Energy monitoring: how to catch off-hours consumption

Energy monitoring collects and analyzes energy data over time to understand how a facility actually runs. One of its uses is spotting consumption after activity has stopped. To decide whether that consumption is excess, you need to connect the reading to schedules, equipment, and needs that persist after closing time.
That check changes the whole job. An empty office, a factory between shifts, and a store with walk-in coolers can all keep drawing power for different reasons. Looking at their curves lets you ask specific questions. Acting on them means knowing what's really behind the numbers.
Start with one time window your team can actually investigate. The goal is to end up with an explained incident and a decision, not a dashboard full of alerts nobody has time to review.
What energy monitoring actually shows you
A monthly total tells you how much energy was used in that period. A series of more frequent readings lets you see when that consumption happens. How much detail you need depends on the question: a bill might be enough for a budget review, but investigating a start-up spike needs a proper time series and clear equipment identification.
The U.S. Department of Energy describes EMIS as a set of devices, data services, and applications used to monitor, analyze, and control energy and system performance. That's a description of possible capabilities, and each one needs to be verified against the actual solution and project.
To study how a facility actually operates, you want to connect the reading to the calendar and to the variables that affect consumption. The 50001 Ready guidance on data collection covers energy sources, end uses, and relevant variables, along with the precision and repeatability of measurements. The data has to be good enough to answer the team's actual question.
Separate power from energy early on. kW describes power; kWh describes energy accumulated over an interval. A power spike and many hours of moderate consumption can point to very different problems. Check what the chart's axis actually represents before you decide on a fix.
A curve's shape gives you clues, not an automatic explanation. An overnight load could be a genuine ongoing need, equipment left on a schedule, or a fault. The analysis should end with a confirmed cause. Labeling all off-hours consumption as waste can lead you to shut down a service that was actually needed.
Monitoring and control are also two different things. Being able to see readings doesn't mean the tool can send commands to equipment. The ability to act depends on integration, permissions, and technical conditions. An EMS can offer both functions, but they need to be evaluated separately.

Proposed workflow. Does not represent results from any client
How to investigate off-hours consumption
Start by defining what off-hours actually means at that site. Factor in cleaning crews, maintenance work, and one-off activities. Across a portfolio, each location can run on a different schedule. Applying one single rule to every site tends to generate a lot of irrelevant alerts.
Next, identify exactly what the meter you're looking at measures. If it covers the whole building, its curve blends several different uses together. A sub-meter can help you pin down a zone or system, as long as you know how it relates to the main meter. Keep the metering diagram on hand so you don't pin a whole-building swing on the wrong piece of equipment.
Check the clock on the data too. The timestamp in a file can differ from the schedule used at the site. A time offset can make a normal start-up look like it happened off-hours. Note the time standard you're using, and check for gaps in the readings before you interpret a flat curve. Missing data means you need to check the connection or the metering, it doesn't prove the equipment was off.
Pick several comparable periods and note the differences between them. A holiday and a regular business day can be a useful comparison, but they need to be flagged as such. Add any context that could explain a variation: outside temperature, production volume, or a change in how many spaces were in use.
DOE/FEMP's building re-tuning screening approach uses operating data and separates occupied periods from unoccupied ones. It also flags how much the condition of sensors and controls matters. Before you blame the schedule, confirm that the reading and the equipment actually support that interpretation.
When you find a discrepancy, write it up as something that can actually be investigated. Include the site, the equipment or supply point, the time window, and the evidence. Spell out the open question: what activity took place, who changed the setpoint, or whether the schedule still matches current use. That record should save whoever picks up the alert from having to start the search over.
Priority depends on impact and on how easily you can act. A critical facility may need an immediate review even if the extra consumption is small. Another deviation can wait for the scheduled visit. Agreeing on that criterion with maintenance helps separate a useful signal from the noise of a system that flags everything.
If a fix is proposed, record who approved it and what conditions have to be preserved. There may be comfort limits, product preservation needs, or process continuity to respect. Whoever acts on the alert needs to know about them before changing a schedule. The same procedure should spell out how to revert if the test doesn't work out.
The investigation is done once you have a documented cause, or a pending action with a named owner. An annotated chart, a work order, and a follow-up review can be enough of a record to start with. How detailed that record needs to be should match the size of the decision being made.
Three sectors, three different questions
In offices, a good starting point is comparing HVAC operation against occupancy and whichever services stay active. DOE/FEMP's facility checklist includes matching HVAC schedules to occupancy. Applying it means reviewing the building and its own conditions, not copying one standard shutoff time across the whole portfolio.
A useful working question: which equipment needs to be ready for the first people in, and which is still running from the day before. The outcome might be adjusting a start-up time, justifying a continuous load, or tracking down a manual override. These are possible outcomes of the investigation, not results attributed to any specific client.
In manufacturing, consumption during a shutdown can help you review auxiliary systems. The Better Plants program [documents PLENCO's work on compressed air: the company measured flow rates, investigated demand outside production hours, and made changes to insulation and controls. That's an outside case, unrelated to Smarkia, and it doesn't turn their fixes into a recipe for any other plant.
The industrial question needs both production and maintenance at the table: which demand belongs to a process that has to keep running, and which part calls for investigating leaks or unnecessary use. Measuring electricity consumption can flag a time window; measuring the right auxiliary service can help explain the cause. Shutting down a line requires knowing its dependencies and the safe sequence for the process.
In food retail, a closed store still has product to keep safe. DOE's EMS guidance for food service applications covers refrigeration and points to temperature tracking and fault detection as relevant applications. Opening hours alone aren't enough to decide what can be switched off.
Here the investigation has to connect energy use to the condition of both equipment and product. A deviation might call for maintenance, a status check, or a review of the control system. The fix gets decided within whatever preservation conditions apply. The operational goal is using energy better while keeping the service running, with evidence for both.
Across sectors, the method stays the same: find a time window, identify a need, and check the effect of a fix. What changes is which condition needs protecting. That difference should show up in the rules you set and in the handoff to whoever acts on them.
What to ask of an EMS before rolling it out to more sites
Bring a real task to the demo. Ask the vendor to locate a reading, identify the meter, and connect consumption to the schedule. Then check how an incident gets created and what information reaches the person who has to resolve it. That walkthrough is what tells you whether the system is actually useful for your operation.
DOE/FEMP's description of EMIS capabilities separates out meter analytics, fault detection, supervisory control, and measurement and verification. That breakdown is useful for framing your questions: which function is included, what integration it needs, and which part still depends on a human review.
If the system can act on equipment, ask to see the permissions, the limits, and the change history. Check how a setpoint gets tested and how the facility team gets control back. A small, well-documented pilot lets you learn the dependencies before rolling automation out across many sites.
Follow-up needs to check both performance and the energy result. DOE/FEMP includes verification and communication as part of operating an EMIS. Finding an opportunity and confirming that the fix was actually applied are two different jobs.
To back up a savings claim, keep a proper baseline and consistent comparison conditions. If you're evaluating cost, separate out the effect of price. If you're evaluating energy, keep your units and boundary consistent. Closing one incident doesn't, on its own, prove an annual savings percentage, that still needs the proper evaluation.
Smarkia offers monitoring, analytics, remote management, and integration with existing systems. These are platform capabilities whose fit still needs to be checked against the site's own data and controls. That same evaluation should identify whatever access or metering is still missing.
To get started, bring one consumption curve and the schedule it's supposed to match. We can go through it with your team and find exactly where the path breaks down between spotting a deviation and resolving it. The first decision can be as concrete as choosing which piece of equipment to look into, and what information maintenance needs to do it.