probabilities for each branch of the event trees gives the outcome frequency for each scenario. A complete list of these scenarios and their frequencies is provided in Table 8.6g and Table 8.6h.

Table 8.6g Outcome Event Frequencies (Tank farm, Marine and Transport Scenarios)

Section No.	Section Code	Description	Equipment Tag	Outcome Frequency (per year)
				Leak					Rupture
				Bund fire	Pool fire (unconfined)	Flash fire	VCE	Toxic	Bund fire	Pool fire (unconfined)	Flash fire	VCE	Toxic
Tank farm
1	T1/T2	Crude grease trap waste storage tank	B100300/ B100400	1.6(10^-6					6.0(10^-8	2.0(10^-8
2	T3	Cleaned grease trap waste storage tank	B100500	8.0(10^-7					3.0(10^-8	1.0(10^-8
3	T4/T5	Crude waste cooking oil storage tank	B100600/ B100700	1.6(10^-6					6.0(10^-8	2.0(10^-8
4	T6	Cleaned waste cooking oil storage tank	B100800	8.0(10^-7					3.0(10^-8	1.0(10^-8
5	T7	PFAD storage tank	B100900	8.0(10^-7					3.0(10^-8	1.0(10^-8 (f) 1.59x10^-6 (g)
6	T8	Crude animal fats storage tank	B100100	8.0(10^-7					3.0(10^-8	1.0(10^-8
7	T9	Cleaned animal fats storage tank	B100200	8.0(10^-7					3.0(10^-8	1.0(10^-8
8	T10	Methanol storage tank	B102400	4.0(10^-6		2.0(10^-7	3.8(10^-6	9.2(10^-5	5.0(10^-8	1.5(10^-7 (f) 2.59x10^-6(g)	2.5(10^-9 (a) 1.5(10^-8 (b)	2.0(10^-9 (a)1.8(10^-7 (b)	1.15(10^-6
9	T14	Infineum (additive) storage tank	B102200	8.0(10^-7					2.0(10^-8	2.0(10^-8
10	T15/ T16	Quality biodiesel storage tank	B200500/ B200600	1.6(10^-6					6.0(10^-8	2.0(10^-8
11	T17	Biodiesel-Europe storage tank	B200700	8.0(10^-7					2.0(10^-8	2.0(10^-8
12	T18	Biodiesel-Hong Kong storage tank	B200800	8.0(10^-7					3.0(10^-8	1.0(10^-8
13	T19	Glycerine storage tank	B202100	8.0(10^-7					2.0(10^-8	2.0(10^-8
14	T21	Bioheating oil storage tank	B202200	8.0(10^-7					3.0(10^-8	1.0(10^-8
15	T22	Gas oil storage tank	B102300	8.0(10^-7					3.0(10^-8	1.0(10^-8
16	T24	Crude waste cooking oil	B101000	8.0(10^-7					2.0(10^-8	2.0(10^-8
17	G01	Biogas buffer tank	V601			1.0(10^-4					5.0(10^-6
Marine scenarios
29	M01	Biodiesel/PFAD barge	-		7.68(10^-8 (c)					6.14(10^-7
					2.05(10^-7 (d)
30	M02	Methanol barge	-							2.04(10^-6 (e)
31	M03	Methanol ISO-tanker	-		7.15(10^-10	7.15(10^-10		1.42(10^-7		2.28(10^-9	2.28(10^-9		5.24(10^-8
Transport scenarios
32	L01	Unloading line from jetty to tanks (biodiesel/PFAD)	-		4.41(10^-8					1.40(10^-8
33	L02	Unloading line from jetty to tanks (methanol)			3.86(10^-8	3.86(10^-8		2.49(10^-6		1.22(10^-8	1.22(10^-8		2.80(10^-7
34	RT1	Road tanker for fats & oils	-		1.50(10^-8					4.80(10^-8
35	L03	Road tanker unloading line	-		6.33(10^-7					1.92(10^-7
Notes: (a) Flash fire frequency for a vapour cloud from bund area (b) Flash fire frequency for vaporisation from a liquid pool that overtopped the bund (c) Small leaks releasing 5% of inventory within 20 minutes (d) Large leaks releasing 15% of inventory within 20 minutes (e) Since methanol is miscible with water, leakage into sea will have no effect. This scenario is therefore modelled as a tank fire on the barge (f) Generic outcome frequency for rupture of tanks T7 and T10 leading to unconfined pool fire (g) Outcome frequency for rupture of tanks T7 and T10 triggered by VCE inside process building, leading to unconfined pool fire (derived from Table 8.6f)

Table 8.6h Outcome Event Frequencies (Process Area)

Section No.

Section Code

Description

Equipment Tag

Outcome Frequency (per year)

Leak

Rupture

Isolation failure, ventilation success

Isolation failure, ventilation failure

Isolation success, ventilation success

Isolation success, ventilation failure

Isolation failure, ventilation success

Isolation failure, ventilation failure

Isolation success, ventilation success

Isolation success, ventilation failure

18

P01

Esterification with catalyst

B10200, B10400

VCE

9.00x10^-8

2.00 x10^-8

1.62 x10^-7

9.00 x10^-8

1.80 x10^-7

8.00 x10^-8

NA

NA

Toxic

1.79x10^-5

1.98 x10^-6

1.62 x10^-4

1.79 x10^-5

1.78 x10^-5

1.92 x10^-6

NA

NA

19

P02

Methanol buffer tank

B111100

VCE

4.50x10^-8

1.00 x10^-8

8.10 x10^-8

4.50 x10^-8

9.00 x10^-8

4.00 x10^-8

NA

NA

Toxic

8.96x10^-6

9.90 x10^-7

8.09 x10^-5

8.96 x10^-6

8.91 x10^-6

9.60 x10^-7

NA

NA

20

P03

Pipe reactor

SP112000

VCE

6.30x10^-7

1.75 x10^-7

1.42 x10^-6

6.30 x10^-7

1.26 x10^-7

3.50 x10^-8

2.84 x10^-7

1.26 x10^-7

Toxic

1.51x10^-5

1.58 x10^-6

1.40 x10^-4

1.51 x10^-5

3.02 x10^-6

3.15 x10^-7

2.81 x10^-5

3.02 x10^-6

21

P04

Methanol recycle tank

B72300

VCE

4.50x10^-8

1.00 x10^-8

8.10 x10^-8

4.50 x10^-8

9.00 x10^-8

4.00 x10^-8

NA

NA

Toxic

8.96x10^-6

9.90 x10^-7

8.09 x10^-5

8.96 x10^-6

8.91 x10^-6

9.60 x10^-7

NA

NA

22

P05

MEK buffer tank

B20500, B20600

VCE

9.00x10^-8

2.00 x10^-8

1.62 x10^-7

9.00 x10^-8

1.80 x10^-7

8.00 x10^-8

NA

NA

Toxic

1.79x10^-5

1.98 x10^-6

1.62 x10^-4

1.79 x10^-5

1.78 x10^-5

1.92 x10^-6

NA

NA

23

P06

GLP settling tank

B60300

VCE

4.50x10^-8

1.00 x10^-8

8.10 x10^-8

4.50 x10^-8

9.00 x10^-8

4.00 x10^-8

NA

NA

Toxic

8.96 x10^-6

9.90 x10^-7

8.09 x10^-5

8.96 x10^-6

8.91 x10^-6

9.60 x10^-7

NA

NA

24

P07

GLP collection tank

B50100

VCE

4.50 x10^-8

1.00 x10^-8

8.10 x10^-8

4.50 x10^-8

9.00 x10^-8

4.00 x10^-8

NA

NA

Toxic

8.96 x10^-6

9.90 x10^-7

8.09 x10^-5

8.96 x10^-6

8.91 x10^-6

9.60 x10^-7

NA

NA

25

P08

Acidulation tank

B50400

VCE

4.50 x10^-8

1.00 x10^-8

8.10 x10^-8

4.50 x10^-8

9.00 x10^-8

4.00 x10^-8

NA

NA

Toxic

8.96 x10^-6

9.90 x10^-7

8.09 x10^-5

8.96 x10^-6

8.91 x10^-6

9.60 x10^-7

NA

NA

26

P09

FFA buffer tank

B50900

VCE

4.50 x10^-8

1.00 x10^-8

8.10 x10^-8

4.50 x10^-8

9.00 x10^-8

4.00 x10^-8

NA

NA

Toxic

8.96 x10^-6

9.90 x10^-7

8.09 x10^-5

8.96 x10^-6

8.91 x10^-6

9.60 x10^-7

NA

NA

27

P10

Neutralization tank

B60100

VCE

9.00 x10^-8

2.00 x10^-8

1.62 x10^-7

9.00 x10^-8

1.80 x10^-7

8.00 x10^-8

NA

NA

Toxic

1.79 x10^-5

1.98 x10^-6

1.62 x10^-4

1.79 x10^-5

1.78 x10^-5

1.92 x10^-6

NA

NA

28

Internal explosion due to maintenance scenario

1.08 x 10^-6 (rupture frequency due to internal explosion)

8.7 Consequence Analysis

8.7.1 Physical Effects Modelling

In this study, the physical effects models included in the PHAST v6.51 suite of models were used to assess the effects zones for the hazardous outcomes of concern:

· Pool fire;

· Flash fire;

· Vapour cloud explosion in the process area; and

· Toxic releases of methanol.

Pool Fires

The pool formed from a release will initially spread due to gravitational and surface tension forces acting on it. As the pool spreads, it will absorb heat from its surroundings and vaporise. The pool will continue to spread until it is confined by a bund or, for unconfined releases, until an equilibrium is reached between discharge rate and vaporisation rate.

Bund Overtopping

The tank farm was assessed in particular detail in the current study. While the bund capacity is large enough to contain spills from the storage tanks, the bund may still be overtopped due to the dynamic effects in the case of an instantaneous failure and release of the full contents of one of the tanks. In such a case, if a large and full tank is located close to the bund wall and if the release is at least partially directed towards the wall, part of the inventory could spill over the bund wall and cause a pool fire outside the bund. The consequences of such a release would be more severe, since the pool fire would then cover a larger area.

The fraction of inventory that overtops the bund was modelled using the relationship of Thyer et al ([45]):

Q = 0.044 — 0.264 ln (h₁/H) — 0.116 ln (r₁/H)

where h₁ is the bund wall height, H is the tank liquid level and r₁ is the distance from the centre of the tank to the bund wall. Applying this relation to the storage tanks gives the calculations shown in Table 8.7a; the results are summarized in Table 8.7b. Overtopping of the site boundary wall, at a distance of r₂ from the tank centre, is also considered where there is a significant direct path from the tank to the boundary wall, not obstructed by other tanks, buildings or additional bund walls. The height of the boundary wall h₂ is taken as 2m.

This modelling was done based on experiments with water. It can be argued, however, that the higher viscosity of biodiesel and feedstock oils used in the proposed plant would result in greatly decreased overtopping. The grounds for this argument are as follows. Theoretical studies by Greenspan and Young([46]) indicate that the extent of overtopping is related linearly to the velocity of the wave front from the tank to the bund wall, as follows:

where H is the tank liquid level, R is the tank radius, T is the elapsed time since the rupture, u_w is the mean velocity of the wave front, and c_w is the height by which the wave height exceeds the bund wall height. If it is conservatively assumed that c_w for a viscous liquid will be equal to that for a free-flowing liquid, it can be deduced that Q is directly proportional to u_w

Next, we consider the definition of the dynamic viscosity, µ:

where τ is the shear force between liquid layers in a flowing liquid, and dc/dy is the difference in velocity between two layers separated by a depth y. Given a constant τ, which will be induced by the head in the tank, there is an inverse relationship between shear velocity and viscosity. If we extrapolate from shear velocity between layers of liquid (dc/dy) to wave front velocity u_w (i.e. the velocity difference between the top and bottom of the flowing liquid layer), it is reasonable to deduce an approximately inverse linear relationship between Q, the fraction of overtopping, and µ.

Typical viscosities for the combustible liquids in the proposed plant at 40 C are as follows: water, 0.658 cP; methanol, 0.39 cP ([47]); biodiesel, 3.5 cP; vegetable oil feedstock, 30 cP; Infineum, 661 cP. Thus, the overtopping values derived from Thyer’s formula above may be adjusted by a factor of (0.658 / 0.39) = 1.69 for methanol, (0.658 / 3.5) = 0.19 for biodiesel, (0.658 / 30) = 0.022 for vegetable oils, and (0.658 / 661) = 0.0010 for Infineum. The adjusted values of Q are shown in the final column of Table 8.7a. No adjustment has been applied to crude grease trap waste as this material is around 85% water.

