Biorefineries and platform chemicals: Economics and real cases

5.1 Feedstock and yield dominate the cost of a bio-based commodity

Dale and Kim state the rule: about 60 to 70% of the cost of making a commodity comes from raw materials. Dean and colleagues give the carbon source as often over 50% of direct production cost in fermentation; Kent puts feedstock at up to 40% of lignocellulosic ethanol cost. For PLA, Gruber and colleagues estimate corn sugar at 25 to 40% of total cost.

On the purchase price, biomass looks cheap. Thoen and Busch list average 2000-03 prices of 175 euro per tonne for petroleum, 80 for corn, 20 for straw and 180 for sugar. Dale and Kim calculate that corn at 3 dollars per bushel matches oil at 35 dollars per barrel on an energy basis but costs less than half on a mass basis, and that crop residues cost a fifth to a half of oil per unit of energy. Wet millers also use 'net corn cost', after credits for oil and protein, typically about 70% of the purchase price.

The catch is yield. Sugar loses mass on its way to most products, and every point of lost selectivity raises sugar cost per tonne, reactor size and separation load. Kent adds that composition variability in stover alone shifts ethanol cost by at least 0.30 dollar per gallon, because the maximum yield is proportional to carbohydrate content.

The Biofine cost table shows the structure for a chemical route. Feedstock at 40 dollars per tonne is 14 million of the 44.3 million dollar gross cost; the purchased ethanol for esterification is almost as large. By-product credits trim the net to 291 dollars per tonne of ethyl levulinate, before capital charges.

• Kamm (2006), Vol 1 Ch 2 Biomass refining global impact · pdf pp. 80-82
• Kamm (2006), Vol 1 Ch 3 Development of biorefineries · pdf p. 100
• Kent (2012), Ch 33 Biomass conversion · pdf pp. 1260, 1273
• Kamm (2006), Vol 2 Ch 12 Industrial chemicals from biomass, Table 12.1 · pdf p. 809
• Kamm (2006), Vol 2 Ch 14 Polylactic acid · pdf pp. 855-856

5.2 Scale, integration and by-products decide who survives

Biorefineries cannot simply get bigger. Johnson shows the usual economics in corn refining: doubling capacity raises investment only about 50% and needs little extra labour, while utilities scale one-for-one. Investment per unit levels off around 60,000 to 80,000 bushels a day. But a 100,000 bushel per day plant already uses more than a square mile of corn every day and 28 railcars of grain. Clements estimates a 1,000 to 2,000 t/day multi-product biorefinery at about half a billion dollars, drawing on a 150 to 300 km radius.

Integration with an existing industry is the cheapest way in. Dale and Kim expect new fuels and chemicals to appear first where capital already exists: pulp mills, corn mills, sugar mills. The cases bear this out. Rossell's PHB analysis found integrating with a sugar mill cut production cost from 2.50-3.00 to 2.25-2.75 dollars per kg. Danisco's xylose process uses hydrolysed sugars already in a pulp mill's spent liquor. Task 42 adds that if lignocellulose can yield carbohydrates at 300 euro per tonne instead of over 600 on the open market, a 'clip-on' to an existing plant may pay.

By-products can make or break the plant. Lignin is the largest: up to 30% of the weight and 40% of the energy of lignocellulose. Pye estimates over 70 million tonnes of kraft lignin a year, more than 99% burned in pulp-mill recovery boilers; perhaps 100,000 t is sold for its chemical value. Lignosulfonates from sulphite mills, below 800,000 t a year after 2001, go to concrete admixtures, binders and dispersants. Burning lignin is not waste: it powers the plant, which is why lignocellulosic ethanol can run with little fossil energy.

• Kamm (2006), Vol 1 Ch 15 The corn wet and dry milling industry · pdf p. 377
• Kamm (2006), Vol 1 Ch 5 The lignocellulosic biorefinery · pdf p. 147
• Kamm (2006), Vol 1 Ch 2 Biomass refining global impact · pdf pp. 85-86, 88
• Kamm (2006), Vol 1 Ch 10 Sugar-based biorefinery (Rossell et al.) · pdf p. 251
• Task 42 (2012), Sections 4.7, 6 and 7 · pdf pp. 6, 10, 15, 25
• Kamm (2006), Vol 2 Ch 5 Industrial lignin production (Pye) · pdf pp. 629-633

5.3 Cases that worked: NatureWorks PLA and DuPont's 1,3-propanediol

Both successes were built next to a corn wet mill, on a novel polymer rather than a cheaper copy of an existing chemical.

