Transparent 3D-printed flow cells for porous media studies with proof-of-concept on two-phase flow-reversal asymmetry

arXiv:2610.04105 · physics.flu-dyn · Submitted 2026-10-02 · Read on arXiv

Listen

Transcript

Introduction to the show: ident: Physics. The day in general physics: instruments, methods and results across the field.

Kai: I'm Kai.

Mira: And I'm Mira, here with Lev.

Kai: We're looking at the paper: "Transparent 3D-printed flow cells for porous media studies with proof-of-concept on two-phase flow-reversal asymmetry".

Mira: Here's the gist of the paper. Transparent 3D-printed flow cells for porous media studies with proof-of-concept on two-phase flow-reversal asymmetry provides a reproducible and open-source protocol to fabricate transparent monolithic porous flow cells by stereolithography 3D printing The triangular geometry produces a clear directional response, whereas the cylindrical geometry remains symmetric within experimental uncertainty This work demonstrates that our fabrication protocol can link digitally designed pore geometries to macroscopic pressure response and steady-state two-phase flow organisation.

Kai: Fabrication Protocol Capabilities The protocol offers a workflow consisting of three main steps: digital design and 3D printing of the monolithic cell, flushing of the internal channels to remove residual uncured resin, and external resin coating to improve optical transparency.

Mira: The geometries are generated in FreeCAD 0.20.2 using a Jupyter notebook that creates the porous structure directly from user-defined geometrical parameters such as chamber dimensions, pillar shape, and target porosity.

Kai: For the proof-of-concept experiments, Hele-Shaw-type quasi-two-dimensional porous cells were fabricated with a fixed gap height of h = 1 mm.

Title and authors: Kai: So we're looking at this paper, "Transparent three dee-printed flow cells for porous media studies with proof-of-concept on two-phase flow-reversal asymmetry" again <ref:2610.04105#pg1,Transparent 3D-printed flow cells for porous media studies with proof-of>. The title tells you immediately that they are using transparent three dee prints to study how two different fluid flows behave when the direction is reversed <ref:2610.04105#pg1>.

Mira: It’s about linking the digital design of these porous structures to actual pressure responses and steady-state organization of those fluids. That suggests they aren't just building pretty objects; they are testing a mechanism.

Lev: The mechanism here seems to be how the pore geometry itself can introduce a directional dependence into the constitutive relation between pressure drop and flow rate in steady two-phase flow.

Kai: They investigate this by comparing steady oil-air two-phase flow in both AB, which is A to B, and BA, which is B to A directions at the same imposed total flow rate.

Mira: And they quantify this asymmetry using two metrics: a pressure-based quantity called IP and an image-based metric called IS.

Lev: That’s what you need to measure the asymmetry directly, both through pressure drops and through visual analysis of the flow patterns inside the cell.

Kai: The key finding for T85 is that IP, which is the corrected pressure-based flow-reversal asymmetry I*P, increases with increasing flow rate.

Mira: They also found that IS remains small at low flow rates but increases rapidly around three hundred mL h-one and stays elevated at higher rates <ref:2610.04105#pg1>.

Lev: That observation about the image-based analysis showing a rapid increase in IS around three hundred mL h-one seems to be where the real insight is coming from for how you might design something practical <ref:2610.04105#pg1>.

Kai: In contrast, they found that C85 and BL geometries remain symmetric within experimental uncertainty.

Mira: That symmetry is important because it shows that the effect isn't universal; it really depends on the specific pore geometry you choose to print.

Lev: It suggests that if you want to get a directional response, you need to design the obstacles in a way that breaks mirror symmetry at the pore scale.

The paper's summary: Kai: To summarize what they did with this paper, they provided a reproducible and open-source protocol for fabricating transparent monolithic porous flow cells using stereolithography three dee printing <ref:2610.04105#pg1,a reproducible and open-source protocol>.

Mira: They emphasize that the flushing and coating steps are central to making the cells functional for imaging, as those steps clear out uncured resin and make the surfaces optically transparent.

Lev: The whole point is to provide a practical route for researchers to adapt these three dee-printed flow cells to different experimental applications using consumer-grade stereolithography printers <ref:2610.04105#pg1>.