Table 8.7a Bund overtopping calculations

Tank ID	Description	Bund ID	Bund wall height h1 (m)	Maximum tank liquid level relative to bund floor H1, H2 (m)	Shortest distance from centre of tank to bund wall r1 (m)	Site wall height relative to bund floor h2 (m)	Shortest distance from centre of tank to site boundary r2 (m)	Fraction Over-topping Bund Q1	Fraction Over-topping Site Wall Q2	Fraction Reaching Outside Site Boundary Q1(Q2	Q1(Q2 adjusted for viscosity
T1	Crude grease trap waste storage tank	2A	3.15	13.8	7.4	3.5	30.1	0.51	0 (Obstructed)	0	0
T2	Crude grease trap waste storage tank	2A	3.15	13.8	7.4	3.5	17.8	0.51	0.38	0.19	0.19
T3	Cleaned grease trap storage tank	2A	3.15	13.8	7.4	3.5	17.8	0.51	0.38	0.19	0.004
T4	Crude WCO storage tank	2E	1.85	7.6	3.4	2.2	43.7	0.51	0 (Obstructed)	0	0
T5	Crude WCO storage tank	2E	1.85	7.6	3.4	2.2	36.3	0.51	0 (Obstructed)	0	0
T6	Cleaned WCO storage tank	2A	3.15	13.8	7.4	3.5	17.8	0.51	0.38	0.19	0.004
T7	PFAD storage tank	2A	3.15	13.8	7.4	3.5	53.6	0.51	0 (Obstructed)	0	0
T8	Crude animal fat storage tank	2A	3.15	10.9	7.4	3.5	17.8	0.42	0.29	0.12	0.003
T9	Cleaned animal fat storage tank	2A	3.15	10.9	7.4	3.5	N/A	0.42	0 (Obstructed)	0	0
T10	Methanol storage tank	2C	4.55	9.2	5.2	4.9	N/A	0.30	0 (Obstructed)	0	0
T14	Infineum (additive) storage tank	2F	1.85	4.7	3.3	2.2	13.7	0.33	0.12	0.04	0.000
T15	Quality biodiesel storage tank	2B	6.35	9.8	8.0	6.7	18.4	0.18	0.07	0.01	0.002
T16	Quality biodiesel storage tank	2B	6.35	9.8	8.0	6.7	18.4	0.18	0.07	0.01	0.002
T17	Biodiesel-Europe storage tank	2B	6.35	17.4	8.0	6.7	18.4	0.40	0.29	0.12	0.022
T18	Biodiesel-Hong Kong storage tank	2B	6.35	13.2	6.5	6.7	N/A	0.32	0 (Obstructed)	0	0
T19	Glycerin storage tank	2A	3.15	9.8	5.3	3.5	17.8	0.41	0.25	0.10	0.000
T21	Bioheating oil storage tank	2B	6.35	7.2	6.5	6.7	N/A	0.09	0 (Obstructed)	0	0
T22	Gas oil storage tank	2F	1.85	5.5	4.5	2.2	42.9	0.36	0 (Obstructed)	0	0
T24	Crude WCO storage tank	2B	6.35	14.7	7.4	6.7	18.2	0.35	0.23	0.08	0.002

Table 8.7b Bund Overtopping Results

Tank ID	Description	Q1(Q2 adjusted for viscosity	Maximum inventory in tank TV (m^3)	Spill offsite = TV (Q1(Q2 (m^3)	Offsite Pool area (m^2)	Offsite pool diameter (m)
T2	Crude grease trap waste storage tank	0.19	1420	269.9	14103	134
T3	Cleaned grease trap storage tank	0.004	965	3.9	380	22
T6	Cleaned WCO storage tank	0.004	965	3.9	380	22
T8	Crude animal fat storage tank	0.003	477	1.4	154	14
T15	Quality biodiesel storage tank	0.002	427	0.9	154	14
T16	Quality biodiesel storage tank	0.002	427	0.9	154	14
T17	Biodiesel-Europe storage tank	0.022	2111	46.5	4301	74
T24	Crude WCO storage tank	0.002	1143	2.3	254	18

The data in Table 8.7a/b is based on the conservative assumption of the tanks being 100% full. Lower liquid levels will produce substantially less overtopping, according to Thyer’s formula given above. Also, it may be noted that crude grease trap waste storage tank, which is the feedstock contains a high proportion of water (up to 80%). Hence this liquid is not combustible. However, this liquid has been assumed as combustible for the analysis here due to the uncertainty in the water content and as worst case assumption.

The modelling of leaks from storage tanks were modelled as follows:

· Small leaks were assumed to be contained by the bund;

· Bund overtopping was considered for tank rupture scenarios. The likelihood of overtopping was estimated based on release orientation with respect to the bund wall (see Section 8.6.4);

· Single overtopping of the bund was considered for all tanks. The resulting pool fire was then modelled as an unconfined pool with contents equal to the liquid fraction that overtopped the bund. In cases where the pool size extended beyond the site perimeter wall, the pool size was limited by the site wall. Essentially, it was assumed that the perimeter wall would contain these large pools;

· Double overtopping of both the bund and site wall was considered only for those tanks listed in Table 8.7b. Double overtopping is modelled as an unconfined pool within the plant and a separate unconfined pool outside the perimeter wall, each with their own ignition probability.

The effects of pool fires and bund fires are modelling using the same methodology as adopted in the PAFF EIA study ([48]). The flame drag distance is calculated using the Moorhouse correlation in the HSE review ([49]).

Fr₁₀ = Froude number of pool fire based on windspeed at a height of 10m.

U₁₀ = windspeed at a height of 10m.

D’ = flame dragged diameter of pool fire (m).

D = pool fire diameter (m).

The flame drag effects as estimated above are considered. Where the flame drag is not significant, a minimum drag distance of 3 m is taken, which is a conservative estimation. Any person outdoors caught within the flame envelope is assumed to be fatally injured (ie 100% fatality). For person indoors (ie inside a building that lies within the flame envelope considering drag), a fatality probability of 10% is assumed to account for potential secondary fires. Anyone outside the flame envelope can escape and hence no fatality is expected.

Geometry and Location of Offsite Pool Fires

The area of liquid pools offsite arising from overtopping of the site boundary wall has been modelled using PHAST. The equivalent circular pool diameter is calculated. The location of pools offsite is assumed to be adjacent to the site boundary and directly opposite the source of the spill. Examples for T2 (crude grease trap waste) and T17 (biodiesel) are shown in Figure 8.7a.

Figure 8.7a Model of Offsite Pool Geometry

Note: the illustration shows notional pool zones of the same area as the circular pools modelled by PHAST.

Flash Fires

Only methanol has sufficient vapour pressure to produce a flammable vapour cloud. For methanol releases, if there is no immediate ignition, the vapour may disperse before subsequently encountering an ignition source. The vapour cloud will then burn with a flash back to the source of the leak. A flash fire is assumed to be fatal to anyone caught within the flash fire envelope, although the short duration of a flash fire means that radiation effects are negligible. The fatality probability is therefore zero for persons outside the flash fire envelope. Persons indoors are assumed to be offered some protection by the building and a fatality probability of 0.1 is applied.

Dispersion modelling is employed by PHAST to calculate the extent of the flammable vapour cloud. This takes into account both the direct vaporisation from the release, and also the vapour formed from evaporating pools. The hazard footprint was obtained by simulating the dispersion to the lower flammability limit concentration (LFL).

The consequence modelling demonstrated that flash fire distances are not large, extending a maximum of 45m from the methanol storage tank and 53m from a leak in the unloading hose. These are insufficient to affect offsite areas but are nevertheless retained in the modelling.

The possibility of a vapour cloud dispersing from the ventilation exhaust on the roof of the process building was also considered but consequence distances were less than 20m. If ignited, there would be no impact at ground level and the fire would flash back to the process building resulting in a confined vapour cloud explosion (VCE). The consequences of the VCE are more severe than the flash fire from the vent and hence these events were modelled as a VCE.

Vapour Cloud Explosions

The process building is confined and congested with process vessels. A VCE is possible if flammable concentrations of methanol occur and is ignited. VCE in the process building is modelled using the TNO Multi Energy explosion model included in the PHAST suite. The multi-Energy Method is a relatively simple model to determine the blast from VCE as a function of explosion characteristic to distance to the explosion sources and is widely used to obtain a conservative quantification of the explosion strength. The following highlights the key conservative assumptions of the model:

· Based on a review of experimental data by Kineslla([50]), the maximum overpressure generated from an explosion relates also to ignition energy. This is particularly true for less reactive material such as methanol which requires a larger ignition energy to produce a strong explosion.

· TNO-ME also assumes that the potential explosion source is filled with a stoichiometric fuel-air mixture and the corresponding quantities of combustion energy is then applied for estimating the explosion characteristic. This is a very conservative approach as it assumes the entire energy contributing to the blast whereas only a portion of the vapour cloud will be at the optimum stoichiometric condition in reality. Indeed, the TNO report⁽²⁾ also suggests that a gas explosion shows a gradual development under most circumstances. Such a development implies that a portion of the flammable mass is burned at a low combustion rate which will not contribute to the blast of the specified high strength.

Based on the method suggested by GAME (guidance for the application of the multi-energy model) ([51]) ,TNO curve 9 is used in the calculation considering the size and confinement of the potential explosion site, obstacle size, volume blockage ratio and material characteristic.

The walls of the process building are metal clad. It is assumed that these will do little to contain explosion overpressures and hence the effects of explosions will have impact outside the building.

Process vessels containing methanol tend to be at slightly elevated temperatures near the boiling point of methanol. Calculations demonstrated that the vaporisation rate upon release can be significant. The modelling took the following approach:

· If emergency ventilation is successfully activated (in 90% of cases), the methanol vapour cloud generated will be diluted and vented from the building at a rate of minimum of 5 air changes per hour [It is noted that higher air change may be required as per design codes, such as IP 15]. The concentration within the building was then calculated by solving the differential equation:

where V is the volume of the building (m³), is the release rate of methanol (kg/s), b is the ventilation rate (m³/s) and C is the concentration of methanol vapour within the building (kg/m³). This gives the concentration as a function of time and ignition is assumed to occur at the highest concentration.

· If the emergency ventilation fails to activate, no mechanical ventilation is assumed to continue during an incident. Also, there is assumed to be no natural ventilation. Natural ventilation is not practicable in Ex-classified zones (that is, zones where a flammable atmosphere may exist, requiring the use of non-sparking electrical equipment and tools), because flammable vapours may escape into non-flameproofed zones outside.

· The calculated concentration is then compared to the LEL (Lower Explosion Limit) concentration of methanol. Ignition or explosion is not considered if the calculated concentration is below the LFL concentration. However, it is possible that local pockets of methanol vapour may remain at a concentration above LFL, due to uneven ventilation. In view of this, explosion has been considered for all the methanol release scenarios, including scenarios for which the calculated concentration is below LEL, to account for any localized high concentration. For these scenarios, the flammable mass is estimated based on the average concentration inside the building, which is lower as compared to the worst case scenario of ventilation failure

· For vessel ruptures, the entire contents of the vessel are assumed to be released instantaneously. This forms a liquid pool on the floor, which then vaporises. The pool is assumed to be confined to the process building due to the provision of curbed areas and the vaporisation rate was determined from PHAST. All vapour is assumed to be methanol since this is the only volatile component used in the process building.

· For leaks from process vessels, the discharge rate of material is calculated based on pressure (including head of liquid within vessel). The procedure is then similar to that adopted for ruptures. A liquid pool is formed from which vaporisation occurs and the transient concentration within the building is calculated.

The fatality probability for VCEs is taken from CIA guidelines ([52]). The fatality probability is higher indoors because of the increased risk from flying debris such as breaking windows (see Table 8.7c).

