Scale, capacity and feedstock: which chemical plants get built: Technology, learning and feedstock: where the cost floor is set
Understanding Chemicals / Module 7: Scale, capacity and feedstock: which chemical plants get built Part 4 of 5Technology, learning and feedstock: where the cost floor is set
Scale and utilisation decide how efficiently a plant spreads its fixed costs. Technology and feedstock decide what those costs are in the first place. This part looks at how process technologies improve and get replaced, why feedstock prices give some regions a lasting edge, and how the IEA's scenarios might redraw the map of where ethylene, ammonia and methanol get made.
4.1 Technologies follow S-curves, and experience lowers costs only up to a point
Brennan describes a process technology's life as an S-shaped curve. In the initiation phase it can take ten years or more from discovery to commercial production, and five to ten more before production is clearly profitable; first plants are often small to limit capital at risk. In the growth phase the process improves, costs fall and plants get bigger. In maturity, cost reduction and scale growth slow sharply (pdf pp. 182-185). A newer technology may then overtake the old one: sulphuric acid moved from lead chambers to the contact process and then to double absorption; chlor-alkali moved to titanium anodes in the 1960s and to membrane cells from the early 1980s (pdf p. 183).
A learning or experience curve links real price or cost to cumulative production: S = d x X^q, with q negative. The Boston Consulting Group observed real cost falls of 20-30% for each doubling of cumulative production. A 20% fall corresponds to q of about -0.32, since 2^-0.32 = 0.80 (pdf pp. 192-193). Taylor and Craven found the characteristic price decline in chemicals only for products with fast-growing demand.
Brennan is cautious about the curve. Prices can fall because inputs get cheaper, not because the industry learned: the real price of Ethylene fell about 7% a year in the 1960s partly because feedstock and energy got cheaper, then jumped after the 1973 oil shock (pdf p. 194). The curve also ignores advances from outside the industry, such as metal anodes and membranes in chlor-alkali. He prefers to track measures over time: capacity, raw materials, energy, staff and capital per tonne (pdf p. 194).
| Technology | Period | Growth in maximum capacity (% a year) |
|---|---|---|
| Ethylene from naphtha | 1950-70 | 13.0 |
| Ethylene from naphtha | 1970-85 | 1.5 |
| Chlorine and caustic soda, mercury cell | 1940-75 | 8.7 |
| Chlorine and caustic soda, mercury cell | 1975-85 | 0.0 |
| Aluminium smelting | 1900-80 | 6.3 |
| Catalytic cracking of petroleum | 1913-53 | 15.0 |
| PSA oxygen | 1971-76 | 33.3 |
- Brennan (2020), Ch 7, 7.2-7.8 · pdf pp. 182-185, 192-196
4.2 Feedstock price is the main reason one region's plant beats another's
The IEA states the rule plainly: the overarching factor in choosing a feedstock and route is production cost. Equipment costs are fairly similar worldwide because there are few vendors, so the big differences come from feedstock (pdf p. 31). Crude oil sells at a similar price everywhere, but natural gas prices can differ by a factor of four; gas tends to cost more than coal, and oil products more than gas. Feedstock can be 15% to 85% of the levelised cost of high-value chemicals (HVCs: ethylene, propylene and the aromatics benzene, toluene and xylenes) (pdf p. 32).
Brennan's 2017 prices show the spread. Natural gas was $3.0/GJ at Henry Hub in the United States and $5.6/GJ in Europe, while Japan's $8.1/GJ was built up from Henry Hub plus $3 for liquefaction and $2.1 for shipping. Crude was $51/bbl for West Texas and $53 for Dubai. Coal was $62/t in the United States and $85/t in China and Europe (pdf pp. 41-42).
Ammonia shows how directly this flows through: gas is both the source of hydrogen and the furnace fuel, and Kent puts it at 70% to over 85% of cash cost (pdf p. 907). Efficiency has improved, from 55.9 GJ/t in 1950 to 26 GJ/t in 1982 (Shreve, pdf p. 325), but no efficiency gain can offset a gas price two or three times higher.
Coal is the alternative where gas is scarce. Coal-based ammonia equipment costs more than twice as much per unit of output as gas-based, and biomass-based nearly seven times (IEA, pdf p. 31). In China, coal is cheap enough to give lower costs for ammonia and methanol anyway, though coal-based plants have tended to run at low utilisation (pdf pp. 33-34).
