https://doi.org/10.5061/dryad.w3r2280xj

Java versions (.java) of COMSOL (.mph) files providing fluid dynamics simulations of modeled archaeocyath taxa in a sweep of fluid velocities. COMSOL files have been cleared of meshes and solutions for space. Necessary Archaeocyatha taxon 3D geometry files are also provided (Favilynthus mellifer, Archaeolynthus porosus, and the Balsam and Vogel model). Archaeolynthus porosus model is labeled PartialPelta-bisected.step. The following text is cut/pasted from the methods section of the in-review manuscript:

2.1 Digital model construction
We used Autodesk Fusion 360 to construct digital models of three archaeocyaths (figure 1). All models were constructed as non-uniform rational basis spline (NURBS) geometries and are provided as supplemental data (electronic supplementary material, S1a–c). Two single-walled archaeocyath taxa were chosen (cylindroconical-shaped Favilynthus mellifer and chalice-shaped Archaeolynthus porosus) based on their availability of detailed systematic descriptions, body and pore morphological descriptions, and published fossil photographs [8,30] (figure 1a–e; electronic supplementary material,table S1). We also replicated the double-walled idealized aluminium models used by Balsam & Vogel [12] (‘BVM’; figure 1e,f).

CFD Setup     
Archaeocyaths are reconstructed to have lived in variety energy conditions within open reefs, sheltered and/or lagoonal settings, and even more sheltered cavity dwellers [3,7,8,31,32]. Such disparate flow conditions suggest average ambient currents <10 cm s−1 for more sheltered conditions [33], and faster currents for reef crest communities [7] that would have been closer to 30–35 cm s−1 [33]. To account for this environmental variability, we simulated inlet flow velocities of U = 1, 5, 10, 15 and 30 cm s−1 for the three scaled models, informed by the characteristic flow velocities reported from modern coral reefs.

Three-dimensional, incompressible fluid flow simulations were performed using COMSOL Multiphysics. CFD is optimal for investigating small-scale flow patterns both within and surrounding the cavity and pores, and allows experimentation with multiple morphotypes and simulated flow velocities. We used hexahedral flow domains (electronic supplementary material, table S1) with the organism placed approximately one third of the length along the lower surface. The fluid properties of water [density (ρ) = 997 kg m–3, dynamic viscosity (μ) = 0.001 kg s–1 m–1] were assigned to the domain surrounding the model.

A stationary Reynolds Averaged Naiver Stokes (RANS) shear stress transport (SST) turbulence model was used with a slip condition applied to the upper surface of the flow domain and sides parallel to flow. No-slip conditions were prescribed along the seafloor and organism model where velocity approaches 0 cm s−1. The flow domain was discretized into tetrahedral elements in the far field with hexahedral boundary layer elements in the vicinity of no-slip boundaries. A mesh sensitivity analysis was performed with an inlet velocity of U = 30 cm s−1. Our final meshes selected based on a compromise between computational efficiency and accuracy (electronic supplementary material, figure S1). Because of the geometric and computational complexities of the A. porosus CFD simulations, the domain and model were bisected down the model’s centre, permitting its symmetry to be used necessitating only half the domain to be solved.

To demonstrate comparability between methods, we replicated Balsam and Vogel’s [12] physical experiments using their same Reynolds number corresponding to a water inlet velocities of 9 cm s−1 (that would have been a fourteen fold increase to 126 cm s−1 for air) (electronic supplementary material, material, table S2; [12]). Because Balsam and Vogel’s [12] experiments were inherently time dependent, we also completed 9 cm s−1 large eddy simulation (LES) CFD simulations for comparability using similar boundary conditions implemented in the RANS SST simulations [34]. Instead of a depth-averaged velocity profile prescribed to the inlet, we used the flow field from the RANS solution as our inlet condition. Flow fields were solved for 180 s to permit flow to fully develop before we exported flow fields every 0.01 s for 30 s.