Visual Decision Support Tool for Supporting Asset Management Performance, Risk, and Cost Analysis

Transcription

1 Infrastructure Visual Decision Support Tool for Supporting Asset Management Performance, Risk, and Cost Analysis IN COLLABORATION WITH GWINNETT COUNTY DEPARTMENT OF WATER RESOURCES Co-published by

2 INFR5R12 VISUAL DECISION SUPPORT TOOL FOR SUPPORTING ASSET MANAGEMENT PERFORMANCE, RISK, AND COST ANALYSIS IN COLLABORATION WITH GWINNETT COUNTY DEPARTMENT OF WATER RESOURCES by: Bradley Jurkovac GHD, Inc. Helena Alegre Sergio T. Coelho LNEC 2015

3 The Water Environment Research Foundation, a not-for-profit organization, funds and manages water quality research for its subscribers through a diverse public-private partnership between municipal utilities, corporations, academia, industry, and the federal government. WERF subscribers include municipal and regional water and water resource recovery facilities, industrial corporations, environmental engineering firms, and others that share a commitment to cost-effective water quality solutions. WERF is dedicated to advancing science and technology addressing water quality issues as they impact water resources, the atmosphere, the lands, and quality of life. For more information, contact: Water Environment Research Foundation 635 Slaters Lane, Suite G-110 Alexandria, VA Tel: (571) Fax: (703) werf@werf.org This report was co-published by the following organization. IWA Publishing Alliance House, 12 Caxton Street London SW1H 0QS, United Kingdom Tel: +44 (0) Fax: +44 (0) publications@iwap.co.uk Copyright 2015 by the Water Environment Research Foundation. All rights reserved. Permission to copy must be obtained from the Water Environment Research Foundation. Library of Congress Catalog Card Number: IWAP ISBN: This report was prepared by the organization(s) named below as an account of work sponsored by the Water Environment Research Foundation (WERF). Neither WERF, members of WERF, the organization(s) named below, nor any person acting on their behalf: (a) makes any warranty, express or implied, with respect to the use of any information, apparatus, method, or process disclosed in this report or that such use may not infringe on privately owned rights; or (b) assumes any liabilities with respect to the use of, or for damages resulting from the use of, any information, apparatus, method, or process disclosed in this report. GHD Inc., LNEC. The research on which this report is based was developed, in part, by the United States Environmental Protection Agency (EPA) through Cooperative Agreement No. CR with the Water Environment Research Foundation (WERF). However, the views expressed in this document are not necessarily those of the EPA and EPA does not endorse any products or commercial services mentioned in this publication. This report is a publication of WERF, not EPA. Funds awarded under the Cooperative Agreement cited above were not used for editorial services, reproduction, printing, or distribution. This document was reviewed by a panel of independent experts selected by WERF. Mention of trade names or commercial products or services does not constitute endorsement or recommendations for use. Similarly, omission of products or trade names indicates nothing concerning WERF's or EPA's positions regarding product effectiveness or applicability. ii

4 About WERF The Water Environment Research Foundation, formed in 1989, is America s leading independent scientific research organization dedicated to wastewater and stormwater issues. Throughout the last 25 years, we have developed a portfolio of more than $134 million in water quality research. WERF is a nonprofit organization that operates with funding from subscribers and the federal government. Our subscribers include wastewater treatment facilities, stormwater utilities, and regulatory agencies. Equipment companies, engineers, and environmental consultants also lend their support and expertise as subscribers. WERF takes a progressive approach to research, stressing collaboration among teams of subscribers, environmental professionals, scientists, and staff. All research is peer reviewed by leading experts. For the most current updates on WERF research, sign up to receive Laterals, our bi-weekly electronic newsletter. Learn more about the benefits of becoming a WERF subscriber by visiting Visual Decision Support Tool for Supporting Asset Management Performance, Risk, and Cost Analysis iii

5 ACKNOWLEDGMENTS This research is made possible through the sponsorship of WERF and U.S. EPA. In particular, the support, assistance and contributions from Walter Graf, WERF Program Director - Infrastructure Management, and from the Project Steering Committee members, were very important for the success of the project. Special thanks are due to the team from Gwinnett County Department of Water Resources staff, who provided all data necessary and collaborated actively in every stage of development. Suggestions and comments made were crucial to focus the research, to tailor the new software modules to the U.S. context and actual utility needs, and to assess results. Finally, authors and project team acknowledge sponsors, partner organizations and team members of concurrent projects AWARE-P ( igpi ( and TRUST ( Research Team Principal Investigator: Bradley Jurkovac GHD Inc. Sérgio T. Coelho Helena Alegre LNEC Project Team: Diogo Vitorino Addition David Kerr Crystal McNeely Varun Sekar GHD Inc. Rita Salgado Brito João Paulo Leitão Pedro Santos Maria do Céu Almeida Maria Adriana Cardoso Luís Mesquita David LNEC iv

6 WERF Project Subcommittee Members Barry J. Buchanan, P.E. Buchanan and Associates Wayne Green, P.E. Regional Municipality of York James Hall, P.E. Advanced Engineering & Controls Lewis Rossman (Retired) U.S. Environmental Protection Agency Andrew Wood, Ph.D., P.E. Metro Vancouver Innovative Infrastructure Research Committee Members Stephen P. Allbee (Retired) Daniel Murray Michael Royer U.S. Environmental Protection Agency Traci Case Water Research Foundation Peter Gaewski, MS, P.E. (Retired) Tata & Howard, Inc. Kevin Hadden Orange County Sanitation District David Hughes American Water Kendall M. Jacob, P.E. Cobb County Jeff Leighton City of Portland Water Bureau Steve Whipp (Retired) United Utilities North West Walter L. Graf, Jr. Water Environment Research Foundation Amit Pramanik, Ph.D., BCEEM (IIRC Chair) Water Environment Research Foundation Water Environment Research Foundation Staff Director of Research: Amit Pramanik, Ph.D., BCEEM Program Director: Walter L. Graf, Jr. Visual Decision Support Tool for Supporting Asset Management Performance, Risk, and Cost Analysis v

7 ABSTRACT AND BENEFITS Abstract: Managing urban water infrastructures faces the challenge of jointly dealing with assets of diverse types, useful life, cost, ages and condition. Service quality and sustainability require sound long-term planning, well aligned with tactical and operational planning and management. In summary, the objective of an integrated approach to infrastructure asset management is to assist utilities answer the following questions: Who are we at present? What service do we deliver? What do we own? Where do we want to be in the long-term? How do we get there? The AWARE-P approach ( offers a coherent methodological framework and a valuable portfolio of software tools. It is designed to assist water supply and wastewater utility decision-makers in their analyses and planning processes. It is based on a Plan-Do-Check-Act process and is in accordance with the key principles of the International Standards Organization (ISO) standards on asset management. It is compatible with, and complementary to WERF s SIMPLE framework. The software assists in strategic, tactical, and operational planning, through a non-intrusive, web-based, collaborative environment where objectives and metrics drive IAM planning. It is aimed at industry professionals and managers, as well as at the consultants and technical experts that support them. It is easy to use and maximizes the value of information from multiple existing data sources, both in data-rich and data-stressed environments. This project aimed at testing, validating, and tailoring the AWARE-P approach and software in the U.S. wastewater services context; developing new open-source software tools. The project produced: A portfolio of tools for the Advanced Water Management and Rehabilitation-Portugal (AWARE-P) software validated at Gwinnett County Department of Water Resources (GCDWR), the main pilot case, and at two other U.S. cases, where a large amount of GIS and inspection data from a major utility, and GIS and operational data from a small utility were extensively explored; A new, enhanced version of the tool to assess infrastructure value and support long-term reinvestment policies; and A new set of tools directed at data mining and condition-based inspection analysis and prediction. Benefits: Promotes an integrated perspective in Asset Management planning. Provides a software platform that can act as an integrator for established as well as new assessment methods. Provides communication between experts and non-technical decision makers. Provides widespread dissemination of general access products, meeting U.S. EPA s general policy. vi

8 Provides broader validation due to the open source nature of the tools. Provides use of synergies between WERF, U.S. EPA, IWA, Consultants, researchers and utilities. Keywords: Infrastructure asset management, decision support, long-term planning, condition assessment, performance assessment, risk assessment, data mining, case studies, Plan-Do-Check- Act, reinvestment policies, strategic plan, tactical plan. Visual Decision Support Tool for Supporting Asset Management Performance, Risk, and Cost Analysis vii

9 TABLE OF CONTENTS Acknowledgments... iv Abstract and Benefits... vi List of Tables... x List of Figures... x List of Acronyms... xi Executive Summary...ES Introduction The AWARE-P Approach and Software Background The AWARE-P Suite AWARE-P Suite Support to Strategic Planning Support to Tactical Planning Software Overview General Outline AM Planning Tool Spatial and Water System Tools Asset System Analytics and Prediction Tools Note on Integration Leveraging of the AWARE-P Framework in the U.S. Context Relationship with SIMPLE AWARE-P and the ISO Asset Management Standards Effective Utility Management and Other U.S. Specific Framework Directive and Data Standards Software Development for Generic U.S. Deployment Step-by-Step Implementation Guideline for U.S. Utilities Development Tools and Pilot Application Pilot Utility: Gwinnett County Department of Water Resources Available Data and Project Expectations Data Analysis AWARE-P Software Deployment GCDWR Plan: Decision-Making Environment The Planning Framework FAIL: Failure Analysis Tool Pipe Failure Analysis and Forecasting Poisson and LEYP Implementation IVI: Infrastructure Value Index Concept and Basic Formulation Software Implementation viii

10 3.5 INSP: Inspections Analysis Tool The Inspections Analysis and Prediction Module Case Study Applications Network Topological Prototype Mode Conclusions Appendix A... A-1 Appendix B...B-1 References...R-1 Visual Decision Support Tool for Supporting Asset Management Performance, Risk, and Cost Analysis ix

11 LIST OF TABLES 2-1 The AWARE-P Tools Relationship Between SIMPLE s 10 Steps 1-2 and AWARE-P Software Tools Relationship Between SIMPLE s 10 Steps 3-7 and AWARE-P Software Tools Relationship Between SIMPLE's 10 Steps 8-10 and AWARE-P Software Tools Pilot Area Selection Project Assessment According to the Main Stakeholders Involved: Global Approach Project Assessment According to the Main Stakeholders Involved: AWARE-P Software Pre-Existing Tools Project Assessment According to the Main Stakeholders Involved: AWARE-P New Software Tools LIST OF FIGURES 2-1 The IAM PDCA Loop Strategic Planning Use Case with Typical Software Workflows Tactical Planning Use Case with Typical Software Workflows Gwinnett s AWARE-P Software Deployment A Single Planning Time Frame Within the AWARE-Plan Tool Example of Curve Used to Specify Reference Values for a Metric Example of Ranking Result for Three Alternatives of Intervention Under Study GCDWR System of Objectives and Criteria GCDWR Metrics for Objective Poisson Estimated Failure Rates and Probabilities for Backup/Overflow Events LEYP Estimated Failure Rates and Probabilities in the Year LEYP Estimated Failure Rates and Probabilities in 2020 and 2060 Year Evolution IVI Over Asset Useful Life AWARE-P Tool: Input Data Worksheet IVI and Reinvestment Needs Over Time IVI and Reinvestment Needs Over Time Three Different Reinvestment Strategies Descriptive Statistics Total vs Inspected Length and Inspected Condition, Per Installation Year Aggregated Results Predicted Probability of Being in a Critical Grade Prioritization Performance: Estimate Probability of Finding Critical Pipes Predicted Percentage of Total Length of Sewer in Critical Grade Color Thresholds of Critical Percentage of Total Zone Length Pipes Color Coded Based on the Criticality x

12 LIST OF ACRONYMS AM AWARE-P CARE-S CARE-W CCTV CIMP CIP DIP FAIL FIN FMECA GCDWR GIS IAM INSPT IRR ISO IVI IWA kpi LEYP NASSCO NETW NPV ODIM O&M PACP P-D-C-A PI PVC PX R&D ROI RPM Asset Management Advanced Water Management and Rehabilitation Portugal Computer Aided Rehabilitation of Water networks Computer Aided Rehabilitation of Sewer networks Closed Circuit TV Component Importance Capital Improvement Plan Ductile Iron Pipe Failure Analysis Financial Project Failure Mode, Effects, and Criticality Analysis Gwinnett County Department of Water Resources Geographical Information Systems Infrastructure Asset Management Inspection Analysis Investment Return Rate International Standards Organization Infrastructure Value Index International Water Association Key Performance Indicator Linear Extended Yule Process National Association Sewer Service Companies Network Analysis and Visualization Net Present Value Optimized Investment Decision making Operations and Maintenance Pipeline Assessment and Certification Program Plan-Do-Check-Act Performance Indicators Polyvinyl Chloride Performance Indices Research and Development Return on Investment Reinforced Plastic Mortar Pipe Visual Decision Support Tool for Supporting Asset Management Performance, Risk, and Cost Analysis xi

