Project Title/ID Number Data Management for OpenSees Simulations—4152002
Start/End Dates 10/1/02—9/30/03
Project Leader Kincho Law (Stanford/Faculty)
Team Members Jun Peng (Stanford/Post Doc)
Project goals and objectives

This research project, in collaboration with researchers at PEER center, aims to develop a software platform that assists the application of performance-based earthquake engineering (PBEE) assessment and design methodologies. The project focuses on supporting testbed applications, data and project management, and archival capabilities. The objective is to provide a persistent storage for the analysis results for OpenSees, to support project management, and to archive testbed projects data. The data management system is to be designed to manage information related to seismic inputs, simulation models, and simulation results so that performance metric, demand parameters, and damage measures can be derived from the simulation results. Finally, an easy-to-use user interface is to be developed so that data can be queried via a web interface or an application program such as MATLAB.

Role of this project in supporting PEER’s vision

We envision that the data and project management infrastructure developed in this project will become a common tool for which PEER researchers can use for the testbed applications. This work involves an internet-enabled architecture that would allow users to access the platform on-line as well as allow researchers to develop and to integrate their works with OpenSees. With the Internet-enabled framework, users and developers from anywhere can have access to the PEER’s platform for earthquake analysis and demand assessment. In addition, a data management system is integrated with the framework for archiving simulation results and other testbed-related information so that researchers can share their results and models.

Methodology employed
  • Commercial and publicly available database systems are integrated with OpenSees to provide persistent data storage. Both Oracle and MySQL are tested with the prototype system.
  • The Internet is utilized as a data delivery vehicle. Users can access OpenSees, simulation results, and testbed-related data from web browsers or other application programs. The architecture of the data access system is shown in Figure 1.
  • Instead of using the traditional data storage method of dumping all the interim and final analysis results into file systems, we propose an alternative approach to store only selected data in the database systems. One example of selected data storage strategy is SASI (Sampling At a Specified Interval) for nonlinear analyses, where the selected state information, instead of the entire analysis results, is saved in a specified interval (e.g. every 10 steps or other appropriate number of steps). Compared with traditional data storage method, the total amount of required storage space is substantially reduced. Additionally, significant computational savings can be achieved when data need to be retrieved and recomputed using OpenSees.
  • The internal data structure of OpenSees is organized in an object-oriented fashion, thus cannot be easily mapped into a relational database. Instead of adopting the typical approach of creating a table for each type of objects that needs to be stored, we choose to use matrix-type data (ID, Vector, and Matrix) to represent object state. The advantage of this approach is that new classes can be introduced without the creation of new tables in the relational database. In other words, the database schema can be defined statically. The database schema is shown in Figure 2.

Figure 1. Data Access System Architecture
Larger View


Figure 2. Database
Larger View

Brief description of past year’s accomplishments and more detail on expected Year 6 accomplishments

This research project has focused on the data and project management aspects of OpenSees. The goal was to provide the facilities that can support testbed applications. Thus, the first task was to develop an interface to link OpenSees with database systems; in this work, both Oracle 8i and MySQL systems have been employed for the backend data storage. This task involved the design of a database schema and the design of a selective data storage scheme for OpenSees. After the database schema was defined, implementation efforts included building a communication channel between OpenSees and the Oracle and MySQL database systems, designing a data query language, and implementing user interfaces. In order to support the collaborative efforts of PEER researchers, a project management mechanism was provided and suitable interfaces were introduced. In summary, we have accomplished the following tasks for the data management and archiving:

  1. Data and Storage Structure
  2. Internal and External Data Representation
  3. Data Input and Query Processing
  4. Project Management Support

The year 5’s research was focused on implementation, and year 6 is expected to be concentrated on calibration and deployment. We will collaborate with the researchers from UCSD to test the data management system on the Humboldt Bay Middle Channel Bridge project. Both the data archival and data access capabilities of the system will be thoroughly tested. After the data management system is fine-tuned, it can be applied to other testbeds.

Other similar work being conducted within and outside PEER and how this project differs

A project entitled “ParCYCLIC: Internet-Enabled Simulation of Earthquake Liquefaction Response on Parallel Computers”, which was approved and funded by NSF, is a joint effort with Prof. Ahmed Elgamel of UCSD. The focus of the project was on the development of sparse matrix solution algorithms, parallel algorithms, and distributed program development. The results of the project will be very useful towards the next phase of OpenSees. First, it will enhance future development of OpenSees; especially the developed parallel solver can greatly improve the performance of OpenSees simulations. Furthermore, it will enhance further incorporation of geotechnical analysis capabilities into the OpenSees platform.

Plans for Year 7 if this project is expected to be continued

We plan to continue the current work to extend the OpenSees platform in the following three aspects:

  1. Project Management. The current project system will be extended and improved. The project management system can be employed to be the central archive of testbed projects data with version control and access control capabilities.
  2. Lanczos Eigensolver. A robust Lanczos procedure has been developed and implemented with OpenSees, and is ready to be released. The Lanczos procedure can greatly improve the performance of eigen-analysis for OpenSees.
  3. Parallel and Distributed Computing. We intend to integrate a parallel solver with OpenSees to improve the solution of large-scaled engineering problems, particularly suitable for geotechnical analyses. The parallel solver is already developed and is currently being tested by the researchers at UCSD.
Describe any instances where you are aware that your results have been used in industry

Expected milestones

First Quarter:

  • Design of database schema for project management purpose.
  • Integration of database (both Oracle and MySQL) with OpenSees.

Second Quarter:

  • Design and implement a data query language.
  • Develop an online access interface using web technology.
  • Develop an interface to allow external application program (specifically MATLAB) to access the simulation results from OpenSees.

Third Quarter:

  • Evaluate the data management schemes by conducting a test with a specific testbed application project (Humboldt Middle-Channel Bridge project).
  • Re-design and re-implement the data management interface, if necessary.

Fourth Quarter:

  • Deploy the data management framework to support other testbed applications.
  • A preliminary design of database schema has been finished and applied to both Oracle 8i and MySQL database systems.
  • A data query language has been defined and a string parser has been implemented according to the defined data query language.
  • Both web-based interface and MATLAB-based interface to the data management system have been constructed and tested.
  • A project management system with version control and access control is being implemented and tested.