Kumar Mahinthakumar

The Department, in conjunction with Dr. Kumar Mahinthakumar acting in the capacity of a modeling advisor, will conduct a review of six groundwater flow and transport (F-T) models that are being developed by Dr. Falta at Clemson University to help select basin-closure options and evaluate groundwater remediation alternatives. The Division of Water Resources (DWR) will lead a small review team (probably two representatives from DWR and a representative from the Division of Waste Management [DWM]) that will: 1) work with Dr. Mahinthakumar to review the models, 2) request information from Dr. Falta as/if needed, and 3) make recommendations to the Department regarding the quality of the models for their intended use. Both Dr. Mahinthakumar and the Department??????????????????s review team will have the capability to run the models and conduct sensitivity analyses. After initial discussions associated with each model, the groundwater model advisor will provide the Department/review team with a technical memorandum for that model. Assistance may be requested from Dr. Falta to initiate model runs so that the Department??????????????????s time is spent on the review process itself rather than dealing with software.

Recent surveys of the national water industry warn of looming costs for capital improvements for drinking water systems in the coming decades [1,2]. One AWWA report estimates that over $1 trillion are needed over the next 25 years, and $1.7 trillion over the next 40 years; about half of this investment would cover the renewal and replacement of aging pipes, and half would pay for system expansions to accommodate population changes [1]. At the same time, SCADA and sensor systems, monitoring and modeling software are facilitating real-time operational decision making in water utilities as never before. These short-term operational decisions, as well as capital improvements such as system expansions, equipment or technology upgrades, and price structures, affect system performance with respect to long-term master planning goals such as system resilience, cost and resource sustainability. Currently, no systematic decision support methodologies exist to optimize the timing of the needed investments over the 25 or 40 year horizon, assessing the criticality of these decisions for overall system vulnerability and resource optimization. The proposed project aims to build a resilience modeling framework using a water utility??????????????????s in-house data and models to bridge the gap between short-term operations and the medium- and long-term decisions that influence master planning objectives such as system resilience. The objectives of this research are: develop a general purpose resilience modeling framework that integrates computational tools and data to expand the optimization capacity of available data and infrastructure component models from short-term operational to long-term planning horizons; build, demonstrate and test the modeling framework using a case study water utility??????????????????s data and models to provide in-house decision support for optimal timing and balance between short-term performance and long-term objectives; and develop an open-source water system resilience library that is compatible with a standard resilience modeling language and water infrastructure system modeling tools.

Leakage in bulk water pipelines is a major problem for many water utilities as it leads to significant economic losses and cause service disruptions threatening public safety. Most utilities currently employ intrusive methods based on acoustic or infrared signals that can be expensive, time consuming, and require trained personnel. Non-intrusive methods that currently exist require significant computational resources and are not suitable for real-time application. We have developed extremely fast leakage detection algorithms that are suitable for real-time application. These methods rely on a hydraulic model and routine pressure measurements and were developed as part of an NSF project funded from 2011-2016. In partnership with DC Water, Lakewood City (CA), TAGO (an Internet of Things (IoT) company based in Raleigh, NC), and Citilogics (a smart water analytics company based in Cincinnati, OH) we will conduct a proof-of-concept study for isolated sections of each utilities?????????????????? water networks. Data from existing pressure sensors as well as new sensors installed at strategic locations in the network will be used to validate our algorithms using experimentally simulated leakage scenarios. Citilogics will provide necessary expertise and software for processing the real time data suitable for inputs into the hydraulic model from AMI, SCADA, and CMMS systems. TAGO will build an IoT framework for acquiring the pressure data (from sensors) in real-time and for integrating our algorithms and associated filters on cloud resources. Partner utilities will install the pressure sensors as needed and execute the experiment. NCSU will design the experiment, build the hydraulic model and the leak detection analytics, test the IoT architecture, and evaluate the potential for commercialization.

