Factory electrical energy monitoring: what to measure and how to read it

What factory electrical energy monitoring means

Electrical energy monitoring is the periodic or continuous measurement of consumption and electrical parameters at selected points in a factory’s power network, followed by processing the results into information that can support decisions. There may be only one measuring point, the utility connection, or dozens, reaching down to the panel of every large machine.

Monitoring is different from simply reading a meter. Reading a meter produces a number, while monitoring connects that number to its context: operating hours, shifts, production counts, ambient temperature, and maintenance work. Without context, consumption that rose this month could mean waste, or it could mean that orders are strong.

In an energy management system under ISO 50001, monitoring, measurement, and analysis of energy performance are a core part of the continual improvement cycle. The organization sets indicators and baselines, plans which data to collect, and then evaluates its performance at regular intervals^[1]. A plant with no intention of pursuing certification can still use the same logic.

Why one main meter and a monthly bill are not enough

Almost every factory already has one energy measuring point: the utility’s kWh meter at the service connection. Its data is complete for billing, but limited for running the plant. It has three limitations.

First, spatial resolution. The main meter adds up every load. A leaking compressor, a chiller whose setpoint was lowered, and an injection machine left heating its barrel all weekend show up as the same single number. The increase in consumption is visible, but its source is not.

Second, time resolution. The monthly bill arrives after the energy has been used. A problem that began in the first week only becomes visible several weeks later, by which time nobody remembers what changed. Data at short intervals, hourly or finer, lets a change be tied to events on the shop floor on the same day.

Third, no production context. The bill does not know how many units were made. The month with the highest consumption may well be the most efficient one, if production rose faster than consumption. This is discussed further in the section on energy performance indicators.

The parameters to measure and what they mean

A modern digital power meter measures dozens of quantities, but not all of them need to be watched every day. The table below summarizes the parameters most often used for decisions in a factory, before each is discussed in more detail.

Energy, power, and demand

Energy (kWh) is the quantity that is billed and that energy performance indicators are calculated from. Power (kW) is the rate at which it is used. A 50 kW machine running for two hours uses 100 kWh. A 100 kW machine running for one hour uses the same energy, but loads the network twice as heavily during that hour.

Demand is the average active power over a set interval, whose length is fixed by the metering system or the tariff rules. Demand matters because transformers, cables, and contracted capacity must be able to serve the peak, not the average. Two plants with the same monthly consumption can have very different peak demand, depending on whether their large machines are started together or in turn.

Voltage and current

Voltage shows supply quality. Voltage that is too low makes a motor draw more current for the same load, so the motor runs hotter. Voltage imbalance between phases, even a small one, also adds heat in three-phase motors. Current shows loading: how close a feeder is to the limits of its cable and breaker, and whether the load is shared evenly across the three phases. On a motor, current that creeps upward at the same production load can be one of the early signs of a mechanical problem, such as deteriorating lubrication.

Power factor and reactive power

Induction motors and transformers need reactive power to build their magnetic fields. This power flows back and forth between source and load without doing work, but its current still flows through cables and transformers. Power factor is the ratio of active power to apparent power. The lower the power factor, the more current is needed to deliver the same active power.

IEEE 1459 distinguishes total power factor from the power factor of the fundamental component^[2]. The distinction matters in plants with many power electronic loads. A capacitor bank can correct the power factor of the fundamental component, but it does not correct the part caused by harmonics.

Harmonics in brief

Variable frequency drives (VFDs), rectifiers, switching power supplies, and welding machines draw current that is not sinusoidal. Such current can be broken down into a fundamental component and harmonic components at multiples of its frequency. Harmonics add heat in transformers and neutral conductors, and they can interact with capacitor banks. For routine energy monitoring it is usually enough to watch the total harmonic distortion (THD) of current and voltage at the main panel, and to carry out a more detailed power quality survey when there are symptoms, such as capacitor banks that fail repeatedly or breakers that trip for no clear reason.

One more thing needs attention before reading any number: the capability of the meter itself. IEC 61557-12 sets performance requirements for power metering and monitoring devices (PMDs), including a performance class for each of their measuring functions^[3]. A meter that is accurate for active energy does not necessarily perform as well for reactive energy or harmonics, so its specification needs to match the purpose it will serve.

