The ECS-Chip series: From development to test

That is a milestone worth marking on its own. The team set out to build a predictive corrosion sensor and the digital infrastructure to support it, and that work has reached the point where real hardware can go into real environments with real users. What comes next is about evaluating the current system in representative operating environments and using the results to further develop, adapt and validate the solution.

Five industrial test partners

The industrial partners joining the test phase: Siemens, RISE, Volvo, Danfoss and Delta Electronics, are not a new addition. They were part of the project from the start and even supported the original funding application to EUDP with letters of recommendation, confirming that a system like ECS-Chip would bring real value to their operations.

Onboarding runs from September through November, with the goal of having every test site up and running by then. From there, the plan is four to six months of continuous testing, feedback and further development based on findings from all five field sites, complemented by ongoing in-house testing at DTU and PAJ.

What testing actually means here

Extensive laboratory testing has been carried out by DTU to evaluate the sensor principles, measurements and models and prepare the system for field testing. The project is now extending this work into representative operating environments, where the full system can be evaluated with the people who will actually use it day to day, and where further refinements can be identified.

Some partners have also asked for a known reference method alongside the sensor. Corrosion is traditionally assessed using small metal samples, called coupons, that are placed at a site and later analyzed in a lab to see how much they have degraded. It is a slow, after-the-fact process, and it is exactly what ECS-Chip is designed to remove from the equation, since the sensor delivers corrosion insight in real time instead of after the fact. Even so, running a known method side by side lets partners build trust in the new approach before relying on it fully.

The first thing partners will actually experience

The first real test is not the sensor itself. Installing ECS-Chip at a partner site means the partner also has to log into the system and configure it, so usability is being tested from the very first interaction, not just measured after the data starts flowing.

Once a site is running, the focus shifts to the long term: continuous data collection, corrosion calculations, and dashboards that need to make sense to the people reading them. Only after the system has been running for an extended period will the sensor hardware itself go through a full validation, checking that the packaging holds up and that measurements stay accurate over time. That is something the project has not been able to test for this long before now.

Real environments come with real constraints

Moving from a controlled setup to five independent industrial field sites has already surfaced practical challenges the team is actively solving. These findings are an important part of the development process and will be used to adapt the system configuration, hardware, software and usability during the project. The project's initial design point was an always-connected sensor: a central gateway collects data from multiple sensors on site and streams it into the cloud for continuous monitoring. Several test sites, though, do not have reliable power or connectivity in place. Mobile connectivity solves part of that, but power is the harder constraint. The individual sensors are built to run on battery for long stretches, but the central gateway needs a steady power connection to keep functioning. Where that is not available, sensors can still log data locally, but it does not automatically make its way into the full system, so continuous monitoring is not yet possible at every site.

What Antire is watching for

Antire's role in this phase is centered on the system the user actually interacts with: how data gets collected, processed, and visualized. The sensor hardware comes from PAJ Group and the underlying AI/ML model comes from DTU, so the team's focus is on whether the dashboards give people the right information in a way that is easy to use and trust.

That means testing will lean heavily on direct feedback sessions with the people using the system day to day, not just on whether the numbers come through correctly. A fully working, polished-looking system carries its own risk here: partners may be reluctant to suggest changes to something that already looks finished. Getting honest, specific feedback, rather than a general "it looks good," is considered a bigger success than the technical baseline of getting reliable data through without interruption.

Bridging lab models with a live system

One of the key development challenges so far has been translating DTU's models, developed and evaluated using laboratory data, into something the live system can run and interpret in practice. That meant working out the exact inputs the models needed and, just as importantly, interpreting their output in a way that gives real value to a non-technical user rather than just a research-grade result.

What comes next

With field testing now getting underway, the project is moving into an important development and validation phase in representative operating environments. The coming months will focus on onboarding the test sites, evaluating system performance, gathering user and technical feedback, and using the results to further adapt and mature the solution.

This EUDP funded project brings together a collaborative team: PAJ Group, leading the ongoing hardware and embedded software development, prototype manufacturing, testing and industrialization activities; DTU, focusing on R&D, testing, qualification, calibration, and developing the AI/ML model to interpret sensor data; and Antire, ensuring the stable and secure dataflow between the sensor and the AI/ML model, and the end users.

Catch up on the series so far: part one introduces the full project, and part two covers the move into development.