Key Points
- Start with rotating assets that have no installed spare, such as the agitator gearbox on a single-train reactor.
- Wireless sensors rated for Class I Div 1 areas avoid new conduit runs and the permits that come with them.
- Trend data on each monitored asset lets planners move repairs into the next turnaround instead of a mid-run shutdown.
Condition monitoring for chemical plants means putting continuous sensors on the rotating equipment that keeps a unit running, then reading the trends early enough to plan the repair. In most specialty and commodity chemical sites, the list starts with reactor agitators and their gearboxes, centrifugal process pumps, process gas and plant air compressors, and cooling tower fans.
This guide is for the reliability engineer or maintenance manager who has to decide which of those assets get sensors first, how to install them in classified areas, and how the data feeds turnaround planning. It includes an asset table with failure modes and signals, and a step-by-step walkthrough of a failing agitator gearbox.
What Does Condition Monitoring Cover in a Chemical Plant?
In a chemical plant, condition monitoring covers the machines whose health can be measured while they run: motors, gearboxes, pumps, compressors, fans, blowers and dryers. Sensors track vibration, ultrasound, temperature, speed and electrical behavior, and software compares each reading with that machine's normal baseline.
Static equipment such as vessels, piping and heat exchanger shells sits under a different program, built on thickness gauging, corrosion coupons and inspection intervals. The two programs meet in the mechanical integrity file, where a pump bearing trend and a nozzle thickness reading both show the state of the same process unit.
The gap in most chemical sites is between vibration routes. A technician with a handheld collector may reach a pump in a classified area once a month, while a seal flush problem or a lubrication fault can develop in days.
Which Chemical Plant Assets Should You Monitor First?
Monitor first the assets whose failure stops a unit or releases product. Four questions sort the list quickly, and an asset that answers yes to two or more belongs in the first wave.
- No installed spare: a single agitator drive, a single-train charge pump or one cooling tower cell that limits summer rates.
- Batch at risk: an agitator that stops mid-reaction can leave a batch off spec or force a hold that ties up the reactor for days.
- Covered process: equipment inside a unit covered by OSHA Process Safety Management, where mechanical integrity records already apply.
- Seal-leak consequence: pumps and agitators handling flammable, toxic or high-temperature fluids, where a failed mechanical seal is an environmental or safety event before it is a maintenance job.
Plant air compressors deserve a place on the list even though they sit outside the process, and compressor monitoring on them reads the same signals as on process gas machines. When instrument air pressure drops, air-operated control valves move to their fail position and the units they serve go with them.
Chemical Plant Assets, Failure Modes and Signals
Each asset type in a chemical plant fails in a small set of predictable ways, and each failure mode shows up first in a specific signal.
| Asset | Common failure modes (general industry examples) | Signals that show them first | Typical area classification need |
|---|---|---|---|
| Reactor agitator and gearbox | Gear tooth wear, bearing wear from overhung shaft loads, lubrication breakdown, mechanical seal leaks, shaft misalignment | Ultrasound for friction, vibration at gear mesh and bearing frequencies, housing temperature, motor current | Often Class I Div 1 or Div 2 around the reactor top and seal |
| Centrifugal process pump | Cavitation, bearing wear, mechanical seal failure, misalignment, running far from its best efficiency point (BEP) | Broadband vibration and vane pass frequency, ultrasound for cavitation and lubrication, bearing housing temperature | Div 1 or Div 2 when the pump handles flammable liquids |
| Reciprocating and screw compressors | Valve and rider band wear on reciprocating units, rotor and bearing wear on screw units, looseness, lubrication faults | Vibration at running speed and harmonics, ultrasound, temperature, RPM | Classified in process gas service; usually unclassified for plant air |
| Cooling tower fan and gearbox | Gear wear, unbalance from fouled or damaged blades, bearing wear, drive shaft misalignment, structural looseness | Vibration at fan speed and gear mesh, gearbox temperature, RPM | Usually unclassified; wet, outdoor and hard to reach |
| Blowers | Unbalance, bearing wear, belt wear, looseness | Vibration at running speed, ultrasound, bearing temperature | Depends on the gas or vapor being moved |
| Rotary and fluid bed dryers | Support roller and trunnion bearing wear, drive gear wear, fan unbalance | Vibration, temperature, RPM | May involve combustible dust or solvent vapor classification |
Area classification in the last column varies by site and is set by your area classification drawings, not by the asset type. Use the column to flag which assets will need a hazardous-location sensor before you plan the install.
