Choosing filter sand involves more than verifying that the grains fit a nominal size range. A material can meet that range when delivered and still behave differently after handling and repeated cleaning. Attrition concerns the loss or fragmentation generated when particles interact. Its practical assessment must distinguish initially present fines, fractured grains and material simply carried out of the vessel.
South Africa has a documented technical foundation for this question. A Water Research Commission report examined locally supplied granular filter media and the diversity of methods used to characterise them. It specifically distinguished hardness, crushing resistance and attrition resistance, and cautioned against translating unlike tests directly into long-term filter performance.1
The approach below is a proposed comparative qualification programme. It connects a repeatable laboratory challenge with observations from the intended filtration system. It does not turn a short accelerated test into a prediction of years of operation.
1. Define the failure being investigated
Begin with the observation that prompted the investigation. Sand in the discharge, a changing bed height, an altered grading curve and persistent cloudy rinse water are different observations. They may occur together, but should not be collapsed into “poor hardness” before the material and the installation have been examined.
For qualification, describe a measurable question: does candidate A generate more new undersize material than reference B under the same defined challenge? For an operating filter, ask a different question: where is the missing material, and has its size distribution changed?
These questions lead to different sampling points. The first needs comparable incoming samples and controlled treatment. The second needs bed samples, collected wash discharge and an inspection of outlets or retention components. Neither can be answered reliably by a photograph of wet sand alone.
Identify the particle size boundary used to call material “fine”. That boundary should have an application reason, such as the lower bound of the proposed grading specification. An arbitrary sieve selected only because it is available can give a reproducible number with little relevance to the filter.
2. Separate hardness, crushing and attrition
Hardness describes one kind of material response; it is not a certificate of complete mechanical durability. Crushing and abrasive wear impose different conditions. A grain that resists scratching can still fracture at a flaw, while a rounded grain can shed a surface coating without undergoing bulk fracture.
The South African WRC comparison is useful precisely because it does not treat these measurements as interchangeable.1 Keep the source’s historical scope clear: the report is a technical study, not a current universal specification for every South African installation.
In the proposed programme, label the applied challenge by what actually happens. A compression test between surfaces, a tumbling test and a water-column wash test should not all be reported as the same “abrasion index”. Include the apparatus and calculation beside the result.
A supplier comparison is only meaningful if both materials face equivalent conditions. Differences in starting particle distribution can also affect the outcome. Decide whether the comparison represents commercially supplied products or matched size fractions; these are both useful comparisons, but they answer different purchasing questions.
3. Establish an incoming reference before stressing the sample
Measure the initial distribution before deliberately generating wear. If a batch already contains undersize particles, finding undersize material after a test does not establish that the test created all of it. Preserve an untreated portion of the same sample for later comparison.
Sampling and splitting require their own procedure. ISO 14488 addresses particulate-material sampling and sample splitting; its published scope identifies that task without providing, on its public page, a complete filter-sand test recipe.4 Apply an agreed sampling method appropriate to the lot rather than use a convenient handful as the batch reference.
For the proposed trial, divide material into test portions and a retained portion using a method that avoids preferentially selecting coarse or fine particles. Record dry mass and initial grading. If the material is first cleaned or screened, declare that preparation as part of the tested condition.
Test both “as supplied” and “prepared for service” when the distinction matters commercially. The first evaluates delivery quality; the second helps isolate behaviour after the prescribed preparation. Do not quietly remove fines from one candidate while testing the other as received.
4. Keep the mechanical challenge reproducible
Choose a challenge the laboratory can describe and repeat. Specify the apparatus, sample mass, duration, liquid conditions where relevant, and the imposed motion or flow. A statement such as “strong washing” is insufficient: a later laboratory cannot reconstruct it or know whether differences are material-related.
For a pilot water-column comparison, keep the geometry, starting bed condition and operating sequence common to all candidates. Record the actual flow rather than assume a valve position provides the same condition. If air is used, identify when and how it enters the sequence.
These are proposed controls for a comparison, not a substitute for the filter manufacturer’s instructions. A pilot designed for investigation must not be copied into an operating pressure vessel. The equipment guide governs permitted operation and safe handling.2
If several intensities are tested, distinguish them explicitly. A deliberately severe challenge may help rank candidates, but its result needs that label. It should not be reported as equivalent to one ordinary operational wash without evidence connecting the conditions.
5. Collect particles rather than watch water colour
Colour and cloudiness can guide observation, but they do not quantify lost sand. Fine mineral particles, retained contaminants and residues from previous treatment can produce different discharge appearances. The mass comparison needs a defined collection and separation process.
In the proposed test, collect the discharge over the entire agreed sequence or document a representative collection method. Determine whether the retained particles are original media, deposited solids or another material. Keep a blank or background measurement where the apparatus can contribute particles.
Use compatible mass bases. Compare dry media mass with dry recovered mineral mass, not with the wet sludge mass from a bucket. If the recovery process includes filters or sieves, account for their tare and any particles not captured by the chosen method.
A low recovered mass can mean low generation, incomplete recovery or passage below the collection boundary. Describe the collection efficiency or limitation before treating the result as a material property. The objective is to explain a balance, not merely produce a small number.
6. Calculate newly generated fines explicitly
A simple fictional example illustrates why the incoming reference matters. Start with 1,000 g of dry material containing 5 g below the agreed boundary. After a defined challenge, a complete recovered set contains 13 g below that boundary. The apparent increase is 8 g, or 0.8% of the starting dry mass.
That calculation is valid only if the initial and final fractions were measured comparably and recovery was sufficiently complete. It is not automatically a standardised attrition value. Name it “increase in undersize fraction under the defined test” rather than give it an undefined commercial label.
