Techno-economics

Almost every cost figure quoted for carbon capture is incomparable to almost every other one, and the reasons are systematic rather than accidental. Reading this literature is mostly a matter of knowing which question a given number answers [1].

Two denominators, two different numbers

The first thing to establish about any figure is what sits under the division bar.

Cost of CO2 captured divides the added cost by the tonnes of CO2 the capture equipment removes from the gas stream. It is an equipment-level number.

Cost of CO2 avoided divides the added cost by the reduction in emissions relative to a reference plant without capture. Because capture consumes energy, a plant with CCS burns more fuel per unit of output, so the emissions avoided are always fewer than the tonnes captured. Avoided cost is therefore always the larger number, and it depends on which reference plant you compare against, a choice that is external to the technology being costed.

Avoided cost is the more meaningful measure for policy and the more frequently quoted, which is precisely why the reference plant needs to be stated alongside it.

What actually drives the spread

Published estimates of the cost of CO2 avoided run from 46 to 99 dollars per tonne for coal with post-combustion capture, 59 to 143 for natural gas combined cycle, and 38 to 84 for coal gasification with pre-combustion capture [2].

Those bands are not measurement error. They reflect real differences in plant location and site conditions, assumed capacity factor, plant size, fuel price, financial assumptions, and whether results are reported in constant or current dollars [2]. Two studies can be individually careful and still differ by a factor of two because they answered slightly different questions. This is the problem a proposed common costing methodology was written to address [3].

First-of-a-kind versus nth-of-a-kind

The single largest source of confusion is the distinction between what the first plant costs and what a mature plant is projected to cost. First-of-a-kind costs for real projects run significantly above estimates for a mature nth-of-a-kind plant, and the magnitude of that gap is very difficult to predict for any particular project or technology [2].

A number without that label attached is close to uninterpretable. Advocates tend to quote nth-of-a-kind; critics tend to quote first-of-a-kind; both can cite a real source.

Learning rates, and the limits of extrapolating them

The usual bridge between the two is a learning rate: the fractional cost reduction per doubling of cumulative capacity. The empirical literature on learning rates across electricity supply technologies shows wide variation in the rates observed and in how they are estimated [4]. Applying a rate borrowed from a manufactured, mass-produced technology to a large bespoke chemical plant is an assumption, not a measurement, and it does most of the work in any projection that shows costs falling steeply.

Two illustrations from elsewhere on this site show how much rides on it. E-fuel mitigation costs are 800 to 1,200 euros per tonne today, with a projected 2050 range of 20 to 270 euros per tonne conditional on large-scale deployment, a future span of more than an order of magnitude [5]. And a carbon-neutral chemical industry via utilisation would require more than 18.1 petawatt-hours of low-carbon electricity, roughly 55 percent of projected global generation in 2030 [6], which makes electricity price the dominant term in that cost rather than any equipment learning curve.

Open challenges

• Non-comparability is the default. Assume two cost figures are incomparable until the metric, reference plant, boundary and dollar year are established [1][3].
• Transport and storage are often excluded. Capture cost and full CCS cost are different quantities, and which one is quoted is not always stated [2].
• Learning rates carry the projections. Where a forecast shows steep cost declines, the learning rate assumption is usually doing more work than the engineering [4].
• Energy price dominates for electricity-intensive routes. For conversion and direct air capture, the cost is largely a pass-through of low-carbon electricity price, so it is bounded by the energy system rather than by the process [6].