Quick answer: an average only works when the batch has one population

If a batch of polyphosphate balls contains two distinct size populations — say a coarse fraction near 20 mm and a fines fraction near 10 mm — the arithmetic mean lands in the gap between them. That single number can describe almost none of the balls in the drum. You then design a screen, a bed, or a dissolution assumption around a diameter that barely exists, and the equipment behaves in a way the datasheet never predicted.

This is why polyphosphate ball diameter data should always be read as a distribution, not a scalar. The practical fix is straightforward: request the size distribution, state the weighting basis, and set limits on the fractions you cannot tolerate. The rest of this article explains how bimodal distributions arise, what they change inside real equipment, and what to put in a specification so the problem never reaches your production line.

How two size populations appear in the same batch

Bimodality is rarely intentional. It usually enters through one of these routes:

• Screening cut-offs that are too wide. A supplier who screens “10–22 mm” and reports the midpoint as the average is technically passing a range, but the resulting batch can show a coarse peak and a fine peak with a trough in between.
• Blending of production runs. Leftover material from an earlier campaign is sometimes blended into a new lot. If the two runs had different target sizes, the blend is bimodal by construction.
• Attrition during handling. Polyphosphate crystal is a glass-like material. Vibration in transit, repeated transfer, and rough valve operation generate chips and fragments that collect in the fine range.
• Forming variability. Where media is cast or formed rather than precisely machined, mold wear and cooling differences can produce two recurring sizes instead of one tight band.
• In-service dissolution. Once installed, polyphosphate balls dissolve and shrink. A bed that started monomodal becomes progressively bimodal, because balls dissolve at different rates depending on their position and the local flow they see.

The last point matters for troubleshooting: if you only measure the average diameter of a used charge, you cannot tell whether the media arrived bimodal or became bimodal in service. Sample the incoming lot and keep the record.

Number-weighted and mass-weighted averages are different numbers

For spheres, mass scales with the cube of diameter. A ball of 20 mm carries roughly eight times the mass of a ball of 10 mm. That single relationship explains most of the confusion in size data:

• A count-based average is dominated by the many small balls.
• A mass-based average is dominated by the few large balls.
• Surface area, and therefore the rate at which polyphosphate enters the water, tracks the small fraction far more strongly than the large one.

So two suppliers can both state “average 16 mm” for genuinely different materials, and both statements can be arithmetically correct. Never accept an average diameter figure that does not say whether it is number-weighted or mass-weighted, and how it was measured.

What bimodality actually changes in your equipment

Pressure drop and flow distribution

Small balls fit into the voids between large ones. That reduces the void fraction of the bed, raises the resistance to flow, and can create preferential paths. The measured pressure drop of a bimodal charge is usually higher than a monomodal charge of the same average diameter — sometimes substantially. If your housing was sized on the average diameter alone, the pump curve may no longer match the system.

Release rate and effective service life

Dissolution of polyphosphate media is largely a surface-driven process. A fines fraction delivers a burst of phosphate early, while the coarse fraction continues to release slowly. The result is an uneven treatment profile: over-dosing at the start of the charge and under-dosing at the end. Where the objective is scale inhibition by sequestration of hardness ions, that variation directly affects how reliably the threshold dose is maintained.

Screen and retainer compatibility

Any ball smaller than the downstream screen opening will leave the housing. Before accepting a distribution, compare the fine tail against the actual screen aperture in your equipment, not against the nominal media size. A distribution where 4–5% of the mass sits below the screen opening is a carry-over problem waiting to happen.

Dose and refill calculations

Refill quantities are usually calculated from bed volume and bulk density. A bimodal bed packs differently from a monomodal one, so a fixed “kilograms per litre” figure taken from an unrelated product may be wrong for your charge. Bulk density should be measured on the actual lot, in the same way it will be packed.

