On-site Oxygen Generation vs. Liquid Oxygen (LOX) for Commercial RAS

1. Size the oxygen demand before comparing supply

Every credible comparison starts from a demand profile, not a vendor brochure. Planning practice is to allow roughly 0.35–0.5 kg of oxygen per kilogram of feed delivered, then add biofilter and heterotrophic demand, degassing losses and transfer inefficiency, and finally a peak-load margin for the hours after feeding at maximum standing biomass.

As a scale reference: a facility feeding 1 t/day typically sits around 350–500 kg O₂/day, and a 1,000 t/yr salmon RAS in the 1,200–2,000 kg O₂/day band. That number, together with your electricity tariff, decides the whole comparison — a supply option that is obviously right at 1,500 kg/day is usually wrong at 80 kg/day.

Run a first-pass estimate with the aquaculture calculators (feed load, biomass and oxygen demand), then hold that profile constant while you price both supply routes.

2. How each supply option works

On-site generation separates oxygen from ambient air. Pressure swing adsorption (PSA) passes compressed air through a molecular sieve that retains nitrogen; vacuum-pressure swing adsorption (VPSA) does the same at lower pressure with a vacuum blower and is materially more energy-efficient at larger capacities. Output is typically 90–95% oxygen, buffered in a receiver and fed to cones, low-head oxygenators or a side-stream saturator. The farm owns a plant room, a compressor, a maintenance schedule and an electrical load.

Liquid oxygen (LOX) is produced in an industrial air separation plant and delivered by tanker into a cryogenic bulk tank on site, then vaporised on demand at 99.5%+ purity. Capital exposure is minimal — the tank is usually vendor-owned under a supply agreement — and there is essentially no on-site energy cost or rotating equipment. What you buy instead is a delivered price per kilogram and a dependency on a tanker route.

In practice most commercial RAS facilities above the crossover point end up hybrid: generation as base load, LOX as automatic standby. That is the configuration lenders and insurers expect to see documented in the life-support design.

3. CAPEX: what you actually commit

Planning-grade (class-4, ±30–50%) ranges for a commercial facility. Replace them with quoted package prices before the numbers reach a lender.

Two items are routinely underestimated in generation budgets. The first is the electrical connection: a continuous compressor load frequently pushes a site past its existing supply capacity, and a transformer or feeder upgrade can rival the cost of the oxygen skid itself. The second is the backup: a generation-only design is not a bankable design, so the standby LOX tank or cylinder bank belongs in the same capital line.

4. Cost per kilogram of oxygen

The arithmetic is simple once the demand profile is fixed. Multiply kWh/kg by your tariff, add maintenance and an allowance for compressor downtime, and compare against the delivered LOX price including rental and any minimum-offtake charge. At 0.6 kWh/kg and USD 0.10/kWh, generated oxygen lands near USD 0.06/kg in energy plus maintenance — comfortably under most delivered LOX prices. At USD 0.20/kWh the same plant is close to parity, and the capital no longer pays back inside a normal appraisal horizon.

Model part load explicitly. RAS oxygen demand follows the production cycle, and a generator sized for peak biomass spends part of the year running well below its efficient point. Vendors quote kWh/kg at rated output; ask for the curve, and check whether the design uses multiple smaller trains to hold efficiency at turndown.

5. Reliability, redundancy and failure modes

Generation fails mechanically and electrically. Compressor faults, valve or sieve degradation, dryer failure and grid outages all interrupt supply, and a high-density system consumes its dissolved oxygen reserve within minutes. Mitigations are engineering ones: N+1 trains or a duty/standby compressor, an adequately sized buffer receiver, automatic changeover to LOX, standby generation on the compressor circuit, and alarms with escalation to a named responder.

LOX fails logistically. The equipment is passive and highly reliable, but supply depends on a tanker arriving. Weather, road closures, driver shortages, regional demand spikes and single-supplier markets have all interrupted deliveries. Mitigations are commercial and volumetric: sufficient storage for 7–14 days at design demand, contractual delivery windows and telemetry-based auto-replenishment, plus a second supplier where the market allows.

Whichever route you choose, specify the emergency oxygen path independently: a cylinder bank on a separate regulator, feeding diffusers on the critical tanks, with a documented protocol and periodic drills. Insurers increasingly ask to see it, and it is the cheapest line item in the entire life-support budget.

6. Remote sites and delivery logistics

Distance from the nearest air separation plant is the variable that most often overrides the cost comparison. Delivered LOX pricing carries a freight component that scales with route length and inversely with drop size, so a small facility 400 km from a depot can pay two to three times the price a large facility next to one pays for the same product.

Beyond price, ask the harder question: what happens when the tanker cannot arrive? Ferry-served islands, seasonal roads, single-supplier regions and cross-border routes with customs exposure all convert a cost decision into a continuity decision. On those sites, on-site generation with generous buffer storage is frequently selected even when the pure per-kilogram comparison is marginal, precisely because it removes an external dependency from the life-support chain.

The inverse also holds. An urban or industrial-corridor site with two competing gas suppliers within an hour's drive has genuine price tension and short response times — conditions in which LOX often remains the better commercial answer well above the usual crossover volume.

7. Which option fits your project

Reduced to three questions: how many kilograms of oxygen per day at peak, what does a kilowatt-hour cost at your meter, and how far is the nearest tanker route. Answer those and the decision usually resolves itself — then price both anyway, because quoted spreads on a defined scope regularly differ from the planning ranges above.

8. What to put in the RFQ

• Peak and average oxygen demand in kg/day, with the feed load and biomass assumptions behind them.
• Required purity and delivery pressure at the oxygenation package.
• Guaranteed kWh/kg at rated output and at 50% turndown, with test conditions stated.
• Redundancy configuration: number of trains, duty/standby arrangement, automatic changeover to backup.
• Buffer storage volume and hours of autonomy at design demand.
• Ambient design conditions — temperature, humidity, altitude — which materially affect compressor output.
• Maintenance scope, sieve replacement interval, spares availability and local service response time.
• For LOX: delivered price per kg at your drop size, tank rental, telemetry, minimum offtake, escalation clause and delivery SLA.
• Commissioning, operator training and warranty terms for both routes.

Quotes built on that specification are comparable. Quotes built on "we need an oxygen system" are not, and the difference typically shows up as a six-figure variation during construction.