A pump’s pressure readings, a chiller’s operating temperatures, and a motor’s existing vibration measurements can all help a maintenance team decide what needs attention. The first question is which readings reveal a developing problem, and how to turn them into a timely inspection.

Building a predictive maintenance program starts with that connection between equipment condition and maintenance decisions. Where useful measurements and accessible interfaces already exist, you can begin without buying new hardware. This guide explains how to reuse PLC, building management system (BMS), and energy meter data, connect different protocols, and build a practical monitoring and response workflow.

What Do You Need for a Predictive Maintenance Program?

Start with one important asset and a condition you can measure. A useful pilot needs:

• An equipment priority: A pump, compressor, motor, or cooling system whose deterioration would disrupt operations.
• Relevant readings: Existing temperatures, pressures, vibration measurements, electrical measurements, or operating status, with access to their data sources.
• A maintenance decision: The condition that warrants investigation, who investigates, and what they check.
• A connection and monitoring platform: A route from the equipment’s interface to dashboards, rules, and alarms.
• A pilot owner: Someone responsible for checking data quality, testing notifications, and recording inspection results.

Condition monitoring provides the foundation: observe measured behavior, identify meaningful changes, and use the evidence to plan maintenance. Thresholds and trends help prioritize investigation; inspection findings then improve the program. This approach complements scheduled preventive maintenance and can grow as your equipment history develops.

Step 1: Condition Monitoring Using Existing PLC and BMS Data

List the measurements already available for each priority asset. Record the source, point name, units, update interval, timestamp, and communication protocol. Ask the controls team which values can be exported through MQTT, Modbus, BACnet, or OPC UA.

Match each point to a useful question. Pump discharge pressure can help identify a change in operating performance. Compressor temperatures can support investigations into abnormal operation. Energy meter data can highlight consumption changes when compared with similar production loads or operating hours.

Keep the operating context alongside the measurement. A stopped pump and a running pump need different interpretations of low pressure. A motor’s load and speed matter when interpreting its vibration. Check that the reporting interval captures the behavior you want to observe.

Collect a baseline across representative operating cycles, including startup, normal production, and shutdown. For a cooling system, compare readings under similar load and ambient conditions. Keep the maintenance team’s observations with that history. These notes help explain whether a change reflects expected operation, a measurement problem, or equipment that deserves closer inspection.

Pilot deliverable: A short asset-to-data list showing which maintenance questions existing readings can answer and where an additional measurement would add value.

Step 2: Predictive Maintenance Thresholds and Failure Modes

Define the failure mode before choosing an alarm threshold. For a pump, the concern might be deteriorating performance; for a motor, a change in vibration; for refrigeration equipment, an operating temperature that needs investigation.

Build warning criteria from manufacturer guidance, the machine’s normal operating history, and applicable engineering standards. Record the measurement location, units, operating state, and required response alongside each limit. This gives the technician a reason for the alarm and a starting point for inspection.

ISO 17359 provides general guidance for machinery condition monitoring. The relevant part of ISO 20816 can inform vibration evaluation when its machine type and measurement conditions match your application. ASHRAE thermal guidance can inform monitoring of applicable IT equipment’s inlet air; compressor discharge temperature needs limits appropriate to that compressor.

Set the warning point to leave a useful inspection window. Use separate activation and clearing criteria where appropriate, and choose a persistence period that filters normal fluctuations while preserving enough time to respond.

Step 3: Industrial IoT Gateways for Modbus, BACnet, OPC UA, and MQTT

The Kilo IoT Platform brings existing equipment feeds and wireless sensors into a shared environment for monitoring, rules, and alarms. Its MQTT connector provides a practical destination for suitably configured PLC, BMS, and meter data.

Choose the connection path that fits the installation:

• Existing MQTT feed: Configure the connection, topics, credentials, and payload mapping so measurements arrive with the correct identities and units.
• Modbus, BACnet, or OPC UA equipment: Use a compatible industrial IoT gateway or software bridge to read the required points and publish them over MQTT. An existing edge computer may already provide the hardware needed to run that bridge.
• Equipment needing a wireless connection: A suitable Modbus-to-LoRaWAN or Modbus-to-mioty adapter can transmit selected equipment readings through the corresponding gateway or base station into the platform.

