The right question: does the new sand suit the already qualified process?
Changing a silica-sand supplier involves more than comparing a price and a SiO₂ percentage. The objective is to establish whether the new material can be received, stored, dosed and processed with quality compatible with current production. A satisfactory sample is one step in that decision, not a demonstration of consistency across all subsequent deliveries.
Morocco’s Ministry of Industry describes Casablanca-Settat as the country’s leading industrial centre 1. That context justifies addressing material continuity for buyers in the region. It does not establish local silica consumption, stock availability or any particular delivery capacity.
The protocol presented here is a practical qualification framework proposed by Minerallum. It uses documented principles of sampling and experimental comparison. It does not replace a specification for the finished product. Its purpose is to make decisions explicit: what must remain identical, what may vary and what requires another trial.
1. Separate substitution from reformulation
Two different projects may lie behind a supplier change. In substitution, the aim is to maintain the existing formulation and settings. In reformulation, changes to proportions, water, binder or process parameters are accepted to take advantage of the new material.
The first question is therefore: “Are we seeking material that is interchangeable under current conditions, or material that becomes usable after adjustment?” Without this distinction, a team may qualify a sand using a corrected recipe, then introduce it into production with the old recipe. The trial result no longer answers the decision actually being made.
Define the authorised scope in writing: application, product reference, equipment, sand fractions and parameters that may be changed. A necessary modification is not automatically a commercial failure. It must simply be documented, assessed and approved as an adjustment, including its effects on cost and production.
2. Establish a usable current reference
The reference is the sand or production against which the candidate will be compared. It must be recent enough, identified and correctly preserved to represent the situation being investigated. An old bag remainder that has absorbed moisture or been mixed with another lot is not a sound reference.
Associate the reference with its origin, lot, receiving date, packaging and test results. Add results for finished products made with that material where traceability exists. Retain manufacturing observations as well: anomalies, setting changes, stoppages and unusual conditions.
This preparation separates “the candidate differs from an isolated result” from “the candidate deviates from normal operation”. If the current material itself shows a drift, that drift must be identified before the material is treated as a reference. Otherwise, a new material may inadvertently be selected to reproduce a defect.
3. Compare lots, not just certificates
The ISO 14488 description addresses obtaining a sample representing a defined quantity of particulate material 2. In supplier qualification, this means connecting each analysis to the lot actually represented. The supplier’s name alone does not define that lot.
Ask how the commercial sample was assembled and whether it corresponds to the product offered for sale. Clarify its preparation: untreated, washed, dried, classified or a blend of fractions. Check whether several certificates refer to separate production runs or several samples from the same stock.
Multiple portions of a delivery allow that delivery to be investigated. They do not replace observations of independent lots when assessing supply continuity. The number of lots and tests depends on product risk and expected variability. No universal number can be inferred from the word “sand” alone.
Our guide to silica-sand sampling develops this distinction.
4. Reduce the sample without selecting the preferred grains
A laboratory generally works on a smaller portion than the original sample. ASTM C702’s public description explains that reduction should limit differences between the test portion and the larger sample 3. It also identifies difficulties associated with rare constituents.
For the comparison, ask the laboratory to use a procedure appropriate to the material and analyses. Do not prepare a sample by visually choosing the cleanest grains. Do not send only material from the top of a bag when the sample is intended to represent the whole bag.
The quantity retained depends on the planned tests and possible repeat analyses. Identify retained portions, closure arrangements and storage conditions. For moisture measurement, preservation and elapsed time form part of the protocol: a portion that dries during transport does not answer the same question as material in its received condition.
5. Write a matrix of relevant characteristics
The comparison must address the properties that explain the expected behaviour. The list below is a preparation tool, not a set of limits imposed on every application.
| Category | Question to resolve | Information to obtain |
|---|---|---|
| Grain distribution | Are critical fractions still controlled? | Curve and passing or retained fractions |
| Delivery condition | Is dosing comparable? | Moisture and declared preparation |
| Composition | Are troublesome constituents introduced? | Analyses appropriate to the application |
| Handling | Does the material behave correctly in the installation? | Unloading and dosing observations |
| Finished product | Does the result satisfy requirements? | Contractual or validated internal tests |
For sands with different gradings, compare actual distributions rather than commercial designations alone. Material labelled “fine” is not a specification. If the reference uses several fractions, also check their blending proportions and agreed tolerances. The industrial silica-sand guide places this information within the overall requirement.
6. Set acceptance criteria before seeing the results
Before testing, classify characteristics into three categories: a non-negotiable requirement, an allowed variation and a monitoring parameter. For each acceptance criterion, identify the method, unit, limit and action if a result lies close to that limit.
Do not turn a historical average into a contractual limit without justification. An average describes observed results; a limit expresses a requirement. They may be related, but they are not interchangeable. A currently compliant product may, for example, have a greater margin than is necessary for operation.
The protocol must also identify who decides. Purchasing can compare commercial terms, the laboratory can produce results and production can report unusual behaviour. Final acceptance must bring this information together rather than allowing each department to interpret trial success separately.
