Fluids are complicated. The equations that govern them are generally very tough to solve, and cover such a broad range of possible situations that no single one-size-fits-all method exists to accurately predict the relevant behaviours. And these predictions are relevant: from cosmetics to construction, understanding the details of how fluids mix and move around particles turns out to be crucial for creating the products we need to drive innovation. In fact, the problem is so hard and so useful that there’s a million-dollar Millenium Prize waiting for the person who solves it.
Why simulation technology matters
So what specifically do we simulate, and why is it important? At Electric Ant Lab we’re mostly concerned with the mesoscale: physics on the scales from tens of nanometers to millimeters.
On this scale, complicated structures such as interfaces between fluids, droplets, and particles form and influence the way mixtures flow through a system. All these effects go into simple questions when creating new products such as “will these two liquids mix?” or “if they don’t mix, can we make them mix by adding something else?”.
Until now, the solution has been “simple”: go to the lab, mix some liquids and particles together, and voilà! But is it so simple in reality? Companies spend millions on R&D every year, with the money going into fully stocked labs full of expensive equipment, and a lot of manpower to do everything from creating the liquids, taking the measurements, all the way through to simple equipment maintenance.
A tool to accelerate innovation
This is where we come in. With cutting-edge advances in computational physics, we are getting to the point where we can usefully simulate the systems encountered in industry. This not only takes a lot of the cost out of creating new products out of the whole process, but also means we can understand what’s happening in greater detail, hence accelerating future innovation.
How do we do it? Well, there are various approaches to model the physics on the mesoscale ranging from traditional mesh-based methods, to methods that are based on statistical physics, and many hybrids of the two. The thing they all have in common is that we have to break the problem down into solvable chunks. Computers are very bad at storing continuous mathematical functions, but they’re very good at storing single numbers, so we break a fluid down into points, then put more and more points in until we can build up a consistent picture (similar to how computer generated graphics work).
Smoothed Particle Hydrodynamics (SPH)
The codes developed at Electric Ant Lab use one of the above conceptual hybrids: smoothed particle hydrodynamics (SPH). In this method, we break the fluids down into “blobs”, and these blobs are free to move around, push each other around, and exchange material. In other words, we have broken the continuous fluid down into points (i.e. the particles in “smoothed particle hydrodynamics”), and used a smoothing function to turn this into blobs, hence the name SPH.
The SPH method has many advantages, including the ability to introduce molecular interactions, and the ability to solve transport and flow fields with clear separation of advection and diffusion. SPH is also easily parallelized, so that we can usefully scale up the computational power and hence get an answer within a reasonable time, while taking full advantage of GPU acceleration.
Providing the full picture
One of the reasons that SPH is so accurate is because it is mesh-free, meaning that the blobs are not constrained to specific points on a grid, and therefore do a better job of representing, as a whole, the domain of the physical problem. Since its invention in 1977, the SPH method has been improved and extended to a diverse range of problems in both fluid and solid mechanics, including novel state-of-the-art developments made at Electric Ant Lab. Because of its consistent theoretical framework, SPH allows a straightforward handling of very large deformations and properly conserves quantities like energy, momentum, mass, and entropy, in contrast to other methods where these can break down.
The work we do with SPH at Electric Ant Lab is redefining what is possible for individual groups to simulate in a wide range of industries, and our continuous developments and innovations ensure that this range is constantly expanding. We also strive to make the simulations as simple as possible for the user, by making the expert choices needed for our users, while retaining the flexibility needed. We hope that our user-friendly interface will bring the SPH method, and mesoscale simulations in general, to the people and industries who most need them.
An SPH simulation of a mixture of decane and water, where each sphere is a blob of fluid. Each blob interacts with its neighbours, dictating how the various forces push and pull the fluid.
A simplified diagram of two SPH blobs (black circles) with their smoothing functions (black lines) centred at the block dots. The overlap between the two blobs (shaded area) is where the forces are calculated by considering the physics of molecular interactions.