Table 8.7c End Point Criteria for Vapour Cloud Explosions

Overpressure (psi)	Fatality Probability (outdoors)	Fatality Probability (indoors)
5	0.09	0.55
3	0.02	0.15
1	0.00	0.01

Toxic Releases

Methanol is mildly toxic and so the toxic effects were modelled for methanol vapour clouds where ignition does not occur. The lethal concentration in 50% of rats (LC₅₀) is reported as 64,000ppm for a 4 hour exposure. To interpret this in terms of toxicity to humans, the method of Lees ([53]) was adopted. This is based on a probit equation of the form:

where Y is the probit, k₁ and k₂ are constants with k₂ taken to be unity, n is assumed to be 2 and t is the exposure time. For a 10 minute exposure, the LC₅₀ concentration becomes 313,500 ppm. This is scaled by a factor of 0.25 to convert rat dose to human dose, giving an LC₅₀ for humans of 78,384 ppm/10min. 50% fatality occurs for a probit of 5.0. The probit equation may therefore be solved for this boundary condition to obtain the constant k₁. This results in the following probit equation used in the current study:

where C is in ppm and t is in minutes. This gives the fatality probabilities summarised in Table 8.7d, assuming 10 minute exposure.

Table 8.7d End Point Criteria for Toxic Releases

Methanol Concentration (ppm)	Fatality Probability (outdoors)	Fatality Probability (indoors)
190,000	0.99	0.099
113,000	0.9	0.09
59,000	0.5	0.05
31,000	0.1	0.01
18,500	0.01	0.001

Smoke Dispersion

The tank farm consists of about a dozen large storage tanks containing heavy hydrocarbons including crude vegetable oil, biodiesel, and glycerine. In case of ignition on loss of containment, incomplete combustion of crude oil and biodiesel will generate thick black smoke and potentially hazardous gases including carbon monoxide, nitrogen oxides and sulphur oxides. In the case of large diameter bund fires, a substantial volume of smoke may be produced.

However, smoke from such fire will be buoyant and tends to rise and has minimum impact on ground level population. This was observed in the Buncefield incident, for example ([54]). In the following analysis, the approach taken in the PAFF report ([55]), was adopted whereby the smoke plume is considered to rise at an angle determined by the prevailing wind speed.

Based on Tseung Kwan O meteorological data (2003-2007) ([56]), the wind speed is 3 m/s or below for 95% of the time. Thus, a 5 m/s wind is assumed, to assess the worst-case impact of the smoke plume on the adjacent buildings. The results are summarized in Table 8.7e below.

In terms of existing buildings, the closest high rise building in the vicinity is the Dream City development in Tseung Kwan O area 86. It is about 800m north of the biodiesel plant, so any smoke plume will be well clear of the development once it has travelled that distance. Smoke will, therefore, pose no significant risk to the current population of the surrounding area.

For future developments, general buildings in the industrial estate are assumed to have a maximum height of 30 m. Therefore, they may be affected by the smoke plume at a distance of 10 m from the biodiesel plant boundary.

Table 8.7e Buildings adjacent to the biodiesel plant affected by smoke plume

Distance of building from the biodiesel site boundary (m)	Minimum height of the building that can be affected by the smoke plume (m)
0	0
5	6
10	13
20	26
30	39
40	52
50	66

The occupants inside the building in the vicinity of the biodiesel plant could be exposed to potentially toxic smoke due to the combined incapacitation effects of CO₂ (causing hyperventilation) and CO (toxic narcosis). The composition of smoke plume of heavy hydrocarbons is estimated as about 11.8% CO₂ and 800 ppm of CO ([57]). At 800 ppm CO, the time required for incapacitation is about 48 seconds and at 300 ppm, the time required is 20 min.

The time between arrival of the smoke plume and harm to the building’s occupants is estimated as follows. The smoke could penetrate and diffuse inside the impacted building through the heating, ventilating and air-conditioning (HVAC) system. The air exchange rate of a typical office building is estimated as ([58]):

( = 0.87 + 0.13 u_m for an exposed site,

( = 0.88 for a sheltered site (u_m < 4.2 m/s),

( = 0.22 u_m for a sheltered site (u_m > 4.2 m/s).

Where ( is the building air exchange rate and u_m is the wind speed.

To calculate the variation of indoor concentration of toxic gases with time, it is assumed (1) the outdoor concentration is steady and the same as the concentration in the smoke plume, (2) any gas that penetrates the building is mixed immediately and perfectly with all the air in the building to produce a uniform concentration. Under these conservative assumptions, the rate of change of the indoor concentration is simply proportional to the concentration driving force, given by

Where C_i is the indoor concentration (ppm), C_o is the outdoor concentration (ppm), and ( is the building air exchange rate (air changes per hour). The indoor concentration is thus given by

C_i = C₀ (1-exp(-(t))

Where t is the time elapsed since the arrival of the smoke plume.

From this analysis, the minimum time to reach 300 ppm CO inside the building (CO in the smoke plume: 800 ppm) can be determined as 32 min (( > 0.87). Considering the time required for the incapacitation at 300 ppm CO is 20 min, there is no fatality in the first 20 min starting from the impact of the smoke plume. Furthermore, 20 min is usually sufficient to evacuate the occupants inside the building. Therefore, the risk of fatalities caused by the smoke plume is assessed to be negligible. This analysis on smoke impacts is applicable to buildings outside the flame envelope. For those buildings inside the flame envelope, a fatality probability of 10% is assumed as described in the earlier paragraphs under pool fire/ bund fire impacts.

Smoke from Offsite Pool Fire

Under certain scenarios, combustible liquid (chiefly biodiesel) spilt onsite may exceed the site boundary and possibly ignite, causing an offsite pool fire. Smoke from this fire could also affect surrounding buildings, and this could potentially cause a greater effect than onsite fire because the pool fire may be closer to the affected buildings.

To analyze this situation, we have considered the duration of the worst-case offsite pool fire. The maximum pool volume is calculated as 270 m³ (see Table 8.7a), leading to a pool area of 1.41 x 10⁴ m², and this burns at a rate 0.039 kg s^-1 m^-2 ([59]) Thus, the duration of the worst pool fire is 894kg/m³ x 270 m³ / (0.039 kg s^-1 m^-2 x 1.41 x 10⁴ m²) = 7.3 min. Since this is considerably less than the 20 min required to build up a hazardous concentration of CO indoors, the offsite pool fire can be considered to pose negligible risk.

8.7.2 Consequence Analysis Results

Detailed results of the consequence analysis conducted for this risk assessment are shown in Table 8.7f, which tabulate the effect zones associated with various end points of the hazardous outcomes considered. Consequence results are presented in terms of:

· d: maximum downwind distance;

· c: maximum half-width;

· s: offset distance between source and effect zone; and

· m: downwind distance at which the maximum width, c, occurs.

These dimensions (Figure 8.7b) are utilized within the risk integration software to define the footprint area of the hazard and calculate the number of people affected based on the meteorological data and population distribution.

