The potential of digital twins for water resource recovery facilities

19 December 2024

With water utilities looking to extract value from their data, these digital representations of the physical world can help address the growing challenges of nutrient management and staffing

As water and wastewater utilities look to balance compliance, cost efficiency, labour availability and sustainability, where are the opportunities for digital twins? One key area is nutrient management at water resource recovery facilities (WRRFs). Operational digital twins are emerging as a powerful tool to help utilities address both workforce limitations and achieve increasingly complex treatment goals for wastewater treatment works.

In this setting, digital twins are digital process simulations that mirror an existing facility. They integrate and use near real-time data (laboratory and SCADA) to provide a more accurate current and future process view of the facility for operations and maintenance. These digital twins may also include some level of auto-calibration to existing conditions, as well as forecasting, to help operators determine the optimum operational settings for the facility.

Live digital twins for nutrient control

Nutrient management is a growing challenge worldwide as communities tackle excess nutrients in water bodies that threaten aquatic ecosystems and drinking water supplies. With discharges from WRRFs a source of nutrient pollution, there is growing regulatory pressure in many regions to improve effluent quality. Solutions that optimise nutrient control systems, and help to reduce energy consumption and operating costs have an important role to play – this includes live digital twins.

Jacobs recently explored the implementation and benefits of live digital twins in the Water Research Foundation’s (WRF) Project 5121: Development of Hybrid Digital Twins for Predictive Nutrient Control. The project developed and demonstrated a hybrid nutrient management digital twin at three full-scale facilities, combining machine learning and mechanistic modelling to optimise nutrient removal processes.

A hybrid approach

By blending machine learning with mechanistic modelling, the hybrid digital twin developed for Project 5121 was able to harness the strengths of both approaches. Machine learning provides valuable forecasting, emulation and optimisation functions, enhancing the overall effectiveness of the digital twin, while the mechanistic modelling handles the inherent uncertainties of wastewater systems and provides assurance that the digital twin is providing sound guidance based on decades of experience.

The project found that existing telemetry and laboratory data collected by WRRFs contain valuable information that can be leveraged to reduce costs and improve operations. This means that facilities do not necessarily need to add more physical complexity, such as costly new sensors, to realise the benefits of advanced control and digital twins.

The project also demonstrated that dynamic forecasting of WRRF flows and concentrations over an ensuing 24-hour period is sufficiently accurate to enable proactive operational decisions. This capability allows facilities to anticipate and respond to changes more effectively.

At one pilot site, the tool recommended dissolved oxygen concentration setpoints and surplus biomass wasting rates to minimise overall energy usage while maintaining compliance with the treatment goals. At another, the tool provided recommendations to minimise energy and chemical usage while maintaining low effluent total nitrogen levels.

Each pilot facility was able to create high-quality process performance and forecast information, updated every 24 hours, that estimated critical flows and concentrations ahead of laboratory results. Despite some forecast errors due to inaccurate weather predictions, the overall forecast error statistics for all three facilities were within the target 20 per cent error margin.

'Soft’ sensors

WRF Project 5121 demonstrated the potential of live digital twins to revolutionise nutrient management at WRRFs. By combining machine learning with mechanistic modelling, these tools provide actionable insights that can improve efficiency, reduce costs and enhance overall facility performance. The possibility of reducing the level of physical instrumentation needed with digital twins is significant, as it reverses the industry trend of needing additional instrumentation to achieve optimisation goals.

Jacobs is building on this learning with a new Hybrid Optimizer tool – a unique live wastewater process digital twin, integrated with Jacobs’ process expertise, that’s designed to extract the most value out of existing data by providing actionable and time-relevant insights into plant operations. It accomplishes this by integrating ‘soft’ sensors, process intelligence and machine learning-based forecasting on a cloud computing platform and operator-focused user interface. Soft sensors are virtual sensors that can accurately predict values where no physical sensors exist, thus reducing the need for more expensive hardware sensors. They process many measurements simultaneously using algorithms to predict parameters that are difficult or expensive to measure directly.

Once configured and deployed to a treatment plant, the Hybrid Optimizer provides comprehensive insight into the system and can send predictive operational recommendations directly to the smart phones or tablets of field staff. These actionable insights and recommendations help frontline staff optimise performance and realise efficiency improvement opportunities.

Where next?

The implementation of live digital twins and hybrid controls at WRRFs offers numerous benefits for utilities. Operational digital twins provide the best feasible estimates of current process information and future values, enhancing operational insight and decision making. By forecasting future operations, digital twins enable staff to make proactive decisions. Automated integration and analysis of complex, disparate data sources, creating user-centric, actionable insights allows them to focus on critical tasks.

For these tools to be effective, their design must consider how the user wants to receive the information. From simple notifications, such as emails or texts, to integration with work management systems, recommendations can augment operations staff to match their preferences.

The potential for a short return on investment (ROI) period makes digital twins an attractive investment for utilities, offering significant benefits without the need for additional physical or human infrastructure.

As the technology evolves, the benefits of digital twins will continue to grow, offering new opportunities for innovation in water resource recovery. A digital twin is not something that is complete once built – it is ‘born’ and can then grow in maturity, scale, connectivity and application. As described by the SWAN Digital Twin Readiness Guide: “A true implementation is scalable and iterative and will likely be phased over many years, depending on the unique needs, budget, and starting point of each utility.”

Author: Bruce Johnson, PE, BCEE, IWA Fellow is a wastewater technology senior fellow with Jacobs, located in Denver. He has been doing wastewater treatment design for over 35 years, the last 29 of which has been with Jacobs, where he has held the roles of wastewater process and simulation global technology leader.