A digital twin implementation timeline from offline analysis to live operations

A district-heating operator can have a network model while dispatchers still work from separate pressure and temperature displays. The gap is the connection between calculated behaviour and current operating data: without it, simulation cannot inform live monitoring. Implementation therefore starts with the heat demand and network behaviour behind operating decisions, then adds data exchange, monitoring, forecasts and, where appropriate, controller-issued setpoints.

The offline model and the live digital twin

An offline model lets engineers simulate a proposed operating condition before changing the network. It can help assess how a different heat load or supply setting might affect flow, pressure and temperature, but its results describe the conditions entered into the calculation. They do not tell a dispatcher what is happening now.

A district-heating digital twin connects that physics-based simulation to operating data so the calculated state can be compared with current conditions. The distinction matters when specifying the first stage: a useful offline model does not become a live operational asset merely because its results appear on a dashboard. The connection must keep the model’s inputs relevant to the decisions operators need to make.

The network behaviour to model before connection

Start with heat demand: where it occurs, how it changes and which operating conditions the model must cover, including the load allocation, control and management challenges that DISTRHEAT identifies for district heating and cooling, particularly under variable climate conditions. Smart-meter heat-consumption data from different UK dwelling types illustrates one possible source of demand information. Such data informs assumptions; it does not by itself describe heat transport through a district-heating system.

Transport matters because a change at production does not reach every customer at once. In a Grenoble district-heating study, models used for setpoint scheduling accounted for both heat transport and thermal storage in the network. Those effects are relevant when assessing when a supply-temperature change will affect demand, not only whether the temperature is high enough.

Hydraulic behaviour completes the initial picture. Model flow and pressure at the points where a decision may constrain delivery; a district-heating prototype produced simulated values for both. Differential pressure was one of the operating setpoints scheduled in the Grenoble study. A deviation from its expected pattern is useful only if operators can relate it to the components and operating conditions the model describes.

Connecting calculations to operating data

Integration changes the model’s job. Instead of asking engineers to enter a scenario and inspect its output, connected operation brings current measurements into the calculation and returns indications that dispatchers can interpret alongside field values. Map each incoming point to a model location, identify its units and timestamp, and decide how a missing or doubtful measurement will be displayed. Otherwise, a precise-looking result can rest on stale input.

The district-heating prototype returned simulated flow and pressure to supervisory control and data acquisition (SCADA) for visualisation. Its test used a virtual controller, so that connection illustrates data exchange rather than a continuously synchronised live heating system. For an operational rollout, the acceptance test is whether the exchange stays dependable under the conditions dispatchers actually encounter.

What connected monitoring gives dispatchers

Connected monitoring puts measurements and calculated indications in the same operating context. A Warsaw dispatcher dashboard displays current and historical pressure and temperature data. In a Romanian energy-system case, operators view live SCADA temperature, flow and pressure values. These are examples of field data a dispatcher can use to understand the present state and examine what preceded a change.

A twin can add calculated behaviour where a comparison helps diagnose that change. The display must identify a measured pressure as a field value and a calculated pressure as a model indication. That separation lets an operator investigate a mismatch without mistaking a simulation for confirmation at an unmeasured point.

Forecasts and operational setpoint decisions

Monitoring answers what the available measurements show now; forecasting supports a decision about what to do next. In the Grenoble study, a heat-load forecast informed schedules for supply-temperature and differential-pressure setpoints. The models accounted for transport and thermal storage, which matter when the effect of a setpoint change arrives later than the change itself.

The researchers assembled the models into a decision-support tool called PEGASE. They also proposed a model-predictive controller for scheduling setpoints, but the operator was still evaluating that proposal. A forecast schedule presented for dispatcher assessment is already an operational use of the model; it does not mean the model is changing setpoints.

The boundary between decision support and automated control

The boundary is the action taken on the recommendation. If a dispatcher reviews a forecast and chooses a setpoint, the twin supports a human decision. If a controller issues the setpoint update, it has entered automated control. At the Ferrara hospital case, a controller updated operating setpoints, illustrating that further step.

Automation calls for an explicit decision about which setpoints the controller may change, the limits it must observe and how operators regain control when inputs or outputs cannot be trusted. It need not be the destination for every district-heating digital twin. Where monitoring and forecast-supported scheduling resolve the operating question, those stages have a purpose of their own.

Platform requirements depend on the intended stage

A digital twin platform should be assessed against the stage it must support, rather than a product ranking. We would use the following checklist before committing to digital twin technology for operations:

  • Integration: Can it map the required measurements and setpoints to the model and exchange them with existing operational systems? Test missing data and timestamps, not only a successful connection.
  • Deployment: Can the proposed arrangement meet the organisation’s requirements for access and response time, whether calculations run on site, in the cloud or across both?
  • Analytics: Does it support the physics-based calculations needed for the chosen decision? Live monitoring, forecast-supported scheduling and automated control place different demands on update frequency and outputs.
  • Security: Who can view data, alter the model or authorise a setpoint change? Check how those permissions and actions are recorded before connecting control.
  • Capacity: Can it handle the intended network scope and calculation frequency as more assets and operating decisions are added?

The first purchase decision is therefore about an operating use, not a claim to have every possible digital twin capability from the outset.

Each connection adds a distinct operational use

An offline model can test network behaviour; connected data makes its calculations relevant to current conditions. Monitoring then supports dispatchers, forecasts support future setpoint decisions, and automated control begins only when a controller updates those setpoints. That sequence makes it possible to judge each stage by the decision it improves, distinguish simulated indications from field measurements and assess platform integration, deployment, analytics and security against the intended stage.

For a practical assessment of which stage fits your district-heating operation, contact Fluidit.

Related Articles

© Fluidit 2026