Member Spotlight

Devin Doring, Technical Services Supervisor – City of Salem, OR

The City of Salem, Oregon is located in the Pacific Northwest of the US. As a public entity, the City is responsible for drinking water, flooding, stormwater, and wastewater treatment, while embarking on a smart water journey. As one of the newest SWAN utility Members, our SWAN Americas Alliance Intern Nishanth Senthilkumar (Staff Engineer – Water Operations at HR Green, Inc.) interviewed Devin about the digitisation of the water sector.

The City of Salem, Oregon has notably leveraged machine learning and artificial intelligence to successfully predict algae blooms. How have operators responded to/appreciated such predictive analytics tools?

The operators collectively felt that the aspect of AI/ML was a neat science experiment, but they were not sure how it applies to them. Humans in general don’t like to put their trust into something that they don’t understand. Since AI/ML are indeed ‘Black Box’ approaches to problem solving, there is a huge emphasis to make the algorithms more transparent for operators to comprehend. This is almost more challenging than the aspect of implementing smart solutions itself.

Further, as an organisation, the City of Salem, Oregon is evolving and learning to incorporate predictive analytical tools in our operations. As such, we are a typical organisation on the operational, managerial, and philosophical front. We are in the process of deploying smart water technologies to improve our operations, and rolling them up into a decision framework is a long road ahead. Setting up the system and making the algorithm work is really the first step. It will probably be a theme of our careers – how we implement intense data analytics into operations, analogous to how it took decades to incorporate internet into organisations.

What are the challenges that you faced with the quality of data or suitability of technology for adopting smart water systems?

The challenges we faced during implementation of smart water systems were really practical. It is often issues such as data quality and data communication that challenge us. Algorithms and the technology are secondary compared to the practicality of getting the data back from the field. For instance, with remote monitoring you’re pretty much left to depend on satellite networks or you might be in regions where you depend on batteries. There is only so much infrastructure, and getting the data back from the field to implement analytics is the harder part. These are challenges we are up for, and we will keep working on it. The technologies will improve with time.

How crucial is the capability/upgrade of SCADA systems to handle the level of integration and interoperability associated with smart water technology?

I’d say SCADA systems are a very critical part of smart water systems. Folks that operate SCADA systems are concerned about the security, and no one wants to run science experiments on things that control the infrastructure. We implemented the PI system, a product of OSIsoft (now AVEVA), a long-time SWAN Member, and it really empowered us to pull data from multiple systems and put them into a single cohesive database. You’re going to struggle if you don’t have a single source of truth, or a single database. Trying to build analytics on top of three or four different systems would require cumbersome and siloed data transfers.

Can you discuss a few recent internal efforts and lessons learned on predicting cyanotoxins in water which is a work in progress unlike algal bloom prediction?

Even though we have been able to predict algal blooms to a good extent, we haven’t been able to crack really well the cyanotoxin aspect of it yet. We’re putting in detectors, and it is a journey trying to understand the complex science of cyanotoxin release. Further, in talking to researchers and mathematical modelers, we realised that there isn’t going to be one model that could address this completely. When you are trying to predict something as complex as nature, it is probably unlikely that you are going to find any model that fits perfectly.

Something new we applied is that we currently have thousands of different models and they are all essentially grouped together by means of a Bayesian model averaging framework. From very simple linear regressions to complex neural networks, the Bayesian average model looks at the cohesive result of all of them and selects the best predictions. Until we make some giant leaps in quantum computing, there is going to be a limit to how well AI/ML effectively handles highly complicated systems. I hope to see it within our times.

There are new ways of approaching problem solving in the water sector, including the emergence of new management and procurement models. Can you touch on how much of your smart water projects rely on external consultants/contractors versus in-house capacity?

We believe in finding an optimal balance between projects running on specific timeframes and more long-term oriented programs. As a public entity, we do need to be careful and ensure that funds are being utilised efficiently regardless of any proposed smart water project. We value our hard working staff and constantly explore ways to recognise those successful in project delivery and implementation, including offering performance-based mobility and training/upskilling opportunities. We do recognise that certain projects do require outside help, especially when the private sector has amassed unique expertise in a specific area relevant to our work.

Lastly, with the increasing digitisation of the water sector, what do you see is the role of young professionals in this shift? What advice do you have for rising water engineers such as myself to supplement our engineering skill sets to be best prepared to accelerate in the smart water sector?

Obviously, there is huge role for the younger generation in carrying forward this transition to digital water systems. The aspect of data analytics isn’t a buzzword anymore – it is a natural progression of where technology is going, and we are going to need people in the workforce that understand the technology and can apply it.

The hardest aspect for young professionals is making the leap from school into the workforce, where things get very practical. They find themselves often dealing with lots of irrelevant data, budgetary restrictions, and most challenging, people who are doubtful or skeptical of technologies and innovative approaches. While having strong technical knowledge is a solid foundation, interpersonal and communication skills are going to ultimately lead you to success and continued professional growth. Further, a healthy dose of curiosity and willingness to experiment (involves making mistakes in the process) is definitely helpful and a way to challenge yourself.