Alternate: Leveraging the Cloud to Power Resiliency: Alexandria's Flood Defense

Learning Objectives This paper presents a case study in Alexandria, VA of the approach, assumptions, and lessons learned using an optimization process to complete a flood mitigation alternatives analysis. The flooding for current and no-action future climate conditions, optimization process, results of the analysis, and recommended alternatives will be summarized and presented. Abstract Background The City of Alexandria, Virginia (the City) is experiencing increased flooding from high-intensity precipitation events associated with climate change and the City's natural low-lying position adjacent to the Potomac River. A previous study identified inadequate capacity in the Hooffs Run Culvert (HRC) for the 10-year design storm as a cause of localized flooding throughout the central portion of the 1,600-acre HRC watershed. The City selected improvements to the HRC and the central portion of the HRC watershed as a key flood mitigation project to move into implementation. The alternative analysis was conducted using an optimization process and software to leverage cloud power. Optimization processes and automation tools provide robust alternative analysis capabilities. However, intensive input data validation and results interrogation are required to maximize the efficiency of the optimization process. During the field investigation undertaken for this project, a previously unidentified constriction was observed, which lead to a significant increase in the range of alternatives requiring analysis. This paper summarizes how the optimization process, following best practices, was scaled to accommodate an increased scope of alternatives. Objective The study's objective was to propose design alternatives that (1) would convey the 10-year future rainfall event through the HRC network without flooding above 6 inches and (2) not adversely impact flooding elsewhere. This modeling study was completed in March of 2024. Methods Analysis of design alternatives was conducted using a third-party optimization software and followed the following steps. 1)Model updates and field verification 2)Alternative identification and pre-screening 3)Definition of objectives and model simplification (2D to 1D dual drainage) 4)Preliminary optimization 5)Final optimization 6)Additional feasibility analysis 7)Final alternative selection Design elements included gravity and force main alignments or upsizes (F1), deep storage (pump out), and modular storage (F2). Design elements were screened prior to inclusion in the optimization analysis by considering previous studies, utility conflicts, and available land. To meet the project objectives, the optimization software assessed each alternative based on project cost and remaining flooding after project implementation, represented in the program as a 'penalty'. The flood penalty was formulated using modeling metrics that reflected the City's priorities. The penalty score calculated volume and added additional weight to flood depths greater than 6 inches. To effectively integrate the City's priorities into the optimization objective functions, the optimization was conducted in phases. The Final Optimization identified alternatives representing a range of flood mitigation and cost profiles. Each design alternative combined multiple design elements to form a comprehensive potential solution. To select alternatives, output from the optimization software was interrogated using automated post-processing and PowerBI. High-performing scenarios selected from the optimization software output were further validated using a 1D/2D SWMM model and cloud computing. Finally, likely scenarios were screened through further feasibility considerations such as technical limitations, property ownership, and social impacts. The team worked closely with client representatives to finalize a selection of three alternatives that would be carried forward for stakeholder engagement prior to design. Results\Conclusion This project presented unique challenges, including integration of a 1D/2D model into the optimization process, a highly developed urban watershed, and significant feasibility complications, including a capacity restriction in the HRC identified during field verification. Prior to field verification, only King Street to Timber Branch culverts, and modular storage were expected to be required (F1, F2). Field verification resulted in solutions extending much further downstream, and the magnitude of predicted flooding significantly increased. Thanks to a cloud-based optimization software and post-optimization techniques used to interpret results using PowerBI, the optimization was performed with minimal impact to project schedule. PowerBI was used for visualization of the tradeoff between cost and penalty (F3), and to visually convey how various solutions impacted the study area (F4). Not all, but most flooding greater than 6 inches in the ROW and adjacent to buildings was resolved (F5). The final alternatives consisted of three primary strategies: 1)Comprehensive flood reduction (F5) 2)Comprehensive flood reduction removing high-difficulty elements 3)Early action flood reduction using elements common to Strategies 1 and 2 The following topics will be emphasized in the presentation: - Climate change assumptions and impacts - Optimization process for a flood mitigation analysis - Lessons learned for efficient use of the optimization process - Study findings and recommendations