(26c) Adaptive Meshing – The Future of Explosion Modeling

Computational modeling (CFD) approaches and capabilities have made great strides in combating the stigma that this type of modeling is too expensive and time intensive. This has mainly been overcome by the continued exponential growth in computing power. Yet, as our computational capabilities has grown, as has the demand for the number and complexity of problems that are simulated in a CFD study. Thus, we have traded hardware efficiencies for an increase in the number and type of models simulated and, in many cases, this has resulted in a net zero effect. In an effort to improve both the image of CFD as a cost effective and efficient tool and the value of our solutions, we must continue to look to new ways to improve our methods.

Despite these improvements, current approaches to vapor cloud explosion modeling with CFD still rely on traditional cartesian based mesh techniques in which a “knowledgeable” user is required to build suitable meshes. Meshing is a critical step in any CFD problem and can present a significant bottleneck. In a traditional approach an experienced user would follow the basic process of: 1) Review the problem, model, and critical questions intended to be answered to identify a base meshing strategy. This step relies heavily on user experience and knowledge of tools being employed.; 2) Construct a starting mesh which provides refinement in proper areas, abides by all meshing rules specific the solvers being used (interface transitions, geometry alignments, stretching, etc.).; and 3) Conduct critical mesh sensitivities analyses were refinement and mesh rules are tested and verified.

This meshing process may have to be conducted again and again, for various vapor cloud locations and shapes, when changes to geometry are made, new ignition locations, etc. But, in contrary to this process, there are alternative approaches – adaptive meshing. Adaptive meshing routines (AMR) can be used to auto-refine meshes around geometry and critical solution effects. Much like Artificial Intelligence, AMRs can be guided by simple rule sets and allow the model to make live mesh changes as necessary during the simulation. The end result is the ability to reduce meshing requirements, setup time, while maintaining and in many cases improving the solution accuracy.

The goal of this paper will be to provide a look into what adaptive meshing is, how it works, and show how it can be used to increase our modeling efficiency (i.e. reducing costs and time). A sample of process modules will be used to demonstrate the capability and provide visuals of the AMR effect.

Computational modeling (CFD) approaches and capabilities have made great strides in combating the stigma that this type of modeling is too expensive and time intensive. This has mainly been overcome by the continued exponential growth in computing power. Yet, as our computational capabilities has grown, as has the demand for the number and complexity of problems that are simulated in a CFD study. Thus, we have traded hardware efficiencies for an increase in the number and type of models simulated and, in many cases, this has resulted in a net zero effect. In an effort to improve both the image of CFD as a cost effective and efficient tool and the value of our solutions, we must continue to look to new ways to improve our methods.

Despite these improvements, current approaches to vapor cloud explosion modeling with CFD still rely on traditional cartesian based mesh techniques in which a “knowledgeable” user is required to build suitable meshes. Meshing is a critical step in any CFD problem and can present a significant bottleneck. In a traditional approach an experienced user would follow the basic process of: 1) Review the problem, model, and critical questions intended to be answered to identify a base meshing strategy. This step relies heavily on user experience and knowledge of tools being employed.; 2) Construct a starting mesh which provides refinement in proper areas, abides by all meshing rules specific the solvers being used (interface transitions, geometry alignments, stretching, etc.).; and 3) Conduct critical mesh sensitivities analyses were refinement and mesh rules are tested and verified.

This meshing process may have to be conducted again and again, for various vapor cloud locations and shapes, when changes to geometry are made, new ignition locations, etc. But, in contrary to this process, there are alternative approaches – adaptive meshing. Adaptive meshing routines (AMR) can be used to auto-refine meshes around geometry and critical solution effects. Much like Artificial Intelligence, AMRs can be guided by simple rule sets and allow the model to make live mesh changes as necessary during the simulation. The end result is the ability to reduce meshing requirements, setup time, while maintaining and in many cases improving the solution accuracy.

The goal of this paper will be to provide a look into what adaptive meshing is, how it works, and show how it can be used to increase our modeling efficiency (i.e. reducing costs and time). A sample of process modules will be used to demonstrate the capability and provide visuals of the AMR effect.