Djinious
District energyEnergy

Euroméditerranée District Energy Network

A seawater-source district-heating network: two intake pumps and two seawater exchangers feeding three heat pumps, two backup boilers and a stratified thermal store, distributing through two network pumps and a differential-pressure valve to thirty consumer substations. The network is simulated; the city under it is not — every substation sits on its real IGN BD TOPO® building footprint in the Euroméditerranée quarter. It is a demonstration network on open data, not any operator's plant.

DjiniousCC
The DjiniousCC 3D city twin over Marseille: 3,356 extruded IGN BD TOPO building footprints on live IGN orthophoto imagery, with thirty consumer substations picked out in amber and ringed by red alarm halos, the network tree on the left and a live time scrubber below.
site code
MEDIsite code
assets
44assets
live tags
301live tags
authority class
C4authority classoperational action

Site MEDI, Marseille, France. These are not mock-ups: all five sites run side by side in the same DjiniousCC instance — 99 assets and 693 live tags, acquired at 1 Hz, historised, alarmed, and each with a procedure that has been dry-run and proven on an isolated twin.

A cold snap asks for more heat than the network was built to deliver

The thirty substations sum to 1,319,287 W/K of heat-loss coefficient. At −4 °C that demands far more than the plant's installed capacity — three heat pumps derated by seawater temperature, two 1,800 kW boilers and a store that can only discharge at 2,000 kW. The supply header sags 21.2 K below its compensation setpoint and every substation downstream starves. You cannot serve everyone; you have to decide who gets trimmed, and be able to show why.

What the platform models here

Each family has a behavioural model that the live simulator and the proving twin share, so a dry-run behaves the way the plant does.

Seawater intake pumps · Seawater exchangers · Heat pumps · Backup boilers · Stratified thermal store · Network pumps · Differential-pressure valve · Consumer substations

The procedure: District Peak Shaving

Carry the network through a cold-snap capacity shortfall without letting the supply header collapse: bring on backup boilers, discharge the store, raise the compensation curve, then stage valve-limit setbacks across the lowest-criticality substations. Authority class: C4 · operational action.

  1. Step 1

    Start backup boiler B01

  2. Step 2

    Raise supply compensation curve → +6 K

  3. Step 3

    Duty-engineer confirmation

  4. Step 4

    Stage 1 setback — SST16…SST30 valve limit 70 % (fifteen substations, one command each)

  5. Step 5

    Notify control room

The proof: Cold Snap Capacity Shortfall

Ambient falls from +2 °C to −4 °C five minutes into a thirty-minute run, on a seeded clock — the same snap, tick for tick, every time it is run.

Baseline — without the procedure

  • Supply header ≥ 67 °C — measured 61.63 °C
  • Peak capacity shortfall ≤ 23 000 kW — measured 24 145 kW
  • Six largest substations hold ≥ 23 K ΔT — measured 19.93 K

With the procedure — every assertion holds

  • Supply header ≥ 67 °C — measured 72.50 °C
  • Peak capacity shortfall ≤ 23 000 kW — measured 22 419 kW
  • Six largest substations hold ≥ 23 K ΔT — measured 27.66 K

Seventeen commands intercepted on the isolated twin. Note what this does not claim: a 22,419 kW shortfall remains after mitigation. This is peak shaving, not peak solving — the network is genuinely undersized for this snap, and the procedure's job is to hold the header up and triage the load, which the measured numbers show it doing.

What the control room sees