Adaptive Digital Twin
MODICA ADT
A live model of your bioprocess that runs next to your bioreactors — from any vendor that offers OPC UA or MQTT: hybrid process models, simulation and replay, and one view of every run. On top, our EU co-financed R&D project adds adaptive capabilities: the twin knows its hardware, adapts when equipment fails and diagnoses problems.
Works with the bioreactors you already have
Any vendor that offers OPC UA or MQTT: the twin connects to your existing equipment, so you don’t replace your bioreactors to use it.
Runs natively on QB Systems modular bioreactors: qb.systemsWhat you get
The value of a digital twin next to your bioreactors
Where the twin pays for itself: fewer lost batches, faster development, steadier quality and quicker answers for QA.
Fewer lost batches
Deviations and equipment faults are caught while the batch can still be steered, and a failed sensor doesn’t have to end the run.
58 weeks
average time between batch failures at a bioprocessing facility: the problem the twin targets
Industry survey: BioPlan Associates, 2022- Real-time monitoring
- Anomaly detection
- Alternative configuration
Faster process development
Parameter variants are tested on the model first, so fewer costly wet-lab runs go to options that were never going to work.
~40%
fewer experiments with model-based DoE
Literature: von Stosch & Willis, Eng. Life Sci. 2016 (simulation)- Simulation and in-silico DoE
Steadier quality
Every batch is compared with a reference trajectory and key values are estimated live, so quality is steered during the run, not found at release.
+22%
prediction accuracy across 18 CQAs, with 3× fewer runs
Vendor case study: DataHow × Bristol Myers Squibb- Reference trajectory
- Soft sensors
Quicker investigations and audits
The deviating variable is named, late lab results are folded in, and every decision is logged for QA.
Annex 11 · Part 11 · ALCOA+
An audit trail of every algorithmic decision and operator action that cannot be edited, mapped to EU GMP Annex 11 and 21 CFR Part 11
- Root-cause hints
- Audit trail
More from the equipment you have
The twin works with any bioreactor that offers OPC UA or MQTT, so the value comes without replacing hardware.
Examples of bioreactor controllers with OPC UA
- Sartorius · BIOSTAT STR, Ambr
- Eppendorf · DASGIP, DASbox
- Getinge Applikon · Livit Flex
- Infors HT · Minifors
- Thermo Fisher · HyPerforma SUB
- Vendor-agnostic connectivity
Knowledge that stays
What one run, scale or operator learned stays in the model and is reused, instead of leaving with people.
1:50
model carried from a 300 mL shaker to a 15 L bioreactor, NRMSE ~11% for viable cells
Literature: Bayer et al., Front. Bioeng. Biotechnol. 2021- Hybrid models
- Transfer across scale
Figures come from literature and industry sources, not from MODICA projects. The effect on your process is measured in a pilot.
In daily work
How your team works with the twin
Five moments in the life of a batch, from connecting a bioreactor to reviewing the finished run. The operator stays in charge; the twin prepares, watches and explains.
1. Day one · Connect a bioreactor
You
Connect a bioreactor from any vendor that offers OPC UA or MQTT.
The twin
Recognises the hardware — sensors, pumps, actuators — and configures its model of the vessel, without a manual mapping project.
Value for you
A new vessel joins the twin without a manual mapping project, and you keep the equipment you already own.
2. Before the batch · Plan the batch in silico
You
Enter the parameter variants you are considering: temperature, pH, feeding, induction time.
The twin
Simulates each variant and shows the predicted trajectory and outcome, so only the most promising ones go to the wet lab.
Value for you
Fewer wet-lab runs spent on variants that were never going to work, and a documented reason for each choice.
3. During the batch · The twin watches and assists
You
Run the batch as usual.
The twin
Compares it live with the reference trajectory, estimates biomass with a soft sensor, warns of a deviation and names the variable. It recommends set-points; a physics check blocks unsafe ones, and the decision stays with you.
Value for you
Deviations caught while the batch can still be steered, not after release testing.
4. When something fails · Equipment fault mid-batch
You
Get a diagnosis instead of a bare alarm.
The twin
Detects a failing sensor or pump, explains the likely cause and proposes an alternative configuration so the batch can continue.
Value for you
Fewer lost batches, and operators who know what to do instead of guessing.
5. After the batch · Replay the batch and find the cause
You
Replay the run and compare it with earlier batches.
The twin
Folds in late lab results by correcting earlier estimates, shows where the run left the reference, and keeps an audit trail for QA.
Value for you
Faster investigations that start from a hypothesis, and a record QA can sign off.
How the twin is built
One twin between your bioreactors and your team
Click a layer to see its capabilities, then a capability to see what it does and what you get.
Structure follows the ISO 23247 digital twin reference model
Physical space
Modica digital twin
People and systems
Cross-domain · across all layers
Click a layer or a cross-domain block to see its capabilities.
Test the twin on your own process
An 8–12 week pilot on your process, with success criteria agreed before we see any results and first results after 4 weeks.
Apply for a pilot