A grading curve is a formulation tool, not a universal recipe
Selecting silica sand for polymer concrete means preparing a mineral skeleton that will be bound by a resin system. The requirement is not simply to choose fine grains or reproduce a curve used in cement concrete. The practical question is how to construct a combination of fractions compatible with mixing, moulding and the expected product properties.
Casablanca-Settat is described as Morocco’s leading industrial hub by the Ministry of Industry 1. This context supports a qualification guide for regional formulators without assuming a particular factory, local polymer-concrete consumption or already measured demand.
The following method organises the necessary data, blend calculations and trials. It does not supply a ready-to-manufacture formulation. It complements the guides on epoxy repair mortars and particle size with a different priority: constructing and then verifying the curve of a skeleton intended for polymer concrete.
1. Define the system before selecting fractions
Polymer concrete uses a polymer binder to bind aggregates. It should be distinguished from cement concrete modified with a polymer. Confusing the two would transfer formulation rules between systems whose constituents and hardening differ.
The initial record should identify the binder family and reference, its components, aggregates, fillers and any other constituents. Distinguish the ratio between binder components from total binder content in the composite. The first concerns the prescribed chemical formulation; the second is a blend variable where its investigation is permitted.
An experimental study by Kerrida and colleagues compared gravel/sand ratios and epoxy contents in a micro-polymer-concrete system 2. It confirms the value of investigating these parameters together without turning the combinations retained in that research into a general optimum for every Moroccan sand.
2. Start from product and process constraints
Describe the component or product before calculating a curve: geometry, thicknesses, filling areas, finish and loading. Add the mixing and forming method. A grain acceptable in a large volume can become incompatible with a narrow area or a particular finish.
Then define the responses to follow. Compressive strength may matter but does not represent every function. Depending on the product, flexure, surface condition, porosity, dimensions or a precisely specified service performance may need consideration.
This document should separate mandatory requirements from optimisation objectives. A cheaper combination that fails a compulsory criterion is not a better formulation. This hierarchy prevents premature selection of the curve with the highest dry packing measurement.
3. Obtain actual curves for available fractions
For each fraction, request a measured distribution, lot identification and method. A “0.2/0.6” designation indicates a commercial class; it does not automatically reveal the percentage passing every sieve used in the calculation.
Establish a common table of diameters and passing percentages. Where curves do not use the same points, arrange a comparable measurement or identify interpolations. An interpolated result must not be presented as a direct measurement.
The particle-size distribution guide explains common indicators. Keeping the complete curve is particularly useful for a blend: distributions with similar medians can contain different proportions of fines or coarse grains.
4. Calculate a blend curve on an explicit basis
Where fractions are combined in dry mass proportions, the blend’s passing percentage at a particular diameter is the sum of each fraction’s passing percentage multiplied by its mass proportion. Proportions must total 100%. This relationship is a mass balance rather than a strength prediction.
Teaching example with fictional data: at the 0.5 mm sieve, fraction A has 80% passing and fraction B 20%. A blend containing 40% A and 60% B gives 0.40 × 80 + 0.60 × 20 = 44% passing. Repeating the calculation for every diameter constructs the predicted curve.
This prediction should be compared with a measurement of the blend actually prepared. It assumes consistent data and preparation conforming to the proportions. A weighing, lot or sampling error can explain a difference before discussing the model.
5. Distinguish mass proportion from volume proportion
Where constituents have different densities, equal mass proportions do not represent equal volumes. Converting mass to material volume requires a particle density compatible with the calculation; bulk density also includes spaces between grains.
For a given mass, material volume equals mass divided by that density. Calculated volumes can express constituent proportions. They do not automatically represent the compacted blend’s external volume, which may include pores.
Research by Smoleń and colleagues specifically examined resin volume fraction and different quartz aggregate distributions 3. It highlights the importance of the formulation basis. A mass-percentage content must not be compared directly with a volume-percentage content.
6. Measure packing without confusing it with performance
A dry filling measurement can help compare combinations. It must follow a reproducible protocol: container, mass or volume, filling method, any vibration and calculation. Changing these conditions between candidates distorts comparison.
Measured packing alone does not describe resin wetting, retained air or actual forming. A combination that packs densely when dry can be difficult to mix under available conditions. The result must therefore be associated with trials using the selected binder.
To organise selection, classify the dry measurement as a screening indicator, then verify composite responses. This status avoids excessive extrapolation and allows a combination promising only in a laboratory container to be rejected.
7. Link fines to a formulation question
Adding a fine fraction changes the skeleton’s distribution. This does not mean “finer” improves every property. Quantity, grain surface area and binder system must be investigated together.
In the trial record, distinguish a fine granular fraction from a filler prepared for a particular function. The word “silica” establishes neither the same size nor the same function. Substitution between sand, quartz powder and another filler should be documented as a constituent change.
Where the comparison concerns fines, keep other fractions and the preparation procedure as constant as possible. Measure useful effects: mixing suitability, filling, surface, pores and relevant mechanical properties. The conclusion should come from these together rather than a single particle-size indicator.
8. Build a trial design answering the right question
NIST distinguishes comparison, factor screening and optimisation when selecting an experimental design 6. For a formulator, this means a trial comparing two curves does not necessarily answer the search for an optimum.
