How to Size a Filter for Your Koi Pond

1. There Are Three Different Meanings of “Filter Size”

When koi keepers say a filter is 'too small,' they may be describing three very different limitations. Understanding which limit exists is essential because adding more biological media will not solve a hydraulic bottleneck, and installing a larger pump will not solve a biological reactor that has insufficient active capacity.

A well-designed koi system has to satisfy all three simultaneously. In high-performance ponds, gas exchange and oxygen delivery should be treated as a fourth design constraint because fish, heterotrophic bacteria and nitrifying biofilm all consume oxygen. A biofilter can be physically large yet biologically weak if its water is oxygen-poor or overloaded with decomposing organics.

2. Start by Defining the Pond You Are Actually Designing For

Before choosing equipment, define the design case. Filtration should be sized for the heaviest realistic load the pond is expected to carry, not for today's minimum stock. A new pond containing small tosai may seem lightly loaded, but those fish can become several times heavier while their feeding rate increases substantially. A dealer pond may receive a large shipment overnight. A show-quality collection may be fed aggressively in warm months and much less in winter.

Record the following design inputs before making equipment decisions:

3. Measure Pond Volume Properly

Pond volume is not enough to size the whole filter, but an accurate volume remains essential. It is used to calculate turnover, interpret water-exchange rates and dose salt, medication or other treatments safely. Guessing the volume of an irregular pond can create both engineering and health-management errors.

For a simple rectangular pond, volume is length × width × average water depth. For circles, curves, shelves and irregular walls, divide the pond into simpler shapes and add the volumes. The most reliable practical method after construction is often a calibrated water-meter fill, provided the pond can be filled from known zero and all connected chambers are accounted for consistently.

Do not confuse nominal construction dimensions with actual operating volume. Freeboard, sloped floors, rounded corners, filtration chambers and the normal running water level can change the number materially. If a pond is advertised as 20,000 litres but actually contains 16,500 litres, a one-hour turnover target requires 16,500 litres per hour of true delivered flow, not 20,000. Conversely, medication dosed for the wrong volume may expose koi to an unintended concentration.

4. Biomass Matters More Than Fish Count

Ten 80 cm koi do not create the same load as ten 25 cm koi. Fish count is therefore a poor sizing metric unless size is also considered. Total biomass—the combined mass of all fish—is a better way to describe how much living animal tissue is being supported by the system.

Biomass does not need to be known to the nearest kilogram for a hobby pond. Its value is in forcing the designer to think about the difference between many small koi and a smaller number of very large koi. When detailed weights are unavailable, estimate conservatively from known fish lengths and body condition, then add a future-growth allowance. Dealer and grow-out systems should revisit biomass estimates frequently because stock changes rapidly.

Biomass is especially useful for oxygen and emergency planning. However, for biological filtration, peak daily feed is often even more useful because the nitrogen entering the system is strongly linked to protein intake and feeding rate. Two ponds with similar biomass can impose different filter loads if one is maintained lean and the other is fed heavily for growth.

5. Peak Daily Feed Is the Most Useful Practical Sizing Variable

A koi's waste begins with what it eats. Feed that is digested and metabolised contributes dissolved nitrogenous waste, much of which ultimately appears as total ammonia nitrogen (TAN). Feed that is not retained as growth also contributes faecal and dissolved organic load. Uneaten food adds an even more direct burden.

This is why commercial recirculating-aquaculture design frequently expresses treatment capacity relative to kilograms of feed per day. For koi ponds, the same concept is extremely useful: calculate or estimate the highest daily feed the system must safely process, then verify that the mechanical and biological stages are capable of handling that load.

Peak means a realistic sustained peak, not an unusual one-off handful of extra pellets. For a growth pond, it may be the warm-season feeding level used for several weeks. For a dealer facility, it may be the normal maximum stock immediately after an import. For a private collection, it may be the feeding rate after all current koi reach their intended mature size.

If feed is never measured, filtration is being managed partly blind. Weighing one day's ration periodically is simple and gives the keeper a powerful reference point: water quality can then be related to an actual loading number instead of impressions such as 'I feed quite a lot.'

