Voltage Monitoring System 2026 | Stop Losing Lines to Voltage Sags

Most plants that start looking at a voltage monitoring system do so the morning after a report like this one. “Around 3 pm yesterday three lines stopped at the same time. There was no power outage.” The utility has no record of anything. The machine builders reply that their equipment is working as designed. Nobody can prove whose side the problem is on, and the same thing happens again the following month. This article sets out a design sequence for turning that situation into numbers you can actually work with.

A voltage monitoring system is not a power monitoring system

The first thing worth separating is that these two are both “systems that measure electricity”, yet they differ in purpose, in the quantity they measure, and in where they are installed. If you go out for quotations while the two are still mixed together, what comes back is an expensive proposal with overlapping functions.

A power monitoring system measures energy and demand in order to bring the electricity bill down. A measurement interval of 15 or 30 minutes is enough. Its job is to convert what is happening into money. We have already covered that side in Factory Power Monitoring System — Cut Thai Electricity Cost, including the cost breakdown and specific energy consumption management, so if the electricity bill is what you want to settle first, start there.

What a voltage monitoring system measures is the voltage waveform itself, and the events in which that waveform leaves its normal range. Measurement happens on a half-cycle basis, which on Thailand’s 50 Hz system means every 10 milliseconds. The goal is not a lower bill but a smaller loss from stopped production. It is worth noting that the same phrase “voltage monitoring” is sometimes used for watching voltage trends on a 1 to 15 minute interval. We separate those two uses from a sensor selection standpoint in Factory IoT Sensor Types 2026 – Sampling Rate Decides, Not Accuracy. This article deals with the first meaning, voltage monitoring for catching sags.

The last row of that table is the biggest practical difference. Power monitoring starts from “it’s expensive, so let’s measure”. Voltage monitoring starts from “it stops, so let’s measure”. Because the motive is different, the alternatives your capital request gets compared against are different too.

What stops your line is not an outage, it is a voltage sag

When Japanese expatriate managers think about what stops a plant, a blackout is the first thing that comes to mind. In reality, what stops production far more often is not an outage but a momentary voltage drop, commonly called a sag.

A sag is a short-duration dip in which voltage falls below 90 percent of nominal without ever reaching zero. Under the IEEE 1159 classification the duration runs from half a cycle to one minute, and the event is expressed as the percentage of voltage that remains. According to Power Quality Bulletin No. 3, a technical document from the North American distribution utility Pacific Gas and Electric, a typical sag lasts 3 to 10 cycles, which is 50 to 167 milliseconds. To the human eye it looks like nothing more than a brief flicker of the lights.

The problem is how often it happens. Another technical document from the same utility, Short Duration Voltage Sags can Cause Disruptions, states that while an urban customer experiences one to two outages a year, that same customer experiences more than 20 sags a year. The reason the numbers do not match lies in the structure of the distribution system. When a ground fault occurs on one feeder out of a substation, every parallel feeder on the same bus sees its voltage collapse until the breaker clears the faulted circuit. The fault itself happened on somebody else’s circuit, but the voltage dip still arrives at your plant.

Power Quality Bulletin No. 3 also cites, as a result of the Distribution Power Quality (DPQ) project run by the Electric Power Research Institute (EPRI), a figure of 12 utility-caused sags per year on average for a typical customer. On top of that, Short Duration Voltage Sags can Cause Disruptions notes that on systems fitted with automatic reclosing breakers or reclosers, a single fault can deliver three sags in succession. From the equipment’s point of view, it gets hit three times in quick succession.

So the picture is this. While everyone is discussing standby generators for the one or two outages a year, more than 20 sags a year are quietly eating into production without anybody recording them. That is the starting point for a voltage monitoring system.

Why equipment stops when 70 percent of the voltage is still there

“The power did not actually go off, so why did the machine stop?” This question comes up on the shop floor and at head office without fail. The answer is that what stopped is not the machine itself but the components that control it.

Contactors and relays hold their contacts closed through coil excitation. If the voltage drops for a few tens of milliseconds, the holding force is lost and the contacts open. Once they open, power to the motor is cut, and the contactor will not re-close by itself when the voltage returns. Short Duration Voltage Sags can Cause Disruptions states plainly that relays and contactors inside motor starters are sensitive to sags, and that when they drop out the process stops.