Learn more about pump monitoring.
Failure Walkthrough: A Reactor Agitator Gearbox
A reactor agitator gearbox usually announces trouble in a fixed order: lubrication first, then gear and bearing vibration, then heat and load. The sequence below is an illustrative example of how that plays out on a top-mounted agitator with a right-angle gearbox, and what the maintenance team does at each step.
- Ultrasound rises on the output bearing. Friction from thinning oil film raises the ultrasound level well before vibration changes. The planner checks the lubrication record and the technician confirms oil level and condition at the next safe access.
- Gear mesh energy grows in the vibration spectrum. Sidebands appear around the gear mesh frequency, spaced at the speed of the worn shaft. The reliability engineer compares the trend with the batch schedule, because load changes between recipes can mask or exaggerate the pattern.
- Housing temperature climbs under full load. Temperature follows once friction and wear generate enough heat to read through the casing. The team orders the gear set and bearings, books crane time and writes the job for the next scheduled reactor cleanout.
- Motor current shifts. A gearbox dragging under wear asks more of the drive motor, which shows in current readings from the motor circuit. By this point the repair is planned, and the reactor runs at reduced speed or finishes its campaign under close watch.
- Planned swap and verification. The gearbox comes out during the cleanout instead of in the middle of a batch. After restart, the same sensor confirms that ultrasound, vibration and temperature returned to baseline.
The value is in steps one and two. A route reading taken monthly could land after step three, which turns a planned repair into an emergency crane lift with product still in the vessel.
Monitoring in Classified Areas
Monitoring in classified areas works when the sensor carries a hazardous-location rating that matches the area classification and installs without hot work or new conduit. Wireless, battery-powered sensors meet the second point by design, because they need no power cable or signal wiring to the control room.
A wired vibration system in a Class I Div 1 area needs conduit with seals, junction boxes rated for the area, and often a hot work permit to fasten brackets or run new cable tray. On a pump row inside an operating unit, that work can wait months for a turnaround window.
Tractian Smart Trac Ultra sensors are Class I Div 1 ATEX/IECEx/NFPA compliant, with an IP69K sealed housing rated from -40°F to 250°F. Each sensor bonds with adhesive or a drill-and-tap stud and is up and running in minutes. Check the marking on any device against the gas group and temperature class on your classification drawing before it goes into the unit.
Learn more about hazardous location wireless condition monitoring.
Combining Mechanical and Electrical Signals
Combining mechanical and electrical signals shows whether a fault starts in the machine or in the motor that drives it. A Smart Trac Ultra sensor on the asset reads vibration, ultrasound, temperature and RPM, and Energy Trac installs in the electrical panel that feeds the motor to track current, voltage and power factor.
On a process pump, that pairing separates cases that look alike on a single sensor. Rising vibration with steady current points toward a bearing or alignment issue, while voltage imbalance or signs of rotor, stator, winding and insulation faults point the investigation toward the motor.
Together the two layers make up Tractian multimodal monitoring. Tractian's industrial AI, Janus, correlates the sensor signals with operational data, maintenance history and technical context to narrow down likely causes. Some machines need a single layer, and Tractian recommends the mix for each machine based on its failure modes, criticality and operating context.
How to Roll Out Condition Monitoring Across a Chemical Site
Roll out condition monitoring one process unit at a time, starting with the unit whose assets scored highest on criticality. A focused first unit gives the team enough alerts to build a response routine before sensors spread across the site.
- Rank the assets. Score each machine on spare availability, batch impact, PSM coverage and seal-leak consequence, then pick the highest-scoring machines in one unit.
