Moulage à mousse perdue : plages de conditions de fabrication, coûts et avantages par rapport au moulage au sable

Ce document s'adresse aux personnes chargées de spécifier les pièces, et non aux modeleurs. Chaque tableau ci-dessous constitue une référence pratique plutôt qu'un résultat de laboratoire : densité du modèle en grammes par litre, épaisseur du revêtement en millimètres, temps de remplissage en secondes, coût en dollars américains. La terminologie relative aux procédés suit les pratiques publiées par le Société américaine de la fonderie; les données relatives aux alliages proviennent des registres tenus par le Association de l'aluminium. Avant de comparer les devis, il est utile de voir où se situe le moulage par mousse perdue par rapport aux autres procédés — commencez par consulter notre Guide sur la fonderie des métaux, puis revenez pour les chiffres.

Comment fonctionne le moulage à mousse perdue, étape par étape

Le principe physique du moulage à mousse perdue est simple : on enfouit une réplique de la pièce en polystyrène expansé dans du sable sec et non aggloméré, on y coule du métal, et celui-ci remplace la mousse au fur et à mesure que celle-ci se vaporise. Ce qui change d'un point de vue commercial, c'est tout ce qui disparaît de votre plan : les empreintes de noyaux, les boîtes à noyaux, les bavures au niveau de la ligne de joint et la majeure partie de la pente que vous aviez prévue pour extraire le modèle du sable.

• Développez et faites vieillir le motif. Les billes d’EPS sont soufflées à la vapeur dans un moule en aluminium pour former grossièrement 16 à 24 g/L, puis laissez-les vieillir afin que les perles finissent de gonfler et se stabilisent. Découpez la pièce en autant de morceaux que vous le souhaitez et collez-les pour former un ensemble. L'outillage nécessaire à la réalisation de cet ensemble représente le coût initial le plus important — consultez notre détail des modèles de moulage et outillage de moulage.
• Collez le système de roulement. La tige de coulée, les canaux de coulée et les points d'injection sont également en mousse, collés à chaud sur l'ensemble. Le principe d'écoulement est le même que pour tout autre procédé de moulage : dimensionnement du col de coulée, règles relatives aux élévateurs et à l'alimentation cela reste vrai — mais l'ensemble du réseau est sollicité à chaque prise.
• Appliquer le produit et laisser sécher. Une ou deux immersions dans un revêtement réfractaire à base d'eau permettent d'obtenir un film sec total d'environ 0,3 à 0,8 mm. Cette couche remplit deux fonctions contradictoires : elle permet aux gaz de pyrolyse de s'échapper et empêche le sable de s'effondrer dans le métal. C'est là que la plupart des problèmes de procédé trouvent leur origine.
• Tassez dans du sable sec. Du sable de silice propre, dont la finesse granulométrique AFS est généralement comprise entre 40 et 70, sans liant ni eau, tamisé par vibration jusqu’à obtenir une densité apparente stable proche de 1,5 g/cm³. L'absence d'eau élimine les problèmes liés à l'humidité ; l'absence de liant permet de tamiser le sable et de le réutiliser presque indéfiniment.
• Versez à une température plus élevée que vous ne le pensez. La paroi métallique utilise sa propre chaleur pour décomposer la mousse ; c'est pourquoi l'aluminium introduit dans le sable vert à une température comprise entre 710 et 750 °C est généralement coulé à 760–790 °C ici.
• Secouer et terminer. Le sable s'écoule tout simplement ; il n'y a pas de noyaux à retirer ni de bavures au niveau de la ligne de joint à poncer. C'est là que réside l'essentiel des économies réalisées par pièce.

Le chiffre qui explique toutes les règles ci-dessous

Prenons un exemple concret : un A356-T6 corps de vanne en aluminium d'un poids de pièce moulée de 6,5 kg. La densité de la pièce moulée est de 6,5 ÷ 2 700 kg/m³ ≈ 2,4 L ; avec le système de commande, l'ensemble occupe un volume d'environ 2,9 L de mousse, qui, à 20 g/L, ne pèse que 58 g.

Ces 58 grammes représentent 0,91 TP3T de la masse métallique — or, si les produits de pyrolyse se comportent à peu près comme le monomère de styrène (masse molaire de 104 g/mol), cela correspond à 58 ÷ 104 ≈ 0,56 mole de gaz, qui occupe environ 13,6 L à température ambiante et environ 46 L at 1000 K. Spread over a 12-second fill through roughly 0.28 m² of coated surface, that is a mean superficial gas velocity of about 1.4 cm per second that the coating has to vent, every single shot, without letting the sand move.

Everything you specify about lost foam casting follows from that number. The coating must be permeable enough to pass that gas; the sand must be dense enough not to move while it happens; the metal must arrive hot enough to keep advancing while losing heat to the reaction. Lower pattern density, a thinner coat, a coarser flux target, and vacuum assist are all just ways of managing those 46 litres.

Process Windows That Control Lost Foam Casting Quality

These are the parameters buyers should see on a lost foam casting supplier’s process sheet before approving first article. Ask for the numbers in writing; if a shop cannot produce this table, expect dimensional drift between batches.

Two items deserve a caveat. First, the slower cooling of a dry silica flask produces a coarser secondary dendrite arm spacing than a permanent mould does, so T6 response is not automatically the same as for the same alloy cast elsewhere — plan your casting heat treatment cycle against the actual section, not the datasheet. Second, because there is no core box and no parting line, most of the dimensional scatter you are used to paying for simply disappears; the remaining tolerance sensitivity sits in pattern consistency and coating thickness, which our ISO 8062 tolerance guide covers in detail.