Bridges also help extend the BMS itself. French manufacturer Nexelec’s MASTER CORE converts LoRaWAN device data into BACnet/IP points for building controllers and BMS software. It illustrates how wireless measurements can enrich existing building automation. A separately configured BACnet-to-MQTT bridge can then publish selected BMS points to Kilo, linking the new measurements with the wider monitoring program.

Document the direction of data flow and the points being exchanged. Configure encrypted MQTT connections, appropriate credentials, and authorized network access. Keep existing building controls in place while adding the measurements and operational visibility your maintenance team needs.

Step 4: Wireless Condition Monitoring with LoRaWAN and mioty Sensors

Add a sensor when a useful maintenance question needs a measurement the asset does not already provide. Examples include vibration and surface temperature on a motor, pressure on a pump line, or a leak detector beneath refrigeration equipment.

LoRaWAN and mioty offer wireless paths for suitable condition-monitoring measurements across buildings and industrial sites. Kilo Cloud includes managed LoRaWAN network-server functionality and an integrated mioty service center, bringing these additions into the same environment as MQTT equipment feeds.

Choose sensors around the actual measurement: temperature range, mounting, accuracy, vibration characteristics, environmental protection, and reporting frequency. For vibration monitoring, select a device whose sampling and onboard measurements suit the machine and the diagnostic question. Check radio coverage at the installed position and gateway connectivity to the platform.

LoRaWAN, mioty, and other IoT sensors and gateways can be sourced through Kilo Electronics. Match the hardware to the required interface and deployment conditions, including battery performance in cold rooms or outdoor locations.

Step 5: Predictive Maintenance Software: Building Rules with Kilo

Turn the maintenance criterion into a repeatable response. Kilo’s visual rules engine uses BPMN flowcharts with CEL expressions for precise conditions, calculations, and branching. Triggers can evaluate selected devices independently and start a rule immediately or after a condition persists for a configured duration.

For example, monitor a compressor’s reported discharge temperature against its documented warning limit. Configure an appropriate duration, then raise an alarm directing the technician to check the equipment and its operating conditions. Use available operating-state data to distinguish normal startup behavior from a condition that warrants attention.

Build separate rules for separate maintenance questions. Version history and deployment controls help teams review changes and restore an earlier version when needed. The AI assistant can help author, test, and deploy rules from plain-language instructions, making the setup more approachable for the people who know the equipment.

Dashboards provide charts and gauges for reviewing readings, while the Digital Building Twin can place live sensor values in a building view. Configured device commands can also form part of an operational response on compatible equipment, coordinated with the site’s control strategy.

Step 6: Equipment Monitoring Alarms and Escalation

Give every alarm an owner and a clear action. Kilo provides five severity levels to classify urgency, alongside configurable escalation steps that specify recipients, delays, and notification channels. Email, push, and SMS options help match the response to the team’s working arrangements.

A pump warning might first notify its maintenance technician. If the event remains unresolved, a later step can contact a backup or supervisor. Resolution stops further escalation. Notification schedules and repeat settings help teams manage delivery across shifts, while the centralized inbox keeps events visible.

Include enough context for the recipient to act: asset name, location, observed condition, relevant limit, and the first inspection step. Agree which conditions require immediate attendance and which can enter the planned maintenance queue. Review those decisions with the people who actually respond.

Step 7: Testing Condition-Monitoring Rules and Alerts

Test the rule with representative input before expanding the pilot. Kilo’s debugger lets you step through execution, inspect variables, and see which branch a condition takes. Include normal readings, values at the threshold, missing fields, and mismatched units in your checks.

Use the debugger’s Mock or Skip options for actions while checking logic. Then run a deliberate delivery test with agreed recipients and equipment. This separates checking a calculation from verifying an actual notification or command.

Also test the complete path from a device or gateway through ingestion, the trigger’s persistence period, the rule, and the alarm response. Confirm that recovery clears the condition as intended and that resolution ends escalation. Record the result so another technician can repeat the test after a configuration change.

Step 8: Predictive Maintenance and CMMS Integration

A computerized maintenance management system (CMMS) organizes work orders, assignments, and service history. Connect that workflow to condition monitoring by defining how a reviewed equipment finding becomes a maintenance request.

