Skip to content
Index ☰
Independent engineering research · Open source

Show the working.

Independent engineering software, numerical experiments and publicly documented energy research—what works, what has been measured, and what remains uncertain.

I'm Abhijith Sivaprasadan. Open Energy Methods is my public engineering workbench: where I build things, test assumptions, document failures and share methods that others can inspect, reproduce and improve.

Explore documented projects Read the field notes

Created and maintained by one engineer · October 2026 onward

01 / On the workbench

Four real research records.

Current software and analysis, linked to real tests, methods and source records. EnerLint is developed within OEM. The other three are separately maintained personal repositories by the same creator—not transferred OEM modules.

OEM repository / EnerLint v0.1.0a0 / Unreleased alpha

EnerLint

Can a small, independent check tell us whether exported model values obey their declared mass and energy balances?

Implemented now: steady control-volume mass/first-law checks, one-interval thermal storage accounting, strict SI JSON inputs, reports and fault-injection tests. The initial benchmark is synthetic; a deliberately compensated compressor error still escapes. No independent external review, public package release or DOI yet.

Source and installation Benchmark results
4 / 4

Hand-constructed baseline cases pass their declared checks.

6 / 6

Deliberately faulty cases fail as expected.

1 escape

A compensated operating-point error still passes. No detection-rate claim.

Evidence: 27 Oct benchmark record and fault protocol. Developer-selected synthetic examples, not external scientific validation or operational fault-detection accuracy.

Other active personal research

EVIDENCE CHECKED · 27 OCT 2026 · SEPARATE REPOSITORIES
Gas turbines · Fortran 2008 · Source-only beta

ThermoTwin-F

Brayton/combined-cycle simulation, degradation, transient and diagnostics work. A tagged source beta is archived with a Zenodo DOI; it is not validated on commercial plant measurements.

v2.0.0-beta.1 · 10/10 release CI jobs · DOI 10.5281/zenodo.23679426
Evidence and limits ↗Source release ↗
Industrial steam · Python / Modelica · Synthetic

OpenSteamOpt

A two-boiler steam-and-power demonstrator with Pyomo/HiGHS scheduling, shadow dynamics and FMU integration. Reported cost differences use an author-built baseline, not a real plant.

3 reproducible scenarios · 3.34–8.95% lower modelled costs vs its own comparator
Evidence and limits ↗Source ↗
Kerala electricity · Observations / ERA5 · 2026 evidence freeze

Kerala2040

Official-statistics reconciliation, daily SLDC data auditing and retrospective weather–demand analysis. Operational hourly network results are not established by the available sources.

354/365 reported SLDC days · 353 observed daily targets matched to ERA5
Evidence and limits ↗Research site ↗

Want the numbers, not the summary? The research catalogue contains dated evidence, reproduction steps, negative results, missing validation and direct source links for all four projects.

Open research catalogue

All three projects have substantial code and documented evidence. They remain separate personal repositories. Figures shown are bounded by their own study designs, reference inputs and release status; linking them here does not transfer code ownership or validate EnerLint.

02 / How I work

Keep the method inspectable.

The workbench is a place to publish both answers and the route taken to reach them. Not every experiment will become software, and not every result will be positive.

01 — ASK

Start with a real question.

State the physical problem, the quantities involved and what would count as a useful answer.

02 — TEST

Show the calculation.

Keep inputs, units, methods and tests visible. Use analytical references or independent evidence when the question requires them.

03 — SHARE

Publish the edges.

Document what passed, what failed, what was not checked and what someone else would need to reproduce the work.

03 / A small calculation

Try a balance.

A browser-only illustration of the energy residual for one steady control volume. Change a value and watch the balance change. This is an educational sketch—not the EnerLint package or a validated simulation.

STEADY TURBINE / ONE INLET · ONE OUTLET / SI UNITS DECLARED THRESHOLD: 1 W
kg/s
J/kg
J/kg
W
W
rE = ṁ(hin − hout) + Q̇in − Ẇout
● PASS · DEMO ONLY r = 0 W

The energy residual is within the 1 W demonstration threshold.

NOT CHECKED — AND STATED

Equations of state, phase validity, whether the chosen boundary is correct, measurement truth, kinetic or potential energy and independent verification of the supplied operating point.

Runs entirely in your browser, with no inputs sent to a server. Equal inlet/outlet mass flow is assumed; only the energy residual is calculated. A passing result is not proof of the model. Why a balance can pass for the wrong reason ↗

04 / Field notebook

Notes from the working.

Not press releases. Specific examples, tests, limitations and open questions, each linked to the material behind it.

NOTE 001 · OCT 2026

Why a passing energy balance is not proof

Two compressor operating points, the same zero residual, and the evidence that is missing.

 METHOD / REFERENCE

How the first fault tests are defined

Hand-constructed baselines, deliberate mutations and the case conservation cannot catch.

 OPEN QUESTION / ISSUE

Do these checks add value beyond native diagnostics?

A comparison still to be performed—not a finding already established.
05 / The person behind the work

One workbench.
Many questions.

I'm Abhijith Sivaprasadan, an engineer working across sustainable energy engineering, thermodynamics and computational modelling.

Open Energy Methods connects public work that already exists: EnerLint’s conservation checks, a citable ThermoTwin-F source beta, OpenSteamOpt’s synthetic plant scheduling studies and Kerala2040’s source-audited electricity research. This site points to the original repositories so the evidence can be checked rather than merely described.

A finished run is a starting point for scrutiny, not necessarily the end of the work.

OEM is not a registered company, a university research group or a certification body. There is no requirement for it to become any of those. For now the objective is simply to make worthwhile engineering work open, understandable and reproducible.

Abhijith Sivaprasadan · Creator & Maintainer

Personal GitHub ORCID