Why correctly-sized components still build systems that fail and what the Excel farmer and the catalogue engineer have in common
Walk into almost any recirculating system that is struggling and you will find the same thing: every component, taken on its own, is correctly sized. The drum filter matches the solids load. The biofilter carries the surface area the calculation called for. The oxygen cone, the pumps, the degasser, each bought to specification, each defensible on paper. And your fish are still telling you that there is something wrong.
The industry has become genuinely good at sizing components. Thirty years of design tables, supplier datasheets and well-worn engineering texts mean almost nobody undersizes a biofilter by accident anymore. That is real progress and it deserves to be said plainly. For most species at most loadings, the engineering of the individual unit process is a solved problem.
But a recirculating system is not simply the sum of its components and that is the line the catalogue will never print. Every vendor sizes the box they are selling. Nobody is paid to own the water as it travels between the boxes, the hydraulics, the transitions, the head loss, the dead zones, the few seconds of contact time that decide whether a gas actually leaves the water or simply rides along to the next tank. A RAS does not fail on the spec sheet. It fails in the seams. And the seams have no supplier.
It fails there because we design for a fish that does not exist. I have made this case before about the spreadsheet, the seductive, forest-green world where dissolved oxygen sits politely at 9 mg/L, feed conversion holds at a bankable 1.1, and mortality arrives on a tidy linear schedule. Biology does not follow a linear VLOOKUP and a farm that only works on a screen does not work at all. But the spreadsheet is only half the story because the same hubris that builds the Excel farm also sizes the steel. The Excel farmer and the catalogue engineer are the same person in different software: both model the clean average and schedule the catastrophes politely.
Run the mass balance at nominal biomass and nominal feed load and the numbers behave. Reality does not. A kilogram of 32 percent protein feed releases roughly thirty grams of ammonia-nitrogen into the water (Ebeling et al., 2006); but fish do not feed in a smooth line, they feed in pulses and a nitrifying biofilter is a living population that cannot be told to accelerate. The bacteria lag. Dissolved carbon dioxide, now recognised as one of the real ceilings on land-based intensification and the parameter most often left under-stripped, climbs during those same pulses, because the degasser that looked adequate at average flow was never asked the question at the peak. Summerfelt and others have spent two decades making exactly this point. None of it is a component failure. Every unit is doing its job. The system is failing in the space between the jobs.
The cost of that gap is rarely dramatic, which is precisely why it is expensive. It is the suppressed growth that never trips an alarm. It is the feed conversion that drifts from the spreadsheet’s 1.1 to a real-world 1.25 over a season before anyone links it to an afternoon pH curve. It is the mortality after grading that the post-mortem quietly blames on the fish. It is capital sunk into correctly-sized equipment that underperforms because nobody designed the relationships between the parts. A farm does not lose money on the components it bought; it loses money on the integration it assumed it was getting.
So, before the next system is specified, the more useful question is not whether every component is sized correctly. It almost certainly is. The question is who owns the seams. Because a recirculating system is not a shopping list and it is not a spreadsheet, it is a single hydraulic and biological argument that has to hold together at the worst hour of the worst day. That is not something you buy off a datasheet or model in a cell. It is something you design and then you go and listen to the fish.
-ends-
