Every dimension on a drawing is a cost decision, most of them made before a foundry ever sees the file. Investment casting is uniquely forgiving of complex geometry — that’s why it wins over sand casting on detail and over fabrication on part consolidation — but it still obeys the physics of solidification, shell strength and metal flow. The difference between a casting that runs at a 2% scrap rate and one that runs at 18% is usually not clever tooling; it’s ten design decisions made in CAD during the first week. This guide walks through those ten rules as cost levers: what each one does to your piece price, your tooling bill, your lead time, and your in-service reliability.
The Challenges That Inflate Casting Cost
Challenge 1: Cost Is Locked Before the Quote
Roughly 70–80% of a component’s manufactured cost is committed at design release. A wall thickness you could have changed with one mouse click in week one costs a tooling modification, a new pattern, a re-sampled first article and eight weeks by week twenty.
Challenge 2: “Net Shape” Is Assumed Rather Than Designed
Investment casting routinely achieves net shape on some features. Assuming all features are net-shape leads to drawings with no machined datums, unsupported tolerances, and higher scrap when those unmachined features inevitably drift.
Challenge 3: Tolerances Are Copied From Machining Habits
A ±0.05 mm habit carried from a machining drawing onto a casting drawing multiplies inspection time and scrap. Casting processes have their own tolerance standards (ISO 8062-3 tolerance grades, commonly DCTG 4–6 for silica sol investment casting) and fighting them costs money.
Challenge 4: The Foundry Is Consulted Too Late
DFM feedback delivered after tooling release can only fix cosmetic issues. Every substantive change after that point is a change order.
The 10 DFM Rules
Rule 1: Keep Wall Thickness Uniform
Non-uniform sections are the number-one cause of shrinkage porosity, distortion and hot tears. Thin sections freeze first; thick sections freeze last and pull liquid metal from their neighbours, leaving voids.
Action: aim for walls within a 2:1 range on any single casting. Where sections must change, transition gradually — a 3:1 length-to-thickness taper is a safe rule of thumb.
| Wall Condition | Typical Result | Cost Impact |
|---|---|---|
| Uniform, ≤2:1 ratio | Predictable solidification, low scrap | Baseline |
| Localized mass 3× adjacent wall | Shrinkage cavity at the mass | +3–8% scrap, added radiography |
| Abrupt step change (no taper) | Hot tear at the junction | Scrap spikes, possible redesign |
| Thin section below ~1.5–2 mm | Misrun / cold lap | Process limit; may need alloy change |
Rule 2: Add Generous Fillet Radii
Sharp internal corners create stress concentration in service and turbulence during pouring and shell cracking during dewaxing and firing. External sharp edges are less problematic but fragile in handling.
Action: internal radii should be at least 0.3–0.5× wall thickness, and never less than roughly R1 mm for small parts. Use the largest radius the design tolerates — this one change frequently eliminates a recurring crack failure in the field.
Rule 3: Specify Tolerances by Process, Not by Habit
Every tolerance you tighten reduces yield. Every tolerance you specify as “cast + machine” implicitly accepts the casting as oversize — which is fine, provided the machined allowance is real.
Action: leave non-critical surfaces to general casting tolerance (ISO 8062-3 DCTG grade appropriate to size). Apply tight tolerances only to features that genuinely affect assembly or function, and mark those as machined features with clear datums. As a rough guide, tightening one feature one ISO tolerance grade can add 5–15% to that operation’s inspection cost alone.
Rule 4: Consolidate Parts
This is where investment casting returns the most dramatic savings. A weldment of six plates, three bosses and four gussets is six fixtures, twelve welds, three distortion-correction steps and a stress-relief cycle — or one casting.
Action: count your current assembly’s piece count, labour minutes and inspection operations. If the total exceeds roughly 15–20 operations, model it as a single casting and compare. Typical consolidation returns 30–50% total cost reduction even when the casting itself looks expensive per kilo.
Rule 5: Design Draft and Undercuts Deliberately
Investment casting handles undercuts far better than permanent tooling processes — but some require ceramic cores or collapsible wax tooling, both of which add cost and cycle time.
Action: where the part’s function permits, add 1–2° draft to surfaces parallel to the draw direction, or orient the part in consultation with your foundry so that critical faces machine cleanly. Internal passages that must be cored should be specified early; adding a ceramic core after tooling is released is significantly more expensive than designing it in from the start.
