How a chemical process works: reactions, separations, catalysts and flowsheets: Reading a flowsheet, and safety as a design input

5.1 Walking through the contact process step by step

Shreve calls a skilfully drawn flowchart the most concise description of a process that exists, and recommends that it show utilities and labour as well as materials. Read a flowsheet from left to right, asking at each box what changes, at what temperature and pressure, and where heat goes in or out.

In a sulphur-burning Sulphuric acid plant, molten sulphur is kept liquid in steam-heated lines and sprayed into a furnace. Shreve notes that sulphur-burning plants are the simplest and cheapest because the gas needs no special cleaning. Combustion air is first dried in a tower of 93 to 98% acid, because water vapour would later condense as corrosive acid. Burning gives a gas of about 10 to 12% sulphur dioxide, and the heat raises steam in a waste-heat boiler.

The gas then enters a converter of four or five fixed beds of vanadium pentoxide catalyst on silica. Kent gives the operating range for this catalyst as about 410 to 440 C, with caesium-promoted versions working at 360 to 400 C. Each bed heats up as it reacts, and boilers, superheaters or gas-gas exchangers between beds bring the temperature back down. After two or three beds the gas passes to an intermediate absorber, where sulphur trioxide dissolves in circulating 98% acid at 60 to 80 C. Kent explains that 98.3% acid is used because the vapour pressures of SO₃, H₂SO₄ and water are lowest there. The gas is reheated, passes over the last beds and goes to a final absorber.

Water is added to the absorber acid to hold its strength, and product acid is drawn off. Acid coolers remove the heat of absorption into cooling water. Shreve reports that vanadium catalysts can last up to 20 years and resist most poisons, apart from fluorine and arsenic.

• Shreve (1984), Ch 2 Flowcharts · pdf p. 26
• Shreve (1984), Ch 19 Contact process · pdf pp. 347-354
• Kent (2012), Ch 25 Sulfur and Sulfuric Acid · pdf pp. 1016-1017

5.2 The material and energy balance tells you what the plant earns

A material balance tracks every kilogram in and out. Shreve calls it the essential first step in any process study, because cost is driven by how materials are used. An energy balance does the same for heat and power. For the contact process both can be worked from source figures.

Materials first. One tonne of pure acid contains 0.327 tonnes of sulphur, 0.490 tonnes of oxygen taken from air, and 0.184 tonnes of water added in the absorbers; these add up to 1.001 tonnes after rounding. Losses are small. A single absorption plant converting 98% of its sulphur dioxide would release about 13 kg of SO₂ per tonne of acid (2% of 0.327 tonnes of sulphur, doubled because SO₂ weighs twice as much as the sulphur in it). At 99.7% conversion this falls to about 2 kg. Shreve reports that US rules of the time required new sulphur-burning plants to match 99.7% conversion, which is why double absorption became standard.

Energy next. The three reaction heats add to about 5.4 GJ per tonne, matching Kent's 63 MW for a 1,000 tonne-per-day plant. Kent says 55 to 60% is typically recovered as high-pressure steam, about 3.0 to 3.2 GJ per tonne, and Shreve puts the steam at more than 1.3 tonnes per tonne of acid in large plants. Most of the rest leaves in cooling water.

The balance turns into money. At Brennan's 2018 price of $10 to $16 per tonne of high-pressure steam, 1.3 tonnes is worth $13 to $21 per tonne of acid, a real credit against the cost of sulphur. Shreve gives the example of a Swedish 2,200 tonne-per-day plant supplying 52 MW to district heating. The catalyst inventory is also set by the balance: Kent gives 150 to 200 litres per tonne-per-day of capacity, or 150 to 200 cubic metres for a 1,000 tonne-per-day plant.

