Clock 18:00Cooling load 0 MWFlow 0 m³/hDelta-T 0 KPumps 0 kWChiller 0 kWCOP 0—FPS 0

Drag to orbit, wheel or pinch to zoom. Blue pipes carry cold supply water from the plant at the left, orange pipes carry the warmer return water back; arrows move faster when the flow is higher and pipe thickness follows the diameter slider. The pipe runs stand for 0.8, 1.2 and 0.9 km of real network.

District Cooling Network

One chilled-water loop, three buildings, one day. Delta-T sets the flow, friction sets the pump power. Virtual model for learning, not a design tool.

Network

Weather and storage

Time

Live readouts

Cooling load by building (MW) and chiller output

Pump and chiller electric power (kW)

Effective delta-T (K) and COP

Missions

Experiments

Each button runs a whole day for several values (all other sliders as set) and reports daily energy.

Independent numerical checks

Controls

Drag / one finger: orbit; wheel / pinch: zoom; four camera presets.
Sliders recompute the whole day at once; the clock then replays it. Time scale is simulated hours per real second.
The thin vertical line in the charts marks the clock.

The science

Flow from load and delta-T. A building with cooling load Q and water temperature rise ΔT needs the mass flow m = Q / (c_p ΔT) (c_p = 4186 J kg⁻¹ K⁻¹), V̇ = m/ρ. The design ΔT falls at part load and with fouling: ΔT = ΔT_d max(0.35, PLR^0.3)(1 − fouling) (the exponent and floor are model values), the well-known "low delta-T syndrome": the same cooling needs more flow.

Pipe hydraulics. Darcy-Weisbach: Δp = f (L/D) ρ v²/2, v = 4V̇/(πD²), Re = ρvD/μ with μ = 1.4 mPa s (about 7 °C). Laminar f = 64/Re; turbulent Swamee-Jain f = 0.25 / [log₁₀(ε/(3.7D) + 5.74/Re^0.9)]² with roughness ε = 0.045 mm (steel, model value). Three runs carry the sum of the loads downstream (diameters D, 0.85 D, 0.7 D, represented lengths 0.8, 1.2 and 0.9 km); supply and return are equal. Head: H = 2 Σ Δp/(ρ g) + 10.2 m (about 100 kPa at the farthest heat exchanger, model value).

Pump and chiller. P_pump = ρ g V̇ H / η, η = 0.75 (variable-speed, model value). The chiller uses a second-law fraction of the Carnot cooling COP, COP = η_II T_e/(T_c − T_e), T_e = T_supply − 3 K, T_c = T_ambient + 8 K, η_II = 0.55 [перевірити за джерелом]. At constant friction factor and no static head, pump power grows with the cube of flow, which is why a larger ΔT pays off.

Thermal store. Usable energy E = ρ c_p V ΔT for 10 K (1000 m³ ≈ 11.6 MWh). The store charges when the load is below the daily mean and discharges above it, so the chiller runs flat; it shaves the peak but does not save energy by itself (no losses modelled).

Simplifications. One branch line without loops, ideal insulation, constant water properties, quasi-static hydraulics, no valve control, no cooling-tower detail, building ΔT set by a simple rule, daily load shapes are illustrative. This is a learning model, not engineering advice, and not for real network design. Building colours and the dusk scene are illustrative. Assets built with Blender; scene by Three.js.