Facility Managers: Generator Monitoring NFPA Compliant and NIST Secure

Table of Contents

• What to monitor: critical telemetry parameters and thresholds
• Alarm strategy and notifications that avoid false fatigue
• Maintenance, testing and NFPA 110 compliance in practice
• Telemetry architecture: RTUs, gateways, sensors and master stations
• Securing remote telemetry against NIST and NCCoE guidance
• Turning telemetry into useful records and trend data
• Building a deployment checklist and maintenance calendar
• Fuel management and storage monitoring
• Integrating generator monitoring with BMS or SCADA
• Predictive maintenance from monitoring data
• Emissions and environmental compliance monitoring
• Training and documentation for monitoring personnel
• What experienced monitoring deployments get wrong
• How PODTECH supports enterprise generator telemetry
• FAQ
• Sources

What to monitor: critical telemetry parameters and thresholds

Generator telemetry is not worth the cost unless it monitors the values that predict failure. PNNL's standby-generator guidance lists oil pressure, coolant temperature, current, voltage, and frequency as the essentials every system should record, in addition to run hours and start attempts.

Every alarm has a reason to exist. If oil pressure falls below the manufacturer's low-pressure specification, it is time to pull the plug immediately because the engine may suffer catastrophic damage if it runs too long. Coolant temperature exceeding the normal operating range indicates a cooling system issue, usually a failed fan clutch or low coolant condition. Battery voltage and charger output are more important than many teams realize: weak batteries are listed on countless facility inspection sheets as the number one reason generator starts fail, which is why these measurements deserve their own alarms instead of a generic diagnostic code.

Fuel level ties the picture together. If it drops between loads without additional run hours, the likely explanation is theft or leakage rather than normal consumption.

Build your monitoring around this set:

• Run hours and start attempts: indicate units nearing service intervals or having difficulty starting.
• Oil pressure: low-pressure alarms should trigger an immediate maintenance dispatch.
• Coolant temperature: sustained highs indicate cooling system faults.
• Battery voltage and charger status: often the first indication of a failed start before the failure occurs.
• AC voltage, frequency, and alternator current: confirm the generator is producing usable power rather than simply running.

Battery failure is the number one reason generators fail to start on facility checklists. Because of this, voltage and charger monitoring is one of the highest-value signals in the entire telemetry stream.

Alarm strategy and notifications that avoid false fatigue

An alarm that fires constantly is useless. By tiering alarms into informational, action required, and critical, you allow on-call personnel to triage at a glance and avoid treating every alert like a fire drill.

Map alarm tiers to real-world events. A successful weekly self-test is informational. An increasing coolant temperature over a few exercise cycles indicates action is needed. Low oil pressure causing a shutdown is critical. Derived alarms help here. Instead of alarming every time there is an exercise cycle, configure the system so it ignores expected self-tests and only alarms on out-of-tolerance sequences, such as trying to start but not achieving rated frequency within the allowed time.

1. Classify every alarm into informational, action required, or critical before deployment.
2. Route each tier to the appropriate channel: dashboard log, email or SMS, or phone escalation.
3. Put a duty roster in place so critical alarms are sent to a named person within minutes, not to a generic inbox.
4. Audit nuisance alarms monthly and fine-tune thresholds that trigger too often so alarms retain significance.

Tip: Configure derived alarms to trigger on sustained conditions, such as three readings outside the acceptable band, rather than a single data point. This reduces false positives while still alerting on actual faults.

Maintenance, testing and NFPA 110 compliance in practice

NFPA 110 establishes the testing cadence facility managers are ultimately held to. It defines weekly inspections and monthly or annual test frequencies with written documentation that an authority having jurisdiction can audit.

A weekly visual check should verify oil level, coolant level, battery charger status, control panel mode, and day-tank fuel level, with abnormalities noted at the time of testing rather than at the end of the month. The monthly load test is where many maintenance programs fail: allowing the unit to run unloaded for several minutes proves very little. A proper monthly load test runs the generator for an appropriate amount of time under a substantial percentage of nameplate load so the engine and alternator are sufficiently stressed and wet stacking is avoided.

Annual testing should include, and go beyond, a complete ATS functional transfer with load applied. On a three-year or manufacturer-recommended basis, more comprehensive load-bank testing should verify capacity across load points.

