ACQEOS

The AI agent for your heating plant.
Forecast-based. Market-optimised. Explainable.

energicos operates complex plant configurations with a forecast-based control instance and an AI agent acting as a digital plant engineer. The system orchestrates all generators, storage units and consumers at 15-minute resolution — driven by AI forecasts and current market prices. The agent monitors operations, investigates anomalies and explains every optimisation transparently: cause, action, impact in euros.

Request a pilot plant Download product data sheet

Peak-load management

Forward-looking storage management cuts power peaks

A buffer tank alone does not prevent a load peak. A rigid charging schedule can merely shift peaks — or even create new ones. ACQEOS recognises recurring patterns from space heating, reheating and domestic hot water and combines them with weather forecasts, occupancy times and the current storage state.

Ahead of an expected peak, only the energy actually needed is stored. During the peak, the storage supplies the building alongside the generators, keeping the required generator capacity limited. At the same time, return-temperature-led operation optimises spread and flow rate — within defined comfort, supply and plant limits.

Max. generator capacity
without optimisation

250 kW
full load peak · reference

Max. generator capacity
with optimisation

174 kW
−31% saving

Reduced peak load
vs. no storage

76 kW
static: 20 kW

Cost savings
per year

6,879 €
per kW of provided capacity

Storage energy
provided per day

300 kWh
discharged into the peaks

Controls

250 kW
300 kWh
00:00

Power draw over the day — static vs. dynamic

Storage
static
0%
Storage
dynamic
12%
Heat demandPower draw staticPower draw dynamicConnected load dynamic
Same storage, same building: statically the connected load only drops to 230 kW (−8 %), dynamically to 174 kW (−31 %). The intelligent control extracts 3.8× the peak shaving from the same hardware.
Illustrative example simulation — the result depends on the plant, load profile, storage capacity and user structure. Peak shaving lowers the peak load but does not automatically reduce energy consumption.

Illustrative simulation on a daily basis (24 h). Static operation: fixed night charging & constant daytime discharge. Dynamic operation: load-dependent capping of the peak (peak shaving) down to the capacity-limited threshold. Capacity-cost saving = reduced peak load × capacity price (cost per kW of provided capacity — installed capacity for gas/heat pumps, demand charge for district heating). Provided storage energy = energy discharged into the load peaks per day. Without validated field data, all values are pure model values, not a guarantee.

How it works

Forecast. Optimise. Control. Monitor.

The Asset Manager works in a continuous four-step cycle. Depending on the maturity level, it delivers recommendations, transmits approved changes — or controls automatically within defined limits.

1

Forecast

Neural networks reliably forecast heat and electricity demand based on weather data, historical patterns and occupancy data.

2

Optimise

Deterministic algorithms compute the cost-optimal schedule for all generators — factoring in power exchange prices, grid fees and storage capacities. Auditable and versioned.

3

Control

The schedule is transmitted automatically to the plants — direct plant control or SCADA integration. No manual intervention required.

4

Monitor

Real-time monitoring in the customer portal with KPIs, operating data and schedules. The AI agent investigates deviations, raises alerts automatically and explains what is behind them.

The AI agent

Intelligence with guardrails.

The AI agent explains results, investigates anomalies and plans optimisation experiments — like a plant engineer who looks at every plant every day. The mathematical optimisation and all limit values remain deterministic, auditable and versioned. No setpoint reaches your plant without the safety and approval layer.

System architecture — from measurement to setpoint

Building & heating plant
Weather, tariffs, forecasts
Measurement & control data
Digital plant model
Optimisation algorithm deterministic
Safety & approval layer limits are defined
Recommendation
→ to plant operations
Automatic setpoint dispatch
→ back to the plant

AI agent outside the control loop

The AI agent reads data, findings and model results. It diagnoses causes, explains every action in plain language and quantifies it in euros — but it never writes setpoints itself. That is done exclusively by the verified control path.

Safety takes precedence

Comfort limits and drinking-water hygiene are enforced as hard limit values. On poor data quality, communication errors or comfort violations, the system automatically falls back to the existing control.

Typical optimisation goals

Reduce fuel and electricity consumption
Optimise flow and return temperatures
Adjust heating curve and night setback
Avoid unnecessary pump runtimes
Dispatch boiler, heat pump and CHP economically
Reduce load peaks and cycling
Lower return temperatures for condensing operation and heat networks
Reliably maintain comfort limits and drinking-water hygiene
Detect faults: sensor deviations, stuck valves, poor hydraulic distribution

Step by step to autonomy

Five maturity levels. No blind flight.

Autonomy is not promised — it is earned: every plant passes through five maturity levels, from a pure diagnostic report to limited autonomous operation. The transition to the next level only happens once the previous one has demonstrably proven reliable.

1

Data & diagnostics agent

No intervention

Reads historical measurements, detects data errors and produces a structured plant report. No intervention in the control yet — just an honest picture of the current state.

2

Recommendation agent

Human implements

Proposes concrete measures — such as a flatter heating curve or different operating hours — and calculates the expected savings, comfort impact and risk for each recommendation.

3

Shadow mode

Computes along, never intervenes

The system continuously computes its own setpoints but does not send them to the plant. Its proposals are compared against actual operation — the proof that it would control better, before it is allowed to control.

4

Approved optimisation

With human confirmation

Individual changes are transmitted to the plant after human confirmation — via MQTT, BACnet, Modbus or a controller API. Every change is documented and reversible.