A sub-metering hierarchy: from the incomer to the machine

Sub-metering means installing meters below the main meter so that consumption can be broken down by area, system, and equipment. The arrangement should follow the plant’s single line diagram, from top to bottom, so that each level can be reconciled with the sum of the meters on the level below it.

A common question is how deep sub-metering needs to go. ISO 50001 uses the concept of significant energy use (SEU): energy use that accounts for a substantial share of consumption or offers considerable potential for improvement^[1]. SEUs are the first candidates for separate metering. In many plants the list includes air compressors, cooling systems, machines with large heaters such as injection molding machines and ovens, and large pumps and blowers.

A tiered arrangement also brings a benefit that is often overlooked: the balance check. The consumption of the meters on one level should add up to close to that of the meter above them. A large gap, or one that suddenly changes, usually points to one of three things: a load that is not metered, a meter that is set up wrongly, or a meter that has stopped sending data.

Energy performance indicators and baselines

Absolute consumption is a number that is easy to collect and easy to misread. Consumption can rise because production rose, the weather was hotter, or the product mix changed, without anything getting worse. ISO 50006 provides a framework for measuring energy performance with two tools: the energy performance indicator (EnPI) and the energy baseline (EnB)^[4].

An EnPI is the measure of energy performance that the organization chooses. ISO 50006 describes several forms, from the simplest measured energy value, through a ratio of measured values, to statistical or engineering models that account for several variables at once^[4]. The ratio form most widely used in factories is specific energy consumption (SEC).

An energy baseline is a quantitative reference from a defined period, against which performance in later periods is compared. ISO 50006 stresses two kinds of factors that should be recorded with a baseline. Relevant variables are factors that change routinely and affect consumption, such as production volume or ambient temperature. Static factors are conditions that normally stay fixed but would affect consumption if they changed, such as the number of shifts or the type of product^[4]. A baseline should be normalized for relevant variables and revisited when its static factors change.

The SEC ratio has one weakness worth understanding. Almost every line has a base load that keeps running whatever the production volume: lighting, cooling, heaters holding temperature, compressors topping up leaks. Because the base load is spread over more units when production rises, SEC falls on its own even when nothing has been improved. This is why a simple regression model of energy against production is often more honest than a ratio.

Reading the load profile

A load profile is a chart of power against time, usually in 15-minute or hourly intervals, over a day or a week. It is often the source of the first findings, because a plant’s energy use follows its working pattern, and departures from that pattern are easy to see.

Start with consumption outside production hours. Look at power during a night with no shift, on a Sunday, or on a holiday. This figure is the plant’s base load. Part of it is genuinely needed, such as warehouse cooling or security systems. The rest is habit: compressors left running to top up leaks, production area lights left on, or injection machines whose barrels are kept hot because starting from cold is seen as a nuisance. Comparing the base load of one weekend with the next is one of the easiest checks there is.

Next, look at the peaks. When do they occur, and which loads form them? Peaks often appear at the start of a shift, when many machines are switched on at almost the same time and heaters run flat out to reach operating temperature. Staggering the start-up order between machines does not reduce the energy used, but it can lower peak demand.

Finally, compare the shape of the profile across comparable days. Two working days with similar production should have similar profiles. If one day shows a higher floor all day long, or a spike at an unusual hour, that is where the investigation starts: who was working, which machines were running, and what was different.

Industrial electricity tariffs in Indonesia

How energy data should be read also depends on how electricity is billed. For PLN industrial customers in certain tariff classes, energy use is recorded separately for peak hours (waktu beban puncak, WBP) and off-peak hours (luar waktu beban puncak, LWBP), at different prices. As a result, the same number of kWh can cost different amounts depending on when it is used. Reading the load profile alongside the WBP hours helps judge whether any load can be shifted, such as filling water tanks or batch processes that are not tied to a shipping schedule.

A PLN bill can also include a charge for excess reactive energy (kVArh) when the customer’s average power factor falls below a set limit. This is where measuring power factor per panel pays off: it shows which loads draw the most reactive power and whether the installed capacitor banks are still working as they should.

Verifying savings against a baseline

Once an improvement has been made, such as fixing compressed air leaks or changing heater settings, the next question is how much energy was actually saved. Comparing this month’s bill with last month’s is almost always misleading, because production, weather, and product mix change as well.

