UNTIRTA · ENGINEERING RESEARCH

CFD Cavitation Lab

Baseline MRF · Cavitation · Physical Transient

CFD pompa · progres solver lokal

Progres run pompa tersedia pada CFD Cavitation Lab di laptop worker.

Ini adalah pemantauan perhitungan OpenFOAM. Kontur tekanan/kecepatan belum tersedia di halaman ini; angka sementara tidak membuktikan konvergensi.

Buka pemantauan CFD lokal

Native OpenFOAM · educational benchmark

Run a fixed educational benchmark on the paired laptop. The laptop must be on and connected. These runs use their own fixed inputs; they do not calculate this study's pump performance or NPSHr.

Administrator · connect owner worker

Loading native job controls… Python study calculations remain independent.

No native job submitted in this tab.

Jobs in this approved session

Inspect jobs belonging to this approved session while it remains active. Opening a job does not run it again. Re-pairing creates a separate session and does not grant access to earlier sessions' jobs.

Scope, limits and recovery

Approved-session pilot: one executing job, two active jobs globally, eight new jobs per UTC day; one active and two new jobs per session per UTC day. Pairing lasts one hour and stays in this tab's memory. Reload ends access to that session; save the report before reloading. A lost execution lease interrupts the job; execution is never retried automatically. Reports are retained for two days. Native case files remain on the laptop.

The report records numerical smoke checks and original input/reference checksums. It does not establish exact-pump validation, mesh/time independence, NPSH3, or experiment agreement. Start the independent native worker on the owner laptop and approve the request with its launcher. No paid compute fallback is used.

Input → Equation → Substitution → Result

CFD studies use their own internal geometry, mesh, operating conditions and solver evidence. Main steady-state results provide a reference. Level 1/2 remain available in the startup Lab.

Frozen rotor/MRF provides a steady baseline. Startup requires physical time. Native case/data files or contours alone do not establish convergence or NPSH3.

01

Study input and reference

Load a complete educational study to explore each Python stage, or use an empty template for your own study. Source inputs, immutable references and derived outputs are stored separately.

Use a copied Current reference from Main

Select a snapshot already stored in this Lab. Python copies the available flow, RPM and properties into a CFD draft; unavailable geometry and inputs remain unfilled.

No study input loaded. This page has not run a solver.

Inputs stored in the Lab workspace

Loading inputs does not restore old results as Current.

03

Equation stages from Python

Validate inputs to view equations and required operands. Missing inputs limit only their dependent branches.

04

Solver evidence and reduction

Import evidence JSON separately from inputs. Declared or contract evidence does not establish validated CFD. A 3% head reduction requires a baseline at the same Q/RPM/fluid and a crossing bracketed by the pressure sweep.

No solver evidence loaded.

Raw values, provenance and reduction results
05

Native checkpoints and restart

Case/data pairs must match the mesh, input revision, snapshot, physical time, step and rotor position. Import alone does not establish Current results or physical validation.

Validate the study first. The browser import limit is 25 MiB; native outputs remain in the Lab workspace.

Manifest and restart checks
DEMO

Educational CFD demonstration

Explore checked native solver observations and published assembly drawings. Selecting a demonstration displays its evidence; it does not run a solver or change a study.

Explore time-step evidence and source drawings

Loading the checked educational presentation.

Full pump workflow · theoretical framework and reference

The official PyFluent frozen rotor example provides a full impeller–volute workflow with rotating and stationary domains and their interface.

  1. Steady baseline: prepare the full impeller–volute fluid domain, mesh and interface, then solve the frozen rotor/MRF baseline at a fixed flow, speed and fluid.
  2. Cavitation: run an inlet-pressure sweep against that matched baseline and inspect vapor formation and the head-reduction crossing.
  3. Physical transient: use sliding/moving-mesh physical time, record rotor position and histories, then refine the time step on the same mesh and operating conditions.

The example requires Fluent 2025 R2 or later. Its 200 iterations are steady solver iterations, not physical time steps. The checked demonstrations here use OpenFOAM; the educational 30° sector and supplied assembly drawings do not establish an exact Grundfos full-pump CFD model.

06

Mesh and time-step verification

Import an independent numerical evidence packet from a controlled three-run series. Python checks matching geometry, physics, boundaries, sampling definitions and physical time before assessing sensitivity. Source declarations and numerical convergence remain distinct from physical pump validation.

No numerical series loaded. This assessment does not run a solver or modify the selected pump study.

Raw series, numerical criteria and provenance

PYANSYS · PYTHON SOLVER INTERFACE

PyFluent API

Local inspection required

Inspect the PyFluent SDK in the independent Python worker. The SDK provides the Python interface for Fluent; solver availability, connection, license and CFD evidence are checked separately.