6. A Practical Feed-to-Ammonia Design Concept

Aquaculture engineering references commonly estimate ammonia production as a percentage of feed input for design calculations. The exact fraction varies with feed protein, digestibility, species, growth, temperature and other factors. Older recirculating-aquaculture design examples commonly use approximately 2.5 to 3 percent of feed weight as TAN-N for rough calculations. This should be treated as an engineering estimate, not as a universal biological constant for every koi food.

The calculation is useful because it converts daily feed into a mass load that a biofilter must oxidise. For example, if a system is fed 1.0 kg per day and a conservative design estimate of 3 percent TAN-N is used, the estimated TAN-N production is about 30 g per day. A pond fed 2.0 kg per day would therefore be designed around roughly 60 g TAN-N per day under the same assumption.

The important step is what comes next: match that load to a biofilter performance rate that is valid for the chosen technology and operating conditions. Do not divide it by a random internet figure for 'surface area required.' Biofilter performance is affected by TAN concentration, dissolved oxygen, pH, alkalinity, temperature, organic loading, media type, hydraulic loading and biofilm condition.

A useful conceptual equation is:

Required biological capacity ≥ peak daily TAN-N production × safety factor

The safety factor should reflect uncertainty, fish growth and the consequences of failure. A lightly stocked garden pond may tolerate a smaller reserve than a valuable dealer system in which one missed feeding-related load calculation can expose a large number of koi to ammonia or nitrite.

7. Do Not Size a Biofilter by Media Volume Alone

Statements such as '100 litres of moving bed is enough for 500 g of food' can be useful only if the underlying conditions are known. Media volume by itself does not tell you how much ammonia will be converted. Different media have different protected surface areas, void spaces and biofilm behaviour. Different reactors provide different oxygen transfer, mixing, contact patterns and solids control. Even the same media can perform differently in two installations.

The strongest sizing data are performance ratings expressed as a mass of TAN-N removed per unit of media, reactor volume or effective surface area per day, under stated conditions. Peer-reviewed design literature and reputable manufacturer testing are more useful than unqualified marketing surface-area numbers.

For example, SRAC design literature reports very different volumetric nitrification rates for different recirculating-aquaculture biofilter types. That difference illustrates why a cubic metre of one medium cannot automatically be treated as equivalent to a cubic metre of another. The reactor configuration is part of the filter.

When a media supplier provides a feed-capacity rating, ask what it assumes: feed protein, water temperature, TAN concentration, oxygen, pH, reactor fill fraction, prefiltration and whether the value is a conservative operating rating or a laboratory maximum. A filter should be designed around dependable operating capacity, not a theoretical best-case number.

8. Mechanical Filtration Must Be Sized for Mass Load and Flow

Mechanical filtration is often chosen by micron rating alone: 70 microns, 100 microns, 200 microns and so on. Micron size matters, but it is only one dimension of capacity. A screen that captures very fine particles can still be undersized if it has too little screen area for the incoming flow and solids load.

Mechanical capacity depends on at least four variables: the actual water flow through the unit, the mass of solids arriving per hour, the particle-size distribution, and how quickly the captured waste is physically exported from the system. A high-stock koi pond that feeds heavily can overwhelm a small screen even if the stated maximum flow rate looks adequate on paper.

For an RDF, sizing should therefore consider both the manufacturer's recommended flow at the intended water-level differential and how often the drum is expected to rinse at the design waste load. A drum that is continuously washing at normal operation may technically pass the flow but has little solids-handling reserve, wastes more water and wears components faster. The same logic applies to sieve screens, static-bed filters and settlement chambers: capacity has a maintenance dimension as well as a hydraulic one.

The best mechanical filter is normally located early in the treatment train so captured faeces and food are exported before they fragment, dissolve and become oxygen-consuming organic load inside the biofilter.

9. Turnover Rate: Useful, but Not a Complete Design Rule

Turnover is the time required for a volume of water equal to the pond volume to pass through the main filtration circuit. If a 20,000-litre pond receives a true 20,000 L/h through the treatment system, its nominal turnover is one hour. At 10,000 L/h, nominal turnover is two hours.

