Neurotic
Menu

News

LNG

Your plant in one conversation. From PI tags to firewall rules.

AVEVA PI System, PLCs, OT networks, Fortinet, Samsara and Azure, handled by people who have worked inside LNG and midstream facilities. OT and IT security is part of the job, not a separate quote.

Typical issues PE faces in LNG

Not the plant's problems. The four that land on the fund, and that we find at almost every terminal we open up.

  1. Availability is the case, and you cannot evidence it

    The model was built on train availability. What the business can actually produce is a spreadsheet compiled a week later, which cannot tell you whether a train stopped or ran derated.

  2. The turnaround is the year

    The largest controllable cost and the largest schedule risk in the hold, scoped from maintenance records nobody fully trusts and planned in spreadsheets nobody else can open.

  3. Standing vendor access to a control network

    Laptops admitted under an exception granted years ago, on a site that is usually designated critical infrastructure. It reaches you through a regulator, an insurer or the next buyer's diligence.

  4. Nobody will touch the safety system

    Change near a safety instrumented system is slow and expensive, so nothing changes. The debt you bought is the debt you sell, and it is priced both times.

What it runs on

Six layers, and our engineers have configured systems in all of them on real lines. The names are what a plant actually has, not a wish list.

  • Process historian

    • AVEVA PI System
    • PI Vision
    • PI Asset Framework
  • Control and HMI

    • iFIX
    • Ignition
    • Emerson DeltaV
    • Honeywell Experion
  • PLCs and safety systems

    • Allen-Bradley
    • Siemens S7
    • Triconex
  • OT network and security

    • Fortinet
    • Claroty
    • Nozomi Networks
  • Fleet and field

    • Samsara
    • IBM Maximo
  • Cloud and identity

    • Microsoft Azure
    • Entra ID
    • Microsoft 365

Third-party names are listed so you can see what we work in. They are the property of their respective owners and imply no endorsement.

What we do about it

In this order, because the third of these is not possible without the first.

  1. Collect what the historian is licensed for

    Tags reviewed against the licence, the buffer watched, and the gaps closed at the interface rather than explained in a report.

  2. Separate the control network

    Segmentation, vendor access that expires on its own, and monitoring that knows a PLC from a laptop and will not patch one.

  3. Make availability readable daily

    Train performance read off the historian the same morning, so a derate is a question that day rather than a finding next week.

  4. Scope the turnaround from condition

    Maintenance history and tag data in one place, so the work list is argued from evidence rather than from memory.

You have control engineers already. This is not their job.

They are keeping a terminal running, and they are good at it. Every hour spent on tag hygiene is an hour not spent on the plant, and the plant wins that argument correctly, every time.

  • Instrumentation is not networking

    And a network engineer who has never stood in front of a safety system should be nowhere near one.

  • Vendor management is not engineering

    The standing exceptions exist because refusing one costs a shutdown, and nobody wants that conversation.

  • Turnarounds do not leave slack

    The team is fully committed for months either side, which is exactly when this work would have to happen.

  • They cannot audit themselves

    A finding is worth more from someone with nothing to lose by writing it down.

Value creation

What we do across an LNG hold, and where each one lands in the numbers a buyer checks.

  • Availability you can evidence

    The historian already collects it. Reading it properly supports the number in your model and survives the next diligence.

  • Turnaround scoped on data

    The largest controllable spend in the hold, argued from condition and history rather than from the last one.

  • OT security that survives diligence

    Segmented, access that expires, monitored. It is an insurance and regulatory question before it is ever an IT one.

  • Licence and estate rationalisation

    The historian licence you are already paying for, the Microsoft estate, and the tools bought once per project.

Neurotic AI Platform, on your terminal

The same product every portfolio company gets, carrying the screens a control room opens and nobody else does. Asked in the words used on the plant.

How the platform answers
Neurotic AI Platform LNG Example view

Ask Show me compressor train availability this week.

Train B ran derated from Tuesday. Suction pressure is the tell.

  • Compressor train A availability, week available
  • Compressor train B availability, week derated
  • Feed gas PI tag healthy
  • Boil-off recovery PI tag healthy

Train B suction pressure

Interface and buffer health

  • Collecting
  • Buffering

The change is that availability stops being something reconstructed a week later. A derate is a question the same morning, asked by whoever noticed, and the answer names the tags it came from so it can be checked rather than believed.

AI intelligent dashboards

We build these quickly, and to your business: dashboards that let you watch availability, derates and tag health across the terminal, and see a compressor train drifting before it trips — with the alerts and notifications that go with them, so a number that moves finds you rather than waiting to be found.

  • Maintenance reports

    What was done, what is due, and what has come back more than once.

  • Anomalies

    Readings drifting away from where they have historically sat, raised while there is still time to act.

  • Compressor trains

    Availability and performance by train, from the historian, the same day rather than the next week.

  • Tag health

    What is licensed, what is collecting, and what has quietly stopped.

  • OT access log

    Who reached the control network, from where, and whether the exception that let them still applies.

These are the five a terminal asks for first. The screens are built to the asset, not chosen from a menu, so a dashboard you need and cannot find anywhere on this list is the ordinary case rather than a special one.

Where to start

A two week audit. One document. No obligation.

Fixed scope, fixed price. It reads your systems and tells you what is wrong, what each fix costs, and what to do first. You keep the report either way.

Readiness assessment 2 weeks

What it turns up

  • Licences paid for and not used
  • Firewall rules nobody has reviewed since install
  • Administrator accounts with no owner
  • Reports built on a source that stopped updating
  • An integration failing quietly, nobody alerted

What it reads live

  • Licences, systems and what they cost waiting
  • Network, remote access and segmentation waiting
  • Cloud tenant, identity and admin rights waiting
  • Core systems and how they connect waiting
  • Reporting, data quality and access waiting

One document: what is wrong, what it costs to fix, and what to fix first.

Other industries