An intelligent energy management system should help a team answer three questions: where is energy being used, what should change, and did that change work? Buying decisions become clearer when each question has a measurement, an accountable operator and a testable result.
This guide sets out a practical buying brief for buildings, industrial sites and portfolios. The requirements below are project-selection guidance. They are not a claim that every product sold as an IEMS includes every function.
What is an intelligent energy management system (IEMS)?
An intelligent energy management system connects energy measurements with analysis and an operational response. It can support operator decisions or commissioned automatic control. The term alone does not establish that a product includes AI, closed-loop control or verified savings.
Keep four functions distinct when comparing proposals:
| Function | What it delivers | Evidence to request |
|---|---|---|
| Monitoring | Measurements associated with known assets, units and times | A reading traced from its meter to the displayed and exported value |
| Analysis | Comparisons, calculations or alerts that help explain operation | A reproducible result with its inputs, assumptions and missing-data treatment |
| Operational decisions | Someone investigates a finding and approves a change | A named owner, action record and follow-up measurement |
| Closed-loop control | An automatic action uses feedback to manage a measured condition | An approved control sequence, actual equipment feedback and tested failure behaviour |
A threshold alert can be useful without automatic control. A command sent to a controller does not prove that the equipment responded. A building management system (BMS) may already own HVAC schedules and plant sequences; agree where energy analysis informs that system and who retains authority to change it.
Start with the decision and measurement boundary
Write the first operational question before selecting meters or software. “Reduce energy use” is too broad to commission. “Identify why the compressor runs outside production hours” names a load, an operating condition and someone who can investigate it.
Choose a measurement boundary that matches the decision. A supply meter helps explain purchased electricity. A compressor meter helps explain that compressor. If photovoltaic generation or a battery sits behind the supply meter, reduced grid imports alone do not establish reduced site consumption. Record generation, storage and import/export flows needed to explain the balance.
| Decision | Useful measurements and context | What the data does not establish alone |
|---|---|---|
| Investigate out-of-hours operation | Equipment power or energy, operating state and an agreed schedule | Whether the load is unnecessary or safe to stop |
| Compare sites | Matching meter boundaries and periods, operating hours and relevant activity | That the lowest kWh per square metre is the most efficient site |
| Evaluate a plant improvement | Energy into the selected system and the useful service it delivers | Equipment efficiency from electricity consumption alone |
| Manage billed demand | Supply-point interval data and the applicable tariff | A bill reduction from an instantaneous power peak |
| Allocate consumption | Meter-to-tenant mapping, complete periods and documented adjustments | Approval to use the readings for billing |
For plant work, the equipment measurement-boundary guide explains why energy inputs and useful outputs must describe the same system. The compressed-air monitoring guide applies that decision to electrical power, air delivery and operating conditions.
Specify accuracy for the complete installed measurement chain. Sensor range, CT selection and installation, phase association, scaling and the operating load all matter. A meter's accuracy class alone does not establish the uncertainty of a calculated saving. Electrical installation and verification require qualified personnel.
Choose the site and portfolio architecture
An IEMS can combine existing meters, new sensors, a local Gateway, site software and a remote platform. The architecture should state where acquisition, history, analysis, rules and user access run, and which network each function needs.
Does an IEMS require a cloud platform?
No. A site can collect measurements, retain history, display dashboards and run configured rules locally. A remote platform can add portfolio or specialist workflows. Specify which functions must remain available during an internet outage and test them on the proposed system.
For an existing BMS or meter network, request the actual point list and interface documentation. Protocol support is only the start: confirm the device model, readable values, units, update intervals, permissions and any writable points. For new wireless measurements, survey the site around the required locations and reporting cadence.
Ask each supplier to document:
- Local operation: which readings, dashboards and rules remain available without the WAN, and how staff reach them.
- Storage and recovery: the recording policy, retention, queue capacity, maximum planned outage and behaviour when storage fills.
- Data delivery: authentication, source timestamps, retries, duplicate handling and evidence that the destination stored the right values.
- Portfolio consistency: stable site and asset identifiers, shared units and naming, and a controlled process for meter replacement.
- Portability: usable exports of readings, metadata and configuration, plus access and costs when a service contract ends.
Request separate evidence for local collection, successful transport and acceptance by the receiving platform. A connected status or successful send cannot establish all three. The integration guides describe these boundaries for documented EpiSensor connection paths.
Make data quality visible
A dashboard should distinguish a valid zero, a missing reading, an old reading and an estimated value. Set a freshness limit and an expected reporting interval for each point class. Keep estimates identifiable and retain the original observations when correcting or aggregating data.
The buying brief should define how the system handles:
- Time: source time versus arrival time, clock synchronisation, time zones, daylight-saving changes and interval boundaries.
- Meaning: units, multipliers, import/export direction, cumulative counters versus interval values, and the provenance of calculated points.
- Continuity: gaps, duplicate arrivals, late backfill, counter rollover and meter resets or replacement.
- Aggregation: which operation applies to each quantity. Adding cumulative meter readings does not calculate interval consumption; averaging power can conceal short events.
- Health: alerts for stale points or collection failure, with an owner who can investigate them.
Define completeness as valid observations divided by expected observations for a stated point and period. Report it per critical point as well as across the estate, so one failed meter cannot disappear inside a good portfolio average. Agree how scheduled shutdowns and deliberate recording changes affect that denominator.
Use a commissioning sample that includes a known gap, a late reading and a meter reset. Ask the supplier to show both the original records and the resulting chart or calculation. These are acceptance scenarios to agree, not assumptions about a platform's default behaviour.