Figure 8.7b Presentation of Consequence Results

Table 8.7f Consequence Distances

Section No.	Section Code	Description	Leak Size	Outcome	End Point Criteria	Weather State 1.5F				Weather State 3B				Weather State 3D				Weather State 6D
						d	c	s	m	d	c	s	m	d	c	s	m	d	c	s	m
1	T1/T2	Crude GTW storage tank	Leak	Bund fire	Flame zone	20	20	-20	0	20	20	-20	0	20	20	-20	0	20	20	-20	0
F					Drag zone	23	23	-23	0	23	23	-23	0	23	23	-23	0	25	25	-25	0
			Rupture	Bund fire	Flame zone	20	20	-20	0	20	20	-20	0	20	20	-20	0	20	20	-20	0
					Drag zone	23	23	-23	0	23	23	-23	0	23	23	-23	0	25	25	-25	0
			Rupture	Pool fire	Flame zone	75	75	-75	0	75	75	-75	0	75	75	-75	0	75	75	-75	0
			(over- topping)	(unconfined)	Drag zone	78	78	-78	0	79	79	-79	0	79	79	-79	0	87	87	-87	0
			Rupture	Pool fire	Flame zone	67	67	-67	0	67	67	-67	0	67	67	-67	0	67	67	-67	0
			(over- topping)	(unconfined, outside site)	Drag zone	70	70	-70	0	71	71	-71	0	71	71	-71	0	78	78	-78	0
2	T3	Cleaned GTW storage tank	Leak	Bund fire	Flame zone	20	20	-20	0	20	20	-20	0	20	20	-20	0	20	20	-20	0
F					Drag zone	23	23	-23	0	23	23	-23	0	23	23	-23	0	25	25	-25	0
			Rupture	Bund fire	Flame zone	20	20	-20	0	20	20	-20	0	20	20	-20	0	20	20	-20	0
					Drag zone	23	23	-23	0	23	23	-23	0	23	23	-23	0	25	25	-25	0
			Rupture	Pool fire	Flame zone	75	75	-75	0	75	75	-75	0	75	75	-75	0	75	75	-75	0
			(over- topping)	(unconfined)	Drag zone	78	78	-78	0	79	79	-79	0	79	79	-79	0	87	87	-87	0
			Rupture	Pool fire	Flame zone	11	11	-11	0	11	11	-11	0	11	11	-11	0	11	11	-11	0
			topping)	(unconfined, outside site)	Drag zone	14	14	-14	0	14	14	-14	0	14	14	-14	0	15	15	-15	0
3	T4/T5	Crude WCO storage tank	Leak	Bund fire	Flame zone	8	8	-8	0	8	8	-8	0	8	8	-8	0	8	8	-8	0
F					Drag zone	11	11	-11	0	11	11	-11	0	11	11	-11	0	11	11	-11	0
			Rupture	Bund fire	Flame zone	8	8	-8	0	8	8	-8	0	8	8	-8	0	8	8	-8	0
					Drag zone	11	11	-11	0	11	11	-11	0	11	11	-11	0	11	11	-11	0
			Rupture	Pool fire	Flame zone	48	48	-48	0	48	48	-48	0	48	48	-48	0	48	48	-48	0
			(over-topping)	(unconfined)	Drag zone	51	51	-51	0	52	52	-52	0	52	52	-52	0	57	57	-57	0
4	T6	Cleaned WCO storage tank	Leak	Bund fire	Flame zone	20	20	-20	0	20	20	-20	0	20	20	-20	0	20	20	-20	0
F					Drag zone	23	23	-23	0	23	23	-23	0	23	23	-23	0	25	25	-25	0
			Rupture	Bund fire	Flame zone	20	20	-20	0	20	20	-20	0	20	20	-20	0	20	20	-20	0
					Drag zone	23	23	-23	0	23	23	-23	0	23	23	-23	0	25	25	-25	0
			Rupture	Pool fire	Flame zone	75	75	-75	0	75	75	-75	0	75	75	-75	0	75	75	-75	0
			(over-topping)	(unconfined)	Drag zone	78	78	-78	0	79	79	-79	0	79	79	-79	0	87	87	-87	0
			Rupture	Pool fire	Flame zone	11	11	-11	0	11	11	-11	0	11	11	-11	0	11	11	-11	0
			(over- topping)	(unconfined, outside site)	Drag zone	14	14	-14	0	14	14	-14	0	14	14	-14	0	15	15	-15	0
5	T7	PFAD storage tank	Leak	Bund fire	Flame zone	20	20	-20	0	20	20	-20	0	20	20	-20	0	20	20	-20	0
F					Drag zone	23	23	-23	0	23	23	-23	0	23	23	-23	0	25	25	-25	0
			Rupture	Bund fire	Flame zone	20	20	-20	0	20	20	-20	0	20	20	-20	0	20	20	-20	0
					Drag zone	23	23	-23	0	23	23	-23	0	23	23	-23	0	25	25	-25	0
			Rupture	Pool fire	Flame zone	75	75	-75	0	75	75	-75	0	75	75	-75	0	75	75	-75	0
			(over-topping)	(unconfined)	Drag zone	78	78	-78	0	79	79	-79	0	79	79	-79	0	87	87	-87	0
6	T8	Crude animal fats storage	Leak	Bund fire	Flame zone	20	20	-20	0	20	20	-20	0	20	20	-20	0	20	20	-20	0
F		tank			Drag zone	23	23	-23	0	23	23	-23	0	23	23	-23	0	25	25	-25	0
			Rupture	Bund fire	Flame zone	20	20	-20	0	20	20	-20	0	20	20	-20	0	20	20	-20	0
					Drag zone	23	23	-23	0	23	23	-23	0	23	23	-23	0	25	25	-25	0
			Rupture	Pool fire	Flame zone	75	75	-75	0	75	75	-75	0	75	75	-75	0	75	75	-75	0
			(over-topping)	(unconfined)	Drag zone	78	78	-78	0	79	79	-79	0	79	79	-79	0	87	87	-87	0
			Rupture	Pool fire	Flame zone	7	7	-7	0	7	7	-7	0	7	7	-7	0	7	7	-7	0
			(over-topping)	(unconfined, outside site)	Drag zone	10	10	-10	0	10	10	-10	0	10	10	-10	0	10	10	-10	0
7	T9	Clean animal fats storage	Leak	Bund fire	Flame zone	20	20	-20	0	20	20	-20	0	20	20	-20	0	20	20	-20	0
F		tank			Drag zone	23	23	-23	0	23	23	-23	0	23	23	-23	0	25	25	-25	0
			Rupture	Bund fire	Flame zone	20	20	-20	0	20	20	-20	0	20	20	-20	0	20	20	-20	0
					Drag zone	23	23	-23	0	23	23	-23	0	23	23	-23	0	25	25	-25	0
			Rupture	Pool fire	Flame zone	75	75	-75	0	75	75	-75	0	75	75	-75	0	75	75	-75	0
			(over-topping)	(unconfined)	Drag zone	78	78	-78	0	79	79	-79	0	79	79	-79	0	87	87	-87	0
8	T10	Methanol storage	Leak	Bund fire	Flame zone	7	7	-7	0	7	7	-7	0	7	7	-7	0	7	7	-7	0
F		tank			Drag zone	10	10	-10	0	10	10	-10	0	10	10	-10	0	10	10	-10	0
				Flash fire	LFL	29	6	0	25	20	1	0	6	26	1	0	14	23	1	0	12
				VCE	5 psi	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0
					3 psi	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0
					1 psi	22	22	-22	0	13	13	-13	0	14	14	-14	0	14	14	-14	0
				Toxic release	190,000 ppm	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0
					113,000 ppm	13	1	0	5	12	1	0	5	12	1	0	5	12	1	0	5
					59,000 ppm	33	11	0	28	24	1	0	7	34	1	0	20	28	1	0	18
					31,000 ppm	43	24	0	33	37	2	0	8	58	3	0	40	42	3	0	28
					18,500 ppm	51	35	0	40	50	2	0	8	74	4	0	50	51	4	0	38
			Rupture	Bund fire	Flame zone	7	7	-7	0	7	7	-7	0	7	7	-7	0	7	7	-7	0
					Drag zone	10	10	-10	0	10	10	-10	0	10	10	-10	0	10	10	-10	0
				Flash fire	LFL	23	17	-15	6	24	13	-12	4	26	14	-13	4	36	11	-9	3
				VCE	5 psi	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0
					3 psi	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0
					1 psi	65	65	-65	0	63	63	-63	0	65	65	-65	0	65	65	-65	0
				Toxic release	190,000 ppm	10	9	-8	2	12	11	-10	2	12	11	-10	2	13	10	-8	3
					113,000 ppm	14	11	-10	3	14	11	-10	3	14	11	-10	3	17	11	-9	3
					59,000 ppm	28	20	-16	7	33	15	-14	5	33	16	-14	6	49	13	-10	10
					31,000 ppm	43	31	-23	11	58	21	-16	14	62	22	-16	14	89	17	-13	25
					18,500 ppm	55	41	-29	11	68	26	-19	25	81	27	-19	25	114	23	-14	46
			Rupture	Pool fire	Flame zone	75	75	-75	0	75	75	-75	0	75	75	-75	0	75	75	-75	0
			(over-topping)	(unconfined)	Drag zone	78	78	-78	0	79	79	-79	0	79	79	-79	0	87	87	-87	0
				Flash fire	LFL	38	25	-16	12	32	14	-12	5	38	15	-12	8	45	13	-9	5
				VCE	5 psi	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0
					3 psi	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0
					1 psi	85	85	-85	0	65	65	-65	0	65	65	-65	0	65	65	-65	0
				Toxic release	190,000 ppm	11	10	-9	0	10	10	-9	0	10	10	-9	0	14	11	-9	3
					113,000 ppm	20	13	-10	6	15	10	-10	2	17	12	-11	2	19	13	-10	3
					59,000 ppm	45	30	-20	14	45	16	-13	13	52	18	-13	13	58	15	-11	11
					31,000 ppm	69	50	-31	17	69	24	-17	33	81	28	-17	38	106	22	-15	38
					18,500 ppm	186	67	-47	23	87	34	-20	44	106	38	-20	54	133	30	-17	85
9	T14	Infineum storage tank	Leak	Bund fire	Flame zone	6	6	-6	0	6	6	-6	0	6	6	-6	0	6	6	-6	0
F					Drag zone	9	9	-9	0	9	9	-9	0	9	9	-9	0	9	9	-9	0
			Rupture	Bund fire	Flame zone	6	6	-6	0	6	6	-6	0	6	6	-6	0	6	6	-6	0
					Drag zone	9	9	-9	0	9	9	-9	0	9	9	-9	0	9	9	-9	0
			Rupture	Pool fire	Flame zone	60	60	-60	0	59	59	-59	0	59	59	-59	0	59	59	-59	0
			(over-topping)	(unconfined)	Drag zone	63	63	-63	0	64	64	-64	0	64	64	-64	0	71	71	-71	0
10	T15/	Quality biodiesel storage	Leak	Bund fire	Flame zone	18	18	-18	0	18	18	-18	0	18	18	-18	0	18	18	-18	0
F	T16	tank			Drag zone	21	21	-21	0	21	21	-21	0	21	21	-21	0	23	23	-23	0
			Rupture	Bund fire	Flame zone	18	18	-18	0	18	18	-18	0	18	18	-18	0	18	18	-18	0
					Drag zone	21	21	-21	0	21	21	-21	0	21	21	-21	0	23	23	-23	0
			Rupture	Pool fire	Flame zone	75	75	-75	0	75	75	-75	0	75	75	-75	0	75	75	-75	0
			(over-topping)	(unconfined)	Drag zone	78	78	-78	0	79	79	-79	0	79	79	-79	0	87	87	-87	0
			Rupture	Pool fire	Flame zone	7	7	-7	0	7	7	-7	0	7	7	-7	0	7	7	-7	0
			(over-topping)	(unconfined, outside site)	Drag zone	10	10	-10	0	10	10	-10	0	10	10	-10	0	10	10	-10	0
11	T17	Biodiesel-Europe storage	Leak	Bund fire	Flame zone	18	18	-18	0	18	18	-18	0	18	18	-18	0	18	18	-18	0
F		tank			Drag zone	21	21	-21	0	21	21	-21	0	21	21	-21	0	23	23	-23	0
			Rupture	Bund fire	Flame zone	18	18	-18	0	18	18	-18	0	18	18	-18	0	18	18	-18	0
					Drag zone	21	21	-21	0	21	21	-21	0	21	21	-21	0	23	23	-23	0
			Rupture	Pool fire	Flame zone	75	75	-75	0	75	75	-75	0	75	75	-75	0	75	75	-75	0
			(over-topping)	(unconfined, inside site)	Drag zone	78	78	-78	0	79	79	-79	0	79	79	-79	0	87	87	-87	0
			Rupture	Pool fire	Flame zone	37	37	-37	0	37	37	-37	0	37	37	-37	0	37	37	-37	0
			(over-topping)	(unconfined, outside site)	Drag zone	40	40	-40	0	41	41	-41	0	41	41	-41	0	45	45	-45	0
12	T18	Biodiesel-Hong Kong	Leak	Bund fire	Flame zone	18	18	-18	0	18	18	-18	0	18	18	-18	0	18	18	-18	0
F		storage tank			Drag zone	21	21	-21	0	21	21	-21	0	21	21	-21	0	23	23	-23	0
			Rupture	Bund fire	Flame zone	18	18	-18	0	18	18	-18	0	18	18	-18	0	18	18	-18	0
					Drag zone	21	21	-21	0	21	21	-21	0	21	21	-21	0	23	23	-23	0
			Rupture	Pool fire	Flame zone	75	75	-75	0	75	75	-75	0	75	75	-75	0	75	75	-75	0
			(over-topping)	(unconfined)	Drag zone	78	78	-78	0	79	79	-79	0	79	79	-79	0	87	87	-87	0
13	T19	Glycerine storage tank	Leak	Bund fire	Flame zone	20	20	-20	0	20	20	-20	0	20	20	-20	0	20	20	-20	0
F					Drag zone	23	23	-23	0	23	23	-23	0	23	23	-23	0	25	25	-25	0
			Rupture	Bund fire	Flame zone	20	20	-20	0	20	20	-20	0	20	20	-20	0	20	20	-20	0
					Drag zone	23	23	-23	0	23	23	-23	0	23	23	-23	0	25	25	-25	0
			Rupture	Pool fire	Flame zone	75	75	-75	0	75	75	-75	0	75	75	-75	0	75	75	-75	0
			(over-topping)	(unconfined)	Drag zone	78	78	-78	0	79	79	-79	0	79	79	-79	0	87	87	-87	0
14	T21	Bioheating oil storage tank	Leak	Bund fire	Flame zone	18	18	-18	0	18	18	-18	0	18	18	-18	0	18	18	-18	0
F					Drag zone	21	21	-21	0	21	21	-21	0	21	21	-21	0	23	23	-23	0
			Rupture	Bund fire	Flame zone	18	18	-18	0	18	18	-18	0	18	18	-18	0	18	18	-18	0
					Drag zone	21	21	-21	0	21	21	-21	0	21	21	-21	0	23	23	-23	0
			Rupture	Pool fire	Flame zone	75	75	-75	0	75	75	-75	0	75	75	-75	0	75	75	-75	0
