SteadfastProcess Simulation Book a demo

Steadfast Dynamic OTS – Cryogenic Gas Plant

Practise the upset before it happens.

A real-time dynamic simulator of a GSP turboexpander gas plant, on classic DCS screens. Your instructor trips the expander, fails a valve or cuts the refrigeration; your operators work it through, and nothing in the real plant is at risk. It starts as a generic plant and is set up with your site's own tags, tuning, alarms and design data.

Cold box 2 screen seconds after an expander trip: the expander shows TRIPPED, inlet guide vanes at 0 percent, V-101 pressure PI-120 at 1,004 psig in alarm and expander flow FI-131 at zero.
Cold box 2, five minutes in, seconds after an expander trip. Guide vanes shut, V-101 pressure up to 1,004 psig in alarm, the JT valve taking the gas. Every number on screen comes from the running model.
294malfunction scenarios an instructor can set, in four categories
20theoretical stages in the demethanizer, each solved four times every simulated second
43controllers and hand stations, with about 165 live indicators behind them
10components in a Peng–Robinson model, from nitrogen and CO2 to C6+

Why a simulator

A cryogenic gas plant leaves no room to learn on the job.

A turboexpander plant runs gas at pipeline pressure, chills it far below the temperature at which carbon steel turns brittle, and drives a machine at thousands of revolutions per minute, all with a flammable inventory. When it goes wrong, it goes wrong in minutes: the expander trips, pressure climbs toward the relief valves, the column loses its cold and product goes off specification.

The events that matter most are the ones operators see least. An expander trip, a refrigeration failure or a cold start may come once or twice a year, and a newer operator can go years without handling one before facing it alone, at night, with the board full of alarms.

And the people keep changing. Experienced operators are retiring, new hires get a few weeks of onboarding, and plants struggle to keep crews staffed. Refining has long trained operators on simulators of their own units; cryogenic gas processing largely has not.

"20 people were needed across four teams. I watched 31 people come and go through those positions in four years."

A gas plant operator, in Pipeline & Gas Journal, August 2026
1,000 psiginlet gas pressure, with relief valves to flare above it
−140 °Fand colder in the cold box and the top of the demethanizer
~100 °F/hthe cooling rate brazed aluminium exchangers tolerate before the joints crack
18,000 rpmexpander-compressor shaft, tripped at 110 % speed
  • Rehearse the rare eventTrip the expander or lose the refrigeration as often as it takes, until the response is automatic.
  • Learn the plant's cuesBuild pattern recognition on the real alarms, trends and interactions before taking the board.
  • Train every crew the same wayThe same scenarios for each new operator, whoever happens to be on shift to mentor them.
  • Practise start-ups safelyCool the plant down from ambient without risking an exchanger or the expander.
  • Keep what veterans knowTurn experienced operators' war stories into repeatable scenarios before they retire.
  • Train on your own plantYour tags, tuning, alarms and design, so nothing has to be translated back to the real board.

Malfunction library

Every failure your operators fear, on demand.

The instructor picks the fault and its size from the Instructor station. Trainees see only the effects: a failed control valve still shows its controller's output, so they have to diagnose it from flows, pressures and alarms, as they would in the plant.

294scenarios in the base GSP plant
Feed disturbances 4Valve failures 102Transmitter failures 165Equipment failures 23
4

Feed disturbances

  • Field gas supply, stepped or ramped at a set rate
  • Feed richness
  • Field gas CO₂ content
  • Inlet gas temperature

Set the size and rate of each change

102

Valve failures

  • Fails closed
  • Fails open
  • Stuck at any position, in %

34 valves × 3 failure modes: inlet compression and flare, amine, mole sieve regeneration, chiller, cold separator, expander guide vanes, JT, GSP reflux, reboilers, NGL product, compressor capacity and anti-surge, refrigeration, hot oil, stabilizer, sales

165

Transmitter failures

  • Fails high or fails low
  • Frozen at the present value
  • Bias, % of span
  • Drift, % of span per minute

33 controller measurements × 5 failure modes

23

Equipment failures

  • Compressor and pump trips: inlet K-401, refrigeration K-701, residue C-202, expander EX-101, lean amine, hot oil
  • Liquid slug at the slug catcher
  • Amine foaming, amine leak, lost cooler fans
  • Wet mole sieve bed on line, stuck switching valves, aged sieve, regeneration heater trip
  • Propane leak, condenser fans lost, hot day
  • Hot-oil heater trip, E-101 fouling, sales pipeline pressure rise

Capacity losses set in %, pressure rise in psi

Cold start from ambientCooldown of the whole plant, with exchanger cooldown-rate and thermal-shock alarms.
Snapshots and backtrackSave named starting points; step back a few minutes and rerun a scenario.
Session journalEvery operator and instructor action, with its simulated time.
Several at onceFaults stack, so a transmitter can fail in the middle of a trip.

Counts are for the base GSP configuration. Plant options change them: an upper side reboiler adds a valve and a controller; an ISS plant has no GSP reflux valve.

