Technology | cloudfluid — Tailored Flow Simulation

Gas and liquid flows occur in a wide variety of technical and non-technical systems. The associated phenomena of fluid dynamics are extremely diverse and complex. The evaluation of flow problems using experiments with complex prototypes is therefore very expensive in many cases, if it is possible at all.

Computational fluid dynamics (CFD) enables the rapid, digital development of products:

fewer experiments, less expensive prototypes,
experimentally inaccessible parameters can be recorded and
physical effects can be considered in isolation.

CFD has therefore established itself as a valuable tool in engineering research and development. However, due to the large number of models required for the respective special cases and complex parameters, CFD has so far been a tool that must be operated by specialists in order to achieve the desired accuracy and reliability of the simulation results. In addition, enormous computer capacities are required to solve the computationally intensive algorithms. This has limited the use of CFD - despite its advantages. The personnel, software and hardware costs associated with flow simulation have been a decisive factor. The cloudfluid offer addresses these central problem areas. Overcome these hurdles with our innovative Simulation as a Service:

Customized solutions

Precise flow simulations accessible to non-specialists through user-friendly and customized software solutions

performance

Highly efficient implementation on high-performance graphics cards in a secure cloud

Pay-per-use

Flexible pay-per-use concept replaces large annual licenses, expensive hardware investments and fixed personnel costs

The Lattice Boltzmann Method

Our cloudfluid solver uses the innovative Lattice Boltzmann Method (LBM) to solve flow problems. As the name implies, the physically well-founded Boltzmann equation is solved numerically by discretizing it on a lattice.

Solving the discretized Boltzmann equation leads to a completely different algorithm than solving the Navier-Stokes equation, THE basic equation in flow simulation, using, for example, the finite volume or finite element method. The LBM can be considered to lie between methods based on molecular dynamics with individual collisions and methods that solve the macroscopic conservation equations in the Navier-Stokes equations. This approach is often referred to as mesoscopic.

The LBM algorithm is characterized by very simple parallelization and high numerical efficiency, so that time-resolved flow simulations in particular can be carried out significantly faster than with other solution methods. By incorporating further models and algorithms, various physical processes such as thermodynamic processes in fluids or solids, flows in porous media or multiphase flows can be calculated using LBM.

Its efficiency leads to a qualitatively new level of understanding, as it enables the solution of problems that previously could not be tackled at all, only with greatly simplified model approaches or with enormous computing times.

Advantages of the LBM

Computing capacity and scaling

Traditional discretization methods such as the Finite Volume Method (FVM) or the Finite Element Method (FEM) require a great deal of expertise and large computing capacities. The Lattice Boltzmann method, on the other hand, offers excellent precision of results with an easily controllable choice of model parameters. This algorithm is characterized by excellent parallelizability and has proven to be highly performant in comparison:

In a benchmark study, we compared our cloudfluid solver against OpenFOAM, an open source solver based on the FVM. Based on the transient simulation of a flow in a combustion engine, a speedup of up to 250 was determined for the simulation with equivalent results compared to a high-precision experiment - cloudfluid is more than two orders of magnitude faster:

With our technology, the simulation is executed on high-performance graphics cards (GPU). By using LBM on GPUs, cloudfluid is one of the fastest fluid solvers in the world, delivering accurate results for even complex flow problems overnight. Our use of GPUs in cloud systems means that neither in-house computing capacity nor costly maintenance is required. This reduces hardware costs to a minimum.

Automated workflows

With our robust and fully automatable workflow, we hide the complexity of a fluid simulation and develop a customized solution based on the wishes of our customers. The graphical user interface can be adapted to individual ideas and tasks. All you need to do is select your process parameters and geometry - numerical models and associated parameters are pre-configured by us and validated for your application.

Pre- and post-processing

Conventional solutions have required an experienced user and complicated algorithms for grid generation, resulting in time-consuming preprocessing. With cloudfluid, the meshing of the computational domain is fully automated and extremely fast - grids with up to 500 million cells can be created in just a few minutes.

In addition, we offer you user-specific post-processing. We record the desired results during the simulation, whereby we start evaluating the data during the simulation - this allows for an efficient, reproducible and highly automatable process. The simulation results are then prepared in tables and graphics so that you always have the relevant data at a glance.

Security and Cloud

The latest hardware thanks to hyperscalers

With our cloud-based simulation environment, you always benefit from the latest hardware generation of powerful GPUs without having to operate your own computing resources. We use leading hyperscalers such as AWS and Google Cloud to provide the graphics cards so that you can run a large number of simulations in parallel. This allows you to carry out extensive parameter studies or Design of Experiments (DOE) in a fraction of the time and significantly shorten your development cycles.

Maximum security through encryption

The security of your sensitive data is a top priority for us and is an integral part of our infrastructure. Your data is protected by state-of-the-art cryptographic security concepts during transmission and storage. All data transfers are made via HTTPS and are end-to-end encrypted. At rest, we use the AES256 encryption algorithm to protect data from unauthorized access.

Isolated instances for maximum security

Each simulation runs in its own isolated instance. This strict separation ensures that no computing resources are shared. The instances are systematically generated from a standardized template and deleted immediately after completion of the simulation. This ensures a high level of reproducibility and protects your data from unauthorized access during the simulation.

Regional servers for optimal performance and compliance

We offer two separate deployments with servers in the EU and the USA. This means that your data always remains in your region during the simulation and subsequent storage, which makes it easier to comply with local legal regulations on data protection and data security. Data is only exported from the respective region for support purposes and only after prior consultation. You also benefit from higher transfer speeds as the servers are located close to you.

Still unsure how our technology can help you? We will be happy to analyze this with you during a free initial consultation. Simply make an appointment with us.

Free Analysis of your use case