Koi literature often uses a turnover between roughly one and two hours as a practical design starting point, with faster turnover commonly used on intensively stocked or high-performance systems. It is better to regard this as a hydraulic framework than a biological sizing formula. Sending the pond through an undersized biofilter twice per hour does not magically create biological capacity, and a huge biofilter with insufficient circulation can still leave parts of the pond poorly swept.

Turnover also assumes the water is well mixed. In a real pond, some water can short-circuit from a return back to an intake while other areas move slowly. Bottom-drain placement, skimmers, returns, pond geometry and directional currents determine how effectively waste reaches the filtration system.

The correct question is therefore not simply 'What is my turnover?' but 'What true flow is passing through each treatment path, and does that flow collect waste from the whole pond while staying within the operating range of each filter component?'

10. Use Actual Delivered Flow, Not the Number on the Pump Box

Pump labels normally state maximum flow under very favourable test conditions. Once a pump has to lift water, push through pipework, valves, elbows, UV units, heat exchangers, pressurised filters or waterfalls, the delivered flow decreases. This is head loss.

Koi Organisation International has published practical examples in which a pump delivered substantially less water in a real filter installation than its nominal rating. The lesson is fundamental: turnover calculations must use measured or curve-derived system flow at the actual operating head, not the pump's zero-head headline number.

Use the pump performance curve and estimate static plus friction head during design. After commissioning, verify flow wherever practical with an inline flow meter, calibrated vessel/timing method on suitable outlets, or another dependable measurement. If a supposedly one-hour-turnover pond is actually moving only 60 percent of the expected flow, every downstream assumption changes.

Oversizing pipework can reduce friction dramatically, but gravity-fed lines also need sensible velocities and geometry so waste does not settle in long horizontal runs. Hydraulic design should therefore be treated as a system: pipe diameter, line length, bends, valves, elevation changes, filter resistance and pump curve all interact.

11. Gravity-Fed and Pump-Fed Systems Are Sized Differently

In a gravity-fed koi system, bottom drains and skimmers normally feed prefiltration by gravity. The pump is positioned after the gravity-fed stages and returns treated water to the pond. Flow through the gravity lines is created by a small water-level difference between pond and filter chambers. Each line must be able to supply the intended flow without excessive drawdown or solids settlement.

In a pump-fed system, the pump lifts dirty pond water before mechanical separation. The mechanical filter must therefore accept the pump's delivered flow and deal with solids that may have been broken into smaller particles by the impeller. Pump-fed layouts can work well, but the filter sequence and maintenance burden are different.

Do not apply the same flow rating blindly to both arrangements. An RDF, sieve or chamber may have different operating constraints depending on whether it is gravity-fed, pump-fed, installed with a particular water differential, or placed at a specific elevation. Use the manufacturer's hydraulic specification for the exact configuration.

12. Size Each Intake and Return Circuit, Not Just the Total Flow

Large koi ponds often use several bottom drains, one or more skimmers and multiple returns. The total flow can look adequate while one circuit is underperforming. For example, a 30,000 L/h system split across three bottom drains is not automatically providing 10,000 L/h through each drain; pipe resistance, line length and valve settings may make the distribution uneven.

A practical design balances flows so each waste-collection point operates effectively. Bottom drains need enough draw to transport dense waste. Skimmers need enough surface flow to capture floating debris. Returns need enough velocity and correct direction to create pond circulation without producing stressful jets or short-circuiting water back to the intakes.

For dealer facilities with multiple ponds, avoid assuming a shared filter distributes capacity evenly. The hydraulic path of least resistance may dominate. Flow meters or balancing valves on branches make actual allocation visible and controllable.

13. Oxygen and Gas Exchange Can Become the Hidden Limit

More fish and more feed do not only increase ammonia. They also increase oxygen demand. Koi consume oxygen directly, heterotrophic bacteria consume it while decomposing organic matter, and nitrifying organisms consume oxygen while oxidising ammonia. Complete oxidation of ammonia-nitrogen to nitrate has a substantial stoichiometric oxygen demand, and real pond systems have additional respiratory loads beyond that reaction alone.