A trip, start to finish

The plant answers the way a real one does.

An expander trip, run in the simulator at 102 MMSCFD. Nothing below is scripted: it is what the model computes, second by second.

  1. The trip valve shuts and the guide vanes close. Expander flow FI-131 drops to zero and the shaft starts to coast down.

  2. Gas has nowhere to go. V-101 pressure rises past the JT setpoint and PIC-130 starts opening the JT valve.

  3. The inlet header backs up, so the wells deliver less. Plant feed falls to about 59 MMSCFD while the JT catches up.

  4. The JT carries the load. Feed recovers to about 98 MMSCFD, and the residue compressor takes over the booster's duty.

  5. The column settles warmer. The JT gives far less cooling than the expander, so ethane recovery drops and the side reboiler works back to its setpoint as tray compositions resettle.

Then change one setting.With the plant's own option "JT quick-open on expander trip" set at 90 %, the same trip holds feed at about 94 MMSCFD or more and raises no process alarms. Trainees see the difference a design choice makes.
Expander and booster trend group: V-101 pressure PIC-131 rises from 950 to about 1,010 psig and settles at 966; plant feed FIC-100 dips from 102 to about 60 MMSCFD and recovers to 98; booster flow FIC-215 dips and recovers.
Trends, Expander and booster group. One-second history: pressure up, feed down and back, booster flow following. Each instrument has its own chart, with setpoint, output and mode above it.
A guide-vane failure, 20 simulated minutes later
MeasurementBeforeAfter
JT valve flow to stage 5 FI-151047.4 MMSCFD
Ethane recovery85 %78 %
Residue compressor C-202 power4,100 hp4,690 hp
Column top temperature−132.6 °F−126.1 °F

From generic to your plant

Your tags. Your tuning. Your alarms.

The simulator ships as a generic GSP plant. Setting it up for a site follows the order a project team already works in, and every step shows your own tag names. Most of it is loading spreadsheets you already have.

  1. Tag names

    Download the CSV template, fill in your plant's tags in Excel and load it back. Equipment, controllers, indicators and valves take your names on every graphic, faceplate, alarm and trend. Indicators you do not have can be marked not installed and disappear from the screens.

  2. Tuning

    Load your controllers' gain, integral, derivative and action from CSV, or type them in. The plant tuning is applied again after every reset, rebuild and snapshot.

    KcTiTdAction
  3. Alarms

    Limits for every controller and indicator, from your DCS alarm list. A blank field means that alarm is not used. Deviation alarms wait after a setpoint change, as DCS PID blocks do.

    HIHIHILOLOLODEVHIHIDEVHIDEVLODEVLOLO
  4. Process design

    Feed composition, inlet compression, amine, dehydration, inlet supply, cold box and expander, guide-vane control, demethanizer, reboilers, residue compression, refrigeration and hot oil, and thermal response: about 60 design inputs on twelve tabs.

  5. Apply and rebuild

    The steady state is re-solved from your design data and the plant settles before it starts, in about 20 to 40 seconds. Every screen is redrawn to match the equipment you have.

  6. Save the plant file

    One file holds the whole plant: design, tag names, tuning, alarms and your named starting snapshots. Load it on any training computer. A checksum refuses a file that has been changed or damaged.

Configuration screen, Alarms tab: one row per instrument with HIHI, HI, LO, LOLO and four deviation alarm columns, buttons to load and download CSV.
Configuration, step 3: Alarms. One row per instrument, loaded from CSV or edited on screen, in force the moment it is saved.

Plant options built in

GSP or ISS process Side reboiler E-201, and an upper side reboiler Hot-oil or inlet-gas reboiler heating Rate-and-pressure or pressure-only expander control GSP reflux ratio station JT quick-open on expander trip Trays or packing Ethane recovery or rejection

Inside the model

Built by a process engineer, for operators.

Training only works if the plant responds the way the real one does. These are the details that make it behave.

A rigorous column

Every theoretical stage of the demethanizer carries its own liquid holdup, composition and energy balance, and solves its Peng–Robinson bubble point four times a second. Tray temperatures move with composition, so a disturbance travels through the column as it does in the plant.

A real control system

ISA-form PID controllers in velocity form, cascades, the HS-101 guide-vane selector, the JT valve as expander backup, and anti-surge on both compressors with setpoints that follow the surge line.

What operators actually see

Analyzer dead time and sample cycle, lines that drift to ambient when their flow stops, exchanger metal temperatures with cooldown-rate and thermal-shock alarms, relief valves that lift to flare.

Classic DCS screens

Overview, inlet compression and stabilizer, amine, dehydration, two cold box screens, demethanizer, residue, propane refrigeration and hot oil, drawn in the black-screen style operators know. Click any controller for its faceplate; click any alarm or trend tag to jump to it on its graphic.

Trends that answer questions

Eighteen trend groups in process order, one chart per instrument, four hours of one-second history, plus six custom groups. Hide what you do not need; open any group as a table.

