Shriji Barot

Third-party technology evaluation: an emissions reduction claim

Problem

An outside inventor submitted a retrofit device for small internal combustion engines, claiming it largely eliminated carbon emissions while multiplying fuel efficiency, alongside several exotic physical claims about what was happening inside it. I ran the evaluation program and owned the determination.

Designed and ran

A matched control-versus-retrofit campaign across multiple engine sizes and load levels, using identical fuel volumes, two independent exhaust gas analyzers from different manufacturers, and component-by-component disconnection to isolate what each part of the device actually contributed. I also tested a post-removal state — device stripped off, engine otherwise unchanged — to confirm the baseline returned.

What the data showed

The retrofitted configuration produced exactly what the claim predicted: dramatically lower carbon species and much higher oxygen in the exhaust, consistently, across every load and both engines. The fuel efficiency claim did not hold at any load.

Where it broke

The analyzers measured concentration, not mass flow. The device introduced a large volume of additional air into the exhaust path, and the apparent emissions improvement scaled with how much air it pulled. Without intake and exhaust flow measurement there was no way to convert concentration into total emitted mass — which is the quantity the claim was actually about. A device that dilutes its own exhaust looks identical, on a concentration-only instrument, to one that eliminates emissions.

Also documented

A leak in the device that could recirculate exhaust back to the intake, inconsistent probe placement between runs, and heat damage to one analyzer's probe partway through the campaign. Each was written into the report as a stated limitation on the data rather than smoothed over, because a reader needs to know which conclusions the apparatus could actually support.

Outcome

The determination was that the claim was not demonstrated, together with a specification of exactly what would demonstrate it: intake and exhaust flow measurement, fuel consumed by mass rather than volume, in-situ mass spectrometry to detect alternative carbon products, before-and-after mass measurement of the device to test for deposition, and calibrated power measurement at the load. I designed that follow-on protocol. The point was never to dismiss the technology — it was to make the next round capable of producing an answer either way.

* Details of the technology, the parties involved, and the measured values are confidential and have been generalized here. What is described is the validation approach and the reasoning behind the determination.

Flow calorimetry resolving ±0.05°C

Problem

Measuring small thermal power changes on experimental energy systems in electrically and thermally noisy lab conditions, where the effect under investigation was close to the noise floor.

Built

A water flow calorimeter resolving ±0.05°C differential temperature — roughly ±1.5 W on a 100 W system — with full error analysis, baseline drift correction, and uncertainty quantification built into the measurement rather than estimated afterward.

Verified

Calibration runs against known resistive input power across the operating range, with mass flow cross-checked gravimetrically against the flow meter's own reading.

Charged particle detection below 100 keV

Problem

Distinguishing genuine charged particle events below 100 keV from electrical noise and detector artifacts, where a false positive would have been indistinguishable from a real result.

Built

A solid-state detector chain with RF amplification, using pulse-shape discrimination and shielding geometry to separate signal from background below 100 keV, paired with custom Python peak discrimination software. Also worked with nuclear track detectors, diamond detectors, and gamma spectroscopy across the same programs.

Verified

Shielding geometry validated against modeled attenuation; extended background characterization runs analyzed with identical discrimination logic to establish the false positive rate.

Distributed DAQ across six laboratory sites

Problem

Synchronized, trustworthy data from many concurrent experiments across six sites, at a hardware cost the program could actually carry. The deeper problem was fragmentation: instruments accumulated over years, each with its own vendor software, file format, and clock, reconciled by hand in spreadsheets afterward — which is exactly where errors enter and stay invisible.

Built

Distributed data acquisition and control in Python, SQL, and LabVIEW, with custom multi-channel isolated instrumentation replacing commercial equipment that would have consumed a large fraction of the budget.

Centralized control

The part that mattered most was one piece of software that talked to every instrument on the bench regardless of vendor or protocol, streamed everything into a single timebase, and let an operator configure and run an experiment without touching six different applications. Parallel experiments ran under independent control, with structured session and channel schemas and a browser-based interface for remote monitoring. Simple enough to use correctly under pressure, structured enough that a dataset from eighteen months earlier could still be traced to the configuration that produced it.

Verified

Per-channel calibration frameworks, with configuration control and version-managed releases through GitHub CI/CD so any recorded dataset could be traced to the exact software and hardware revision behind it.

Gas dosing and pressure regulation for under $300

Problem

Precise gas dosing and pressure control for electrolytic and metal hydride platforms, without commercial controllers priced well beyond what the experiment could justify.

Built

A dosing and regulation system using orifice gaskets, low-cost solenoid valve actuators, and a custom Python control layer — specified against what the experiment actually needed to resolve rather than against a catalog.

Verified

Dosing and pressure accuracy characterized directly on the platforms it served, under operating conditions.

Calibrated four-wire PT1000 probes for under $20 each

Problem

Programs needing dozens of accurate temperature probes, where catalog prices made adequate sensor coverage unaffordable and teams were tempted to under-instrument.

Built

Custom four-wire PT1000 RTD probes using top-tier sensing elements, assembled and calibrated in-house for under twenty dollars each, with shielded cabling designed for the electrical environment they operated in.

Verified

Calibrated against a reference standard before deployment, with periodic recalibration built into the lab's operating procedure.

High-voltage plasma experiments on UHV platforms

Problem

Running novel plasma experiments safely and repeatably on systems with genuine hazard potential and no established operating procedure.

Built and ran

High-voltage plasma experiments including AC plasma and pulsed capacitor-bank discharges, on ultra-high-vacuum and gas handling platforms. I owned the experimental design, the safety procedures, and the failure mode analysis.

Verified

Full system validation before each campaign, with interlocks and procedure review as a precondition for operation.

Note

Specifics on voltages, energies, and pressures are covered by confidentiality.

Materials characterization supporting experiment interpretation

Problem

Understanding what physically changed in a sample after a run, to distinguish genuine material effects from contamination, handling damage, or surface artifacts.

Worked with

SEM/EDS, TOF-SIMS, XRD, Raman spectroscopy, and surface profilometry, alongside sample preparation including plating, annealing, etching, and rolling.

Verified

Pre- and post-run characterization on matched samples, with control specimens processed identically but not exposed to the experimental condition.