Continually increasing water demand (due to population growth) and fuel costs threaten the reliability of water and energy systems and also increase operational costs. In addition, both natural climatic variability and the impacts of global climate change increase the vulnerability of these two systems. For instance, reservoir systems depend on precipitation; whereas power systems demand depend on mean daily temperature. Currently, these systems use seasonal averages for their short-term (0-3 months) management, which ignores uncertainty in the climate, thereby resulting in increased spillage and reduced hydropower. While seasonal climate forecasts contain appreciable levels of skill over parts of the US in both winter and summer, the uptake of these forecasts for water and energy systems management has been limited due to lack of a coherent approach to assimilate probabilistic forecasts into management models. We systematically analyze various scenarios that aim at improving the performance of these systems utilizing the multimodel climate forecasts and a high performance computing (HPC) framework.

The Department, in conjunction with Dr. Kumar Mahinthakumar acting in the capacity of a modeling advisor, will conduct a review of six groundwater flow and transport (F-T) models that are being developed by Dr. Falta at Clemson University to help select basin-closure options and evaluate groundwater remediation alternatives. The Division of Water Resources (DWR) will lead a small review team (probably two representatives from DWR and a representative from the Division of Waste Management [DWM]) that will: 1) work with Dr. Mahinthakumar to review the models, 2) request information from Dr. Falta as/if needed, and 3) make recommendations to the Department regarding the quality of the models for their intended use. Both Dr. Mahinthakumar and the Department??????????????????s review team will have the capability to run the models and conduct sensitivity analyses. After initial discussions associated with each model, the groundwater model advisor will provide the Department/review team with a technical memorandum for that model. Assistance may be requested from Dr. Falta to initiate model runs so that the Department??????????????????s time is spent on the review process itself rather than dealing with software.

The main objective of the proposed research is to understand, identify and quantify uncertainties related to the freshwater sustainability under near-term climate change and population growth by incorporating adaptive responses, feedbacks as well as the needs of hydro-ecological systems and human-environmental systems through a cross-regional synthesis. The specific tasks associated with this objective are: 1) Reduce the uncertainty in targeted hydro-ecological variables under near-term climate change by combining projections from multiple GCMs (from AR5), regional climate models (from NAARCAP) and hydrological models (VIC, SWAT and MODFLOW) 2) Ingest the hindcasts and future projections (i.e., time series) of the targeted variables in reservoir models, groundwater models and ecological models to quantify the resilience and vulnerability of water infrastructure systems and ecological systems over the target basins 3) Identify key policy interventions, supply augmentation and demand reduction measures that will ensure freshwater sustainability under projected population growth and climate change 4) Incorporate the identified consumer feedbacks and societal adaptations with the water infrastructure models through an agent-based model 5) Perform an integrative cross-regional analyses to identify critical tipping points and feedback mechanisms that will reduce the uncertainty in freshwater sustainability under near-term climate change and population growth

NCSU??????????????????s CCEE department will work with Sensus and Town of Cary (ToC) in designing and executing a series of experiments to evaluate NCSU optimization algorithms (Phases 1-3). Phases 1-3 are expected to take 1 year and will require 1 graduate student from the NCSU CCEE department. Depending on the success of these experiments, additional phases may be carried out with the goal of developing software modules for efficient management and real time operation and control of urban water distribution systems. Activities beyond the experimental phase will be contingent on developing a successful intellectual property agreement between NCSU and Sensus.