13 SAMP SAMSG SIMPLE SP SSO TAMP TEAMQF 3 TP UL UNMET U.S. EPA WERF WS WWS Strategic Asset Management Plan Strategic Asset Management Specialist Group Sustainable Infrastructure Management Program Learning Environment Strategic Plan Sanitary Sewer Overflows Tactical Asset Management Plan Total Enterprise Asset Management Quality Framework Tactical Plan Underwriters Laboratories Unmet Demand United States Environmental Protection Agency Water Environment Research Foundation Water Supply Wastewater (sanitary or stormwater) xii

14 EXECUTIVE SUMMARY Managing urban water infrastructures involves the challenge of dealing with assets of diverse nature (e.g., pipe networks, treatment plants, pumping stations, and storage structures), useful life, cost, age and condition. Service quality and sustainability require sound long-term planning, well aligned with tactical and operational planning and management. In summary, the objective of an integrated approach to infrastructure asset management is to assist water and wastewater utilities in answering the following questions: Who are we at present, and what service do we deliver? What do we own in terms of infrastructure? Where do we want to be in the long-term? How do we get there? The AWARE-P approach, initially developed in Europe ( builds on from a number of previous R&D efforts, such as the European Union Research projects CARE- W and CARE-S (5 th framework Program, ), and continued to be developed in the scope of other R&D projects, such as igpi ( , and TRUST ( , AWARE-P offers a methodological framework and a powerful portfolio of software tools. It is designed to assist water supply and wastewater utility decision-makers in their analysis and planning processes, based on a Plan-Do-Check-Act process and in accordance with the key principles of the new ISO standards on asset management. It is compatible with, and complements WERF s pre-existing SIMPLE framework. AWARE-P is a novel proposition in the way that it enables the quantified, long-term appraisal, comparison and prioritization of capital investment projects and continuous operations and maintenance (O&M) interventions over multiple dimensions (i.e., incorporating the objectives, analyses and strategies of financial, asset management, engineering, urban planning and other stakeholders) in a defendable, repeatable and transparent way. The portfolio includes some of the most advanced analyses tools available on the market, for the first time encapsulated in a usable framework that sheds academic complexity in favor of direct, pragmatic applicability. The AWARE-P software contains tools for data management, data mining and analyses from many perspectives, performance assessment, forecasting and multi-criteria decisionmaking. It assists in strategic, tactical and operational planning, through a non-intrusive, webbased, collaborative environment where objectives and metrics drive IAM planning. It was designed for water supply, sewer and stormwater systems, and is aimed at industry professionals and managers, as well as at the consultants and technical experts that support them. It is easy to use and maximizes the value of information from multiple existing data sources, both in datarich and data-stressed environments. AWARE-P has already been used by over 30 utilities in the development of corporate strategic and tactical IAM plans. This includes utilities of diverse organizational models, levels of sophistication and size, including several small utilities with limited data availability. The process and the software are designed to make the most of the available information to maximize Visual Decision Support Tool for Supporting Asset Management Performance, Risk, and Cost Analysis ES-1

15 the usefulness of the decisions to be made. The core of the software does not require the existence of vast amounts of analytical information, although several of its modules will extract the best out of rich data, if available. The IAM approach developed in the AWARE-P project was the winner of the International Water Association (IWA) s 2014 Europe and West Asia Project Innovation Award1 in the Planning category (PIA 2014), recognizing innovation and excellence in water engineering projects around the world. The award specifically rewarded the project s originality and innovation; future value to the engineering profession; social, economic, and sustainable design; complexity; and exceeding client/owner needs. AWARE-P has also been awarded the 2014 edition of the biennial Muelheim Water Award 2, recognizing "an excellent contribution to structured infrastructure asset management in water companies." The main objectives of this project were: To test, validate and tailor the AWARE-P approach and software in the U.S. context, for the wastewater services. To develop new open-source software tools. The project aimed more generally at 1) creating incentives for the implementation or enhancement of strategic asset management in U.S. wastewater utilities, by demonstrating integrated, effective and usable leading-edge tools; 2) leveraging the impact of existing and new tools by bridging SIMPLE, AWARE-P and WaterID through a combined usage perspective; 3) promoting and disseminating open-source or free usage leading-edge software in the U.S., leveraging WERF s and the U.S. EPA s impact, and internationally, taking advantage of IWA s Strategic Asset Management Specialist Group (SAMSG). These objectives were fully achieved: The AWARE-P methodology and software have been comprehensively and successfully tested on the Gwinnett County Department of Water Resources (GCDWR) gravity sewer system. Selected key tools were also tested by U.S. project team members not familiar with the software, using available data from GCDWR in the U.S. Based on this case study, it can be stated that the AWARE-P model is simple to apply to even small utilities. The development and validation of the newly developed INSP Inspections Analysis software module 3 counted on a large inspection database and GIS information from another U.S. region. Significant, novel and leading-edge knowledge has been generated, particularly with regard to inspection data analysis and decision-making processes. Several new modules and features of the AWARE-P software have been developed to a professional grade (e.g., IVI2 Infrastructure Value Index 2 and INSP Inspections Analysis), with validation by GCDWR. Several others (e.g., a topological model) were explored in the first part of the project, in order to give GCDWR, GHD and WERF the possibility of choosing the most (as part of the AWARE-P tool portfolio) ES-2

16 relevant paths to proceed within the framework of the project. The results of that exploration process are detailed in the report. The recommendations from the Project subcommittee were also taken into account. The development of the new tools comprised a substantial number of advances, including a prototype topological model with the ability to validate GIS/asset record data; a system to evaluate the mutual information (gain) contained in a preset of user-selected covariates; two distinct predictive models (Random Forest and GompitZ); the development of a zone-based approach to inspection prioritization; and a GIS-based and chart-rich system for inspections and predicted condition result visualization. These tools were leveraged by a survey of procedural steps in WERF s SIMPLE where AWARE-P can provide support or offer a coherent linkage between tools on both frameworks. The Sustainable Infrastructure Management Program Learning Environment, or SIMPLE, provides a web-based asset management learning environment to help utilities implement best practice asset management programs. A key characteristic of AWARE-P tools is the combination of algorithm and methods sophistication with ease of use, both in terms of data input and result interpretation. This has value for utilities as illustrated in statements by GCDWR senior staff such as: One of the most common-sense parameters that I think I have ever overlooked, about the Infrastructure Value Index (IVI) module. It is great to have the possibility to add factors without a preconceived notion of what should drive the inspection, on the Inspection module. I prefer the classification methods such as Random Forest (implemented in AWARE- P/INSP), as it does not require preconceived assumptions about co-variates. On the other side, it does have the hard-core rigorous methods behind it (AWARE-P/Fail LEYP, Poisson). The ability to export to MS Excel and then move into a shapefile is useful; ability to import and handle shapefiles means that geo-referenced information can be displayed and made useful, about the shapefile features available in the data management tool. The project partners were GHD, Inc., as the lead partner, LNEC (National Civil Engineering Laboratory, Portugal) and Addition (Portugal). The project was funded as part of WERF s Innovation and Research for Water Infrastructure for the 21st Century cooperative agreement with the U.S. EPA. Visual Decision Support Tool for Supporting Asset Management Performance, Risk, and Cost Analysis ES-3

17 ES-4

18 CHAPTER 1.0 INTRODUCTION The project was designed to 1) create incentives for implementing strategic asset management in U.S. wastewater utilities, by demonstrating integrated, effective, and usable leading-edge tools; 2) leveraging the impact of existing and new tools by bridging SIMPLE, AWARE-P and WaterID in a combined usage perspective; 3) promoting and disseminating open-source or free usage leading-edge software in the U.S., leveraging WERF s and the U.S. EPA s impact, and internationally, taking advantage of IWA s Strategic Asset Management Specialist Group (SAMSG). The project comprised the following main tasks: Development of the overall Project Plan for the utility case. Testing of the applicability of the pre-existing applicable AWARE-P tools to U.S. wastewater systems, with particular focus on the planning and decision-making environment (PLAN tool), failure analysis (Poisson and LEYP models) and long-term reinvestment (IVI). Development of a new version of the IVI tool to support long-term reinvestment policies. Development of a set of tools to analyze inspection results and to prioritize future actions. Testing of the failure analysis tool (FAIL) and preliminary prototyping of a new tool for component importance in wastewater systems, based on topological modeling. Dissemination Plan (Appendix A) This report summarizes the work carried out and outlines the more relevant outcomes. The report focuses on the AWARE-P components that were tested and developed in this project. The Appendices contain the aforementioned Dissemination Plan (Appendix A) and the complete set of software Quick Guides (Appendix B), including the newly developed tools. There were three goals of this project: A validated portfolio of tools for the AWARE-P software (Alegre et al., 2013; Baseform, 2013; Coelho et al., 2012) in a U.S. context and specifically for wastewater networks, based primarily pilot implementation selected by WERF at Gwinnett County Department of Water Resources (GCDWR), and on complementary tests carried out in other cases, including inspection data from a large and a small utility within the United States. A new, enhanced version of the IVI module to assess infrastructure value and support longterm re-investment policies (Alegre et al., 2014). The new INSP Inspection Analysis module, mainly directed at condition-based inspection analysis and prediction (Vitorino et al., 2014). The latter comprised a substantial number of advances, including a prototype topological model with the ability to validate GIS/asset record data; a system to evaluate potential information gains (mutual information) in the predictive potential of user-selected geo- and asset-related data (covariates); two distinct predictive models (Random Forest and GompitZ); the Visual Decision Support Tool for Supporting Asset Management Performance, Risk, and Cost Analysis 1-1

19 development of a zone-based approach to inspection prioritization; and a GIS-based and chartrich system for inspections and predicted condition result visualization. These tools were leveraged by a coherent linkage with WERF s SIMPLE framework. The Sustainable Infrastructure Management Program Learning Environment, or SIMPLE, knowledge base was developed to provide a web-based asset management learning environment that would help utilities implement best practice asset management programs. SIMPLE was designed to help organizations better direct limited budget resources toward the sustained performance of their assets. Furthermore, improved asset management practices and work processes raise the confidence of stakeholders that sound decisions are being made. SIMPLE includes a number of essential tools to help utilities apply asset management concepts and philosophies in a practical manner. The tools were designed to enable data to be integrated with more advanced asset management systems as industry capabilities matured to include complex asset management decision support systems. The new AWARE-P modules were designed to bring its portfolio of assessment methods for wastewater networks in line with its water supply system capabilities 4. The project partners were GHD Inc., as the lead partner, LNEC (National Civil Engineering Laboratory, Portugal) and Addition (Portugal). The project was funded as part of WERF s Innovation and Research for Water Infrastructure for the 21st Century cooperative agreement ( ) with the U.S. EPA. 4 Fully documented online at (Documentation section, and How-to section of AWARE-P) 1-2

20 CHAPTER 2.0 THE AWARE-P APPROACH AND SOFTWARE 2.1 Background The AWARE-P project ( , Alegre et al., 2013, Alegre and Covas, 2010; Almeida and Cardoso, 2010) aimed at providing water and wastewater utilities with the know-how and the tools needed for efficient decision-making in infrastructure asset management (IAM) of urban water services. All project results from best practice handbooks to business cases, training courses and e-learning materials have been placed in the public domain and are freely distributed as they became available. AWARE-P was developed based on a number of previous R&D efforts, such as the CARE-W and CARE-S EU 5th FP research projects (Sægrov, 2005, 2006), and continued to be enhanced in the scope of other R&D projects, such as igpi ( , and TRUST ( , The infrastructure asset management approach developed is a broad management process that addresses the need for a fundamental plan-docheck-act cycle at a utility s various decisional levels strategic, tactical, operational aiming at the alignment of objectives, metrics and targets, as well as effective feedback across levels. It incorporates the principles generally recommended and adopted in IAM by leading-edge research, consultant and utility organizations (Hughes, 2002; INGENIUM and IPWEA, 2006; Sægrov, 2005 and 2006; Sneesby, 2010). The IAM approach developed in the AWARE-P project was the winner of the 2014 Europe and West Asia Project Innovation Award in the Planning category (PIA 2014, awarded by the International Water Association). Visual Decision Support Tool for Supporting Asset Management Performance, Risk, and Cost Analysis 2-1