			(over-topping)	(unconfined)	Drag zone	78	78	-78	0	79	79	-79	0	79	79	-79	0	87	87	-87	0
15	T22	Gas oil storage tank	Leak	Bund fire	Flame zone	6	6	-6	0	6	6	-6	0	6	6	-6	0	6	6	-6	0
F					Drag zone	9	9	-9	0	9	9	-9	0	9	9	-9	0	9	9	-9	0
			Rupture	Bund fire	Flame zone	6	6	-6	0	6	6	-6	0	6	6	-6	0	6	6	-6	0
					Drag zone	9	9	-9	0	9	9	-9	0	9	9	-9	0	9	9	-9	0
			Rupture	Pool fire	Flame zone	74	74	-74	0	74	74	-74	0	74	74	-74	0	74	74	-74	0
			(over-topping)	(unconfined)	Drag zone	77	77	-77	0	78	78	-78	0	78	78	-78	0	86	86	-86	0
16	T24	Crude WCO storage tank	Leak	Bund fire	Flame zone	18	18	-18	0	18	18	-18	0	18	18	-18	0	18	18	-18	0
F					Drag zone	21	21	-21	0	21	21	-21	0	21	21	-21	0	23	23	-23	0
			Rupture	Bund fire	Flame zone	18	18	-18	0	18	18	-18	0	18	18	-18	0	18	18	-18	0
					Drag zone	21	21	-21	0	21	21	-21	0	21	21	-21	0	23	23	-23	0
			Rupture	Pool fire	Flame zone	75	75	-75	0	75	75	-75	0	75	75	-75	0	75	75	-75	0
			(over-topping)	(unconfined)	Drag zone	78	78	-78	0	79	79	-79	0	79	79	-79	0	87	87	-87	0
			Rupture	Pool fire	Flame zone	9	9	-9	0	9	9	-9	0	9	9	-9	0	9	9	-9	0
			(over-topping)	(unconfined, outside site)	Drag zone	12	12	-12	0	12	12	-12	0	12	12	-12	0	12	12	-12	0
17	G01	Biogas buffer tank	Rupture	Flash fire	LFL	3	3	-3	0	4	3	-3	0	3	3	-3	0	6	3	-3	0
18	P01	Esterification with catalyst	Rupture	VCE (isolation failure, ventilation success)	5 psi	51	51	-51	0	51	51	-51	0	51	51	-51	0	51	51	-51	0
					3 psi	70	70	-70	0	70	70	-70	0	70	70	-70	0	70	70	-70	0
					1 psi	165	165	-165	0	165	165	-165	0	165	165	-165	0	165	165	-165	0
			Rupture	VCE (isolation failure, ventilation failure)	5 psi	57	57	-57	0	57	57	-57	0	57	57	-57	0	57	57	-57	0
					3 psi	79	79	-79	0	79	79	-79	0	79	79	-79	0	79	79	-79	0
					1 psi	184	184	-184	0	184	184	-184	0	184	184	-184	0	184	184	-184	0
			Leak	VCE (isolation failure, ventilation success)	5 psi	22	22	-22	0	22	22	-22	0	22	22	-22	0	22	22	-22	0
					3 psi	30	30	-30	0	30	30	-30	0	30	30	-30	0	30	30	-30	0
					1 psi	72	72	-72	0	72	72	-72	0	72	72	-72	0	72	72	-72	0
				VCE (isolation failure, ventilation failure)	5 psi	45	45	-45	0	45	45	-45	0	45	45	-45	0	45	45	-45	0
					3 psi	62	62	-62	0	62	62	-62	0	62	62	-62	0	62	62	-62	0
					1 psi	146	146	-146	0	146	146	-146	0	146	146	-146	0	146	146	-146	0
				VCE (isolation success, ventilation success)	5 psi	13	13	-13	0	13	13	-13	0	13	13	-13	0	13	13	-13	0
					3 psi	19	19	-19	0	19	19	-19	0	19	19	-19	0	19	19	-19	0
					1 psi	44	44	-44	0	44	44	-44	0	44	44	-44	0	44	44	-44	0
				VCE (isolation success, ventilation failure)	5 psi	25	25	-25	0	25	25	-25	0	25	25	-25	0	25	25	-25	0
					3 psi	35	35	-35	0	35	35	-35	0	35	35	-35	0	35	35	-35	0
					1 psi	83	83	-83	0	83	83	-83	0	83	83	-83	0	83	83	-83	0
19	P02	Methanol buffer tank	Rupture	VCE (isolation failure, ventilation success)	5 psi	25	25	-25	0	25	25	-25	0	25	25	-25	0	25	25	-25	0
					3 psi	34	34	-34	0	34	34	-34	0	34	34	-34	0	34	34	-34	0
					1 psi	80	80	-80	0	80	80	-80	0	80	80	-80	0	80	80	-80	0
				VCE (isolation failure, ventilation failure)	5 psi	55	55	-55	0	55	55	-55	0	55	55	-55	0	55	55	-55	0
					3 psi	76	76	-76	0	76	76	-76	0	76	76	-76	0	76	76	-76	0
					1 psi	178	178	-178	0	178	178	-178	0	178	178	-178	0	178	178	-178	0
			Leak	VCE (isolation failure, ventilation success)	5 psi	17	17	-17	0	17	17	-17	0	17	17	-17	0	17	17	-17	0
					3 psi	24	24	-24	0	24	24	-24	0	24	24	-24	0	24	24	-24	0
					1 psi	57	57	-57	0	57	57	-57	0	57	57	-57	0	57	57	-57	0
				VCE (isolation failure, ventilation failure)	5 psi	36	36	-36	0	36	36	-36	0	36	36	-36	0	36	36	-36	0
					3 psi	50	50	-50	0	50	50	-50	0	50	50	-50	0	50	50	-50	0
					1 psi	118	118	-118	0	118	118	-118	0	118	118	-118	0	118	118	-118	0
				VCE (isolation success, ventilation success)	5 psi	11	11	-11	0	11	11	-11	0	11	11	-11	0	11	11	-11	0
					3 psi	16	16	-16	0	16	16	-16	0	16	16	-16	0	16	16	-16	0
					1 psi	38	38	-38	0	38	38	-38	0	38	38	-38	0	38	38	-38	0
				VCE (isolation success, ventilation failure)	5 psi	21	21	-21	0	21	21	-21	0	21	21	-21	0	21	21	-21	0
					3 psi	29	29	-29	0	29	29	-29	0	29	29	-29	0	29	29	-29	0
					1 psi	69	69	-69	0	69	69	-69	0	69	69	-69	0	69	69	-69	0
20	P03	Pipe reactor	Rupture	VCE (isolation failure, ventilation success)	5 psi	41	41	-41	0	41	41	-41	0	41	41	-41	0	41	41	-41	0
					3 psi	57	57	-57	0	57	57	-57	0	57	57	-57	0	57	57	-57	0
					1 psi	134	134	-134	0	134	134	-134	0	134	134	-134	0	134	134	-134	0
				VCE (isolation failure, ventilation failure)	5 psi	57	57	-57	0	57	57	-57	0	57	57	-57	0	57	57	-57	0
					3 psi	79	79	-79	0	79	79	-79	0	79	79	-79	0	79	79	-79	0
					1 psi	184	184	-184	0	184	184	-184	0	184	184	-184	0	184	184	-184	0
				VCE (isolation success, ventilation success)	5 psi	32	32	-32	0	32	32	-32	0	32	32	-32	0	32	32	-32	0
					3 psi	45	45	-45	0	45	45	-45	0	45	45	-45	0	45	45	-45	0
					1 psi	105	105	-105	0	105	105	-105	0	105	105	-105	0	105	105	-105	0
				VCE (isolation success, ventilation failure)	5 psi	57	57	-57	0	57	57	-57	0	57	57	-57	0	57	57	-57	0
					3 psi	79	79	-79	0	79	79	-79	0	79	79	-79	0	79	79	-79	0
					1 psi	184	184	-184	0	184	184	-184	0	184	184	-184	0	184	184	-184	0
			Leak	VCE (isolation failure, ventilation success)	5 psi	41	41	-41	0	41	41	-41	0	41	41	-41	0	41	41	-41	0
					3 psi	57	57	-57	0	57	57	-57	0	57	57	-57	0	57	57	-57	0
					1 psi	134	134	-134	0	134	134	-134	0	134	134	-134	0	134	134	-134	0
				VCE (isolation failure, ventilation failure)	5 psi	57	57	-57	0	57	57	-57	0	57	57	-57	0	57	57	-57	0
					3 psi	79	79	-79	0	79	79	-79	0	79	79	-79	0	79	79	-79	0
					1 psi	184	184	-184	0	184	184	-184	0	184	184	-184	0	184	184	-184	0
				VCE (isolation success, ventilation success)	5 psi	32	32	-32	0	32	32	-32	0	32	32	-32	0	32	32	-32	0
					3 psi	45	45	-45	0	45	45	-45	0	45	45	-45	0	45	45	-45	0
					1 psi	105	105	-105	0	105	105	-105	0	105	105	-105	0	105	105	-105	0
				VCE (isolation success, ventilation failure)	5 psi	57	57	-57	0	57	57	-57	0	57	57	-57	0	57	57	-57	0
					3 psi	79	79	-79	0	79	79	-79	0	79	79	-79	0	79	79	-79	0
					1 psi	184	184	-184	0	184	184	-184	0	184	184	-184	0	184	184	-184	0
21	P04	Methanol recycle tank	Rupture	VCE (isolation failure, ventilation success)	5 psi	29	29	-29	0	29	29	-29	0	29	29	-29	0	29	29	-29	0
					3 psi	40	40	-40	0	40	40	-40	0	40	40	-40	0	40	40	-40	0
					1 psi	95	95	-95	0	95	95	-95	0	95	95	-95	0	95	95	-95	0
				VCE (isolation failure, ventilation failure)	5 psi	57	57	-57	0	57	57	-57	0	57	57	-57	0	57	57	-57	0
					3 psi	79	79	-79	0	79	79	-79	0	79	79	-79	0	79	79	-79	0
					1 psi	184	184	-184	0	184	184	-184	0	184	184	-184	0	184	184	-184	0
			Leak	VCE (isolation failure, ventilation success)	5 psi	18	18	-18	0	18	18	-18	0	18	18	-18	0	18	18	-18	0
					3 psi	26	26	-26	0	26	26	-26	0	26	26	-26	0	26	26	-26	0
					1 psi	61	61	-61	0	61	61	-61	0	61	61	-61	0	61	61	-61	0
				VCE (isolation failure, ventilation failure)	5 psi	38	38	-38	0	38	38	-38	0	38	38	-38	0	38	38	-38	0
					3 psi	53	53	-53	0	53	53	-53	0	53	53	-53	0	53	53	-53	0
					1 psi	125	125	-125	0	125	125	-125	0	125	125	-125	0	125	125	-125	0
				VCE (isolation success, ventilation success)	5 psi	12	12	-12	0	12	12	-12	0	12	12	-12	0	12	12	-12	0
					3 psi	17	17	-17	0	17	17	-17	0	17	17	-17	0	17	17	-17	0
					1 psi	40	40	-40	0	40	40	-40	0	40	40	-40	0	40	40	-40	0
				VCE (isolation success, ventilation failure)	5 psi	22	22	-22	0	22	22	-22	0	22	22	-22	0	22	22	-22	0
					3 psi	31	31	-31	0	31	31	-31	0	31	31	-31	0	31	31	-31	0
					1 psi	73	73	-73	0	73	73	-73	0	73	73	-73	0	73	73	-73	0
22	P05	MEK buffer tank	Rupture	VCE (isolation failure, ventilation success)	5 psi	24	24	-24	0	24	24	-24	0	24	24	-24	0	24	24	-24	0
					3 psi	33	33	-33	0	33	33	-33	0	33	33	-33	0	33	33	-33	0
					1 psi	78	78	-78	0	78	78	-78	0	78	78	-78	0	78	78	-78	0
				VCE (isolation failure, ventilation failure)	5 psi	53	53	-53	0	53	53	-53	0	53	53	-53	0	53	53	-53	0
					3 psi	73	73	-73	0	73	73	-73	0	73	73	-73	0	73	73	-73	0
					1 psi	171	171	-171	0	171	171	-171	0	171	171	-171	0	171	171	-171	0
			Leak	VCE (isolation failure, ventilation success)	5 psi	17	17	-17	0	17	17	-17	0	17	17	-17	0	17	17	-17	0
					3 psi	23	23	-23	0	23	23	-23	0	23	23	-23	0	23	23	-23	0
					1 psi	55	55	-55	0	55	55	-55	0	55	55	-55	0	55	55	-55	0
				VCE (isolation failure, ventilation failure)	5 psi	35	35	-35	0	35	35	-35	0	35	35	-35	0	35	35	-35	0
					3 psi	49	49	-49	0	49	49	-49	0	49	49	-49	0	49	49	-49	0
					1 psi	115	115	-115	0	115	115	-115	0	115	115	-115	0	115	115	-115	0
				VCE (isolation success, ventilation success)	5 psi	11	11	-11	0	11	11	-11	0	11	11	-11	0	11	11	-11	0
					3 psi	15	15	-15	0	15	15	-15	0	15	15	-15	0	15	15	-15	0
					1 psi	35	35	-35	0	35	35	-35	0	35	35	-35	0	35	35	-35	0
				VCE (isolation success, ventilation failure)	5 psi	21	21	-21	0	21	21	-21	0	21	21	-21	0	21	21	-21	0
					3 psi	29	29	-29	0	29	29	-29	0	29	29	-29	0	29	29	-29	0
					1 psi	68	68	-68	0	68	68	-68	0	68	68	-68	0	68	68	-68	0
23	P06	GLP settling tank	Rupture	VCE (isolation failure, ventilation success)	5 psi	21	21	-21	0	21	21	-21	0	21	21	-21	0	21	21	-21	0
					3 psi	29	29	-29	0	29	29	-29	0	29	29	-29	0	29	29	-29	0
					1 psi	70	70	-70	0	70	70	-70	0	70	70	-70	0	70	70	-70	0
				VCE (isolation failure, ventilation failure)	5 psi	47	47	-47	0	47	47	-47	0	47	47	-47	0	47	47	-47	0
					3 psi	65	65	-65	0	65	65	-65	0	65	65	-65	0	65	65	-65	0
					1 psi	152	152	-152	0	152	152	-152	0	152	152	-152	0	152	152	-152	0
			Leak	VCE (isolation failure, ventilation success)	5 psi	11	11	-11	0	11	11	-11	0	11	11	-11	0	11	11	-11	0
					3 psi	16	16	-16	0	16	16	-16	0	16	16	-16	0	16	16	-16	0
					1 psi	38	38	-38	0	38	38	-38	0	38	38	-38	0	38	38	-38	0
				VCE (isolation failure, ventilation failure)	5 psi	25	25	-25	0	25	25	-25	0	25	25	-25	0	25	25	-25	0
					3 psi	35	35	-35	0	35	35	-35	0	35	35	-35	0	35	35	-35	0
					1 psi	81	81	-81	0	81	81	-81	0	81	81	-81	0	81	81	-81	0