Kai: They are showing that this approach can link digitally designed pore geometries directly to macroscopic pressure response and steady-state two-phase flow organization, which was a demanding test case because the macroscopic pressure response depends on the pore-scale organization.

Mira: This is important because traditional methods for making these porous models often have difficulty reproducing the specific pore-scale configuration they need to control.

Lev: So, what this paper offers is a practical path toward creating compact passive flow-control elements, like a two-phase fluidic diode, by using these three dee printed structures <ref:2610.04105#pg1>.

Kai: They are demonstrating that the directional dependence in pressure drop versus flow rate can be introduced purely by the geometry of the obstacles themselves.

Mira: And they show how you can use AI to classify porous media geometries by measuring that image-based flow-reversal asymmetry IS, which lets an AI predict the directional pressure response before physical testing even happens.

The paper's improvements: Lev: They suggest improvements mainly through the measurement of these two metrics, IP and IS. IP is the pressure-based quantity comparing mean steady-state pressure drops at the same imposed flow rate in both AB and BA directions.

Kai: And IS is the image-based metric that compares active air-flow widths measured in the AB and BA directions. These are how you see what's happening visually inside those porous structures.

Mira: The improvement here is moving beyond just measuring a single pressure drop to quantifying the asymmetry using both pressure and visual flow pattern comparisons simultaneously.

Lev: That’s useful because it gives you two different ways to probe the same underlying physical phenomenon, which helps confirm that the directional dependence isn't just an artifact of one measurement technique.

Kai: They also show that for T85, IP increases with flow rate, and IS shows a rapid increase around three hundred mL h-one <ref:2610.04105#pg1>. That specific quantitative relationship is what they found to be most interesting in terms of control potential.

Mira: If we can use this data to train generative models, we can use the fabrication protocol to design optimal pore structures for specific flow control goals, like creating that fluidic diode mentioned earlier.

Lev: That would mean you could design a structure first, then print it, and then you have a physical device that responds directionally without needing complex modeling beforehand.

Conclusion: Kai: So to wrap up this paper on "Transparent three dee-printed flow cells for porous media studies with proof-of-concept on two-phase flow-reversal asymmetry," the main implication is that pore geometry can introduce a directional dependence into the pressure drop versus flow rate relationship in steady two-phase flow <ref:2610.04105#pg1,Transparent 3D-printed flow cells for porous media studies with proof-of>.

Mira: They showed this using a transparent, three dee printed method, and they quantified it with both pressure and image metrics to confirm that the asymmetry is linked to broken mirror symmetry at the pore scale <ref:2610.04105#pg1>.

Lev: The practical application here is developing compact passive flow-control elements, like a two-phase fluidic diode, where you rely on geometry for control rather than complex active systems.

Kai: They also provided a way to lower the barrier for fabricating these transparent millifluidic flow cells, showing that this protocol can be used with consumer-grade stereolithography printers and common equipment.

Mira: The method they offer is open source, providing the design files, processing scripts, and analysis software so other researchers can adapt it for their own needs.

Lev: From an engineering standpoint, it shows you how to build a physical device where the directional response is governed by microstructural arrangement rather than just external forces.

Kai: It’s interesting work because it moves the goal of creating transparent millifluidic flow cells from being a specialized facility thing to something accessible using this three dee printing protocol <ref:2610.04105#pg1>.

Mira: This paper gives us a concrete, measurable way to test and control these fluid dynamics in a physical system built with readily available tools.

Department of Physics, PoreLab, Norwegian University of Science and Technology (NTNU) · Departments of Geoscience and Physics, PoreLab, Njord Centre, University of Oslo · Department of Physics, University of Cambridge

physics.flu-dyn

Submitted: 2026-10-02

Updated: 2026-10-28

Comments: 30 pages, 11 figures. v3: corrected and expanded bibliographic information; revised author list and acknowledgements; added article highlights and an author-contribution statement to meet journal submission requirements. Scientific content and results unchanged. Accompanying design files, data, and analysis code are available on Zenodo

