Reverse Engineering Chlor-Alkali Unit Economics | Chapter 1: The Stasis

As much as one may be interested in the $50bn chlor-alkali industry, which produces caustic soda, a chemical critical for alumina refining, soap & paper industry, wastewater treatment, desalination, pharmaceuticals and more, one will find it an expensive challenge to find data and understand the inner economic workings of this industry.

No chlor-alkali plant operator is open to discussing the market price or the unit economics. I’ve talked to several. And I’m not paying $6000 to a consulting firm.

The market price of caustic soda (NaOH) is $450-700/ton. What is the production cost? There’s no clear answer to this, and even LLMs are not useful. So, I had to reverse engineer the chlor-alkali unit economics. This is part one.

A quick ChatGPT query to pull up the reported cost of a chlor-alkali plant and its production capacity is a good starting point. Here’s what it says:

How far can we go with this data? Surprise, surprise, it gives us a lot. (TPA is tons per annum.)

Let’s assume a few things.

1. Cost of electricity = $60/MWh (the cheapest industrial grid price in the US)
2. Yield of NaOH = 2.5 MWh/ton-NaOH
3. Lifetime of plant = 10 years (I got this from a chlor-alkali plant operator)
4. Non-electric operating cost = $169/ton-NaOH (salt, steam, membranes, labor, maintenance, overheads — built up line by line from the European BAT reference document for chlor-alkali)

Now, digging in…

What is the plant’s CAPEX per ton-NaOH per year? The $/TPA?

This is straightforward.

$/TPA = reported cost / production capacity

The points fall into two bands. The Indian projects and the Egyptian project sit at $155-$661/TPA, whether they are brownfield expansions or full plants. The Western newbuilds and conversions — Westlake, Covestro, Cape Igarassu — plus the Indonesian feasibility study sit at $1,060-$1,806/TPA. Geography and project type, not size, decide which band you land in.

Within either band, $/TPA doesn’t seem to be inversely proportional to production capacity and goes against the economies of scale. The smallest datapoint here, a 16 ktpa expansion at $155/TPA, is the cheapest per ton of capacity on the chart, and the 500 ktpa study is among the most expensive. Something interesting is going on. If $/TPA is truly independent of production capacity, consumers with small-scale NaOH consumption can entertain a new, smaller chlor-alkali plant closer to them to avoid paying for transportation from far-away plants or storage. That is good news and is probably an indication of the stasis in chlor-alkali plant costs.

Moving on.

What is the plant’s load CAPEX/kW? The $/kW-load?

$/kW-load = ($/TPA) [$-year/ton] * (365 * 24) [h/year] * (1 / yield) [ton/MWh]

What is the production cost of NaOH/ton? The $/ton-NaOH?

$/ton-NaOH = CAPEX + electricity + everything else
= ($/TPA) [$-year/ton] * (1 / lifetime) [1/year] + (yield * electricity-cost-per-ton) [MWh * $/MWh-ton] + $169

An earlier version of this post set that last term to zero and looked only at capital and power. That was too generous. Salt, the steam needed to concentrate 32% cell liquor up to merchant-grade 50% caustic, membrane and gasket replacement, maintenance, operating labor, plant overheads, insurance and corporate allocation are all real, and they add up to about $169 per ton before you buy a single kilowatt-hour. One caveat worth stating: the fixed lines in that build-up come from a 500 kt/y plant, so applying the same $169 to the small expansions below flatters them. Real small-plant labor and overhead per ton are higher.

These are the key unit economics numbers. The $/kW-load tells us the cost of the plant based on the power input. The $/ton-NaOH tells us how expensive the caustic soda is to make and how much margin you can make at a certain market price.

Production cost lands between $335 and $500/ton across this set. Against a market price of $450-700/ton, that is a real but unspectacular margin — and the plants at the top of that range are underwater at the bottom of the price range. This is a commodity business behaving like one.

The load CAPEX / kW for overall plants, not expansions, splits into two clusters rather than one range. The Indian projects sit at $1,100-$2,300/kW. The Western newbuilds and conversions — Westlake, Covestro, Cape Igarassu — and the Indonesian feasibility study sit at $3,700-$6,300/kW, three to six times higher for the same chemistry. Whichever cluster you believe applies to you, that CAPEX number has some interesting implications for whether you need batteries if you want to use a solar/batteries play—more on this in part two.