The International Performance Measurement and Verification Protocol (IPMVP) provides a widely used framework for answering this question. Savings cannot be measured directly, because they are energy that was not used. What can be done is to compare the energy measured in the reporting period with the energy that would have been used under the same conditions had the improvement not been made^[5].

IPMVP also calls for a clear measurement boundary: whether savings are measured on the improved equipment alone, or on the meter for the whole facility. The protocol describes several options for this, from isolating a single piece of equipment to measuring the whole facility^[5]. This is where sub-metering earns its place. An improvement on one line is hard to prove from the main meter, because a small change is lost in the variation of the whole plant.

A worked example

The figures below are illustrative, not data from any particular plant. A line has a baseline model from a regression on monthly data, E = 10,000 + 2.0 × P, with E in kWh per month and P in units per month. After an improvement to its heating system, the line is measured again for one month.

1. 1Comparing bills: 54,000 − 50,000 = 4,000 kWh more. Taken at face value, the line appears to have gotten worse.
2. 2Comparing SEC: (2.50 − 2.25) × 24,000 = 6,000 kWh. This overstates the savings, because part of the drop in SEC comes only from the base load being spread over more units.
3. 3Comparing with the adjusted baseline: 58,000 − 54,000 = 4,000 kWh. This is the honest savings estimate for the month: the energy that would have been used at a volume of 24,000 units without the improvement, minus the energy actually used.

The 2,000 kWh gap between the second and third methods is the base load effect. At a volume of 20,000 units, the 10,000 kWh base load contributes 0.50 kWh per unit; at 24,000 units, its contribution drops to about 0.42 kWh per unit. That drop happens without any improvement at all. The example also shows why an estimate from a single month should be confirmed with several months of data before it is used as the basis for the next investment decision.

Steps to start energy monitoring

1. 1Collect the bills and the single line diagram. Twelve months of bills show seasonal patterns, the cost structure, and the power factor record. The single line diagram shows where meters can be installed.
2. 2Identify the significant energy uses. Estimate each large system’s share of consumption from nameplate data, operating hours, and spot measurements. Start with the few largest.
3. 3Install meters from the top down. The incomer and main panel first, then the utilities and lines that count as significant energy uses. Choose meters whose performance matches the purpose they will serve.
4. 4Verify every measuring point at commissioning. Check the CT ratio set in the meter, the orientation of each CT, and that the voltage and current inputs are on matching phases. Compare readings with a clamp meter and with the meter on the level above.
5. 5Link to production data. Energy per line without production counts is only half the story. Set the same output unit and period for both.
6. 6Set EnPIs and baselines. Choose an indicator for each significant energy use, collect baseline period data, and record its relevant variables and static factors.
7. 7Assign owners and a review schedule. Every important meter needs a person responsible for reading its data, and every week there should be time to discuss deviations with the production and maintenance teams.

Common mistakes in energy monitoring

• Monitoring without owners. A dashboard that nobody owns will rarely be opened. An energy finding only matters when someone has the authority to decide what to do about it, such as a line supervisor or a utilities engineer.
• No normalization for production volume. Rising consumption is read as waste and falling consumption as success, when both may simply follow volume. As the example above shows, even a simple ratio can mislead when the base load is large.
• CT ratios left unchecked. A meter configured with a CT ratio different from the CT actually installed reads wrong by a fixed factor, so its data looks plausible and the error goes unnoticed. A CT installed backward or on the wrong phase can produce negative power or an odd power factor on one phase.
• Comparing periods with different static factors. A month with a new machine added, a change in the number of shifts, or a new product cannot be compared directly with an old baseline without adjustment.
• Data completeness left unchecked. A meter that stops sending data for a few days makes monthly consumption appear to fall. Balance checks between meter levels and monitoring for missing data should be routine.

Conclusion

Useful electrical energy monitoring is defined not by the number of meters or parameters, but by the questions it can answer. The main meter and the monthly bill answer how much. Sub-metering answers where. The load profile answers when. Performance indicators normalized for production answer whether performance is improving.

Plants that get value from energy data usually start with a few measuring points on their largest energy uses, make sure the readings are right from commissioning, set an honest baseline, and name someone to read the data every week. Savings, where there are any, are verified against an adjusted baseline, not inferred from a lower bill.