Sulphur removed so it does not poison the nickel catalyst
2 Primary reformerCH4 + H2O → CO + 3H2 over nickel catalyst in fired tubes
3 Secondary reformerAir added at 28-30 bar and 955-1,025 C; burns some gas and brings in nitrogen
4 Shift conversionCO + H2O → H2 + CO2, more hydrogen per unit of gas
5 CO2 removal and methanationCO2 absorbed by solvent; traces of CO and CO2 turned to methane
6 Compression and synthesis loopN2 + 3H2 → 2NH3 at about 20 MPa over iron catalyst; ammonia condensed, purge gas recovered
Outputs Ammonia1 t Carbon dioxideby-product, used to make urea Steamabout 4 t raised per t of ammoniaGas is 70% to over 85% of cash cost, so the gas price dominates plant economics.
Source: Shreve (1984) Ch 18; Kent (2012) Ch 22; IEA (2018), Shreve pdf p. 325; Kent pdf pp. 892, 905-907; IEA pdf p. 29- IEA (2018), Ch 2 The present petrochemical landscape · pdf pp. 31-34
- Brennan (2020), Ch 2 Table 2.4 · pdf pp. 41-42
- Kent (2012), Ch 22 · pdf pp. 905-907
- Shreve (1984), Ch 18 · pdf p. 325
- Wittcoff (2004), Ch 1 · pdf p. 57
4.3 Ethane gives the cheapest ethylene, naphtha the broadest product slate, and scenarios may shift the map
Ethylene is made by steam cracking: heating a hydrocarbon very briefly to a high temperature so its molecules break into smaller ones. The lighter the feed, the more becomes ethylene. Ethane needs only about 1.2 t of feed per tonne of HVCs and gives about 50% more HVCs than naphtha, but nearly all of it is ethylene. Naphtha, a light petrol fraction from refineries, gives a balanced mix of ethylene, propylene and aromatics (IEA, pdf pp. 31-32).
Where ethane is cheap, it wins. The IEA found Middle East ethane crackers the cheapest route to HVCs in both 2010 and 2017 (pdf pp. 32-33). Saudi associated gas, which comes up with oil, had no other use, and the Saudis costed ethane extraction at $0.50-0.75 per million Btu in 2001, when US gas hit $10 (Wittcoff, pdf pp. 52, 57). US shale gas then changed the picture. From 2010 to 2017 US ethane prices fell by about two-thirds, and producers left up to 400-600 kb/d of ethane in the gas stream ('ethane rejection') until about 4 Mt of new cracking capacity arrived in 2014-17 (IEA, pdf p. 76).
Regions without gas liquids crack naphtha. China uses naphtha for over 90% of its HVCs and has added coal-based methanol-to-olefins (MTO) plants. Before the 2014 oil price slump, MTO undercut the cheapest naphtha cracker by nearly $300 per tonne of HVC; by 2017 the gap was almost zero (pdf p. 33). A feedstock advantage moves with the oil price.
The IEA's Reference Technology Scenario (RTS) follows current policies; its Clean Technology Scenario (CTS) works back from sustainability goals. Neither is a forecast. In the RTS, ethane supply peaks in the late 2020s and naphtha supplies over half of HVC output to 2050 (pdf p. 76).
| Route | Capital cost (USD per t HVC) | Product slate | Where it dominates |
|---|---|---|---|
| Ethane steam cracking | 1,500 | Mostly ethylene; about 1.2 t feed per t HVC | United States, Middle East |
| Naphtha steam cracking | 2,050 | Balanced ethylene, propylene, aromatics | Europe, Asia (China over 90%) |
| Methanol-to-olefins (MTO) | 1,000 plus upstream methanol | Balanced propylene and ethylene | China, coal-based methanol |
Capital assumptions from the IEA's simplified levelised cost comparison, with an 8% discount rate and 25-year life.
Source: IEA (2018), Ch 2, Figure 2.6 notes, pdf pp. 31-33- IEA (2018), Executive summary; Ch 2; Ch 4; Ch 5 · pdf pp. 12, 31-34, 69, 75-79, 92, 99, 112
- Wittcoff (2004), Ch 1 · pdf pp. 52, 57
See it in the table
Oil Crude oilGas Natural gasTi TitaniumH2SO4 Sulphuric acidH2 HydrogenNH3 AmmoniaUr UreaC2H4 EthyleneC3H6 PropyleneBz BenzeneTol TolueneMeOH MethanolCheck your understanding
Brennan cautions against reading the 7% a year fall in ethylene's real price in the 1960s as pure learning. Why?
Part of the price fall came from falling real unit costs of feedstock and energy, not from improved performance, and prices jumped after the 1973 oil shock.Why can two identical ammonia plants have cash costs differing by more than a factor of two?
Gas prices vary by a factor of up to four between regions (IEA), and with Kent's 40-50 multiplier, $3 versus $8.5 per million Btu gas gives roughly $126-158 versus $342-427 per tonne.What is the main drawback of ethane as a cracker feed compared with naphtha?
Ethane gives the highest HVC yield but mostly ethylene. Naphtha gives a more balanced slate of ethylene, propylene and aromatics.