Intelligent Light and FieldView

Intelligent Light and the CFD 2030 Vision - Live presentation at NASA Aug 17

Intelligent Light and the CFD 2030 Vision​

Steve M. Legensky
and Brad J. Whitlock
Intelligent Light

NASA Ames Research Center
August 17, 2017
Video Archive Available Here

ABSTRACT

Over the past thirty years, computational simulation of fluid dynamics has made huge strides in meshing of complex geometries, computational efficiency and most importantly, greater fidelity in physics models. Current trends include greater adoption of unsteady methods via LES, higher order methods and alternatives to classical CFD such as Lattice Boltzmann methods. However, the majority of engineering applications remain constrained by computational and storage resources as well as schedule and time pressure. The CFD 2030 Vision highlights the need for improved Knowledge Extraction and Data Management tools that address the scale and fidelity of exascale class problems. Perhaps most interesting, these tools will enable non-deterministic engineering (NDE) in which variations in boundary conditions, discretization and models can more realistically predict the behavior of complex aerodynamic, propulsion or power generation systems.

This talk highlights recent activities at Intelligent Light that include how to manage the massive data flows resulting from ensembles of unsteady CFD calculations, new tools that support engineering use of uncertainty quantification techniques and the latest HPC-FieldView software that combines the familiar interface of FieldView with the scalability of VisIt. Through the lens of post-processing and data analysis, we have gained a unique perspective on research and engineering use of CFD from the late 1980's through today. The talk touches on this history while giving examples of state-of-the-art petascale CFD and surveys the challenges that exascale computing is intended to address.

Speakers' Bio

Steve M. Legensky - Founder and General Manager of Intelligent Light

Steve M. Legensky is the founder and general manager of Intelligent Light, a company that has delivered products and services based on visualization technology since 1984. He attended Stevens Institute of Technology in Hoboken, New Jersey and received a BE degree in electrical engineering in 1977 and a MS degree in mathematics in 1979. Steve's passion is applying computer graphics and data management to difficult engineering problems. Steve is an Associate Fellow of the American Institute of Aeronautics and Astronautics (AIAA) and has published and presented for AIAA, IEEE, ACM/SIGGRAPH and IDC.

Brad J. Whitlock – Visualization and Post-Processing Engineer

Brad Whitlock is a visualization and post-processing engineer in Intelligent Light's Applied Research Group and a founding developer of VisIt. VisIt is a powerful, massively parallel visualization tool designed for high performance computing environments. Brad joined Intelligent Light in 2013 to develop a commercially-hardened version of VisIt that serves as an enabling technology for Intelligent Light's HPC FieldView product. Brad's interests lie in scientific visualization, in situ processing, and parallel programming.

Video Archive is available here

For previous topics in AMS seminar series and to view recorded sessions: http://www.nas.nasa.gov/publications/ams 

0 Comments

Visions of Exascale CFD: ASME Fluids Engineering Division Summer Meeting

IL Founder & General Manager, Steve Legensky, will speak at the ASME Fluids Engineering Division Summer Meeting​ at the Hilton Waikoloa Village, Waikoloa, Hawaii

Steve M. Legensky, Founder and General Manager, Intelligent Light

Visions of Exascale CFD

Over the past thirty years, computational simulation of fluid dynamics has made huge strides in meshing of complex geometries, computational efficiency and most importantly, greater fidelity in physics models.Current trends include greater adoption of unsteady methods via LES, higher order methods and alternatives to classical CFD such as Lattice Boltzmann methods.However, the majority of engineering applications remain constrained by computational and storage resources as well as schedule and time pressure.The Department of Energy's Exascale Computing Project (ECP) offers the capacity for scale, fidelity and perhaps most interesting, non-deterministic engineering (NDE) in which variations in boundary conditions, discretization and models can more realistically predict the behavior of complex aerodynamic, propulsion or power generation systems.

Intelligent Light is participating in research activities aimed at managing the massive data flows resulting from ensembles of unsteady CFD calculations and implementing tools to support engineering use of uncertainty quantification techniques.Through the lens of post-processing and data analysis, we have gained a unique perspective on research and engineering use of CFD from the late 1980's through today.The talk will touch on this history but have a primary focus on state-of-the-art petascale CFD in aerospace, combustion and wind energy and survey the challenges that ECP is intended to address.

To request a meeting during the conference, please send email to events@ilight.com.

0 Comments

Uncertainty Quantification (UQ) at ASME V&V Symposium

Uncertainty Quantification (UQ) of CFD data.

Earlier this month at the ASME V&V symposium, Seth Lawrence, a graduate student at our University Partner Northern Arizona University, presented his Master's thesis work on "Verification, Validation and Uncertainty Quantification of Turbulent Twin Jets". Seth was advised by our own Dr. Duque who maintains an adjunct Faculty position @ NAU. This event was Seth's first outing at a major international technical symposium. He did a great job of presenting (and defending) his work to the leaders in the field of V&V/UQ, such as Oberkampf, Roy, Celik and Eca. The work was a Challenge Problem sponsored by the ASME V&V 30 Committee.  GREAT JOB SETH!    

In Uncertainty Quantification (UQ), engineers utilize standardized procedures such as the ASME V&V 20 and V&V 30 guidelines to account for the effects of probabilistic inputs to a CFD simulation to arrive at a non-deterministic answer. Through UQ, an engineer could state with 95% certainty answers to their design question while justifying and documenting how they arrived at their answer.

