Ignition, moisture and safety

Thermit has an unusual hazard profile. It is hard to start: a match will not light it, and it needs a special booster to reach the temperature at which the reaction begins. Once started, it cannot be stopped, and it releases its whole energy in seconds as molten metal and slag at more than 2,000 °C. And it has one enemy above all others: water, which turns a controlled pour into a violent eruption.

This chapter explains why thermit is hard to ignite, calculates what a small amount of water does when it meets molten iron, and sets out how the crusher-frame job was planned around these hazards. It is an explanation of hazards for people who plan and supervise such work, not a set of instructions for carrying it out.

Why thermit is hard to start

The aluminothermic reaction releases a great deal of heat, but it must first be started. Aluminium is protected by its oxide film, and the solid ingredients touch only at points, so the reaction does not begin until part of the charge is heated to a very high temperature. Viall wrote in 1921 that it takes over 1,300 °C to ignite thermit, which is why a special ignition powder had to be used to start it, and why he considered the mixture itself safe to store and handle.

The history of ignition shows the same difficulty. Goldschmidt’s breakthrough, around 1895 or 1896, was a fuse of barium peroxide set off by a storm match; magnesium powder or ribbon was later used. Hart described heaping a spoonful of barium compound on the thermit, lighting it with a storm match and getting away as soon as it caught. TWI’s modern description is that the reactants are usually triggered with a spark from a flint lighter, but that rail welding needs a high activation energy, so a booster powder such as magnesium or a very hot flame source is also used.

Figure 1How a thermit charge is started

1. 1Boostera small, easily ignited high-temperature mixture or igniter
2. 2Ignitionspark, match or flame sets off the booster
3. 3Local reactiona small region of the charge passes about 1,300 °C and reacts
4. 4Propagationthe reaction's heat ignites the surrounding charge
5. 5Whole charge reactswithin seconds; it cannot be stopped once started

Hard ignition is a safety feature in storage and transport, but it brings its own risk at the moment of firing: the booster and the igniter are the sensitive part of the system. Viall warned that the ignition powder had to be kept away from heat, and that its box should be tightly closed before the reaction started, so that no spark could drop into it. Modern igniters are designed for the job and handled under the supplier’s instructions, but the principle is the same: the booster is kept separate, protected and away from the reaction until the moment it is used.

Once started, it cannot be stopped

Chapter 2 estimated that the crusher-frame charge would release about 375 MJ, at an average of the order of 19 MW if it reacted in about 20 seconds. Nothing practical can interrupt that once it has begun. There is no valve to close and no switch to turn off. The only controls are those applied before ignition: the right charge, a sound and dry crucible, the right position over the mould, an exclusion zone, and people standing well clear.

Hart described what the reaction looks like: dense white smoke, flame and showers of glowing slag thrown out as the charge burnt, a spectacle he compared to the blowing of a Bessemer converter. TWI notes that the exothermic process can create a lot of smoke and even explosions, and that protective equipment is required. The ejected drops and the intense light and heat are expected parts of a normal reaction, which is why the exclusion zone exists even when everything goes right.

Water: the hazard above all others

Viall warned in 1921 that all thermit materials must be kept dry, and that thermit which has once become wet cannot be restored to its original condition by drying. Modern suppliers say the same, and they add the more dangerous point: water anywhere near the reaction or the molten iron, in the charge, the crucible, the mould, the slag pan or the ground underneath, can cause a violent eruption.

The reason is the expansion of water into steam. One gram of water occupies one millilitre as a liquid. Treated as an ideal gas, it occupies about 1.7 litres as steam at 100 °C and atmospheric pressure, and much more at higher temperatures, because gas volume rises in proportion to absolute temperature. When molten iron at over 2,000 °C covers a pocket of water, the water flashes to steam almost instantly, trapped under the metal, and the steam blows the metal outwards.

