- \n
- The printed route deletes two steps from the schedule \u2014 pattern making and core box making \u2014 and those two steps account for most of the lead-time difference.<\/li>\n\n\n\n
- Tooling cost is not reduced, it is eliminated: a printed mold is a digital file, reprinted on demand.<\/li>\n\n\n\n
- Design revisions cost hours instead of weeks, which changes the crossover quantity far more than the arithmetic suggests.<\/li>\n\n\n\n
- Printed molds are not stronger than pattern-made molds \u2014 they are more precise and more repeatable, which is a different advantage.<\/li>\n\n\n\n
- Melting, pouring, shakeout and heat treatment are unchanged. The printed route does not make a foundry faster at being a foundry.<\/li>\n<\/ul>\n\n\n\n
What each process actually does<\/h2>\n\n\n\n
The two routes diverge at exactly one point: how the cavity gets made. Everything after the mold is built is identical \u2014 the same alloys, the same melting practice, the same pouring, the same inspection standards.<\/p>\n\n\n\n
Traditional: pattern \u2192 core box \u2192 mold \u2192 pour<\/h3>\n\n\n\n
A pattern is machined or hand-built to the shape of the casting, then mounted on a plate. A core box is built for every internal passage. Sand is packed around the pattern, the pattern is withdrawn, cores are set, and the mold closes. Each step needs its own tooling, its own lead time and its own storage.<\/p>\n\n\n\n
That tooling is an asset with a life. Patterns wear, warp and need repair; a worn pattern produces drift that shows up in the casting and eventually forces a replacement. Over a long production run the tooling is cheap per part, but it is never free and never zero-maintenance.<\/p>\n\n\n\n
Printed: CAD \u2192 printed mold and core \u2192 pour<\/h3>\n\n\n\n
A sand printer lays down sand and binder, layer by layer, following the digital mold. Side walls, sprue, runners, cores and every internal passage are built in one piece. There is no pattern to withdraw, so there is no draft angle, no parting line and no core assembly.<\/p>\n\n\n\n
![\"Sand](\"https:\/\/www.lyafs.com\/wp-content\/uploads\/2026\/03\/3D-Sand-Printing-Process.jpg\")
From CAD to a casting-ready mold: the printed route skips pattern making and core box making entirely.<\/figcaption><\/figure>\n\n\n\n Step<\/th> Traditional pattern casting<\/th> Printed sand mold<\/th><\/tr><\/thead> Pattern making<\/td> Required \u2014 machined or hand-built<\/td> None \u2014 the model is the pattern<\/td><\/tr> Core box<\/td> One per internal passage<\/td> None \u2014 cores print with the mold<\/td><\/tr> Draft angle<\/td> Required to withdraw the pattern<\/td> Not required<\/td><\/tr> Parting line<\/td> Required<\/td> Not required<\/td><\/tr> Core setting<\/td> Manual, with chaplets and core prints<\/td> Built in, no assembly<\/td><\/tr> Storage<\/td> Physical pattern store, wear and repair<\/td> Digital files only<\/td><\/tr> Revision<\/td> Modify or remake the pattern<\/td> Edit the file, reprint the same day<\/td><\/tr><\/tbody><\/table><\/figure>\n\n\n\n Cycle time, step by step<\/h2>\n\n\n\n
Cycle time is where the two routes differ most, and where published comparisons are usually least useful \u2014 because they quote the printing time and omit everything around it.<\/p>\n\n\n\n
Step<\/th> Traditional<\/th> Printed<\/th> What changes<\/th><\/tr><\/thead> Tooling<\/td> 5\u201315 days for a pattern, more for core boxes<\/td> None<\/td> This is the whole gap<\/td><\/tr> Mold and core making<\/td> 1\u20133 days<\/td> Printing runs parallel to planning<\/td> Overlaps instead of following<\/td><\/tr> Printing the mold<\/td> \u2014<\/td> Hours to 1 day for a 1,800 \u00d7 1,000 \u00d7 700 mm mold<\/td> Layer time, not geometry complexity<\/td><\/tr> Cure and depowder<\/td> \u2014<\/td> Several hours<\/td> Real, and usually omitted from comparisons<\/td><\/tr> Pour and shakeout<\/td> Same<\/td> Same<\/td> Unchanged<\/td><\/tr> Cut, heat treat, machine<\/td> Same<\/td> Same<\/td> Unchanged<\/td><\/tr> CAD to first mold<\/strong><\/td> 2\u20138 weeks<\/strong><\/td> Hours to 1 day<\/strong><\/td> <\/td><\/tr><\/tbody><\/table><\/figure>\n\n\n\n The 2\u20138 week range on the traditional side is dominated by tooling availability, not by mold making. A busy pattern shop is often the real constraint.<\/p>\n\n\n\n
