Qualifying a resource requires more than a favourable analysis
A silica sand resource may have interesting potential for glass without immediately supplying qualified material. Between sampling and ordering, the sample needs to be connected with the resource, impurities to be removed must be identified, yields checked and the resulting product compared with the glassmaker’s requirements. A staged programme avoids committing to a complex process before knowing which obstacle it needs to resolve.
Jordan provides documented context for this approach. In its announcement of 27 March 2025 about Dabat Hanout, the Ministry of Energy describes work involving mineralogy, microscopic observation, particle size distribution and exploration of resource layers 1. This communication locates a particular project. It qualifies neither every Jordanian sand nor a Minerallum offer.
This guide proposes a test programme for preparing glassmaking qualification. Unlike simply choosing between washing and separation, it organises decisions from representativeness to trials with the user. The stages are proposed working arrangements to adapt to the resource and target product, without announcing an unmeasured performance.
1. Define the target glass before target purity
Start with the user’s specification: glass type, composition constraints, required fraction, impurity limits and control criteria. “As pure as possible” defines neither the useful product nor processing economics. A chemical improvement can lose its value if it destroys the required distribution or excessively reduces saleable quantity.
The public scope of ASTM C146-26 covers chemical analysis of glass sand and checking composition against a specification 4. It documents the analytical scope, rather than a universally acceptable composition for every glass. Limits must therefore connect with the user’s requirements and the agreed method.
Separate mandatory criteria from those still under discussion. Identify properties that can be checked on sand and those requiring examination within the glassmaking process. This avoids treating a chemical certificate as a complete demonstration of melting suitability.
2. Organise samples around the question they represent
The official description of ISO 14488 concerns representative sampling and splitting of particulate materials 3. For a resource programme, the first question is what each sample represents: a zone, layer, delivery or particular preparation.
Build a matrix distinguishing individual samples from composites. A composite may help explore a process but conceal differences between layers. Retaining origin information allows the project to revisit material that produced an atypical result. Available mass must also support analyses, trials and retained references.
Do not attribute the entire resource’s variability to a small sample. The programme should explain how subsequent stages will incorporate variability. If initial material was selected for apparent quality, record that: its result does not automatically describe other zones.
3. Separate composition from the occurrence of impurities
Chemical composition reports quantities, whereas mineralogical and microscopic investigation asks how constituents occur. The Jordanian announcement confirms the importance of these observations within its project 1. Connect both types of information in the qualification record.
For the proposed trial plan, distinguish separate particles, surface coatings and associations with quartz. This classification helps formulate processing hypotheses. It should not be inferred solely from a total iron or aluminium concentration. Materials with similar compositions may require different investigations.
The laboratory should explain its observation method and scale. Absence observed in a few fields does not establish absence of a rare constituent throughout a lot. Preserving detection limits and examination conditions strengthens the decision without turning the record into a promise of absolute purity.
4. Analyse individual particle size classes
Separating several classes allows comparison of where useful mass and impurities occur. For every class, measure dry mass and relevant concentrations, then reconstruct the feed balance. This avoids assuming that all contamination is concentrated in fines.
The proposed procedure should specify cut sizes and their rationale. If the user requires a particular range, include that range in interpretation. A very clean class that is too fine for the application does not automatically have the same value as a fraction meeting requirements. Connect quality with quantity.
Check balance closure before interpreting differences. If recovered masses do not sum to the feed, investigate losses or inconsistent moisture conditions. A favourable concentration obtained after poorly monitored preparation may provide a poor basis for selection.
5. Read a Jordanian example without generalising it
SGS report 19097-03 published by the ministry examines three samples in a processing programme. Final residues analysed gravimetrically contained 99.66%, 99.80% and 99.58% SiO₂, below the 99.9% objective. The document presents a proposed flowsheet requiring validation 2.
This illustrates the difference between target, trial result and validated process. Values concern the examined samples, not all Jordanian resources, and are not Minerallum product specifications. Displaying additional result digits does not remove the question of its scope.
Use such a case as a methodological example, rather than a reproducible recipe for another project. Another resource may have different mineral associations, distributions or glassmaking requirements. The first decision remains explaining what must be removed and what treatment must retain.
6. Compare initial operations using the same feed
Start comparative trials from a common reference. Describe each preparation and preserve charges allowing an untreated reference to be compared with variants. Comparing two samples of different origin cannot isolate the processing effect.
At each stage, measure at least recovered mass, product grading and chemical parameters answering the question. An improved concentration without yield information is incomplete. Where an operation produces several streams, identify them and retain each destination in the balance.
The programme may stop after a simple stage if the product meets verified requirements. Conversely, non-compliant concentration should trigger targeted investigation rather than automatic addition of every available operation. The purpose is to identify the necessary process, not build the longest treatment train.
7. Calculate yield and recovery separately
Mass yield is dry product mass divided by dry feed mass. Constituent recovery compares its mass in the product with its mass in the feed. These results answer different questions.
In a fictional example, 100 kg of feed contains 95% SiO₂. An 80 kg product at 99% SiO₂ represents an 80% mass yield. It contains 79.2 kg SiO₂, so silica recovery is 79.2 / 95, approximately 83.4%. Concentration increased, but some silica no longer belongs to the retained product.
