Cybersecurity: keeping your automation system safe

Despite the many benefits of introducing process automation solutions at industrial wastewater treatment plants (WWTP), a concern that always comes up during the initiation of any new digital project is around cyber security. Continuing in our series of articles about advanced process control (APC) solutions for industrial WWTP, Niels Tiemessen, product expert and business developer at Royal HaskoningDHV, takes a look at cybersecurity and plant safety with regards to system hacking.

Threats from cybercrime

Any industrial or municipal WWTP, or any other asset that is connected to a network, can be subject to threats from cybercrime, with tremendous potential damages, especially when local control is compromised. The most obvious solution to eliminate any type of remote risk is to have a local control system (SCADA/DCS) that is not reachable through the internet. However, with more businesses enabling remote working, especially since Covid, and also not wanting to have to staff the control room outside of office hours, this is not an option anymore.

You only need to look at the Wikipedia page of Cybersecurity, to see that the list of threats and defenses is rather overwhelming, requires expert knowledge and would be impossible to cover in a short blog. To gain a more detailed insight on our view on this matter, please read our Whitepaper on managing cybersecurity.

Operational safety and cybersecurity are different things

While cybersecurity measures need to ensure that an organisation stays safe from malicious intentions, this doesn’t automatically mean that the software itself is safe and robust to use. Since actual setpoints, and not just insights, are provided to the local control system, it is highly advisable to determine whether the various necessary built-in safety measures are present.

These include a Watchdog that monitors the connection between the APC software and SCADA/PLC. When this connection is lost, process control should automatically revert to traditional, local control. Since all APC solutions depend on online measurements, extra safeguarding for when these measurements are unreliable or missing is crucial as well. Anomaly detection could provide additional insights on when things like sensor drift start to occur and help to improve maintenance and data quality. Also, the fundamental type of machine learning model and algorithms determines the robustness of an APC solution. Despite providing great modelling insight, black box models often need long (re)training times and are influenced by different actions from operators, leading to potentially higher downtime, especially when the line-up of a system changes. A more grey-box type of approach could be more favourable. This is where the fundamental working principles, based on deep domain knowledge, are ingrained in the algorithms themselves, and not learned by the AI.

Find out more about how Twinn Aqua Suite can help improve industrial WWTP performance and help the industry tackle the skills shortage. Or, get in touch with Niels Tiemessen for more information.