Duckweed Cultivation Cost: What the Real Numbers Say

Ask what duckweed cultivation cost looks like and the literature answers three times, in three different orders of magnitude.

A transparent cost model built on Flemish price quotes puts production at EUR 3.55 per dry kilogram at a yield of 10.7 dry tonnes per hectare per year. That is EUR 3,550 per dry tonne. An industry deck aimed at protein buyers quotes $400 to $700 per dry tonne unprocessed at a productivity of 25 to 35 tonnes per hectare per year. And the most rigorous techno-economic analysis in the field lands on a minimum duckweed selling price of $7.69 per dry megagram, which is $7.69 per dry tonne.

Roughly a factor of 460 between the top and the bottom. Nobody is lying. They are pricing four different machines against four different denominators, and one of them is quietly counting a revenue line the others do not have.

Here is what actually separates them, and what it means for costing a system on your own site.

The credit is the single biggest lever

The Penn State analysis models three 100 hectare ponds treating wastewater, using nth-plant economics, which assumes several plants of the same design have already been built and the cost overruns of a first-of-a-kind are behind you. Pond construction accounts for 55.6 percent of capital expenses and land for 15.8 percent. Harvesting accounts for 90.4 percent of operating expenses. Land is valued at $3,000 per acre, plant life at 30 years, IRR at 10 percent.

Then comes the clause that almost never travels with the number. The $7.69 figure holds only if the construction and operation of the activated sludge stage the duckweed pond replaces are credited back to the system. The same paper’s sensitivity analysis makes the size of that assumption explicit: the by-product credit moves the mean selling price by plus or minus 382 percent, more than any other input including IRR and labour cost.

The Flemish figure has no such credit. It is a full production cost, everything in, nothing subtracted. So EUR 3,550 per dry tonne and $7.69 per dry tonne are not two estimates of the same quantity. One is what it costs to grow the plant. The other is what you would have to sell it for, after the water treatment has already paid for most of the pond.

The denominator is a design choice, not a convention

The second reason the numbers disagree is that cost per dry kilogram is only one of at least three legitimate ways to express this.

Oron’s 1994 facility assessment for a 30,000 cubic metres per day plant reports cost per cubic metre of wastewater treated: about $0.0413/m3, against a return of $0.0418/m3. Net benefit, $0.0005 per cubic metre. That is a rounding error, and it should be read as one: the honest conclusion of that study is that the system roughly broke even, not that it printed money.

Design guidelines for duckweed-based treatment express it per person instead, putting capital cost at 7.9 to 9.7 USD per capita on 1.5 to 1.8 square metres per capita with a retention time of 15 to 18 days.

Cost per kilo, cost per cubic metre, cost per capita. Three questions. If you have not decided which one your project is answering, no figure you find will settle anything, because you can move the apparent cost by a factor of ten just by changing what you divide by.

The five line items that actually move the number

Strip out the framing and the same five items dominate every model in the set.

Land and earthworks

The largest single block, everywhere. Pond construction and land are 71.4 percent of capital expenses in the Penn State model. In the Flemish lagoon calculator, land, excavation and foil together are 70 percent of annual cost, in a model whose maximum annual figure is EUR 38,883. Oron prices the components directly: land at $1,000 per hectare and excavation at $1.30 per cubic metre, in 1994 money.

If the site already has ponds, lagoons or a decommissioned basin, a large fraction of the published cost simply does not apply to you.

The liner

This is the finding most worth knowing and the least widely repeated. In the Flemish model, foil lining is responsible for about 40 percent of the cost of Lemna. Remove it and production falls from EUR 3.55 to EUR 2.13 per dry kilogram.

Whether you can remove it is a soil permeability and permitting question, not a budget question. It is also why duckweed systems price above plain stabilisation ponds: the same guidelines find duckweed treatment about 25 percent more expensive than waste stabilisation ponds on both investment and operating cost, specifically because of seepage prevention and the need for skilled labour.