• Weekly: visual inspection, fluid levels, battery charger status, and fuel level.
• Monthly: load test run for an adequate length of time at a substantial percentage of nameplate capacity.
• Annual: ATS functional verification under load and full documentation review.
• Triennial or as specified: extended load-bank testing across load bands.

Maintain documentation in an audit-ready state at all times. If you have to recreate records later, the program has already failed its purpose.

Telemetry architecture: RTUs, gateways, sensors and master stations

Everything beneath the dashboard determines the actual confidence level of the data. The RTU sits at the generator and converts controller data into something a master station can understand, typically Modbus for legacy controllers and SNMP for newer network-aware controllers. Multiprotocol support is important on mixed fleets because forcing a single protocol often means replacing perfectly good controllers.

Network media affects both reliability and security. Cellular connections are ideal for temporary or remote locations where laying cable is cost prohibitive. Wired Ethernet provides more reliable throughput for permanent sites. If connecting remotely through a VPN gateway, placing the gateway at a defined network boundary instead of inside the generator's internal network simplifies future security segregation.

Sensor selection deserves the same attention as the network layer. Pressure or temperature sensors that drift out of calibration silently reduce alarm accuracy long before anyone notices. Periodic checks against a known reference, plus a signal verification step that compares sensor output to a manual reading, should be included in the same maintenance cycle as the generator itself.

• RTUs with multiprotocol support bridge legacy Modbus controllers and modern SNMP-capable units.
• Cellular or wired connectivity, selected based on site permanence and cable access.
• VPN gateways placed at a defined network boundary, not inside the generator LAN.
• Scheduled sensor calibration and manual verification, to keep alarm thresholds meaningful.

Securing remote telemetry against NIST and NCCoE guidance

Remote access is what makes monitoring valuable, and it is also what makes monitoring attractive as a target. NIST SP 1800-32 recommends network monitoring, strong authentication and access control, and immutable logging of control commands as a baseline for distributed energy resource telemetry security, including generator fleets.

CISA's guidance on controlling remote access is clear about what to remove and what to replace it with: eliminate standing vendor modems, consolidate on VPN-based access, and open vendor connections just in time instead of leaving them permanently exposed. Installing a jump box between the corporate network and the generator control network creates a single, logged chokepoint for all remote sessions.

Logging failed authentication attempts and unusual successful logins is foundational detection guidance for remote access abuse. This is a lower bar than many teams assume: most monitoring platforms and RTUs already generate this data, but it still has to be collected and reviewed.

• Establish a cyber demarcation point between the generator control network and the corporate network.
• Require multi-factor authentication for every remote session, whether vendor or internal.
• Use just-in-time vendor access, opened only for the duration of the support session.
• Collect logs centrally through SIEM or syslog and treat command logs as immutable records.

Suspicious activity includes abnormal device behavior, multiple authentication failures, or off-hours logins.

Our industrial IoT security roadmap explains how to prioritize these controls across an entire facility's sensor and telemetry assets, not just the generator fleet.

Turning telemetry into useful records and trend data

Raw logs are only half the battle. The important records are timestamped alarms, all attempted starts and their results, ATS transfer timestamps, complete load-test recordings, and fuel analysis if it is performed. Treat PNNL's recordkeeping recommendations as your baseline.

Retention may need to satisfy an authority having jurisdiction conducting an audit, which often means several years of test records rather than a rolling few months. The real value appears in trending: monitoring ATS transfer time and battery cold cranking performance over 12 to 24 months often reveals slow deterioration long before a unit fails outright. Dashboards that highlight relative drift catch issues a simple pass/fail check would miss.

• Log every alarm, start attempt, and load test with a timestamp and outcome.
• Retain records long enough to satisfy AHJ expectations, not just internal review cycles.
• Track transfer time and battery condition as trend lines, not isolated readings.
• Flag fuel consumption anomalies that do not match logged run hours.

Building a deployment checklist and maintenance calendar

Rolling out a new monitoring system works best with a defined order of operations rather than a rolling build. Start with an asset list that identifies every generator, controller type, and sensor already in place. Perform a baseline load test so you know what normal looks like before adjusting alarms. Confirm sensor health by cross-checking with manual readings. Tune alarms so they do not create nuisance alerts. Establish the cybersecurity baseline, including demarcation point, MFA, and jump box, before go-live instead of bolting it on later.

