Select treatment according to the impurity to remove
Washing, attrition and magnetic separation are not three names for the same operation. They address different situations and can be combined. Before investing, establish where impurities occur, how they are associated with grains and what product quality must be obtained. Without this characterisation, a machine can work correctly while answering the wrong question.
The Ministry of Industry presents the Oriental region with industrial and logistics platforms around Nador, including Selouane and Nador West Med 1. This context makes structured raw-material preparation relevant. It does not demonstrate a qualified sand deposit or a particular local composition. This article therefore examines a testing programme for a project in Nador, without announcing a resource or commercial yield.
The British Geological Survey factsheet explains that different silica-sand applications require combinations of properties that are not interchangeable 2. Start with a product specification: required fraction, relevant composition, consistency and packaging. An increase in SiO₂ alone does not demonstrate that every requirement has been satisfied.
1. Define the product before the machine
Record the characteristics the eventual user will examine. Distinguish overall composition, particular contaminants and grading. For glass, an average composition can be insufficient if undesirable grains remain. For another application, an excessively fine fraction may be the main difficulty. The testing programme should follow these differences.
Also establish the results basis. A composition expressed on dry material is not directly comparable with a commercial mass containing water. A yield calculated on all feed differs from one calculated only on a preselected fraction. Without these conventions, two tests can appear contradictory while describing different boundaries.
Date the target sheet and associate it with the application studied. If it changes during the campaign, retain both versions and identify which decision concerns each. A machine selected for an earlier specification should not be presented as automatically suitable for a requirement that has become stricter.
2. Characterise representative feed
Define sampling lots or areas and their variability. Do not base the entire decision on the best available sample. Where materials differ visually or by origin, keep them separate initially. Premixing can conceal a difficult fraction and produce an average that represents no actual usable lot.
Associate each portion with grading and relevant analyses. Mineralogical examination can help distinguish accessory grains from surface coatings. Total iron content alone does not identify which operation will remove that iron. The decision should establish a connection between chemical measurement and the form of occurrence being studied.
Retain reference portions and available masses. The programme may require a repeat test, microscopic observation or independent checking. A properly identified reserve avoids repeating the campaign with different material. The retained sample should remain connected to its original lot, preparation conditions and date.
3. Understand what washing is intended to separate
In the proposed programme, the first test examines particles that can be removed through wet preparation and defined classification. The objective is to measure what enters rejected fractions and what remains in the product. Describe washing through the actual operations performed, rather than the phrase “washed sand” alone.
Analyse the product and outputs where they inform the decision. If chemical improvement accompanies substantial loss of useful grains, the result should show both effects. A photograph of lighter-coloured sand does not replace measured composition. Colour is an interesting observation, but it is not an impurity-content measurement.
Also record water quality, rinsing stages and fines separation. A test using one water supply does not necessarily describe operation with recycled water carrying particles. Managing that water belongs to the industrial programme because treatment creates several streams whose quality must be understood.
4. Distinguish washing from attrition
Attrition adds grain-to-grain rubbing under controlled conditions. It is investigated where adhering material needs to be detached before separation. The CSIR study notice on Cherthala reports a combination of screening, attrition and magnetic separation for identified Indian samples 3. It does not supply a guaranteed yield for another material.
To compare attrition’s effect, plan a reference preparation without that stage and another with it, followed by compatible classification. If the grading cut changes simultaneously, improvement can no longer be attributed to attrition alone. The trial should make the changed operation visible.
Monitor distribution before and after treatment, along with output masses. Surface improvement can accompany grain modification. The programme must examine the useful product, rather than merely a machine’s ability to generate mechanical action. More intensive treatment is not automatically more favourable to the product.
5. Test magnetic separation on appropriate material
Investigate magnetic separation according to existing constituents and their behaviour under the equipment conditions. It is not a universal method for removing every impurity. Identify the feed for this stage: raw, washed, classified or attrition-treated.
The National Metallurgical Laboratory notice describes a test on Ratnagiri sand combining wet preparation, attrition and magnetic separation 4. The result belongs to that sample and flowsheet. Using it as evidence that a combination is worth investigating differs from inferring a composition guarantee for material in Nador.
In the campaign, retain settings and masses of magnetic and non-magnetic fractions. Analyse outputs needed to understand what is removed. A higher product grade can result from effective separation, but should be read alongside useful-material recovery. Rejection percentage alone does not measure performance.
6. Connect the three operations instead of ranking them abstractly
A trial may show that one operation prepares the next. Material detached by attrition still needs separation; classification can remove fines before a later stage. Compare complete flowsheets corresponding to the final product. The best isolated operation does not necessarily form the best circuit.
Build a simple table: feed, stage, product, reject and available analyses. Retain the same conventions for every flowsheet. Where a preparation requires drying, add it to the boundary. Comparing it with another preparation without including that requirement would give an incomplete industrial interpretation.
Do not multiply equipment before identifying each machine’s contribution. A complex flowsheet can be justified, but should bring a measurable result in quality, yield or consistency. The decision should not rest solely on the number of stages or the appearance of a treatment line.
7. Calculate mass yield and recovery separately
Product mass yield is its dry mass divided by the defined dry feed mass. Constituent recovery also accounts for composition. These indicators differ. A product can have a higher grade while containing a smaller recovered quantity if mass loss is substantial.
