NEWS EVENTS
The Price of a Contact Point: How Support Strategy Decides the Surface You Cast
2026-11-23
In most process discussions of vat photopolymerization, supports are treated as a disposable by-product of the slicer: the build completes, the supports come off, the job is done. Put that assumption against a jewellery investment casting workflow and it fails. The small patch where a support tip meets the part surface — the contact point — is where the peel force enters the part during separation, where the material fractures during removal, and where the local UV dose sets how hard that patch actually is. Three different problems occupy the same few hundred micrometres.
This article draws on three journal papers and two conference papers published in 2024–2025. Every figure below is attributed to its source. Where an abstract reports a trend without a number, the trend is stated as a trend and no number is invented.
1. Three Definitions Before the Argument
An overhang is a surface that is inclined relative to the build direction and cannot rest on the previously cured layer. Practice draws the line at 45 degrees: downward-facing surfaces below that angle are treated as needing support. The 45-degree threshold is a process convention, not a material constant — it shifts with resin chemistry, layer thickness and exposure strategy.
The contact point is the junction between a support tip and the part surface. It carries the highest stress concentration anywhere in the support chain, and it collects the highest density of surface defects.
The supported area fraction is the projected area covered by contact points divided by the total projected area of the part. A lower fraction means less removal work and more process risk.
The Support-Free Factor (SFF) is the metric Artemeva et al. (2024) introduced to quantify that trade-off: part projected area divided by support contact area. On their open-architecture ultra-high-resolution vat photopolymerization system, using a commercial photopolymer resin, the minimum workable SFF was measured at 65.19 — equivalent to 1.51 percent of the part's projected area. Below that ratio the part would not stay attached to its supports during the build.
2. Support Count Is Set by Pull-Off Traction
In a bottom-up architecture, every completed layer requires the platform to lift and peel the cured layer away from the release interface. That motion generates a separation force whose peak scales with the cross-sectional area of the layer: the larger the cross-section, the larger the instantaneous pull. The job of the support structure is to route that force down to the build platform in stages, so the part is neither torn off nor distorted.
Under-supporting fails loudly and immediately. Over-supporting fails quietly: it shows up as removal labour, and as the surface degradation that comes with a higher count of contact points. Cao et al. (2025) put it plainly — too few supports cause build failure, too many require manual intervention and degrade surface quality.
The difficulty is that this is not a quantity anyone should tune by trial and error. Change the support layout and the peel-force distribution changes with it; an accurate assessment requires layer-by-layer simulation. Cao et al. cluster layers with coherent cross-sectional shape, identify the representative layer carrying the maximum support traction force, and run full simulation only on that layer, inferring the rest through geometric optimisation. Simulation iterations drop from the order of hundreds of thousands to a few dozen.
The reported experimental outcome: compared with commercial slicing software, the method substantially reduces the number of support contact points while keeping the structure stable throughout the build. The abstract gives no reduction percentage, so none is quoted here.
3. Where the Numbers Land
Artemeva et al. chose a worst case deliberately: a simple bar geometry positioned at 0 degrees to the surface, with support attached underneath. The arrangement strips out most variables, allows per-layer control of mask and process parameters, and magnifies whatever the support strategy does.
The study varied SFF value and process parameters, and observed the UV light intensity at the contact between support tip and part. Intensity at that point correlated with final surface quality. Z-axis dimensional error fell from 20 percent to 6.9 percent. The same experiments found that denser support structures, combined with removing the part from its supports before post-processing, produced finished dimensions closest to the CAD model.
Those two findings have to be read together. Denser support improves dimensional fidelity, at the cost of more contact points. Without the early-removal condition, the denser support hands that cost back in the form of surface marks.
The limits deserve equal emphasis: one geometry, one commercial resin, one machine. The 20-to-6.9 percent improvement does not transfer to another resin or another layer thickness. What transfers is the method — calibrate the process window with two measurable quantities, supported area fraction and UV dose at the contact point.
4. Two Ways Out: Clustered Tree Topologies and Dissolvable Supports
On the generation side, the goal is carrying the same load with fewer supports. The hierarchical clustering evolutionary tree support (HCETS) proposed by Cao et al. (2024) clusters support points hierarchically and handles two distinct classes of points in parallel, optimises the tree topology with an improved multi-objective decomposition algorithm, and applies KD-tree acceleration to inclined support structures to cut computational load. Two selection strategies — intra-group and environment — hold back the selection pressure and diversity loss that come with population expansion.
The conclusion: with print stability maintained, HCETS outperforms known algorithms on support generation speed and material waste, and produces low-volume tree structures. No speed multiplier or material-saving percentage appears in the abstract, so only the qualitative claim is reported here.
On the removal side, there is a different route: make the support dissolvable. Tang et al. (2025) developed a fast-dissolving, recyclable thiol-ene photopolymer for use as a support material in multi-material vat photopolymerization. The work examines how surface-area-to-volume ratio and grayscale projection affect dissolution rate and recyclability, and embeds porous gyroid architectures inside the support volume.
Two results stand out. Dissolution rates reached up to 50 times those of solid supports. After multiple recycling cycles the thiol-ene resin retained its photopolymerization performance and mechanical properties. The team used the route to build microfluidic channels, cactus-inspired spines and interlocked structures — geometries where mechanical support removal is impractical.
The precondition is multi-material capability: support and part must be two different resins, and the machine must switch between them within a layer. A single-material system cannot copy this directly. It should also be noted that the study does not address the burnout stage of investment casting; how support resin residue behaves at high temperature falls outside its conclusions.