				VCE (isolation success, ventilation success)	5 psi	8	8	-8	0	8	8	-8	0	8	8	-8	0	8	8	-8	0
					3 psi	12	12	-12	0	12	12	-12	0	12	12	-12	0	12	12	-12	0
					1 psi	28	28	-28	0	28	28	-28	0	28	28	-28	0	28	28	-28	0
				VCE (isolation success, ventilation failure)	5 psi	15	15	-15	0	15	15	-15	0	15	15	-15	0	15	15	-15	0
					3 psi	22	22	-22	0	22	22	-22	0	22	22	-22	0	22	22	-22	0
					1 psi	51	51	-51	0	51	51	-51	0	51	51	-51	0	51	51	-51	0
24	P07	GLP collection tank	Rupture	VCE (isolation failure, ventilation success)	5 psi	42	42	-42	0	42	42	-42	0	42	42	-42	0	42	42	-42	0
					3 psi	59	59	-59	0	59	59	-59	0	59	59	-59	0	59	59	-59	0
					1 psi	138	138	-138	0	138	138	-138	0	138	138	-138	0	138	138	-138	0
				VCE (isolation failure, ventilation failure)	5 psi	57	57	-57	0	57	57	-57	0	57	57	-57	0	57	57	-57	0
					3 psi	79	79	-79	0	79	79	-79	0	79	79	-79	0	79	79	-79	0
					1 psi	184	184	-184	0	184	184	-184	0	184	184	-184	0	184	184	-184	0
			Leak	VCE (isolation failure, ventilation success)	5 psi	19	19	-19	0	19	19	-19	0	19	19	-19	0	19	19	-19	0
					3 psi	26	26	-26	0	26	26	-26	0	26	26	-26	0	26	26	-26	0
					1 psi	62	62	-62	0	62	62	-62	0	62	62	-62	0	62	62	-62	0
				VCE (isolation failure, ventilation failure)	5 psi	39	39	-39	0	39	39	-39	0	39	39	-39	0	39	39	-39	0
					3 psi	54	54	-54	0	54	54	-54	0	54	54	-54	0	54	54	-54	0
					1 psi	127	127	-127	0	127	127	-127	0	127	127	-127	0	127	127	-127	0
				VCE (isolation success, ventilation success)	5 psi	12	12	-12	0	12	12	-12	0	12	12	-12	0	12	12	-12	0
					3 psi	17	17	-17	0	17	17	-17	0	17	17	-17	0	17	17	-17	0
					1 psi	40	40	-40	0	40	40	-40	0	40	40	-40	0	40	40	-40	0
				VCE (isolation success, ventilation failure)	5 psi	23	23	-23	0	23	23	-23	0	23	23	-23	0	23	23	-23	0
					3 psi	31	31	-31	0	31	31	-31	0	31	31	-31	0	31	31	-31	0
					1 psi	74	74	-74	0	74	74	-74	0	74	74	-74	0	74	74	-74	0
25	P08	Acidulation tank	Rupture	VCE (isolation failure, ventilation success)	5 psi	27	27	-27	0	27	27	-27	0	27	27	-27	0	27	27	-27	0
					3 psi	37	37	-37	0	37	37	-37	0	37	37	-37	0	37	37	-37	0
					1 psi	88	88	-88	0	88	88	-88	0	88	88	-88	0	88	88	-88	0
				VCE (isolation failure, ventilation failure)	5 psi	57	57	-57	0	57	57	-57	0	57	57	-57	0	57	57	-57	0
					3 psi	79	79	-79	0	79	79	-79	0	79	79	-79	0	79	79	-79	0
					1 psi	184	184	-184	0	184	184	-184	0	184	184	-184	0	184	184	-184	0
			Leak	VCE (isolation failure, ventilation success)	5 psi	17	17	-17	0	17	17	-17	0	17	17	-17	0	17	17	-17	0
					3 psi	23	23	-23	0	23	23	-23	0	23	23	-23	0	23	23	-23	0
					1 psi	55	55	-55	0	55	55	-55	0	55	55	-55	0	55	55	-55	0
				VCE (isolation failure, ventilation failure)	5 psi	35	35	-35	0	35	35	-35	0	35	35	-35	0	35	35	-35	0
					3 psi	48	48	-48	0	48	48	-48	0	48	48	-48	0	48	48	-48	0
					1 psi	113	113	-113	0	113	113	-113	0	113	113	-113	0	113	113	-113	0
				VCE (isolation success, ventilation success)	5 psi	11	11	-11	0	11	11	-11	0	11	11	-11	0	11	11	-11	0
					3 psi	15	15	-15	0	15	15	-15	0	15	15	-15	0	15	15	-15	0
					1 psi	35	35	-35	0	35	35	-35	0	35	35	-35	0	35	35	-35	0
				VCE (isolation success, ventilation failure)	5 psi	20	20	-20	0	20	20	-20	0	20	20	-20	0	20	20	-20	0
					3 psi	28	28	-28	0	28	28	-28	0	28	28	-28	0	28	28	-28	0
					1 psi	66	66	-66	0	66	66	-66	0	66	66	-66	0	66	66	-66	0
26	P09	FFA buffer tank	Rupture	VCE (isolation failure, ventilation success)	5 psi	27	27	-27	0	27	27	-27	0	27	27	-27	0	27	27	-27	0
					3 psi	38	38	-38	0	38	38	-38	0	38	38	-38	0	38	38	-38	0
					1 psi	89	89	-89	0	89	89	-89	0	89	89	-89	0	89	89	-89	0
				VCE (isolation failure, ventilation failure)	5 psi	57	57	-57	0	57	57	-57	0	57	57	-57	0	57	57	-57	0
					3 psi	79	79	-79	0	79	79	-79	0	79	79	-79	0	79	79	-79	0
					1 psi	184	184	-184	0	184	184	-184	0	184	184	-184	0	184	184	-184	0
			Leak	VCE (isolation failure, ventilation success)	5 psi	18	18	-18	0	18	18	-18	0	18	18	-18	0	18	18	-18	0
					3 psi	25	25	-25	0	25	25	-25	0	25	25	-25	0	25	25	-25	0
					1 psi	58	58	-58	0	58	58	-58	0	58	58	-58	0	58	58	-58	0
				VCE (isolation failure, ventilation failure)	5 psi	37	37	-37	0	37	37	-37	0	37	37	-37	0	37	37	-37	0
					3 psi	51	51	-51	0	51	51	-51	0	51	51	-51	0	51	51	-51	0
					1 psi	119	119	-119	0	119	119	-119	0	119	119	-119	0	119	119	-119	0
				VCE (isolation success, ventilation success)	5 psi	11	11	-11	0	11	11	-11	0	11	11	-11	0	11	11	-11	0
					3 psi	15	15	-15	0	15	15	-15	0	15	15	-15	0	15	15	-15	0
					1 psi	35	35	-35	0	35	35	-35	0	35	35	-35	0	35	35	-35	0
				VCE (isolation success, ventilation failure)	5 psi	20	20	-20	0	20	20	-20	0	20	20	-20	0	20	20	-20	0
					3 psi	28	28	-28	0	28	28	-28	0	28	28	-28	0	28	28	-28	0
					1 psi	66	66	-66	0	66	66	-66	0	66	66	-66	0	66	66	-66	0
27	P10	Neutralization tank	Rupture	VCE (isolation failure, ventilation success)	5 psi	28	28	-28	0	28	28	-28	0	28	28	-28	0	28	28	-28	0
					3 psi	39	39	-39	0	39	39	-39	0	39	39	-39	0	39	39	-39	0
					1 psi	92	92	-92	0	92	92	-92	0	92	92	-92	0	92	92	-92	0
				VCE (isolation failure, ventilation failure)	5 psi	57	57	-57	0	57	57	-57	0	57	57	-57	0	57	57	-57	0
					3 psi	79	79	-79	0	79	79	-79	0	79	79	-79	0	79	79	-79	0
					1 psi	184	184	-184	0	184	184	-184	0	184	184	-184	0	184	184	-184	0
			Leak	VCE (isolation failure, ventilation success)	5 psi	17	17	-17	0	17	17	-17	0	17	17	-17	0	17	17	-17	0
					3 psi	23	23	-23	0	23	23	-23	0	23	23	-23	0	23	23	-23	0
					1 psi	55	55	-55	0	55	55	-55	0	55	55	-55	0	55	55	-55	0
				VCE (isolation failure, ventilation failure)	5 psi	35	35	-35	0	35	35	-35	0	35	35	-35	0	35	35	-35	0
					3 psi	49	49	-49	0	49	49	-49	0	49	49	-49	0	49	49	-49	0
					1 psi	114	114	-114	0	114	114	-114	0	114	114	-114	0	114	114	-114	0
				VCE (isolation success, ventilation success)	5 psi	11	11	-11	0	11	11	-11	0	11	11	-11	0	11	11	-11	0
					3 psi	15	15	-15	0	15	15	-15	0	15	15	-15	0	15	15	-15	0
					1 psi	35	35	-35	0	35	35	-35	0	35	35	-35	0	35	35	-35	0
				VCE (isolation success, ventilation failure)	5 psi	20	20	-20	0	20	20	-20	0	20	20	-20	0	20	20	-20	0
					3 psi	28	28	-28	0	28	28	-28	0	28	28	-28	0	28	28	-28	0
					1 psi	67	67	-67	0	67	67	-67	0	67	67	-67	0	67	67	-67	0
29	M01	Biodiesel/PFAD barge	Small	Pool fire	Flame zone	43	43	-43	0	43	43	-43	0	43	43	-43	0	43	43	-43	0
				(on sea)	Drag zone	46	46	-46	0	47	47	-47	0	47	47	-47	0	52	52	-52	0
			Large	Pool fire	Flame zone	75	75	-75	0	75	75	-75	0	75	75	-75	0	75	75	-75	0
				(on sea)	Drag zone	78	78	-78	0	79	79	-79	0	79	79	-79	0	87	87	-87	0
			Rupture	Pool fire	Flame zone	193	193	-193	0	193	193	-193	0	193	193	-193	0	193	193	-193	0
				(on sea)	Drag zone	196	196	-196	0	196	196	-196	0	196	196	-196	0	210	210	-210	0
30	M02	Methanol barge	Rupture	Tank fire	Flame zone	13	13	-13	0	13	13	-13	0	13	13	-13	0	13	13	-13	0
					Drag zone	16	16	-16	0	16	16	-16	0	16	16	-16	0	17	17	-17	0
31	M03	Methanol ISO-tanker	Leak	Pool fire	Flame zone	12	12	-12	0	12	12	-12	0	12	12	-12	0	11	11	-11	0
					Drag zone	15	15	-15	0	15	15	-15	0	15	15	-15	0	16	16	-16	0
				Flash fire	LFL	13	7	0	9	8	1	0	5	10	2	0	6	0	0	0	0
				Toxic release	190,000 ppm	8	3	0	5	2	1	0	1	4	1	0	2	0	0	0	0
					113,000 ppm	10	5	0	7	6	1	0	3	7	1	0	4	0	0	0	0
					59,000 ppm	14	9	0	10	9	2	1	6	12	2	1	7	1	1	0	1
					31,000 ppm	19	13	0	12	12	2	0	9	15	3	0	11	3	1	0	2
					18,500 ppm	22	17	0	15	14	3	0	11	18	4	0	14	4	3	0	3
			Rupture	Pool fire	Flame zone	35	35	-35	0	35	35	-35	0	35	35	-35	0	35	35	-35	0
					Drag zone	38	38	-38	0	39	39	-39	0	39	39	-39	0	43	43	-43	0
				Flash fire	LFL	14	8	-5	5	10	4	-4	0	10	4	-4	0	11	4	-4	0
				Toxic release	190,000 ppm	4	3	-3	0	4	3	-3	0	4	3	-3	0	4	3	-3	0
					113,000 ppm	5	4	-4	0	5	4	-4	0	5	4	-4	0	5	4	-4	0
					59,000 ppm	17	10	-7	5	14	5	-5	2	14	5	-5	2	18	5	-5	2
					31,000 ppm	25	18	-10	8	30	8	-6	10	30	9	-6	10	34	7	-6	12
					18,500 ppm	35	23	-15	10	38	12	-7	22	38	14	-7	22	43	10	-7	25
32	L01	Unloading hose/line from	Leak	Pool fire	Flame zone	75	75	-75	0	75	75	-75	0	75	75	-75	0	75	75	-75	0
		jetty (biodiesel/PFAD)		(on sea)	Drag zone	78	78	-78	0	79	79	-79	0	79	79	-79	0	87	87	-87	0
			Rupture	Pool fire	Flame zone	75	75	-75	0	75	75	-75	0	75	75	-75	0	75	75	-75	0
				(on sea)	Drag zone	78	78	-78	0	79	79	-79	0	79	79	-79	0	87	87	-87	0
33	L02	Unloading hose/line from	Leak	Pool fire	Flame zone	75	75	-75	0	75	75	-75	0	75	75	-75	0	75	75	-75	0
		jetty (methanol)		(on sea)	Drag zone	78	78	-78	0	79	79	-79	0	79	79	-79	0	87	87	-87	0
				Flash fire	LFL	47	26	5	31	19	18	5	14	21	21	5	14	18	11	5	15
				Toxic release	190,000 ppm	9	1	6	8	9	1	6	8	9	1	6	8	9	1	6	8
					113,000 ppm	15	10	5	10	9	1	6	8	9	1	6	8	9	1	6	8
					59,000 ppm	58	40	5	40	23	23	5	14	26	26	5	14	22	20	5	15
					31,000 ppm	84	80	5	57	41	38	5	14	47	40	5	15	37	32	5	15
					18,500 ppm	105	111	5	74	51	46	5	14	63	48	5	13	48	40	5	15
			Rupture	Pool fire	Flame zone	75	75	-75	0	75	75	-75	0	75	75	-75	0	75	75	-75	0
				(on sea)	Drag zone	78	78	-78	0	79	79	-79	0	79	79	-79	0	87	87	-87	0
				Flash fire	LFL	53	20	5	15	24	16	5	16	27	17	5	16	24	13	5	18
				Toxic release	190,000 ppm	12	1	6	10	12	1	6	10	12	1	6	10	12	1	6	10
					113,000 ppm	17	9	5	12	12	1	5	10	12	1	5	10	12	1	5	10
					59,000 ppm	70	28	5	49	31	21	5	16	33	22	5	16	31	18	5	18
					31,000 ppm	102	66	5	75	56	33	5	16	63	33	5	16	55	29	5	17
					18,500 ppm	128	100	5	93	78	40	5	15	85	40	5	15	70	35	5	17
34	RT1	Road tanker for fats & oils	Leak	Pool fire	Flame zone	21	21	-21	0	21	21	-21	0	21	21	-21	0	21	21	-21	0
					Drag zone	24	24	-24	0	24	24	-24	0	24	24	-24	0	27	27	-27	0
			Rupture	Pool fire	Flame zone	24	24	-24	0	24	24	-24	0	24	24	-24	0	24	24	-24	0
					Drag zone	27	27	-27	0	27	27	-27	0	27	27	-27	0	30	30	-30	0
35	L03	Road tanker unloading line	Leak	Pool fire	Flame zone	39	39	-39	0	39	39	-39	0	39	39	-39	0	39	39	-39	0
					Drag zone	42	42	-42	0	43	43	-43	0	43	43	-43	0	47	47	-47	0
			Rupture	Pool fire	Flame zone	39	39	-39	0	39	39	-39	0	39	39	-39	0	39	39	-39	0
					Drag zone	42	42	-42	0	43	43	-43	0	43	43	-43	0	47	47	-47	0