7. Organise a trial that actually isolates the sand’s effect
NIST distinguishes designs intended to compare a factor from designs used to identify important factors or optimise a response 4. For substitution, the factor under investigation is initially the candidate material, not an uncontrolled combination of new conditions.
Prepare a trial sheet covering materials, masses, settings, steps and measurements. Record changes made during the trial. If an operator adds water or changes mixing time to correct behaviour, that adjustment is an important observation. It must not disappear from the report.
A smaller trial can help eliminate an obvious incompatibility. However, it represents only the conditions it reproduces. Manual handling alone does not demonstrate the consistency of automatic dosing or stability throughout a long production campaign. Each stage must clearly state the question it can resolve.
8. Avoid confusing supplier with production day
Comparing the previous sand on Monday and the new sand on Friday can introduce differences in temperature, team or complementary materials. NIST describes blocking as a way to account for controllable nuisance factors 5.
In an adaptation for a workshop, a block might correspond to a series conducted under similar conditions. The materials are then compared within that series in a planned order. Residues from the previous material and transitions within the installation must also be considered. A theoretical alternation is not useful if the materials cannot be properly separated.
The person responsible for testing chooses relevant blocks, repetitions and an arrangement compatible with the equipment. The practical principle is not to systematically reserve the most favourable conditions for the candidate, or the least favourable conditions for the reference. The plan must remain feasible and traceable.
9. Read a measured difference in context
A difference may be statistically detectable without practical consequences. Conversely, an imprecise trial may fail to detect an important difference. NIST’s test for comparing means addresses a question of statistical equality under defined assumptions 6.
Writing “no significant difference” is therefore insufficient to demonstrate that two sands are interchangeable. The decision must consider the acceptable margin defined before testing, measurement precision and the amount of information available.
Present individual results, their variability and the associated conditions. An average alone can conceal a trial requiring adjustment or a failing lot. When a result is ambiguous, the useful response is to identify the missing information: an independent repetition, a more appropriate measurement or a trial under a condition not yet represented.
10. Do not confuse one successful lot with capability
Capability compares the performance of a stable process against specification limits. NIST highlights the assumptions and data requirements associated with common indices 7. An index calculated from a few results is not robust evidence of supply continuity.
For initial qualification, separate two conclusions: “the tested lot meets the criteria” and “the production system maintains these criteria over time”. The second requires consistent monitoring of relevant production. Repeatedly measuring the same portion mainly provides information about that measurement, not variability between deliveries.
Retain the criteria after introduction and identify changes in source, classification or preparation. Material can retain the same commercial name while its industrial history changes. Traceability of changes makes it possible to decide when additional qualification is necessary.
11. Prepare introduction into the actual installation
Progressive validation reduces the number of open questions when the change takes place. It may include a document comparison, material analyses, a finished-product trial and controlled introduction into the installation. The extent of each stage depends on risk and process.
Before industrial introduction, specify the stock involved, identification of manufactured products, sampling points and the procedure in case of a deviation. Avoid prematurely mixing candidate material into the reference stock if that prevents observations from being attributed to one or the other.
When several fractions are used, do not change every reference simultaneously unless the protocol allows their effects to be identified. A controlled transition must preserve the ability to explain a result. This is particularly useful for a small series, where little material is available to repeat a trial.
12. Compare useful cost rather than the quoted price
Qualification cost can be examined without inventing a market price. For a given requirement, separately identify material, transport, analyses, preparation losses and process adjustments. Compare offers on the same delivery condition and mass basis.
The following relationship is an internal calculation tool: material cost per usable tonne = cost of the delivered and prepared lot / corresponding usable mass. It does not predict a yield. Numerator and denominator must come from the same lot and the same scope.
For example, a moist bag and dry material cannot be compared solely by gross mass if the application requires a defined quantity of dry material. Likewise, a sand requiring repackaging must be compared with that operation included. For payment terms and quantities, use only commitments actually negotiated.
13. Turn the trial into a documented decision
The final file should allow a person who was absent from the trial to understand the decision. Bring together the objective, lots, analyses, manufacturing conditions, results, deviations and agreed actions. Identify the recipe or setting version to which the conclusion applies.
Four decisions are possible: acceptance within the tested scope, acceptance with an approved adjustment, further trials or justified rejection. A favourable result for one application must not become general approval for every production activity. A conditional decision must identify its condition and the person who checks its fulfilment.
Finally, specify the information to request with subsequent deliveries and the events triggering another review. The best qualification file is not the longest: it connects identifiable material to observed behaviour and a comprehensible decision rule.
Conclusion: qualify continuity, not just the sample
For a buyer in Casablanca, a controlled change starts with a clearly defined industrial requirement. It combines a usable reference, representative lots, predetermined criteria and trials able to distinguish the sand’s effect from other variations.
The useful conclusion is not “this sand appears similar”. It states the application, grading, delivery condition and settings under which the material was accepted, followed by how that acceptance will be monitored. This framework makes supplier discussions more precise while retaining priority for the product actually manufactured.