J. Vikelsùe, E. Johansen / Chemosphere 40 (2000) 165±175
167
other parts of the TOXFIRE project (Jùrgensen and
Smith-Hansen, 1995; Markert, 1996; Hietaniemi et al.,
chosen for technical, economical and safety reasons,
and because they are investigated in other parts of the
TOXFIRE project (Markert, 1996; Hietaniemi et al.,
1997; Lonnermark and Babrauskas, 1997; Blomqvist et
al., 1996; Lonnermark et al., 1999). Technical chemicals
1997; Lonnermark and Babrauskas, 1997; Blomqvist et
were investigated, to simulate products stored in ware-
houses.
al., 1996; Lonnermark et al., 1999). The medium labo-
ratory scale experiments were carried out in a cone
calorimeter (Babrauskas, 1982) modi®ed with con-
trolled ventilation (Mikkola, 1993) at the Technical
Research Centre of Finland (VTT), Espoo. In the cone
calorimeter, 10 g sample was loaded into a sample
holder (for CNBA stainless steel with the dimensions
10 cm  10 cm, for CB ceramic crucible 9.5 cm ;),
which was subsequently placed inside a closed chamber,
into which a stream of air was led at a ¯ow of 60 or 180
1/min, respectively. Above the sample was located a
continuously operating spark ignition device and a cone
shaped device electrically heated to 625°C or 800°C,
exposing the sample to an infrared radiation intensity of
25 or 50 kW/m2, respectively. Under these conditions,
CNBA ignited in 28±80 s and burned out in 90±126 s,
whereas CB ignited in 7±25 s and burned out in 84±135
s. The combustion gasses leaves through a chimney in
the cone centre, followed by an exhaust hood and an
exhaust channel where the isokinetic sampling
of PCDD/F took place. The exhaust ¯ow was about
24 1/s. Each experiment was performed as a double
experiment with common sampling, burning 2 Â 10 g of
substance. For each substance ®ve experiments
were performed, including a special extinction test with
water.
2.2. Experimental
2.2.1. Experiments in DIN-furnace, 2.5 g scale
The small laboratory scale experiments were carried
out in a DIN-furnace (Anon., 1981) at Risù National
Laboratory in Roskilde, Denmark. In this device the
combustion of all substances in Table 1 were investi-
gated. In each experiment, 2.5 g of substance was dis-
tributed in 24 quartz crucibles (having a total active
opening area, or pool size, of 33 cm2) and placed in the
quartz tube of the DIN-furnace, having a length of 1 m
and an internal diameter of 4 cm. The crucibles prevent
melting substances from ¯owing horizontally, much
improving the reproducibility (Smith-Hansen and
Jùrgensen, 1992). During the experiment, which lasted
40 min, pure atmospheric air at ambient temperature
and pressure was led into the quartz tube at a ¯ow of 100
1/h. In total, 18 g O2 was supplied, this is 2±6 times the
amount required for complete combustion, corre-
sponding to an over-ventilated ®re scenario. The tube
was externally heated by means of a 20 cm long movable
annular electric oven, moving upstream at a speed of 1
cm/min. The oven temperature was controlled by a
precision thermostat set to 500°C and 900°C, corre-
sponding to an infrared radiation intensity of 11 kW/m2
and 30 kW/m2, respectively. These two temperatures was
expected to cover the most important dioxin-generating
temperature range of a ®re. At 900°C ignition took place
readily for the majority of substances, initiating a
¯aming combustion, corresponding to a fully developed
®re scenario. At temperatures above 900°C the com-
bustion is more complete, and lower amounts of dioxins
can be expected according to incinerator studies (War-
nùe et al., 1989; Manscher et al., 1990). 500°C was
chosen as the lowest temperature relevant for a ®re,
corresponding to a smouldering ®re scenario. At this
temperature ignition did generally not take place, hence
the reactions may be characterised as non-¯aming
combustion/oxidation combined with pyrolysis and va-
porisation of un-reacted substance. At intermediate
temperatures, the substance will ignite occasionally,
leading to poor reproducibility of the test (Smith-Han-
sen and Jùrgensen, 1992). Hence, such temperatures
were avoided.
2.2.3. Experiments in ISO-room test chamber, 50 kg scale
The large scale experiments were carried out at The
Swedish National Testing and Research Institute SP in
ꢀ
Boras, Sweden, in a test chamber according to ISO 9705-
room test (Anon., 1993). The test chamber is 3.6 m long,
2.4 m wide and 2.4 m high, provided with a vertical
opening in an end wall 0.8 m wide with variable height,
set to 0.45 and 0.89 m, respectively. The in¯ow of air
and the out¯ow of combustion gasses takes place by
convection through the opening, above which is located
a large exhaust hood followed by an exhaust channel,
where the isokinetic sampling of PCDD/F took place.
The exhaust gas ¯ow in the channel was 3.5±4 kg/s.
About 50 kg test-substance was loaded into square steel
pans of sizes 0.5, 0.8 or 1.4 m2, respectively, placed on
the ¯oor of the chamber, and ignited by a ring
shaped propane burner held above the pan. CNBA ig-
nited in 1±2 min, CB in 15±20 s, both substances burned
out in 20±39 min. Three experiments were performed for
each substance testing dierent combinations of sub-
stance weight, pan size and chamber opening. The air
in¯ow can neither be controlled nor measured precisely,
but the fuel/oxygen equivalence ratio / in the ®re-gases
was measured by a special instrument, a phi-meter
2.2.2. Experiments in cone calorimeter, 20 g scale
In the medium laboratory scale and large scale ex-
periments only CB and CNBA were studied. These
chemicals are not sold in signi®cant amounts, but were
(Lonnermark and Babrauskas, 1997; Blomqvist et al.,