OpenFOAM – cloud HPC

OpenFOAM^® on cloudHPC

Run OpenFOAM^® on hundreds of cores – without managing a cluster

Upload your case as it is, pick the OpenFOAM version and the number of cores, and launch. cloudHPC generates the mesh, decomposes the case, runs it in parallel and reconstructs the results for ParaView. You pay only while the machine runs.

Get 300 free vCPU hours OpenFOAM scalability guide

OpenFOAM simulation of a race car running on cloudHPC

v5 → latestopenfoam.org releases

v1706 → latestopenfoam.com releases

192physical cores on a single machine

300 hfree vCPU hours for new users

Every OpenFOAM branch

The version your case was built for, already installed

Pick the release that matches your dictionaries: no porting, no compiling, no environment to set up. New releases are added as they come out.

openfoam.org

The OpenFOAM Foundation releases, from version 5 to the latest.

openfoam.com

The ESI-OpenCFD releases, from v1706 to the latest.

foam-extend and DAFoam

foam-extend, and DAFoam for adjoint-based shape optimisation, including its turbomachinery branch.

How it runs

You upload the case. cloudHPC does the rest.

The platform handles the routine steps that take time on a cluster, and writes everything it does in the log.

Your case, as it is

The usual 0/, constant/ and system/ folders: the solver is read from controlDict
Your own Allrun script runs instead of the standard sequence
Custom solvers and libraries with an Allwmake are compiled before the run

Mesh generated for you

No mesh? blockMesh and snappyHexMesh run in parallel, or cfMesh if there is a meshDict
createPatch, changeDictionary, setFields and checkMesh run when their dictionaries are there
Mesh-only runs to check the grid before solving

Parallel without the bookkeeping

decomposeParDict created or set to the cores you choose
Multi-region and CHT cases decomposed with all regions
Results reconstructed at the end, with case.foam ready for ParaView

Change settings while it runs

Edit controlDict, fvSolution or fvSchemes on a running job: OpenFOAM re-reads them
Live graphs from the postProcessing folder
Stop cleanly at any time, writing the last time step

Restarts and spot machines

Relaunch in the same folder and the run continues from the latest time
If a machine stops while writing, the incomplete time step is removed automatically
Ready for preemptible (spot) machines to cut the cost of long runs

Handy extras

potentialFoam initialisation with one line in controlDict
Steady run followed by a transient one, from an extra fvSchemes-transient file
Mesh scaling, region splitting, extrusions and baffles from the same file

For OpenFOAM users

What changes compared with your workstation

Step	On a workstation or in-house cluster	On cloudHPC
Versions	Compile and maintain each release you need	openfoam.org, openfoam.com, foam-extend and DAFoam pre-installed
Decomposition	Edit `decomposeParDict` for every machine	Set automatically to the physical cores of the machine
Hardware	Limited to your cores and RAM	Up to 192 physical cores per run, with the RAM type you choose
Design variants	One case after another	Every variant on its own machine, at the same time
Monitoring	SSH into the cluster and tail the log	Log, live graphs and settings from the browser or the API
Post-processing	Download gigabytes of results	ParaView on a remote desktop, on the same storage

Pre- and post-processing

GUIs ready on a remote desktop

Prefer a graphical interface? Start a remote desktop in your browser: these tools are installed and read the same storage as your runs.

Ennova new

CAD import and repair, polyhedral meshes with boundary layers and OpenFOAM case set-up. Learn more

cfMesh and CF-MESH+

Automatic Cartesian, polyhedral and tetrahedral meshes with boundary layers, in GUI or batch. Learn more

BARAM

Open-source GUI by NextFoam for meshing and setting up incompressible, compressible, heat transfer and VOF cases.

HELYX-OS

Open-source GUI by ENGYS for the standard OpenFOAM libraries, suited to academic and entry-level projects.

MantiumFlow

Template-based set-up: provide the geometry, choose a template and run.

ParaView

Post-process large results without downloading them, on a machine with the RAM they need. Learn more

Choosing the machine

How many cores does my case need?

OpenFOAM runs one MPI process per physical core: choose highcore or hypercore, where vCPU = cores
Plan for at least 50,000 cells per core: below that, communication eats the gain
hypercore uses the newest CPU generation and is the fastest option
Use the price calculator to compare cost and run time before launching

Turbomachinery impeller simulated with OpenFOAM on cloudHPC

Turbomachinery example run on cloudHPC

Why cloudHPC

Built by OpenFOAM users

Expert support

Our team has worked with OpenFOAM since version 1.5, on industrial projects and research, and can help you set up, debug and scale your cases.

Pay per use

No subscription and no licence: you pay per vCPU hour, only while the machine runs, and spot machines cut the cost further.

Automate with the API

Launch, monitor and download from your own scripts with the cloudHPC API. The platform is ISO 27001 certified.

Get started

Your first OpenFOAM run in five steps

Register on cloudHPC and activate your account.
Upload your case folder to the storage.
Choose the OpenFOAM version and the cores on highcore or hypercore.
Launch and follow the log and the live graphs.
Post-process in ParaView on a remote desktop, or download the results.

Run your next OpenFOAM case in the cloud

Create your account and get 300 free vCPU hours to try it on your own case.

Create your account Read the documentation