This means a biofilter that is theoretically large enough can fail to achieve its rated performance if dissolved oxygen becomes marginal. Warm water makes the problem harder because koi generally eat and metabolise more while the water holds less oxygen.

Moving-bed chambers should have vigorous enough aeration to keep the carriers moving and the biofilm oxygenated. Shower and trickling filters need unrestricted air exchange. Submerged static media should not be allowed to clog with organics and create oxygen-poor zones. The pond itself also needs sufficient aeration so fish are not forced to compete with the filter for the same shrinking oxygen reserve during hot nights or feeding peaks.

For expensive collections, emergency oxygen or aeration should be sized as part of the filtration system rather than treated as an optional accessory. A power failure does not care that the biofilter was oversized on paper.

14. Design for Peak Load, Then Add Safety Margin

A filtration system designed to operate at 100 percent of its theoretical capacity on an ordinary summer day has no resilience. Real ponds experience temporary load spikes, partial clogging, reduced pump flow, ageing equipment, warmer water, increased feeding, new fish, medications, biofilm disturbances and maintenance events.

A safety margin is therefore essential, but there is no single percentage that suits every pond. The margin should increase as uncertainty and consequence increase. A heavily stocked dealer pond, commercial holding system or collection of very valuable koi deserves more redundancy than a lightly stocked private pond with easy access to water changes and spare equipment.

Safety margin can be created in several ways: additional biological reactor volume, unused media fill capacity, parallel filter circuits, a pump with controlled reserve capacity, oversized mechanical screen area, spare aeration, a bypassable UV, and physical space for adding media or another chamber later.

Redundancy is not the same as simply installing one very large device. Two independent air pumps, two biological trains or multiple circulation pumps may provide more operational resilience than one oversized component whose failure stops the entire pond.

15. Future Growth Must Be Included from Day One

Koi are unusual compared with many ornamental pond fish because a successful collection becomes more demanding over time. A 30 cm fish may eventually become 70 cm or larger, and its body mass increases far faster than its length. Meanwhile, the keeper may become more confident and feed more aggressively.

If the pond is being built for long-term koi keeping, size the biological and hydraulic infrastructure around the expected mature collection, not the fish currently in the quarantine tank. It is far cheaper to install adequate gravity lines, chamber sizes and electrical capacity during construction than to discover later that the plumbing itself is the bottleneck.

Where the future is uncertain, build expandability into the layout: spare pipe penetrations, space for a second moving bed, a manifold with capped branches, an electrical circuit with reserve capacity, or an RDF sized above the initial flow. Expansion is easier when planned than when retrofitted around finished landscaping.

16. Worked Example: A 20,000-Litre Private Koi Pond

Consider a 20,000-litre pond intended for a mature collection with a peak daily feed of 1.0 kg. The example below is illustrative; it shows the design logic rather than prescribing exact equipment for every 20,000-litre pond.

Suppose a chosen biofilter technology has credible test data showing a dependable operating nitrification rate of R grams TAN-N per unit of media or reactor per day under conditions comparable to the pond. The theoretical minimum required amount is then 30 ÷ R. The installed amount should be higher after applying a suitable safety factor and considering whether the source data were conservative, whether the pond will run warmer or dirtier, and whether fish load is likely to increase.

The purpose of the example is not to establish 3 percent as an exact koi number or one-hour turnover as a universal rule. It is to demonstrate a transparent chain: feed → estimated nitrogen load → validated biofilter capacity; pond volume → intended turnover → true flow after head loss; feed and stock → solids load → mechanical capacity.

17. Worked Example: A Heavily Stocked Dealer or Grow-Out Pond

Now consider the same 20,000 litres used as a heavily stocked dealer or grow-out pond, with a realistic peak feed of 2.5 kg/day. Using the same illustrative 3 percent feed-to-TAN-N design assumption gives roughly 75 g TAN-N/day—two and a half times the biological load of the first pond despite identical water volume.