Figure 2Steam from 10 millilitres of water

V = n × R × T ÷ P

n

moles of water: 10 g ÷ 18.0 g/mol ≈ 0.555 mol

R

gas constant, 0.0821 L·atm per mol·K

T

absolute temperature, K

1. at 100 °C (373 K): V ≈ 0.555 × 0.0821 × 373 ≈ 17 L
2. at 1,000 °C (1,273 K): V ≈ 0.555 × 0.0821 × 1,273 ≈ 58 L

Ten millilitres of water, about two teaspoons, produces about 17 litres of steam at 100 °C, a 1,700-fold expansion, and about 58 litres at 1,000 °C. That is enough to throw molten metal several metres. Hot iron can also react with water to produce hydrogen, which can burn. This is why moisture control runs through every stage of a thermit job:

• portions are stored sealed, dry and off the ground, and discarded if they have been wet;
• crucibles and moulds are dry and, where the procedure requires, preheated to drive off moisture;
• the slag pan or sand bed is dry, and covered if rain is possible;
• the ground under the work is dry, without puddles or damp sand;
• work stops in rain unless the area is fully protected.

TWI’s description of rail welding notes that the rail ends are preheated to make sure the molten steel is not chilled as it flows; a side effect of preheating is that it also drives off moisture from the rail and mould surfaces.

Fumes, light and heat

The reaction produces a dense white fume, largely aluminium oxide, along with fume from additions, mould materials and any coatings or contaminants on the parts. The intense light includes infra-red and ultraviolet radiation. The radiant heat from the crucible and the pour can burn exposed skin at a distance. Safe Work Australia’s model code of practice on welding processes covers fumes, radiation, burns and fire for welding processes generally, and those controls apply alongside the specific controls the thermit supplier and the work’s risk assessment require: position upwind where possible, respiratory protection as assessed, eye and face protection suitable for the radiation, and heat-resistant protective clothing.

Fire

Molten metal and slag ignite almost anything combustible they land on. In a quarry the hazards include conveyor belting, rubber, oil and grease on machinery, timber, dry vegetation and fuel. The crusher area must be cleared of combustibles within the area that ejected slag and sparks could reach, with fire extinguishers ready and a fire watch kept after the weld, because slag and hot metal stay hot enough to start fires long after the pour. If the site requires a hot-work permit, the thermit weld needs one. Hidden combustibles deserve particular attention: grease inside bearing housings, oil-soaked dust in crevices of the crusher frame and rubber liners on the far side of a plate can all be ignited by heat conducted through the steel, out of sight of the people watching the pour.

Storage and transport

Because thermit is hard to ignite, a sealed portion in a dry store is a stable product. The risks in storage and transport are moisture, damage and confusion: a torn bag that has drawn in damp air, a portion knocked about so that its contents have separated, or the wrong portion picked for the job. Suppliers’ instructions and safety data sheets set out the storage conditions, the classification of the product for transport and the handling precautions, and those govern. In general terms, portions are kept in their sealed packaging in a dry, cool store away from heat sources and incompatible materials, off the floor, used in date order and discarded if the packaging is damaged or they have been wet. Igniters and boosters are stored as their own instructions require, separately from the portions.

For the crusher job, the contractor would bring the portion, igniter and crucible to site on the day, and the quarry’s store would hold nothing. That keeps the storage problem with the party equipped and trained to manage it.

The crucible

The crucible holds the reaction and, in the fusion method, releases the iron through a tap in its base. Its refractory lining must withstand the reaction’s temperature and the erosion of the boiling metal and slag. A lining that is cracked, eroded or damp can fail during the reaction, releasing metal where it is not wanted, or can chill and contaminate the iron. Many modern rail welding systems use single-use crucibles supplied with the portion, which removes the question of how many reactions a lining has survived. Reusable crucibles are inspected, dried and maintained to the supplier’s instructions, and their history is recorded.

