DEV environment · test payments only
ChemEngWorks

Calculators › GridWorks

NETWORK DESIGN

GridWorks

Build incompressible and compressible networks with connected equipment.

Solves steady, single-phase pipe networks of liquid or gas: the pressure at every free node and the flow in every connection, with pumps, fans, compressors, valves, regulators, orifices, heat exchangers, choking, junction losses, heat transfer and mixing.

Prices

RunsPrice
Networks of up to 30 nodes and connections1 credit
Networks of more than 30 nodes and connections3 credits
  • A run that fails because of its inputs, doesn't converge or runs out of time is free.
  • Cancelling a run is free, before or after it starts.
  • A fault on our side is never charged: the run is tried again automatically.

A credit costs 8p to 10p, depending on the pack. Credits don't expire. Packs and prices

Inputs and results

You enter

  • Nodes: pressure and flow boundaries, liquid reservoirs, junctions, bends and tees
  • Connections: pipes and equipment, with curves, set points and fittings
  • The fluid: components from ThermoWorks' database, or a liquid's measured density and viscosity

You get

  • Each node's pressure, temperature and composition
  • Each connection's flow, pressure drop and velocity, and any choking
  • Profiles along each connection and the equipment's operating points
  • A drawing of the solved network, and the solver's convergence

Example: Network solution

CONVERGED

Water at 20 °C between two reservoirs at atmospheric pressure, through two 20 m pipes of 52.5 mm bore and a pump between two junctions.

The starting case every new GridWorks case opens on, run by the engine on our servers and stored (engine 1.1.0 · data 1.0.0+components-1.0.0). Nothing is calculated in your browser, here or in the calculator. Exports and the other tools of a run need an account.

MAX MASS RESIDUAL0kg/s
MAX THERMAL UPDATE0.00163K
SHAFT POWER0.0037 kWSum of rotating equipment
MODEL FLAGS2Inspect before design use

Numerical solution converged. Inspect property and equipment flags below.

Warnings and model qualifications (2)
  • P1: Operating point outside supplied curve — no validated extrapolation
  • Liquid energy model uses the user-specified common Cp for all liquid mixtures.

Results

Export results CSV

PipeFlow statusCritical mass flowCritical pressureMass flow A → BActual inlet flowStandard flow (gas)Static Pg at AStatic Pg at BFriction ΔPMachine riseT outletMax velocityMach maxReDarcy fHeatShaft power
P1Subcritical——27,000 kg/h26 m³/h—2.4 barg2.4 barg0.0 bar0.0038 bar20 °C3.4 m/s0.0023171,46300.0 kW0.0037 kW
P3Subcritical——27,000 kg/h26 m³/h—0.013 barg2.4 barg0.45 bar—20 °C3.4 m/s0.0023171,4620.020680.0 kW0.0 kW
P2Subcritical——-27,000 kg/h26 m³/h—0.099 barg2.4 barg0.45 bar—20 °C3.4 m/s0.0023171,4990.020680.0 kW0.0 kW

Node states and balances

NodeNamePressureTemperatureBoundary supply (+) / discharge (−)Composition mol %
N1Reservoir0.0 barg (surface)20 °C27,000 kg/hWATER 100
N2Reservoir0.0 barg (surface)20 °C-27,000 kg/hWATER 100
N3Junction2.4 barg20 °C-0.0 kg/hWATER 100
N4Junction2.4 barg20 °C-0.0 kg/hWATER 100

Junction port static pressures

JunctionPipe portReference portPort elevationStatic pressureVelocity headJunction loss
N1P3Liquid surface29 m0.013 barg0.057 bar0.029 bar
N2P2Liquid surface19 m0.099 barg0.057 bar0.057 bar
N3P1P30.0 m2.4 barg0.057 bar0.0 bar
N3P3P30.0 m2.4 barg0.057 bar0.0 bar
N4P1P10.0 m2.4 barg0.057 bar0.0 bar
N4P2P10.0 m2.4 barg0.057 bar0.0 bar

Graphs and equipment curves

Connection P1 · pump

Static pressure1.231.832.433.033.6300000 mbarg
Temperature18.8119.4120.0120.6121.2100000 m°C
Velocity2.182.783.383.984.5800000 mm/s
Mach number-1.2-0.600.61.200000 mMach
Pump curve at speed ratio 1-3.477.0517.5628.0838.5906.5813.1619.7426.32 m³/hHead mOperating point

