Selecting Sensors for Utility Infrastructure | UtilityGrid
UtilityGrid · Knowledge Center

Selecting Sensors for Utility Infrastructure

Learn how to select sensors for utility infrastructure based on measurement needs, operating conditions, installation, calibration, and data quality.

⌁
Start with the physical question. Asset · Measurement · Decision · Site conditions
The selection principle

Choose the measurement before the instrument.

Sensor selection begins with the physical question the utility needs to answer. A sensor may measure pump vibration, motor temperature, pipeline pressure, water flow, reservoir level, transformer temperature, or another operating condition. The correct choice depends on the asset, the decision that will use the reading, and the conditions in which the instrument must operate.

For water and wastewater systems, measurements may support pump-station oversight, process monitoring, level management, and maintenance planning. For electric utilities, they may support transformer and switchgear condition assessment, electrical measurements, and substation inspection. In each setting, define the required measurement before comparing sensor models or planning analytics.

01

Define the Measurement Requirement

Start by documenting the asset and the operating question. A request to monitor a pump is not yet a sensor specification. The team must decide whether it needs to detect mechanical vibration, track bearing temperature, confirm suction or discharge pressure, measure flow, or combine several signals to understand operating state.

For each measurement, record the following:

  • Physical variable and engineering unit, such as vibration velocity in mm/s, temperature in °C, pressure in kPa or bar, flow in m³/h, or electrical current in A.
  • Expected operating range, including normal operation, startup, shutdown, transient conditions, and credible abnormal states.
  • Required measurement uncertainty, resolution, repeatability, and response time for the intended decision.
  • Asset identity and measurement point, including the equipment tag, location, mounting surface, and orientation where relevant.
  • Operating context that can affect the reading, such as pump speed, valve position, load, fluid characteristics, ambient temperature, or equipment duty cycle.
  • The person or system that will review the measurement and the action that may follow an abnormal result.
Use equipment nameplates, original equipment manufacturer (OEM) documentation, process and electrical drawings, instrument specifications, maintenance records, and site measurements to establish the expected range. Do not select a range from a generic example when the actual equipment design or operating envelope is available.
Instrument matching

02. Match the Sensor to the Physical Variable

The sensor type should follow the variable and the required measurement method. Similar-looking instruments may differ in measurement principle, range, accuracy, response, installation requirements, and suitability for the site environment.

Utility measurementCandidate instrumentSelection considerations
Pump and motor vibrationAccelerometer or vibration transmitterMounting location, measurement direction, frequency response, machine speed, cable routing, and the diagnostic method.
Bearing, motor, or transformer temperatureResistance temperature detector (RTD), thermocouple, or approved temperature transmitterContact method, installation point, thermal response, range, insulation, and ambient conditions.
Pipeline or pump pressurePressure transmitter or pressure switchGauge or absolute reference, pressure range, process connection, fluid compatibility, pulsation, and overpressure conditions.
Water or wastewater flowElectromagnetic, ultrasonic, or other application-appropriate flow meterPipe material, diameter, fluid conductivity or composition, straight-run requirements, solids, fouling, and installation access.
Tank, wet well, or reservoir levelRadar, ultrasonic, hydrostatic, or other level instrumentFoam, vapor, turbulence, submergence, sensor mounting, material compatibility, and required measurement span.
Electrical loading and power qualityCurrent transformer, voltage transformer, or power-quality instrumentElectrical rating, isolation, measurement bandwidth, installation category, and applicable protection and safety requirements.
Transformer conditionTemperature, dissolved-gas, bushing, or partial-discharge monitoring instrumentTransformer design, available measurement points, monitoring objective, baseline data, maintenance plan, and specialist review.

A sensor that measures one variable should not be treated as a direct measurement of another. For example, motor current can help characterize electrical loading, but it does not directly measure bearing vibration. A temperature trend may indicate a developing issue, but it does not by itself establish the cause. AI models may derive features or estimate a condition from multiple inputs; those outputs should remain distinguishable from the raw instrument readings.