Now suppose another candidate starts with 20 g of fines and ends with 24 g. Its final fines content is higher, but its measured increase is only 4 g. Depending on whether the decision concerns delivery cleanliness or generation during use, the ranking can differ.
Present both the initial fines content and the increase. Reporting only the final value obscures the mechanism. Reporting only the increase obscures whether incoming material already requires considerable preparation before service.
7. Track lost mass separately from fragmentation
Fragmentation can redistribute mass across size classes while leaving most material inside the apparatus. Conversely, whole grains can leave the filter without appreciable breakage. A mass-loss result therefore answers a different question from the proportion shifted to finer classes.
Build two comparisons. First, compare total dry mass entering and recovered across all streams. Second, compare size distributions of the remaining bed and the collected material. Examine the missing portion before attributing it to abrasion.
If recovered discharge contains grains similar in size to the incoming product, investigate retention and hydraulics alongside durability. If a newly enriched fine fraction appears, investigate fragmentation or surface shedding. These observations support a hypothesis; microscopy or another appropriate examination may be needed to distinguish mechanisms.
For purchasing, avoid accepting a single favourable percentage without its denominator. Is it a percentage of the incoming mass, retained mass or recovered discharge? Those quantities can differ significantly and must be named before results are compared.
8. Inspect shape changes with a consistent method
Photographs are useful when they are made comparable. Use the same magnification, lighting and preparation, and identify how the pictured grains were selected. A striking broken grain is not evidence that the entire batch fractured at the same frequency.
For the proposed audit, compare representative fields before and after the challenge. Note observable features such as newly sharp faces, detached coatings or fragments. Keep interpretation separate from observation: “more fragments visible in this field” is narrower than “the quarry produces unstable sand”.
If automated image analysis is used, state its size and shape definitions. Do not mix a count-based image statistic with a mass-based sieve statistic as though they expressed the same distribution. Small grains can dominate a particle count while contributing little total mass.
Return the observations to the purchasing question. Shape changes matter when they accompany a measurable alteration relevant to filtration or handling. A purely aesthetic preference for rounded or angular grains is not, by itself, proof of better attrition resistance.
9. Relate wear observations to hydraulic performance
Mechanical durability is one qualification dimension. A resistant material can still have an unsuitable initial grading or behave poorly in the chosen filter. Conversely, a modest measurable change may have little operational consequence within a specific accepted system.
Use the proposed pilot to compare clean-bed behaviour before and after the challenge under the same defined conditions. Measure the relevant flow and pressure difference rather than compare isolated gauge readings taken at different operating points.
If the particle distribution changes, ask whether the observed hydraulic difference follows that change or a remaining deposit. A washed bed and a contaminated bed are not equivalent reference conditions. Where possible, maintain separate observations of grading, retained deposits and hydraulics.
The WRC study also considered media properties and hydraulic behaviour.1 Here the practical decision remains installation-specific: a laboratory wear ranking should complement, rather than replace, evidence that the chosen media operates acceptably in the intended treatment train.
10. Use repeated portions to judge the ranking
One portion per candidate cannot reveal how variable the result is. Repeat the comparison with independent portions of the sample using the same procedure. If the differences between candidates are smaller than the variation between repetitions, avoid declaring a confident winner.
A fictional result of 0.8%, 0.9% and 0.7% for one candidate, compared with 0.3%, 0.4% and 0.3% for another, presents a different decision situation from two groups whose individual values overlap substantially. These numbers illustrate comparison logic, not acceptance limits.
Include preparation repeatability in the investigation. If incoming fines vary strongly between portions, check the sampling and splitting before blaming the challenge apparatus. If recovery varies, check collection and drying before concluding the grains vary.
Report the repeated values alongside the summary. A mean alone can conceal an abnormal run. Document the reason for any excluded result and repeat the affected condition when the exclusion changes the ranking.
11. Set an acceptance rule tied to the application
DIN EN 12904 covers silica sand and silica gravel used for treatment of water intended for human consumption.3 Its catalogue entry establishes the subject and edition; the full applicable requirements must come from the complete specification adopted for the project.
An original comparative test should not borrow a limit from an unrelated method. Even an identically named “loss” result may reflect different apparatus, preparation or calculations. State the adopted method, acceptance rule and any additional installation trial together.
For a reference-based decision, define the tolerated difference before seeing results. Consider measurement variability and the operational significance of the change. A statistically distinguishable difference may be too small to justify changing supply; a large practical difference deserves investigation even before a precise lifetime model exists.
Keep the acceptance decision conditional on the qualified material and application. Changing the size fraction, preparation or operating sequence may require another comparison. Qualification belongs to a defined product condition, not simply to the word “silica”.
12. Make the specification useful for subsequent deliveries
Translate the successful comparison into a concise purchasing specification: identity of the material, agreed grading, incoming fines criterion, the applicable mechanical test and its reporting basis, and the installation requirements retained by the operator.
Keep laboratory and operational responsibilities distinct. The laboratory reports the defined measurements; the operator determines whether they satisfy the process requirement. A supplier certificate should not be asked to predict performance under unspecified backwash conditions.
For later deliveries, compare the incoming grading and agreed indicators with the retained qualification reference. Investigate meaningful shifts before assuming every lot from the same origin behaves identically. A consistent specification makes this comparison easier without requiring a complete pilot programme for every delivery.
The silica sand page provides a starting point for describing a material requirement. For South African filtration projects, a clear attrition assessment links initial quality, the exact applied challenge, recovered particles and actual filter behaviour. That chain is more useful than a hardness claim or a loss percentage without a method.