Inverters and servo amplifiers monitor their internal DC link voltage and issue an undervoltage trip when it falls below the set point. As protection that is correct behaviour, but on equipment with servo axes a trip means the axes have to be re-homed, and recovery takes time.

Control PLCs and instrumentation are usually fed from a single-phase control supply, so they fall over before the three-phase power side does. The same document points out that modern equipment such as process controllers, programmable logic controllers, variable speed drives and robots has become more sensitive to sags as it has become more complex, not less.

The important point here is that the same sag does not knock out every machine in the plant at once. The result depends on which of the three phases dipped, which phases the equipment is connected across, and how much headroom its power supply unit has. That is why you get the apparently inexplicable outcome where three machines in Building A stop while Building B is untouched. The only thing that can explain that asymmetry is an instrument recording voltage phase by phase.

SEMI F47 draws the line for how deep and how long equipment must ride through

There is an agreed industry benchmark for what equipment should be able to withstand. It is SEMI F47, written for semiconductor manufacturing equipment. It is widely used outside semiconductors as well, as a common language you can write into an equipment procurement specification.

SEMI F47 defines the sags equipment must ride through as pairs of retained voltage percentage and duration. The values below are from the summary table published in Power Quality Bulletin No. 3. Because this is written for Thailand’s 50 Hz system, the cycle counts shown are the 50 Hz column.

You read the table like this. For a sag that falls to 70 percent of nominal and lasts 0.4 seconds, the equipment must keep running rather than stop. Put the other way round, equipment that stops under those conditions does not meet SEMI F47.

The same kind of requirement exists on the international standards side. IEC 61000-4-11 covers equipment with a rated input current of 16 A or less per phase, and IEC 61000-4-34 covers equipment above 16 A, both defining test methods for immunity to voltage dips, short interruptions and voltage variations. Large factory equipment falls into the second category. Amendment 2 to that standard was issued on 7 August 2025, so requirements in this field are still being updated.

In practice, the place to use this is not the modification of existing equipment but the purchase specification for new equipment. Writing a single line such as “shall comply with IEC 61000-4-34” or “shall meet the SEMI F47 requirement curve” into the specification changes your stoppage count five years later. Making existing equipment tougher after the fact costs an order of magnitude more than writing it into the order.

What a voltage monitoring system measures — IEC 61000-4-30 as the common language

Recording events is of little use if every instrument measures them differently, because then you cannot compare or negotiate. The standard that aligns this is IEC 61000-4-30.

The standard defines measurement methods and the interpretation of results for power quality parameters on AC systems. It covers supply voltage dips and swells, voltage interruptions, and also frequency, flicker, harmonics and unbalance. There are two measurement classes, A for “advanced” and S for “surveys”. Class A requires dips, swells and interruptions to be measured over a one-cycle window, refreshed every half cycle.

The fourth edition was published on 24 October 2025, replacing the third edition of 2015, and an amendment followed in July 2026. In other words, the common language in this field was updated only within the past year. Existing instruments built to the third edition do not immediately become a problem, but if you are procuring now it is worth confirming which edition an instrument complies with.

Situations that genuinely require Class A are limited, but where they arise nothing else will do. If you are formally telling the utility that “an event originating on your system stopped our production”, or if a customer asks you to explain a delivery delay, a Class A compliant record functions as evidence to a third party. If all you are doing is isolating causes internally, Class S is sufficient. One Class A instrument at the point of supply with Class S instruments on the internal branches is the realistic balance between cost and effect.

Where to put the measurement points — the point of supply alone cannot separate causes

If your only voltage measurement point is the point of supply, you will know that a sag arrived but not what that sag stopped. If your only points are next to the equipment, you will know the voltage dropped but not whether it came from upstream or was generated inside your own plant. You need both.

Cross-referencing the Level 1 and Level 3 records is what lets you separate causes. If the voltage dropped at the point of supply and also at the equipment, it is a grid-side event. If the point of supply was normal and only the equipment side dropped, the cause is your own wiring, transformer capacity, or a large motor starting nearby. Complaining to the utility without being able to make that distinction gets you nowhere.

