Building Performance Analytics & Digital Twins
Model · the digital twin.
We turn the data a building already produces into decisions it can act on — fault detection, diagnostics, and models that stay useful after the consultant has left.
Most buildings are already instrumented well enough to explain their own faults. The data sits in the BMS historian, unread, because nobody specified what should be watched or what a deviation means.
This is where our research work meets practice. We build the analytics layer that reads that data continuously, and — where it earns its cost — the model that lets you test a control change before deploying it to an occupied building.
What we do
Data analytics and fault detection
Automated fault detection and diagnostics on existing BMS data — simultaneous heating and cooling, stuck valves, schedule overrides, sensor drift.
Digital twins and calibrated models
Models calibrated against measured data, used to test control changes before they reach the building.
AI and data-driven methods
Applied machine learning where it is genuinely better than a rule — occupancy inference, preference modelling, load prediction.
Occupant-centred control
Control strategies informed by measured individual preference rather than a single assumed comfort band.
Energy performance analysis
Where the energy actually goes, quantified against the control functions that drive it.
Applied research and European projects
Research partnership and technical work packages for funded programmes.
How we work
- 01
Data audit
What is already logged, at what interval, and whether it can be trusted.
- 02
Fault library
The specific failure modes worth detecting in this building, and what each one costs.
- 03
Analytics build
Rules and models implemented against live data, tuned to suppress false positives.
- 04
Model and test
Where justified, a calibrated model to trial control changes off-line first.
- 05
Operate and review
Findings reviewed with the operating team on a set cadence, so the analytics stay believed.
- Data quality and instrumentation audit
- Prioritised fault library with cost impact
- Fault detection and diagnostics rule set
- Calibrated performance model where justified
- Control change recommendations, tested first
- Periodic performance review reports
Do we need new sensors before this is worth doing?
Usually not. Most buildings we look at are already logging enough to find their most expensive faults. We audit what exists first, and only recommend instrumentation where a specific, costed question cannot be answered without it.
Is a digital twin worth the cost on a normal commercial building?
Often not, and we will say so. A calibrated model earns its cost when control changes are frequent, risky, or hard to trial in an occupied building. Otherwise good analytics on existing data gives most of the benefit for a fraction of the effort.
What does “occupant-centred control” mean in practice?
Controlling to measured preference distributions rather than to a single assumed setpoint band. It is the subject of our peer-reviewed research, and the basis of the Energy-Wise Archetypes project.
Let’s discuss your project.
Tell us about the building and the systems involved. We reply to every serious enquiry.