• Shreve (1984), Ch 2 Material balances · pdf pp. 25, 29
• Shreve (1984), Ch 19 · pdf pp. 347-349
• Kent (2012), Ch 25 · pdf pp. 1016-1017
• Brennan (2020), Ch 4 Table 4.4 · pdf p. 110

5.3 Inherent safety: remove the hazard before adding protection

Kent separates occupational safety (slips, falls, cuts) from process safety, which is about preventing fires, explosions and toxic releases. The common thread in process accidents is loss of containment: material escaping from the pipes and vessels meant to hold it, through rupture, overpressure, corrosion or human error.

Protection is layered. Kent describes the Swiss cheese picture, where each barrier has holes and an accident happens only when the holes line up. The layers run from the vessel and piping itself, through the basic control system and alarms, to an independent safety instrumented system that trips the plant, then relief valves, and finally dikes, fire protection and emergency response.

Kent ranks strategies by reliability. Least reliable are procedures, which depend on people. Better are active engineering controls, then passive features such as dikes that work without any device operating. Most reliable is inherent safety, a term coined by Trevor Kletz of ICI: designing the hazard out. His principles are to use less hazardous materials, keep inventories small, run at less severe conditions and simplify equipment. They are easiest to apply at the laboratory or concept stage.

Inherent choices involve trade-offs. Kent's example: supplying chlorine from many small cylinders instead of one big tank means more connections and more chances of a leak, but each leak is much smaller. Another: a reactor designed to 50 psig to make release less likely stored so much energy that, when a runaway did burst it, the damage was far greater than a low-pressure vessel would have caused.

Relief systems are the last engineered layer. Designers pick the worst credible scenario, such as fire, runaway reaction or loss of cooling, calculate the flow to be vented, and size the device. Kent reports a key finding of the DIERS industry programme: runaway reactions often vent a two-phase froth that needs two or more times the relief area of vapour alone.

• Kent (2012), Ch 2 Safety Considerations in the Chemical Process Industries · pdf pp. 61-67, 71-72, 96-97

5.4 Flixborough and Bhopal turned inventory into a design variable

Kent's case histories show how management failures turned process hazards into disasters, and how the lessons fed back into design.

Flixborough, England, 1974. The Nypro plant oxidised cyclohexane to make cyclohexanone, a step towards Caprolactam and so Nylon (PA6, PA66). Six reactors in series ran at 155 C and 8.8 bar and held 120 tonnes of cyclohexane, a flammable liquid far above its boiling point. When one reactor cracked, the plant was rushed back into service with a 20-inch pipe bypassing it, supported on scaffolding, with bellows at each end and a dog-leg to cover the level difference. The only drawings were in chalk on the workshop floor, and there was no qualified mechanical engineer on site. The pipe failed. About 30 tonnes of cyclohexane formed a vapour cloud that exploded, killing 26 people, injuring 36 and destroying 1,821 houses. A key conclusion, Kent notes, was that limiting inventory of flammable material should be a specific design objective in major hazard plants.

Bhopal, India. A Union Carbide India plant used methyl isocyanate (MIC), an extremely toxic intermediate, to make the pesticide Sevin. About 120 to 240 gallons of water got into an MIC storage tank. The MIC reacted, heating and pressurising the tank until the rupture disc burst. Roughly 27,000 lb escaped over two hours. The scrubber and flare were sized for 10,000 lb per hour of routine vent flow, not a runaway in storage; the refrigeration was off, the flare was down for repair, and the water curtain could not reach the gas. A shanty town of about 100 people per acre had grown next to the plant. Kent reports about 1,700 to 2,700 deaths, possibly more, and 50,000 people seriously affected. Contributing causes included a larger inventory of MIC than was needed.

• Kent (2012), Ch 2 Case histories · pdf pp. 113-115

See it in the table

Check your understanding

Which strategy does Kent rank as the MOST reliable way to control a hazard?

Why do relief systems for runaway reactions often need much larger vents than simple vapour relief?

In the contact process energy balance, roughly what share of the heat leaves in cooling water from the acid coolers?