Industrial AI – How machine learning models bridge data gaps to ensure compliance

Increasingly stringent laws and regulations on emissions monitoring pose a challenge for shipyards, ship operators and shipowners.  For example, stringent environmental rules mandate continuous monitoring and documentation of vessel emissions:

Regulations governing the shipping industry have already been tightened up considerably. Sulfur caps on marine fuels were introduced worldwide in 2020. And the EU has had rules in place since 2024 that require ships to pay for their CO₂ emissions. Work is also underway on global regulations, with the International Maritime Organization mandating net-zero greenhouse gas emissions across the shipping industry worldwide by 2050. Customers and investors have long been pressing for transparency around the industry’s environmental footprint, so it’s become increasingly important to measure emissions accurately and reduce them rapidly.

In this context, ships running on heavy fuel oil may be fitted with scrubbers to reduce the sulfur content of their exhaust gases. The efficiency of the scrubbers is then analyzed using continuous emissions monitoring systems (CEMS). These systems – such as the MARSIC300 analyzers – work by measuring concentrations of the relevant pollutants in the engine exhaust gases before and after scrubbing. In short, the measurement values not only document the emissions released into the environment but also monitor the process efficiency of the scrubbers and catalytic converters themselves.

Ship operators must continuously monitor and record these emission values and report them to the authorities as required. If a CEMS fails, certain exhaust gas values will be missing, and the ship operator will no longer be able to demonstrate that the ship complies with the limits. They risk non-compliance and may face sanctions because they can no longer prove adherence to the mandated limits.

Normally, only trained specialists can repair a malfunctioning sensor. This exposes the ship operator to unforeseeable risks, ranging from additional manual labor and higher costs due to the use of higher-grade fuel to penalties for failing to comply with regulatory limits.

Maritime Suite MARpems (for “maritime predictive emissions monitoring system”) offers a solution here – it is essentially the digital twin of a CEMS: If a CEMS (independent from the CEMS supplier) malfunctions or fails, the digital solution takes over: Machine-learning models calculate the current emission values based on the available operational data from the ship and the scrubber.

In this way, Maritime Suite MARpems ensures temporary emissions compliance and thus fulfillment of IMO requirements. These machine learning models are the same for every ship, but they are trained individually for each ship using system-specific data from several past months.

Maritime Suite MARpems is on-premise, meaning that its data stays on the shipping company’s servers. It is currently the only solution on the market accepted by leading flag states – such as Liberia, Germany, and Malta – and classification societies such as DNV. Flag states establish the rules governing the supervision of technical and administrative matters on board ships flying their flag. Further approvals from flag states and classification societies will follow shortly. This will gradually advance the solution’s acceptance.

In addition to ensuring compliance while also reducing the time and effort when a CEMS breaks down, Maritime Suite MARpems has another positive effect: the solution increases transparency on the status of the emissions monitoring equipment and scrubbers, which can help to avoid or plan short-term repairs and maintenance. Accordingly, the system boosts process reliability and reduces service calls to a minimum. All in all, MARpems increases operating efficiency, competitiveness and saves money.