21 2.2 The AWARE-P Suite The following section discusses the IAM approach, support to strategic planning, and support to tactical planning IAM Approach Water networks are the most valuable part of the public lifeline infrastructure, and utilities are vested with the responsibility of keeping and expanding them for current and future generations. Both continuous O&M (Operations & Maintenance) and CIP (Capital Improvement Plan) projects translate into a variety of impacts to the infrastructure, its sustainability and its value; to the service provided; to the risks incurred; to the financial management of the utility. These impacts are felt over the long term and in multiple dimensions (service, economics, social). The AWARE-P software suite is designed for infrastructure asset management and planning of urban water systems. It provides the tools for quantified, long-term impact assessment of capital investment projects and continuous O&M interventions in a defendable, repeatable and transparent way. AWARE-P supports service-oriented IAM planning, promoting key ISO requirements and full consideration of system behavior, for both linear and vertical assets of clean water or sewer systems. AWARE-P addresses the need for a plan-do-check-act cycle at the utility s various decisional levels, in order to integrate all infrastructural interventions (both O&M and CIP) as well as those non-infrastructural investments (e.g., in information systems and data) that may have a direct effect on the ability to manage the infrastructural assets (Figure 2-1). Figure 2-1. The IAM PDCA Loop. Importantly, it also provides a solution to the fundamental need for alignment and line-ofsight among the strategic, tactical and operational level-the lack thereof being a frequent, major source of inefficiencies in ROI (return on investment) and use of resources. It is an innovative yet inclusive methodology that materializes two decades of leading-edge R&D, focusing on simplicity for practical application in the industry, and building upon the integration of existing asset management practice. An open-source software system, based on a set of tools and models that assist in the analyses and decision support involved in the planning process, has been developed in order to host the methods mentioned above. 2-2

22 2.2.2 Support to Strategic Planning Strategic planning is developed for the entire organization and aims at establishing the global, long-term corporate directions (typically years). The first stage in developing a strategic plan (SP) is the definition by top management of clear objectives. Through the identification of assessment criteria, a range of metrics is selected to assess them, and targets are set for each metric, translating the attainment of the stated objectives. Realistic objectives and targets require proficient knowledge of the context. If a utility is preparing a strategic plan for the first time, setting up objectives requires taking into account the available context information, even if not yet fully structured and accurate. The second stage is the diagnosis of both the external and internal contexts, in view of the objectives and targets established. The evaluation should be carried out through the planning horizon. The third stage is the formulation, comparison, and selection of strategies that lead to meeting the targets, given the starting point surveyed at the diagnosis stage. These strategies are at the core of the strategic plan, which is then implemented and regularly monitored and reviewed (Figure 2-1). A typical workflow for a strategic planning use case can be summarized through the schematic shown on the left-hand side of Figure 2-2. The right-hand side relates to the software modules and is better understood after reading Section 2.3, particularly for an explanation of the tools and a key to the icons. Figure 2-2. Strategic Planning Use Case with Typical Software Workflows. The tools are presented in Section 2.3. Visual Decision Support Tool for Supporting Asset Management Performance, Risk, and Cost Analysis 2-3

23 2.2.3 Support to Tactical Planning Tactical planning and decision making are framed by the Strategic Plan (SP) and guided by the strategic objectives and targets. The aim of a Tactical Plan (TP) is to establish the intervention alternatives, to be implemented in the medium-term (typically three to five years), that will gradually fulfill the strategies set in the SP. IAM tactical planning is not restricted to infrastructure solutions, as the practitioner should also consider options related to operations and maintenance and to other non-infrastructural solutions. Managing infrastructure has close interdependencies with the management of other assets, such as human resources, information, financial and intangible assets. Not considering them in the SP and TP potentially leads to lack of alignment and inefficiencies in the use of resources. The IAM plan needs to address the noninfrastructure solutions that are critical for meeting the targets and are related to these other types of assets, e.g., investing in a better work orders data system. Typical stage-by-stage workflows for a tactical planning use case are summarized on the left-hand side of Figure 2-3 (as in the previous figure, for software tool names and a key to the icons, see Section 2.3). Refer to Figure 2-2 for the explanation of tool roles at the corresponding stages in the two workflows. Figure 2-3. Tactical Planning Use Case 5 with Typical Software Workflows. The tools are presented in Section NETW, a hydraulic modelling tool (available for water supply systems but not yet for wastewater systems) represents an engineering stage for diagnosis and/or re-engineering of intervention solutions to address the identified pathologies. It feeds the generation of alternatives for intervention. 2-4

24 The software s tools are also ready to be used in stand-alone, direct assessment mode, for the fastest possible path to results. 2.3 Software Overview The AWARE-P software suite is a modularized software package with numerous standalone components developed to help utilities optimize their asset management. Some modules are designed specifically to support strategic asset management while other modules focus on tactical issues. Combined, the various modules constitute a decision support system for asset management that includes several visual tools to help understand and communicate the asset management challenge General Outline The AWARE-P software is a visual suite of tools for infrastructure asset management and planning, supporting the quantifiable impact assessment of AM interventions and promoting the best trade-offs between performance, economics and risk. It is a non-intrusive, web-based, collaborative environment where objectives and metrics drive IAM planning. It was designed for water supply and sewer systems, and is aimed at industry professionals and managers, as well as at the consultants and technical experts that support them. Using available data (asset registry, geo-databases, kpis 6, financial, operational), the AWARE-P suite comprises a modular tool portfolio for AM planning, spatial analysis, and asset system diagnosis and prediction. The software has been designed following a modular paradigm in an open arrangement that allows for its usage with multiple workflows. The tools may be used individually or in combination. The AWARE-P IAM approach is equally applicable to water supply, wastewater or stormwater assets. The methodology and the software were designed with that purpose, and developed for the entire urban water cycle, while considering the specificities of the different types of systems. However, the initial portfolio of tools was tailored to water supply systems instead of wastewater or stormwater systems. A roadmap was put in place to address that balance, and this project represents an important contribution in that regard. PLAN and a number of the current tools (see the next section for an explanation of the acronyms) are designed to be directly applicable to all of those systems, such as performance indicators, IVI, financial analysis, or failure analysis. An important feature of the software and of the AWARE-P IAM approach is its focus on evaluating urban water networks as systems rather than collections of independent components. For this reason, the range of assessment models and methods available draws heavily on the capability to represent and simulate system behavior, whenever possible with support from network simulators. This leads to the capability to produce both component-based metrics and system-wide metrics. 6 kpi: Key Performance Indicator Visual Decision Support Tool for Supporting Asset Management Performance, Risk, and Cost Analysis 2-5

25 2.3.2 AM Planning Tools Infrastructure planning in the short- and long-term, and deciding where to act or which projects to prioritize, are the driving forces behind the AWARE-P methodology. At project, subsystem or entire system level, alignment among strategic, tactical and operational decisions is promoted through a coherent metrics/kpi development process. These metrics make up the assessment system that quantifies the consequence of the prospective decisions on utility objectives in the short-, medium- and long-term-a central framework for the continuous improvement principles of ISO The currently available AM planning software tools are: PLAN Plan. A decision-support environment where planning alternatives or competing projects are measured up, compared and prioritized through objectives-guided metrics PI FIN Performance Indicators. A quantitative assessment of the efficiency or effectiveness of a system is provided, through the calculation of performance indicators. A tool for selection and calculation of KPI based on organized libraries, including industry standards (IWA) and userdeveloped libraries. Financial Project. Assess the net present value (NPV) and the investment return rate (IRR) of any financial project from a long-term/ asset lifecycle perspective Spatial and Water System Tools Water supply and wastewater systems are spatial infrastructures with a direct connection to geographical data and to network hydraulics behavior. Network analysis, spatial analysis and the ability to express results where they happen on a map are key to understanding a system and how it may evolve as a consequence of infrastructure strategies. The available spatial and water system tools include the following: NETW EPANet Network Modeling. An efficient, Java-implemented EPANet simulation engine and natively integrated MSX library for full-range hydraulic and water quality network simulation, with advanced 2D/3D visualization and Google Earth integration. SHAPE SHAPE (Baseform s GIS core). Enables the inclusion of geo-referenced data in the analysis environment, with mapping facilities to help contextualize analysis results. 2-6

26 2.3.4 Asset System Analytics and Prediction Tools The biggest challenge in AM is to take advantage of the available data to reach the best tactical and strategic decisions. A range of modules that make available to everyday use the most effective methods for diagnosing current status and predicting future behavior are included in this category to complement standard AM software and aggregate even more value to their results. The available asset system analytics and prediction tools comprise the following: FAIL CIMP Failure Analysis. Using system component failure records, such as work orders, predict present and future probability of failure of pipes or sewers. Component Importance. Simulate the failure of each individual pipe in a water supply network to measure its hydraulic impact on nodal consumption. UNMet INSP PX IVI Unmet Demand. Quantify water supply service interruption risk through the expected reduced service, calculated as the volume of unmet demand over a given period. Inspection Analysis. Using system condition assessment records, such as from CCTV inspections, predict present and future sewer condition, and intelligently guide the inspection effort. Performance Indices. Technical performance metrics based on the values of certain features or state variables of water supply is provided. Simulationbased tool provides detailed technical performance assessment, related to capacity, level-of-service, network effectiveness and efficiency, water quality and energy system behavior. Infrastructure Value Index. Analyze the ageing degree of an infrastructure as a ratio between the current and replacement values of its components, and project short- and long-term investment needs. Visual Decision Support Tool for Supporting Asset Management Performance, Risk, and Cost Analysis 2-7

27 Table 2-1 identifies the tools that were already available in the AWARE-P platform before this project began; those that were developed concurrently in other, simultaneous projects and in some way contributed to the currently existing tools; and those that were specifically developed by this project. Detailed information concerning the latter is presented in Section Table 2. The AWARE-P tools Table 2-1. The AWARE-P Tools. Pre-existing from Projects AWARE-P and TRUST WS + WW Failure Analysis: Poisson (WS) LEYP (WS) Financial Project (WS+WW) Performance Indicators (WS+WW) Performance Indices (WS) Network Analysis and Visualization (WS) Component Importance (WS) Developed Concurrently through Projects igpi & TRUST New Data Editor (WS+WW) Shape File Viewer (WS+WW) New Help System (WS+WW) I/O Excel Improved and Easier to Use (WS+WW) TRUST PI Library (WS+WW) Tested/ Validated or Developed/Enhanced within INFR5R12 Tested / Validated: WW WW Enhanced: Developed: LEYP (WW) Shape Viewer Data Editor and Import/Export IVI2 (WS+WW) IVI1 (Linear Assets) (WS+WW) Unmet Demand (WS) Data Representation(WW) Mutual Explanation (WW) Random Forest (WW) Gompitz (WW) Help Quick Start WS: Water Supply WW: Wastewater (Sanitary or Stormwater) A Note on Integration AWARE-P forms a diagnosis, analysis and planning layer where data and processes related to AM may be integrated to converge in useful, defensible decision making. One of the main challenges faced by asset managers in most utilities (and particularly in those with more diversified data) is integrating the several, often conflicting, sources of information available on the infrastructure, its condition, its performance, and the various predictive analyses that assist in prioritizing projects or interventions. The software represents a leap forward in integration of: Technology: uses data from any source (GIS, work orders/ maintenance, accounting, asset register, network models). Processes: is able to incorporate results from existing AM processes to inform assessment. 2-8

28 Asset management decision making: promotes alignment between operations, technical management and strategic governance. All infrastructure assets: the software is applicable to linear assets (networks) as well as vertical assets (facilities), and supports both water supply, wastewater and stormwater infrastructures. 2.4 Leveraging of the AWARE-P Framework in the U.S. Context AWARE-P was initially developed in Europe and is currently used there by numerous agencies. The focus of this project is, in part, to bring AWARE-P into use in the United States. This requires a discussion about the integration of AWARE-P with existing asset management initiatives in the U.S. market. To accomplish this, several programs familiar to U.S. utilities are discussed below. AWARE-P software is relevant to each initiative as described below: SIMPLE, ISO 55000, and Effective Utility Management (EUM) Relationship to SIMPLE Developing a method and set of tools for IAM support compatible with the WERF research portfolio demands a coherent linkage with SIMPLE. This is all the more important given the U.S. and international relevance of SIMPLE. The objective of this section is to show the link between SIMPLE and the AWARE-P framework and available tools, including both the tool set developed for this project and the remaining portfolio currently available. SIMPLE was developed to provide a web-based asset management learning environment that can help utilities implement best practice asset management programs. SIMPLE benefits the industry by: Providing a systematic approach to determining cost effective investment. Helping extend the life of assets through use of optimal maintenance and rehabilitation tools. Improving asset investment analysis. Enabling gains in productivity and effectiveness. Improving operational efficiency in work practices. Benchmarking activities to drive efficiencies. SIMPLE was designed to help organizations better direct limited budget resources toward the sustained performance of their assets. Furthermore, improved asset management practices and work processes raise the confidence of stakeholders that sound decisions are being made. The key elements of SIMPLE are: A framework for organizing and applying AM material (TEAMQF). Guidelines outlining best appropriate practices. Case studies. A basic implementation case study and training program. Visual Decision Support Tool for Supporting Asset Management Performance, Risk, and Cost Analysis 2-9