License: http://creativecommons.org/licenses/by/4.0/

Importance score: 74/100

The gist: Transparent 3D-printed flow cells for porous media studies with proof-of-concept on two-phase flow-reversal asymmetry provides a reproducible and open-source protocol to fabricate transparent

Terminology

Hysteresis — 3× in this paper · explained in 3 episodes
Accuracy — 2× in this paper · explained in 1 episode
Core — 2× in this paper · explained in 2 episodes
Energy Efficiency — 1× in this paper · explained in 4 episodes
Generative Model — 1× in this paper · explained in 2 episodes
Generative Models — 1× in this paper · explained in 4 episodes
Generator — 1× in this paper · explained in 2 episodes
Temporal Correlations — 1× in this paper · explained in 2 episodes

Summary

Transparent 3D-printed flow cells for porous media studies with proof-of-concept on two-phase flow-reversal asymmetry provides a reproducible and open-source protocol to fabricate transparent monolithic porous flow cells by stereolithography 3D printing The triangular geometry produces a clear directional response, whereas the cylindrical geometry remains symmetric within experimental uncertainty This work demonstrates that our fabrication protocol can link digitally designed pore geometries to macroscopic pressure response and steady-state two-phase flow organisation.

Fabrication Protocol Capabilities

The protocol offers a workflow consisting of three main steps: digital design and 3D printing of the monolithic cell, flushing of the internal channels to remove residual uncured resin, and external resin coating to improve optical transparency. The geometries are generated in FreeCAD 0.20.2 using a Jupyter notebook that creates the porous structure directly from user-defined geometrical parameters such as chamber dimensions, pillar shape, and target porosity. For the proof-of-concept experiments, Hele-Shaw-type quasi-two-dimensional porous cells were fabricated with a fixed gap height of h = 1 mm. The manufacturing workflow involves printing using Clear Resin V4.1 on a Form 3B+ stereolithography printer, with a layer thickness of 25 µm.

Resin Removal and Optical Quality

The flushing step is crucial for removing uncured resin from internal channels, performed by repeatedly infusing isopropanol through the device using a programmable syringe pump. After flushing, the cell is dried by flowing air continuously through the internal channels to remove residual isopropanol. To improve optical transparency, a thin resin coating is applied to the external optical surfaces using Clear Resin V4.1. This coating fills small printing-induced surface irregularities, thereby producing smoother optical surfaces and enabling clearer imaging of the two-phase flow.

Flow Reversal Asymmetry Metrics

The study investigates flow-reversal asymmetry by comparing steady oil–air two-phase flow in both AB (A → B) and BA (B → A) directions at the same imposed total flow rate. Two metrics are introduced to quantify this asymmetry:

The pressure-based quantity IP compares the mean steady-state pressure drops ∆P(Q) measured at the same imposed flow rate Q in the AB and BA directions, defined as IP (Q) = ∆PBA(Q) − ∆PAB(Q).
The image-based metric compares the active air-flow widths measured in the AB and BA directions, defined as IS(Q) = WBA(Q) − WAB(Q).

Key Findings on Flow Asymmetry

The results show that for T85, the corrected pressure-based flow-reversal asymmetry I∗P increases with increasing flow rate, with a pronounced increase between 200 mL h−1 and 400 mL h−1. The image-based analysis also shows that for T85, IS remains small at low flow rates but increases rapidly around 300 mL h−1, remaining elevated at higher flow rates. In contrast, C85 and BL geometries remain symmetric within experimental uncertainty. The observed breaking of flow-reversal symmetry accompanies the broken mirror symmetry of the triangular obstacles at the pore scale.

