Historical localhost engineering review. This hosted preview now uses browser-scoped cloud storage; no local experiment files were migrated. Return to upgrade

◈ GFIS
CHATAKE INNOWORKS / ENGINEERING RELEASE V2

A plant you can take apart.
A system you can inspect.

Modular 3D engineering, instrument commissioning and time-coupled process scenarios—connected to GFIS’s retained research evidence in an isolated localhost workspace.

10SELECTABLE PLANT ASSEMBLIES
6 + 5CAMPUSES + CITY DISTRICTS
46/46AUTOMATED CHECKS PASSED
8RETAINED REPLAY-MATCHED SCENARIOS
Local engineering simulation, not live physical plant operation. No physical IoT devices or actuators are connected. Actual WebGL, screenshots and mobile visual acceptance remain unverified because browser-control access was unavailable. Tested geometry and DOM behaviour do not prove pixel appearance.
01 / OPEN WORKSPACE ↗

Custom plant

Ten reversible assemblies, instrument ports, cutaway internals and capacity-linked inventories.

02 / OPEN WORKSPACE ↗

City systems

Six campuses, five districts, shared storage, constrained gas dispatch and 72-hour scenarios.

03 / OPEN WORKSPACE ↗

Measured evidence

500,400 archived SCADA records remain separate from generated plant and city states.

04 / OPEN WORKSPACE ↗

Research intelligence

Retained persistence, XGBoost and LSTM comparisons; actual epoch histories, not simulated training.

One source of state, several ways to understand it.

Design & instrument inputsAssembly graph / compatible ports
Validated commandRevision / readiness / mode
Process engine15-minute integration
Retained state3D / plots / alarms / replay

Exploding the model is a display operation. Removing a module is an engineering-state operation. That distinction keeps an attractive animation from masquerading as working plant logic.

Operate the custom plant

  1. Open Plant studio. Create a new plant; select components and inspect their ports.
  2. Drag the exploded-view slider and enable cutaway. Then isolate and disassemble the actual assembly state.
  3. Assemble the plant and attach essential sensors. Set feed, temperature feedback, agitation and gas demand.
  4. Start and advance time. Watch inventories, energy totals and process history change together.
  5. Inject a required instrument fault. Inspect the latched trip, restore the sensor, reset and restart.
  6. Verify replay and export the complete session. Reopen it from Experiment memory.

48-hour modular plant experiments

Identical initial sizing and default process assumptions; one control intervention per variant. Outputs below were newly computed and retained on SSD. They are model responses—not observations or a validation of these kinetic constants.

ScenarioDuration hBiogas Nm³CH₄ Nm³Temperature °CGross electricity kWhReplay
Reference commissioning48.001002.42601.4539.102301.19MATCH
Heating unavailable48.00998.06598.8334.532292.09MATCH
Low agitation48.00692.66415.6039.101654.61MATCH
High gas demand48.001002.42601.4539.102301.19MATCH
Reference commissioning1002.4 Nm³
Heating unavailable998.1 Nm³
Low agitation692.7 Nm³
High gas demand1002.4 Nm³
Gas residual = initial inventory + produced − stored − dispatched − flared
Organic residual = initial organics + input − inventory − destroyed − withdrawn

Biogas is derived from simulated destroyed organics and an assumed methane volume fraction; gross electricity uses fixed heating-value/efficiency assumptions. Net energy, gas cleanup efficacy and plant economics are not established.

72-hour city resilience

ScenarioGas served Nm³Unmet demand Nm³Final storage Nm³Max cumulative residual Nm³Replay
Balanced city44640.000.0013765.253.09e-11MATCH
Peak demand58594.4021757.601740.002.18e-11MATCH
Transfer bottleneck26240.0018400.008700.004.00e-11MATCH
Campus outage43470.271169.739577.732.91e-11MATCH

Supply, demand, storage and link bottlenecks produce the visible city state. The scene is schematic: pipelines show topology, vehicles illustrate collection connectivity, and no pressure/compressor or vehicle-route solver is claimed. City supply profiles and custom-plant kinetics are not yet coupled.

Engineering checks that changed the implementation

Evidence retained, claims kept separate

DS-05 research observations, DS-06 historical total-biogas SCADA, dated persistence/XGBoost/LSTM predictions and saved epochs remain available. The current work did not train a new CNN, perform new LSTM optimisation, validate VFA/ALK or connect a physical plant. Existing gasification remains an educational thermochemical module.

Training history ↗

Open actual retained epochs and configuration, distinct from animation playback.

Original project and evidence ↗

Reach GFIS engineering journey, dissertation, evidence reports and both public portals without replacing them.

Verification record and next gate

Generated from MODULAR_V2_RELEASE_VERIFICATION.json at 2026-09-08T17:29:22.337Z. Node v22.19.0. 17 HTTP asset/API checks passed; 46 automated tests passed. Geometry tests use real Three objects with a stub renderer; browser pixel QA is outstanding.

Next gate: desktop/mobile WebGL acceptance, followed by a calibrated read-only physical telemetry adapter, prospective shadow-mode evaluation and reviewed control authority. No AWS/DNS release occurred.

Pre-upgrade local Git tag: gfis-local-v1-before-assembly. Current stores and previous source remain retained.

Preserved first local review ↗ · GFIS public platform ↗ · India collaboration ↗