For impurities, also calculate the fraction rejected in excluded streams. Use identical units and the same dry basis. A reject with high impurity concentration may account for a small proportion of total impurity mass if the stream is very small. The complete balance is more informative than concentration alone.
8. Organise analysis near acceptance limits
Request a method and precision consistent with the decision. A low-concentration limit cannot usefully be checked with a method incapable of distinguishing the relevant region. ASTM’s public description notes reference-material use to verify analytical procedures 4.
NIST defines a decision rule as how measurement uncertainty is accounted for when stating conformity with a requirement 5. Agree that rule with the laboratory and user before trials. A result close to a limit should not receive an interpretation improvised after reading it.
Also separate elements from oxides: an elemental iron value cannot be directly compared with Fe₂O₃ without conversion and clarification of convention. Keep the reported form in every table. Conversions must remain visible so recipients can verify what was compared.
9. Test variability before increasing scale
In its flotation-programme documentation, SGS describes staged work and variability assessment before more advanced process definition 6. This concerns trial organisation; it does not demonstrate that flotation is necessary for the project’s sand.
For the proposed programme, repeat the promising variant on samples covering the initially identified conditions. Compare quality, yield and operating behaviour. If one zone fails, preserve its result separately instead of allowing it to disappear within a favourable composite average.
Then define progression to a pilot. Explain which laboratory conditions will be retained, which will change and which observations will assess the transition. A small-charge result does not automatically describe continuous operation, water recirculation or management of every stream.
10. Treat chemical routes as complete projects
If trials suggest a chemical route, selection must cover equipment, consumption, effluents and final-product preparation. This guide supplies no acid recipe. The question is the operation’s industrial relevance within the complete treatment train.
Compare useful benefit with the new constraints. If chemical improvement requires grinding that moves the product away from the requested fraction, discuss the resulting product with the user again. Higher concentration alone does not demonstrate a better material for the original application.
Economic balances should use measured consumption or explicitly defined pilot assumptions. Separate investment, operation and stream management. A laboratory-effective operation is not automatically the best solution when the project must produce, deliver and control consistent material.
11. Qualify the product with the glassmaker
After material acceptance, application trials must answer the glassmaker’s specification. Agree the reference batch formulation, process criteria and observations on the glass obtained. The sand certificate and glassmaking-trial result are complementary.
Avoid simultaneously modifying many materials without an explicit plan. If sand changes alongside cullet or another ingredient, identify that situation in the trial. Overall product improvement cannot automatically be attributed to the new sand.
The acceptance record should state what was checked, on which lot and under which conditions. It should also identify material changes requiring another check. Qualification then becomes a documented relationship between sand preparation and a defined glassmaking requirement.
12. Use decision gates to limit unnecessary spending
| Stage | Main question | Expected evidence |
|---|---|---|
| Sampling | What does the material represent? | Origin and sampling plan |
| Characterisation | Which impurities and classes? | Connected analyses and observations |
| Initial trial | Is a simple operation sufficient? | Quality and mass balance |
| Variability | Does the result hold across conditions? | Comparisons using distinct materials |
| Pilot | Can operation be transferred? | Monitored streams and performance |
| Application | Does the product meet glassmaker requirements? | Use trial and accepted criteria |
This table is a proposed project-management tool. An unfavourable stage can prompt target revision or another application without concealing results. At every gate, useful observations should explain the decision.
The silica sand guide helps define supply requirements. For a Jordanian resource, constructing trials before promising a grade is the strongest approach: connect material, processing, useful quantity and behaviour in glass rather than confuse potential with qualification.
13. Check units and the reporting basis
To compare a certificate with a limit, put both on the same basis. A percentage expresses mass per hundred masses of material; parts per million express mass per million. Thus 0.01% equals 100 ppm on the same mass basis. This conversion changes neither analytical method nor uncertainty: it changes only the reporting unit.
A wet-material result must not be compared uncritically with a dry-material specification. Water mass belongs in the denominator of the former. Ask the laboratory to state the calculation basis rather than infer it from decimal places. Apply the same discipline to normalised results: explain whether concentrations were recalculated to a constituent sum.
For iron reported as oxide, conversion depends on the selected chemical convention. State “Fe” or “Fe₂O₃” explicitly. Correcting units can make results comparable, but does not prove they came from equivalent methods. Analytical validation remains a separate question.
14. Close an impurity balance with an example
In a fictional example, 100 kg of dry feed at 0.10% impurity contains 0.10 kg of that impurity. If 80 kg of product at 0.025% contains 0.020 kg, impurity recovery into product is 20%. Excluded streams must then explain the remaining 0.080 kg, subject to measurement precision and losses.
This separates concentration change from mass distribution. Saying only that concentration fell fourfold does not show where the impurity went. Analysing rejects checks whether separation follows the initial hypothesis or whether some mass is missing.
Use the same table for each variant. Place masses, concentrations and calculated constituent masses side by side for review. If closure is inconsistent with measurements, resolve that issue before ranking treatments. A variant should not be selected solely because its final certificate appears favourable.
When several size fractions are produced, calculate balances for each fraction before combining them. A clean fine fraction may represent little recoverable material, while a larger fraction carries most of the usable silica. Recombining fractions changes both chemistry and particle distribution; assess the combined product against its intended use rather than averaging certificates without mass weighting.