The delivered-cost case

Our model builds the cost stack from the mill gate to the Singapore delivery point. Every figure below is declared in the figures manifest with its arithmetic.

At the $20 price the pipeline route earns $7.67/MMBtu over delivered cost and the marine route $3.17. Both are before the 9% of revenue the project pays as a joint-venture fee and administrative charge, which is $1.80/MMBtu, so the pipeline route's margin net of everything is $5.87.

Capital for a 100 tonne-per-day first phase is $32.37M on the pipeline route, of which $15.63M is upgrading and gathering, $4.97M desulfurization, $4.58M compression and $7.19M for a collection fleet of 20 trailer sets. The marine route adds $11.81M of liquefaction for $44.18M.

Two results in that table deserve attention.

The pipeline route shows no economy of scale between 100 and 200 tonnes a day on the way this capital was scaled. Upgrading trains are treated as modular above about 2,000 Nm3/h, compression is treated as modular, and the collection fleet scales with volume, so doubling the hub doubles the capital. That is a convention rather than a measurement: the six-tenths rule is applied inside a cluster and to liquefaction, and not to compression, desulfurization or the fleet. No other answer was reachable once the capital was built that way, and a reader should hold the result as a property of the method until vendor pricing at both scales tests it. The scale economy in this business lives in liquefaction, which the pipeline route does not use. That undercuts the aggregation-to-reach-scale argument: aggregation buys access to enough feedstock, but on this route it does not buy a better unit cost.

Two liquefaction figures deserve the same scrutiny the upgrading capital gets, and they did not receive it in the first pass. The client's liquefaction capital works out at $534 per tonne of annual capacity, against a published small-scale range of $200 to $1,600 (MDPI); the top of that range is three times the figure used. The $4.00/MMBtu liquefaction operating penalty comes from the brief and from no independent source. Neither is placed within its range by anything here: $534/tpa sits in the lower third of a $200 to $1,600 band, so a correction could go either way, and the $4.00/MMBtu operating cost has no published range at all. The gap between the routes is $4.10 to $4.50/MMBtu, so the conclusion that liquefaction is what makes the marine route fail would survive a large correction in either direction, but it rests on a figure taken from the brief and tested against nothing.

The single largest uncertainty in the capital is the upgrading equipment. A published Western benchmark of $1.66M for a 500 Nm3/h pressure-swing plant, peer-reviewed and published in December 2024, scaled at the six-tenths rule, puts the 100 tonne-per-day fleet at $15.63M. Applying the client's own vendor pricing to the same configuration gives $4.34M, 3.31 times lower. At the client's pricing the pipeline route returns 25.0% rather than 16.3%, and the marine route still fails. Which figure is right decides whether this is a comfortable Go or a marginal one, and it can be settled with three vendor quotations.

One modeling choice deserves to be a condition rather than an assumption, because it is what keeps this case above the No-Go floor.

The upgrading capital above assumes mills are grouped four to a site, sharing one upgrading train of 1,876 Nm3/h reached by about 15 km of raw-gas gathering line. Both of those numbers are estimates. The alternative, a train at each mill sized to that mill's own 469 Nm3/h, costs $25.08M instead of $15.63M and carries higher maintenance with it; on that configuration the return falls to 11.1%, still above the 8% No-Go floor but below the 13% Go bar. So whether four mills can be reached by one gathering line is not a modeling detail. It is the difference between a project worth testing and one that is not, and it belongs in the first stage of work rather than in a spreadsheet.

Whether it holds is a question about geography, and the mill census in the testing plan answers it directly: a cluster needs four mills within about 15 km of each other, and Southern Thailand's mills are distributed along a corridor rather than evenly, so some clusters will form and some will not. The honest reading is that the true capital sits between $15.63M and $25.08M, nearer the first where mills are dense around Surat Thani and Krabi and nearer the second at the corridor's edges.

Two cross-checks on the cost stack are worth stating, because the two largest soft numbers in it were each built twice.

The collection cost was derived from the haul rather than borrowed. At 33.3 trailer loads a day over a 150 km one-way haul, with the workbook's own 3-tonne payload, 3.3 km per liter and 32 THB per liter, and all-in truck cost taken at three times fuel, the figure is $1.51/MMBtu. A published virtual-pipeline penalty of $1.50 to $2.50 brackets it. Two methods agreeing within 6% is the reason this line is quoted with any confidence.

The separated carbon dioxide is a real co-product and the client's model books it at zero. Upgrading strips 0.0443 tonnes of carbon dioxide per MMBtu of product, which at the 100 tonne-per-day scale is 174 tonnes a day. Merchant liquid carbon dioxide sells at $335/t in India and $238/t in Japan, the nearest markets with published prices (IMARC). Capturing, liquefying and storing it costs an estimated $16.87M at this scale, so it is not free revenue. At a $50/t netback at the hub fence the stream does not pay for itself: the project returns 14.9% with it against 16.3% without, so at that price it consumes more capital than it earns. It needs $32/t to hold the project at the 13% bar and more than that to be worth building. That figure is 9.6% of the India price. Whether a Southern Thai producer can net it depends on whether an industrial gas partner has customers within reach, and it is not in the base case, because a netback nobody has quoted cannot carry one.

What the gas is worth if it never leaves Thailand sets a floor under all of this. Thai retail natural gas ran at about 333 THB/MMBtu, roughly $10/MMBtu, in December 2025 (Intratec), and the client's own model prices compressed biomethane domestically at $14.42 against liquefied petroleum gas parity. Plant-gate cost here is $10.33. A domestic sale therefore covers cash cost and very little else, which is why the export question exists. It also means the downside of a failed export route is not zero: the plant has a domestic outlet at around cash breakeven rather than a stranded asset.