Worked Example: Selecting Measurements for a Utility Pump

Consider a pump station where the maintenance team wants better visibility into pump and motor condition. The initial question is whether the objective is to identify mechanical changes, confirm hydraulic performance, or understand both. A vibration measurement at an appropriate bearing housing can help characterize mechanical behavior. A temperature measurement can provide a separate view of thermal condition. Suction and discharge pressure, together with flow, can help describe the hydraulic operating point.

Build a practical selection record

Identify the measurement source, unit, required range, uncertainty, physical placement, sampling approach, operating modes, and conditions that could invalidate the reading. For vibration, review machine speed, bearing arrangement, mounting rigidity, measurement direction, and the frequency content needed for the intended analysis.

Understand what a simpler option misses

A temperature-only monitoring point may be simpler to install and adequate for a narrowly defined thermal warning, but it would not provide the vibration information needed to assess mechanical behavior. The preferred configuration depends on the failure modes, available maintenance response, and evidence the utility needs.

Sampling settings should be determined from the diagnostic objective and engineering review, not copied from a generic template.
03

Evaluate Installation and Operating Conditions

A technically suitable sensor can produce poor data if it is installed in the wrong location or exposed to conditions outside its design limits. Review the full installation before approving an instrument.

  • Mounting and contact: Confirm that the surface, process connection, insertion depth, orientation, and mounting method suit the measurement principle.
  • Environmental exposure: Assess moisture, flooding, dust, corrosive atmospheres, chemical exposure, ultraviolet exposure, temperature extremes, vibration, and washdown requirements.
  • Electrical and hazardous-area requirements: Verify power supply, isolation, grounding, enclosure rating, area classification, and applicable site safety requirements.
  • Interference: Consider electromagnetic interference, nearby motors and drives, structural vibration, pipe turbulence, radio obstructions, and sources of heat or acoustic noise.
  • Access and maintainability: Allow for inspection, calibration, cleaning, replacement, cable access, and safe isolation without creating unnecessary maintenance burden.
  • Power and communications: Document wired or wireless connectivity, network coverage, power availability, battery replacement needs, gateway placement, and behavior during communication loss.

Remote water assets

A low-power wireless instrument may reduce cabling but introduce battery, coverage, and message-delay considerations.

Substations

Electrical isolation, electromagnetic conditions, access control, and maintenance procedures may be more important to the installation decision. Confirm these constraints at the actual site.

04

Design for Reliable Data

Sensor selection is also a data-quality decision. A measurement becomes useful only when its value, unit, timestamp, asset identity, operating context, and quality status are preserved. The Sensing Engine should receive enough information to distinguish a real equipment change from an instrument or communications problem.

Calibration and drift

Define the calibration method, reference equipment, calibration interval, acceptance tolerance, and responsible party. Review calibration certificates and traceability where required by the utility’s quality procedures. Sensor drift can develop gradually, so trend checks and periodic reference comparisons may be needed between formal calibrations. Record calibration events and any adjustment or replacement.

Missing readings, outliers, and sensor health

Plan how the system will identify missing, duplicated, stale, implausible, or out-of-range observations. A sudden change may reflect a real operating event, a loose connection, a failed instrument, a process transient, or a communications interruption. Use quality flags and diagnostic context rather than automatically treating every unusual value as an equipment fault.

Sampling and timestamps

Choose the sampling interval and acquisition method according to the physical behavior and decision latency. Slowly changing tank levels and fast-changing vibration signals have different data requirements. Preserve the measurement timestamp, ingestion timestamp, time zone or UTC convention, and any aggregation applied. If data is averaged, filtered, or reduced to a derived feature, retain enough information to understand what the value represents.

Application context

05. Consider Utility-Specific Sensor Choices

Asset type and process conditions shape the measurement plan, instrument materials, installation details, and interpretation requirements.