Start with a limited question: does a new combination improve a defined criterion without degrading other requirements? Where several variables are needed, organise the design to separate their effects rather than changing curve, resin, mixing time and vibration simultaneously.
Blend proportions are linked: increasing one fraction requires another to decrease where the total stays constant. The trial table should make this dependence visible. A statistical discussion treating every proportion as independent can lead to an inconsistent design.
9. Keep a reference and useful repetitions
Include a reference combination prepared under the same conditions as the candidates. It can reveal an unusual day or a change in equipment operation. Without a reference, apparent improvement may result from a different preparation condition.
Distinguish several specimens from one batch from several independent batches. The first describe that preparation; the second help examine manufacturing reproducibility. Identifiers should preserve this relationship.
Establish acceptance criteria and treatment of missing or invalid results in advance. An unfavourable result should not disappear from the table simply because another specimen is better. Document exclusion where the measurement genuinely failed for an identifiable cause.
10. Control preparation with the binder
The procedure should follow instructions applicable to the resin system: component proportions, preparation conditions, usable time and curing. Values from a research paper do not replace the instructions for the binder actually selected.
Record material temperature and ambient conditions where they form part of process control. Preserve introduction sequence, equipment, duration and prepared quantity. This information helps interpret difficult mixing or irregular filling.
Aggregate moisture condition should be known and compatible with the selected system. Establish the control method against binder and application requirements. Avoid transferring a universal drying temperature from an experimental protocol to every material.
11. Assess filling and retained air
A moulded specimen should represent the method being investigated. Document filling, permitted consolidation and surface observations. Where the comparison includes vibration, define its procedure rather than relying on an assessment of “sufficiently vibrated”.
Suitable sections or examinations can help inspect pores and grain distribution, but interpretation must remain linked to observations. One local area does not automatically represent the component’s entire volume.
A formulation giving a good result in a small mould should be tested against production geometry. Transfer is a qualification stage in its own right. Keeping the same curve does not guarantee that air, filling and curing remain identical when scale changes.
12. Select suitable mechanical methods
ASTM C579 and C580’s public entries concern compression and flexure respectively, the latter including elastic modulus, for families including polymer concretes 4, 5. Their published scopes guide method selection; execution requires the complete applicable protocol.
Specify geometry, age, conditioning and units before comparing values. Two strengths expressed in MPa are not directly comparable where methods and specimens differ. This requirement also applies when comparing with a publication.
The performance retained should correspond to the product. Compression alone cannot characterise a component whose function depends on flexure, surface durability or stable dimensions. The validation programme brings together the requirements actually needed.
13. Interpret research without importing a recipe
Studies provide methods, variables and observations under identified conditions. Their value is to show how a comparison was organised and which interaction was observed. Their optimum concerns the system studied.
To use a publication, establish a brief record of binder, aggregates, fractions, content basis, preparation, specimens and responses. Then compare these with the project. Important differences indicate what needs local verification.
The Minerallum guide uses literature to support the approach without announcing a formulation validated by the company. This distinction gives readers a useful technical tool while preserving a clear relationship between published evidence and their own formulation work.
14. Assess the cost of the qualified combination
Compare cost on the same basis: composite mass, product volume or manufactured unit. Include fractions, binder, preparation and observed losses. Comparing sand cost per tonne alone can conceal the cost of a formulation change.
Reducing resin content is a gain only where requirements remain satisfied. Likewise, a cheaper fraction may require additional preparation. The balance should distinguish measured data from price or yield assumptions.
When presenting an economic result, keep dates and calculation bases. This guide does not provide a Casablanca price: it proposes a structure for comparing real offers and results when available.
15. Perform a numerical volume-balance check
A fictional example can clarify the difference between grain density and filling. Suppose a fraction has a particle density of 2.65 kg/L and a bulk density, measured under a defined protocol, of 1.60 kg/L. In one litre of bed, mineral material occupies approximately 1.60 / 2.65 = 0.604 L. The remainder, approximately 0.396 L, represents bed spaces under those measurement conditions.
This is not a prescription for 39.6% resin. Introducing binder, mixing and forming can change the arrangement, while air and additional constituents need consideration. It is a consistency check for understanding what the dry test measures rather than a sufficient formulation method.
In the spreadsheet, identify every datum’s origin: laboratory measurement, material sheet or teaching assumption. Add units to input cells. A ratio calculated from kg/L and kg/m³ without conversion produces a substantial error despite an apparently correct formula. Preserving these labels makes the calculation easier to review and prevents the same numerical value being reused with a different physical meaning.
Finally, verify that percentages of fractions, binder and other constituents use the same basis when added together. Binder content relative to aggregate mass is not content relative to total composite mass. Preserve the denominator in the calculation heading to avoid ambiguity when moving from laboratory work to the manufacturing record. The receiving specification and trial summary should use the same convention so that another operator can reconstruct the formulation without guessing how its percentages were defined.
16. Prepare the receiving specification
The retained combination should be translated into verifiable requirements: fraction identities, distribution, method, preparation condition and traceability. Specify which lot or material changes trigger a check or a new trial.
Industrial silica sand remains a material requiring qualification for this application. A purity designation or commercial class does not replace the record linking grains to the resin system and product.
Finally, retain the selected reference, individual results and limits of the tested domain. A well-constructed grading curve has transparent calculations and effects verified in a reproducible formulation. This continuity between data, trials and receipt makes the decision usable.