Qualify a silica fraction for a specific repair
A refractory repair mix must perform a function in identified equipment and on a defined substrate. Choosing a silica-sand fraction solely by SiO₂ content or an assumed service temperature is insufficient. The grain belongs to a system comprising matrix, binder, application method and thermal treatment.
Germany provides a concrete standardisation context: DIN publishes the German version of the EN ISO 1927 series for monolithic refractory products. Part 1’s catalogue entry distinguishes these products from granular raw materials without binder 1. Sand alone is therefore not qualified as a monolithic mix simply by association with that reference.
This guide complements the general guide to silica and refractories. Its purpose is narrower: establish the information needed to compare a fraction in a repair mix and avoid substitution between chemically similar but technically different materials.
1. Identify the repair area and function
The first decision concerns the defect to address: joint, crack, eroded surface, volume reconstruction or temporary protection. Describe its geometry and access, together with the existing material. A thin surface layer and volume reconstruction do not impose the same demands on a granular skeleton.
Define the intervention timing and intended process. Cold installation, gunning and specialised hot repair involve different systems. Required grading must be linked to the product and its permitted method rather than a general request for “furnace sand”.
Resco’s coke-oven repair range distinguishes several families, including mortars, gunning mixes and fused-silica products 2. This diversity illustrates the need to identify a function; it recommends none of these products for an unexamined installation.
2. Separate chemical identity from silica condition
Crystalline quartz in sand and fused silica must not be confused. SiO₂ content may be high in both, but chemical similarity does not establish identical thermal behaviour.
RHI Magnesita describes its COMPAC COKE blocks as fused-silica products designed for coke-oven repairs 3. Their advertised characteristics concern these industrial blocks. Attributing them to ordinary quartz sand would be an incorrect extrapolation.
The material request should therefore specify the expected identity and preparation: quartz sand, another processed siliceous aggregate or a formulated product. Where a specification requires fused material, replacing it with merely washed or graded sand changes the system. Grading cannot compensate for that difference in identity.
3. Link distribution to repair dimensions
A commercial class does not describe all grain proportions. To compare two fractions, retain passing or retained percentages at the same diameters using a consistent method. An advertised maximum grain does not describe the actual coarse tail or fine proportion.
The formulator needs installation geometry: thickness, passages, corners and contact surfaces. The grading criterion follows from the qualified formulation and these constraints. This guide proposes no universal ratio between thickness and maximum grain for every repair method.
Where the mix contains several fractions, document their combination. Changing the fine fraction can change the system even when the coarse fraction remains identical. Control should follow the prepared product’s distribution, not merely bag labels.
4. Describe skeleton and matrix separately
The formulation record should distinguish skeleton aggregates from finer matrix constituents. Overall chemical analysis can conceal a changed distribution between these functions.
For each constituent, record fraction, lot, preparation and proportion on a defined basis. Total SiO₂ content alone does not reveal which constituent contains that silica. Comparison should preserve material identities rather than reduce the entire mix to an oxide sum.
This separation helps organise a substitution trial: change an identified fraction, keep a reference and examine responses. Where several constituents change together, the conclusion concerns the new combination rather than an isolated sand effect.
5. Examine substrate compatibility
A repair works with the existing material rather than alongside an abstract specimen. Identify substrate composition, surface condition and changes during service. A contaminated or transformed surface can make a trial on a new substrate unrepresentative.
The qualification programme should address the interface: intended preparation, bonding, dimensional changes and exposure. Good strength measured within the repair volume does not automatically establish performance at its contact with the old lining.
Where comparison uses laboratory-prepared substrates, specify what they represent and what they do not reproduce. An intermediate stage can examine geometry and contact condition before service testing. The objective is to avoid presenting raw-material validation as assembly validation.
6. Define thermal conditions rather than temperature alone
Maximum temperature is only one item. Add the traversed range, heating and cooling stages, durations and continuous or cyclic operation. Distinguish measured process temperature from temperature genuinely representative of the repair area.
German research published at UNITECR 2023 compared refractory materials under different thermal-shock modalities and observed rankings dependent on test conditions 4. It concerns its specimens, including fused-silica bricks, and does not qualify commercial sand.
The consequence is to choose tests corresponding to the intended loading. A good result in one range does not guarantee another. The protocol must preserve this relationship between cycle, geometry and result.
7. Do not turn cold strength into service life
An ambient-temperature mechanical measurement can contribute to prepared-material control. It does not alone represent behaviour under load, chemical contact or hot cycling. The results table should identify each specimen’s thermal condition before measurement.
DIN EN ISO 1927-6’s catalogue entry describes a testing scope for certain monolithic products before and after firing 6. Respect its scope limits: it does not provide a universal procedure for every welding powder or specialised mortar.
Choose methods with the formulation manager and laboratory. Preserve method, dimensions, conditioning and units. A higher value becomes an advantage only where the property corresponds to a useful criterion and other requirements remain satisfied.
8. Treat the binder as an active system constituent
The binder does more than hold grains before use. Its identity, quantity and preparation belong to qualification. A fraction selected for one system does not automatically become interchangeable in another bonding system.
Product instructions should determine permitted additions and procedure. Adding sand to a ready-made mix to reduce cost changes its proportions; that modification requires the formulator’s validation. It is not merely a volume adjustment.