1. Inventory every unit, controller protocol, and existing sensor.
2. Run a baseline load test to set reference values.
3. Verify sensor accuracy against manual instrument readings.
4. Tune alarm thresholds and tiers before go-live.
5. Apply the cybersecurity baseline: segmentation, MFA, and logging.

• Weekly: visual checks, fluid status, and battery status.
• Monthly: load test and alarm review.
• Annual: ATS test, full documentation audit, and sensor recalibration.

Perform an unscheduled load-bank test any time you see recurrent cycles of light-load exercise in your run logs, since repetitive light-load operation can cause wet stacking, or if the unit does not achieve rated output during a monthly exercise.

Fuel management and storage monitoring

Fuel is often the most under-measured input in generator monitoring programs, yet it determines how long backup power will actually last. Day-tank and bulk-storage level sensors should feed the same telemetry stream as engine parameters so a level drop outside expected consumption triggers the same alarm matrix as a coolant fault.

Diesel eventually degrades through water contamination and microbial growth. Monitoring storage tanks should therefore include occasional sampling for fuel quality in addition to level measurement, especially on generators that rarely run under load. A tank can read full and still contain unusable fuel.

Simple fuel gauging compared with logged run hours can quickly expose discrepancies. If fuel consumption does not match the expected burn rate for the hours logged, there is likely a leak, metering error, or unauthorized use. Integrating this into the same dashboard as engine data helps ensure fuel management is not overlooked.

Fuel logistics matter as well. Where facilities rely on generator backup for longer outages, scheduling fuel deliveries should be part of the monitoring program rather than treated as a separate procurement process. A generator with pristine telemetry but an empty tank still provides no power. Generator vendors that take responsibility for installation and site electrical work, including Guardian' generator services, often find fuel system design to be a common deficiency during commissioning.

Integrating generator monitoring with BMS or SCADA

Generators that send data only to their own dashboards force personnel to watch multiple displays when trouble arises. During an outage, every second counts, and staff needs to work from a single operational picture. Integrating generator telemetry into the building's existing BMS or SCADA places emergency power information alongside HVAC, electrical, and life-safety data on one platform.

Integration most often uses the same channels already implemented for telemetry gathering, with Modbus or SNMP being the most common. The generator RTU or gateway exposes datapoints that the BMS can poll directly. The practical advantage appears during an actual outage: an operator watching the BMS sees the ATS transfer, the generator voltage stabilize, and the building load shift as one coherent sequence rather than three separate alarms.

Correct integration usually means involving the monitoring team during system design rather than after implementation. You can read more in our article on early monitoring involvement in mobilisation. Forcing integration onto a generator set that was not designed with accessible data points usually costs more.

Predictive maintenance from monitoring data

Years of run-hour, load-test, and trend data can turn a monitoring system into a maintenance planning tool rather than a reporting tool. Machine learning models applied to historical data can flag units whose performance is drifting from normal, catching subtle decline in battery condition or transfer time before it leads to failure.

Predictive analytics proves its worth most clearly on larger fleets where manually reviewing trend lines for dozens of units becomes impractical. On a single-generator site with disciplined baseline telemetry and monthly load tests, careful manual trend review often captures much of the same value without the overhead of predictive modeling.

One area where predictive maintenance clearly adds value is prioritization: ranking which units across a fleet should receive attention first based on multiple signals, such as battery trend, transfer time drift, and fuel anomalies, rather than any single reading. You can read our guide to predictive maintenance models for how to move from raw telemetry to that kind of prioritized alerting.

Regardless of how notifications are configured, the premise remains the same: alerts are only as useful as the telemetry powering them, so the parameter list and alarm tiering discussed earlier are foundational either way.

Emissions and environmental compliance monitoring

Standby generators are often regulated by air permit conditions that limit annual run hours and, in some jurisdictions, emissions during testing. Tracking run hours against permit limits should be integrated with the maintenance telemetry already being gathered, because the same run-hour log used for maintenance scheduling is usually the logbook regulators want to see.

Installations with multiple units, or those performing frequent load-bank testing, should keep cumulative totals by unit rather than by site because permit conditions are typically specified per engine. An engine nearing its annual testing-hours limit should be identified long before it reaches the cap, not discovered during annual compliance calculations.