Harvesting

Ninety point four percent of operating expenses in the Penn State model. Not a detail, the entire operating cost structure. The Flemish model reaches the same place from the other direction, noting that labour for harvesting with pumps and feedbags is a major part of operational cost and that automated conveyor systems are what would be needed to compete with soy at scale.

This is where a cost model meets hardware, and where the harvest regime you chose for biological reasons turns into a labour line. Before quoting duckweed harvesting equipment, price the labour it is meant to displace, because that number is what justifies the machine.

Nutrients

On a clean stream, fertiliser is a real cost. The Mirzapur programme in Bangladesh spent about $1,800 per hectare per year on fertiliser to sustain its crop, at 1992 prices.

On an effluent stream, this line item inverts. Somebody is already paying to remove that nitrogen. The nutrient is not an input you buy, it is a liability you are being paid to consume, which is the whole reason duckweed wastewater treatment economics look different from duckweed farming economics.

The yield you assumed

Every figure above is a numerator over an assumed productivity, so the yield assumption silently sets the answer.

The Flemish model separates operating cost as EUR 646 per dry tonne, fixed per tonne, from capital cost, which is fixed per hectare. Doubling productivity from 10 to 20 tonnes per hectare per year therefore halves the capital component per tonne and drops the price per kilo by about 40 percent, the same magnitude of saving as removing the liner entirely.

Which makes the yield number the most dangerous input in the model. A 2026 review of duckweed in controlled environment agriculture finds that short-term laboratory experiments frequently overestimate long-term productivity, with relative growth rates declining as cultivation duration and system complexity increase. A cost model built on a headline duckweed growth rate from a short indoor run is not conservative or aggressive. It is measuring a regime a working plant never stays in.

Cost per kilo is usually the wrong question

For most operators reading this, biomass revenue will not close the business case, and the published models say so bluntly.

The Flemish analysis compares a production cost of EUR 2 to 4 per kilogram against a fresh feed sale price of EUR 0.22 per kilogram and concludes that over 90 percent of the value lies in nitrogen and nutrient recapture. Selling the crop recovers a small fraction of what it cost to grow.

Now price the same system as a treatment asset. A modelled case for a Flemish pig farm builds duckweed lagoons for EUR 38,003 per hectare covering both capital and operating expenses, reaching EUR 7,400 in first-year total outlay for that operation once coagulant and dredging are added. The feed value offsets part of it, bringing the annual treatment cost to about EUR 970. Against the EUR 23,850 the 1,500 pig place farm was already paying to treat and spread its manure, that is roughly 4 percent, with payback in under a year.

Same ponds. Same plant. The economics close because the comparator changed from a commodity price to an avoided compliance cost. That is the reframe that decides most duckweed projects, and it is why costing this system starts with your current treatment invoice rather than with a yield table. The same logic drives the harder case of duckweed on manure, where the incumbent cost is highest.

What it is competing against, honestly

Duckweed treatment is not universally cheaper. The Bangladesh experience puts design and construction at about one tenth the cost of a conventional system, but on roughly three times the land. Adding multiple inlets and recirculation raises land requirement by about 30 percent and capital by about 20 percent.

So the shape of the answer is consistent across every source here. Duckweed wins where land is cheap or already owned, labour is affordable or automatable, and the nutrient load is a cost the operator is currently paying to remove. It loses where land is expensive, the climate shortens the season, or the only revenue on offer is biomass at commodity prices.

The revenue side has a hard constraint

A cost article naturally reaches for the feed line to close the gap. On a remediation stream, that line is zero.

At SERAPH this is published policy rather than preference: remediation biomass is never sold or blended as food or feed, and the two streams are separated from the first sensor reading. The design guidelines reach the same caution from the technical side, warning that cultivation in wastewater ponds has a negative effect on duckweed’s use as animal feed because the biomass may contain toxic organic compounds and heavy metals, so it never leaves the remediation stream.