Runs anywhere today

Today the simulator runs in a web browser, at real time or faster, for demonstrations. The product runs as a Windows desktop app in your training room, with nothing else to install.

Screens

What your trainees work with.

The product

A training room, not just a simulator.

The simulator you see here is the engine. The product puts it in your training room: a Windows application licensed to your site, an instructor console, and trainee stations on your own network.

In development: the desktop app, site licence and logins are built and in test; the networked instructor console follows. The engine and screens shown on this page run today.

Desktop appInstalls on Windows and runs offline. Plant files, snapshots and trend data are saved and opened like documents, where you choose.
Site licenceOne licence per site, renewed yearly, unlimited trainees. Your configured plant opens only at your site.
Instructor consoleOne screen shows every trainee's plant: alarms, key readings, recovery, trips and the last action taken. Open any trainee's screens live.
Trainee stationsEach trainee logs in with their own ID and runs their own plant. No instructor screens, no malfunction list, nothing to give the answer away.
ScenariosBuild multi-step exercises: a malfunction at a random time, a second one ten minutes later, a third when a reading crosses a value. Save them in a library and reuse them.
DebriefA timeline per trainee: when each fault came in, the alarms that followed, every action taken and how long the first correct action took.
Your plantYour tags, tuning, alarm limits and design data, set up in the simulator's own configuration screens and saved as your plant file.
Your manualThe operating manual issued with your plant's tag names, plus an instructor guide: the expected response to every malfunction, from the model.

How training runs

The instructor drives. The trainees work it out.

A training server in the room runs one live plant for each trainee. The instructor console and the trainee stations connect to it over your local network; nothing leaves the building.

Set upThe instructor picks a starting point (design rate, a snapshot or a cold plant) and a scenario, for one trainee or the whole class.
Hand overTrainees run their plants from the DCS screens. The faults fire on schedule or at random; nobody is told what they are.
WatchThe console shows each plant's alarms and readings, flags trips, and opens any trainee's screens live. The instructor can freeze or add a fault.
DebriefEach trainee's timeline is replayed against the expected response from the instructor guide: what they saw, what they did and how long it took.

Beyond gas processing

We build simulators for other process industries too.

The same approach carries to other units: rigorous thermodynamics, tray-by-tray column dynamics, real control schemes and classic DCS screens, configured to your plant's data. Tell us about your unit.

NGL fractionation and LNG

Deethanizers, depropanizers and debutanizers, refrigeration loops and liquefaction trains.

Refining

Distillation and stabilisation units, with the column dynamics that drive product quality.

Olefins and petrochemicals

Ethylene recovery trains, cold sections and splitters.

Chemicals and solvent recovery

Continuous and batch distillation, reflux and steam-rate control, column start-up.

Gas treating and dehydration

Amine and glycol units upstream of the cryogenic plant.

Your unit

Bring your design data and P&IDs; we model the process and the controls your operators actually run.

Book a demo

See your next trip before your operators do.

In 30 minutes I'll trip the expander, fail a valve and cut the refrigeration on the live simulator, then show how your own tags, tuning and alarms go in. Bring your toughest training scenario.

Rajiv Kalra

Chemical engineer: distillation modelling and controller tuning on operating columns, then 30 years delivering large systems projects.

To arrange a demo:
steadfastsim@gmail.com
+1 609-356-9711

About the founder →

The founder

Built by an engineer who has done this before.

Rajiv KalraFounder, Steadfast Process Simulation

Steadfast Process Simulation was founded in 2026 to bring real-time operator training to the process plants that need it.

Rajiv holds a master's in chemical engineering from the City University of New York, where he specialized in mathematical modeling of chemical processes and process control, and a bachelor's in chemical engineering from IIT Roorkee, India.

Process simulation is where his career began, and Steadfast is a return to it. The company plans real-time simulators for natural gas and LNG, refining, olefins and petrochemicals, and other process industries. The first product, Steadfast Dynamic OTS – Cryogenic Gas Plant, is built for natural gas processing.

  1. 1989–1991

    ABB Simcon

    Senior Process Lead

    Led development of real-time operator training simulators for large ethylene plants at Osaka Petrochemical, Mitsubishi Petrochemical and Maruzen Petrochemical, projects of $4–6 million each that are still in use at those sites in Japan. Also built a high-fidelity xylene isomerization model for Mobil in Singapore.

  2. 1992–1998

    FMC Corporation

    Senior Process Engineer, Technical Services

    Built mathematical models of plant processes and used them to stabilize complex batch distillation units and retune controllers that were causing frequent trips. At the Baltimore plant this raised output of a proprietary agricultural chemical from 600,000 to 2.4 million lb a year. FMC sites in the US and Canada, including Prince George and Bessemer City, called him in to stabilize their operations.

  3. 1998–2026

    ERP and supply chain

    28 years in life sciences, consumer products and the US federal government

    Led SAP and supply chain programs that bring ERP, planning and plant data together, so supply chain, manufacturing and finance work from the same numbers. Most recently at AbbVie in Chicago.