5. From Wax Pattern to Casting: Where the Contact Mark Ends Up
The preceding sections stop at the part surface. Jewellery investment casting asks for one more step: the surface of the wax pattern — or the photopolymer pattern standing in for it — is replicated through the rubber mould and the plaster shell, and reappears in metal. The contact mark is not left on the prototype; it is transferred downstream.
Dakhli et al. (2025) document the full chain. They printed a single-piece investment casting pattern for an Inconel 718 jet fuel distributor on a masked stereolithography (MSLA) system, iterating the model design through numerical casting simulation before printing, treeing, shelling and pouring. The result was a porosity-free casting with a microporosity index below 0.54 percent and shrinkage deformation under 6.6×10⁻² mm; casting yield reached 69.17 percent and pattern yield 88.6 percent, the latter attributed to pre-processing calibration and refined slicing parameters.
Its boundaries are just as clear: a single aerospace part, comparatively thick metal sections, an MSLA system — not the thin walls and cosmetic-face demands of jewellery. What transfers is the sequence: simulate to fix the model, then fix the supports, then fix the removal method. Run it the other way and the contact point turns out to sit on a cosmetic face only at the removal stage.
Jewellery adds two constraints. First, the cosmetic face is the final face — there is no machining allowance to turn a mark away, so contact points must be placed on the sprue side, inside a setting recess, or under an area a stone will later cover. Second, pattern walls often sit in the sub-millimetre range, so a nick or micro-crack from mechanical support shearing becomes a stress raiser in the casting. The same peel force acting on a thinner section produces more deflection, which is why thin-walled work is more sensitive to peel force than heavy sections.
At the equipment level this maps onto layer thickness. The Totus UR and URC (Ultra-Rapid Continuous) series offer three layer thicknesses: 10, 25 and 50 μm. Thinner layers add transition layers between support tip and part, which makes the cure gradient at the contact point easier to control layer by layer — at the cost of build time. That is a statement about machine capability, not an experimental finding.
6. What the Evidence Covers, and What It Does Not
The sources fall into three tiers. Academic tier: all five studies are published in journals or conference proceedings, with stated experimental conditions and retrievable DOIs. Their shared limitation is narrow sampling — one worst-case bar specimen, two algorithm-led papers with limited experimental validation, one dependent on a dedicated multi-material resin, one single aerospace part. None of their numbers should be generalised.
Public technical literature tier: slicer documentation and industry white papers. This tier supplies engineering convention, such as the 45-degree overhang threshold. It ranks below peer-reviewed work and rarely discloses test conditions.
Product specification tier: published layer thickness options, build envelopes and similar data. This tier answers what a machine can do. It does not answer whether the result is good.
Working from those three tiers, the practical order of decisions is: identify the cosmetic faces first and exclude them as contact-point locations; set support density from peel force rather than geometric intuition, densifying where the cross-section changes abruptly; complete removal before post-processing so that post-cure does not harden a contact point into a permanent bump; for thin walls, prefer a dissolving or low-peel-force route over simply increasing shear force; and after any change of resin or layer thickness, recalibrate the exposure dose at the contact point.
The single claim worth keeping: the support problem is not about how many supports there are, but about where the contact points sit, how much UV they receive, and how they break away. All three are measurable, and all three can be specified at the slicing stage. Treat them as process parameters instead of software defaults, and surface variation starts to converge.
References
[1] Artemeva M, Lopez Navas J, Calaon M, Basso A, Nadimpalli V K, Bue Pedersen D. Towards Scarless Support Removal in Vat Photopolymerization Additive Manufacturing for Mass Manufacturing of Precision Components. Procedia CIRP, 2024, 130: 47–52. DOI: 10.1016/j.procir.2024.10.054. Indexed in EI, Scopus.
[2] Cao L, Yan X, Liu L, Liu J, Jiang T, Lu L. Auto-support: traction force constrained support structures optimization for bottom-up vat photopolymerization. Virtual and Physical Prototyping, 2025, 20(1). DOI: 10.1080/17452759.2025.2466189. Indexed in SCI, EI, Scopus, Web of Science.
[3] Cao X, Yu M, Zhang S, Zhang T, Chen Y, Wang Y, Han X. Hierarchical clustering evolutionary tree-support for SLA. Journal of Manufacturing Processes, 2024, 125: 189–201. DOI: 10.1016/j.jmapro.2024.07.056. Indexed in EI, Scopus, Web of Science.
[4] Tang T, Nawab P K, Prabhudesai P, Alfarhan S, Pesqueira I, VanWormer K, Li J, Li D, Jin K, Li X. Sustainable vat photopolymerization of complex functional structures with fast dissolvable and recyclable supports. Virtual and Physical Prototyping, 2025, 20(1). DOI: 10.1080/17452759.2025.2499452. Indexed in SCI, EI, Scopus, Web of Science.
[5] Dakhli C E, Ben Fathallah B, Soula M, Aouadi B. MSLA 3D Printing of Wax Pattern for Rapid Investment Casting Inconel 718-Superalloy Aeronautical Part. Advances in Additive Manufacturing: Materials, Processes, and Applications II (AIAM' 2024), 2025: 115–132. DOI: 10.1007/978-3-031-94889-3_10. Indexed in EI, Scopus, Web of Science.
Keywords: vat photopolymerization,support structure,contact point,support removal,surface quality,DLP,investment casting,wax pattern
Related News
F2 Building 6A Jintong National Industrial Park No.8 Xihu Road Wujin District Changzhou City Jiangsu Province China
Follow Us
Copyright © 2022 Jiangsu Totustec. All Rights Reserved. 苏ICP备17013684号-1 Powered by:www.300.cn SEO
Online Message
Message
cn@totustec.com
+86 18112582262