The ECHELON Plug-in for Petrel

Petrel^TM is a popular tool among geo-modelers and reservoir engineers used to build dynamic simulation models and analyze their results. Engineers invest time and effort into complex workflows and frequently exchange their Petrel projects rather than the resulting simulator data decks. Stone Ridge Technology developed the ECHELON® plugin for Petrel to give engineers the seamless ability to run their Petrel exported models in Eclipse^TM format using ECHELON. The ECHELON plug-in allows Petrel users to keep their Eclipse based workflows while experiencing the unparalleled performance of ECHELON.

The ECHELON® Plug-in for Petrel can launch ECHELON® on a local workstation, a networked HPC cluster, or the SRT Cloud.

The plug-in instructs Petrel that the simulation model should be exported in Eclipse format, which can be natively read without modification by ECHELON. The user can then define a new simulation case selecting ECHELON as the simulator. The plug-in enables the user to launch ECHELON locally on the Petrel Workstation, on a networked HPC cluster, or on the SRT cloud. Each is described below in more detail. Petrel’s workflows and uncertainty quantification cases, as well as other third-party plug-ins such as ResX^TM have also been successfully tested with the ECHELON plug-in.

Users running Petrel on a local workstation that has one or more GPUs are able to launch ECHELON simulations on the same machine. In this case, they would install the Windows version of ECHELON (Petrel is only available on Windows) and add the executable to the path. When launching the new case, the ECHELON plug-in waits for the simulation process to terminate before it loads the results, so they can be examined by the user in precisely the same way they are accustomed to working.

Larger organizations may have access to a networked HPC cluster with ECHELON installed. In this case, the user running Petrel on their local workstation would select the cluster option with the ECHELON plug-in. The only requirement is that the login node of the HPC cluster must be reachable via ssh from the Windows workstation.

The user will input the login and remote queue information in Petrel’s “queue definition” tab. The plug-in will use that information provided by Petrel to connect via ssh to the login node and submit, monitor, and possibly cancel the simulation job associated with ECHELON.

Using the plug-in’s GUI, the user can map Windows paths to Linux directories, if those are available. Otherwise, the plug-in can copy (via scp or sftp) the simulation data decks to the cluster and copy-back the results once the job is finished.

Currently, three queueing systems are supported: IBM’s LSF, PBSPro, and UGE (Univa Grid Engine). The plug-in will issue the corresponding commands for those schedulers; submitting jobs, checking their statuses, and killing jobs if directed by the user. For unconventional cluster configurations, it’s also possible to generate a custom script for submitting jobs.

A third possibility for running ECHELON is to use the SRT Cloud based on AWS. The user can create a cloud account using the plug-in’s interface or a provided command line tool. Similarly to the HPC case, Petrel’s files will be copied via scp to the Cloud. Once the job is submitted, it will be monitored until it is completed, then the results will be copied back. The user can purchase SRT Cloud computing hours using the plug-in’s GUI.

CONCLUSION

The ECHELON Petrel plug-in seamlessly integrates ECHELON into existing Petrel workflows, it allows users to experience unparalleled performance with no change to the project’s workflow. ECHELON can be launched locally on the same machine where Petrel is running, on a locally accessible HPC cluster or on the SRT Cloud. For questions on the plug-in and its availability please email us at info@stoneridgetechnology.com.

*Petrel and Eclipse are trademarks of SLB
*ResX is a trademark of HAL
*ECHELON is a registered trademark of SRT