Where the time actually goes<\/h3>\n\n\n\n
In the traditional route, most of the calendar time is spent waiting for tooling that is not being made yet \u2014 queued at a pattern shop, waiting on a material delivery, or being reworked after a first-article check. The mold itself is quick once the pattern exists.<\/p>\n\n\n\n
In the printed route, that queue disappears. The model goes to the printer, and the printer either has capacity or does not. Capacity is a scheduling question, not a tooling question.<\/p>\n\n\n\n
What still takes time in the printed route<\/h3>\n\n\n\n
It would be misleading to present printing as instant. Four things still take real time:<\/p>\n\n\n\n
- \n
- Printing itself.<\/strong> Layer-based, so a tall mold is slower than a short one at the same volume. A 1,800 \u00d7 1,000 \u00d7 700 mm build runs at 18\u201323 seconds per layer, with a maximum throughput around 153 L\/h.<\/li>\n\n\n\n
- Cure and depowder.<\/strong> The printed mold must be cured and cleaned out before it can be poured.<\/li>\n\n\n\n
- Melting and pouring schedule.<\/strong> A foundry runs heats on a schedule; a mold that is ready on Tuesday may still be poured on Thursday.<\/li>\n\n\n\n
- Heat treatment and machining.<\/strong> Unchanged from the traditional route, and often the longest remaining step.<\/li>\n<\/ul>\n\n\n\n
Cost comparison by quantity<\/h2>\n\n\n\n
The cost question has two halves, and they point in opposite directions: printed molds win decisively at low quantity and lose gradually as quantity rises.<\/p>\n\n\n\n
Quantity of identical molds<\/th> Traditional<\/th> Printed<\/th> Better route<\/th><\/tr><\/thead> 1<\/td> Tooling US$5,000\u2013150,000+ plus mold cost<\/td> No tooling, mold cost only<\/td> Printed, by a wide margin<\/td><\/tr> 10<\/td> Tooling amortised over 10<\/td> No tooling, ten mold builds<\/td> Printed in most cases<\/td><\/tr> 50<\/td> Tooling amortised over 50<\/td> No tooling<\/td> Close \u2014 depends on part size<\/td><\/tr> 200<\/td> Tooling largely amortised<\/td> No tooling, 200 builds<\/td> Traditional<\/td><\/tr> 1,000+<\/td> Tooling fully amortised<\/td> Mold production becomes the cost<\/td> Traditional<\/td><\/tr><\/tbody><\/table><\/figure>\n\n\n\n Why the crossover sits where it does<\/h3>\n\n\n\n
The crossover is set by the ratio of tooling cost to the per-mold cost difference. For a large pattern set at the top of the range \u2014 a complex pattern plus several core boxes \u2014 the breakeven can sit well past 200 molds. For a small, simple part with an inexpensive pattern, it can sit below 50.<\/p>\n\n\n\n
Two variables move it further than arithmetic does:<\/p>\n\n\n\n
- \n
- Number of core boxes.<\/strong> Every internal passage is a separate core box in the traditional route, and each one adds tooling cost without adding casting value. A part with five internal passages has a very different crossover from a solid block.<\/li>\n\n\n\n
- Probability of design change.<\/strong> A modified pattern costs almost as much as a new one. A modified CAD file costs nothing. If the design has a real chance of changing, the printed route stays economic far beyond the arithmetic breakeven.<\/li>\n<\/ul>\n\n\n\n
Hidden costs on the traditional side<\/h3>\n\n\n\n
Pattern storage, transport between pattern shop and foundry, pattern repair, and the scrap from a worn pattern are all real costs that rarely appear in a per-mold quote. They are small per unit and large per year, and they vanish entirely when the mold is a file.<\/p>\n\n\n\n
Tolerances and surface finish<\/h2>\n\n\n\n
