Hello, A new project on the LCRC cluster has been requested. Please forward the information on to the LCRC Allocation sub-committee. Applicant's name: Yuepeng Zhang Applicant's institution: ANL Applicant's division: Energy Systems Division (ES) Project Name: WATER_EAM Project title: Application of high performance computing to manage water infrastructure networks Associated funding: This project is associated with a CRADA project currently under development and negotiation with California Water Service Company (Cal Water) Other Systems: Our collaborators at Cal Water has regular access to the company’s internal business network. The network is mainly comprised of dual-core and quad-core HP desktop computers, equipped with a variety of modeling and data analytical software. However, this network does not include any regular form of (or access to) high-performance computing (HPC) capabilities, except through project-specific requests and contract services; that is the basis for this request. Science: The primary science objectives for this project is to examine the intersection/overlap of big data and high performance computing capabilities, and long-range programmatic planning and optimization of water system infrastructure. As noted in the description below, specific topics will encompass code refinement and optimization within the context of HPC, code implementation, sensitivity analysis, and rich visualization. Project description: Infrastructure is the physical framework upon which the American economy operates, and the nation’s standard of living depends. As highlighted by recent reports issued by both the American Society of Civil Engineers (ASCE) and the American Water Works Association (AWWA), the aging of these physical assets represents one of the most critical technical, managerial, financial, and safety issues currently facing the nation. According to the most recent ASCE report, the cost of building new infrastructure to service increasing populations and expanded economic activity, as well as to maintain or rebuild existing infrastructure can be as high as one trillion dollars by 2020 and roughly five trillion dollars by 2040. Within the water sector alone, these funding shortfalls are expected to be $84 billion and $144 billion, respectively. Drawing upon these considerations, of note and significance is that enterprise asset management (EAM) programs can be a significant and effective form of mitigation against the above issues and cost. More specifically, asset management programs that incorporate asset criticality, risk, life-cycle analysis, grid optimization, and other advanced techniques, can produce significant insight into infrastructure management and improvements. Our current LCRC request is framed around addressing the computational requirements associated with comprehensive EAM programs. For example, a medium to large water system may have as much as 1,000 miles of water main, 50-200 pumps, 20-100 storage tanks, hundreds of line and control valves, etc. Applying criticality, risk, and optimization techniques to systems of this scale quickly becomes computationally unapproachable for most water utilities. When taken to full national scale, this unapproachability is amplified even further. The objectives of this application are multi-dimensional, and are intended to provide industry reflection and future guidance as motivated by suitably-chosen case studies. Specific topics will encompass code refinement and optimization within the context of HPC, code implementation, sensitivity analysis, and rich visualization. Collectively, these steps will help measure the associated technical, analytical, and procedural steps and improvements (“lessons learned”) that can be used as industry guidance for future and similarly situated EAM considerations. Programming model includes algorithmic design and use, combined with considerations of modularity (function, subroutine, and library calls). This approach largely springs from the modular and customizable design of the EPANET code/engine, combined with an objective to optimize any newly-developed or modified code. The main experiments planned for this work will be a relatively small number of example water network models that can be used for testing proposed computational frameworks, methods, and solutions. The water network models will be provided by the industry partner. No field-based or laboratory-based experiments are envisioned at this time. Given the somewhat pilot nature of the project, there will be no scalability issues anticipated at this stage. The software requirements for this project will be associated with EPANET, a public domain hydraulic network analysis engine, written by the USEPA. Supporting software will also be required as follows: - Standard MS Office Suite for simple data analysis and documentation - ANSI C Compiler - FORTRAN compiler - MATLAB and/or Mathematica for post-processing - Python or similar scripting language - ESRI ArcGIS or similar/compatible package that can read spatial data Regarding performance evaluation, no previous requests of this computational scale have ever been proposed by the industry partner. Thus, no specific performance benchmarks or metrics are available. Developing these benchmarks and metrics will be one of several deliverables produced by this project (combined iteratively with some code optimization). As noted in the core-hour request, this project envisions a total demand of slightly over 23,000 core-hours. While some initial testing will be required, such as code dimensionality and efficiency, most runs are expected to be complete within 24 hours or less. An estimated 30 runs will be required on 16 cores, 2 nodes. Code optimization and efficiency improvement will be one aspect of this project. For instance, it is not immediately known whether the EPANET code/solver is compatible with a parallel computing or other high-performance computing (HPC) framework. Thus, it is likely that some amount of code development, refinement, and debugging within an HPC framework will be required. Over 200 core-hours have been allocated to this task. The overall project team has four people, in the roles of framework and model development as well as code development and debugging. Two people will primarily submit job requests to the computing cluster. The Argonne PIs are Yuepeng Zhang and Kaizhong Gao. The industry contact is Jonathan Keck at the California Water Service Company. Industry partnership: California Water Service Company, San Jose, CA, USA Project URL: Requested allocation: 26220 Q1: 5414 Q2: 7846 Q3: 7776 Q4: 5184 Justification: Not Applicable; this request falls below 500,000 core-hours. Storage requirements: 1 TB is adequate; no additional storage resources are required. The requester has used undetermined amount hours of their initial startup project. In addition to approving an initial amount, please specify a Category and Subcategory for this project. For a list of the current selection of approved categories, please see: https://wiki.lcrc.anl.gov/wiki/Processes/Categories Once the Allocation committee has approved the project, please go to the Project Management page to create it: https://accounts.lcrc.anl.gov/projects.php Thank You, The LCRC Accounts System