Keep the same calculation basis between references. A proportion relative to dry material cannot be directly compared with one relative to the prepared mix. Liquid constituents, where present, should remain identified in the balance.
9. Verify the mix actually installed
Delivered material, prepared material and material retained in the repair correspond to three control levels. Qualification should identify the level to which each sample and measurement applies.
For an application trial, collect useful observations: homogeneity, equipment behaviour, deposit consistency and layer condition. Where rejects or rebound are measured, distinguish them from mass actually retained. Mass loaded into equipment is not necessarily mass deposited in the area.
A teaching balance can be written: prepared mass = deposited mass + recovered material + identified losses + change in material held in the circuit. An unexplained difference should not automatically become an application yield. Define boundaries and units before comparing two trials.
10. Prepare specimens for the property investigated
DIN EN ISO 1927 Part 5 concerns specimen preparation and treatment, including storage, drying and firing, for families defined within its scope 5. Its public entry explains this organisation without supplying every execution instruction.
The trial record should preserve the complete preparation applied to the product. Comparison between two mixes is interpretable only where preparation differences are known. Record the time between mixing and forming where it matters to the method.
Specimens prepared differently from the actual repair can remain useful for standardised control. Present their results for that function. They do not directly demonstrate that the field application will reproduce the same material.
11. Preserve the commissioning thermal treatment
Drying and initial heating belong to the selected system. Their procedure depends on formulation, dimensions and equipment. Proposing a common heating schedule for every quartz-containing mix would be incorrect.
The record should distinguish manufacturer instructions, a laboratory programme and the procedure actually used in service. Where these differ, identify the validation permitting transfer. A common final temperature does not make two thermal histories equivalent.
Preserve available readings and procedural deviations. A repair failing after a different startup does not directly isolate a grading effect. Investigation should include installation, treatment and operating conditions.
12. Read results through a qualification matrix
| Level investigated | Data to preserve | Possible conclusion |
|---|---|---|
| Raw fraction | Identity, preparation, curve and lot | Material conformity with defined criteria |
| Prepared mix | Constituents and procedure | Behaviour of the investigated combination |
| Treated specimen | Geometry and thermal history | Result under the test conditions |
| Substrate interface | Substrate condition and preparation | Performance of the tested assembly |
| In-service repair | Installation and operation | Observed performance within that domain |
The table prevents stages being skipped. A sand certificate does not validate a repair. Conversely, satisfactory repair with several changed materials does not demonstrate one fraction’s isolated superiority.
13. Check fraction and deposited-material balances
Consider a fictional teaching calculation: a dry blend contains 30 kg of fraction A and 70 kg of fraction B. At a given diameter, measured passing percentages are 90% and 10% respectively. Predicted blend passing is therefore (30 × 0.90 + 70 × 0.10) / 100 = 34%. This relationship checks the grading table; it does not determine an acceptable refractory formulation.
Then compare a measurement of the prepared blend with that prediction on the same basis. Where it differs, check lots, masses and sample representation before adjusting the recipe. A fraction separating in the hopper can make the final sample differ from the calculated average. Keeping individual results allows this hypothesis to be examined.
For the application balance, a second fictional example separates flows: 100 kg prepared, 75 kg retained and 15 kg recovered leave 10 kg to explain through identified losses or change within the circuit. The 75/100 ratio describes a retained fraction within this boundary. It measures neither service life nor chemical material performance.
14. Link mass, volume and geometry without mixing units
When estimating consumption per area, thickness must follow the same measurement convention. In a fictional uniform example, 2 m² covered at 20 mm represents 2 × 0.020 = 0.040 m³. A measured bulk density of 2,000 kg/m³ would give 80 kg for that condition. The values illustrate the calculation only.
An irregular substrate, varying thickness or density established in a different condition prevents this estimate being used directly as actual retained mass. For fieldwork, specify the volume-survey method and condition to which density refers: fresh, dry or after treatment. A particle-density value does not replace the product’s bulk density.
When comparing two trials, also preserve the areas actually included. A layer extending beyond the useful area can increase consumed mass without increasing the required function. The balance should expose this difference rather than assigning all consumption to a theoretical surface.
These numerical checks make the record easier to audit. They reveal conversion and boundary errors before consumption is presented as an advantage of the new fraction.
15. Compare repairs providing the same function
For an operational balance, retain the area actually treated, thickness or volume, intervention duration, retained quantity and rework. Data should come from measurements or an explicit record. Purchased mass alone does not describe repair effectiveness.
Compare cost for an equivalent function with known commissioning conditions. Material savings may be cancelled by rework or additional downtime; they should not be announced before examining the result.
Likewise, observed duration must be linked to loading and the end-of-service criterion. A repair removed at a scheduled shutdown has not necessarily reached the same limit as a degraded repair. Preserving this distinction makes comparison more useful than isolated durations.
16. Write the fraction request without promising a finished product
The raw-material supplier request should specify mineral identity, fractions, preparation and checks. Final mix characteristics remain in the formulator’s record. This separation prevents properties a sand certificate does not measure being attributed to it.
The silica-sand page provides the general material framework. For refractory use, selection must remain linked to a defined formulation and relevant trials.
Sound qualification therefore connects three things: the received fraction, the manufactured repair and the conditions actually represented. This continuity makes a material change usable rather than assuming every silica with the same composition is equivalent.