The classic in-plant cause is the starting of a large motor or a compressor. These produce a gradual voltage drop lasting more than 0.5 seconds, so the waveform tells them apart from a sag caused by a grid fault. A grid fault clears in tens of milliseconds and the voltage snaps back, whereas a motor start sags gently and recovers gently.

Without an event record, the discussion about countermeasures never begins

What is happening in most plants is not a technical problem but a record-keeping problem. The fact that the line stopped is written in the daily report, but nobody has recorded what the voltage was doing at that moment. Asking the utility the next day does not help either, because the granularity of grid-side records does not match what the plant needs.

Three things are needed to get out of this state. First, an event record on the voltage side. Second, a stoppage record on the equipment side. Third, time synchronisation so the two can be lined up against each other.

The third is missing in a very large number of plants. The voltage instrument runs on its own clock, the PLC on its own clock, and the production management system on yet another server’s clock, each drifting by tens of seconds. When a sag lasts 0.2 seconds, a discrepancy of tens of seconds makes causality unreadable. NTP time synchronisation across every device is a precondition. How to capture the stoppage record on the equipment side, including the design of tag counts and polling intervals, is covered in PLC Data Collection 2026 – Tag Count and Polling Interval Decide It, Not the Protocol.

There is one more point. A record that never reaches a person is worthless. If the shop floor knows that a sag has just occurred, the first minutes of recovery look different. We have set out how to design who gets notified through which route in Equipment Alert Notification System 2026 | Design for MTTR.

Separately, power quality for information systems such as servers and network equipment needs a different design from production equipment. A sag shallow enough for production equipment to ride through can still reboot a server. That area is handled in Factory System Monitoring 2026 | Detection Time Drives the Loss, so keep those countermeasures separate from the ones discussed here.

How to count what you are losing

To make an investment decision you need to know how much you lose per sag event. A useful reference here is a field survey of 33 small and medium-sized plants in Brazil.

A 2021 paper by Motoki and co-authors in the journal Energies visited 33 small and medium enterprises across 12 industry sectors, connected to 11.9 kV and 13.8 kV medium-voltage systems, and used a questionnaire to compile the losses caused by voltage sags and short interruptions. The average headcount was 349 employees, so the survey population is close in scale to a Japanese-owned plant in Thailand. Note that the figures are in US dollars for Brazil as of 2021, so the right way to use them is to read the differences between sectors and the structure behind them, rather than the absolute level.

The headline result was an average loss of 7,364.75 US dollars per event, with a median of 4,733 US dollars. Per kilowatt of interrupted load it came to 6.72 US dollars per kilowatt, with most companies clustered in the range of 2 to 6 US dollars per kilowatt.

The differences between sectors are clear as well.

What deserves attention in this table is not the size of the amounts but a second finding from the same paper. The correlation coefficient between contracted demand and loss per event was only 0.221. In other words, a bigger plant does not necessarily lose more. The paper suggests that what determines the loss is not the scale of the installation but the added value of the product. It also states that company size is probably one of the variables influencing the loss amount, so it does not claim that scale is irrelevant.

This has a direct bearing on how you write a capital request. The judgement “our contracted demand is small, so we do not need countermeasures” is not supported by this survey. A plant making high-unit-price products, with processes that are laborious to restart, will show large losses regardless of its size.

Putting a number on annual loss with a model case

As an example you can substitute your own figures into, here is a build-up for a hypothetical automotive parts plant in Chonburi, Thailand. The following is our own estimate, not measured data. US dollar conversions use 32 baht to the dollar.

The assumptions are a contracted demand of 1,500 kilowatts, three lines from machining through assembly on the same system, 20 sags a year caught by voltage monitoring, of which 8 a year actually trip equipment and stop production, and an average recovery time of 75 minutes per event. Recovery time includes restart, re-homing and first-article verification.

That 251,260 baht converts to roughly 7,850 US dollars, which is almost identical to the 7,974.75 dollar automotive sector average in the survey above. It is corroboration that the assumptions in the build-up are not extreme.

At 8 events a year, the annual loss is 2,010,080 baht, roughly 62,800 US dollars. That is the figure against which the investment gets compared.

Countermeasures come in three layers — measure, ride through, compensate

Ask for quotations on sag countermeasures and you may be offered a multi-million-baht compensation device straight away. But countermeasures have a clear layered structure, and skipping the lower layers to reach the upper one makes the cost jump.

