Severity and duration

Ammerman, et al. [5] found that the casualty data derived from 15 years (1979 through 1993) of Lloyd’s casualty data and 2 years (1988, 1993) of Lloyd’s port call data contained very little information about accident severity. Their analyses were carried out using all fires in the 15 years of data, without regard to their severity.

Selway [3], after having established estimates for the frequency of a “severe fire” on the INF 2 class freight ferry, Nord Pas-de-Calais, sought to obtain estimates of the fire durations, with particular reference to Ro-Ro ferries.

Only two of a number of organizations dealing with shipping accidents were able to provide specific information on the duration of fires on ships. These were the IMO and the Marine Accident Investigation Branch (MAIB).

The IMO was found to have the largest number of fire reports which gave times to control, and times to extinguish, fires on ships. Over a period of 25 years, IMO has received a total of 382 fire casualty records reporting on ship fires from all over the world. The shortest fire recorded was extinguished in one minute and the longest in seventy-one days. These data produced an estimated average time for a fire of 26 hours while the ship was in port and 19 hours while under way. However, with such a range of duration times, these average figures are only of mathematical interest and could not provide any reliable indication of the time for which cargo in any particular position could have been exposed to fire of any defined severity.

Having thoroughly investigated all the available reports of fires, it was concluded that at present, there is insufficient historical data to reach a definitive conclusion on the time period that a fire on a ship would be considered to be intense. Calculation approaches or practical tests must be resorted to, in order to quantify ship fire durations and in turn to relate these to fire severities for particular fire scenarios.

5.2.2.1. Severity and duration of fire on the INF 2 ship Nord Pas-de-Calais

Using fire modelling techniques, Selway, et al. [11] investigated the growth of fires initiating on the rail deck and in the engine room of the Nord Pas-de-Calais in order to establish the likely temperatures to which a flask of irradiated fuel might be subjected.

To determine the type and size of fire on the rail deck, a study was undertaken of the imported cargo inventories which the Nord Pas-de-Calais carried recently. This established that of the eight wagons that could surround the flask, two would likely contain flammable commodities (e.g. timber, chipboard, plastic tubes), four would likely contain non-flammable commodities (e.g. ash slag, steel tubes, mineral water), and the remaining two would likely be empty.

The HAZARD I computer code, developed by the National Institute of Standards and Technology in the USA, was used to model three fire scenarios on the rail deck. From the three fire scenarios considered for the Rail Deck, the restrictions on the ventilation, as a result of the enclosed deck, produced peak temperatures in the upper gas layer of 450°C after twenty minutes. The depth of this upper layer was about 5 m so it would encompass the flask, but heat transfer could not induce a temperature greater than 450°C on the surface of the flask.

The four fire scenarios modelled in the engine room gave a wide range of temperatures, again being dependent on the ventilation, which varied depending on whether the ventilation dampers and/or the fire doors were operated to close off the area. The scenarios involving the fire doors closing, whether the ventilation dampers closed or not, resulted in ceiling temperatures of no more than 160°C. The scenario involving a fire in the engine room with the end fire door staying open achieved a ceiling temperature of 400°C after an average fire duration of 2¹/2 hours.

The sensitivity analysis of this final engine room fire scenario extended the fire to a duration of 11 hours and found that the ceiling temperature reached a maximum of 440°C after 8 hours.

This temperature is well below that at which the integrity of the engine room ceiling would be considered to be threatened.

In none of these seven fire scenarios would the flask be exposed to conditions more severe than those specified in the IAEA Regulatory Thermal Test for a Type B package.

5.2.2.2. Severity and duration of fire — experimental measurements

Ammerman, et al. [5] conducted eight practical fire tests aboard the US Coast Guard fire test ship Mayo Lykes at Mobile, Alabama. Tests aboard this break-bulk type cargo ship consisted of heptane spray fires simulating engine room and galley fires, wood crib fires simulating cargo hold fires, and pool fires staged for comparison to land-based regulatory fire results.

Primary instrumentation for the tests consisted of two pipe calorimeters that simulated a typical package shape for radioactive material packages (though of much smaller thermal capacity than that typical of heavily shielded flasks used to carry spent nuclear fuel or high level radioactive waste). These fire tests were then modelled with the methods of computational fluid mechanics to confirm that analytical models can successfully predict the shipboard fire environment.

Two holds, holds 4 and 5 at the aft end of the ship, were selected for the tests. Level 1 of these holds, immediately below the weather deck, was used for all fires and measurements. In all cases the fires were set in hold 4. Steel pipe calorimeters representing simulated radioactive material packages were placed in both holds 4 and 5. Fires included ignited heptane sprays impinging on the steel bulkhead between holds 4 and 5, and wood crib fires representing combustible cargo fires. The general experimental arrangement is shown in Figure 1.

