CO2CoControl – A Human-AI Co-Governed Digital Twin for the Operational Reliability and Optimization of Carbon Capture and Wastewater Processes

CO2CoControl – A Human-AI Co-Governed Digital Twin for the Operational Reliability and Optimization of Carbon Capture and Wastewater Processes
Humans and AI working together to anticipate disturbances and improve process reliability


How can an industrial process remain under control when temperature, process load or water quality changes unexpectedly?

The CO2CoControl project is developing a pilot and test environment in which a researcher, artificial intelligence and a digital twin work together to support the management of bio-based, carbon capture and wastewater processes.

The aim is to anticipate disturbances, assess the effects of changing conditions and test suitable responses before they are implemented in a real process.

The development work is carried out in a pilot environment based on a microalgae bioreactor. Continuous measurements, IoT technologies, process data, scientific knowledge and artificial intelligence are combined through a digital twin that helps create an up-to-date picture of the process.

A key principle is that AI does not replace the human decision-maker. The researcher evaluates the recommendations generated with the help of AI and approves changes before they are implemented. This allows the project to explore how human expertise and the ability of AI to process large amounts of information can complement each other.


Why is this being studied?

Climate change can increase variations in temperature, process loads and water quality in bio-based, water and wastewater processes. These changes can reduce process reliability and increase the unnecessary use of energy, water and raw materials.

Another challenge is that measurement data, scientific knowledge, process models and the practical experience of personnel are often kept in separate systems or forms. As a result, disturbances may be detected late and finding the right response may require several trial runs or adjustments.

CO2CoControl aims to bring these sources of information together so that changes can be detected earlier and their potential effects assessed before measures are implemented in the actual process.


How the system works

The approach studied in CO2CoControl is based on a continuous cycle of measurement, assessment and learning.

Sensors and IoT devices connected to the microalgae bioreactor produce data about the state of the process. The measurements are stored in a database and combined with other information, such as scientific publications, patents and open data.

Using this information, the researcher and AI assess the current situation and jointly plan the next experimental setup. The digital twin can be used to examine in advance what might happen if, for example, the temperature rises, the process load changes or water quality deteriorates.

Once the researcher has approved the experimental setup, the agreed settings can be transferred to the pilot equipment. The process is monitored continuously, and the results are returned to the database to support the next assessment and development cycle.


A digital twin allows changes to be tested before they are made in the real process


A digital twin is a digital representation of a real process.

In CO2CoControl, the digital twin represents the microalgae bioreactor and its experimental environment. It can be used to monitor the state of the process, assess transitions and simulate changes before they are implemented in the physical equipment.

For example, the effects of rising temperature or changing process loads can first be examined with the model. This can reduce unnecessary trial runs and provide better information for decision-making under changing conditions.


AI as a partner for the researcher

The project develops a model for human-AI co-governance using an AI-based environment called Raven.

Raven can bring together information from process measurements, recent scientific publications, patents and open data. The researcher can use the AI as a dialogue partner when planning the next experiment, assessing possible risks and considering suitable process parameters.

The aim is not to create an autonomous autopilot. The human remains responsible for the decisions. AI-generated recommendations, their reasoning and the researcher's approval are recorded to make the decision-making process traceable and as transparent as possible.


A microalgae bioreactor provides a controlled test environment

Human-AI collaboration is tested in a pilot environment based on a microalgae bioreactor.

A microalgae process provides a useful testing environment because a real biological process can be monitored continuously and different disturbance scenarios can be introduced in a controlled manner and on a limited scale. The environment also makes it possible to study phenomena related to carbon capture, nutrient use and water treatment.

In CO2CoControl, the microalgae bioreactor is primarily a test platform. The aim is to develop an approach that can later be applied to other bio-based, water, circular economy and industrial processes as well.


Preparing for disturbances before they occur

The project will test at least three disturbance or change scenarios associated with changing operating conditions.

These may include:

• rising temperature
• variations in process load
• changes in water quality.

The aim is to investigate how the digital twin and human-AI collaboration can help anticipate changes, select appropriate responses and support the recovery of the process after a disturbance.


Where are we now?

CO2CoControl started in September 2026 and will continue until the end of August 2029.

According to the project plan, the first stage focuses on building the measurement and data infrastructure for the pilot environment. This will be followed by the development of the digital twin and the co-governance model between the researcher and Raven. Later stages will focus on practical pilots, evaluation and the transferability of the solutions to company processes.


From the pilot environment to company applications

The aim is not to develop a solution that remains limited to a laboratory experiment. The methods and tools created in the project are intended to be transferable to companies and other RDI actors.

The planned project outputs include an operational pilot and test environment, a scenario model for the digital twin, an approval and traceability procedure for AI-generated recommendations, and an implementation and testing package for companies.

The implementation package will describe areas such as the technical architecture, data requirements, acceptance testing and the steps needed to progress from piloting towards later application.

The project will also establish an active cooperation group of at least eight companies. The group will contribute to defining the goals of the pilot, relevant risk scenarios, evaluation criteria and pathways for applying the results.


Project objectives

The aim of CO2CoControl is to develop and pilot an approach in which humans, AI and a digital twin work together to support the management of dynamic processes.


The project focuses in particular on:

• building a measurement and data infrastructure for the microalgae bioreactor
• developing a digital twin that supports real-time situational awareness
• anticipating changes and disturbances through scenario testing
• developing joint decision-making between the researcher and Raven AI
• making AI recommendations, reasoning and human approval transparent and traceable
• testing climate-related disturbance and change scenarios
• creating an implementation and testing approach that can be transferred to companies.

The project aims to improve the operational reliability of bio-based, water and circular economy processes and to help companies adopt new low-carbon and digital solutions in a more controlled manner.


Would you like to learn more about the project? Get in touch!

Elja Kallberg
Senior Advisor
Institute of New Industry
School of Technology
+358 50 427 3078
firstname.lastname@jamk.fi