Local Python inspection required. The cloud page does not install the SDK or execute Fluent. GitHub installation evidence is available below.

SDK versionNot inspected
Python import and entry pointsNot inspected
SDK runtimeIndependent Python worker
Native Fluent solverNot inspected
License verificationNot checked
Fluent sessionNot checked

Open the local CFD Lab and inspect its Python worker. This inspection does not start a solver or change study inputs.

A successful SDK installation does not establish a Fluent session or a verified CFD run.

Local API inspection and connection prerequisites

The Python worker reports which API entry points were imported. No launch or connection is attempted by this panel.

No local API inspection has been requested.

Fluent requires its own compatible installation and license. The available PyFluent profile is a steady frozen rotor baseline; physical transient and NPSHr claims require their corresponding solver evidence.

OFFICIAL PYFLUENT EXAMPLE · SOURCE REVIEW

Impeller–Volute Frozen Rotor workflow

Steady baseline example

Study the supplied frozen_rotor_workflow.py and notebook alongside the API. The source sets up a rotating impeller and stationary volute in Fluent. Its notebook has no executed cells or stored results; displaying this workflow does not execute it.

The verified learning package includes both actual official HDF5 meshes, source, notebook and original license notices. Inspect upstream provenance and SHA-256. It contains no solver results.

Actual mesh inventory · Student FREE fit

Exceeds Student cell limit

Python read the actual HDF5 cell metadata and stored ranges in both unmodified source meshes. The combined count below is a metadata sum; no joined Fluent case has been generated.

Impeller stored cells773,352
Volute stored cells368,411
Combined stored cells1,141,763
Student cell limit1,000,000

Input

Impeller: 773,352 cells
Volute: 368,411 cells
Criterion: 1,000,000 cells

Equation

Ntotal=Nimpeller+Nvolute

Substitution

773,352+368,411=1,141,763

Result

1,141,763>1,000,000Nexcess=141,763 cells

The cell-count criterion is not met.

The official Ansys Student fluid limit is one million cells/nodes. This assessment compares cells only: the unmodified impeller–volute example exceeds that cell limit by 141,763 cells. A lighter mesh must be derived and checked separately; the original meshes remain unchanged.

This is Python metadata arithmetic, not a physical CFD or NPSH result. Meeting a cell limit would not establish mesh quality, sufficient RAM, a Fluent license, convergence or CFD readiness.

Inventory provenance and numerical source
Cell-count evidence
ConfirmedFromStoredRanges · impeller zone 36 · volute zone 2
Read-only GitHub inspection
Run 38152095914
Raw inventory SHA-256
7c33b681d30be88ad2c5396d27e4990d34cf2463e28ab546ce52d01a27ff65e4
Python projection SHA-256
51f2cab88f64db972c3398d82032a992e1f0a0ba478cdfdbc4c609780062415b

No solver session, native case, mesh-quality check or CFD calculation was performed by this inventory.

Download actual mesh inventory JSON

Local OpenFOAM preparation · 29 October 2026

Full-pump run not ready

Both official meshes have been converted through the native OpenFOAM importer. The original HDF5 files remain unchanged. This dated preparation record is separate from the Fluent inventory above.

Impeller native meshMesh repair required

773,352 cells

Volute native meshMesh OK

368,411 cells

Impeller corrective action: 11,310 cells below the native determinant threshold. Repair the derived mesh and repeat the full native geometry/topology checks before solving. Volute passed those checks.

  1. Fluent access: the isolated SDK is installed; a licensed native session remains unverified. Complete the official academic download verification.
  2. Mesh capacity: the format-only OpenFOAM derivatives retain all 1,141,763 original cells. Memory, mesh quality and runtime must still pass their own checks.
  3. Full-pump application job: rotor/stator coupling, declared fluid/pressure conditions and the fixed execution route require verification before Run can be enabled.

No full-pump flow solve, converged head, cavitation, NPSHr or NPSH3 is established by this preparation. Main and Level 1/2 calculations retain their independent acceptance.

Remesh and interface review · 29 October 2026

3 of 5 local correction candidates retained every original boundary facet; 0 passed the full native mesh checks. The volume rebuild attempts did not produce an accepted mesh within their bounded runs.

Interface geometry: no zero-coverage faces. Area-weighted coverage is 99.999858% on the impeller and 99.999603% on the volute. These geometric observations do not establish conserved flow.

Next gate: an accepted impeller volume mesh, then combined coupling, flux conservation and residual/head convergence. No full-pump flow solve has run.