Start with a consistent handoff: asset identifier, location, observed condition, measurement time, priority, and recommended inspection. Kilo’s public REST API provides access to device and sensor data, including measurement history. A configured integration can retrieve relevant readings and use the CMMS’s supported interface to attach evidence or create a work request according to your approval process.

Map identifiers between systems, prevent repeated readings from creating duplicate requests, and assign ownership of integration errors. Have technicians record what they found and what work they completed. Those outcomes help refine thresholds and show which monitoring rules produce useful maintenance actions.

Common Predictive Maintenance Mistakes

• Starting with sensor purchases: Define the maintenance question and check existing measurements before adding instrumentation.
• Treating every alert as a diagnosis: Use the alarm to prioritize investigation, then record the technician’s findings.
• Copying thresholds between machines: Match limits to equipment type, operating state, measurement position, and units.
• Ignoring data freshness: Include timestamps and reporting intervals in the pilot checks so teams know how current a reading is.
• Skipping the response procedure: Agree who investigates, who provides backup, and how the event is resolved.
• Dropping scheduled maintenance: Combine condition-based decisions with required preventive tasks, inspections, and manufacturer service requirements.
• Expanding before reviewing results: Check alarm usefulness and response quality on the pilot asset before repeating the setup across sites.

Predictive Maintenance Tools and Resources

Kilo brings MQTT ingestion, LoRaWAN and mioty connectivity, dashboards, visual rules, and alarm escalation together for a practical equipment-monitoring program. Its Free plan covers up to five devices, one dashboard, and one rule, with no card required and no expiration. Starter is €25 per month for up to 25 devices. Gateways are unlimited on every tier; SMS uses paid credits. Budget separately for any hardware and integration work your installation needs.

Keep an engineering reference pack for each pilot: manufacturer limits, measurement definitions, normal operating records, and the applicable parts of ISO 17359, ISO 20816, or ASHRAE guidance. The U.S. Department of Energy’s Operations & Maintenance Best Practices Guide provides broader context for condition-based maintenance within an operations program.

Predictive Maintenance Program FAQ

Can you start predictive maintenance without buying new sensors?

Yes, when existing instrumentation provides the measurements needed for the chosen maintenance question. Reuse accessible PLC, BMS, or meter feeds, then configure the connection and rules. A software bridge on an existing computer can connect suitable industrial interfaces without adding a gateway appliance.

What is the difference between predictive and preventive maintenance?

Preventive maintenance follows scheduled intervals or usage limits. Predictive maintenance uses equipment-condition evidence and analysis to anticipate maintenance needs. A practical program combines both, using monitoring to guide investigations while retaining required service tasks.

How does Kilo turn equipment data into maintenance actions?

Kilo evaluates connected measurements through triggers and visual rules, then routes alarms to designated recipients. Teams can add calculations, contextual information, and configured device commands to the response. The AI assistant helps build and test the workflow.

Can Modbus and BACnet equipment connect to an IoT platform?

Yes. A compatible gateway or software bridge can publish selected Modbus registers or BACnet points over MQTT for ingestion into Kilo. OPC UA equipment can follow the same pattern through a suitable bridge. Choose the interface around the existing system and required data.

How much does a predictive maintenance pilot cost?

Kilo’s Free plan supports up to five devices, one dashboard, and one rule. Start with existing equipment feeds to keep the initial hardware requirement small. Add any gateway, sensor, connectivity, or integration costs according to the pilot’s actual scope.

How do you choose an equipment-monitoring threshold?

Start with manufacturer guidance and verified normal operation. Apply an engineering standard when its scope matches the equipment and measurement. Record the threshold’s reasoning, then review inspection results to improve its usefulness under real operating conditions.

Start a Predictive Maintenance Pilot with Kilo

Choose one important asset, identify the condition worth investigating, and connect the measurements already available. Build and test the rule, establish a response owner, and review the resulting inspections before expanding across more equipment or sites.

Measure the pilot by data availability, useful alerts, response times, and completed inspections. Record whether each inspection found a developing issue and whether the warning arrived early enough to schedule work. Review nuisance alarms with technicians and adjust the documented criteria. This creates an evidence base for deciding which assets to connect next and which measurements will bring the most value.

The Kilo IoT Platform gives that program a shared home for equipment data, wireless additions, dashboards, rules, and alarms. For teams planning maintenance improvements in 2026–2027, it provides a practical way to extend existing infrastructure and turn useful readings into coordinated action.