Establish a baseline before claiming savings
Savings require a comparison with what consumption would have been without the intervention. EVO's public IPMVP principles explain the need for consistent boundaries and adjustments between baseline and reporting conditions. A lower bill or a before-and-after chart alone does not isolate the intervention's effect.
Choose a baseline that represents the relevant operating cycle. Record energy use alongside the factors that explain it, such as weather or production volume. Agree routine adjustments for those factors and a process for non-routine changes, such as altered floor area or a changed production process. Retain the original baseline and a versioned record of adjustments.
For example, a plant could use less electricity because it produced fewer units. The analysis must account for that change before attributing a saving to new controls. A kWh-per-unit ratio can help, but may mislead when fixed loads or product mix change.
ISO 50015 provides general guidance for measurement and verification of organisational energy performance. IPMVP supplies a framework with different measurement approaches. The U.S. Department of Energy's summary distinguishes retrofit isolation with key parameters measured (Option A), retrofit isolation with all parameters measured (Option B), whole-facility analysis (Option C), and calibrated simulation (Option D). More meters do not automatically make one option appropriate.
Before a performance-based procurement, ask the M&V practitioner to agree the boundary, method, baseline period, reporting period, explanatory variables, treatment of missing data, uncertainty and review responsibilities. Make the resulting report reproducible from retained inputs. A generic IEMS report is not itself an IPMVP-adherent M&V plan. Consult EVO's current protocol library for the project method; the public 2018 principles explain the fundamentals.
Keep energy, cost and emissions results separate. Cost calculations need the relevant tariff and periods. Emissions calculations need an explicit accounting boundary, factor source and factor date. A percentage saving should always identify its baseline, scope, period and method.
Assess demand management and automatic control
Energy use, often billed in kWh, and billed demand or capacity, commonly expressed in kW or kVA, answer different questions. Establish the exact quantity, interval and charging rule in the site's tariff. ESB Networks' current statement of charges, for example, uses kVA for some demand and capacity charges. The DOE rate guide shows why shifting a load can affect energy and demand charges differently. Its examples describe U.S. tariffs; the site's actual contract governs the calculation.
The interval-demand tool helps explain interval averaging. An engineering estimate is not a replacement for utility settlement data. Demand-response participation adds programme-specific baselines, metering, availability and verification requirements; assess these with the service provider through the demand-response project journey.
For automatic control, require a sequence that names the trigger, permitted action, maximum duration, equipment feedback and recovery state. It should also cover stale inputs, lost communications, a Gateway restart, a failed command and manual override. Keep equipment interlocks and approved safety controls authoritative.
Test a rule's physical result, not only its execution log. A successful command response and a verified reduction in load are separate observations. Make one system responsible for each controlled function so that a BMS schedule and an energy rule cannot silently compete.
If a proposal includes AI or forecasting, request a comparison against a simpler method on representative data withheld from model development. Agree forecast error, false-alert burden, model-change records and fallback behaviour. An “AI-powered” label is not an acceptance criterion.
Agree governance, commissioning and acceptance
Name the energy owner, site operator, installer, IT owner and M&V reviewer. State who can view readings, change point mappings, approve control rules and revise baselines. Include access removal, software updates, backup restoration, incident escalation and configuration history in the handover.
Does installing an IEMS provide ISO 50001 certification?
No. ISO 50001 addresses the organisation's energy management system, including responsibilities and continual improvement. Metering and software can support its evidence, but installing a product or exporting a report does not establish conformity or certification.
Turn the specification into signed acceptance criteria before the pilot begins:
- Point acceptance: reconcile the installed point list with physical assets. Check units, direction, scaling and timestamps against an appropriate reference within an agreed tolerance.
- Data acceptance: meet defined freshness and completeness limits over a representative operating period. Inspect gaps, duplicate handling and counter events at the intended destination.
- Outage acceptance: demonstrate the agreed local functions during a controlled network interruption. Check retained records and delivery recovery within the specified capacity and recovery time.
- Control acceptance, where included: witness authorised actions, actual feedback, overrides and failure recovery against the approved sequence.
- Analysis acceptance: reproduce a selected report from exported inputs, including its baseline version, adjustments and excluded data.
- Handover acceptance: provide configurations, point mappings, access ownership, operator training and a tested recovery procedure.
Set the numeric limits for the application rather than adopting an unexplained supplier default. A weekly investigation, a tenant allocation and a rapid control loop need different evidence. Technical acceptance of the monitoring system should be separate from a later finding that an efficiency measure delivered its expected saving.
Compare total project cost over the chosen service life: installation, site surveys, communications, licences, integration, calibration or verification work, maintenance and support. Include the staff time needed to investigate findings. A low software subscription cannot compensate for a measurement plan that cannot answer the site's question.
Where EpiSensor fits
EpiSensor Edge provides local device management, dashboards and historical data analysis. Configured automation can run on the Gateway when the required extension and device integrations are installed. Local history depends on recording settings, storage health and retention; external notifications and onward data delivery need their network connection.
Use Edge as the site system, or connect a chosen platform where portfolio analysis, specialist M&V or another workflow warrants it. Third-party software is optional. Choose the hardware from the actual measurements and existing meter interfaces, then qualify the complete path for the project.
The CoolPlanet manufacturing case study shows how circuit data helped identify boiler trips and out-of-hours loads. CoolPlanet combined those measurements with its software and engineering work at one client. It is evidence of that project's approach, not a transferable savings or payback guarantee.
For an estate, the Royal Mail case study describes a Cardiff pilot that retained the existing EnergyCAP platform and provided the basis for a wider programme. Use the multi-site application guide to define point naming, autonomy and rollout checks for your own sites.
Start an energy-management project with the site question, existing meter list, required reporting interval and person who will act on the result. Those details make the first installation useful and its acceptance criteria clear.