The sequence of eight fires conducted aboard the Mayo Lykes is shown in Table XVIII. A brief description of each type of fire and major fire characteristics follows. Hold 4 measured 17.6 m wide by 21 m long by 3.8 m high. Hold 5 dimensions were 17.6 m wide by 16 m long by 3.8 m high.

For all tests the calorimeter in hold 5 was located with its centreline 0.4 m above the deck and 2 m aft of the hold 5-4 bulkhead. To avoid potentially explosive conditions with the heptane spray and in-hold pool fires, adequate oxygen was supplied to hold 4 via openings in the hull.

Measurements indicate that oxygen levels in the vicinity of the fire were usually near normal atmospheric content.

In sealed ship hold fires at sea, oxygen would be more limited, leading to smouldering fires with even lower heat flux levels than experimentally measured. The experimental fires reported here represent conditions more typical of a fire that could occur during ship loading or unloading in port, where both on ship and off ship fire fighting equipment and personnel

Fire Location Bulkhead

Calorimeter

Calorimeter

HOLD #4 HOLD #5

Fire Types:

Heptane Spray (Engine room fire) Wood Crib (Cargo fire)

Fig. 1. Fire test arrangement.

For comparison to the in-hold fire test, a 3 m × 3 m pool was built on the weather deck of the Mayo Lykes on the port side amidships. The pool was constructed to closely follow the dimen-sions of the pool built in hold 4. The calorimeter from hold 5 was centred above the pool, 1 m above the fuel surface at the start of the test. A depth of 13 cm of diesel fuel gave a 32 minute burn, typical of a regulatory pool fire. Calculation of the recession rate for this fire led to an estimated average thermal output of 18.8 MW, i.e. somewhat higher than the thermal output of the in-hold pool fire test (15.7 MW).

TABLE XVIII. FIRE TEST SEQUENCE

Test No. Date, time and duration Type of test Peak thermal power, MW 5037 95/9/12, 2:09 PM CDT, 60 min Two-burner heptane spray test 2.2 5040 95/9/14, 9:13 AM CDT, 20 min Wood crib fire test with 17 L heptane

accelerant 4.1

5041 95/9/14, 12:21 PM CDT, 60 min Two-burner heptane spray test with diesel fuel in drip pans for smoke

2.2

5043 95/9/15, 8:26 AM CDT, 20 min Wood crib fire test with 17 L heptane accelerant

4.1

5045 95/11/13, 12:02 PM CDT, 60 min Four-burner heptane spray test 5.6 5046 95/11/13, 2:46 PM CDT, 60 min Four-burner heptane spray test with

diesel fuel in drip pans for smoke

5.6

5048 95/11/14, 3:09 PM CDT, 27 min Diesel pool fire in hold 4 15.7 5049 95/11/15, 2:20 PM CDT, 32 min Diesel pool fire on weather deck 18.8

From this experimental work, heat flux measurements as a function of time were measured and may be compared with the heat flux implied by the IAEA fire test (65 kW/m²). For heptane and diesel pool fire tests, it is not possible to obtain any direct information on fire duration, since in these tests there was control over fuel supply such that the fire could be terminated at will. However, in the case of the diesel pool fires, information on the pool level recession rate is given by the Ammerman, et al. paper, which may be used to estimate fire duration for any particular case in which total fuel availability in the vicinity of a flask is known.

In the case of the wood crib tests, fire duration was limited by the available combustible material in the 20-A size crib design specified in UL Standard 711, to approximately 20 minutes.

In summary, heat fluxes and temperatures measured during the tests were as follows:

Four-burner heptane spray tests:

Calorimeter in hold adjacent to fire: Max temp. rise above ambient temp.: ~25°C at 70 min

Heat flux <0.8 kW/m²

Wood crib tests:

Calorimeter in same hold as fire: Max temp rise above ambient temp ~215°C at 30 min

Heat flux (max.) 22 kW/m². Peak at ~5 min at worst position on calorimeter

Heat flux (min.) ~17 kW/m² over period ~20 min at worst position on calorimeter

Pool fire test in hold:

Calorimeter in same hold as fire: Max temp. rise above ambient ~800°C at 24 minutes Heat flux (peak) 190 kW/m². Peak at ~1.5 minute at worst position on

calorimeter

Heat flux (max., discounting peak) ~80–100 kW/m² at ~2.5 minutes at various positions on calorimeter, falling quasi inverse exponentially Heat flux (min.) ~25 kW/m² at ~24 minutes at various positions on

calorimeter

Flame temperature ~1000°C at 2 minutes, falling to ~900°C at 24 minutes

Flame temperature ~1000°C at 2 minutes, rising to ~1100°C at 24 minutes

Pool fire test in hold:

Calorimeter in hold adjacent to fire: Heat flux = 0 kW/m² at 0 min, rising quasi-linearly to ~1.5 kW/m² at worst position on calorimeter at 24 minutes

5.3. Assessment of risk