Polylactic acid was first made by Carothers in 1932 but stayed a costly niche, over 2 dollars per pound, used mainly for sutures. Cargill and Dow formed Cargill Dow in 1997 to cut cost and build markets; the plant at Blair, Nebraska started in 2001-02 at a declared 300 million pounds a year, alongside a 400 million pound lactic acid plant. Task 42 gives NatureWorks, by then wholly owned by Cargill, a capacity of 140,000 t. The key was a solvent-free continuous process that makes lactide and polymer in the melt. Gruber notes that PLA cannot be made economically from petrochemicals at all, because the polymer's properties depend on controlling the optical isomer, which fermentation does naturally.

DuPont's route to 1,3-propanediol (PDO), monomer for its Sorona polyester, replaced chemistry from ethylene oxide or acrolein. Natural bacteria convert glycerol to PDO, but glycerol was expensive and conversion incomplete. Genencor and DuPont built both pathways, glucose to glycerol and glycerol to PDO, into E. coli. After analysing over 70 genes and modifying 18, the strain made 135 g/L at 3.5 g/L/h and 51 wt% yield on glucose. Kent reports the bioprocess needs 40% less capital and has 25% lower operating cost than the chemical route. DuPont Tate and Lyle's plant at Loudon, Tennessee, has 45,000 t capacity, with a 35% expansion announced; Bachmann and Riese put DuPont's Sorona investment at over 500 million dollars.

• Kamm (2006), Vol 2 Ch 14 Polylactic acid from renewable resources · pdf pp. 835-845, 855-856
• Task 42 (2012), Section 9.3 C3 compounds · pdf pp. 19-20
• Kent (2012), Ch 30 Industrial biotechnology · pdf pp. 1164-1165
• Kamm (2006), Vol 2 Ch 16 Industrial biotech (Bachmann & Riese) · pdf p. 902

5.4 Drop-ins, succinic acid and cellulosic ethanol: scale and status by 2012

Task 42 estimated bio-based chemicals and polymers at about 50 million tonnes a year, excluding biofuels, against about 330 million tonnes of petrochemicals. Much of the 50 million is old: non-food starch, cellulose fibres, tall oils, fatty acids, ethanol and citric acid.

The easiest new entrants are 'drop-ins', identical molecules such as ethylene, which need no new customer approval. Braskem built a 200,000 t polyethylene plant at Triunfo, Brazil, in 2010 for about 300 million dollars, about 1,500 dollars per annual tonne, making it from cane ethanol; Task 42 notes vapour-phase dehydration of ethanol at 400 C gives over 99% conversion and selectivity. At 1.7 t ethanol per t ethylene, the plant needs about 340,000 t of ethanol a year. Brazilian cane yields about 85 t per hectare, and bagasse fires the mills.

New molecules move slower. Task 42 notes a new plastic typically needs over 6 years to reach large scale, and one with a complex supply chain 20 to 40 years to pass 100,000 t. Succinic acid had a world market of about 30,000 t in 2009, under 5% bio-based. BioAmber and Mitsui planned 17,000 t at Sarnia in 2013, rising to 35,000 t plus 23,000 t of butanediol.

Cellulosic ethanol remained the hardest. Iogen ran a demonstration from April 2004 at up to 2,000 gallons a day; a full plant would be 85 times larger. Bachmann and Riese judged it too inefficient to compete with grain ethanol. History warns too: Scholler-process wood hydrolysis plants such as Tornesch made about 200 litres of ethanol per tonne of wood, yet most German plants shut after the Second World War when cheaper petroleum ethanol arrived.

• Task 42 (2012), Sections 1, 7, 8 and 9 · pdf pp. 4, 14, 16-26
• Kamm (2006), Vol 1 Ch 10 Sugar-based biorefinery · pdf p. 238
• Kamm (2006), Vol 1 Ch 9 Iogen's demonstration process · pdf p. 221
• Kamm (2006), Vol 2 Ch 16 Industrial biotech · pdf p. 903
• Kamm (2006), Vol 1 Ch 1 Biorefinery systems - an overview · pdf p. 42
• Kamm (2006), Vol 1 Ch 6 Lignocellulosic feedstock biorefinery: history · pdf p. 161
• Wittcoff (2004), Ch 14 Carbohydrates · pdf p. 483
• Kent (2012), Ch 33 Biomass conversion · pdf p. 1254

See it in the table

Check your understanding

Why did integrating PHB production with a Brazilian sugar mill lower its cost?

In the BIOREF-INTEG study, why did adding 1,3-propanediol to a biodiesel plant raise the required biodiesel price?

Why does Gruber argue PLA cannot be made economically from petrochemicals?