Rule 6: Avoid Large Flat Areas
Flat plates warp. Large flat surfaces also show shell marks and require heavier finish grinding.
Action: break any area larger than roughly 100 × 100 mm with ribs, steps, a shallow crown or decorative recession. Ribs should be about 0.6–0.8× the base wall thickness — thinner than the wall they sit on, so they don’t become new hot spots.
Rule 7: Locate the Parting Line and Gate Marks Where They Don’t Matter
Every casting has marks where the wax was injected, where the assembly met the tree, and where the gate was cut off. These are not defects; they’re process signatures.
Action: nominate specific zones on your drawing where gate remnants and parting-line flash are acceptable, usually non-sealing, non-cosmetic surfaces. Also specify your cosmetic requirements explicitly — a part that must be visually perfect on all sides costs considerably more than one with defined cosmetic and non-cosmetic zones.
Rule 8: Design Machining Datums and Allowances In
Machining allowance exists for a reason: it gives every subsequent operation a repeatable reference. A casting with no defined datum forces the CNC shop to improvise.
Action: identify three datum features that are (a) cast in a single tooling half where possible, (b) accessible to a cutter, and (c) stable. Provide 0.5–1.5 mm machining allowance depending on part size and wall stability, and dimension from those datums. Also avoid asking for machining on surfaces that sit over thick sections unless you accept that the cutter may open subsurface porosity — a real and frequently disputed failure mode.
Rule 9: Consider the Material Early — Not Just the Grade
The alloy influences minimum wall, achievable tolerance, machinability and heat treatment cost. Some grades pour thin and clean; others are hot-short or demand solution annealing.
Action: specify the grade and its purpose — corrosion, wear, elevated temperature, pressure containment. Your foundry may propose a near-equivalent that casts better and machines faster. Be cautious with leaded free-machining grades and certain high-sulphur variants; they complicate welding repair and are restricted under RoHS and ELV in many markets.
Rule 10: Plan Inspection Into the Geometry
Ask a simple question for every critical dimension: how will this be measured? If the answer is “with difficulty,” either the feature or the tolerance needs to change.
Action: provide accessible surfaces for CMM contact, allow clearance around features that need gauging, avoid dimensions taken across a parting line or between features formed in different tooling halves (they carry the worst natural variation), and state your measurement method expectations — CMM, functional gauge, or scan — before the first article.
Putting DFM Into Your RFQ
DFM works best as a conversation, not a checklist. The most effective sourcing teams send a STEP file plus a marked-up drawing showing critical-to-function features and explicitly invite comment. Ask your supplier to return three things: proposed gating and parting-line arrangement, recommended tolerance relaxation on non-critical features, and any suggested material alternative. A supplier who returns silence has likely designed nothing; a supplier who returns ten questions has probably read your drawing properly.
Frequently Asked Questions
Q1. What is the realistic minimum wall thickness for investment casting? For silica sol investment casting, walls down to roughly 1.5–2 mm are achievable in favourable alloys and small parts, with 2–3 mm being comfortable for general production. Very thin walls depend heavily on section length, flow path and alloy fluidity, so validate early with a prototype pour rather than assuming a catalogue figure applies to your geometry.
Q2. How much does part consolidation really save? Savings vary, but consolidating a welded or machined assembly into one investment casting commonly reduces total cost by 30–50%. The drivers are fewer piece numbers, eliminated fixtures and welding, fewer inspection operations, and reduced inventory handling. The per-kilogram cost of a casting almost always looks higher than raw plate — the system cost is what falls.
Q3. Can investment casting hold ±0.1 mm on every feature? Achievable on small features in favourable orientations, but unnecessary on most surfaces and expensive to verify everywhere. Reserve tight tolerances for features that control assembly or function, hold everything else to an ISO 8062-3 tolerance grade appropriate to the casting size, and machine the features that truly need precision.
Q4. When should I bring my foundry into the design? At concept stage if possible, and no later than the point where geometry is frozen. DFM input received before tooling release can change walls, radii, datums and even part splits at zero cost. After tooling release, the same changes become change orders with both cost and schedule attached.
Q5. Do I need a prototype or can I go straight to production tooling? Prototypes are strongly recommended for complex geometry, thin walls or consolidated parts converting an existing assembly. The cost of one prototype cycle is modest compared with discovering a shrinkage problem in production tooling, where correction means welding or remaking a die.