Can a compliant incoming curve change before dosing?
Silica-sand segregation is the uneven distribution of constituents during handling. To investigate it between storage, transport and dosing, compare receiving-lot results, observations during transfers and samples at the dosing point. Sand may show a satisfactory distribution on receipt and subsequently feed the process with varying proportions of fine and coarse grains. This difference deserves investigation before changing a formulation or attributing a defect to the supplier.
The Ministry of Industry’s project sheets document Berrechid’s Ecoparc industrial park 1. They provide local industrial context without measuring silica-sand consumption. This guide addresses teams organising receipt, storage and granular-material feeding in that area.
The guide proposes an observation and trial approach for the handling circuit. It complements the particle-size distribution guide: knowing a curve is one step; maintaining feed compatible with that curve is another.
1. Distinguish three problems before correcting anything
A material variation means genuinely different lots arrive at the site. Segregation means a material’s constituents become unevenly distributed during handling. Sampling bias means a sample does not adequately represent what is being measured. All three situations can produce differing particle-size results.
The initial investigation should therefore compare information answering distinct questions: the result associated with the received lot, observations during transfers, and results at the dosing point. A surface sample from a pile cannot establish the composition of the entire delivery on its own.
Writing the question before each sampling operation reduces mistaken conclusions. “Characterise the lot” is not “detect variation between the start and end of discharge”. The first question requires lot representation; the second requires temporal information to be preserved.
2. Understand the mechanism without imposing a universal pattern
Grains can separate according to size and other properties during flow. An experimental study by Samadani, Pradhan and Kudrolli observed segregation in a quasi-two-dimensional silo containing glass beads of different sizes 3. It illustrates a physical phenomenon without directly predicting an industrial hopper’s behaviour with angular sand.
It would therefore be incorrect to turn a laboratory diagram into a rule that “large grains are always in this position”. Geometry, filling and discharge must be examined in the actual circuit. The useful approach is to identify an observed distribution, then test whether it explains production variation.
Jenike & Johanson’s guidance distinguishes funnel flow and mass flow 4. The latter can reduce certain segregation effects; achieving it requires design adapted to material properties rather than simply a slope assumed to be sufficient.
3. Map the material chain through to the process
Represent the steps in order: delivery container, unloading, stock, reclaiming, conveying, buffer hopper, feeder and mixer. For each step, record geometry, filling method, residence time and whether lot identification can be preserved.
This drawing must include returns and mixing. Material recovered during cleaning and reintroduced into a hopper belongs to the chain. An old charge remaining at the bottom of a container can also obscure comparison between two deliveries.
The diagram need not be a complete engineering drawing. Its purpose is to show where an observation can be linked to a step. Adding authorised, accessible sampling points allows the investigation to be organised without improvising actions while equipment is operating.
4. Define the property that actually matters
Particle size plays different roles according to the application. In a formulation, the question may concern proportions retained on particular sieves; for a finish, a coarse fraction may become visible; with volumetric dosing, several material characteristics may matter.
Define indicators from the validated requirement, then retain them throughout the investigation. An isolated D50 can remain similar while the distribution’s tails change. Comparing the complete curve and sensitive fractions avoids missing a relevant variation.
The measurement method must remain consistent. The NIST guide addresses the principles and limitations of several characterisation techniques 2. A sieve result and a diffraction result do not become interchangeable because both are displayed in micrometres.
5. Separate overall receipt from the discharge profile
ISO 14488’s scope concerns sampling and division of particulate materials 5. This reference frames the subject; a public catalogue entry does not supply every detail of an installation-specific plan.
For the investigation, arrange two series where resources allow: a representation of the received material and an output profile preserving positions within discharge. The protocol should define what those positions represent: elapsed time, mass already extracted or another measurable reference.
Start, middle and end are practical categories, not automatically equal proportions. Where flow varies, equal time periods can represent different masses. Recording the chosen basis prevents a misleading average when reconstructing the lot.
6. Establish a record for every sample
A minimal record contains the lot identifier, sampling point, position in the sequence, date, configuration and preparation method. Add sampled mass, operator observations and the laboratory or person responsible for measurement.
Keep individual results before calculating an average. Combining every increment into one composite can be useful for overall characterisation but can erase precisely the variation the investigation seeks to detect.
Sample reduction should also be documented. If two diverging results come from poorly divided portions, they do not establish segregation in the circuit. Preparation control belongs to the evidence chain just as much as the choice of sampling point.
7. Use a weighted balance correctly
Where several portions represent known masses, an average proportion can be calculated by weighting. For a given fraction, multiply the measured proportion in each portion by the mass it represents, add these contributions and divide by the total represented mass.
Teaching example, without site measurements: three periods represent 100, 200 and 100 kg respectively, with 10%, 20% and 30% of the tracked fraction. Contributions are 10, 40 and 30 kg, giving 80 kg in 400 kg: 20%. The values are fictional and demonstrate the calculation only.
This average does not remove temporal variation. The process may have received different compositions successively even when the overall balance appears compliant. Keeping the weighted average alongside the output profile preserves this distinction.
8. Observe filling without changing several parameters
Document the entry point, drop, distribution and filling levels. During a corrective trial, compare the current configuration with one identified modification while keeping other parameters as constant as possible.