Remesh measurements and claim limits
  • impeller-r1: 5,856 small-determinant cells; 2 failed native checks; boundary facets changed. Rejected.
  • impeller-a1: 396 small-determinant cells; 10 failed native checks; boundary facets changed. Rejected.
  • impeller-b1: 4,026 small-determinant cells; 10 failed native checks; boundary facets unchanged. Rejected.
  • impeller-c1: 11,274 small-determinant cells; 2 failed native checks; boundary facets unchanged. Rejected.
  • impeller-s1: 9,252 small-determinant cells; 6 failed native checks; boundary facets unchanged. Rejected.

Native quality thresholds remain unchanged. Boundary comparison retains exact coordinates, winding, patch names and duplicate-face counts. Gmsh volume rebuilds use a separate one-CPU, 2 GiB, 240-second process scope.

Review SHA-256: 179c1dbdf53fa4c4e13b63349d4b5fbc8043fd69860494cd54a3f6f95facf201

Native preparation evidence and rejected remesh candidates

OpenFOAM Foundation 12 / build 12-86e126a7bc4d / WSL Ubuntu 24.04. Full checkMesh geometry/topology checks; unchanged quality thresholds. Native conversion/remesh jobs were capped at one CPU, 2 GiB RAM, no swap, 16 tasks and 240 seconds, without network access.

impeller-dual-no-interior: 10 failed native checks; not selected. impeller-dual-concave: 10 failed native checks; not selected.

Impeller log SHA-256
2e1b0d300c3b02609d71ebe3cabbe3fe74981a962192a7d8661e704c5bbce042
Volute log SHA-256
12505ab23ec2ebd41215d124c5620d07ddcd3994c02090652648aed090ee03c2
Preparation record SHA-256
4173fd871d2968d429c3ba128c1ee243b56f8caca5099013624ce3ce81ea1b9c

1 · Mesh and domains

Impeller mesh
impeller.msh.h5
Volute mesh
volute.msh.h5
Example source folder
pyfluent/examples/pump-volute
Domain interface
imp-volute-interface · No-Pitch-Scale

The original ZIP contains Python and notebook source. The two native meshes are separate downloads; they are not calculated from Main inputs.

2 · Fluid and rotating frame

Material
water-liquid · Fluent database
Turbulence
k–ω SST
Speed declared in source
1450 rpm
Axis origin / direction
[0, 0, 0] / [0, 0, 1]
Head-report constants
ρ = 998.2 kg/m³ · g = 9.81 m/s²

ω = 1450 × 2π / 60 rad/s is the source conversion. Verify the active Fluent setting and units before execution.

3 · Boundaries and pressure reference

Inlet source setting
supersonic_or_initial_gauge_pressure = −100
Outlet mass flow
90 kg/s
Outlet zone name in source
mass-flow-inlet-11, converted to mass-flow-outlet
Pressure reports
Mass-averaged total pressure at inlet and outlet

The source setting above is not an absolute suction pressure. Check the active inlet boundary and operating-pressure datum before any NPSHA or cavitation comparison.

4 · Solve and inspect native outputs

Required Fluent release
2025 R2 or later
Source process count
4
Initialization
Hybrid · absolute reference frame
Source iteration count
200 steady iterations
Pseudo-time scale factor
10
Contour plane
XY plane · z = −0.015 m

Neither iteration count nor pseudo-time defines a physical startup time step. Convergence must be checked from actual residuals, reports and balances.

Learning sequence · source settings to solver evidence
  1. Inputs: obtain both meshes, inspect named zones and check the internal impeller–volute interface.
  2. Setup: inspect water properties, rotating frame, wall motion, pressure reference and the specified mass-flow outlet.
  3. Report equation: the source defines H = (pt,out − pt,in) / (998.2 × 9.81), using mass-averaged total-pressure reports. No head result is supplied by this source package.
  4. Steady solution: initialize, iterate and inspect native convergence and mass balance before accepting the baseline.
  5. Native records: the source plans pump_voulte_setup.cas.h5 before calculation and pump_volute_setup_solved.cas.h5 with case/data afterwards. Those names identify planned files, not files generated by this page.
  6. Research extension: cavitation requires a matched vapor model and pressure sweep; physical transient requires its own motion, time history and time-step verification.
Source fidelity and engineering checks

Retain the original script and notebook. The source declares viscosity 0.001002 kg/(m·s) and copies water from the Fluent material database; check the actual material properties used by the solver. Record the Fluent release, mesh identities, interface, pressure datum and applied settings in the independent study. The official example is a learning baseline; it supplies no cavitation sweep, NPSHr curve or transient CFD history.

Worker and execution limits

OpenFOAM is the primary adapter. Optional PyFluent requires a compatible Fluent installation. GitHub workflows run fixed development benchmarks within the FREE allowance; user jobs run through separate services.

Solver inspection has not been requested.

Worker capabilities