8.8 Risk Results

This section presents the risk results which are derived by combining the frequency of hazardous outcome events with the associated consequences. This is known as Risk Summation and was conducted using in-house software RISKPLOT^TM.

Societal Risks

The overall off-site Potential Loss of Life (PLL) value from the biodiesel plant was estimated as 6.83x10^-7 per year for the existing population, increasing to 1.59x10^-5 per year with the future population, once the industrial estate is fully developed. The main contributors to the societal risks are summarised in Table 8.8a and Table 8.8b. Vapour cloud explosions within the process building are the main contributors to the societal risk. Pool fires from tank leaks and unloading activities also make some contribution. The toxic effects of methanol are minimal and do not make significant contributions to the risks. Similarly, flash fires are only small contributors to the overall risk because of small dispersion distances.

Table 8.8a Main Contributors to Potential Loss of Life (PLL)

Current Population Scenario

Scenario	Scenario Description	Main Hazardous Chemical	PLL (per year)	%
P03 Pipe reactor	Leak leading to VCE (isolation success, ventilation failure)	Methanol	1.64 x10^-7	24.0%
P03 Pipe reactor	Leak leading to VCE (isolation failure, ventilation success)	Methanol	7.59 x10^-8	11.1%
P03 Pipe reactor	Leak leading to VCE (isolation success, ventilation success)	Methanol	5.99 x10^-8	8.8%
P03 Pipe reactor	Leak leading to VCE (isolation failure, ventilation failure)	Methanol	4.55 x10^-8	6.7%
M01 biodiesel/PFAD barge	Rupture leading to pool fire	Biodiesel	3.67 x10^-8	5.4%
P01 Esterification reactor with catalyst	Rupture leading to VCE (isolation failure, ventilation success)	Methanol	3.57 x10^-8	5.2%
P03 Pipe reactor	Rupture leading to VCE (isolation success, ventilation failure)	Methanol	3.27 x10^-8	4.8%
T10 Methanol tank, T7 PFAD storage tank	Tank rupture leading to pool fire with overtopping (VCE escalation)	Methanol, PFAD	2.35 x10^-8	3.4%
P01 Esterification reactor	Rupture leading to VCE (isolation failure, ventilation failure)	Methanol	2.08 x10^-8	3.0%
P10 Neutralization tank	P10 rupture leading to VCE (isolation failure, ventilation failure)	Methanol	2.08 x10^-8	3.0%
Total (including all scenarios)			6.83 x10^-7	100%

Future Population Scenario

Scenario	Scenario Description	Main Hazardous Chemical	PLL (per year)	%
P03 Pipe reactor	Leak leading to VCE (isolation success, ventilation failure)	Methanol	2.97 x10^-6	18.7%
Maintenance error	Internal explosion in process building	Methanol	2.16 x10^-6 ^(a)	13.6%
T10 Methanol tank, T7 PFAD storage tank	Tank rupture leading to pool fire with overtopping (VCE escalation)	Methanol, PFAD	1.71 x10^-6	10.8%
T10 Methanol storage tank	Rupture leading to pool fire with overtopping (VCE escalation)	Methanol	1.08 x10^-6	6.8%
P03 Pipe reactor	Leak leading to VCE (isolation failure, ventilation failure)	Methanol	8.25 x10^-7	5.2%
P03 Pipe reactor	Leak leading to VCE (isolation failure, ventilation success)	Methanol	8.18 x10^-7	5.1%
P03 Pipe reactor	Leak leading to VCE (isolation success, ventilation success)	Methanol	5.98 x10^-7	3.8%
P03 Pipe reactor	Rupture leading to VCE (isolation success, ventilation failure)	Methanol	5.94 x10^-7	3.7%
P01 Esterification reactor with catalyst	Rupture leading to VCE (isolation failure, ventilation success)	Methanol	5.47 x10^-7	3.4%
P01 Esterification reactor with catalyst	Rupture leading to VCE (isolation failure, ventilation failure)	Methanol	3.77 x10^-7	2.4%
Total (including all scenarios)			1.59x10^-5	100%
Note: (a) This figure is double the frequency value derived from the event tree in Figure 8.6l because two fatalities have been assumed (see Page 46).

Table 8.8b PLL Breakdown by Systems (Future Population)

System	PLL (per year)	%
Process	1.14 x 10^-5	72%
Tank Farm	3.59 x 10^-6	22%
Transport	8.76 x 10^-7	6%
Total	1.59 x 10^-5	100%

The FN curve for current population and future population are presented and compared with the Hong Kong Risk Guidelines in Figure 8.8a and Figure 8.8b.

For the current population scenario, there is only one office building of Gammon with a population of about 50 that lies within the consequence zone of 1 psi. This results in fractional fatalities as the fatality probability is 1% of persons indoor within the 1psi zone. For the current population scenario, the marine population is also affected by pool fire events from loading/ unloading operations. This scenario also results in fractional fatalities as the population is distributed over an area in Junk Bay. These outcomes with fractional fatalities have been included in the PLL estimation. These scenarios have also been represented in the FN curve by converting the PLL into an equivalent frequency of one fatality and the same is shown on the FN curve in Figure 8.8a.

Considering the potential future development in the vicinity of this plant (although of industrial in nature), the FN curve for future case is higher than the current case. Also, the predicted fatality range is wider.

The risks for both current and future population case lie within the acceptable region of the Hong Kong Risk Guidelines.

Figure 8.8a FN Curves for Current and Future Population and their Comparison with the HK Societal Risk Guidelines

Figure 8.8b FN Curves of Future Population and their Comparison with the HK Societal Risk Guidelines (Breakdown by Operational Areas)

Individual Risk

The individual risk contours for the biodiesel plant are shown in Figure 8.8c. These contours express the risk to a hypothetical individual present outdoors 100% of the time. As seen from the figure, the 10^-5 per year contour does not extend beyond plant boundary. Individual risk from the plant is therefore considered acceptably low and complies with the Hong Kong individual risk guidelines which state that the individual risk off-site should not exceed 10^-5 per year.

Figure 8.8c Individual Risk Contours for Biodiesel Plant

10^-5 per year

10^-6 per year

Cumulative Risk

The risks presented in Figure 8.8c represent the risks due to the biodiesel facility. The neighbouring Hong Kong Oxygen facility, some 400m away, also contains hazardous equipment and storage of compressed gases that may add to the risks from the biodiesel facility. Although no details of the Hong Kong Oxygen facility are available, it is expected that the main risks will originate from the storage of a large number of compressed gas cylinders. Consequence distances associated with such equipment have a very limited range because the inventory within each cylinder is small. Hence it may be expected that the risks will decrease rapidly with distance from the facility. The risk levels at the biodiesel facility, being 400m away, will be negligible. The scenario of projectiles from the Hong Kong Oxygen facility was addressed in this QRA and found to present insignificant incremental risks to the biodiesel facility.

Similarly, as seen from Figure 8.8c, the risk of hazards from the proposed biodiesel plant impacting on the Hong Kong Oxygen facility is insignificant. The risks decrease quickly offsite.

8.9 Mitigation Measures Assumed in the Base case

The following assumptions have been made in the base case analysis and hence these will require taking forward during the design and operations and need confirming:

· The process plant building will be provided with adequate number of gas detectors distributed over the various areas of potential leak sources to provide adequate coverage. A coverage factor of 90% for 1 out of N detectors for alarm to be ensured (i.e. the system will be designed so that at least one detector (out of the N detectors provided) triggers in 90% of occasions when a high concentration of flammables is present).

· Additional leak detection systems based on process parameters will be considered such as low pressure or others as applicable.

· Upon gas detection, the process system will be isolated. All pumps, motors will be stopped. Also, emergency shutdown valves will be provided at the liquid outlet connections of major equipments holding significant inventory of methanol (>5m3). Emergency shutdown system to meet a performance target of 90% for the reliability of the overall shutdown system

· Emergency ventilation system will be provided in accordance with relevant design codes for adequate ventilation of process areas inside buildings, to ensure that the ventilation rate is sufficient to bring down the concentration to 50% of lower explosive limit (LEL). Also, the ventilation system will be designed to avoid any stagnant pockets. Relevant codes include IP 15 : Area Classification Code for Installations Handling Flammable Liquids, IEC 60079, Part 10 :2002 Electrical Apparatus for Explosive Gas Atmospheres and NFPA 30 : The Flammable and Combustible Liquids Code. A performance target of 90% for the reliability of the ventilation system is to be achieved

· All electrical equipment inside the building will be classified in accordance with the electrical area classification requirements. No unclassified electrical equipment will be used during operations or maintenance.