Fan chart pressure is at reference density. Actual pressure rise scales with suction density. Shaft power: 0.0037 kW

Visual report

Download SVG

Untitled network · Converged · 2 flags

P1: flow N3 → N4 · 27,000 kg/hPump · design orientation A → B P127,000 kg/hΔP 0.0 bar · Rise 0.0038 bar P3: flow N1 → N3 · 27,000 kg/hP3: flow N1 → N3 · 27,000 kg/h P327,000 kg/hΔP 0.45 bar P2: flow N4 → N2 · 27,000 kg/hP2: flow N4 → N2 · 27,000 kg/hP2: flow N4 → N2 · 27,000 kg/h P2-27,000 kg/hΔP 0.45 bar N1 · ReservoirSurface 0.0 barg · z 30 mTake-off z 29 m · depth 1.0 mP3 outlet at reservoir · Static P 0.013 bargDepth 1.0 m · Hydrostatic head 0.099 bar N2 · ReservoirSurface 0.0 barg · z 20 mTake-off z 19 m · depth 1.0 mP2 inlet at reservoir · Static P 0.099 bargDepth 1.0 m · Hydrostatic head 0.099 bar N3 · JunctionP 2.4 bargStatic ref: P3 N4 · JunctionP 2.4 bargStatic ref: P1

Element colours · ▶ Arrows follow solved mass flow. ΔP = integrated pipe friction and fitting losses only, excluding elevation and acceleration. Pump rise is shown separately. Gas equivalent-nozzle loss split is unavailable. Node pressures are static gauge pressures. Arrows appear after convergence; none for closed or zero-flow connections. ■ Red = CHOKED (overrides colour map)

ThermoMix properties

Node N1

PropertyValue
P0.0 barg
T20 °C
Density1,007.84554 kg/m³
Viscosity1.042629 cP
Cp4,180.067 J/kg·K
Z1
Specific enthalpy-20.799 kJ/kg
Sound speedNot calculated for liquid
Molecular weight18.015 kg/kmol

At the node's solved state. The network uses the ThermoMix component database (ThermoWorks'). Liquid Cp is the common input; steam uses the ThermoMix approximation, not IAPWS.

Calculation audit

Hydraulic convergence-13.37-9.05-4.72-0.43.9214.758.512.2516 Iterationlog₁₀ scaled residual
Audit quantityValue
Hydraulic residual0.000001437
Node mass residual0 kg/s
Thermal mixing update0.0016314 K
Composition update0
Solution accepted numericallyYes
ConnectionEnergy residual WΣ fitting KConductance W/(m·K)Outlet density kg/m³
P100—1,007.83789
P3-00—1,007.8379
P2-00—1,007.83084

Mass and momentum

Displayed element ΔP integrates (f/D + K/L)ρv²/2 along the branch. It excludes hydrostatic and acceleration contributions. Pump pressure rise is reported separately; a pump-only element has zero pipe-friction ΔP. Junction-node losses are reported separately in the port table. Gas equivalent-nozzle restrictions do not resolve an irreversible pressure-loss split and display unavailable.

Signed mass flow is positive from node A to B. Each unconstrained node satisfies Σṁout − Σṁin + demand = 0. Pressure boundaries supply or absorb the remaining balance.

dP/dx + G² d(1/ρ)/dx = −(fD/D + ΣK/L) G²/(2ρ) − ρg dz/dx

G = ṁ/A. Darcy friction uses 64/Re in laminar flow, iterated Colebrook above Re = 4000, and a linear blend between Re = 2300 and 4000. Fittings are specified as explicit K values, distributed along the equivalent branch. Individual bend, tee and area-change nodes apply losses at outgoing ports using their selected reference velocity. Pipe fitting inventories remain separate. Avoid entering the same loss twice.