The consequence for a spreadsheet is exact. If the water is wastewater, manure or anything contaminated, delete the feed revenue row and make the case close on treatment alone. If you want a feed output, the clean stream that produces duckweed protein is a separate system with its own water, its own containment and its own capital cost. It is not a downstream option on the dirty one, and costing it as though it were is how a business plan fails an auditor rather than a spreadsheet.

What we can and cannot tell you

SERAPH is at TRL 3, proof of concept, with outdoor field validation still ahead. So we do not publish a cultivation cost figure of our own, and we will not until we have measured one. Every number above belongs to somebody else, with its year and its currency attached, because 1992, 1994 and 2021 prices are all in this article and quoting any of them as current would be the most expensive mistake available here.

What we do have is the layer underneath the cost model. Our knowledge base indexes 256 duckweed papers across 31 research areas in 5 domains, each area carrying a coverage weight, because the evidence in this field is unevenly distributed and cost models inherit that unevenness without noticing. And we build measurement first: our vision system reconstructs a cultivation tray in 3D from a single capture using 170,000 surface points drawn from 5.9 million captured and tells species apart in a live frame.

That exists for the reason this whole article circles. Assumed yield is the largest uncertainty in every cost figure published, and you cannot narrow it by sampling occasionally. AI proposes, the bench validates.

How to cost your own site

Nine inputs turn the question into a number you can defend:

1. The denominator: per dry tonne, per cubic metre treated, or per capita served.
2. The credit: what treatment cost does the pond displace, and is it in the model.
3. Land price, and whether excavation is needed or a basin already exists.
4. Liner or no liner, decided on soil and permit grounds before cost.
5. The harvest labour rate, and how frequency multiplies it across a season.
6. The nutrient source, and whether it is a purchase or a liability someone pays you to take.
7. The yield assumption, with a note on where it came from and how long that run lasted.
8. The containment decision, which sets whether a revenue row exists at all.
9. Discount rate and asset life, since a 25 or 30 year horizon quietly does much of the work.

Answer those and the conflicting figures stop conflicting. They become what they always were: other people’s sites, priced honestly, in currencies and years that are not yours. That is also the point where duckweed cultivation is a design problem before it is a budget stops being a slogan, and where a duckweed cultivation business plan can be written from evidence rather than from a template.

FAQ

Is duckweed cheaper than soybean meal as a protein source?

Not at published European cultivation costs. The Flemish model puts operating cost at EUR 646 per dry tonne against a 2024 soybean price range of EUR 350 to 450 per tonne, and expects scale plus a decade of cultivation research to close that gap rather than treating it as already closed. The industry figure of $400 to $700 per dry tonne unprocessed sits in similar territory.

Do you need a pond liner to grow duckweed, and what does it cost?

In the Flemish cost model, foil lining is responsible for about 40 percent of the cost of Lemna, taking production from EUR 3.55 to EUR 2.13 per dry kilogram when removed. Whether it can be removed depends on soil permeability and the discharge permit, not on the budget, and seepage prevention is also part of why duckweed systems price above plain stabilisation ponds.

How long does a duckweed treatment system take to pay back?

In the modelled Flemish pig farm case, under one year. That result is set by the EUR 23,850 the farm was already spending annually on treating and spreading manure, not by anything the duckweed sells for. A site with a low incumbent compliance cost will not reproduce it.

Is duckweed treatment cheaper than conventional wastewater treatment?

The Bangladesh experience puts design and construction at roughly one tenth of a conventional system, on about three times the land. Compared with waste stabilisation ponds specifically, duckweed runs about 25 percent more expensive on investment and operating cost because of seepage prevention and skilled labour. The price of land usually decides which comparison matters.

Why do published duckweed cost estimates vary by so much?

Mostly because of what each one credits and counts. A minimum selling price calculated with a wastewater treatment credit applied is not comparable to a full production cost with no credit, and the sensitivity analysis in that same study shows the credit alone shifts the price by plus or minus 382 percent. Currency and vintage do the rest.