SCADA-informed emissions & operational intelligence
Detection, cause, impact and confidence on every event — and what to fix first.
AroIQ turns the SCADA data operators already collect into operational and emission events — each with a measured impact (volume, downtime or cost) and a likely cause, ranked by what they cost. Nothing to install.
- No new field hardware
- SOC 2 Type II
Compressor blowdown
Station 07 · Segment B · 38 min
490
MCF · 95%
390–620
245
tCO₂e · AR6
0.94
COD · detection
What we think happened
Planned depressurization for maintenance isolation on compressor K-201.
Cause notes · ranked
- 1 · Planned shutdown sequence Strong · 0.86
- 2 · ESD valve XV-1180 discharge excursion Fair · 0.41
- 3 · Instrument-air loss on isolation Weak · 0.22
Pattern
14 similar events
Cost
$1,240 per event
Triage
#03 Watch
Illustrative event record
What AroIQ reads
One SCADA feed. Two kinds of event.
The pressures, temperatures, flows and valve states you already record carry both stories: what left the site, and what the equipment was doing when it did.
Emission events
- Compressor blowdowns
- ESDs vs planned blowdowns
- PSV / PRV releases
- Tank venting & PVRV lifts
- Open thief hatches
- Flaring volumes & anomalies
- Unlit flare periods
- Segment depressurizations
- Engine methane slip
- Pneumatic device events
Operational events · the same data
- Compressor trips
- Stuck or hung dump valves
- VRU back-pressure & trips
- Setpoint & operating changes
- Compressor health drift
- Pump drift
- Process upsets
- O₂ / H₂S excursions
Patterns, not piles of events
Find it once. Find it everywhere.
Most events aren't one-offs. AroIQ groups every event into the recurring pattern behind it — the same fingerprint, the same mechanism. Validate the cause once and the whole pattern carries it: back through years of history, onto live data, and to your sister sites.
Your team reviews a handful of events. AroIQ explains the rest.
Dump valve hung open
Separator · Battery 3 · 3-sensor fingerprint
Cause validated once, by your engineer
- Equipment · LV-310 dump valve
- Condition · liquid slug
- Cause · worn valve trim
1
event reviewed
74
events explained
4
sites
-
01Across your history
24 months
Every past occurrence found and labelled: the ones in your records, and the ones that never made it there.
-
02On live data
from here on
The next time the fingerprint shows up, it arrives already explained — with its cause and its cost.
New match · Battery 3 · 04:12 worn trim · known -
03At your sister sites
same design
Your best-instrumented sites set the pattern — a cause validated where there's a sensor carries to sister sites without one.
- Battery 3 41 where it was validated
- Battery 7 18 sister site
- Pad 12 9 sister site
- Battery 11 6 no tank pressure sensor
Inside AroIQ
A workflow for every team that touches the event.
Start with what costs the most. Open the evidence behind it. Make the numbers your own. Three zones over one record — so operations, reliability and emissions debate the fix, not the data.
Where to act, across every site.
Operations, reliability and emissions open the same four pages. Each answers the question the last one raised.
- 01 Where to look first
Triage
Every site, one ranked list. Patterns sorted by cost, each with a likely cause and a status — act now, watch, or normal operation.
- 02 What keeps happening
Patterns
Events that share a cause, grouped and costed as one thing to fix — one work order, not a stack of alarms.
- 03 What changed, and did it work
Sites
One facility's operating history. Every setpoint change, reroute and repair on the same timeline as the emissions that followed.
- 04 What to fund
Investment Plan
What each cause costs to fix and what fixing it returns. Replace or observe — sized with your own numbers.
The evidence behind every call — and your say in it.
See exactly why AroIQ thinks what it thinks. Then validate it with what only your people know.
-
01
Events Investigation
The SCADA traces as the model saw them, the cause notes, the narrative and the calculation audit — one record per event.
-
02
RCA Workbench
A short review queue — lowest-confidence and highest-impact first. Confirm, override or flag; we do the rest of the labelling.
-
03
Validate once. Reuse everywhere.
Validate what an event really was, in your own words. AroIQ applies that answer to every event with the same fingerprint — across your history, on live data and at your sister sites. Your crew's know-how stays with the site, even when people move on.
Your costs. Your methodology.
- Costs & KPIs
- Your own cost per event type, per hour, per tonne — and count-based KPIs where there's no carbon price.
- Emission Parameters
- Gas compositions, GWPs, combustion efficiencies and site overrides — versioned, with every recalculation tracked.
- Datasets
- Historian exports, live feeds and gas analyses in one place — ingested, staged and versioned.
Follow one pattern
Illustrative
-
1 Triage
Dump valve hung open tops the list — Battery 3, 41 events, $86k a year.
-
2 Events Investigation
LV-310 hangs open after each liquid slug.
-
3 RCA Workbench
An engineer confirms worn trim; the other 40 events update.
-
4 Investment Plan
A $4,800 trim job against $86k a year.
-
5 Sites
After the repair, the recurring spikes stop.
Enterprise
- SSO / SAML Azure AD / Entra ID
- Org-scoped tenancy Row-level security in the app
- Public status page status.arolytics.com (opens in a new tab)
How AroIQ thinks
Before we detect anything, the data has to be trustworthy.
Six steps, from raw tag to ranked decision. Ingest and clean come first, because drift, gaps and inconsistent tag naming are the normal condition of SCADA data — not the exception.
- 01
Ingest
“Send everything.” Every tag, at the finest interval you have — historian exports or a live historian feed (PI, AVEVA and others). No tag list or tag templates needed to start.
- 02
Clean
Self-tested: the model hides readings it already knows, fills them back in and scores itself — thousands of times — before it trusts a gap fill. Telemetry codes stripped, drift corrected. SCADA writes on change, and the model knows quiet from missing.
- 03
Fingerprint
Learns each site's normal, then the shapes that break from it. Any recurring pattern in any sensor becomes a signature it finds again. Tag names don't matter; behaviour does.
- 04
Match causation
A person confirms the first cause — our analyst or your engineer — and the model applies it to every event with the same fingerprint. Equipment → operating condition → underlying cause.
- 05
Quantify
Volumes with a 95% range — your methodology or ours, GHG and VOC separate, every assumption recorded.
- 06
Triage
Events grouped into patterns, costed against your KPIs and ranked: what to act on, what can wait. Into AroIQ, or into your own stack.
Illustrative · Dots are raw readings; the line is the repaired signal. Temperature can't be 100 °C if the equipment is off — the model knows that.
- Months of patchy telemetry. The model fills the true gaps and keeps tracking the signal's behaviour through them.
- Multi-day idle gaps. On/off state identified from neighbouring sensors — equipment can't be running when the sensors around it say it's down.
- The real blowdown at 15:13 is kept. The spike at 16:48 is a calibration glitch — the speed channel never moved, so it's filtered.
Illustrative · cleaning on sparse SCADA
Two detection paths
per source, per site
Yes · direct
Read the sensor.
Compare it against its setpoint. The crossing is the event.
No · pattern-based
Infer it from everything else.
No sensor on the source. The sensors around it tell the story.
Two paths in. One record out.
Learns across sites
illustrative
- Station 01 first of its designonboarding
- Station 02 same designonboarding
- Station 03 same designonboarding
Every confirmed cause makes the next site faster.
Bring a facility export.
Leave with ranked causes.
The fastest way to evaluate AroIQ is on your own data — 12–24 months of SCADA history from a handful of facilities.