29 SIMPLE includes a number of essential tools to help utilities apply asset management concepts and philosophies in a practical manner. The tools were designed to enable data to be integrated with more advanced asset management systems as industry capabilities matured to include complex asset management decision support systems. The tools developed for the SIMPLE program are listed below: Asset Hierarchal Tool. Condition Assessment Tool. Remaining Effective Life Tool. Life Cycle Costing Tool. Level of Service Tool. Business Risk Exposure Tool. Benefit Cost Tool. End of Asset Life Tool. Business Case Tool. Capital Investment Validation and Prioritization Tool. Asset Management Plan Tool. SAM-GAP Asset Management Gap Analysis Tool. The 10 steps of asset management in SIMPLE are used as the basis to establish the existing links, which are summarized in Tables 2-2, 2-3, and 2-4. Table 2-2. Relationship Between SIMPLE s Steps 1-2 and AWARE-P Software Tools. Fundamentals of Asset Management 10 Steps in SIMPLE Step 1. Develop Asset Registry System layout; data hierarchy standards, and inventory Step 2. Assess Performance, Failure Modes Condition assessment protocol; rating methodologies Related AWARE-P Tools Contribution from AWARE-P Tools to Implementing SIMPLE s IAM Steps The development, and above all, the update, maintenance and actual use of a good asset registry depend on the ease of integration of multiple data sources. The Data Manager module (including its import/export features) and the Shape Viewer were designed with this purpose. AWARE-P derives from many years of research and practical implementations of performance assessment systems by LNEC s team. It benefits directly from this know-how through the incorporation of several performance indicator libraries generally accepted as key international references (e.g., IWA PI systems). The PI and the PX tools also allow for tailor- made libraries. The FAIL and INSP modules are state-of-the-art tools for estimating failure probabilities and for assessing the likelihood of condition-critical assets over time. 2-10

30 Table 2-3. Relationship Between SIMPLE s Steps 3-7 and AWARE-P Software Tools. Fundamentals of Asset Management 10 Steps in SIMPLE Step 3. Determine Residual Life Expected life tables; decay curve Step 4. Determine Life Cycle & Replacement Costs Valuation; life cycle costing Step 5. Set Target Level of Service Demand analysis; balanced scorecard; performance metrics Related AWARE-P Tools Contribution from AWARE-P Tools to Implementing SIMPLE s IAM Steps Residual life depends on functional performance, on asset condition and on the expectation of their respective evolution over time. PI, PX, FAIL, and INSP may significantly contribute to this step. Life cycle costing requires a good integration of information from many different sources in the organization. Data Manager may be very useful in this regard. Asset valuation can easily and effectively be carried out using IVI. Other tools such as PLAN, PI, and FIN support sensitivity analyses and enable using the life cycle and replacement costs to assist in planning and decision making. The selection of the right performance metrics and the establishment of reference levels and targets is a critical path in IAM and a rather developed centerpiece in AWARE-P. PLAN, PI, and Px are its main supporting tools. Step 6a. Determine Business Risk ("Criticality") Step 6b. Optimized Investment Decision Making (OIDM) FMECA; Business Risk Exposure; Delphi techniques Step 7. Optimize Operations & Maintenance (O&M) Investment Root cause; RCM; PdM; RDM The current version of AWARE-P does not include yet any specific tools to assess business risk, particularly in the scope of wastewater systems. Plan, PI, Fail, and INSP may be indirectly applied and useful. PLAN was designed to support the ranking or selection of intervention alternatives, being O&M, Capital Investment or a combination of these. This process may also benefit from a financial assessment of the projects (FIN) and by the analysis of the long-term evolution of IVI. Visual Decision Support Tool for Supporting Asset Management Performance, Risk, and Cost Analysis 2-11

31 Table 2-4. Relationship Between SIMPLE s Steps 8-10 and AWARE-P Software Tools. Fundamentals of Asset Management 10 Steps in SIMPLE Step 8. Optimize Capital Investment Confidence level rating; strategic validation; ORDM Related AWARE-P Tools Same as previous. Contribution from AWARE-P Tools to Implementing SIMPLE s IAM Steps Step 9. Determine Funding Strategy Renewal annuity Same as previous. Step 10. Build Asset Management Plan Asset mgmt plan; policies and strategy; annual budget Template strategic plan The AWARE-P product portfolio goes beyond the software and contains guidelines and other written publications. A Strategic Plan template was made available in the scope of this project 7, complement to the SIMPLE guidelines directed to tactical IAM planning AWARE-P and the ISO Standards on Asset Management The 2014 publication of the international standards (ISO) Asset Management is an important milestone that challenges the compatibility of the presented AM framework and tools. A thorough analysis of the requirements in these standards and how the AWARE-P principles and tools relate to them has been carried out. The analyses pointed to full compatibility, but also confirmed that AWARE-P fulfills relevant gaps in the previously available AM software offer in terms of transparency, accountability and repeatability of the decision-making processes, which are at the core of the standards Effective Utility Management and Other U.S. Specific Framework Directives and Data Standards In addition to SIMPLE, an important dimension of this project lies in establishing the points of contact and aligning, to the extent possible, the methods and tools described in this report with the best practice U.S.-specific frameworks and guidelines, such as Effective Utility Management and Lean (U.S. EPA, 2012; U.S. EPA et al., 2008). It is also important to note that data requirements related to the software developed in the project are made to comply to the extent possible with U.S. data standards, particularly as defined by NASSCO for pipeline assessment (NASSCO, 2010). 7 Benefitting from the experience collected in the development of over 30 strategic AM plans in the igpi project ( 2-12

32 2.4.4 Software Development for Generic U.S. Deployment An online deployment of the AWARE-P software has been made available to Gwinnett County (GA) Department of Water Resources and to the project team, including the open-source modules in the current AWARE-P portfolio, the newly developed features and the new modules as they become available, as well as other modules that may be needed to support development. The final result of the project is a software deployment for WERF that includes the new open-source modules and features developed under the current project as well as all of the currently available open-source AWARE-P modules. Integration or combination with SIMPLE tools and workflows will be highlighted or flagged at WERF s online SIMPLE/AWARE-P frameworks whenever possible Step-by-Step Implementation Guideline for U.S. Utilities The software and its various components are fully documented online in the baseform.org site 8, through a comprehensive help system, including quick start guides, how-to tutorials, workflows and examples. This documentation was developed as a step-by-step implementation guideline for U.S. utilities and is included in this report as Appendix B. A section is included for each software tool, specifying how to use it. This information constitutes the software user guide. 8 Baseform.org is a software platform hosted by project partner Addition. The AWARE-P software is distributed and supported through that platform. Visual Decision Support Tool for Supporting Asset Management Performance, Risk, and Cost Analysis 2-13

33 2-14

34 CHAPTER 3.0 DEVELOPED TOOLS AND PILOT APPLICATION 3.1 Pilot Utility: Gwinnett County Department of Water Resources With the assistance of WERF, member utilities were recruited for this project. Gwinnett County Department of Water Resources (GCDWR) responded to that recruitment effort and volunteered to participate. The County, which is located northeast of Atlanta, Georgia, agreed to participate and to host several onsite work sessions throughout the project. Their participation included providing the project team with access to data and information and the commitment of resources to participate in three onsite multi-day work sessions as well as time between sessions during which GCDWR management and staff provided a significant number of hours to test and refine the software during development. The project team is grateful for GCDWR s participation Available Data and Project Expectations The Gwinnett County Department of Water Resources (GCDWR) 9, providing water, wastewater, stormwater and reusable water services to Gwinnett County (GA), was selected by WERF as the pilot wastewater utility for the project. The project was focused on GCDWR s separate sanitary sewer system. The county owns and maintains approximately 2,600 miles of collection system pipes, ranging in size from 6-72 inches in diameter, as well as approximately 74,600 maintenance access structures (GCDWR, 2013). Due to the expansion of the collection system in the 1980 s and 1990 s, the median year of installation for the network is 1994, making the average pipe approximately 19-years old. The gravity collection system spans most of the unincorporated county, comprising a service area of 437 square miles. Many portions of the County connect to this gravity sewer through the use of pumping stations and their associated force mains. The system currently serves 151,000 sewer customers (Gwinnett County, 2013). Onsite sessions took place at GCDWR s facilities in Lawrenceville, GA, during the project. The utility made available a large amount of documentation and data pertaining to the gravity sewer system, containing namely (but not limited to): 2011 Gravity Sewer System Strategic Asset Management Plan (SAMP) Gravity Sewer System Tactical Asset Management Plan (TAMP) including all appendices Field Operations sheets. Sewer Gravity Risk Matrix v6 / Model Data Dictionary v The 2011 Tamp was also supplied Visual Decision Support Tool for Supporting Asset Management Performance, Risk, and Cost Analysis 3-1

35 Work Orders database (WO_CAData mdb), including pipeline and manhole repairs, backups/sso, and inspections. GIS shapefiles for sewer gravity mains; sewer manholes; sewer pressurized mains; sewer pump stations; sewer service areas; sewer treatment plants; sewer valves; stormwater drainage basins; centerlines; force main dump manholes; land lots; parcels; sewer abandoned pipes; sewer abandoned structures. CCTV inspection records (Inspection_Data.accdb and dbo_swtvobsr). Access to the FlowWorks monitoring system (flow and rainfall continuous monitoring stations). GCDWR has an operational-level hydraulic model available for the gravity sewer system. The model is based on the SewerGems software and corresponds to a skeletonized description of the system, comprising the most important pipes. During the on-site working sessions, the GCDWR team conveyed their expectations for the project as a meaningful contribution towards an integrated solution for IAM planning that joins available results as well as new results, helping to circumvent the current limitations of running several unrelated models in separate technologies. Some of the potential goals that were most emphasized by the GCDWR team included: Decision-making integration around failure analysis, IVI and other aspects of IAM planning. (Data) systems integration that would help reduce the number of different models (currently up to 5) involved in the decision-making process. A solution where the entire network can be addressed, and where 30,000-foot views, as well as detailed views, are possible, including the ability to address and evaluate long-term analyses (15-20 years). Validation of GCDWR s current procedures through AWARE-P. Contributions towards short-term operational decisions, and their coherence as regards the tactical and strategic levels Data Analysis The availability of GIS records (shape files), pipe inspections and flow and rainfall monitoring data was well above average; the data was of generally high quality and easily accessible. Work order records were available since The key sewer and manhole data needed was mostly available, with minor gaps 11 which did not appear to be significant overall. A descriptive statistics report, developed using external statistical analysis tools and intended to support project team sensitivity gain, was included in the Project Plan, as Appendix A. Until a certain stage of development of the project, it had been agreed that the AWARE- P methodology and software, including the modules developed during the project, would be tested in one pilot area. The pilot area selection process pointed to the Patterson area, as shown 11 Missing or invalid installation dates for fewer than 8% of the sewers and fewer than 10% of the sewers lacking a consistent value for one or both invert levels; these are low levels that reflect the high quality of the original data. 3-2

36 in Table 3-1. However, the evolution of the work showed that it was feasible to continue embracing the whole network in the study. Table 3-1. Pilot Area Selection. Basin: Patterson Ezzard Existing SSO (sanitary sewer overflow) issues (flooding or discharges) Just minor Historically relevant South Buford Just minor Selection criteria Existing inflow and infiltration issues Existing condition problems Potential for design improvement (layout/diameters) Potential for medium term concentrated rehab needs Yes Yes Historically relevant Historically relevant Yes Yes Limited No Limited High No High Potential for operational modes improvement High Limited High Data availability Very good Very good Very good Basin size Large Small Medium AWARE-P Software Deployment GCDWR Analysis of the potential usage of AWARE-P software tools for GCDWR s sewer system, and for sewer systems in general, was an expected outcome of the project. A specific deployment of the baseform.org AWARE-P software was created to support the Gwinnett County pilot project, and made available at: gwinnett.baseform.org (Figure 3-1). Figure 3-1. Gwinnett s AWARE-P Software Deployment. Visual Decision Support Tool for Supporting Asset Management Performance, Risk, and Cost Analysis 3-3

37 The deployed software was a private, secured instance and included the full available portfolio as well as the beta (and eventually final) versions of the project-developed modules. The software deployment was used by Addition and LNEC s project team since the beginning of the project, to store and share data and documentation, and to perform preliminary analyses as described along this report. The software is proven as a collaborative environment, not only to perform calculations but also to share information, much as offered by cloud storage environments. Access to this deployment has been extended to all project team members, including GCDWR s project team and designated users. 3.2 Plan: Decision-Making Environment As a decision support system AWARE-P is intended to help utility staff determine the optimum choices for spending limited resources on many competing projects. This planning aspect is vital to achieving long-term sustainability for many agencies The Planning Framework PLAN is a decision-support environment providing organized assessments and comparisons for any number of competing projects, solutions or alternative designs, which can be assessed and pitched against each other numerically as well as visually. PLAN was designed as the central planning framework of the AWARE-P methodology, where planning alternatives or competing projects are measured up and compared, through selected performance, risk and cost metrics, through an interactive numerical and 2D/3D graphical information display. It may also be used to compare different systems or sub-systems for diagnosis and prioritization. It was created as a technical tool, but just as importantly as a negotiation and communication vehicle. As in the AWARE-P approach, emphasis is placed on evaluating impacts over the longterm and in multiple dimensions (service, economics, social), and on quantifying the impact of those interventions in a defendable, repeatable and transparent way. PLAN is based on the three main axes that characterize the assessment and comparison process: a time frame, a set of alternatives under comparison, and the chosen metrics (Figure 3-2). PLAN's flexible 2D/3D cube display gives the user total control of which dimensions and viewpoints are required for analysis, providing a global view of the impact of a decision. Figure 3-2. A Single Planning Time Frame (Left) and a Cube of Results (right) from AWARE-P s PLAN Tool. 3-4