1/√fD = −2 log₁₀[ε/(3.7D) + 2.51/(Re√fD)]

Energy and heat transfer

d(h + v²/2 + gz)/dx = q′/ṁ

q′ = U′(Tambient − T)
1/U′ = 1/(hᵢ2πrᵢ) + ln(rₒ/rᵢ)/(2πkwall) + ln(rins/rₒ)/(2πkins) + 1/(hₒ2πrins)

Adiabatic: q′ = 0. Isothermal: specified pipe temperature, with required heat calculated from the endpoint energy balance. Heat loss: cylindrical wall and insulation resistances; user-specified film coefficients (external radiation may be included in effective hₒ). A simple exchanger uses fixed duty or UA against a fixed utility temperature.

Mixing uses incoming component mass rates and mass-weighted enthalpy. Junction mixing transfers incoming enthalpy and kinetic energy to the node reference port. Port static pressure is P_ref + α ρv_ref²/2 − α ρv_port²/2 − Δp_loss. The reference is the largest-flow incoming port, or largest-flow outgoing port at a supply. Junction transfer uses a low-Mach approximation and is flagged above M = 0.3; no high-speed tee or transition choking model is implied. Liquid Cp is a common user input, including mixed-liquid cases.

Properties and equipment

Gas: ideal gas or Peng–Robinson with kᵢⱼ = 0, ideal-gas Cp polynomials from ThermoMix, and PR enthalpy departure. Density and enthalpy derivatives are numerical. Liquids: additive specific volumes, logarithmic viscosity mixing, or measured constant density/viscosity. No excess enthalpy or activity-coefficient model.

Pump is a lumped two-port machine with no pipe section, fittings or pipe heat transfer. Its elevation sets both connection-node elevations; connected pipes carry the elevation changes. Legacy pump pipe properties are ignored. Pump curve: actual suction volume flow versus head. Fan curve: suction volume flow versus pressure rise at reference density. Compressor curve: suction volume flow versus pressure ratio, with ideal-gas isentropic work/efficiency. Speed scaling uses affinity approximations. Curve extrapolation is flagged and invalidates design acceptance.

Valve Cv uses liquid-equivalent resistance, with density updated along the branch. Gas valves and orifices use a compressible equivalent-nozzle law, with upstream total pressure/temperature estimated from the adjoining pipe. Orifices use Cd × bore area; valves use effective area from Cv. Gas k and Z are frozen at the restriction inlet. Vendor xT, pressure recovery and IEC sizing are not included. Regulators throttle excess available pressure; fully open behaviour applies when the setpoint cannot be reached.

Choked-flow solver and scope

Each passive gas branch is solved as a pressure-to-mass-flow characteristic. Pipe capacity is found by bracketed integration to the critical momentum/energy condition. Adiabatic pipes approach sonic flow; isothermal pipes reach their isothermal critical condition (approximately M = 1/√γ for an ideal gas). Heat-loss pipes use the coupled property derivatives. The network solves continuity with the resulting capped flow.

r* = [2/(γ+1)]γ/(γ−1)
ṁ* = CdA P₀ / √(Z RsT₀) · √γ · [2/(γ+1)](γ+1)/(2(γ−1))

For restrictions above r*, ṁ = CdA P₀ √{2γ/[Z RsT₀(γ−1)] × [r2/γ − r(γ+1)/γ]}; below r*, use the capacity plateau. PR cases use a frozen-Z restriction approximation; this is not a rigorous real-gas critical-nozzle calculation. Pipe profiles stop at the critical exit state. A downstream node represents a mixed receiver; jet kinetic energy is included in its incoming total enthalpy. Critical pressure and downstream receiver pressure are reported separately. An infeasible fixed demand is rejected by the mass/flow residuals, not silently clipped.

Visual output: choked items override all colour maps in red and carry a CHOKED note, limiting mass flow and critical pressure. If the network has not converged, these remain provisional capacity indications.

Liquid reservoirs

Fixed surface pressure and elevation define a steady-state liquid reservoir. Connected pipes can share a take-off elevation or use individual depths/elevations and entry/exit losses. Hydrostatic pressure at rest is P_surface + ρg(z_surface − z_takeoff). For discharge into a pipe, subtract (1 + K_entry)ρv²/2. For inflow from a pipe, add (K_exit − 1)ρv²/2. Density comes from the selected liquid properties. Pipe elevation changes start at the take-off, avoiding double-counting reservoir depth. The surface is treated as negligible velocity. No level depletion or gas-space compression is simulated.

