# How to simulate water, falling under gravity, and impinging on a curved surface, which is kept/present in a domain, containing air?

TL;DR: How do I simulate a hole, at the bottom of a (full) water tank?

I am attempting to simulate water, flowing out of a hole/slit, at the bottom of a tank (Water Domain) (under the influence of gravity); entering into a domain (vertically downward), that contains air (Air Domain), and impinging on a curved surface. I am using COMSOL Multiphysics 5.4/5.3a. The idea is to simulate (in 2D) and obtain data on:

1. The distribution of water, as it flows along the surface, on two sides of the surface (for a 2D Simulation), for different initial points of impingement.

2. The (distribution of) stress produced, due to the impinging water, on the (different points of the) curved surface.

I have looked through a plethora of papers, as well as tutorials and none of them include, “liquid/gas, flowing out, under gravity, into another domain, containing a different liquid/gas”. I am assuming Laminar Flow and No Slip Boundary Condition, at the hole & walls and the curved surface.

Here, as far as I know, I cannot define an Inlet at the interface of the two domains. I have tried Multiphase (Phase Field), with no inlet and 1 outlet in the Air Domain (situated vertically downwards), but it did not converge. A Surface Plot of Fluid Velocity showed water, entering the Air Domain, but after a few more time steps, COMSOL threw a Did not converge error, thereby preventing the water from reaching the curved surface. I have tried the following, hoping to fix the issue:

1. Turning the “hole”, into a faucet or nozzle (of sorts), with a smooth downward curve: No change in outcome observed - Still not converging.

2. Defining a Pressure Point Constraint (1 atm) at the top of the Fluid Domain - No change in outcome observed - Still not converging.

Any help or hint is much appreciated. Also, if this problem (First paragraph - in Bold text) can be done in any other way (preferably in COMSOL), I’d like to hear that, as well.

• I have seen an approximation being done by applying Newtons laws on small voxels using GPU compute shaders some years ago. But have not found a solution with as accurate physics as one would like. – Emil Jan 18 '19 at 6:14
• @Emil Actually, I need the stress distribution over the surface. While I can simulate the system, like you mentioned, I don’t think, I can obtain the distribution of stress, over the surface. – P_0 Jan 18 '19 at 6:18