A reduced drop height or different distribution is an improvement hypothesis to measure. It should not be presented as an established result before testing. The objective is to verify composition at dosing, not merely to obtain a visually more uniform pile.
Also record constraints: access, required flow, dust-control equipment and operating rules. A quality solution that creates an impracticable operation cannot be adopted unchanged. The comparison should include these conditions from its preparation.
9. Examine reclaiming and residual volumes
Reclaiming can mobilise some zones before others. Where stock or a hopper retains an older volume, identify whether the observed discharge corresponds to a single delivery or a mixture. The history should distinguish this from variation between suppliers.
The circuit diagram can then be supplemented by a sequence of fillings and withdrawals. It becomes possible to relate a particle-size change to a change of source or level. This relationship is a lead; samples and results must confirm it.
A structural hopper modification requires material data and a design assessment. Neither a generic angle nor an accessory selected from a photograph guarantees mass flow for every sand.
10. Do not confuse flow variation with composition variation
A feeder can deliver varying mass without a particle-size change, or varying composition at constant mass. The amount delivered and grain distribution therefore need separate tracking where both matter.
For volumetric dosing, record the filling method and material condition. Moving to a mass-based approach does not remove the need to control composition. Accurate weighing answers “how much?”; it does not automatically answer “which grains?”
The record should identify instruments and their verification, together with the duration or mass covered by each result. This prevents comparison of a cumulative shift mass with a single particle-size measurement and an artificial correlation being inferred.
11. Keep moisture and preparation in the investigation
A moisture difference can complicate mass comparisons and change sample preparation. Record analytical conditioning and the reporting basis: as-received, dry or another explicitly defined convention.
This is not a reason to add water to prevent separation without understanding the application. A moisture modification may be incompatible with storage or formulation. Any material modification should be treated as a change requiring qualification.
The moisture-control guide complements this point. Within a segregation investigation, this information makes comparisons consistent without turning the record into a new drying specification.
12. Build a causes-and-evidence matrix
| Hypothesis | Observation sought | Useful verification |
|---|---|---|
| Different incoming lots | Variation already present at receipt | Representative samples from independent deliveries |
| Separation during filling | Variation associated with positions or sequences | Comparative trial of an identified configuration |
| Selective discharge | Changing output profile during withdrawal | Sequential sampling with represented masses |
| Analytical bias | Divergences associated with preparation | Controlled repetition of division and measurement |
| Mixed stocks | Transition consistent with reclaiming | Charge history and residual volumes |
The matrix prevents a hypothesis being treated as a demonstrated cause. A “verification not performed” entry is useful: it shows what is missing for a decision and prevents an intuition being presented as a technical conclusion.
13. Define success before the correction
Establish comparison criteria before modifying the circuit: acceptable range for a critical fraction, profile consistency, product effects and operating constraints. Tolerances come from the application; this guide does not propose a universal segregation limit.
A successful trial should demonstrate improvement under representative conditions, not just during a selected discharge. Repetition on independent operations allows the effect’s stability to be examined. Several portions of one operation do not represent several independent operations.
Keeping unfavourable results is necessary. They establish whether the modification works only for a particular fraction or condition. A partial solution may remain useful, but its operating domain must be clearly documented.
14. Translate findings into usable instructions
Final instructions should describe the retained configuration, filling method, reclaiming management, checks and exceptions. They should be understandable to the team performing the operation, not just the team investigating the problem.
Plan for unusual situations: a different delivery, extended shutdown, cleaning, container change or loss of traceability. Instructions should identify the verification that allows a return to normal operation.
The silica-sand page connects this approach with overall material qualification. A selected sand and a controlled circuit are two components of the same industrial requirement.
15. Compare results without concealing variability
A difference in proportion between two samples must be interpreted alongside measurement precision and how the samples were obtained. ASTM C136/C136M’s catalogue entry places sieving within determination of aggregate distribution 6. The laboratory protocol should specify the steps actually applied rather than simply citing a standard reference.
Where several analytical repeats originate from one portion, they provide information about that measurement’s repeatability. They do not directly describe variation between delivery days. Keep these levels separate in the table: lot, discharge operation, sample and analytical repeat.
For a contractual indicator, record the result, unit, calculation basis and uncertainty where available. A value near a limit calls for the agreed decision rule rather than convenient rounding to display compliance. Apply the same rule to the current supplier and the corrective configuration.
Finished-product monitoring should retain its own criteria. If the material shows measurable variation but an effect on the process has not been established, the record must preserve that distinction. Conversely, a persistent product defect despite stable grading may indicate that another property deserves investigation.
This approach avoids confusing a detected difference with an industrially important difference. It also allows subsequent checks to focus on information that actually contributed to the decision. A check added without a defined question burdens operation; a targeted check preserves the necessary evidence without multiplying analyses unnecessarily.
16. Preserve evidence through to the finished product
The useful outcome is feed compatible with the process and its product, rather than an isolated curve that looks better. Linking material samples to actual production allows examination of whether the observed variation explains the defect being investigated.
The final record should retain the diagram, methods, individual results, weightings and decision. A circuit modification can then be reassessed without restarting the investigation from scratch.
In Berrechid as in other industrial areas, particle-size control does not stop at the delivery certificate. It continues through transfers, stock and dosing, with reproducible observations and demonstrated corrections.