Water and wastewater assets

At pump stations, lift stations, treatment facilities, and distribution assets, the measurement plan may include pump vibration, motor and bearing temperature, suction and discharge pressure, flow, wet-well or tank level, and selected water-quality or environmental variables. Sensor materials, process connections, submergence, solids, corrosion, condensation, and cleaning practices can affect instrument life and data quality. Confirm that the selected instrument is suitable for the fluid and installation point.

Electric utility assets

At substations and other electrical facilities, measurement needs may include transformer oil and winding temperature, bushing condition, dissolved-gas indicators, partial discharge, current, voltage, and ambient conditions. These measurements have different installation and interpretation requirements. Some require specialist equipment, established baseline practices, or integration with existing protection and monitoring systems. A monitoring sensor should not be assumed to replace protective relays or other independent protection.

06

Establish an Acceptance and Verification Plan

Before a sensor is accepted for operational use, define how the utility will verify that it measures the intended variable under representative conditions. Acceptance should cover the instrument, installation, data path, and the way the measurement will be used.

  • Confirm the instrument model, measurement principle, range, units, uncertainty, response characteristics, environmental rating, and approved installation details.
  • Inspect mounting, process connections, wiring, grounding, power, enclosure, labeling, and access for maintenance.
  • Compare readings with an appropriate reference instrument or documented reference method across relevant operating conditions.
  • Check repeatability, stability, response, data gaps, timestamp alignment, and behavior during startup, shutdown, load changes, and expected transients.
  • Verify that readings are associated with the correct asset, measurement point, unit, and operating mode in the receiving system.
  • Document calibration evidence, configuration, firmware or device settings where relevant, test results, deviations, corrective actions, and approval responsibilities.
  • Define how operators or maintenance personnel will review alerts, investigate uncertain results, and record the outcome.
Acceptance limits should come from equipment requirements, instrument specifications, site procedures, and the intended decision. Record any conditions that invalidate the selection, such as an unsuitable mounting surface, process fluid outside the instrument’s compatibility limits, insufficient sampling for the diagnostic objective, or communication coverage that cannot support the required reporting interval.
07

Preserve Sensor Assumptions in Analytics and Simulation

The same selection fields should be carried into analytics, pilot evaluation, and synthetic-data work. A data model should record the measured variable, unit, asset and measurement point, timestamp, operating mode, sampling interval, expected range, uncertainty assumptions, calibration status, and quality flag. This helps technical teams distinguish an observed value from a derived feature or an estimated equipment state.

For synthetic data, model the underlying operating condition separately from the imperfect sensor observation. Include plausible measurement noise, drift, missing values, outliers, timestamp mismatch, delayed messages, and sensor faults where relevant. Keep hidden ground-truth states separate from the data available to the candidate method. Clearly label synthetic measurements as simulated; they are not calibrated field readings and do not establish actual sensor performance.

Practical review

Sensor Selection Checklist

Use this checklist to review whether the measurement, installation, data, and operational workflow have been defined.

Before approving a sensor

0 of 7 items completed
Knowledge center

Frequently Asked Questions

What should be decided before choosing a sensor?

Define the physical variable, the decision it supports, the expected operating range, the required uncertainty and response, and the conditions at the installation point. These requirements narrow the instrument choices before supplier or model comparisons begin.

Can one sensor monitor every aspect of a pump?

No. Vibration, temperature, pressure, flow, and electrical loading describe different aspects of pump operation. A monitoring design may combine measurements when the maintenance or operating question requires them.

How often should a utility sensor be calibrated?

There is no single interval for every sensor or application. Set the interval using instrument requirements, site quality procedures, operating conditions, calibration history, risk, and the consequences of measurement error. Review drift and reference-check results when adjusting the interval.

How can a utility tell whether an unusual reading is a real equipment issue?

Check sensor health, calibration history, asset identity, timestamps, operating mode, and related measurements. Compare the observation with an appropriate reference or inspection finding before treating it as a confirmed equipment condition.

Start a conversation

Review an Asset or Measurement Requirement

To review a specific asset or measurement requirement, contact UtilityGrid with the equipment class, operating question, available data, and site constraints.

Contact UtilityGrid ↗