38 The time frame comprehends both a planning horizon (i.e., the time frame of the intervention) and an analysis horizon (the long-term time frame for impact assessment). Particular care was given to the process for selection of metrics and their targets. Stating objectives up front, and understanding which criteria to use for their assessment, is a crucial step in developing effective metrics. Standardized metrics reflect the impact of the alternatives on the set objectives in performance, risk or economic terms. The metrics development procedure in PLAN is a powerful means for understanding the impact of the measured criterion on the overall decision. It uses a standardization of the realworld values of the chosen metric into a 0-3 continuous scale. This scale has an explicit qualification into desirable (green, >= 2), inadequate (yellow, >= 1, < 2) or undesirable (red, <1) levels (Figure 3-3). Figure 3-3. Example of a Curve Used to Specify Reference Values for a Metric, Converting an Intensive Metric Into a Normalized Dimensionless Graph. In essence, the metric reflects the deviation from the target (green zone), which translates the set objective. In other words, it reflects the consequence or impact of adopting the alternative under evaluation (if used to compare competing solutions) or the shortcoming of the object under evaluation (if used to diagnose and prioritize subsystems or sets of assets). A simple weighting system provides a mechanism to balance the different metrics selected for the assessment process. The system can be further fine-tuned through timedependent weights or an exclusion criterion that precludes alternatives not complying with a specific individual target. The diagnosis stage should be carried out based on the metrics selected, for the present situation and through the planning horizon. There is often the need to adopt a progressive system-based screening process, aimed at prioritizing system sectors, using the set of metrics selected. The most problematic sectors are focused on and analyzed in more detail. For those that do not display significant overall problems, there is the need to confirm that they do not have relevant localized problems. If they do, these localized areas need to be retained as well for detailed analysis. This screening process leads to the identification of priority areas of intervention. For these, the diagnosis needs to be more detailed, so that the causes of the problems are fully understood. Feasible intervention alternatives are compared and ranked in PLAN. For each Visual Decision Support Tool for Supporting Asset Management Performance, Risk, and Cost Analysis 3-5

39 subsystem, the intervention alternatives that best balance the set of metrics for the chosen objectives (evaluated via a ranking procedure), over the long-term, will be selected (Figure 3-4). Figure 3-4. Example of a Ranking Result for 3 Alternatives of Intervention Under Study. The main steps of the planning process are the following: Begin by setting objectives and the assessment criteria. Define the time steps: the planning horizon and the analysis horizon. If required, define scenarios, to translate potential external contexts that may impact the system or alternatives under consideration. Chose the metrics, to quantify the degree of achievement of a given objective, and define its reference values. Describe the alternatives under assessment. Introduce data related to all the metrics in each of the alternatives. Once the problem is formulated, PLAN compares and ranks the different alternatives. As referred in Section (Tables 2-3 and 2-4), PLAN can contribute to implementing SIMPLE s AM Step 5. Set Target Level of Service; Step 7. O&M; Step 8. Optimize Capital Investment; Step 9. Determine Funding Strategy and Step 10. Build Asset Management Plan. 3-6

40 Software Implementation Joint sessions with GCDWR allowed for a pilot prioritization of the utility s tactical-level projects for a specific catchment area. Under GCDWR s stated corporate vision- Uninterrupted excellent service -a detailed development of a system of objectives, criteria, metrics and targets was carried out with the GCDWR team, following the AWARE-P IAM planning methodology. The GCDWR objectives and criteria can be seen in the screenshot in Figure 3-5. Figure 3-5. GCDWR s System of Objectives and Criteria. GCDWR metrics for Objective 2 Criteria are presented, as an example, in Figure 3-6. Figure 3-6. GCDWR Metrics for Objective 2. This was an essential stepping stone in the ability to assess and compare alternative strategic and tactical interventions, such as when prioritizing competing projects for a given system area. Since the above-mentioned joint onsite working sessions, the AWARE-P modules PLAN and PI have been detailed, tested and used by the utility team for the continued development of the assessment system and for validation of the approach. Despite the quantity and quality of GCDWR s data, the detailed preliminary data analysis made it evident that the relative young age of the system (a pipe age median around 17 years) would be a limiting factor in failure- and reliability-related analyses. In other words, the pipe Visual Decision Support Tool for Supporting Asset Management Performance, Risk, and Cost Analysis 3-7

41 population is too young to allow for a full test of some of the statistical models under development or pre-existing. Upon meeting with Gwinnett County, the project team realized that there were not enough data available for assessing the aging effect in infrastructure. Additional data from another utility were requested and retrieved from large a sized utility. This was a very large dataset with an appropriately long history. Data were studied and processed; however, it was clear to the project team that data collection had not been driven by the same criteria as in the case of GCDWR. This fact prevented extrapolation of the aging effect for the case of Gwinnett. However, the large utility data was very valuable for testing and validating the several inspection-related tools. Given the non-disclosure agreement in place, detailed information on this data analyses are not included in this report. However, the data were instrumental in further testing the validity and resolving power of INPS s data mining routines, in developing and finetuning the relevant data views, and in highlighting key issues in geo-data and asset-register data that carry the most important reliability threats to the significance of the analysis results. 3.3 FAIL: Failure Analysis Tool In the vast majority of instances, conducting preventive maintenance is less costly than reacting to infrastructure failures. As a result, the ability to use data and information to help predict failures can have a significant effect on utility operating and capital budgets. This tool supports failure forecasting for improved infrastructure cost management Pipe Failure Analysis and Forecasting A key goal of the project was to have the same set of tools available for wastewater/stormwater drainage systems that were available in the AWARE-P portfolio for water supply. Among those, some of the most needed methods concerned risk assessment, through the combination of probability or likelihood of failure estimates and the corresponding estimation of consequences, in their multiple dimensions. Some risks can be assessed quantitatively (probability x numerical measure of consequence), but for in most cases a qualitative approach is required, often through a risk matrix. The project tackled both quantitative and qualitative assessment of probability and consequence. It must be noted that there was no intention to artificially convert tools developed for water supply in order to suit wastewater/stormwater. Each system has its specificities and failure analysis or component importance for wastewater systems were considered from the ground up for that purpose. According to U.S. and international best practice, the probability of sewer failure can be estimated from CCTV inspection observations, evaluated through standard condition assessment protocols, such as PACP or similar (NASSCO, 2010; WRC, 2004), or by applying statistical analysis and prediction of failure events, e.g. as recorded in work orders. When data for any of these alternatives are not available, sewer age and pipe material are often used to obtain an estimate of the likelihood of failure. The AWARE-P portfolio already contained two tools for structural pipe failure analysis and forecasting, developed for water supply networks: Poisson and LEYP. 3-8

42 Poisson is based on the statistical model of the same name, for direct failure rates and estimated probabilities. The Poisson is a counting process in which the events occur independently at a constant rate, and where the number of events is assumed to follow a Poisson distribution (Martins, 2011). It is assumed that the rate of the counting process is proportional to the length of each pipe. The failure rate is estimated by the maximum likelihood method. The predicted number of failures in each pipe is obtained using the expected value of the Poisson distribution, whereas the failure probabilities are obtained using the Poisson probability function. LEYP, or Linear Extended Yule Process (Le Gat, 2009; Martins, 2011), is a more sophisticated prediction model that accounts for individual asset s history, a counting process where the intensity function depends on the age of the pipe, the number of past events and a vector of covariates potentially predictive variables-such as pipe diameter and, in this implementation, the logarithm of pipe length. As no conceptual reason appeared to prevent the successful application of these models to sewer networks, their applicability was explored in this project Failures may be divided into those that can be repaired or solved, and those that correspond to an end-of-life condition. Systems that are predominantly made up of newer pipes, such as Gwinnett s, will not have significant numbers of the latter case; furthermore, any pipes that may have been replaced during this initial short life are not normal members of the population and further bias those results. Some of the most relevant failure modes are related to structural condition problems: roots, blockages, and other capacity-hampering causes such as the accumulation of grease. As referred to in Section (Tables 2-2 and 2-3) FAIL (and LEYP in particular), can contribute to implementing SIMPLE s AM Step 2 (Assess Performance, Failure Modes); and Step 3 (Determine Residual Life). Statistical prediction of future failures is to be utilized with caution, as a complement to other forms of inference, such as based on inspections. Realizing this, the model implemented by the statistical analysis tool includes self-testing mechanisms to quantify model fitness Poisson and LEYP Implementation A validated dataset of gravity sewers was used. It included 71,200 pipes these are obtained by retaining only the ones containing information on installation date, material, diameter and length, out of the 72,512 mains of the prototype topological model (see Section 3.4). Important note: the events that this type of analysis applies to must be reactive and not issued from planned actions. A preliminary analysis indicated that some of the failure modes represented in GCDWR s data was recorded through proactive intervention, a common, often unavoidable feature of work order systems. Further work with GCDWR s team made it possible to partially filter the data in order to concentrate on reactive events as much as possible. Such data validation is crucial in making sure that the results of failure analysis are meaningful. Conversely, those results must always be read in light of such considerations regarding the available data. The analysis was based on the failure data contained in the Backups_SSO table of the work orders database (WO_CAData mdb), including backups and overflows, both on manholes and homes/businesses, potentially causing flooding and damage. This file assigns Visual Decision Support Tool for Supporting Asset Management Performance, Risk, and Cost Analysis 3-9

43 backups directly to gravity sewers IDs and overflows to manhole IDs, and therefore relates both to frequency and location through geodatabase overlay. Using the topological model to identify downstream relations, a simplified approach was employed to assign the manhole-recorded overflows to the sewer that is immediately downstream, in order to isolate a sewers-only dataset. Raw failure data includes 3,349 backup/overflow events over a 3.4-year period (from Jan 2010 onward). The filtered backup/overflow failure dataset, including only those backup/overflow events on gravity sewers or manholes marked "complete," and which are included in the abovementioned validated gravity sewers dataset, included 3,312 events (98.89% of the total number of events). This backup/overflow failure dataset has not been checked for duplicates (particularly pipe backups and manhole SSO that may reflect the same event). This check would require direct work with the GCDWR field teams and has been deferred for future work. The analysis carried out included Poisson estimated failure rates and probabilities, by material and by individual sewer (Figure 3-7). All data files are available in the project software deployment. Figure 3-7. Poisson Estimated Failure Rates and Probabilities for Backup/Overflow Events. 3-10

44 The LEYP model includes as covariates the pipe s previous events, as well as its age, diameter, and length. The AWARE-P LEYP implementation includes a fitness test to validate the resulting model parameters against each covariate, shown in grey or red in the screenshot; whenever a covariate fails to describe the behavior of the event data the whole line is show in red. The model can use any set of explanatory covariates for which data are available, and separate the assets into any group of cohorts. The software implementation used in AWARE-P focuses on commonly available data: material, installation date, diameter, length, age at each failure and number of failures. As a further development, other data may be explored as suitable covariates. This LEYP analysis included 2013, 2020 and 2060 forecasting (Figures 3-8 and 3-9a) and b), respectively). It showed good LEYP results for RPM, DIP and PVC, covering the majority of the system (86.2% of the total number of sewers). Results for the remaining materials were inconclusive either due to the small size of the samples or the short recorded histories for the failure modes analyzed. The red-colored parameters in Figure 3-8 reflect that effect. Figure 3-8. LEYP Estimated Failure Rates and Probabilities in the Year Visual Decision Support Tool for Supporting Asset Management Performance, Risk, and Cost Analysis 3-11

