The IWA Water Resource Recovery Modelling Seminar, held earlier this year, showcased the latest thinking on modelling and its potential to transform water and resource recovery operations. David Ikumi outlines current directions.
Mathematical models have great potential to contribute towards addressing current and foreseen challenges in water and sanitation through their predictive power. By generating critical data, they enable different scenarios to be considered and inform decision-making with regards to future systems and known or anticipated challenges.
Models are essential tools that will allow us to navigate system intricacies, both for individual unit operations and for integrated, system-wide water and sanitation infrastructure, enabling more holistic solutions to be adopted. Water and resource recovery facility (WRRF) models are continuously being refined with the aim of achieving the most environmentally sustainable solutions, while at the same time minimising the costs associated with climate, sources of waste, population dynamics, and social perceptions.
The University of Cape Town’s (UST’s) Water Research Group hosted the prestigious 8th IWA Water Resource Recovery Modelling Seminar (WRRmod2022+) on 18-22 January 2023. The seminar brought together academics, engineering experts and scientists specialising in modelling wastewater treatment processes. It provided a platform for sharing innovative research, practices and prospects in modelling WRRF, highlighting recent engagement on changing paradigms and globally relevant topics, building on the consensus of the WRRF modelling community. The event also included a special gala dinner held in honour of, and attended by, renowned global wastewater expert George Ekama, who contributed so much to this field, including to IWA; George, who was Professor of Water Quality Engineering at UCT, sadly died on 19 February. This was the first time that this biennial event was hosted in Africa, which was celebrated as a positive step in the quest to make the seminar internationally diverse and representative of expertise across the globe. Strongly supported by the IWA Modelling and Integrated Assessment (MIA) Specialist Group and the Water Institute of Southern Africa (WISA), WRRmod2022+ attracted a sizeable number of high quality abstract submissions and workshop proposals, which resulted in rich discussions of current challenges and opportunities in the modelling of WRRF.
The key themes discussed at the seminar included:
- Integrated, system-wide approaches to modelling WRRFs for improved resource recovery and climate resilience
- The development of models for the adaptation of nature-based solutions to wastewater treatment
- The development of hybrid models and digital twins (DTs) for WRRFs.
Integrated system-wide mathematical models are valuable tools for promoting a circular water economy and addressing water and sanitation challenges in a holistic way. They could be useful, for example, for determining whether decentralised or centralised options are more promising, and for providing guidance on high-level planning to promote resource recovery, resilience to climate change, and efficient use of renewable energy resources.
For this to be achieved it is necessary to understand how the various unit operations that make up a WRRF interact.
This can be a challenging process. Moreover, while mechanistic models for primary and secondary wastewater treatment processes are well established, tertiary treatment technologies and water reuse processes still require a lot of investigation to improve their predictive accuracy. Hence, there are ongoing developments in the improvement of water treatment and reuse models, which has led to research on the tracking of emerging contaminants, including viruses, bacteria, protozoa, and organic and inorganic micropollutants, within integrated system-wide WRRF models.
Nature-based solutions (NbS) are often considered to provide a low technology solution, but they have multiple uses and high process complexity, which could pose a challenge for the development of better modelling tools. Many new models focus on using the opportunities provided by NbS, such as the use of natural systems like wetlands and aquifer recharge for water treatment and storage.
Within the WRRF modelling community, several modelling approaches are being considered. However, from discussions at the WRRmod2022+ seminar, it was agreed that improved models should begin by having the following properties:
- A mechanistic basis for water column/sediment changes
- Powerful insights into material mass balances over the system
- The ability to vary the water column as sediments accumulate and simulate the benefits of desludging
- To be built from common process units
- Provide effective simulation of biological activities and settling processes.
Digital twins (DTs) are creating a wide range of new possibilities for the operation and maintenance of WRRFs, from the effective use of data generated by the systems through to reduced operating costs and better engagement of qualified staff for the improvement of plant operations and maintenance. The use of DTs for WRRFs is relatively new, and there are currently few examples of their practical implementation and limited possibilities for adoption in developing regions, which already have major challenges with minimal maintenance of instrumentation and less potential in terms of human resources.
However, during the WRRmod2022+ seminar, there were some discussions regarding the design aspects required for the effective application of DTs. From these discussions it was noted that there is a need for good sourcing and storage of clean data, to be available at reasonable frequency. Good data structures are required to provide adequate links between different pieces of information. In addition, capacity is required to secure qualified personnel, including good operators for maintenance support of DTs. Process engineers with some background in data science and domain expertise are required, along with technology to support good communication protocols and favourable model implementation.
Artificial intelligence and machine learning
The use of artificial intelligence (AI)-enabled data and machine learning (ML) could be used to develop more accurate and efficient hybrid models (HMs) that integrate data-driven and mechanistic approaches for DTs. This is because they can allow for automation of the modelling process, reducing the time and effort required to develop and validate models, and accurately simulate real-world scenarios and identify potential problems before they occur.
However, the future application of HMs may require some improvements, including:
- Common application programming interfaces
- The development of effective and harmonised scientific terminologies
- The establishment of good modelling practice frameworks for HMs
- Improved methods to deal with uncertainty
- The development of methods to obtain good data quality and meta-data
- Continuous refinement of architecture for ML algorithms.
There is a need to encourage platforms for information sharing, including data from different types of systems and case studies on the application of HMs and DTs.
The work by Serrao et al (2023) is an example of recent studies towards development of smart tools for WRRF digitalisation and was used as the basis of an interesting debate topic for the WRRmod2022+ seminar. This work involved a consolidation of research developments by modelEAU at Laval University towards the effective use of sensors and data-driven tools.
To improve decision-making processes, reduce costs, and improve efficiency in the water and sanitation sector, it is important to continue investing in the development of mathematical models and the necessary tools of support.
The WRRmod2022+ seminar discussed prospects for WRRF modelling, which included several key points, many of which were presented by Young Water Professionals.
First, it was observed that there is a need for more structured ways to implement models as a collaborative process that involves all stakeholders, such as system operators, regulators, academics, process engineers, utility leaders, municipality decision-makers, IT and IT security experts, instrumentation technicians, software developers, DCS/SCADA system manufacturers, and programmers.
The structure of general modelling workflow typically involves data collection, data pre-processing, data storage and access, data mining, modelling (including data-driven, mechanistic and machine learning components), comprehension of model results, and proposed action. For this process to be impactful, the levels of engagement with public and municipal decision-makers needs to be well established, particularly in developing regions. This is to ensure that the public is more knowledgeable about the challenges and solutions regarding water and sanitation systems, empowering them to influence policy and improve the corporate appetite for investment in the most suitable technologies, to address equality in service provision.
Second, information sharing will be the new norm, with increased platforms for capacity building and information sharing encouraging creative problem solving via transdisciplinary collaborations. This will require protocols to be followed to ensure that data is clean, and that cybersecurity is maintained.
Finally, rigorous evaluation processes will be required to ensure that the more complex and fully automated systems that may operate in smart cities of the future provide socio-economic and environmental improvements to help resolve the global water crisis. •
David Ikumi is Associate Professor at the University of Cape Town in South Africa.