For a monitored constituent, recovery uses product mass and content, divided by feed mass and content. Use compatible units. Specify whether contents are fractions or percentages. Conversion factors must be identical in the numerator and denominator.
Then check balance closure from collected outputs. An unexplained difference can arise from sampling losses, stock remaining in equipment or incompatible moisture bases. Identify these before publishing a yield. Precision to several decimal places does not correct a balance with poorly defined streams.
8. Examine quality by size class
An overall analysis can conceal differences between classes. If the final product retains only particular fractions, analyse these where appropriate. Treatment can concentrate an impurity in part of the distribution or remove a commercially important fraction. The decision should connect quality with usable grading.
Keep the class-separation method consistent. Wet and dry sieving should not be compared as if they automatically represent identical preparation. Specify conditions and the treatment of agglomerates. The aim is to describe material rather than silently changing its definition between tables.
For a customer enquiry, present the curve and analyses of the proposed product. Our silica-sand page complements this interpretation with general properties. The treatment dossier should retain the specific feed and product data on which investment rests.
9. Prepare the laboratory-to-pilot transition
A laboratory test checks material response under defined conditions. It does not directly demonstrate industrial throughput, water consumption or line consistency. Before purchase, define questions the pilot must resolve: variable feed, product stability, stream management and sustained operation.
Select several materials representing observed variation. Test a realistic operating range rather than one favourable point. Retain startup data separately from regular operation. Transitional stock or adjustments in progress can produce a result difficult to compare with stabilised operation.
The pilot should also identify monitoring and maintenance operations. Describe how drift will be detected and corrected. Satisfactory quality obtained only through exceptional supervision does not necessarily describe regular operation. These conditions should be known before sizing the project.
10. Include water, rejects and handling within the boundary
Treatment creates outputs to manage. Describe water streams, fines, oversize and intermediate products without automatically assigning commercial value. A possible use mentioned in a publication does not mean that your stream has a buyer or satisfies another application’s requirements.
Assess storage, withdrawal and drying needs with the selected flowsheet. A laboratory-qualified product can be changed by poorly organised withdrawal or mixed with another lot. Design should preserve identity and properties up to acceptance. The material circuit forms part of final quality.
For economic data, use quantities actually measured or clearly identified project assumptions. Separate investment, operation, consumables, water, energy and maintenance. No local cost is inferred here from foreign studies. The dossier should allow an assumption to be replaced with a quotation without rebuilding the whole assessment.
11. Compare two flowsheets on the same basis
Suppose flowsheet A provides more product but less favourable composition, while B provides less product with better composition. This is a reasoning example, rather than a study conducted by Minerallum. The decision first depends on specification: more abundant but unacceptable product answers a different question from usable product.
Then compare consistency and sensitivity to feed variation. A flowsheet successful on one favourable portion remains less documented than one evaluated on the lots actually expected. Examine outputs and costs over the same duration, dry basis and preparation boundary.
If neither flowsheet satisfies every requirement, identify the blocking point precisely. It may justify another preparation, another use or reassessment of the resource. Adding a machine without a treatment hypothesis is not a technical answer. The next stage should be designed to resolve an identified question.
12. Record the investment decision and its conditions
The final dossier combines feed, objective, flowsheets, methods, balances, variability and pilot trials. Distinguish measured elements from those still being sized or estimated. The decision identifies the product the line must supply and the conditions in which this capability was examined.
Plan modifications requiring another trial: material origin, fraction proportions, composition or customer specification. A line selected for particular feed should not be considered universal. Qualification becomes more valuable when teams can recognise material outside the studied boundary.
In Nador as elsewhere, the best choice therefore does not declare washing, attrition or magnetic separation a universal winner. It connects a characterised impurity, a verified operation and a useful product with a readable balance. This chain of evidence prepares a more precise investment decision than a purity promise disconnected from the material.
13. Worked example: higher purity can conceal lower recovery
Consider an invented teaching example only, unrelated to Minerallum material or a Nador resource. Dry feed is 100 kg containing 98% SiO₂. Treatment produces 85 kg of dry product containing 99% SiO₂. Product mass yield is 85%. The feed contains 98 kg of SiO₂; the product contains 84.15 kg.
SiO₂ recovery in this product is therefore 84.15 divided by 98, approximately 85.87%. The product is richer than the feed, but some silica occurs in other outputs or losses. This distinction is essential: increasing purity does not mean complete recovery. Examine other streams to understand where material absent from the product has gone.
Now suppose treatment retains an interesting composition but removes a size fraction requested by the user. Overall yield no longer describes usable-product yield sufficiently. Calculate the mass within the fraction meeting specification. One circuit can therefore show a high sand yield and a substantially different yield for a targeted application.
In the trial table, present three distinct results: product composition, mass yield on the defined feed and recovery of the constituent studied. Where necessary, add yield of the fraction actually saleable. Each uses an explicit numerator and denominator. This presentation makes flowsheet comparisons clearer for technical teams and investors.
Finally, a balance calculation does not itself measure sand acceptability. Specification can include other constituents, consistency or accessory grains. The example only checks conventions and mass coherence. Industrial decisions should retain all useful results with their methods, rather than selecting whichever percentage looks most favourable.