Lost Foam Casting vs Green Sand vs Investment Casting

Pick the process from geometry and volume first, then from price. The comparison below assumes a 0.5–10 kg part with internal galleries — the class where this decision actually gets made — read it as a decision filter, not a ranking.

A note on the two processes on either side of this table. Below about 1,000 pieces the foam tooling rarely pays back and green sand stays cheaper, which is why most low-volume complex parts we quote go through conventional moulage au sable or, when the surface demands it, fonderie à cire perdue. Above the low tens of thousands, dies win. Our earlier walk-through of permanent mould casting put its break-even against sand near 1,200 pieces for a comparable part, and against high-pressure die casting near 3,000 — so the useful lost foam casting band is real, but it is a band, not the whole range.

Where the Money Is: Tooling, Break-Even and Piece Cost

Here is a worked lost foam casting example. All figures are US dollars, quoted against a green-sand baseline on the same part — the pump housing above, run in aluminium with three internal cores today.

Break-even is then straightforward: $6,100 ÷ $6.10 = 1,000 pieces. Under a thousand units a year you are paying for tooling you will not recover; at 5,000 units the same part is about $24,400 better off, mostly because three cores and their cleaning simply stopped existing.

Two things are easy to miss when you build your own version of this lost foam casting table. Cluster pattern tools are single-cavity: if you need a rate above roughly 150–250 shots per shift, budget for a second tool rather than expecting the first one to run faster — the cycle is dominated by bead blowing and cluster gluing, not by the casting machine. And the piece saving is only real if your drawing actually exploits the process; converting a part that needs one core will not produce a $6.10 delta. If you want the deeper logic on how these drivers interact, our shop-floor cost breakdown walks through each line item, and the machining side is covered in our CNC machining services.

Defects You Only Meet With Lost Foam Casting

Shrinkage, dross and general misruns behave much as they do elsewhere. The table below is the shortlist that belongs specifically to lost foam casting — names your supplier should be able to say out loud — if they do not use these words, they probably are not controlling them.

Verification is worth specifying explicitly, because several of these only show up after sectioning or X-ray. Fold lines and carbon films are usually invisible on the skin, so put the acceptance method on the drawing rather than relying on a visual check — the options and their limits are set out in our casting inspection methods guide, alongside the routine défauts de moulage that still apply here.

When a Buyer Should Specify This Process

Lost foam casting earns its keep when several of the following are true at the same time. One or two is not enough to justify the tooling.

• You are paying for three or more cores today. Every core is a core box, a print, a setting operation and a fin to grind off. This is the strongest single signal, and it is why cylinder heads, manifolds and hydraulic blocks are the classic applications.
• Internal galleries are curved or multi-planar. Straight drilled cores are cheap to make with other routes; a maze is not.
• Annual volume is above roughly 1,000 pieces. Below that the tooling delta dominates, and below about 200 it rarely gets quoted at all.
• You can hold walls at 4 mm and above for aluminium, 5 mm and above for grey iron.
• You have six to ten weeks of tooling lead available. The cluster tool, not the casting, is on the critical path.

The counter-cases matter too, and they are why you should not treat lost foam casting as a default upgrade. If your part lives or dies on pressure tightness without impregnation, the coarser structure and slower cooling can work against you. If the drawing is dominated by thin fins under 3 mm, look at fonderie à cire perdue instead. And if the whole geometry is a thick-section load path, ask whether casting versus forging is even the right question.

In practice the parts that convert best cluster in a few applications: pump and pièces moulées pour vannes with cross-drilled galleries, pièces moulées pour l'automobile such as intake manifolds and brackets that consolidate several pieces into one, and aluminium castings where machining stock is currently generous because the location datums move between batches. Iron casting runs well here too — but check the next answer before you commit a ferrous part to this route.

FAQ

Does lost foam casting really need no draft?

You still need 0.5–1° to eject the foam cluster from its aluminium tool, but you do not need the 2–3° normally required to draw a pattern out of sand. On a 60 mm deep boss that difference is roughly 1–1.5 mm less taper per face, which often removes a machining step.

Why does it have to be poured hotter?

The advancing front spends part of its superheat decomposing the foam in front of it. If you pour at the conventional green-sand temperature, the front stalls when it reaches a thin rib, and what you get is a fold line rather than a clean fill — hence the 760–790 °C band for A356 instead of 710–750 °C.

Can it run grey iron and steel?

Grey and ductile iron run routinely through lost foam casting. Low-carbon steels are harder: the casting skin can pick up carbon from the pyrolysis residue, so if your surface has to stay at specification you need roughly 1.5–2 mm of machining allowance or a different route. Raise that question before quoting, not after first article.

What tolerances should I expect?

Because there is no parting line and no core shift, the same part usually lands one ISO 8062 grade tighter than it did in green sand. From there the remaining variation is pattern-to-pattern consistency plus coating thickness. Put the numbers in the drawing and cross-check them against our tolerance reference rather than assuming a grade.

Is it more expensive than green sand?

Tooling yes, per piece usually no — once you are past the break-even illustrated above. The honest way to answer it for your own part is the tooling delta divided by the per-piece saving; anything above a volume you can actually commit to within about 18 months is a gamble.

How do I start if I only need twenty parts?

You do not need the production tool. Clusters for small runs are cut and glued from foam stock, which removes most of the tooling cost and lets you validate the design before committing. That is how to convert an existing machined or fabricated part into a casting with the risk kept low.

*Process windows referenced from AFS and Aluminum Association publications, combined with Supro MFG production records; cost figures are illustrative ranges and must be re-quoted against your own drawing.