Implications for Flow Control

The directional response arises from pore-scale geometry and interfacial dynamics, suggesting that asymmetric pore geometries can couple to these processes strongly enough to produce a different hydraulic response when the direction of the imposed flow is reversed. This approach could potentially enable compact passive flow-control elements, such as a two-phase fluidic diode. The protocol aims to lower the barrier to fabricating transparent millifluidic flow cells, while the proof-of-concept experiments demonstrate that pore-scale geometry alone can introduce a directional dependence into the constitutive relation between pressure drop and flow rate in steady two-phase flow. The design files, processing scripts, and analysis software are openly available in the accompanying Zenodo repository. The authors thank Lydéric Bocquet, Marie-Caroline Jullien, Daan Frenkel, and Marcel Moura for many stimulating discussions. The project received funding from the European Union’s Horizon Europe research and innovation programme under the Marie Skłodowska-Curie Doctoral Network FLUXIONIC, grant agreement No. 101119598. The authors declare that they have no competing interests. The complete materials and equipment list, processed data, analysis scripts, design files, LabKit classifiers, and representative image data supporting the findings of this study are available in the accompanying Zenodo repository. The full raw image data are not included in the repository because of their large file size. The references include Anbari et al. 2018, Au et al. 2016, Armstrong et al. 2016, and others. The paper is available on arXiv:2610.04105v2 [physics.flu-dyn] on 19 Oct 2026 The paper is available on page 17 The paper is available on page 8 The paper is available on page 3 The paper is available on page 6 The paper is available on page 9 The paper is available on page 8 The paper is available on page 3 The paper is available on page 7.

--- Page 8 ---

The Reynolds and capillary numbers for phase α, where α = o denotes oil and α = g denotes gas, are defined as Reα = ρuαd / µα, Caα = µuα / σ. The mean phase velocity is estimated as uα = Qα / ϕASα, where ϕ is the porosity, A is the cell cross-sectional area normal to flow, and Sα is the phase saturation. For a porous region of area Ap containing N obstacles with equivalent-area radius robs, the characteristic pore size is estimated as d = r / (Ap πN − robs).

--- Page 9 ---

Fig. 5 summarises the estimated Reynolds and capillary numbers over the investigated flow-rate range. The oil Reynolds number remains well below unity, while the gas Reynolds number remains of order unity or below, indicating weak inertial effects over the investigated range. The wetting properties of the oil on the 3D-printed solid were characterised with a drop-shape analyser, using 0.5 µL sessile drops. Observed contact angles decrease towards zero as a power-law in time, consistent with the silicone oil being fully wetting on the 3D-print resin. Because oil is the wetting phase, the oil-based capillary number is used to summarise the balance between viscous and capillary effects.

--- Page 17 ---

The pressure response is shown in Fig. 10.

Improvements for AI systems

Image-based flow-reversal asymmetry quantification allows AI systems to classify porous media geometries by measuring IS (Image-based flow-reversal asymmetry). This enables an AI to predict the directional pressure response based on a 3D model's geometry before physical testing.
The protocol for fabricating transparent monolithic porous flow cells can be used to train generative models that design optimal pore structures for specific flow control goals, such as creating a two-phase fluidic diode.
AI can be trained to correlate the observed pressure and image-based metrics with specific microstructural arrangements, such as the observation that the observed breaking of flow-reversal symmetry accompanies the broken mirror symmetry of the triangular obstacles at the pore scale.

Abstract

Porous media experiments commonly rely on geometry-controlled and optically transparent experimental flow cells. Examples include studies of flow structure, solute mixing, multiphase flow, bacterial dynamics, and reactive transport. Here, we provide a reproducible and open-source protocol to fabricate transparent monolithic porous flow cells by stereolithography 3D printing. The protocol covers digital design, resin printing, internal flushing, and resin coating. The flushing and coating steps are central to the workflow, enabling enclosed single-material cells to be cleared of uncured resin and made optically transparent for direct imaging. The protocol produces optically transparent, millimetre-scale, and quasi-two-dimensional flow cells suitable for simultaneous pressure and image-based measurements. As a proof of concept, we study flow-reversal asymmetry in steady oil-air two-phase flow in quasi-two-dimensional Hele-Shaw cells containing either triangular or cylindrical obstacles. Directional differences are quantified from pressure measurements and from image-based active-width measurements. The triangular geometry produces a clear directional response, whereas the cylindrical geometry remains symmetric within experimental uncertainty. These results demonstrate that our fabrication protocol can link digitally designed pore geometries to macroscopic pressure response and steady-state two-phase flow organisation. By making the design files, scripts, and manufacturing workflow openly available, we aim to make 3D-printed flow cells readily adaptable to other experimental systems and research fields.