Remaining life
Every app has a Remaining life tab: how long until an indicator reaches its limit, as an approximate range with the limit and its source, never a single date.
Open an exampleLoad your own data. Drop a CSV or historian export and get a printable Health Report: Valves and loops · City water · Buildings and chillers. Needs the CAPMAC service (ask your CAPMAC contact to start it); your file stays on this PC.
455 of 466checks work today
12also passed a test on real plant data
14also passed a fresh test on simulated records
3apps: valves, water, buildings
Real data from real plants
REALReal data from Yorkshire WaterData Mill North, CC BYRepairs confirmed by a night-flow drop: 0.30 vs 0.03 for placebo dates.REALReal data from City of TopekaGIS, free use with source creditTop 10 % of pipes hold 61.7 % of 478 breaks (latest test), vs 52.3 % by material.REALReal data from City of TorontoOpen Government LicenceTop 10 % of pipes hold 47.3 % of breaks vs 31.8 % by material.REALReal data from Figshare 27948192 (VAV building)Figshare 27948192, CC BY18 of 18 stuck dampers found, 1 false alarm in 234 box-days.
The engine
Live checks
Every reading goes through checks that each passed a fair test before they may raise an alarm.
OpenPhysics twins
Plain physics picture of pumps, chillers and valves; the confidence tier is shown on every verdict.
OpenLearning layer
Per-unit healthy levels worked out from your own history, no black box.
OpenConnect
Read-only link to your plant: OPC UA or a CSV upload.
OpenYour equipment
Valves and loops
Control valve
We watch: Sticking, jams, leaking seats, plugged trim, wear.
22checks working
REAL
Real Bauer/SACAC loops: all 3 sticking loops flagged. Small sample; trial on your own records recommended. No licence stated: shown one-to-one only; do not copy or forward.
Open in the appPositioner and air supply
We watch: Air supply loss, miscalibrated travel, diaphragm leaks.
6checks working
OWN SIMULATED
Passed on our own simulated valves; no real or outside-simulator pass yet.
Open in the appPID control loop
We watch: Oscillation, windup, saturation, manual mode, sluggish response, wrong set point.
24checks working
OWN SIMULATED
Sluggish-loop and hidden stick-slip checks passed fresh own-simulated tests; no real-data pass yet.
Open in the appWater
District meter zone
We watch: Night flow, bursts, meter drift, theft, repairs confirmed.
48checks working
REAL
Real data from Yorkshire Water: repairs confirmed 0.30 vs placebo 0.03 on one test, 0.30 vs 0.05 on the next. Alerts are tuned for a low false-alarm rate; trial on your own history recommended.
Open in the appPipe network
We watch: Which mains to renew first (the order is tested; a chance of breaking is not), surge, closed valves, pressure-reducing valves.
21checks working
REAL
Real data from City of Topeka: top 10 % of pipes hold 61.7 % of 478 breaks vs 52.3 % by material alone (latest test).
Open in the appPump station
We watch: Vibration, bearings, curve, cavitation, short cycling.
54checks working
OWN SIMULATED
Passed our own simulated tests and held-out tests on simulated records; no real-plant pass yet.
Open in the appReservoir and tank
We watch: Level balance, overflow, leakage, chlorine decay.
4checks working
OWN SIMULATED
Tank and chlorine checks pass on our own data.
Open in the appBuildings and chillers
Air handler
We watch: Dampers, valves, filters, mixed air, stuck or spiking sensors.
8checks working
REAL
Narrow pass on real Korean office air handlers: injected sensor bias found, nothing false (9 of 19 stretches judged).
Open in the appVAV box
We watch: Stuck damper and airflow bias against the box's own normal level.
2new checks(new, tested on real data, not yet in the main count)
REAL
Real evening set (Figshare 27948192): 18 of 18 stuck dampers, 7 of 8 airflow bias, 1 false alarm in 234.
Open in the appFan coil
We watch: Leaking valves, filters, dampers, fan loss, zone sensor bias.
9checks working
OUTSIDE SIMULATOR
LBNL fan-coil open files: 4 of 4 faults, 0 false calls (easy by construction).
Open in the appChiller
We watch: Fouling, undercharge, flow loss, staging, sensor bias.
26checks working
REAL OUTSIDE SIMULATOR
Real data from the FrostByte rig: all severe fault groups named right on rig data it had not seen, no healthy period alarmed. FrostByte (TH Wildau), CC BY-NC 4.0: shown one-to-one only; do not copy or forward. Outside simulator (LBNL plant): 9 of 10 aimed fault pairs found in each of two later tests, no false calls.
Open in the appCooling tower
We watch: Approach, fan, blowdown, fill fouling, hunting loop.
14checks working
OWN SIMULATED
Passed our own simulated tests and held-out tests on simulated records.
Open in the appData-centre and clean room
We watch: Filters, coils, CRAH fans, HEPA, door, pressure.
20checks working
OWN SIMULATED
Passed our own simulated tests.
Open in the appData centre cooling
We watch: CRAH / CRAC filters, coils, fans, valves, air paths, supply air, free-cooling exchanger; a PUE-style energy view.
11checks working (8 also passed a held-out test)
HELD-OUT TEST, SIMULATED
Real data: none yet from a data centre. Passed held-out tests on simulated records.
Open Data centresBuilding automation
We watch: BMS points by AHU, VAV, fan-coil and chiller; schedule and override audit; simultaneous heating and cooling; set-point drift; trend-log quality.
7check groups
REAL
Real data from a VAV building and office air handlers. LBNL fan-coil lab files are an outside-simulator result, not real plant data. Read-only BACnet/IP connector.
Open Building automation3D, savings and documents
3D walkthrough
Walk through a building, a data centre, a valve plant and a city water network with live check results on the equipment.
Building Data centre Valves WaterSavings calculator
Put in your own prices and sizes; every figure is marked as an assumption you can change.
Open the calculatorDocuments
How your data is read, and how healthy levels are worked out, are explained on the Connect and Learn pages.
Connect Learn