15 Years (2007-2023) of AI Optimization Technology Innovation at Johnson County Wastewater

INTRODUCTION In 2007 Johnson County Wastewater (JCW) was the first US utility to pilot emerging genetic algorithm optimization technology applied to collection system master planning. The technology enabled JCW to perform an exhaustive analysis of complex and interdependent SSO remedial alternatives including conveyance, storage and inflow and infiltration (I/I) reduction. The innovative approach and project success provided the foundations for ongoing technology development and advancements in the associated engineering practices. The most distinct advantage realized by JCW has been the ability to evaluate thousands of potential I/I reduction targets in each catchment and find the most cost-effective strategy that reduces conveyance and storage upgrades based on a holistic consideration of complex system-wide interdependencies, and evaluating the efficiency of rerunning the optimization with varying degrees of assumed I/I reduction effectiveness to identify sensitivities and determine resilient strategies that can be adapted over time. This presentation reflects on the ongoing application of Optimizer across five collection system basins at JCW, the innovative techniques that have been applied to maximize cost savings, the technology innovations that have occurred, and the possibilities for future progression. SIGNIFICANCE - Demonstrate value added to collection system planning by supporting innovation. - Provide a case study for best practice methodologies for using master planning decision support systems. - Provide a data driven approach I/I reduction with continuous updates to refine and reevaluate assumptions at critical milestones. - Share benefits and challenges of implementing formal optimization technology from a client's perspective. METHODOLOGY The Nelson Complex collection system will be the primary case study presented to highlight the following technology and methodology innovations between 2007 and 2022: - In 2007, the optimization engine relied solely on Genetic Algorithms to evaluate SSO remedial alternatives, whereas, in 2022 a range of intelligent algorithms and implicit heuristics customized specifically for collection systems are incorporated in Optimizer. -In 2007, limited empirical I/I reduction data was available. Running the optimization for a broad range of sensitivity analyses helped show areas for I/I reduction that would be cost-effective irrespective of the assumptions. As more pre- and post-I/I reduction flow monitoring was completed, the I/I reduction effectiveness assumptions were verified and refined to allow more accurate I/I reduction optimization over time. - Progressively rerunning the optimization each year as new data becomes available helps refine the roadmap and makes use of the efficiency of having an optimization model that can easily be updated with modular inputs. - The 2020 Nelson Complex master plan update incorporated a novel application of Optimizer to prioritize the sequence of master plan project implementation to maximize return on investment by identifying the project sequence to maximize SSO reduction at least cost. - The methodology for evaluating storage alternatives changed after the 2007 project to enable a more effective analysis. Rather than modeling storages with specific volumes, the storage is modeled at the largest feasible size and the storage is costed based on the volume received which is controlled by weir height alternatives and downstream capacity constraints. - An advancement in the approach to evaluating gravity alternatives included considering asset condition and available easement to determine whether upgrades would be implemented parallel to the existing asset or as a replace and upsize capacity improvement. - Flow diversion alternatives create significant complexity for an optimization algorithm due to introducing local minima within the search space. To help manage this, the optimization technology now allows for 'logical adjusters' to be included such as, if flow-path A is selected, then flow path B must not be selected, in the case where it is only ever logical to construct an upgrade along one of the flow path alternatives. - To reduce cloud computing cost and run time, the model is trimmed to remove upstream assets that don't have capacity constraints and apply time-series boundary conditions. This approach makes the optimization of much larger and complex systems more feasible. In 2007 only the Turkey Creek basin was evaluated, whereas both the Turkey Creek and Mission Main basins could be evaluated in a single optimization and the interdependencies between basins tied to the capacity of the Nelson Wastewater Treatment Complex were thoroughly evaluated. RESULTS The updated Nelson Complex collection system master plan was completed in 2019. JCW completed master plan optimizations for Leawood basin in 2021 and Indian Creek basin in 2022 with continuous improvement with each subsequent project. The Nelson Complex preferred master plan solution was based on the lowest life-cycle cost combination of conveyance, storage, treatment, and I/I reduction strategies. In addition, the engineering team integrated non-cost objectives into the evaluation. These included community impacts associated with large capital projects, operational flexibility, and resilience for the plan to be adapted over time. Total life-cycle cost of the least-cost optimized was approximately $205M and the final preferred master plan solution accounting for non-cost objectives, illustrated in Figure 1, was approximately $220M. To compare solutions with distinctly different characteristics, optimization scenarios were completed with different sets of alternatives included. Once the optimization model was formulated, the range of allowable alternatives were easily toggled on and off so that different solutions would be produced automatically when the cloud computing run was performed. The cost-effective analysis curve in Figure 2 compares solutions for the conveyance-only scenario (i.e. without storage alternatives, with I/I reduction limited to 10%) with the least-cost optimized solution (all alternatives considered) and the solution where treatment plant capacity is constrained to existing capacity. Once the capital improvement plan was optimized, the next task was to prioritize the implementation schedule. Optimizer was used to evaluate thousands of possible scheduling alternatives to determine the sequence that would maximize return on investment based on the following objectives: - reduce basement backups; - reduce modeled manhole flooding; - reduce modeled SSOs at Flow Protection Devices; - reduce flows to downstream agencies; and - reduce discharge at Peak Extraneous Flow Treatment Facilities, (wet weather/high rate treatment facilities). The return on investment achieved in the prioritization analysis is illustrated in Figure 3. The results from the various JCW basin optimization and prioritization projects have helped inform JCW on future capital works leading to maximum benefits to the community and the environment while also providing the operational flexibility and resilience to be adapted over time. CONCLUSIONS While technology innovations and advancements provide engineers with useful tools, the value that can be achieved using technology is dependent on solid foundational engineering skills and advancements in the techniques applied to support technology implementation. The JCW optimization projects completed from 2007 to 2022 highlight the value that can be achieved when both technology and technical application innovations are supported by the industry and continuously improved over time.