Descriptive
Shows the current state of the object in one place, from the same data the operators already collect.
Digital twins
A model of the object, kept in sync with the object itself.
A digital twin is a purpose-built digital representation of a real asset — a plant, a data hall, a network, a field — connected to that asset by a live data link. The value is not the 3D picture: it is the ability to test a decision on the copy before it is applied to the original.
Three levels of maturity
The term is used loosely. These three differ by how data moves between the object and the model.
01
Digital model
A static representation. Data is transferred by hand; changes on either side do not propagate.
02
Digital shadow
A one-way link. The object feeds the model automatically, but the model does not act back on the object.
03
Digital twin
A two-way link. The model receives live telemetry and returns setpoints, schedules and warnings to the operating system.
What a twin is asked to do
Four working types, in the order organisations usually adopt them.
Diagnostic
Explains why an indicator has moved: which unit, which parameter, which interaction.
Predictive
Projects wear, load and failure probability forward in time.
Prescriptive
Proposes a specific action — a setpoint, a maintenance window, a load redistribution — and shows its modelled effect.
What a project consists of
The order of work is fixed; the depth of each step depends on the object.
- 1
Survey
Which data already exists, at what interval, in what systems; where the losses are; whether a model is justified at all.
- 2
Model
Physics, statistics or a combination — chosen by the question the object has to answer, not by fashion.
- 3
Synchronisation
A data link from sensors, metering and control systems, with checks on completeness and drift.
- 4
Operation
Scenario runs, alerts and handover: the customer's team operates the model without the contractor.
Where it applies
Objects where metering already exists — a twin cannot be built without data.
Data centres
Airflow and heat simulation, placement of high-density racks, failure scenarios for cooling and power, capacity that is trapped by airflow rather than missing.
Thermal modelCapacityFailure scenarios
Power generation and grids
Equipment condition, load modes, maintenance planning by actual wear instead of a fixed calendar.
ConditionLoad modesMaintenance
Industrial plant
Pumping stations, compressors and process lines: failure prediction and selection of an operating mode without risk to production.
Failure predictionModes
Urban networks
Water, heat and lighting: locating losses and setting the order of replacement.
LossesReplacement order
Case study
Digital twin of the KT Cloud Lab data centre
The data-centre building in the Park of Innovative Technologies, Alatau. Geometry follows the working drawings; inside are 84 racks in seven rows, in-row cooling, power from two 10 kV feeders with a standby generator, and telemetry from the operator's Zabbix.
- 44.85 × 18 m, server hall 221.7 m²
- 84 racks · 14 in-row coolers
- UPS N+1 · 833 kW generator
- PUE, alarms and forecasts live
The standard behind the term
ISO 23247, Digital Twin Framework for Manufacturing, published in 2021, defines the reference architecture, the digital representation of physical elements and the rules of information exchange. Part 6, on composing several twins into one system, was published in 2026. The framework gives a common language for describing what exactly is being built — and what is not.
- Reference architecture
- Digital representation of elements
- Information exchange
- Composition of several twins
Start with a survey
The first step is an assessment of the data an object already produces. It ends with a written answer on whether a twin is justified here.
Contact the team