If local emission standards are enforced through exhaust monitoring during testing, that information should be recorded as diligently as electrical datapoints: timestamped, archived, and linked to the test event it came from. Treating emissions compliance as an add-on report instead of routine telemetry is exactly how compliance programs end up in a last-minute fire drill.

Training and documentation for monitoring personnel

The most sophisticated monitoring system is only as good as the person reading its alarms. Personnel responsible for generator alarm response must understand, from an operational standpoint and not just a technical one, what each alarm level means so an oil pressure failure prompts an immediate site visit rather than a note for the next shift.

Documentation should clearly define three processes: how to interpret each alarm and threshold, how to escalate when alarms are not correcting the issue, and how to document the result once the issue has been corrected. Keep this as a live document that is updated whenever a threshold or alarm tier changes. That avoids the common problem of training material describing how the system used to work.

Ideally, new personnel joining a generator monitoring program should shadow both a real load test and a full alarm pull before taking rotation. Reading a procedure and responding to a live critical alarm are different skills. For operations with more than one facility, this training should be standardized across buildings so a technician is not relearning alarm norms every time they move to a new site.

What experienced monitoring deployments get wrong

The biggest mistake is installing sensors before deciding what action should occur for each alarm. That creates data-drowning scenarios with no action plan. Predictive analytics justifies itself when a fleet is large enough that manual trend review is unrealistic; smaller sites often benefit more from disciplined baseline telemetry first.

Three questions to ask before signing with any monitoring vendor are: what protocols does the RTU natively support, how is remote access secured by default, and what happens to historical data if you change providers later.

— Harry

How PODTECH supports enterprise generator telemetry

Generator monitoring that needs to integrate into facility-wide building and datacenter systems, rather than remain a siloed dashboard, quickly outgrows commercial off-the-shelf solutions.

Relevant capabilities for teams planning or upgrading a monitoring program include:

• Master Systems Integration, merging generator, BMS, and PMS data into one source of truth.
• Datacenter Telemetry and Containment Monitoring, extending the same telemetry discipline across critical infrastructure.
• POD View Liquid Monitoring platform, for sites needing sensor-level monitoring and analytics.
• Alarm routing and escalation workflow software that scales across fleets.

Engagement can be through full teams or staff augmentation depending on how your facility's internal team wants to operate. Check out PODTECH's services overview to learn about scheduling a pilot integration on your generator fleet.

FAQ

What is the 20/20/20 rule for generators?

There is no universally accepted rule endorsed by consensus standards such as NFPA 110. The phrase appears informally among some operators, but definitions vary. Facility operators should reference NFPA 110 written testing intervals rather than rely on a rule of thumb when determining schedule.

What is the 80% rule for generators?

This is a general rule of thumb, not an absolute standard. Always check the specific sizing and loading guidance recommended by the manufacturer of your system.

How can I monitor my Generac generator?

Generac units can be monitored using the manufacturer's proprietary telemetry solutions or a third-party RTU or gateway that reads the controller through its available protocol. Either way, make sure the system monitors key parameters such as oil pressure, coolant temperature, battery voltage, and run hours, and that these values are placed into alarm tiers your team will actually act on.

How often should a generator be checked?

NFPA 110 requires visual inspections once a week and load testing once a month for most emergency power systems, along with annual and periodic extended testing. Fluid levels and battery condition are checked weekly, and the monthly test should operate the unit under load for at least 30 minutes.

Does PODTECH monitor generators directly?

PODTECH does not run generators. Instead, it manufactures enterprise telemetry and monitoring software that links generator data to BMS, PMS, and NMS systems. PODTECH's services are commonly combined with a facility's existing generator maintenance plan, including integration and platform services such as the POD View Liquid Monitoring platform.

Sources

• NIST SP 1800-32 — Vol. B
• Best Practices for Standby Generator Operations and Maintenance — PNNL
• NFPA 110 — Standard for Emergency and Standby Power Systems
• CISA — Managing Remote Access

Recommended

• NVIDIA AI Factory Monitoring Starts at the CDU
• Critical Infrastructure Datacenter Solutions
• PODVIEW Datacenter Monitoring Platform
• Master Systems Integration: The Engine Behind Smart-Ready Buildings