Mechanical loading also rises sharply because more feed means more faecal solids and more fine organic material. The RDF or sieve may need greater screen area, more frequent but still sustainable waste export, or a parallel mechanical path. The biological stage may need a much larger moving bed or additional shower. Aeration becomes more critical, especially if the pond is warm and crowded.

Dealer systems also experience abrupt stocking changes. A filter that slowly adapted to 1 kg/day of feed may not immediately handle 2.5 kg/day when a shipment arrives. Even if the final installed capacity is sufficient, the active nitrifying population still needs time to increase. Operational strategy therefore matters: staged feeding, close ammonia/nitrite monitoring, seeded mature media, redundant filters and the ability to perform large water changes can be as important as equipment size.

This example shows why 'litres per filter' marketing is particularly risky for commercial holding ponds. The loading history and peak feed matter more than the water volume printed on the sales brochure.

18. How to Interpret Manufacturer Filter Ratings

Filter manufacturers need simple product labels, so ratings are often expressed as maximum pond volume. Such ratings can be useful for comparing models, but they are not engineering guarantees unless the assumptions are stated.

Before relying on a rating, ask:

• Is the stated pond volume for an ornamental pond, a koi pond, or a heavily stocked koi pond?
• What daily feed load is assumed?
• What fish biomass or stocking density is assumed?
• What flow range is required through the unit?
• Does the rated flow apply to gravity-fed or pump-fed operation?
• What water-level differential, head loss or media fill is assumed?
• For biological media, is the capacity based on protected surface area, total surface area, empirical feed loading or measured TAN removal?
• At what temperature, pH and dissolved-oxygen level was performance evaluated?
• Is the number a recommended continuous operating capacity or an absolute maximum?

If those answers are unavailable, derate the marketing number rather than designing a valuable koi system exactly at the claimed maximum. Engineering uncertainty should create more margin, not less.

19. UV Size Is a Separate Calculation

An ultraviolet clarifier can control free-floating green-water algae and, at sufficiently designed doses and contact conditions, can reduce certain microorganisms passing through the chamber. It does not replace mechanical or biological capacity and should not be included as if it processes ammonia or removes faeces.

UV sizing depends on lamp output, chamber design, water clarity, flow rate and the target UV dose. If flow is pushed above the manufacturer's specified operating range, exposure can become insufficient even though the plumbing physically passes the water. The UV should therefore be sized to the actual branch flow and intended purpose, not merely matched to total pond litres.

In large systems, a bypass or dedicated branch can make it easier to maintain the desired UV flow while the main circulation rate remains high.

20. Water Changes Do Not Excuse an Undersized Filter

Regular water replacement is valuable in koi husbandry. It helps control nitrate and dissolved accumulations, replenishes minerals where source water is suitable, and provides a way to correct problems. But routine water exchange should not be used to disguise filtration that cannot keep ammonia and nitrite controlled at normal feeding.

If safe water quality depends on emergency-scale water changes every day, the system is either overloaded, immature, malfunctioning or undersized. Water exchange is part of good management, not a substitute for matching waste production to treatment capacity.

21. Common Filter-Sizing Mistakes

22. How to Know Whether the Existing Filter Is Big Enough

An operating pond provides data that a new design does not. If ammonia and nitrite remain non-detectable or acceptably controlled under the pond's normal maximum feeding, dissolved oxygen remains strong, pH/KH are stable, solids are removed without excessive maintenance and there is measurable hydraulic reserve, the system is demonstrating adequate current capacity.

The strongest test is not how the pond behaves after a week of light feeding. It is how the system behaves during its normal warm-season maximum, after the fish have been fed consistently for several days and the mechanical filter is carrying the usual solids load.

Signs that capacity may be running out include:

• Ammonia or nitrite appears whenever feeding is increased modestly.
• The RDF rinses almost continuously or a static mechanical stage clogs far more often than intended.
• Downstream biological media accumulates visible solids despite normal maintenance.
• Dissolved oxygen falls substantially overnight or after heavy feeding.
• The pump must be throttled because downstream equipment cannot accept the intended flow.
• Water-level drawdown in gravity chambers becomes excessive at the target flow.
• One blocked screen, one failed air pump or one pump fault immediately places the pond at risk.
• The current system cannot accommodate the expected mature fish load without operating at its published maximum.