Junction geometry and losses

Bends use a specified K, or user K90 scaled by angle/90. Radius/diameter is recorded for traceability; it does not generate an automatic correlation. Tees use explicit outgoing-path K values and assigned A/B/C ports; geometry angle is recorded and does not alter manual K. Conical area changes use Crane-style angle/area-ratio loss coefficients; abrupt selects the 180° limit. These are not vendor-specific or detailed AFT tee correlations.

For β = small/large bore and s = sin(θ/2), expansion K (upstream basis) = (1−β²)² × (2.6s for θ ≤ 45°, otherwise 1). Contraction K = (1−β²)/β⁴ × (0.8s for θ ≤ 45°, otherwise 0.5√s). Manual K can use incoming or outgoing velocity.

AFT documentation: Crane conical area-change loss model. Abrupt contraction here uses the Crane 180° limit, not AFT’s alternative Idelchik abrupt correlation.

Drawing symbols and orientation

General liquid pump: circle with filled delivery triangle (2301). General compressor: circle with open delivery triangle (2302). Symbol basis: ISO 10628-2:2012 process diagrams and its ISO 14617 references. Fan, valves, heat exchanger and strainer use schematic vector outlines. Pressure/flow boundaries are model annotations. This is a process schematic, not a certified fabrication drawing.

Equipment symbols follow the local pipe tangent in the defined A-to-B direction. Flow arrows follow the calculated mass-flow direction, independently of static-pressure rise or fall. Connected bends and tees align their arms with the isometric pipe routes; assigned tee ports A/B form the run. Moving nodes updates the routes and symbols. Disconnected junctions use a default isometric pose until connected. Visual routing does not change hydraulic lengths, elevations or entered bend angles.

ISO 10628-2 reference

Scope and references

Steady-state, Newtonian, single-phase networks with subcritical and choked pipe/restriction branches. Downstream supersonic jets and shocks are not resolved. Fans and compressors retain their curve models and are not sonic-capacity maps. Pressure range defaults to 0.001 bara to 20 barg. Exact vacuum is excluded. No water hammer, tank inventory dynamics, two-phase hydraulics, solids, non-Newtonian flow, surge or rotating-machine stall maps.

Phase screening retains the supplied approximate vapour-pressure flash. The embedded package is not IAPWS steam tables; near-saturation steam and nonideal liquid mixtures require independent property validation. Red or amber flags must be resolved before using a result for design.

Built-in verification

The page's verification cases run as the engine's tests with every release, not on request: laminar Darcy f at Re = 1000; Hagen–Poiseuille pressure drop in one pipe (within 0.01 Pa) and its flow; the water loop's and the compressed-air network's mass balances (within 10⁻⁶ kg/s) and convergence; the choked orifice converging, flagged and holding its flow from atmospheric to 30 kPa back pressure; the critical ratio for γ = 1.4; and the °F and psig conversions. These checks establish numerical and unit-conversion behaviour, not full validation of every property correlation or item of equipment.

Methods

  • Each connection integrated along its length: friction, elevation, acceleration and heat transfer
  • Newton's method on the nodes' mass balances and the connections' pressures, with mixing at the nodes between solves
  • Gas by Peng–Robinson (or ideal gas) with Wilke viscosity; liquids from ThermoWorks' correlations
  • Choked flow in pipes and restrictions

How every calculator is tested: Methods and validation.

Limits

  • Steady state, single phase and Newtonian
  • No water hammer, surge, two-phase flow or solids
  • Steam from ThermoWorks' correlations, not IAPWS
  • Up to 100 nodes and 150 connections; larger networks take longer to solve

Results are screening estimates from the methods described. A competent engineer must check them before relying on them in design, safety or operating decisions.

How it is checked

Parity with the original
The engine is a port of the original standalone calculator, and reproduces it on 315 generated cases: every quantity within 10⁻¹² relative, except 11 groups of quantities with a recorded tolerance of its own, where the original's numerical methods amplify rounding (iterative solvers, finite differences, residuals near zero).
Validation against reference data
No published reference cases yet.
The original's own checks
The original's verification networks: Hagen–Poiseuille flow, balanced loop and compressed-air examples, a choked orifice and a nozzle's critical pressure ratio.