Measure<\/th> Traditional pattern<\/th> Printed mold<\/th><\/tr><\/thead> Dimensional accuracy<\/td> Varies with pattern wear and ramming<\/td> \u00b10.3 mm up to 300 mm; \u00b10.5 mm above<\/td><\/tr> Repeatability<\/td> Manual \u2014 drifts over a run<\/td> Automated \u2014 same file, same mold<\/td><\/tr> Minimum wall thickness<\/td> About 4 mm, plus draft<\/td> 2\u20133 mm<\/td><\/tr> Draft angle<\/td> Required<\/td> None<\/td><\/tr> Surface as cast<\/td> Depends on pattern finish and sand<\/td> Depends on layer thickness, 0.2\u20130.5 mm<\/td><\/tr> Casting tolerance standard<\/td> ISO 8062 DCTG 5\u20136 typical<\/td> ISO 8062 DCTG 5\u20136 achievable<\/td><\/tr><\/tbody><\/table><\/figure>\n\n\n\n Where printed molds are still behind<\/h3>\n\n\n\n
Layer lines are visible on the mold surface, and they transfer to the casting as a fine texture. For most industrial castings that is irrelevant; for a cosmetic surface it means the mold needs coating or the casting needs more machining allowance. The traditional route, with a polished pattern, produces a smoother as-cast surface.<\/p>\n\n\n\n
Where printed molds are already ahead<\/h3>\n\n\n\n
Repeatability. A printed mold is the same on the first build and the hundredth, because nothing physical wears. In a pattern-based foundry, the hundredth mold is made from a pattern that has been handled, cleaned and re-clamped a hundred times \u2014 and that variation shows up in the casting.<\/p>\n\n\n\n
Design freedom you only get with printing<\/h2>\n\n\n\n
Some geometries cannot be made with a pattern at all, at any cost. That is a different category of advantage from “cheaper” or “faster”.<\/p>\n\n\n\n
Conformal cooling channels<\/h3>\n\n\n\n
Channels that follow the shape of the casting, rather than being drilled in straight lines, are impossible with a rigid pattern and a core box. Printed cores carry them as part of the mold.<\/p>\n\n\n\n
Lattice and hollow cores<\/h3>\n\n\n\n
Internal lattice structures that reduce weight while maintaining stiffness can be printed into the core. A conventional core box cannot form them and cannot be withdrawn from them.<\/p>\n\n\n\n
Consolidated cores \u2014 no assembly, no core shift<\/h3>\n\n\n\n
A part with several internal passages normally needs several cores, set by hand, each with its own print and its own positioning tolerance. Core shift is a real defect mode. When the whole core set prints as one piece with the mold, the shift cannot happen.<\/p>\n\n\n\n
![\"3D](\"https:\/\/www.lyafs.com\/wp-content\/uploads\/2026\/05\/3D-Printed-Engine-Block-sand-casting.jpg\")
A printed mold and core set for an engine block \u2014 the internal passages print as one piece, so there is no core assembly and no core shift.<\/figcaption><\/figure>\n\n\n\n Which machine fits which job<\/h2>\n\n\n\n
![\"3DPTEK](\"https:\/\/www.lyafs.com\/wp-content\/uploads\/2026\/08\/Binder-jetting-sand-3D-printer-3DPTEK-J1800.jpg\")
3DPTEK-J1800 \u2014 1,800 \u00d7 1,000 \u00d7 700 mm build volume, 8 printheads, up to 153 L\/h of mold volume.<\/figcaption><\/figure>\n\n\n\n Two machines cover the entry and mid range of the printed route. Both use the same 1,800 \u00d7 1,000 \u00d7 700 mm build volume; they differ in printhead count, throughput and the footprint they need.<\/p>\n\n\n\n
- Probability of design change.<\/strong> A modified pattern costs almost as much as a new one. A modified CAD file costs nothing. If the design has a real chance of changing, the printed route stays economic far beyond the arithmetic breakeven.<\/li>\n<\/ul>\n\n\n\n
- Cure and depowder.<\/strong> The printed mold must be cured and cleaned out before it can be poured.<\/li>\n\n\n\n
- Printing itself.<\/strong> Layer-based, so a tall mold is slower than a short one at the same volume. A 1,800 \u00d7 1,000 \u00d7 700 mm build runs at 18\u201323 seconds per layer, with a maximum throughput around 153 L\/h.<\/li>\n\n\n\n