23. Upgrading an Existing Pond: Find the Bottleneck First

When a pond is struggling, do not automatically add another box of media. Diagnose which capacity is actually limiting performance.

Fix the bottleneck rather than the most visible piece of equipment. In many koi ponds, improving prefiltration or actual flow through the existing biofilter produces more benefit than adding extra media into a dirty, oxygen-limited chamber.

24. A Practical Filter-Sizing Worksheet

Use the following sequence when designing a new koi filtration system or auditing an existing one:

1. Measure or calculate the true operating water volume.
2. Define the maximum intended fish biomass and stocking pattern.
3. Estimate the realistic peak daily feed at mature stock and warm-season conditions.
4. Convert feed to an approximate TAN-N design load using an evidence-based assumption suitable for the feed/system, and state the assumption explicitly.
5. Choose a biological technology and obtain credible performance data for that reactor/media under comparable operating conditions.
6. Apply a safety factor appropriate to uncertainty, fish value, future growth and consequences of failure.
7. Select the intended turnover/circulation strategy and calculate the required true system flow.
8. Use pump curves and expected head loss to select a pump that will deliver that flow in the installed system.
9. Check every gravity line, chamber, screen, UV, heat exchanger, valve and return against the same true flow.
10. Size the mechanical stage for both the hydraulic flow and the expected solids mass, with a realistic cleaning frequency.
11. Provide sufficient aeration and gas exchange for koi, biofilm and warm-weather peak load.
12. Build in redundancy and emergency aeration for high-value or highly stocked systems.
13. Commission the pond with measured flow, ammonia, nitrite, pH/KH and dissolved oxygen rather than assuming the design numbers were achieved.
14. Record feed and water-quality data so future capacity decisions are based on the pond’s real operating history.

25. Sizing Philosophy for High-Value Koi Systems

For high-value koi, the cost of extra filtration capacity is usually small compared with the value of the fish and the consequences of chronic water-quality stress. That does not mean installing equipment without purpose. It means placing reserve capacity where it improves reliability: adequate gravity plumbing, mechanical screen area, biological headroom, oxygen delivery, independent pumps or air sources, and room to expand.

The most efficient high-performance systems are often not those with the largest collection of devices. They are systems in which each component is matched to the same design load and the waste path is short: waste leaves the fish, reaches a bottom drain or skimmer quickly, is physically removed early, dissolved ammonia reaches a well-oxygenated mature biofilter, and clean water is returned in a pattern that sweeps the pond again.

That design is easier to maintain, easier to troubleshoot and more predictable when stocking or feeding increases. Good sizing is therefore as much about system architecture as component size.

Conclusion

Sizing a koi pond filter is not a matter of matching one equipment label to pond litres. Pond volume determines hydraulic requirements, but fish biomass and especially peak feed determine how much waste the treatment system must process. Mechanical filtration must pass the true system flow while exporting the solids mass. Biological filtration must oxidise the estimated ammonia load under the pond's actual oxygen, pH, alkalinity, temperature and organic-loading conditions. Pumps must deliver the required flow after head loss, not merely at their advertised zero-head rating.

The strongest design process therefore works from load to equipment. Measure pond volume. Estimate mature biomass and peak feed. Translate feed into a conservative nitrogen load. Select a biofilter with validated performance and safety margin. Size mechanical separation for both flow and solids. Calculate true hydraulic flow through the entire treatment train. Then provide the oxygen, redundancy and future expansion capacity that keep the system stable when real life deviates from the ideal calculation.

For koi keepers, this way of sizing filters has one major advantage: every assumption can be tested. Feed can be weighed. Flow can be measured. Ammonia and nitrite can be monitored. Dissolved oxygen can be checked. Mechanical cleaning frequency can be observed. A filtration system is truly large enough when it consistently controls the pond's actual peak load with reserve—not simply because the box says it should.

The next article in this series, FL-005, will focus specifically on K1 moving-bed filters: how the media and reactor work, how aeration controls performance, how to choose fill level and chamber size, and where moving beds fit into a complete koi filtration train.