Optimizing Energy Performance with Digital Twins in Norwegian Buildings

1) Background and framework

In the Norwegian building sector, the energy requirements in the building regulations are performance-based (requirements for results, not just solutions). At the same time, we see a clear movement towards more digitally readable regulations and automatic rule checks – a stated goal in the Norwegian Building Authority’s (DiBK) work on “building regulations for the future”.

About “TEK26” in this case: As of today, TEK17 is the current regulation (legislated via Lovdata and managed by DiBK). In the industry, “TEK26” is often used as a working title/ambition for the next generation of more digitized and stricter building regulations. This case shows how a digital twin can deliver on the TEK17 requirements while being rigged for the TEK26 direction (digital use, traceable documentation, more precise follow-up in operation).

In addition, the case NS 3701 (passive house/low energy for non-residential buildings) adds a level of ambition with clear criteria for, among other things, heat loss, heating and cooling needs, lighting, leakage figures, test procedures and reporting upon completion.

Digital twin is understood here in the SINTEF tradition: a virtual representation of the building connected to data and models for prediction, optimisation, monitoring and decision support – and in mature form, with a two-way data flow between physical and digital “twins”.

---

2) The project (illustrative, but realistic)

Building: “Fjordsporet office + teaching”, climate/weather exposed Norwegian coastal town
Size: approx. 9,000 m² BRA (non-residential building)
Ambition: NS 3701 passive house level + operation that can withstand high energy price volatility and power challenges Energy
system: district heating/heat pump hybrid, ventilation with high heat recovery, solar cells on roof, battery for load management (optional scenario)
Delivery model: BIM + interaction, with digital twin as the “backbone” from design to operation

---

3) Problem statement

The project team experienced a classic challenge:
Paper-fulfilled projected performance is not the same as verified performance in operation.

The goal was therefore twofold:

1. TEK/NS compliance with traceability
   • TEK17 energy measures (e.g., U-values, heat recovery, SFP) could be documented, but the team wanted requirements to also be continuously “monitored” in operation.
2. NS 3701 as an operating contract
   • Not just a “designed passive house” but measured and explained energy use and indoor climate in line with the standard’s requirements for measurement, testing and reporting.

---

4) The solution: Digital twin architecture

The core was to combine three layers:

A. Information model (BIM/IFC) – “what the building is.”

• Rooms, zones, areas, structures, U-values, systems
• Metadata is associated with requirement points (TEK/NS) for later tracking.

B. Data model (IoT + BMS) – “what the building does”

• Temperature, CO₂, relative humidity, pressure/air volumes, energy meters (electrical/heating/cooling), SFP, operating hours
• Data into time series database, quality assured (sensor validation, hole filling, deviation flags)

C. Physics and control models – “why it happens.”

• Calibrated energy model (for heat loss, internal loads, ventilation, solar irradiation)
• Error detection (FDD): “expected vs. actual.”
• Optimization logic: demand control and power limitation (e.g., ventilation/temperature within comfort limits)

SINTEF points out that digital twins become particularly valuable when numerical models are intricately linked to practice and operations.

---

5) TEK26/TEK17 and NS 3701 coupling (requirements → measurement point → action)

The key move was to “translate” requirements into operational KPIs:

TEK (Energy/Technical Performance)

• Heat recovery and SFP: operational monitoring against the projected level (TEK17 guidance shows specific minimum levels for e.g., heat recovery and SFP).
• Envelope/heat loss (indirect): the twin used energy signature and temperature data to reveal whether real heat loss deviated from the projected (sign of leaks, faults in insulation, thermal bridge problems)

NS 3701 (passive houses/low-energy non-residential buildings)

• Requirement areas such as heat loss, heating/cooling requirements and lighting energy were linked to both calculation and verification, and leakage figures were entered as verified key parameters at handover.

TEK26 direction: digital regulation and automation

• DiBK’s goal of more digital use and automatic rule checking was mirrored by the project building a “requirements matrix” that was machine-readable: each requirement point was given an ID, data source, calculation method and acceptance criterion.

---

6) Implementation in three phases

Phase 1: Design (twin “before the building exists”)

• Early simulation of energy demand and power peaks
• Zoning perfected for demand-controlled ventilation.
• “Design for operability”: sensor plan and measurement hierarchy (so that NS 3701/TEK requirements can be verified)

Phase 2: Construction and commissioning (twin “meet reality”)

• Systematic functional testing: ventilation unit, heat recovery unit, VAV dampers, temperature sensors
• Deviations found early: incorrectly installed sensors, incorrect operating times, air flow imbalance.
• The result: fewer “mysterious” comfort complaints after moving in, and less energy operation in “emergency mode.”

Phase 3: Drift (twin “like autopilot with human in the loop”)

• Weekly report: energy use, comfort, deviations, recommended actions
• Automatic fault detection: filter clogging, heat recovery power drop, simultaneous heating/cooling
• Predictive control (weather forecast + thermal inertia): reduced power output during morning/evening peaks

---

7) Results (typical effects – reported as interval)

After 12 months of operation (normalized for weather and usage time), the twin showed the following, compared to “conventional operation” (same building without twin-based follow-up):

• -10-20% energy delivered (mostly from better uptimes, fewer faults, and less simultaneous heating/cooling)
• -10-15% reduction in peak power during peak hours (better load management and preheating/pre-cooling)
• Better thermal comfort (fewer hours outside of operationally defined comfort band)
• Shorter troubleshooting time (FDD pointed to root cause, not just symptom)

The point is not one magic number, but the mechanism: the twin makes energy performance a continuously controlled product, not a one-time calculation.

---

8) What made this TEK26-ready?

Three practices point to a more digital regulatory everyday life:

1. Machine-readable requirements matrix (requirements → data point → acceptance criterion → log)
2. Traceable documentation from projected to measured performance (audit trail)
3. Automated compliance monitoring: the twin alerts when operations slip away from intent (and documents actions)

This harmonizes with DiBK’s stated direction of simplification, digital facilitation and automatic rule checks.

---

9) Lessons learned and recommendations

1) Start with “what do we need to be able to prove?”
NS 3701 and TEK requirements only become real when the measurement plan makes them verifiable.

2) Calibration is everything digital twin without a calibrated model often becomes just a pretty dashboard. The Gemini must be able to explain deviations (causal model), not just show them.

3) Operating personnel are co-developers The best twins are built together with those who will run the building – otherwise you end up with alarms no one trusts.

4) Add the “TEK26 logic” nowEven though the regulations are formally called TEK17 today, it pays to structure requirements and documentation digitally – because it reduces friction in handover and makes follow-up cheaper.

---

10) Short summary

This case shows how a digital twin can work as:

• Documentation engine for TEK compliance and NS 3701 ambition
• Operating motor that keeps passive house performance stable over time
• Bridge to the TEK26 direction, where rules, data and building logic become more machine-readable and traceable.

---