5

Limited autonomous operation

Within hard limits

Automatic adjustments happen only within narrow, technically defined limits. On poor data quality, communication errors or comfort violations, the system automatically falls back to the existing control.

Predictive maintenance

Maintain before something breaks.

Every component that leaves a data trail reveals its degradation long before it fails. The agent continuously tracks the condition of pumps, burners, valves and sensors — and reports when a part is heading into wear. So you replace on schedule in summer instead of via emergency call-out on 2 January. Maintenance is not apportionable — every avoided failure is your money.

Available from day one — read-only, no intervention in the control (maturity level 1)

What we monitor

Circulation pump

high

Signal: Power draw, cycling and runtime patterns

Detects: Bearing wear, short cycling, over-speed

Burner / generator

high

Signal: Start frequency, runtime per start, modulation

Detects: Increasing cycling, ignition failures, efficiency decline

Valves / mixers

high

Signal: Temperature response vs. actuation signal

Detects: Stuck or leaking valves

Heat exchanger

trend

Signal: Return temperature trend

Detects: Scaling, fouling of the condensing surface

Buffer tank

trend

Signal: Stratification, loss rate

Detects: Degrading insulation, stratification faults

The sensors themselves

high

Signal: Plausibility, drift analysis

Detects: Drifting or failed sensors

Honestly scoped: Sealed pressure and corrosion parts such as expansion vessels, anodes or gaskets leave no data trail without an additional sensor. We tell you which sensor is needed — instead of claiming a monitoring capability that does not exist.

Four levers — straight into your pocket

01

No premature replacement

No calendar-based swap of working parts. You only replace once the data shows wear — and save at the top end.

02

No consequential damage

A pump defect caught early costs a fraction of the water or plant damage behind it. The biggest lever.

03

No emergency surcharge

Scheduled replacement in summer instead of an emergency repair with weekend surcharge — and no rent-reduction risk from cold flats.

04

Bundled site visits

Replace several degrading parts in one visit instead of three — fewer call-out fees, less downtime.

Predictive maintenance provides the lead time for a scheduled replacement — not a guarantee of complete failure-free operation. Sudden defects without warning remain possible. The reach of the monitoring depends on the installed measurement and sensor equipment.

Core functions

Everything intelligent plant control needs.

Power exchange connection

Direct connection to the spot market (EPEX) for dynamic pricing. Maximise electricity revenues, use favourable purchasing windows. Up to 25% higher electricity sales revenues.

Storage optimisation

Forecast-based control of heat and electricity storage. Smooth load peaks, use favourable time windows. Automated charging and discharging management.

Generator-mix optimisation

Optimal combination of heat pump, CHP unit, boiler, heat network and hybrid plants. Minimisation of costs per operating hour.

Peak-load control

Reduction of load peaks, better use of grid connection capacity. Multi-charging-point management for e-mobility integrated.

Fault management

Automatic alerts on plant failure. Backup schedules are activated immediately. Automatic notification to electricity trading.

Parameterisable

Minimum runtimes, downtimes, cold-start costs, maintenance windows — all configurable in the customer portal. Full transparency.

Vendor-independent

Any plant configuration. Any hardware.

The Asset Manager is vendor- and model-independent. It orchestrates any combination of generators, storage units and consumers — from the simple heat pump to the complex multi-generator system.

Heat pump
CHP unit
PV system
Storage
Boiler
Heat network

Technical integration

Hardware

  • Own hardware box or existing SCADA
  • Measurement acquisition and setpoint transmission

Protocols

  • MQTT, OPC UA
  • Modbus, BACnet

Data

  • Cloud-based, end-to-end encrypted
  • REST API for system integration

Customer portal

Full transparency. At any time.

The customer portal shows you operating data, KPIs, schedules and forecasts in real time — so you always know what your plant is doing and why.

  • Real-time dashboard with all operating data
  • Schedules and forecasts visualised
  • KPIs: efficiency, costs, CO₂, utilisation
  • Parameterisation: minimum runtimes, maintenance windows, priorities
  • Alerts and fault notifications with escalation chains
  • Export of operating data for your own analyses

Cost-effectiveness

25%

more revenue

Up to 25% higher electricity sales revenue through market-optimised operation — depending on plant and load profile

100%

automated

Forecasting, optimisation, schedule reporting and control run fully automatically

15 min

resolution

Quarter-hour-precise schedule optimisation for maximum market utilisation

BAFA-fundableISO 50001Allocable (BetrKV §2)Redispatch 2.0

System integration

Part of the ACQEOS ecosystem.

The Asset Manager is not a standalone solution — it is fully integrated into the ACQEOS ecosystem and works seamlessly with EMS, EQiTherm and E-Switch.

EMS supplies data

The EnergyCockpit supplies the data basis: consumption histories, generation patterns, cost structures. The Asset Manager uses these for its forecasts.

Learn more

EQiTherm optimises hydraulics

EQiTherm ensures the lowest possible return temperature — the prerequisite for the Asset Manager to operate the heat pump efficiently.

Learn more

E-Switch supplies the plant

The heat pump from the E-Switch programme is controlled by the Asset Manager — forecast-based, market-optimised, fully automated.

Learn more

Ready for intelligent plant control?

In a live demo we show you how the AI agent analyses your heating plant, controls it forecast-based and explains every action — with concrete savings and revenue figures for your plant.

Request live demoCalculate ROI

Get in touch

Ready to activate your portfolio?

In 30 minutes we'll show you what economic potential lies within your portfolio.

No sharing with third parties. No marketing without your consent.