Metrics that describe a distribution properly

How to measure the diameter distribution

1. Define the sample. Draw it from the delivered packaging, not from a single bag at the top of a pallet. A riffle splitter or quartering gives a more representative subsample than scooping. Sample size must be large enough that the smallest fraction is represented by many individual balls — a handful of balls is never sufficient.
2. Choose the method and state it. For larger balls, caliper measurement of individual pieces across at least two axes, or image-analysis systems, are common. For smaller media, a stack of sieves with a defined aperture series is standard practice.
3. Record the mass retained on each sieve or in each size bin. Mass-based data converts directly into the packing and dose figures you actually need. Where surface area matters, convert mass fractions back to an approximate count using the cube relationship.
4. Plot a histogram and a cumulative curve. Look for a trough between two peaks. A cumulative curve with a flat step, or an S-curve that flattens twice, is a strong indication of two populations.
5. Report method, sample size, and weighting. A distribution without this context cannot be compared between suppliers or between lots.

Sieves should conform to a recognised sieve standard agreed between buyer and supplier, and the same aperture series should be used for every lot you compare. Keeping the method constant matters more than which recognised series you pick.

An illustrative example

Consider a hypothetical lot described only as “average diameter 16 mm.” A check on 500 individual balls shows that roughly 60% of them fall between 19 mm and 22 mm, while about 35% fall between 9 mm and 12 mm, with very little in between. The 16 mm average sits in the empty middle of the distribution. The fines fraction accounts for a small share of the mass but a large share of the surface area and an even larger share of the risk if the housing screen opening sits above 12 mm. Nothing in the original statement was false — and nothing in it was useful for design.

The lesson is not that the supplier was careless. It is that a single summary statistic cannot carry the information a packed bed needs.

Specification checklist before you order

• Nominal diameter range, not a single average figure.
• Minimum and maximum allowable diameter, expressed as mass percentages.
• Maximum fines content below the lower limit, and the test method used to determine it.
• Statement of weighting basis and the number of individual balls or mass of sample tested.
• Sampling plan and the point in the supply chain where the sample is taken.
• Bulk density, measured on the packed bed as it will be used.
• Attrition or handling test, if the media travels long distances before installation.
• Sieve aperture series or measurement instrument, stated explicitly.

What polyphosphate media does — and does not do

Glassy polyphosphate media is used for scale inhibition, generally by sequestering hardness ions and interfering with carbonate crystal growth at low doses. It is not a water softener: it does not remove calcium and magnesium from the water, and it cannot be regenerated like an ion-exchange resin. It is also not a filter medium in the mechanical sense, and it is not a disinfectant.

Some polyphosphate materials are formulated with silicate and can contribute to corrosion inhibition by supporting a protective film on metal surfaces, but that behaviour is formulation-specific. Brand names such as Siliphos refer to particular commercial products, and one supplier’s siliphos crystal is not automatically interchangeable with another’s polyphosphate crystal in composition, dissolution behaviour, or size grading. Where drinking water is the intended application, do not rely on a “food grade” description alone — suitability depends on the exact product, the dose, the contact conditions, and the requirements of the destination market, and must be confirmed for that specific use.

Frequently asked questions

Is a bimodal distribution always a defect?

No. Some applications deliberately blend sizes to adjust packing or release profile. It becomes a defect when the distribution is undocumented, when the fine tail escapes the housing, or when the bed was designed on an average that the blend does not actually deliver.

Can I detect bimodality from a bulk density figure?

Not reliably. A bimodal and a monomodal lot can share similar bulk density while behaving very differently in pressure drop and release rate. You need the size data.

Should I measure unused media or media taken from service?

Both, and keep them separate. Incoming data tells you what you bought; in-service data tells you how it is performing. Mixing the two makes it impossible to assign responsibility for a problem.

What if the supplier only provides an average?

Ask for the underlying per-sieve or per-bin data behind that average. A supplier who has measured the material already has it. If the data does not exist, request that a distribution be established before the specification is fixed.

Send the size question with your inquiry

If you are comparing polyphosphate ball or siliphos crystal options and the diameter data you have received is a single average, send us the details that determine what the distribution actually needs to satisfy: your application and treatment objective, the source-water conditions you are working with, the equipment type and screen aperture, the diameter range and fines limit you require, the quantity, and the destination market. With those inputs, the size conversation can be settled on the numbers rather than on a midpoint.