CAPMACApp 1· Valves and control loops
REPLAY OF RECORDED RUNS
IST UTC+5:30 · replay day —
REAL DATA: Bauer/SACAC, ISDB, DAMADICS loops

Is my control valve healthy, and is my loop doing its job?

CAPMAC watches the ordinary records of valves, positioners and control loops and raises a plain-words alarm when a check that has passed a fair test sees a problem. OpenLive dashboardto replay a recorded run,Case studiesfor the tests behind it,Checksfor every working check.

At a glance

What we watch

Control valve
Sticking, play, jam, sizing, tight spots. Needs the loop output and, where there is one, the valve position.
Smart positioner
Sticking hidden behind the positioner, seen from the actuator air pressure.
PID loop
Sluggish, hunting, stuck at a limit, left in manual, outside oscillation, controller too hot.
Level loop
Valve or sensor trouble on tank level loops.

Overview

Who sees what (display only)

Operator
Alarm list, priorities, Ack and shelve, plain recommendation. Evidence and money hidden.
Engineer
Everything an operator sees plus the evidence card, readings, trends and the checks behind each alarm.
Manager
Counts, money at stake (assumed prices), the report and the CSV. Pick the role at the top right; it only changes what is shown.

Site settings

Time zone: India Standard Time (UTC+5:30). Reading units: percent of range for output and position; the indicator is a ratio with no unit. Money in Indian rupees (Rs). Replay time is in days of the recorded run.

Units

Asset tree of the recorded plant: site, area, loop, then its valve and positioner. Click a loop for its card.

Pick a loop.

Live dashboard

Alarms appear on the day the check raised them in the recorded run. Nothing here is live plant data.

Alarm list

PriDayLoopAlarmSureState

Trend against the normal band

Each dot is the recorded reading of one loop (valve behaviour, log scale). Shaded band = spread of the quiet loops in this recording. The recorded files hold one value per loop, not raw time series.

Evidence card

Click an alarm.

Recommendation

Click an alarm.

Audit log of acks and shelving

Replay dayWhoActionAlarm

Case studies

Stories show only what the checks got right.

Real data first

Rigs: our own simulators, outside simulators, held-out tests

Checks

Every check marked WORKING in our latest check record (of; one more is still being fixed). A check is WORKING after a fair pass on simulated data. The badge shows the best evidence beyond that.

#CheckEvidenceNote

REAL = also passed a fresh real-data test. HELD-OUT = also passed a held-out test run by our test team, answers kept there. OWN SIMULATED = fair pass on our own simulator. Where there is no real or held-out pass yet, the badge shows the best evidence we have.

Money

All prices below are ASSUMED Indian defaults, not your plant's. Edit them. Figures are an estimate of energy lost while a faulty valve or loop runs on; they are not a prediction of failure or remaining life.
LoopAlarmRs per 30 daysBasis

Reports

One A4 condition-monitoring report of the alarms open at the current replay day, and a CSV of alarms and the audit log.

Connect your plant

Read-only: OPC UA read, or upload a CSV of loop records. The page here is a recorded replay; nothing leaves your computer.

What a loop CSV needs: time stamp, set point, process value, controller output; for position checks the valve position; for the pressure route the actuator air pressure. See the suite pageconnect.htmlwhen published.

Learn this site

Some checks need a few weeks of the loop's own records to learn what normal looks like (for example 28 days for the torque-profile check). This recorded replay does not learn; the learning layer is described inlearn.htmlof the suite when published.

How it works

  1. Each loop's ordinary records go through checks. A check is built from the physics of a valve or a loop, or from a plain rule.
  2. An alarm is raised only by a check that passed a fair test (see Checks). Each alarm carries which check, which readings, and how sure.
  3. Priority: P1 act now, P2 plan this month, P3 note. Colour is used only for abnormal; amber and red are alarms, purple means not judged.
  4. You Ack an alarm to say you have seen it, or shelve it to hide it for a while. Both go in the audit log.
  5. Alarms report what the readings show now. The Remaining life tab gives an estimate as a range, never a single date.

About

App 1 of the CAPMAC demo suite. Evidence source: our latest check record (2026-10-21) and the real-data evidence table (2026-10-21). Replay data: committed result files our recorded replay files.

Real-data sources: Bauer/SACAC loop repository (no licence stated; one-to-one use), ISDB industrial stiction database (not cleared for commercial use; no figures shown), DAMADICS actuator benchmark (Lublin sugar factory).

Still being proved: sticking on level loops with smaller than 1 percent travel, and non-cycling small valve faults. Remaining-life ranges are on the Remaining life tab.

Load your data

Add your readings and get a printable report. It takes three easy steps.

Looking for the file service...

1Add your readings file

Drop your file here, or choose a file
A spreadsheet saved as CSV. One column for the time, and one column for each sensor.
Download the sample file

2Add your repair or cleaning dates (optional)

This helps CAPMAC learn from your plant and sharpen remaining-life estimates. You can skip it.

    3Get your report

    Add your readings file first.

    Advanced options

    Most people do not need this. CAPMAC picks the columns, units and checks for you. Change them here only if something is wrong.

    What is in the file

    Add a file to see this.

    Which column is what

    Add a file to see this.

    Which checks to run

    Add a file to see this.

    Report wording

    Remaining life: limits and dates (optional)

    A limit is the reading at which you would act. Leave it empty and CAPMAC uses a labelled assumed limit. Dates help it count from the last repair or cleaning.

    Remaining life: valves

    Faults this app caught, by equipment

    Rows with a Replay button come from a recorded run with a named planted fault. Rows without one are checks that caught every planted fault in their scored tests.