Our two-year research project with the University of Exeter has optimized our stacked tray design, and opens up new performance upgrade possibilities for our wastewater and stormwater hydrodynamic separation technologies.
Our Product Development team and the University of Exeter have completed a ground-breaking two-year research project to optimize the performance of one of our core wastewater grit removal technologies.
Combining optimization methods and computational fluid dynamics (CFD), the project was able to derive new designs for hydrodynamic solids removal components that would have been difficult or impossible to achieve using traditional engineering methods.
The research was part of a Knowledge Transfer Partnership (KTP), supported by the UK government’s innovation body Innovate UK, which began in July 2019 and ran until September 2021. It builds on years of collaboration between our Product Development team and the University of Exeter in the field of CFD.
The core of the project involved the optimization of stacked trays that remove high levels of suspended solids in wastewater, with the primary objectives being to improve performance and reduce maintenance requirements.
This was a complex, multi-objective, high-dimensional problem, so the team needed to adopt an unconventional approach in order to solve it. To tackle the problem the project team coupled Bayesian optimization techniques and CFD modeling, using a Bayesian optimization toolset originally developed by the world-renowned Machine Learning Group at the University of Exeter. Bayesian optimization was selected as it can be an order of magnitude more efficient than alternative approaches, such as genetic algorithms, on such complex design problems. The Bayesian optimization toolset was then connected to CFD simulations, allowing the team to use computing power to automate the design and evaluation process.
The team ran the CFD simulations using supercomputers in Exeter and Bristol, with each optimization run modeling some 300 designs at a time. This would have been impossible to achieve using other optimization techniques. The team was one of the first in the world to use these techniques on new GW4 Isambard supercomputing architectures. The team subsequently corroborated simulation results through physical testing of prototypes in our hydraulics laboratory in Clevedon, UK.
The outcome of the project was a new stacked tray design that improves grit removal performance and enables the production of more compact systems that can handle higher wastewater flows and requires less maintenance—an excellent example of combining academic research with commercial incentives to address real-world challenges.
Although the project was focused on improving technologies that will be applied to future iterations of the HeadCell® system, the findings are fundamental to our core technologies, and will be applicable to other products that rely on the same hydrodynamic principles such as Downstream Defender® and First Defense®.
In addition, not only did the project apply methods for complex design optimization already developed by the University of Exeter, but it also drove new developments in multi-objective optimization and constraint handling, resulting in new advances in expertise and understanding. Following completion of the project, Innovate UK assessed the KTP and awarded it a Grade A, or “Outstanding”.
“We have a long history of investing in new science and new technologies to help our customers, and this project continues that tradition,” said Dan Jarman, Hydro International’s Group Technical Manager. “This is the first time that Bayesian optimization and CFD techniques have been applied in the water sector, which represents a significant step forward in product design excellence. In a sector that has been unfairly criticized for being slow to adopt new technologies and techniques, we’ve shown that there are teams out there willing to push boundaries in order to benefit utilities, consumers and the environment.”
We'd like to give special thanks to Prof Jonathan Fieldsend and Prof Gavin Tabor for their invaluable work on this project.