Choose a glass mould as a complete system
To choose mould material for kilnformed glass, connect the forming process, mould material, its condition before firing and the intended result. Granular sand intended to imprint a texture, silica flour used in an investment and the glass being shaped do not perform the same function. Purchasing by the name “silica” alone risks comparing products that are not interchangeable.
This guide addresses studios and project developers in Fez who want to understand these distinctions before testing. The Ministry of Industry describes the industrial activities of Fez-Meknes in its regional profiles 1. That context locates the discussion; it does not measure local sand consumption for artistic glass. The method below concerns mould-material qualification, rather than a supposedly validated local industrial recipe.
The practical objective is to connect four decisions: the shaping process, the mould material, its condition before firing and the result actually required. An attractive surface on a small specimen does not qualify a thick sculpture, a different glass or repeated production. The proposed procedure keeps these differences visible in the trial record.
1. Separate fusing, slumping and kilncasting
Fusing joins glass elements through a thermal treatment. In slumping, a softened sheet takes shape under gravity. In kilncasting, glass flows into a refractory cavity. Bullseye treats these operations separately and explains that their firing programmes are not automatically interchangeable 5.
Before looking for a grain size, draw the piece and identify the surfaces contacting the mould. State whether the mould will be destroyed during removal, reused or simply used as a textured support. An open shape and a closed cavity do not impose the same demoulding constraints. Success as a flat support does not establish suitability as an investment for a deep cavity.
The project sheet should also identify the glass, loaded mass, maximum thickness and intended finish. A deliberate granular texture should not be assessed using the criteria for an optically smooth face. This first clarification avoids rejecting a material for a feature that was deliberately sought, or accepting it for a finish it cannot produce.
2. Do not substitute ordinary sand for silica flour
The plaster-silica investment described by Bullseye consists of casting plaster and silica flour; its product page states that it is not intended for reusable moulds 2. This information concerns the particular product described. It does not imply that every quartz-containing material behaves alike.
For purchasing, separate mineral identity from preparation. “Silica” identifies a chemical family; “flour” indicates a finely prepared material; “sand” describes a coarser grain population. The particle size distribution, delivery condition and complete formulation therefore need to accompany the name. A high SiO₂ percentage demonstrates neither adequate fineness nor suitability for moulding.
If a studio wants to prepare a finer material from sand, that transformation needs its own qualification. Grinding, classification and contamination control can change useful properties. The starting material and prepared product should retain separate references. This makes it possible to distinguish a sand-lot issue from a preparation issue or a mould-mixture issue.
3. A sand bed is a separate route, not evidence of equivalence
Bullseye also presents kilnforming on a sand bed to transfer textures. The public materials list for that lesson includes olivine sand 3. It would therefore be incorrect to use that resource as evidence that any silica sand suits the same project, or that quartz automatically reproduces the illustrated result.
The qualification consequence is straightforward: record the exact reference material, not just the word “sand”. Changing mineral identity, grain distribution or preparation opens a new comparison. If the objective is an imprint, assess relief fidelity, grain adhesion, cleaning effort and stability of the resulting shape.
A textured bed can also support aesthetic exploration. To make that exploration useful, retain a labelled reference plate and photographs taken under the same lighting. A description such as “a more natural surface” is too subjective for lot selection. An organised assessment of texture, adhered particles and deformation helps the artist, kiln operator and buyer reach the same decision.
4. Define the surface before selecting the grains
Discuss particle size in relation to the final face. Visible texture, fine relief and a surface intended for polishing require different criteria. The manufacturer’s FAQ explains that mould texture influences casting texture 5. This relationship does not supply a universal millimetre fraction: it must be assessed for the system actually used.
For a material trial, prepare simple patterns incorporating a flat area, an angle, a groove and a curved zone. These reveal several defect types without committing a complete artwork. Keep dimensions, glass and photography methods the same across variants. Comparing textures produced on different geometries would be difficult to interpret.
Also define the acceptable amount of finishing after firing. A satisfactory imprint requiring lengthy removal of adhered particles may be less useful than a slightly less detailed reference. The criterion is therefore not merely appearance immediately after demoulding. It is appearance after the finishing operation actually planned, together with the associated work time.
5. Qualify the mixture, not just its mineral filler
Bullseye distinguishes binder, refractory and modifiers in casting investments 7. Organise trials around these components as a system. Replacing a filler may require another mixture check; keeping the same recipe name does not guarantee identical preparation.
A manufacturing sheet should record component references, condition, masses, added water and the operating sequence. Record the mixture’s age when used. The aim is not to publish a universal recipe, but to make the studio result reproducible. Without these records, a variation caused by preparation might wrongly be attributed to particle size.
When comparing two fillers, keep a reference formulation unchanged. If a variant simultaneously requires different water addition and a different firing programme, separate those changes in the trial plan. A good piece resulting from several modifications does not identify which change helped. Distinguishing material, mixture and firing enables progress without losing the starting reference.
6. Control water and preparation before firing
The manufacturer’s instructions highlight the water-to-investment ratio, initial saturation and vessel cleanliness when preparing its product 4. These instructions concern that particular mixture. Its proportions should not be transferred to another system with an unknown formulation.
