Scale, capacity and feedstock: which chemical plants get built: How plant sizes grew, and what stops them growing

2.1 For three decades the biggest plants grew faster than their markets

Brennan describes a self-reinforcing loop in a technology's growth phase: falling costs keep the market growing, a growing market justifies larger plants, larger plants cut costs further. For many major chemicals the largest plant capacity grew faster than demand (pdf p. 185). In the 1960s the largest new plants grew by about 13% a year for ethylene, 15% for ammonia, 17% for ethylene oxide and 26% for vinyl chloride monomer (pdf p. 227).

Ethylene shows the arc clearly. Wittcoff gives a typical cracker of 70 million lb/year in 1951 and 1 billion lb/year in 1972 (pdf p. 58). At 0.4536 kg per lb that is about 32,000 t to 454,000 t, a 14-fold rise in 21 years, close to 13% a year, which matches Brennan's 13.0% for 1950-70. Then the curve bent to 1.5% a year for 1970-85 (Brennan, pdf p. 195). Plants of 1.5 billion lb (about 680,000 t) appeared in the early 1990s and of 2 billion lb (about 907,000 t) in 2001 (Wittcoff, pdf p. 58).

Ammonia followed a similar path. Shreve gives the largest US plant in 1984 as 1,500 t/day, with designs to 1,650 t/day made possible by single-train construction and rotary compressors; energy use fell from 55.9 GJ/t in 1950 to 26 GJ/t in 1982 (pdf pp. 323, 325). By 2001, 2,000 t/day was common, and 3,000 t/day plants were being designed in 2002 (Kent, pdf pp. 902, 907).

Sulphuric acid started tiny: the first lead chamber of 1746 was 1.8 m2 (Kent, pdf p. 1012), and chamber plants were small units beside superphosphate works near farms (Shreve, pdf p. 345). A 2,800 t/day plant started in 1982 (pdf p. 358).

• Brennan (2020), Ch 7 and Ch 8 · pdf pp. 185, 192, 195, 227
• Wittcoff (2004), Ch 1 · pdf p. 58
• Shreve (1984), Ch 18 Nitrogen industries; Ch 19 Sulfur and sulfuric acid · pdf pp. 323, 325, 345, 347, 358
• Kent (2012), Ch 10, 22, 25 · pdf pp. 385, 902, 907, 1012

2.2 Equipment limits, capital, project risk and hazard put a ceiling on size

Why did the curves flatten? Brennan lists the forces in a mature technology (pdf p. 192). Each further increase in capacity brings a smaller gain in capital efficiency. Very large plants need capital that is hard to raise; Wittcoff puts the bare minimum to enter basic chemicals at $1 billion (pdf p. 61). Big projects are harder to manage, so delays and overruns grow. A huge plant loses heavily if utilisation falls, through a weak market or poor reliability. Compressors, reactors and furnaces have physical size limits. And larger plants hold larger inventories of hazardous material.

When equipment reaches its limit, designers stop scaling up and start scaling out, adding identical units in parallel. Smelter pots, chlor-alkali cells and ethylene cracking furnaces are built this way (pdf p. 189). Scale-out still gives some economies because units share piping, foundations, design and project management. In Brennan's heat exchanger estimates, two identical shells give an exponent of 1.0 on purchased cost but only 0.68 on installed cost (pdf pp. 190-191).

Scale-out has trade-offs. Each unit holds less hazardous material, so one failure releases less, and the plant can keep running at reduced output if one reactor fails. But more units mean more flanges, seals and relief valves, so more leak points and more monitoring (pdf p. 191). A single-train plant is cheaper but fragile: one failure stops everything.

Pratten described each vintage of plant as having its own scale curve, kinked just above the largest size anyone has built, reflecting the uncertainty of being first; the kink moves outwards as experience grows (pdf p. 185).

• Brennan (2020), Ch 7 · pdf pp. 185, 189-192, 205
• Wittcoff (2004), Ch 1 · pdf p. 61

2.3 Small plants survive where the market, freight, hazard or a feedstock protects them

Despite the pull of scale, plant sizes vary widely: European membrane-cell chlor-alkali plants range from about 20 to 650 kt/year (Brennan, pdf p. 189). Brennan gives four cases where a small plant can be justified: a limited quantity of cheap feedstock; a product whose transport cost is high relative to its price; a hazardous product whose transport is restricted, such as liquefied chlorine; and an innovation that cuts capital cost.

The ceiling on a local producer's price is import parity, the cost of landing the product from abroad. Brennan's example for a polymer shipped from Europe to Australia is 1,200 euros/t free on board, plus 120 duty, 160 freight and 20 landing: 1,500 euros/t (pdf pp. 35-36). A small domestic plant can live under that umbrella if its costs are low enough.

This is also why trade happens further down the chain. Of 180 Mt of ammonia made in 2016, only about 7% was traded, while trade in urea, a solid with more nitrogen value per tonne, was more than 300% higher (IEA, pdf p. 35). Sulphuric acid from a remote mine smelter is seldom saleable, because of the cost and safety of transport, so downstream options such as phosphoric acid and phosphate fertilisers have to be weighed (Brennan, pdf p. 236).

Market size sets the hardest limit. If the local market is small, even fast growth adds too few tonnes to fill a new plant of economic size, so expanding an old plant wins (pdf p. 220). At the far end, fine chemicals are made in batches in multipurpose plants; Kent's rule of thumb is batch below about 500 t/year and continuous above about 5,000 t/year (pdf p. 174).

• Brennan (2020), Ch 2, 7, 8 · pdf pp. 35-36, 189, 220, 233, 236
• IEA (2018), Ch 2 · pdf p. 35
• Pollak (2011), Ch 5 Plant design · pdf p. 74
• Kent (2012), Ch 5 Green Engineering · pdf p. 174

See it in the table

Check your understanding

Wittcoff's ethylene crackers grew from 70 million lb/year in 1951 to 1 billion lb/year in 1972. What annual growth rate does this imply?

Why does scaling out with identical parallel units still give some economy of scale on installed cost?

Why is far more urea traded internationally than ammonia?