Houston Develops and Implements An Automated Process For Collection System Risk Assessment and Mitigation Planning Using An Integrated Analytics Platform, InfoAsset Planner, and Artificial Intelligence

Background
City of Houston is the fourth largest city in the United States of America. Houston Water (HW), within Houston Public Works (HPW) department of the City of Houston, facilitates wastewater services to approximately 2.3 million customers. HW maintains wastewater collection system that includes approximately 31 million LF of sewer pipes, 127,000 manholes, 385 lift stations, and 39 treatment plants. Houston Water Planning (HWP) team within Houston Water develops strategies and plans for O&M and CIP projects.
Houston Water applies cutting edge technologies for forecasting future scenarios utilizing advanced models including Geographical Information System (GIS), Info-Works ICM, Info-Asset Planner, Business Intelligence tools and techniques, and others. HPW recognizes the importance of asset and data centric planning using both capacity and condition for operational and capital decisions to optimize cost and increase level of service and comply with regulatory requirements.
This paper illustrates the process of information integration and system implementation for risk-based asset management of wastewater collection system for the City of Houston. Conventional approach to infrastructure asset management relies on time-based responsive and preventive maintenance, and reactive decisions that involve high rates of inaccuracy and inefficiency. The ever-increasing volume of data from multifaceted sources and technologies and availability of AI/ML based predictive models enhance the opportunities for City of Houston Wastewater Operations to venture into predictive and prescriptive approaches with the objective of lowering cost, reducing system downtime, and improving level of service.
Commonly used approach to detect wastewater asset failures and consequent rehabilitation decisions are based on age-dependent models of pipe attributes, location, repair, and failure history. With the availability of CCTV condition data along with numerous physical and asset performance information, and analytical tools like GIS, AI, ML, BI, InfoAsset Planner, and cloud-based computing environment, Houston Water envisions a data-centric predictive planning framework to optimize comprehensive long-term social, environmental, and economic perspective of planning decisions. The current planning process for operation and maintenance activities are designed at basin-based aggregated level. The proposed planning process will enable micro to macro level planning in roll-up and drill-down fashion capable of addressing the condition of individual asset to the system level from a single platform. Analytical Framework Data-centric predictive planning is built on analysis of asset capacity and condition assessment for sewer system with physical characteristics of the assets, spatial attributes, CCTV data, and other performance data. Spatial and asset related attributes of the wastewater system are stored in GIS system that is regularly updated. HPW plans to conduct CCTV condition assessment of 4 to 5 million LF of sewer pipes every year that is transcribed into digital form using AI based automated process and integrated in the planning module.
Condition based planning system is designed using open and flexible architecture to facilitate seamless modification and expansion. In the initial stage, the system is constructed on ArcGIS platform along with InfoAsset Planner and integrated with business analytics platform (Power BI) and custom modules built on Python. The process is in line with ISO 55000, International Infrastructure Management Manual (IIMM), EPA SIMPLE and NASCCO standards for CCTV condition assessment. The analytical procedure is based on asset risk calculation for individual pipe segments using relevant parameters to estimate Consequence of Failure (COF) and Likelihood of Failure (LOF). The process involved developing a linear model to estimate relative weights for LOF parameters based on the relationship between PACP condition scores and estimated risk for the assets for which PACP data was available. The estimated weights were used for citywide pipe segments risk calculation. The decision trees were developed by City's wastewater operations and planning staff knowledgeable about the system. Decision trees were used in InfoAsset Planner for determining appropriate rehabilitation actions through a consistent and unified process. The individual high-risk pipe segments are grouped for project scoping using rule-based integration and prioritization.
Rehabilitation actions can be agglomerated into integrated projects satisfying constraint that include budget allocation, proximity, action type etc. Furthermore, the projects can be scheduled, automated workorders can be framed, and work-process can be monitored for resource allocation through smart work manager. Also, project costs are estimated and fed into Life-Cycle Cost Analysis (LCCA) module to reach at optimum long-term investment decisions. The process also involves asset deterioration model to capture the effects of timing of rehabilitation decisions. The integrated planning process is depicted in Figure 1 Application of the Condition-Based Planning Process
The designed asset condition-based planning process is being implemented for the Houston's sewer pipe network. Based on the preliminary results, approximately 1.63 and 3.08 percent of the total sewer network is in Extreme and High-risk level respectively. The proposed planning process enable authorities to identify segments that need inspection and take preemptive action to avoid failure.
Conclusion and Plans for the Future Development
The paper presents the preliminary design and implementation for condition-based asset management framework for the wastewater collection system in the City of Houston. The developed system will enable HPW to enforce more efficient predictive actions to reduce pipe failures. The risk-based asset management will be further expanded to include the followings in the next stages of development. - Optimum infrastructure rehab strategy formulation under different scenarios through simulation process. Integrated simulation may be executed through use of scripting language. - Extended analytics for decision making can be implemented through integration of Business Analytics tools (Power BI or similar) and InfoAsset Planner. Automated data-exchange between backend asset condition simulation under different budget scenario and front-end analytics will reduce the gap between analysts and stakeholders and enable judicious and optimal decision making. - Integration of cloud-based data collection and repository can be integrated into real-time monitoring and rehab action activation.