· Reference will also be made to codes of practice and guidance issued in Europe that apply to places where explosive atmospheres may occur (called ‘ATEX’ requirements). These are covered as part of the European Directive: the Explosive Atmospheres Directive (99/92/EC) and the UK regulations, Dangerous Substances and Explosive Atmospheres Regulations 2002 (DSEAR). Where potentially explosive atmospheres may occur in the workplace, the requirements include, identifying and classifying (zoning) areas where potentially explosive atmospheres may occur; avoiding ignition sources in zoned areas, in particular those from electrical and mechanical equipment; where necessary, identifying the entrances to zoned areas; providing appropriate anti-static clothing for employees; and before they come into operation, verifying the overall explosion protection safety of areas where explosive atmospheres may occur. The code of practice and guidance cover among others Control and mitigation measures, Safe maintenance, repair and cleaning procedures.

· If there are any openings from the building, the near vicinity outside of such openings should also be meet with the area classification requirements as per the relevant code example IP 15.

· Gas detection in methanol storage area and emergency shutdown system on liquid inlet and outlet piping of methanol storage tank including automatic shutdown on high level will be provided.

· Online oxygen analysers will be provided in the closed vent system of process equipment located inside the process building to detect any air ingress into equipment due to a maintenance activity or during normal operation, for example due to nitrogen blanketing failure. Appropriate control and shutdown actions on high oxygen alarms will be designed as required. Also, portable gas analyzers will be used to test the internal atmosphere of process equipment after completion of maintenance.

· A preliminary process hazard analysis has been carried out as part of the basic design. Also, as part of this QRA study, a detailed hazard identification has been carried out. Further review of design safety measures will be performed as the design process continues, using a structured hazard identification process such as Hazard and Operability Study.

· The reliability requirements for process safety interlock systems will be determined following a structured process such as Safety Integrity Level determination and verification studies as per IEC 60508 and 60511.

· Safety Management Systems and Procedures will be developed to cover all aspects of operations and maintenance including safe handling of chemicals, safe operating and maintenance practices, operator training, employment of suitably qualified personnel with relevant process industry experience, period audit and review of the safety management systems and procedures etc. The systems and procedures will be in line with the best practices in the process industry and also reflect the high standards adopted in Hong Kong by companies operating in the LPG, flammable liquid and gas industry.

· In the event of any maintenance activity inside a building, procedures will be developed to ensure that flammable concentration build-up does not occur due to draining, opening of vessel or piping etc. The start-up and maintenance operations will be supervised and checked independently of the person undertaking such tasks, to provide a greater oversight.

· Also, the reliability of the nitrogen blanketing system will be ensured to minimise failures in the blanketing system leading to potential vapour releases from enclosed equipment inside the building. All vents from the process equipments inside the building will be routed to a safe location outside the building.

· Detailed emergency response plans will be developed to handle any impacts onsite and offsite due to any incident at the facility during loading/ unloading operations, transfer operations, storage tank farm operations and processing operations.

8.10 Further Mitigation Measures

Since both the individual and societal risks posed by the biodiesel plant to off-site population meet the criteria of Hong Kong Risk Guidelines, no further mitigation measures are required. Nevertheless, further voluntary risk reduction measures may be considered during the detailed design. As an example, consider:

· To minimise the spill vaporisation from a liquid pool, by a spill containment system which may be designed such that spills can be diverted and collected outside of the process building

8.11 Conclusions

The potential risks from the proposed biodiesel plant were assessed using a standard risk assessment methodology. The results show that both the societal and individual risks evaluated for the plant meet the acceptable criteria of the Hong Kong Risk Guidelines.

---

([1]) Engineering Safety and Risk, Environmental Assessment Services for Permanent Aviation Fuel Facility, Environmental Impact Assessment Report, Feb 2007.

([2]) Code of Practice for Oil Storage Installations 1992, Building Authority, Hong Kong.

([3]) The Annual Traffic Census 2007, Transport Department, Hong Kong SAR.

([4]) Lees, F. P., Loss Prevention in the Process Industries, Second Edition, 1996.

([5]) Note that no ammonium- or other nitrogen-containing byproducts will be produced. There is no fire or explosion risk from this part of the process.

([6]) http://www.ethanolmarketplace.com/plant/list/3

([7]) http://www.biodieselmagazine.com/article.jsp?article_id=1779&q=number%20plants&category_id=29

([8]) http://biodieselmagazine.com/plantmap/

([9]) See, for example, www.biodieselmagazine.com/article.jsp?article_id=1127, where a consultant design engineer, Rocky Costello, comments: that safety is most likely to be compromised in smaller, startup plants in which entrepreneurs are trying to get into the industry at minimal costs.

[10] http://www.ktvb.com/news/localnews/stories/ktvbn-jul1406-explosion_cause.113ae8b1.html

([11]) U.S. Chemical Safety and Hazard Investigation Board, “Confined Vapor Cloud Explosion”, Report No. 2007-03-I-MA, p. 68 (2008).

([12]) Hazards of Oil Refining Distillation Units, BP Process Safety Series, Institution of Chemical Engineers, 2008, p.4.

([13]) http://www.biodieselmagazine.com/plantmap/

([14]) Duguid, I. M. “Analysis of past incidents in the oil, chemical and petrochemical industries,” Loss Prevention Bulletin (IChemE), issue 142 (1998), p.3-6.

([15]) Engineering Safety & Risk, Environmental Assessment Services for Permanent Aviation Fuel Facility – Environmental Impact Assessment Report, Feb 2007, section 10.6.2.5

([16]) According to a recent Hong Kong Civil Service report (www.csb.gov.hk/hkgcsb/doclib/showcasing_fsd_e.pdf), 94% of fire emergency calls to urban locations are attended within 6 minutes, while the attendance time is 9 to 23 minutes for isolated locations. Thus, we may conservatively adopt an expected response time of 10 minutes for this location.

([17]) GEO, Seismic hazard analysis of the Hong Kong region, GEO Report No. 65, Geotechnical Engineering Office, Government of the Hong Kong SAR, 2002.

([18]) GCO, Review of earthquake data for the Hong Kong region, GCO Publication No. 1/91, Civil Engineering Services Dept., Hong Kong Government, 1991.

([19]) Byrne, J. P., The calculation of aircraft crash risk in the UK, Health and Safety Executive, HSE\R150, 1997.

([20]) www.ntsb.gov/aviation/table6.htm

([21]) Annual review of aircraft accident data: US General Aviation, Calendar year 2001, National Transport Safety Board.

([22]) Civil Aviation Department, Facts and Statistics, Government of Hong Kong SAR.

([23]) Byrne, J. P., The calculation of aircraft crash risk in the UK, Health and Safety Executive report HSE\R150, 1997.

([24]) www.epd.gov.hk/eia/register/report/eiareport/eia_1252006/html/eiareport/Part2/Section13/Sec2_13AnnexA5.htm

([25]) Lee, B. Y., Report of Hong Kong in the international tsunami seminar in the western pacific region, Tokyo, Japan, 7-12 March 1988.

([26]) Crowl, A. D. and Louvar, J. F., Chemical Process Safety: Fundamentals with Applications, 2nd ed., Prentice Hall, p. 16 (2002).

([27]) www.law.umaryland.edu/marshall/crsreports/crsdocuments/RL31530B_02142005.pdf )

([28]) Lees, F. P., Loss prevention in the Process Industries, Second Edition, 1996

([29]) Guidelines for Quantitative Risk Assessment, The “Purple Book”, Committee for the prevention of disasters, first edition, 1999

([30]) http://en.wikipedia.org/wiki/Methanol

([31]) Engineering Safety and Risk, Environmental Assessment Services for Permanent Aviation Fuel Facility, Environmental Impact Assessment Report, Feb 2007.

([32]) Cox, Lees and Ang, “Classification of Hazardous Locations”, IChemE.

([33]) Engineering Safety & Risk, Environmental Assessment Services for Permanent Aviation Fuel Facility – Environmental Impact Assessment Report, Feb 2007, Table 10.5 and section 10.3.3, page 10-21.

([34]) Engineering Safety & Risk, Environmental Assessment Services for Permanent Aviation Fuel Facility – Environmental Impact Assessment Report, Feb 2007, Table 10.4, page 10.20. The leak sizes are derived from the values shown in the PAFF report (0.3%, 1%, 7%) multiplied by 15 (because the Biodiesel plant barges will have only a single compartment instead of 15 compartments for the PAFF). The size probabilities are taken directly from the PAFF report, with the ‘rupture’ and ‘multiple rupture’ cases combined to give 60% for ‘rupture’.

([35]) Ite, P. J., Fitzpatrick, R. D., and Hurst, N. W., Risk assessment for the siting and developments near liquefied petroleum installations, IChemE Symposium Series No. 110, 1988.

([36]) Lees, F. P., Loss prevention in the Process Industries, Second Edition, 1996.

([37]) Ite, P. J., Fitzpatrick, R. D., and Hurst, N. W., Risk assessment for the siting and developments near liquefied petroleum installations, IChemE Symposium Series No. 110, 1988.

([38]) Blything, K. W. & Reeves, A. B., An initial prediction of the BLEVE frequency of a 100 tonne butane storage vessel, SRD report R488, August 1988.

([39]) TNO, Guidelines for Quantitative Risk Assessment, the “Purple Book”, Report CPR 18E, The Netherlands Organisation of Applied Scientific Research, Voorburg, 1999.

([40]) Engineering Safety & Risk, Environmental Assessment Services for Permanent Aviation Fuel Facility – Environmental Impact Assessment Report, Feb 2007, sections 10.5.4 and 10.5.7.2.

([41]) http://users.wpi.edu/~ierardi/FireTools/pool_fires.html. Transformer oil is used as a proxy for biodiesel in this calculation, as it has a similar flash point to biodiesel.

([42]) Committee for the Prevention of Disasters, Guidance for Quantitative Risk Assessment, the “Purple Book”, first edition, 1999.

([43]) Lees, F. P., Loss Prevention in The Process Industries, Second edition, 1996.

([44]) Center for Chemical Process Safety (CCPS) of the American Institute of Chemical Engineers, Guidelines for Consequence Analysis of Chemical Releases, p.139-140, 1999.

([45]) Thyer, A. M., Hirst, I. L. and Jagger, S. F., Bund overtopping – the consequence of catastrophic tank failure, Journal of Loss Prevention in the Process Industries, Vol. 15, p.357-363, 2002.

([46]) Greenspan, H.P., and Young, R.E. Flow over a Containment Dyke. J. Fluid Mechanics, Vol. 87, p.179-192 (1978).

([47]) Estimated value based on extrapolation from published data at 20 C and 25 C.

([48]) Engineering Safety and Risk, Environmental Assessment Services for Permanent Aviation Fuel Facility, Environmental Impact Assessment Report, Feb 2007.

([49]) Development of pool fire thermal radiation model, HSE contract research report No. 96/1996, p39-41.

([50]) Daniel A Crowl, Understanding explosions, centre for chemical process (CCPS)

([51]) GAME: development of guidance for the application of the multi-energy method, Health & Safety Executive Contract Research Report 202/1998

([52]) Chemical Industry Association, Guidance for the location and design of occupied buildings on chemical manufacturing sites, CIA/CISHEC/9802/CP/500/2M, Feb 1998.

([53]) Lees, F. P., Loss Prevention in The Process Industries, Second edition, 1996.

([54]) Buncefield Major Incident Investigation, Initial Report to the Health and Safety Commission and the Environment Agency of the investigation into the explosions and fires at the Buncefield oil storage and transfer depot, Hemel Hempstead, on 11 December 2005, Buncefield Major Incident Investigation Board.

([55]) Engineering Safety and Risk, Environmental Assessment Services for Permanent Aviation Fuel Facility, Environmental Impact Assessment Report, Feb 2007.

([56]) Hong Kong Observatory, private communication.

([57]) Engineering Safety & Risk, Environmental Assessment Services for Permanent Aviation Fuel Facility – Environmental Impact Assessment Report, Feb 2007, section 10.2.6.2, page 10.17.

([58]) Hong Kong Water Supplies Department, Reassessment of Chlorine Hazard for Eight Existing Water Treatment Works: Methodology Report, Sept 1997.

([59]) http://users.wpi.edu/~ierardi/FireTools/pool_fires.html, assuming burning liquid is equivalent to transformer oil (flash point = 140degC, similar to biodiesel).