A trial can nevertheless apply general discipline: use the same weighing method, prepare vessels comparably and record deviations. If two moulds use different water or tools carrying residues, record that distinction before comparing their surfaces. The trial needs to preserve what was done, rather than just a photograph of the finished piece.
Drying is a separate decision from apparent hardening. Bullseye’s TipSheet 8 stresses removing moisture from kilncasting moulds 7. A surface that looks dry does not measure the internal condition of a thick shape. The programme must come from the selected system’s instructions and account for geometry; a duration copied from a small piece does not qualify a more massive mould.
7. Design a trial that separates causes
Minerallum’s proposed procedure starts with a simple reference piece and one material variation. For every firing, record components, mould dimensions, glass load, kiln position and programme. Assign an identifier before the trial so that photographs, observations and remaining material can subsequently be connected.
The first series can explore differences. A second should check whether the selected variant reproduces its result. Including a reference in that repetition avoids confusing material improvement with a change in kiln operation or preparation. Trial numbers depend on the decision’s importance; they should not be chosen solely to accelerate approval.
Define stopping criteria as well. Mould failure, adhesion preventing demoulding or a geometry outside requirements are not merely aesthetic defects. Conversely, a small irregularity compatible with the requested finish need not trigger complete reformulation. Classifying observations before the series helps avoid contradictory judgements after firing.
8. Read defects without assigning a single cause
| Observation | Questions to document | Next decision |
|---|---|---|
| Rougher surface | Are material, pattern and finishing identical? | Compare geometrically equivalent references |
| Adhered particles | Which face is affected and how was it prepared? | Examine the contact system before trying another lot |
| Mould crack | At what stage was it observed? | Review manufacture, drying, support and programme |
| Glass deformation | Have glass, mass and firing changed? | Separate the material trial from the thermal trial |
| Inconsistent result | Were references and positions recorded? | Repeat with a reference and complete traceability |
This table is a proposed diagnostic tool, not an automatic relationship between defect and cause. A final photograph does not show when the problem appeared. Where possible, preserve observations from preparation, loading and demoulding. Blaming sand too quickly can lead to replacing a correct material while leaving the real variable unchanged.
9. Separate glass compatibility from mould quality
Bullseye explains that a coefficient of expansion alone cannot establish compatibility between two glasses 8. For mould trials, use a glass combination already defined for the project. Introducing another glass combination simultaneously obscures interpretation.
A piece can demould correctly while having a problem related to glass or its thermal cycle. Conversely, mould material may be unsuitable even when the glasses are compatible with each other. The two records should be linked while retaining separate criteria. One describes mould contact and integrity; the other describes the glass and treatment of the piece.
Before moving from a small reference to an artwork, include an intermediate specimen representing the difficult areas: thickness changes, deep relief or prolonged contact. This stage does not guarantee every future dimension. It checks whether the original choice remains appropriate as geometry begins to resemble the intended product.
10. Organise powders and cleaning
HSE publishes guidance on crystalline silica dust control for preparation, finishing and kiln-loading tasks in ceramics 6. These British resources document working situations; they are not presented here as Moroccan regulations.
Include post-firing operations in qualification. Demoulding, residue removal and surface finishing can create dust exposure as well as initial preparation. Extraction, wet methods appropriate to the task and cleaning should be organised with the workstation rather than added after purchasing the material. A powder choice therefore involves more than its result in the kiln.
Identify waste and residues by formulation reference. Bullseye’s instructions exclude disposal of its investment through drains 4. For the studio, routing waste appropriately for its composition avoids treating a complex mixture as clean, recoverable sand. Fired or contaminated material does not automatically regain the status of new material.
11. Write a usable material enquiry
The purchasing record should combine process, material function, required grain distribution, necessary preparation and acceptance criteria. State whether the request concerns sand for a textured bed, material requiring transformation or a formulated investment. This is more useful than a general request for “silica for glass”.
Include results for the selected variant and reference. Provide for a retained sample, lot reference and a rule for dealing with a material change. If a future lot has a different distribution or preparation, the record should allow a decision between a simple check and another firing trial.
The silica sand material guide is the entry point for defining the supply requirement. For this application, purchasing should follow clarification of the moulding system. A documented trial connects a precise material with a precise result. It turns initial success into a usable decision without claiming that every sand or every piece will behave identically.
12. Compare the cost of an acceptable piece
An economic comparison should separate material price, mould preparation, component consumption, demoulding time and finishing. Record rejects with their observed cause rather than automatically assigning every defect to the mineral material. In an exploratory trial, these items can remain observations of time and consumption; no invented monetary value is needed to obtain useful information.
A material with a lower price per kilogram may require more preparation or produce a surface incompatible with the intended use. Conversely, a more elaborate material may be unnecessary for a deliberately coarse texture. The comparison should therefore concern a piece meeting the same criteria, rather than simply a purchased quantity. This rule requires comparable finishing across the variants.
Before repeated production, also check how the material will be stored: closed packaging, readable references, separate residues and an available retained reference. If the trial material was prepared differently from future deliveries, record that distinction. Acceptance should cover what will actually be used, with its normal preparation, rather than only the exceptional sample that produced initial success.
The final record can consist of four documents: a material sheet, a mould-manufacturing sheet, a firing sheet and a result checklist. Each retains the same trial identifier. This organisation remains usable in a small studio and allows new materials to be introduced progressively without reconstructing the entire trial history from memory.