45 Figure 3-9. LEYP Estimated Failure Rates and Probabilities in (a) 2020 and (b) 2060 year. The methods proved to be useful in terms of usability in the operational context of a utility, and to provide results beyond published methods, in terms of both quality and reliability of the predictions. A new AWARE-P open-source module was made available for sewer failure statistical analysis and forecast, including a specification of data and applicability requirements, in particular as regards the types of failure events that can be analyzed e.g., the events must be failures (to which the utility reacted) and not stem from a biased population of data such as resulting from planned actions. Testing of the existing LEYP and Poisson models demonstrated both the applicability and the limitations of this type of statistical models for the available sewer failure data. Part of these limitations relates to data issues pertaining to WO records, the proactive/reactive nature of the recorded works, and the variety and nature of failure modes. This is often unavoidable in WO records and failure analysis and prediction models must be deployed with the best possible knowledge of the failure data used. The LEYP and Poisson implementations present in the software were designed to present the most useful features of both methods and to make the limitations of the dataset as evident as possible to the user. An important complement to failure data (both at GCDWR and in a large number of utilities) is condition assessment data, and it was deemed important to second the above reliability models with a condition-based deterioration model. Early implementations showed the potential at GCDWR, and this was discussed with the GCDWR team at the October on-site sessions. It was considered by the joint team that this type of model is of greater advantage than supplementing the existing statistical-based reliability models (LEYP/Poisson) with a third formulation of the same type. The methods and software provided by SIMPLE for condition assessment (namely in the context of SIMPLE s IAM Step 2) and for end-of- life assessment (EOL tool, in SIMPLE s AM Step 3), combine usefully with AWARE-P s Failure Analysis module, namely the LEYP model included. 3-12

46 3.4 IVI: Infrastructure Value Index Asset Management, when executed properly, can help utilities not only understand what future investments will be required but also to manage those requirements through proactive planning. The IVI tool is designed to help utility managers visualize how various investment strategies can influence rates over time so that sustainability can be achieved Concept and Basic Formulation The Infrastructure Value Index at a given time t (IVIt) is a performance-cost measure that reflects the age of an infrastructure. It is given by the ratio between the current (fair) value of an infrastructure and the replacement cost on modern equivalent asset basis (Alegre and Covas, 2010) as stated in (1). This index is particularly suitable for establishing goals associated to infrastructural sustainability criteria. Infrastructure Value Index (%) = Infrastructure Current (Fair)Value Infrastructure Replacement Cost If all assets of a given infrastructure had the same replacement cost and the same useful life, IVI would represent the residual life (%), i.e., [1- (average age/useful life)] %. In a real-life infrastructure, IVI can be seen as a weighted average of the residual lives (%) of the infrastructure components, where the weights are the component replacement costs. IVI is always referred to a date (year), as a snapshot. The Infrastructure Current Value would be, in a competitive market activity, its market value. In a monopolistic activity, as in urban water services, alternative valuation approaches must be adopted. One possibility might be the use of the accounting value. However, this option is not recommended for multiple reasons. The Infrastructure Replacement Cost is the expected cost of a modern equivalent if the infrastructure were built in the IVI year. IVI can be assessed in many different ways, derived from two main families: Asset-oriented: The calculation is based on the useful life of each asset, on depreciation curves and on replacement costs for each category of assets. Service-oriented: The calculation is based on the performance of functional units of the infrastructure. IVI computation starts with the calculation of the residual life and associated present value for all pipes (or any other system element), individually. Subsequently, the global infrastructure IVI is calculated. The IVI of an individual pipe corresponds to its RUL (remaining useful life), expressed as a percentage of total useful life. IVI requires an additional parameter, the estimated replacement value for each asset. As the estimates were not available at the time of this preliminary analysis, sewer length was assigned to this parameter just for the purpose of being able to produce a base IVI calculation. The IVI of mature, well-maintained infrastructures should be in the vicinity of 50% (40-60%). Higher values point towards one of the following situations: Young infrastructure. Old infrastructure subject to a recent and significant expansion phase. Old infrastructure subject to over-investment in rehabilitation. Visual Decision Support Tool for Supporting Asset Management Performance, Risk, and Cost Analysis 3-13

47 Low IVI values translate essentially the accumulated lack of capital maintenance. Figure 3-10 represents the evolution of the value of a specific asset. Its maximum value occurs on the installation date and the minimum, 0, for the end of its useful life. The mean asset value is 50 % of the replacement. A linear depreciation is assumed. In a mature, well-balanced infrastructure system, assets in all stages of their life cycles coexist, more or less evenly distributed. The global IVI is, according to the law of large numbers, a value around 50%. In practice, not all assets of the same type and age are in the same stage of their life cycle at the same time. If the observation period contains several asset life cycles, the IVI will tend to this central value, regardless of its initial value. Figure Evolution of IVI Over Asset Useful Life. Whenever investments in new assets took place over short periods, the need for reinvestments is also concentrated in certain periods of time. Only after several asset cycles the dispersion of re-investment needs occurs naturally. Modelling IVI evolution over time (e.g., years) for status quo as well as for diverse investment scenarios provides a useful picture of the long-term impact of today s rehabilitation policies. Furthermore, it helps decision-makers to understand future needs and provides additional sensitiveness to alternative strategies under consideration. IVI may be applied in several contexts, in order to: Support long-term reinvestment planning, in the framework of strategic asset management. Assess the standardized value of an infrastructure asset in the beginning and at the end of a concession period. Set up contractual, regulatory of managerial targets aiming at infrastructure asset. Support national or regional rehabilitation policies. The long-term view of the impact of reinvestment policies helps put into context the consequences of reinvestment savings, and compare less costly solutions (e.g., materials or equipment of lower quality and shorter duration) to solutions requiring a higher initial investment but lower depreciation rates due to longer expected lives. In the context of enterprise valuation, it is normal to check whether the level of investment is of the same order of magnitude as the total asset depreciation, i.e., of the investment in new assets or rehabilitation of existing assets. This situation corresponds to stabilized, mature companies, where the global value of the assets is in a steady state. Details on the formulation, the underlying assumptions and applicability can be found in Alegre, et al. (2014). As referenced in Section (Tables 2-2 and 2-3), IVI can contribute to implementing SIMPLE s AM Step 7 (O&M Investment); Step 8 (Optimize Capital Investment); and Step 9 (Determine Funding Strategy). 3-14

48 3.4.2 Software Implementation Initial IVI analysis was carried out using the validated dataset of GCDWR gravity sewers including ID, installation date, length, material and diameter; UL values used were taken from the 2012 TAMP, Table ES-3 Service Life Estimates for Various Pipe Materials (UL for RPM is not given, and a value of 75 years was assumed provisionally, to be updated by GCDWR if or when appropriate. The pre-existing IVI software module (IVI1), although based on the general assumptions and principles presented above, had been designed to assess IVI for linear assets (pipes) only and for a single year. Studying IVI evolution over time required manual adjustment of the input and the collection of year-by-year results elsewhere. Excel import / export were not available and there were no graphical representation capabilities. The current IVI module was redesigned from scratch, taking into account the comments, suggestions and real-life scenario motivations of GCDWR s team. The experience collected through over a year of usage of the previous tool by 30+ utilities worldwide also contributed greatly to the redesign. IVI analysis was refined in collaboration with GCDWR and the project team, to include different service life estimations, replacement costs and the remaining assets of the system, as well as the ability to address all types of infrastructure assets (linear and vertical), specified individually (as from e.g., an asset register) as well as by any cohorts or classes of assets, depending on the data available and on the targeted granularity in the results. The expected useful life of an asset of asset class can now be expressed as a single value or as a range, an option that most utility managers seem to favor. The tool was also expanded to allow long-term IVI analysis, and the new graphical (timechart-based) interface both encourages what-if scenario exploration and is a much more powerful means of expressing results. The understanding of the results of IVI analysis, particularly by non-experts, benefitted greatly from this development. Modeling features include the specification of rehabilitation rate, annual investment or IVI targets, combined with the possibility of requiring automatic end-of-life replacement. The software requires as input data a list of assets (discretized or grouped by cohorts) with the installation date, estimated useful life range and unit replacement cost. Constant prices are used. It assumes linear depreciation, but is flexible in terms of how the user assesses replacement costs: straightforward import/export MS Excel files allow to externally using cost functions or any other support calculations. Assets that include both construction works and equipment may be split into components. Figure 3-11 shows an example input data worksheet in the IVI tool. Figure AWARE-P IVI Tool: Input Data Worksheet. Visual Decision Support Tool for Supporting Asset Management Performance, Risk, and Cost Analysis 3-15

49 When an analysis is based on a target index (IVI, rehab rate or annual investment), the program replaces the assets that are closer to their end-of-life. In case an asset s useful life is expressed as a range, then a uniform distribution is assumed and a random value is picked from the interval. Figure 3-12 shows IVI analysis for GCDWR s infrastructure, mostly installed in the last three decades. It illustrates the IVI (yellow line) and the reinvestment needs (in pink) if assets were replaced at the end of their expected lives. An expected life of years was used in the IVI calculations, according to what the utility managers considered realistic. Figure IVI (yellow line) and Reinvestment Needs (pink bars) Overtime Assets Replaced at the End of the Expected Life. 3-16

50 IVI with a high value, 95% in 1980 and 77% in 2014, demonstrates the infrastructure s youth; IVI decreases to 35% in 2050, when the quality of service is likely to already be affected in terms of asset failure frequency. The almost inexistence of reinvestment needs in the first years contrasts with what will happen from 2050 onward. The utility must be prepared for these needs, while acting to disperse them. Possible strategies are anticipating some of the interventions or changing the maintenance and renovation procedures (in order to change the expected residual useful life). Figure 3-13 shows the same infrastructure submitted to three different rehabilitation strategies: a rehabilitation of 0.5% of the replacement value per year (Figure 3-13a); a target value of 50% for the IVI, with (Figure 3-13c) and without (Figure 3-13b) forcing replacement at the service life. (a) Strategy: rehabilitation rate = 0.5% per year (b) Strategy IVI 50 % without automatic replacement at the end of the service life (c) Strategy IVI 50% with automatic replacement at the end of the service life Figure IVI (yellow line) and Reinvestment Needs (pink bars) Overtime Three Different Reinvestment Strategies. Visual Decision Support Tool for Supporting Asset Management Performance, Risk, and Cost Analysis 3-17

51 Results show that any strategy that involves replacing an asset before the end of its useful life either leads to assets being operated beyond its service life or to a globally more expensive solution than the operate-to-fail approach. Long-term investment planning requires other considerations than total cost. Service quality, reliability (e.g., service interruptions, disruption to third parties due asset failure), as well as organizational aspects (e.g., uneven distribution of human resources and capital needs) need to be taken into account. For instance, in Figure 3-13a, a rehabilitation rate of 0.5% leads to an abrupt decrease in the percentage of assets in service (blue line) and a steady decrease in IVI. Statements from utility managers such as There is one thing I cannot understand with regard to IVI: how could we have managed our utility until now without it? (Marcelo da Velha, Inframoura, Portugal), or IVI is one of the most common-sense parameters that I think I have ever overlooked. (Steve Sheets, GCDWR, USA) illustrate utility feedback about IVI. Apparently obvious, conceptually easy to understand, IVI is proving to be a rather powerful awareness, communication, negotiation and long-term planning tool. This open-source software implementation, which is simple to use and provides visual, easy to understand results, makes it available for application by utilities of all types. 3.5 INSP: Inspections Analysis Tool Infrastructure inspection has proven to be a best practice that reduces failures over time. However, the act of performing inspections themselves is a costly endeavor. Utilities want to invest inspection dollars wisely by optimizing the inspection process. The INSP tool helps managers identify assets that need inspected more often and others that can be inspected less frequently so that inspection dollars are well spent The Inspections Analysis and Prediction Module Proactive maintenance and rehabilitation planning play a crucial role in asset management (AM) of urban water services. Wastewater drainage systems often suffer from inadequate cadastral and monitoring information as compared to their water supply counterparts. On the other hand, they benefit from accessibility to CCTV and direct inspection, which are sources of information on condition that allow for the implementation of condition-based methods to determine deterioration trends. Through the application of standard defect classification and scoring systems and condition assessment protocols such as those defined by Water Research Centre (WRC, 2013) or NASSCO s PACP Pipeline Assessment and Certification Program (NASSCO, 2010), each inspected sewer pipe is assigned a grade, as a result of expert interpretation of a number of direct or video visual observations. There are similar protocols for manholes and for laterals, e.g., in the PACP system, grades of 1 ( minor defect grade ) to 5 ( most significant defect grade ) are assigned in two categories structural (e.g., a wall fracture) and operations and maintenance (e.g., deposits). Depending on the intervention/rehab protocol adopted by the utility, grade 3 usually places the sewer under closer observation; grade 4 slates the pipe for rehab or corrective action in the near future (up to five years); and grade 5 often defines an end-of-life condition. Pipes graded 4 or 5 are commonly labeled critical in this context, as an expression of the seriousness of their condition (not to be interpreted as a consequence-of-failure assessment). Inspections are sufficiently beneficial to inspire full coverage strategies, with a focus on direct evaluation of the assets. They are also often used to feed the ranking of assets based on 3-18

