Digital twin approach to continuous emissions monitoring

Shipping operators are under increasing pressure to continuously monitor and document vessel emissions to comply with tightening environmental regulations. Continuous emissions monitoring systems (CEMS) are central to this task, measuring pollutant concentrations before and after exhaust gas treatment to verify compliance and assess scrubber performance. However, when these analyser systems fail, operators can be left with critical data gaps. This creates compliance risks, increases administrative workload and may expose operators to penalties. The challenge is amplified at sea, where access to specialist personnel capable of repairing CEMS hardware is often limited.

To address this issue, Endress+Hauser developed a software-based solution that maintains emissions monitoring continuity without relying solely on physical sensors. Its maritime predictive emissions monitoring system (MARpems) applies digital twin technology and machine learning to provide a reliable fallback in the event of analyser downtime, enabling operators to maintain compliance even when instrumentation is unavailable.

The system was developed using data-driven modelling techniques, with machine learning algorithms trained on historical vessel operating data, including engine performance, fuel consumption and scrubber behaviour. These models establish a digital twin of the onboard emissions monitoring system, replicating its functional behaviour under varying operating conditions. By correlating available process variables with expected emissions outputs, the software can estimate pollutant levels and compliance status when direct measurements are not available, reducing reliance on redundant sensing hardware.

Designed as part of Endress+Hauser’s Maritime Suite, MARpems is deployed as an on-premise solution, ensuring that operational and compliance data remains within the ship operator’s IT environment. The software integrates with existing vessel systems and CEMS infrastructure, accessing real-time process data streams. When a sensor or analyser fault is detected, the system automatically transitions to predictive mode, generating validated emissions data based on current operating conditions. The solution has also achieved regulatory acceptance from multiple flag states, including Germany, Malta and Liberia, enabling its use as a compliant alternative during equipment outages.

In operation, MARpems mitigates the immediate risks associated with monitoring system failures while delivering broader operational benefits. Continuous visibility into scrubber and exhaust treatment performance improves transparency and supports more informed decision-making. Predictive insights enable condition-based maintenance strategies, helping to reduce unnecessary servicing and lower overall maintenance costs. The ability to maintain uninterrupted emissions records also simplifies reporting and reduces the administrative burden on crew.

“It’s more than an emergency back-up system — it’s also a strategic tool for greater efficiency and competitiveness in the maritime industry,” said Fabienne Jäckle, team lead for digital product management at Endress+Hauser SICK.

For electronics and instrumentation engineers, the solution reflects a broader shift towards hybrid measurement architectures, where software-defined sensing and data modelling complement traditional hardware. Rather than relying exclusively on physical redundancy, digital twins and machine learning provide an alternative pathway to resilience, particularly in remote environments such as maritime operations.

Image credit: iStock.com/choochart choochaikupt