3D Concrete Printing Process Control: What 56,000 Layers Reveal

3D Concrete Printing Process Control: What 56,000 Layers Reveal

And what they don’t.

The RILEM TC 304-ADC inter-laboratory study (ILS-mech), led by researchers at Eindhoven University of Technology, is the largest open dataset connecting commercial 3DCP mix properties to layer-level process parameters. Thirty laboratories worldwide printed and tested specimens from a single commercial mix — Weber 3D 145-2 — generating over 56,000 layer-level records covering geometry, surface roughness, and process parameters, with 4,096 matched compressive strength measurements at the specimen level.

The purpose of the ILS was to characterize the reproducibility of extrusion-based 3D concrete printing across different labs and equipment setups — not to rank participants. We analyzed the full dataset to extract what matters for 3D concrete printing process control and production QC. The results are illuminating — but not always in the way you might expect.

Finding #1: Layer Interval Time in 3D Concrete Printing

Of all the process parameters tracked in the dataset, layer interval time shows the strongest positive correlation with compressive strength: r = +0.408, with a slope of +0.015 MPa per second of added interval.

A printer running 120-second layer intervals produces measurably stronger parts than one running 60-second intervals — same material, same age, same test protocol.

While this is a clear correlation within the dataset, we should not draw general conclusions to all 3D-printable cements. This trend likely reflects a material designed for printing large structures with 10–15 minute inter-layer times, compared to the sub-6-minute layer times tested in this study. The slightly longer intervals within this narrow range may allow marginally better moisture retention at the layer surface, improving interlayer consolidation.

The Weber 3D 145-2 has a 120-minute open time and is specified for optimal cycle times under 10 minutes, with cold-joint risk beginning beyond 45 minutes. The entire ILS dataset sits between 55 and 151 seconds — all within the optimal window, and nowhere near the 5–15 minute layer times typical for residential-scale prints.

What this means for practitioners: The data confirms this mix performs well across short layer intervals. But if you are printing structures with 10+ minute layer times, this dataset does not cover your operating window. Do not assume the positive trend continues beyond the tested range.

Finding #2: Reproducibility Across 3D Printing Labs

The single largest source of variation in compressive strength is not a process parameter — it is which lab printed the specimen.

Across 11 laboratories using identical material and test protocols, average compressive strength ranged from 42.1 MPa to 49.0 MPa: a 7 MPa reproducibility range. All results fall within the manufacturer’s stated C35/45 strength class, but the spread between labs is larger than the effect of any individual process parameter in the dataset.

This is a natural outcome of an inter-laboratory study — it reflects the cumulative effect of different pump systems, nozzle geometries, mixing procedures, ambient conditions, and operator practices across 30 independent institutions. The ILS was specifically designed to quantify this variability, and the 7 MPa range provides a valuable benchmark for what to expect when deploying the same material across different equipment setups.

What this means for practitioners: Before optimizing layer time or print speed, make sure your equipment is calibrated and your procedures are repeatable. Consistent mixing, pump calibration, and nozzle maintenance are the foundation of reliable 3D concrete printing process control.

Finding #3: Static Mixers and Air Void Formation in 3D Concrete Printing

A subset of laboratories used inline static mixers in their pumping setup. These mixers — typically used in two-component (2K) systems to dose accelerators at the printhead — rely on flow-division and radial mixing across internal helical elements.

The results in this dataset show a measurable effect:

• Density: 2,088 kg/m³ with static mixer vs 2,139 kg/m³ without — a 51 kg/m³ drop
• Compressive strength: 42.0 MPa with static mixer vs 46.5 MPa without — a 4.6 MPa reduction

The density reduction is consistent with entrapped air. Unlike cast concrete, 3D concrete printing has no post-extrusion vibration to consolidate the matrix. The folding action of the mixer elements can permanently trap air in the deposited filament.

What the voids look like under the microscope

Cross-section analysis via optical petrography and X-ray micro-CT in related 3DCP studies reveals characteristic patterns associated with mechanical air entrainment:

• Void size shift: A well-mixed batch produces fine entrained air (10–100 μm). Specimens extruded through static mixers can exhibit a bimodal distribution with macro-voids from 100 μm up to 1–3 mm.
• Irregular morphology: Unlike the spherical voids from chemical air-entraining agents, mechanically entrapped voids tend to be irregular and elongated — the fingerprint of shear-induced entrainment.
• Non-uniform distribution: Macro-voids can concentrate along the shear planes of the mixer elements and migrate toward inter-layer boundaries during extrusion — exactly where structural integrity matters most.

A 51 kg/m³ density drop in a 2,200 kg/m³ material corresponds to roughly 2.3% additional air by volume. The concern is not just total air content, but where that air concentrates.

It is important to note that this observation is specific to the particular static mixer designs used by the participating labs in this study. Static mixers vary widely in geometry and performance. For operators running 2K systems with inline accelerator dosing, this variable should be characterized for your specific equipment and material combination.

What this means for practitioners: Monitor density at the nozzle exit. A drop of 50+ kg/m³ from your target density is worth investigating. Cross-section a specimen periodically — even a simple saw-cut and visual inspection will reveal large irregular voids. If they concentrate at layer interfaces, your mixing setup may be worth revisiting.

Finding #4: What Actually Drives 3D Concrete Printing Strength

When we rank all measurable factors by their observed impact on compressive strength within this dataset, the picture becomes clear:

1. Lab equipment & procedures: 7.0 MPa range — The largest source of variation
2. Static mixer presence: 4.6 MPa — Equipment-specific effect
3. Print speed variation: 2.1 MPa — Moderate, controllable
4. Layer interval time: 1.8 MPa — Small effect within the tested range
5. Humidity (24–54%): 0.8 MPa — Negligible in this dataset

The first two items are equipment-related, not process parameters. For production QC, equipment maintenance and mixing consistency matter more than parameter tuning.

Key Takeaways for 3D Concrete Printing Process Control

The RILEM TC 304-ADC ILS-mech dataset is an extraordinary resource — the only published dataset linking a commercial 3DCP mix to layer-level process data across 30 labs worldwide. We are grateful to the TC 304-ADC committee and all participating laboratories for making this data openly available. Our analysis represents one interpretation of their work.

1. Calibrate your equipment first. A 7 MPa reproducibility range from lab procedures alone shows where the biggest gains are.
2. Monitor density, not just strength. A density drop is the earliest indicator of air entrainment, mixing problems, or equipment issues.
3. Characterize your inline mixing equipment. If you use static mixers in a 2K system, verify their effect on your specific material and setup.
4. Don’t extrapolate short layer times to long ones. The 55–151 second range in this dataset does not represent residential-scale printing conditions.

CEMFORGE incorporates these process-property relationships into its digital twin predictions. See how process parameters affect your mix predictions →

---

Data source: RILEM TC 304-ADC ILS-mech dataset v1.1.0 (CC BY 4.0), Eindhoven University of Technology. Weber 3D 145-2 commercial mix, 30 laboratories worldwide.

Disclaimer: This analysis was performed by Sunnyday Technologies and represents our independent interpretation of the publicly available dataset. It is not endorsed by or affiliated with RILEM TC 304-ADC, Eindhoven University of Technology, or Weber/Saint-Gobain. All compressive strength values are specimen-level results (prisms) and fall within the manufacturer’s stated C35/45 strength class. Void analysis references are from related 3DCP studies, not the ILS dataset directly.