52 risk-based approaches crossing an estimation of likelihood of failure (using the condition grade and age/material-based decay curves from the literature), with an estimation of the consequence of failure (measured by compounding several variables such as zoning, sewer proximity to sensitive areas or sewer capacity). However, inspections are costly and it is well worth trying to prioritize them by directing the effort to those sewers most likely to be in a condition that requires action (grades 4 and 5, termed critical by the utility), based on correlations between previous results and potentially explanatory covariates (e.g., age, material), rather than merely following a contiguity strategy when selecting the next sewer zone or sub-basin to inspect. Concurrently, given the need to prioritize and plan O&M and tactical asset management interventions, there is always a need for the best possible snapshot prediction of the network s current condition. The new Inspections module (INSP) comprises a new set of tools, mainly directed at condition-based inspection analysis and prediction. The module includes a prototype topological model; a system to evaluate the mutual information contained in preset user-selected covariates; two distinct predictive models, Random Forest and GompitZ; the development of a zone-based approach to inspection prioritization; and a GIS-based and graph-rich system for inspections and predicted condition result visualization. The module was developed, tested and validated by Addition and LNEC, and was made available in the project s software deployment at the end of October Subsequently, the module underwent further improvement, benefiting from successful backup research into statistical (GompitZ) and classification (Random Forest) algorithms and their testing using the large datasets made available by GCDWR. The usefulness of the tool to prioritize future actions, both to schedule inspections and to plan rehabilitation interventions, has been demonstrated and tested for reliability. GHD provided a second, larger dataset from another U.S. utility, which was useful in varying the scope of data for testing and provided relevant challenges to the processes tested. GCDWR s input to the development of the module in its latter stages was significant, with important contributions to focusing the tool on areas of direct return to the utility, in particular, in moving from asset-by-asset to zonal prioritization, yielding more reliable assessments and more actionable results. A detailed application of the Random Forest (RF) classification/regression-based data mining method was carried out on a GCDWR large updated dataset of sewer condition based on their thorough CCTV inspection program and AM planning. The method was used both for estimating the importance of potential covariates in the forecasting process, predicting condition of sewers and in directing investment in inspections. The level of confidence in the predictions was tested through systematic generation of predictions and comparison with recorded results using mathematical bootstrapping techniques. This implementation of the Random Forest method is discussed in Santos (2013) and in Vitorino et al. (2014). A more general discussion of the method can be found, e.g., in Breiman (2001). The quality of the results was (and continues to be) tested through application at Gwinnett County s ongoing inspection effort, where prioritized sub-basins are compared with the previously used sewer selection procedures. Bulk (accumulated critical length) predictions have proved to be particularly suitable to the utility s operational environment. Visual Decision Support Tool for Supporting Asset Management Performance, Risk, and Cost Analysis 3-19

53 The INSP tool also includes a second analysis/prediction model, GompitZ, to allow for cross-validation and complementary insight. The GompitZ model (Le Gat, 2008) uses a nonhomogeneous Markov chain mechanism to define the relationship between the current state and the expected service time of sewer pipes, from the inspections grades. At the single pipeline level, it models pipe deterioration as a succession of a few discrete condition grades (4 to 6 are usually considered), within the framework of Non-Homogenous Markov Chains. The probability of a pipeline to be in a given condition state or better is formalized as a survival function of the pipe age, derived from the Gompertz distribution. The survival function depends on covariates influencing either the initial condition state or the deterioration rate. GompitZ is made available by the software and constitutes in itself a major addition to the portfolio. It was tested at Gwinnett County with good results and was validated against previously existing datasets and against its own pre-existing software implementation, a research-oriented tool made available by IRSTEA (Le gat 2008). RF is more detailed in this report due to being a novel proposal and due to having yielded in GCDWR s case outstanding results in faster run times than GompitZ. The latter is available in the software and offers an alternative to the users. The inspection analysis tool guides users through the process of adding input data, running the model, and assessing generated information. The RF model is trained with the existing CCTV-based condition assessment data, combined with pipe records and user defined covariates: Pipe Material discreet covariate. Pipe Zone discreet covariate (optional). Pipe Length continuous covariate. Covar 1 user-defined continuous covariate (optional). Covar 2 user-defined continuous covariate (optional). Covar 3 user-defined discrete covariate (optional). Age at previous inspection, if available, a floating point decimal value in years. Condition Class of previous inspection, if available. Age at Inspection a floating-point decimal value in years. Condition class the training classification parameter. The inclusion of a mutual information calculation in the model, providing a measure of the informational gain in incorporating data from a particular source for the predictions, is a key feature in driving the quest for the best covariates (explanatory variables/data). Essentially, it provides a means to decide which variables to include in the analysis in the example in Figure 3-16, sewer age and material appear to be relevant, whereas diameter, zone, slope (user-defined covariate 1) or age at previous inspection were all irrelevant in explaining condition for the sample analyzed. Note that the classification and scoring systems that generate the condition assessment include over one hundred different standard defects from a large variety of causes. The ability of a given covariate to explain the condition of a sewer depends to a high degree on local conditions, and making assumptions in this respect may be plainly misleading. 3-20

54 3.5.2 Case Study Applications A summary of basic descriptive statistics for GCDWR network is shown in Figure 3-14, which allows for important insight on asset distribution by material, inspection effort significance, relative lengths inspected, frequencies of condition-critical pipes found (grades 4 and above were defined as condition-critical by GCDWR), and the time windows inspected. Figure Descriptive Statistics. GCDWR Critical grades were defined as 4 and 5. Figure 3-15 displays other descriptive statistics of the inspection records: total vs. inspected length and inspected condition, per installation year. Figure (TOP) Total vs Inspected Length and (Bottom) Inspected Condition Per Installation Year. Note that the number of sewers inspected in 2014 was very small when this analysis was run, and had incidentally recorded a non-representative proportion of grade 5 sewers. Visual Decision Support Tool for Supporting Asset Management Performance, Risk, and Cost Analysis 3-21

55 The results were calculated globally for the system, and predictions were made for each individual sewer. Figure 3-16 summarizes aggregated results from applying the RF model: mutual information on covariates and (right) estimated distribution of critical sewers (grades 4 and) per sewer pipe material. Figure Aggregated Results: (left) Mutual Information on Covariates and (right) Estimated Distribution of Critical Sewers (grades 4 and 5) Per Sewer Pipe Material. Covariate 1: Pipeline slope; Covariates 2 and 3: not used. The mutual information table in Figure 3-16 shows that for GCDWR specific case, with a large network and a large number of inspections, pipe material or diameter were not found to significantly explain condition. However, these two covariates are often found in the sewer deterioration curves commonly found in the specialized literature that are frequently used to drive rehab/replacement policies in the industry. Individual and zone bulk predictions were produced and made available on the screen or in Excel format. By adding a GIS sewer map, GCDWR can also view estimated conditionalcritical pipes on the map (Figure 3-17). Figure Predicted Probability of Being in a Critical Grade (4 or 5) for Individual Sewers, for a Given Year (current or future), the Color Thresholds of Probability are User Defined. 3-22

56 A measure of the benefit of inspecting a given length of sewers by following the software predictions is graphically depicted on the chart in Figure Non-directed inspections have an expected success rate in finding condition-critical pipes (4 or 5) of 11.31% of the total inspected length (as seen in Figure 3-14), whereas using RF ranking (Figure 3-18) to select the first uninspected 200mi would yield >50% of expected success rate. Following this prioritized strategy, the expenditure per critical find is, in this case, decreased almost five-fold for those first few hundred miles. The process is further informed as more pipes are inspected-particularly as a greater percentage of condition-critical pipes are added to the sample. Figure Prioritization Performance: Estimated Probability of Finding Critical Pipes by Accumulated Inspected Length (miles of sewer) Using the RF Model. Figure Predicted Percentage of Total Length of Sewer in a Critical Grade (4 or 5) Per Network Zone (drainage sub-basin), for a Given Year (current or future). Visual Decision Support Tool for Supporting Asset Management Performance, Risk, and Cost Analysis 3-23

57 The color thresholds of critical % of total zone length are user-defined (Figure 3-20). Figure Color Thresholds of Critical Percentage of Total Zone Length. It is believed that the developed model is a useful addition to the range of tools in AWARE-P, namely for prioritizing future actions, both to schedule inspections and to plan rehabilitation interventions. It has the main advantages of 1) being relatively simple to deploy, based on two simple input files that can be easily extracted from the GIS/asset register and from the maintenance system; 2) providing a good prediction performance; and 3) enabling a learning curve with respect to useful explanatory variables, by the utility, within a relatively straightforward process. Its short-, medium- and long-term predictions of sewer condition are based on direct, measurable, analysis of the utility s own data. The model outcomes can also feed more thorough risk approaches by combining the condition prediction, as a measure of the likelihood of failure, with informed consequence-offailure (CoF) estimation for specific risks (e.g., financial consequences of pipe structural failure; local consequence of blockages). The developed tools and application were also tested using available data from a small utility in the U.S. A small utility is classified as a utility serving population less than 50,000. This small-sized utility has approximately 120 miles of sewer lines, and a majority of the sewer lines are 8-inch PVC or clay pipes. The majority of the utility s sewer lines are structurally sound and operationally functional based on available knowledge. The utility also has recently mapped the sewer system in GIS. A recent CCTV study was conducted on a section of the utility s sewer system and the resulting PACP codes were available. The AWARE-P tool has defined a format on the basis of which a utility s GIS data needs to be uploaded including the field names and the field types for pipe ID, material, length, diameter, installation date, and two covariates. The defined field names and field types for the inspection data include pipe ID, inspection date and inspection score from a range of 1-5. The utility s GIS and inspection data was exported to MS Excel and was processed to make sure it adheres to AWARE-P s format for field names and field types. The utility s processed data was then imported to the AWARE-P tool using the import functions in the tool. The AWARE-P tool was able to quickly process the uploaded GIS and inspection data, perform the random forest simulations, and display the results as color coded map (Figure 3-21) based on criticality. In addition, the resulting data can be displayed in bar charts and/or tables for further analysis. The tool was able to provide statistics of the criticality based on material and installation date by length. 3-24

58 Figure Pipes Color Coded Based on the Criticality. The tool also provides the ability to change the base maps to satellite, street or hybrids with zooming and panning capabilities. The simulation results were also exported as a MS Excel file for further analysis. Based on this case study the AWARE-P model is simple to apply to even a small utilities data with some simple formatting of their existing data. The results can be presented in a graphical format and any resulting data can be exported for further analysis if the user chooses Network Topological Prototype Model Gravity sewer network models are more complex than their water distribution counterparts. Building a model and calibrating it are tasks that require larger quantities of network data and higher accuracy levels on many of those data, such as elevations, pipe sections and operational data 12. It is uncommon to find a utility with a thoroughly complete model of the entire sewer system network, not to mention the lack of flow metering data. This made it worthwhile to look for a simplified modeling approach in the scope of IAM analyses, with fewer data requirements, that may help produce failure consequence analysis metrics for sewer systems. The prototype developed in this project was intentionally simplified, in order to allow the utilities to rapidly establish a system view, rather than an asset-by-asset view. It informs on basins/sub-basins, system dependencies, zones/consumers served by each asset or group of assets, serviceability; therefore, it improves the ability to rapidly generate consequence-of-failure assessments, by crossing with geo-referenced information available (land-use, critical users, zoning, traffic, census, etc.). 12 Flow metering in sewers is also more onerous and seldom as widespread as in water supply; although Gwinnett County s sanitary sewer flow metering system is indeed quite comprehensive Visual Decision Support Tool for Supporting Asset Management Performance, Risk, and Cost Analysis 3-25

59 A topological model was proposed, in order to represent and model hydrology and hydraulics, and consequently to assess network performance (performance indices) and component importance. Component importance evaluates an individual component s failure consequence in the network, used in the assessment of risk. The existing AWARE-P CIMP model calculates a component importance metric for each individual pipe in a water distribution network. The model is supported by a network s hydraulic model, using full simulation capabilities. The topological model has the ability to validate GIS/asset record data. Reading GIS shapefiles, the model uses them to represent the network and perform topological, connectivity, and geographical analyses. The model relates each pipe with linked components (manholes, pumping stations and other basin end-points recorded in GIS layers shapefiles or other records), and automatically generates a coherent topology based network model. This association defines gravity sewer network units by drainage basin. The analysis is purely topological and does not take into consideration the sewer slopes or invert elevations. Often those elevations are not totally available or reliable (in GCDWR s case, in about 10% of the sewers, as mentioned in Section 3.1), and the purpose is to provide a mechanism that is fully scalable to an entire generic system. The implemented procedure works as described below. For the sewers/manholes model: For each sewer, search for manholes on each end, registering the relationship on the sewer and on the manholes. Whenever a manhole is absent (there is a small proportion of cases), a virtual manhole is created to be able to follow the relationship between the current pipe and other(s) that are connected to it. Sewers without any relationship to manholes or other sewers are registered and discarded-in GCDWR s case, this happens only at one sewer, out of a total of 75,864. When the sewers/manholes model is completed, pump stations/basin analysis is the next step: For each pump, find the directly connected sewers. Follow the sewers using the previously constructed model, to associate it with the pump station, generating a pump station/basin model and identifying upstream/downstream manholes. Whenever a topological link is present between two or more pump stations, that information is registered and needs to be solved manually (at GCDWR, there were three of these situations, out of 229 pump stations). In GCDWR s case, by analyzing services areas, three sewers were eliminated, producing a coherent model that comprises 72,512 (95.58%) of all sewers into 216 basins. The remaining 13 pump stations would appear to be exclusively related to pressure sewers systems, and will be checked in conjunction with GCDWR. Initial module development was undertaken and a prototype was made available to the project team. Given time constraints, other tools were given a higher priority by the project steering committee, GHD and GCDWR (as documented in the project s first and second quarterly reports). Given such constraint, the project team opted to focus the efforts on FAIL, IVI 3-26