The crucible’s position matters as well. It must be supported securely over the mould, at the height the procedure specifies, because the tapped stream must enter the mould’s runner cleanly. Hart’s 1910 description placed the tap hole about 4 inches, roughly 100 mm, above the mould’s opening. A crucible that shifts during the reaction can direct the stream outside the mould.

When something goes wrong

Emergency planning for a thermit job covers what happens if the reaction does not start, if metal escapes from the mould, if a fire starts, or if someone is burnt. The supplier’s procedure states what to do if an igniter fails: in general, people stay clear for a set time, because a delayed ignition is possible, and nobody leans over the crucible to look. A mould that leaks, called a run-out, spills molten metal; the dry sand bed, the cleared area and the exclusion zone limit the damage. Burns from molten metal and slag are severe and need immediate first aid and medical treatment, so the job plan includes trained first aiders, a burns kit and a way to get an injured person out of the crusher area quickly.

Planning the crusher-frame weld around its hazards

This is an illustration. The quarry engineer’s job plan for the proposed thermit weld, reviewed with the contractor, listed the hazards and controls:

Figure 3Hazard controls in the crusher-frame thermit job plan

• Avoid: Supplier's sealed portions onlystored dry, off the ground, checked before use
• Avoid: Dry crucible, mould and slag pancovered from rain; ground under the work dry
• Avoid: Exclusion zoneset by the contractor's procedure; marked and enforced during firing
• Avoid: Combustibles removedbelting, oil, grease and timber cleared or protected
• Avoid: Fire watch and extinguishersduring and after the pour
• Avoid: Fume, radiation and heat protectionas the risk assessment requires
• Avoid: Hot-work permitissued by the site
• Do: ? Weatherwork stops if rain threatens and the area cannot be protected

The plan also named the people and their roles: the contractor’s accredited welder in charge of the weld and the firing, a second contractor’s crew member, the quarry’s permit issuer, a fire watch and a first aider, with nobody else inside the exclusion zone during firing. Plant isolation for the crusher, conveyors and feeders was covered by the quarry’s normal isolation procedure before any work in the crusher started.

The engineer’s main concern was the crusher pit. Water collects in the bottom of a jaw crusher pit, and fines that look dry on top can be wet underneath. The plan required the area under the frame member to be pumped out, dried and covered with a bed of dry sand before the mould was built, and checked again immediately before ignition.

A failure: the damp slag pan

This is an illustration. During an earlier rail weld on the quarry siding, years before the present engineer, the contractor’s crew had set the slag pan on the ballast beside the rail after an overnight shower. When the slag ran into it at the end of the pour, there was a sharp report and a spray of slag fragments across the ballast, several metres from the pan. Nobody was hurt, because the crew were outside the exclusion zone, and the weld itself was unaffected.

The cause was a small amount of water that had collected in the bottom of the pan, which had been left uncovered. The contractor’s investigation found that the pre-pour check had covered the crucible and the mould but not the slag pan. Its procedure was amended to include the slag pan, its location and its cover in the pre-pour check, and the quarry kept a copy of the investigation report, which the present engineer found while preparing the crusher job.

Exercise: water in the pit

The quarry engineer is asked how much steam would be produced if 50 mL of water were trapped under the pour in the crusher pit. Estimate the volume of steam at 100 °C and at 1,500 °C, treating the steam as an ideal gas at atmospheric pressure.

Fifty millilitres of water is about 50 g, or 50 ÷ 18.0 ≈ 2.78 mol. At 100 °C (373 K), the volume is about 2.78 × 0.0821 × 373 ≈ 85 L. At 1,500 °C (1,773 K), it is about 2.78 × 0.0821 × 1,773 ≈ 404 L.

A volume the size of a small cup of water would become several hundred litres of steam in contact with molten iron, released in a fraction of a second underneath a mass of liquid metal. That is the calculation behind the plan’s insistence on a pumped, dried and sanded pit, checked immediately before ignition.