This document describes the 2002 SNL Accelerated Strategic Computing Initiative (ASCI) Applications Software Quality Engineering (SQE) Assessment and the assessment results. The primary purpose of the assessment was to establish the current state of software engineering practices within the SNL ASCI Applications Program.
Reliability Engineering and System Safety
The US Department of Energy (DOE) Waste Isolation Pilot Plant (WIPP), located in southeast New Mexico, is a deep geologic repository for the permanent disposal of transuranic waste generated by DOE defense-related activities. Sandia National Laboratories (SNL), in its role as scientific advisor to the DOE, is responsible for evaluating the long-term performance of the WIPP. This risk-based Performance Assessment (PA) is accomplished in part through the use of numerous scientific modeling codes, which rely for some of their inputs on data gathered during characterization of the site. The PA is subject to formal requirements set forth in federal regulations. In particular, the components of the calculation fall under the configuration management and software quality assurance aegis of the American Society of Mechanical Engineers (ASME) Nuclear Quality Assurance (NQA) requirements. This paper describes SNL's implementation of the NQA requirements regarding configuration management. The complexity of the PA calculation is described, and the rationale for developing a flexible, robust run-control process is discussed. The run-control implementation is described, and its integration with the configuration-management system is then explained, to show how a calculation requiring 37,000 CPU-hours, and involving 225,000 output files totaling 95 GB, was accomplished in 5 months by two individuals, with full traceability and reproducibility.
Reliability Engineering and System Safety
The US Department of Energy (DOE) Waste Isolation Pilot Plant (WIPP), located in southeast New Mexico, is a deep geologic repository for the permanent disposal of transuranic waste generated by DOE defense-related activities. Sandia National Laboratories (SNL), in its role as scientific advisor to the DOE, is responsible for evaluating the long-term performance of the WIPP. This risk-based Performance Assessment (PA) is accomplished in part through the use of numerous scientific modeling codes, which rely for some of their inputs on data gathered during characterization of the site. The PA is subject to formal requirements set forth in federal regulations. In particular, the components of the calculation fall under the configuration management and software quality assurance aegis of the American Society of Mechanical Engineers (ASME) Nuclear Quality Assurance (NQA) requirements. This paper describes SNL's implementation of the NQA requirements regarding software quality assurance (SQA). The description of the implementation of SQA for a PA calculation addresses not only the interpretation of the NQA requirements, it also discusses roles, deliverables, and the resources necessary for effective implementation. Finally, examples are given which illustrate the effectiveness of SNL's SQA program, followed by a detailed discussion of lessons learned.