This challenge problem was the only one at the symposium to focus on UQ. It is a key area of interest for those seeking to capitalize on information gleaned from verification & validation work in new design studies. 

To combine CFD and UQ analysis, Mr. Lawrence created an automated workflow using FieldView to post-process the results of Fluent solutions and pass data to Dakota (Sandia National Lab) and then pass data from Dakota as input to Fluent in an iterative process. FieldView was also used to visualize the CFD data to create images for the presentation and 3D PDF to share results.

"Seth did a great job presenting to the leaders in the VVUQ community. His work was well received and cited by other presenters later in the symposium.  It was gratifying to hear statements among veteran symposium participants including 'This is the first time I've seen error bars on a CFD result, very impressive.'"

Earl P. N. Duque, PhD Manager of Applied Research at Intelligent Light

Mr. Lawrence noted that he enjoyed the chance to see how the experts in this field approached the benchmark ASME turbulent twin jet numeric model validation problem.

Professor Tom L. Acker from NAU and Intelligent LIght's Earl P.N. Duque served as advisers on the project.

​Mr Lawrence used the 3D PDF export capability in FieldView throughout the development of the CFD model, allowing him to easily share results of his grid convergence study (CGI) and in the observed order of the solver (p-obs). 3D PDF files are downloadable below.

"Throughout the development of the CFD model, I made good use of the 3D PDF generator that is available in the new FieldView 16.1 package. This was very helpful in the presentation of model results, and provided the ability to easily send detailed model results of large CFD datasets in the form of a small file via email, and the recipient does not need any special software to view the 3D PDF results - fantastic!"

Seth Lawrence, Northern Arizona University

Download 3D PDF. 

Numerical uncertainty in y-velocity.

Download 3D PDF.

Observed y-velocity.

Adobe Acrobat Reader recommended for viewing. Do not attempt to open these 3D PDF documents from your web browser. It will fail, showing a black window with a colorbar. Adobe Acrobat Reader is needed to properly display the 3D content contained in these documents.  Once loaded in Adobe Acrobat Reader, the following message may appear, depending on your security settings: "3D content has been disabled. Enable this feature if you trust this document". From the Options menu, select "Trust this document…" and click anywhere in the document to cause a refresh of the graphics.


0 Comments

Workshop: Extract-based and In Situ Methods for HPC Enabled CFD

To be held August 3, 2017, in conjunction with the ASME Fluids Engineering Division Summer Meeting​ at the Hilton Waikoloa Village, Waikoloa, Hawaii

Workshop Objectives

This workshop will assess the state-of-the-art in CFD Visualization and Post-Processing in High Performance Computing (HPC) environments. Specifically the issues of data transfer from remote HPC resources, a comparison of remote graphics vs. extract methods, and the trend toward in situ methods. Real world examples of industrial CFD workflows will be presented that offer a path to greater fidelity and faster throughput enabling an increased use of unsteady simulation.

Presenters

Prof. Kozo Fujii, Tokyo University of Science/ ISAS-JAXA 
Revisit the Role of Research Graphics in HPC Post Processing

Yves-Marie Lefebvre, FieldView Product Chief, Intelligent Light
Unblocking Data Bottlenecks in HPC CFD with Extract-based Workflows

Steve Legensky, Founder and General Manager, Intelligent Light
In Situ: Real and in Production

Read presentation abstracts

Who should attend

Researchers and practitioners in aeroacoustics, combustion, LES and large scale CFD and CFD solver code technologists.

Thursday, August 3, 2017
Lunch for all FEDSM participants sponsored by Intelligent Light
Workshop from 1:30 to 4 PM

This event is free, but seating is limited and advanced registration is required.

To register, please send email to events@ilight.com with your full contact information and the number of people in your party. Location details will be provided with your registration confirmation.

0 Comments

Making Collaborative CFD Fly

"This method of aggregating data and standardizing the interrogation of the data, especially large scale, full 3D multiphyisics data is very, very necessary and very, very useful. This partnership is exciting and we are exploring the boundaries of what we can achieve with this."

Nathan Hariharan
Chair, AIAA Helicopter Hover Prediction Workshop

The world's helicopter community came together to study the complex physics of rotorcraft in hover using CFD. They needed a HPC collaboration hub that minimized researchers' efforts to collaborate in full 3D. We made it work for the AIAA Hover Prediction Workshop and we can make it work for you.

Join Dr. Nathan Hariharan, Chair, AIAA Helicopter Hover Prediction Workshop, Dr. Earl Duque, Manager of Applied Research at Intelligent Light, and Michael Senizaiz, ‎Chief Technology Officer at R Systems, as they explore how work done to support the Hover Prediction Workshop is yielding valuable lessons for any organization looking to streamline its CFD workflow.

Learn how Intelligent Light and R Systems provide the CFD expertise and on-demand HPC capability to make high-performance CFD workflow a reality:

  • Enable data sharing among many disparate members of the organization: CFD Engineers, Designers, Managers and Customers
  • Compare many different cases, even those with different meshes or from different solvers, etc.
  • Derive higher order features, like vortex path, automatically for comparison
  • Standardize the workflow so that input from many groups can be compared side by side, both visually and numerically
  • Utilize on-demand HPC resources to run automated post-processing workflows across all the datasets producing standardized outputs and enabling collaboration

A live Q & A with the speakers will follow the presentation.

When: Tuesday, May 9, 2017, 2:00 PM Eastern

0 Comments