60 and INSP, and the topological model as an independent module was consigned for further development. The topological model algorithm was incorporated in the INSP module, for which it plays a crucial role. This prototype is a candidate to provide the basis for the network modeling procedure that will support component importance, consequence of failure and the inspections module, as well as for all the other tools that need a network representation. Visual Decision Support Tool for Supporting Asset Management Performance, Risk, and Cost Analysis 3-27

61 3-28

62 CHAPTER 4.0 CONCLUSIONS In view of the results described in this report and made available as project deliverables, it is the project team s view that the objectives of the project have been fully achieved. The AWARE-P methodology and software have been comprehensively and successfully tested on Gwinnett County Department of Water Resources gravity sewer system. Selected key tools were also tested by U.S. project team members not familiar with the software, using available data from a small-sized utility in the U.S. Based on this case study, it can be stated that the AWARE-P model is simple to apply to even small utilities data. The development and validation of the new inspection tools counted on a large inspection database and GIS information from another U.S. region. Significant, novel and leading-edge knowledge has been generated, particularly with regard to inspection data analysis and decision making processes. Several new modules and features of the AWARE-P software have been developed and tested for use in a professional context (IVI2 Infrastructure Value Index 2 and INSP Inspections Analysis), with validation by GCDWR. Several others (e.g., topological model) were explored in the first part of the project, in order to give GCDWR, GHD and WERF the possibility of choosing the most relevant paths to proceed within the framework of the project. The recommendations from the Project Steering Committee were also taken into account in this process. Tables 4-1, 4-2, and 4-3 summarize the assessment carried out by WERF, by GHD (as a key consultant in the U.S. market and lead partner) and by GCDWR during the final on-site project meeting. GCDWR GHD WERF Suggestion for further developments Table 4-1. Project Assessment According to the Main Stakeholders Involved: Global Approach. AWARE-P Global Approach Merit, Shortcomings, Potential for Gwinnett, for GHD, for the U.S. Three dimensional approach is very beneficial especially at the planning level stages. A very good helpful tool for tool building as we mutual attempt agreement to relate the before various going components into the of decisional level of service process. to monetization of our project options. The tool is a good mechanism for thinking about the planning and operations in a more system-wide basis rather than the asset-by-asset or tactical level view. Allows you to look at investment options from different angles, the cube slides. A grayscale option? (Because of colorblind operators). Visual Decision Support Tool for Supporting Asset Management Performance, Risk, and Cost Analysis 4-1

63 If analyzed in conjunction with Table 2-1, Table 4-2 demonstrates that the project benefited from other concurrent projects, running in parallel, which produced modules also considered of great practical value (e.g., Shapefiles). Table 4-2. Project Assessment According to the Main Stakeholders Involved: AWARE-P Software Pre-Existing Tools. PLAN GCDWR GHD WERF Failure Analysis GCDWR GHD PI GHD Shapefiles GCDWR GHD Assessment of Pre-Existing AWARE-P Tools I like the red-yellow-green but would prefer a 5-scale (as in NASSCO), which is becoming widespread in many tools in the U.S. A tool to structure the process, absent in other approaches. Agree with the structure point, lends itself to work with the various parameters that can be calculated. Gets the discussion started, get people thinking about the AM process. Recommendation: Make sure that users understand the cube may have unlimited number of alternatives. We always show cubes with three levels. Prefers the classification methods such as Random Forest (implemented in AWARE-P/INSP), as it does not require preconceived assumptions about co- variates. On the other side, it does have the hard-core rigorous methods behind it (LEYP, Poisson). Pre-existing tools were limited and mostly custom built based on available utility information. Provides users with something other than a blank sheet of paper. Educates users on what performance is about and gives users ideas on what to build on. Ability to export to MS Excel and then move into a shapefile is useful; ability to import and handle shapefiles means that geo-referenced information can be displayed and made useful. Data is your friend, but interactive data is useful (as in, when the spreadsheet talks to the geo display). This is an important feature that has to be in there. The ability to visualize and to export data is not useful if you cannot visualize it geographically. 4-2

64 Table 4-3 recognizes the value of the topological model, recommended for future development. Table 4-3. Project Assessment According to the Main Stakeholders Involved: AWARE-P Software New Tools. Topological model (proof of concept) GCDWR GHD New IVI GCDWR GHD WERF Inspections GCDWR GHD Suggestions for further developments Assessment of the New Tools Can see where it will be useful at many different levels; it would be cool to have. Useful in itself; understandable that effort is channeled to other tools within the project. One of the most common-sense parameters that I think I have ever overlooked. Simple, but not simplistic? Usually people talk about RUL of one asset or an asset class; but not about the whole set of assets in the utility. Agree with GCDWR s statement. Consistent with the tools available in SIMPLE. What needs to be addressed today to avoid a crisis 40 years later. The most basic underpinning of the whole AM program. Any asset intensive industry has to use this as a guide. Looking forward to testing Random Forest (RF) against own perception of likelihood of criticality. Possibility to add factors without a preconceived notion of what should drive the inspection. Assigns value to getting the 2nd round of inspections and start having data that may reflect deterioration. Looking forward to testing RF against own perception of likelihood of criticality.* Like the RF as it avoids the preconceived issues. Concern about forecasting beyond range of historical data available (important to warn users to not extrapolate using RF, as it is not an inference model). Another advantage is the ability to consider extra covariates. Likes the fact that it is possible to bring up inspections and GIS. Ability to add discrete covariates. Ability to generate bar charts and pie chart for the statistics and the results. Ability to filter results on map by any of the covariates. * GCDRW tested Random Forest against their perception of likelihood of criticality in the last period of project. The feedback received was rather positive. Visual Decision Support Tool for Supporting Asset Management Performance, Risk, and Cost Analysis 4-3

65 4-4

66 APPENDIX A DISSEMINATION PLAN A dissemination plan was detailed during the team meetings held adjacent to the March onsite meeting. A-1 Dissemination in Conferences A list of presentation opportunities in forthcoming technical conferences has been developed, including specifically: WDSA (Bari, Italy) July 2014; TRICON (abstract accepted) August 2014; Utility Management Conference (Feb. 2015); ACE (AWWA Annual Conference and Exhibit) (June 2015); WEFTEC (September / October 2015); LESAM 2015 (Japan). WDSA (Bari, Italy) (two papers submitted and orally presented) July Champions: Diogo Vitorino, Sergio Coelho and Helena Alegre. TRICON (abstract accepted; no written paper) August Champion: Brad Jurkovac. Utility management conference (February 2015): focused on AM; utility managers; CEO; focus of the paper should strategic planning. Tentative title: You want me to pay what? - Tools to justify and communicate the need for asset investment. Champions: Brad Jurkovac and Dave Kerr. ACE (AWWA Annual Conference and (to be submitted) (June 2015). Champions: Brad Jurkovac and Dave Kerr. Target public: water professionals; focus: failures. Contents: present models and results for drinking water and wastewater. Champions: Brad Jurkovac and Dave Kerr. WEFTEC (Sept / October 2015); Include Gwinnett; focus on the Gwinnett business case. Tentative title: Common-sense parameters that I overlooked: The case study of AWARE-P in Gwinnett. Champion: Brad Jurkovac and Dave Kerr. Conference of Mayors. LESAM 2015 (November 2015, Japan). Champions: Helena Alegre and Sérgio Coelho. Other presentations / papers on the RF work (to be defined at a later date). A-2 Dissemination for the PSC A web seminar for the PSC was carried out to present the final project results. A-3 Dissemination for the U.S. EPA Sergio Coelho, Helena Alegre, Diogo Vitorino, resuming the conversations in Cincinnati three years ago. A-4 Main Means of Communication from WERF WERF website. Visual Decision Support Tool for Supporting Asset Management Performance, Risk, and Cost Analysis A-1

67 Laterals newsletter. LinkedIn. Survey to WERF subscribers. EPA Water Headlines: Communications with WEF. A-5 Other Means of Communication AWARE registered users. Portugal: ERSAR, Água&Ambiente, PPA, project websites, UWCommons. Australia: WSA Australia IPWEA (Walter Graf will take care of these contacts). A-6 Communication Within GHD Quarterly newsletter (U.S. version and global) Samples: Americas_issue4_English.pdf. Electronic system. A-2

68 APPENDIX B SOFTWARE USER GUIDE: TOOLS QUICK STARTS Visual Decision Support Tool for Supporting Asset Management Performance, Risk, and Cost Analysis B-1

69 B-2

70 Visual Decision Support Tool for Supporting Asset Management Performance, Risk, and Cost Analysis B-3

71 B-4

72 Visual Decision Support Tool for Supporting Asset Management Performance, Risk, and Cost Analysis B-5

73 B-6

74 Visual Decision Support Tool for Supporting Asset Management Performance, Risk, and Cost Analysis B-7

75 B-8

76 Visual Decision Support Tool for Supporting Asset Management Performance, Risk, and Cost Analysis B-9

77 B-10

78 Visual Decision Support Tool for Supporting Asset Management Performance, Risk, and Cost Analysis B-11

79 B-12

80 Visual Decision Support Tool for Supporting Asset Management Performance, Risk, and Cost Analysis B-13

81 B-14

82 Visual Decision Support Tool for Supporting Asset Management Performance, Risk, and Cost Analysis B-15

83 B-16

84 Visual Decision Support Tool for Supporting Asset Management Performance, Risk, and Cost Analysis B-17

85 B-18

86 Visual Decision Support Tool for Supporting Asset Management Performance, Risk, and Cost Analysis B-19

87 B-20

88 Visual Decision Support Tool for Supporting Asset Management Performance, Risk, and Cost Analysis B-21

89 B-22

90 Visual Decision Support Tool for Supporting Asset Management Performance, Risk, and Cost Analysis B-23

91 B-24

92 Visual Decision Support Tool for Supporting Asset Management Performance, Risk, and Cost Analysis B-25

93 B-26

94 Visual Decision Support Tool for Supporting Asset Management Performance, Risk, and Cost Analysis B-27

95 B-28

96 Visual Decision Support Tool for Supporting Asset Management Performance, Risk, and Cost Analysis B-29

97 B-30

98 Visual Decision Support Tool for Supporting Asset Management Performance, Risk, and Cost Analysis B-31

99 B-32

100 Visual Decision Support Tool for Supporting Asset Management Performance, Risk, and Cost Analysis B-33

101 B-34

102 Visual Decision Support Tool for Supporting Asset Management Performance, Risk, and Cost Analysis B-35

103 B-36

104 Visual Decision Support Tool for Supporting Asset Management Performance, Risk, and Cost Analysis B-37

105 B-38

106 Visual Decision Support Tool for Supporting Asset Management Performance, Risk, and Cost Analysis B-39

107 B-40

108 Visual Decision Support Tool for Supporting Asset Management Performance, Risk, and Cost Analysis B-41

109 B-42

110 Visual Decision Support Tool for Supporting Asset Management Performance, Risk, and Cost Analysis B-43

111 B-44

112 Visual Decision Support Tool for Supporting Asset Management Performance, Risk, and Cost Analysis B-45

113 B-46

114 Visual Decision Support Tool for Supporting Asset Management Performance, Risk, and Cost Analysis B-47

115 B-48

116 Visual Decision Support Tool for Supporting Asset Management Performance, Risk, and Cost Analysis B-49

117 B-50

118 Visual Decision Support Tool for Supporting Asset Management Performance, Risk, and Cost Analysis B-51

119 B-52

120 Visual Decision Support Tool for Supporting Asset Management Performance, Risk, and Cost Analysis B-53

121 B-54

122 Visual Decision Support Tool for Supporting Asset Management Performance, Risk, and Cost Analysis B-55

123 B-56

124 Visual Decision Support Tool for Supporting Asset Management Performance, Risk, and Cost Analysis B-57

125 B-58

126 Visual Decision Support Tool for Supporting Asset Management Performance, Risk, and Cost Analysis B-59

127 B-60

128 Visual Decision Support Tool for Supporting Asset Management Performance, Risk, and Cost Analysis B-61

129 B-62

130 Visual Decision Support Tool for Supporting Asset Management Performance, Risk, and Cost Analysis B-63

131 B-64

132 Visual Decision Support Tool for Supporting Asset Management Performance, Risk, and Cost Analysis B-65

133 B-66

134 Visual Decision Support Tool for Supporting Asset Management Performance, Risk, and Cost Analysis B-67

135 B-68