Urban water distribution systems (WDS) are prone to leaks, deterioration, and contaminant intrusion. The risks associated with these conditions can be spatially diverse due to flow conditions, network characteristics, and external factors. The key to long term sustainability of these systems is to identify high risk regions and develop sound operational procedures and preventive maintenance plans. This project will develop an adaptive leak detection and contaminant intrusion framework that utilizes real time pressure, flow, and water quality data driven by a high performance simulation optimization engine. The research will also develop a risk analysis capability that could be used to analyze economic and public health risks associated with gradual leaks and contaminant intrusion occurring in day to day operations. In a recently completed project funded by NSF, the research team has developed an adaptive high performance simulation-optimization engine and associated optimization methodologies for contaminant source identification and sampling design in water distribution systems. While the proposed work will build on this work to an extent, a number of new developments are planned including: (a) a new simulation-based leak detection and contaminant intrusion detection methodology that uses routinely measured pressure, flow and water quality measurements (e.g., from a SCADA system), (b) a methodology for incorporating spatially varying macro indicators (e.g., pipe attributes) to improve the search process, (c) a methodology for incorporating demand uncertainty in leak detection, and (d) a risk assessment methodology that targets economic losses from leaks and public health risks from contaminant intrusion. The developed framework could be used by decision makers to make operational decisions and to develop long term maintenance and expansion plans. The system can also be used by decision makers to evaluate the risk reduction potential of different response and maintenance actions. In collaboration with a local utility, the leak detection and risk assessment framework developed through this research will be applied and validated using data from a mid-sized urban area. The major objectives are to: (a) develop a parallel simulation-optimization leak detection and contaminant intrusion methodology that uses routinely measured pressure and water quality data; (b) develop a Markov Chain Monte Carlo (MCMC) methodology to incorporate prior information and demand uncertainty into the leak and contaminant intrusion detection framework; (c) develop a methodology for evaluating economic and public health risks associated with leaks and contaminant intrusion; (d) test and evaluate the computational framework and the associated components for a mid- sized urban area; and (e) integrate and disseminate the research results in classroom teaching for college students, as well as continuing education and short courses for practitioners. Intellectual Merit. While complementing other related NSF funded efforts, the proposed research hinges on a number of paradigm shifting and transformational developments, including the conjunctive use of routine operational measurements and external factors for generating leak and contaminant intrusion maps in real time, incorporating uncertainties in a probabilistic framework, and the use of high performance computing technologies to enable real time computation. Broader Impact. Besides addressing one of the nation?s critical infrastructure security priorities in a real setting, the project contributes in general to the development of adaptive simulation methods and optimization algorithms that can harness high end computing resources. The computational framework developed through this proposal will set the stage for using emerging automated data collection systems for real time characterization.

The overall goal of this SciDAC-2 project is to develop and maintain a world class Performance Engineering Research Institute (PERI) that will address the growing challenges of achieving good performance of grand challenge applications on highly complex emerging high-end computing (HEC) systems. This center will function on the basis of a tripartite research plan encompassing: (1) performance modeling and prediction; (2) automatic performance optimization; and (3) performance engineering of high profile applications. The work performed in this proposal will primarily address the third component with appropriate interactions with the first two components. The major thrust of this third major component, application engagement, is to maintain direct interactions with SciDAC applications through passive, active, or ?tiger team? liaisons. Approximately 30% of the total PERI resources are devoted to this activity, including tiger teams that will focus on particular codes.

This project entails the development and optimization of software algorithms that read/write data sets from/to parallel file systems in an efficient and scalable manner. In this context, scalable means that the simulators read/write performance does not degrade significantly as the number of cores grows. It is accepted (and well understood) that parallel I/O will not scale as efficiently as the parallel scientific algorithms in the simulator due to hardware limitations beyond the scope of the project. The products generated by the project will include software code composed of functions and/or subroutines that read and write data sets in parallel and perform the read/write operations associated with simulation checkpoint/restart.

