An interactive numerical data processing and visualization tool for large simulations of ground motions
|Affiliation||Civil and Environmental Engineering|
|Project title||An interactive numerical data processing and visualization tool for large simulations of ground motions|
|Background||One of the most constant pursuits of geotechnical earthquake engineers is that of predicting the response of soil masses and any associated infrastructure to earthquake loading. Soil deposits are highly spatially variable, expensive to test in-situ, and challenging to sample. At the same time, the greatest challenge comes from earthquakes which are unpredictable in nature and impose high demands on our geosystems. The approach we follow is to build experience as well as robust numerical tools with which we run hundreds of analyses of our systems in 1, 2 and the full 3 dimensions. Interestingly enough, we then become limited in our ability to process and interpret all the data these produce. Unless we have a tool that can readily process large ground motion datasets from analyses we cannot move forward.|
|Description||Students working on this project will assist in the design and implementation of an interactive tool capable of acting as a post-processing and visualization unit, ideally capable of employing Machine Learning algorithms to decipher the secrets of massive amounts simulated ground motions. This tool will be used by graduate student researchers at UC Davis as well as researchers and practitioners nationally and internationally. We are willing to support the team as much as needed along the way as well as contribute the discipline-specific expertise needed. Ideally the tool could be an extensible application such that we can implement newer findings later. The tool should be able to visualize the data in a flexible and interactive format and we expect that it will be tested and improved with our feedback. We already do some of the desired functions via Matlab and Fortran codes but they are inefficient, not integrated, and not-sustainable.|
|Deliverable||We envision a working prototype of the tool alongside with a manual for users as well as the tool’s source in Github. The resulting program could be stand-alone (on Mac and/or PC).
1) Ability to process massive data (most likely running several input motions and getting numerous results)
2) Basic ability to see and get data from single or multiple records, e.g. for selecting ground motions for some analysis (filtering, baseline correcting, Fourier Amplitude Spectra, response spectra, power spectra, phase spectra, phase distribution, CAV, Arias intensity, etc.). In this scenario, we want to be able to plot single or multiple records at the time for comparison purposes and get the data to generate those plots in other software, and get summary of parameters.
3) Ability to check results from a site response analysis (1D, 2D or 3D, main point is we will have some input motions and numerous responses). In this scenario, we have the additional need to compute parameters that involve input/output motions: transfer functions, and phase shifts.
4) In both of the above we would welcome the ability to apply filters in order to be able to isolate motions based on criteria e.g. peak ground acceleration higher than 0.5.
5) Ability to generate spatially variable motions very useful for geographically distributed systems like pipelines. For this, we have a Fortran code available and would like to translate it.
|Skill set desirable||Experience with Python, MatLab or a similar system for input and analysis of continuously measured data. The ideal team will have experience with reading and understanding Fortran code (but not writing Fortran), database APIs (retrieving data), pre-processing data, and data visualization tools. Signal processing skills and knowledge of machine learning algorithms are also desirable but not required.|
|Client time availability||30-60 min weekly or more|
|IP requirement||Open source project|