- Map the classification. Mark which of those assets sit in Div 1, Div 2 or unclassified areas, and confirm sensor ratings and mounting methods with the unit's safety lead.
- Decide who answers alerts. Route alerts into the CMMS the plant already runs and name an owner for each severity level, so a gearbox warning on night shift reaches someone who can act on it.
- Feed the turnaround list. Before each turnaround or reactor cleanout, pull the assets with active faults into the scope and drop the overhauls that trend data shows are not due.
- Review after the first quarter. Compare caught faults, missed faults and false alarms, then extend coverage to the next unit using the same criticality scoring.
The turnaround step is where condition data pays back in a chemical plant. Learn more about maintenance planning around PSM requirements and turnarounds.
How Tractian Supports Chemical Plants
Tractian supports chemical plants with wireless sensors for rotating equipment, AI diagnostics that say what is wrong and what to do next, and connections to the maintenance and business systems a site already uses. Tractian focuses on condition monitoring and asset health rather than replacing the plant's CMMS.
- Fast to deploy: sensors install in a day with no cabling and no dependency on plant WiFi, sending data over 4G/LTE, and teams can expect full visibility within the first 14 days as the models learn each machine.
- Long service between visits: the battery lasts up to 5 years, which matters for assets inside a unit where access needs a permit.
- Human review when it counts: on high-impact alerts, supervised analysis adds expert-vetted insights to the AI diagnosis.
- Open to the plant's systems: Tractian integrates with most CMMS, ERP, SCADA and APM systems via API, including MaintainX, Fiix, UpKeep and Limble on the CMMS side and SAP and Oracle NetSuite on the ERP side.
- Reports built to the site's needs: Power BI and SQL connectors support custom reporting on downtime, failure trends and costs.
At its San Diego facility, CP Kelco faced a familiar agitator risk: a failure on its fermenter agitators could stop production for 3 to 4 days and require cranes and manual gearbox replacements. Tractian now monitors 280+ critical assets at the site. Within months, the reliability team counted $446.9K in total cost savings, $380K of it in production uptime, plus 84 hours of avoided downtime, including a serious unbalance on a dryer bearing.
Frequently Asked Questions
What is condition monitoring in a chemical plant?
It is the continuous measurement of machine health on rotating equipment such as agitators, process pumps, compressors and cooling tower fans while the unit keeps running. Each reading is compared with that machine's own baseline, so a developing fault shows up with time left to plan the repair.
Can wireless vibration sensors go into Class I Div 1 areas?
Yes, when the device carries a hazardous-location rating that matches the gas group and temperature class of the area. Tractian sensors are Class I Div 1 ATEX/IECEx/NFPA compliant. Because they run on batteries and transmit by radio, they avoid new conduit and signal wiring into the classified area.
Does condition monitoring help with PSM mechanical integrity?
It can support it. The OSHA Process Safety Management standard calls for documented inspection and testing of covered process equipment, and time-stamped trends on pumps, compressors and agitators add evidence of equipment condition between inspections. The plant's own mechanical integrity procedures still define what the program requires.
Which signals catch agitator gearbox problems early?
Ultrasound tends to move first, as friction from a thinning oil film rises. Vibration around the gear mesh frequency follows as the teeth wear, then housing temperature and motor current change. Watching those signals side by side gives the team the longest lead time to schedule the gearbox job.
How long does it take to get condition monitoring running in a chemical plant?
Tractian sensors mount with adhesive or a threaded stud and are up and running in minutes. The rest is a learning period, and teams should plan for full visibility within the first 14 days while the models learn how each machine normally runs. Permits for unit access often set the pace more than the hardware does.
See What Tractian Would Monitor in Your Unit
Bring the equipment list for one process unit, its area classification drawing and the date of the next turnaround. A Tractian specialist will walk through which agitators, pumps, compressors and fans to cover first, how the sensors mount in your classified areas and how alerts reach your CMMS. Explore Tractian condition monitoring for chemical plant assets.