The overall objective of this project is to develop a set of tools to assist design engineers in developing effective, reasonably efficient systems for distributing aqueous amendments for in situ treatment of groundwater contaminants. At this time, the primary applications for the tools will be for design of in situ chemical oxidation systems using permanganate and in situ anaerobic bioremediation systems using soluble substrates and emulsified oil. However, as technology evolves, this general approach should be applicable to distribution of other aqueous amendments. Specific objectives of this project are listed below. 1. Use currently available numerical models to understand the effects of site conditions (e.g. permeability, contaminant distribution, site heterogeneity) and design variables (location of wells, injection rates, volumes, amount of reagent, etc.) on reagent distribution and associated contact efficiency. Develop simple design curves relating reagent distribution efficiency to amount of fluid/reagent injected. To the extent possible, present the results in a non-dimensional form (e.g. ratio of reagent injected to theoretical demand). 2. Develop a simple, spreadsheet-based design tool to assist junior to mid-level engineers in planning injection systems for in situ aquifer treatment with MnO4-, soluble substrates and emulsified oil. This design tool will allow designers to evaluate the effect of different design variables (well spacing, amount of reagent, injection rate, etc.) on remediation system cost and expected performance. Different worksheets will be developed for each of the major design alternatives (e.g. type of reagent and type of injection system). Experienced users who have already compiled the input data for their site (e.g. permeability, natural oxidant demand, contaminant concentrations) should be able to develop a preliminary design for one alternative (e.g. 5-spot injection grid with MnO4-) in about an hour. 3. Develop materials to train junior to mid-level engineers with no modeling experience in the use of the design tool. These materials will consist of: (a) a detailed guidance manual; and (b) a series of 15 to 30 minute PowerPoint tutorials that walk new users through each worksheet. Chemical oxidation and anaerobic bioremediation are being used to treat thousands of DoD and private sites. At some sites, the process works very well, resulting in substantial reductions in contaminant concentration and mass. However at too many sites, the remediation process does not meet cleanup objectives. We believe the most common problem is poor delivery of the chemical reagent to the treatment zone. If we can provide design engineers, with a simple, easy to use tool for planning aqueous injection systems, we can substantially improve the performance and reduced costs at DoD sites.

Effective threat management of urban water distribution systems, a critical civil infrastructure, is a challenging problem due to the dynamic nature of data and management requirements. Recent home-land security concerns have underscored the need for a dynamic feedback and control system that will automatically adapt to ever changing conditions and also devise optimal threat management strategies. Effective threat management requires that the source be characterized and contained and strategic shutdown and decontamination procedures implemented. An automated sensor network and an infrastructure that can use this data and control the data collection process is an important if not essential component in this process. How these systems are managed on a day to day basis and spontaneous threat management strategies will have profound socio-economic implications that may affect local city authorities as well as consumers. Adaptive management strategies will enable dynamic control of valves and pumps to control flow to meet demand conditions as well as to respond to accidents such as leaks and breakages. This in turn will lead to reliable and secure systems with increased operational life of various components (pipes, pumps, valves, sensors etc.). While some current systems have a level of automation in the way the major valves and pumps are controlled, they mostly rely on data provided by static sensor networks and do not provide any feed back to the sensors. This multi-disciplinary DDDAS proposal will attempt to develop a cyberinfrastructure system that will both adapt and control changing data, model, computer resources, and management needs.

The supplement will increase minority participation in the project by supporting an african american PhD student. This student will develop parallel hybrid optimization approaches to enable realtime processing of environmental characterization problems. This student will also participate in various instructional and outreach activities to reinforce diversity as a resource for enriching education in environmental engineering.

This work plan outlines North Carolina State University's participation in the performance evaluation SciDAC project entitled "PERC-2, High-End Computer System Performance: Science and Engineering". The project involves 8 major institutions: LBNL, LLNL, ORNL, ANL, UTK, UNC-CH, UMD, and SDSC. The NCSU portion is being subcontracted through ORNL (Dr. Patrick Worley, Technical Leader of the project for Applications and Performance Evaluation). NCSU activities under PERC-2 are a continuation of PERC-1 activities which lasted from 10/24/02 to 09/30/04. The total duration of this PERC-2 is 2 years but the NCSU subcontracts will be awarded on an annual basis with a separate subcontract for each year. This request is for the first year of PERC-2 from 10/01/04 to 9/30/05. The progress for the first 2 years of PERC-1 was detailed in two final reports submitted to ORNL. A final report for the 3rd year of PERC-1 is in preparation and will be submitted in October 2004. The results are also posted on a regular basis on an internal web site.