Layer 2 looks unglamorous but it is where the effect is. Power Quality Bulletin No. 3, cited above, lists nine ways to raise voltage sag immunity yourself. Choose a power supply whose voltage setting range places the nominal towards the upper end. Connect single-phase supplies phase to phase to gain headroom. Move load off the power supply unit onto another system. Increase the capacity of the power supply unit itself so its load factor drops. Replace single-phase supplies with three-phase supplies. Feed from a DC bus. Review the trip thresholds and time delays on protective relays. Use contactors with greater mechanical mass, or hold-in accessories. And, as a last resort, install a voltage regulator. The document positions the ninth item explicitly as a last resort.

One thing about Layer 1 should be stated plainly. Layer 1 on its own does not reduce the loss by a single baht. Fitting instruments does not stop machines stopping. You therefore cannot calculate a payback period for Layer 1 alone. There are still two reasons to put it first. One is that you cannot decide where to apply Layer 2 without knowing which machines drop out at which sag depth. The other is that whether a Layer 3 compensation device is worth installing depends on the distribution of event depth and duration. Booking Layer 1 as the design cost for Layer 2 matches what actually happens.

Comparing the payback

Here is a comparison between installing Layers 1 and 2 together and adding Layer 3 on top. We assume Layer 2 reduces the 8 stops a year to 3. With Layer 3 added, we assume only the extremely deep sags remain and the count falls to 0.5 a year.

There is one thing to take from this table. Adding Layer 3 multiplies the annual saving by 1.5, but it multiplies the initial investment by 3.6. As a result the payback stretches from 1.6 years to 4.3 years.

For most plants, therefore, the right answer is to stop at Layer 2. A Layer 3 sag compensation device is justified only where a single stoppage costs an order of magnitude more than usual. Continuous furnaces, plating baths, continuous resin extrusion and certain cleanroom processes qualify. Conversely, a plant built around assembly and machining that starts its evaluation at Layer 3 will almost certainly never recover the money.

What happens when you move the assumptions

The estimate above leans heavily on three assumptions. Let us move each of them.

Assumption 1 is the stoppage count. Suppose it is 4 a year rather than 8. Plants in industrial estates with a stable grid, or plants supplied by two routes from separate substations, will be closer to this. Note that drawing two circuits from the same bus at the same substation barely reduces sags at all, because a voltage dip on the bus reaches both circuits equally. What that arrangement reduces is outages, not sags, and the distinction matters here too. If the same reduction rates apply, 4 a year becomes 1.5 a year with Layer 2 and 0.25 a year with Layer 3 added. In that case the payback for Layers 1 and 2 is 3.4 years, and adding Layer 3 gives 10.9 years. Layer 3 simply does not stand up.

Assumption 2 is the recovery time. Suppose it is 40 minutes rather than 75. That describes a plant with standardised restart procedures and proper written work instructions for re-homing. The point to watch here is that shortening recovery time only reduces lost production opportunity and attendance labour. The scrapped work in progress and the equipment damage allocation do not fall, because the work in progress is determined at the instant of the stop and contactor wear is determined by how many sags the plant takes. Under this condition, lost production opportunity is 3 lines x 40 minutes x 900 baht per minute, or 108,000 baht, and attendance labour is 12 people for 40 minutes each, which is 8 person-hours at 320 baht per hour, or 2,560 baht, while the 28,000 baht of scrap and the 15,000 baht equipment damage allocation remain unchanged. Loss per event becomes 153,560 baht, and the payback for Layers 1 and 2 stretches to 2.7 years.

Assumption 3 is that Layer 2 works less well than expected and the 8 stops a year only fall to 5. This is what happens in plants with old equipment where there is little room to modify the control panels. The payback for Layers 1 and 2 is then 2.8 years. It still pays back, but in this situation it is cheaper overall to replace equipment with IEC 61000-4-34 compliant units as each one reaches its renewal point, rather than to invest further in Layer 2. For the general approach of leaving equipment in place and adding capability externally, the thinking in Legacy Equipment IoT – 5 Cost Layers of a Thai Plant Retrofit is a useful reference.

Of the three assumptions, the one you are least certain about in your own plant is Assumption 1. That is exactly why the sequence is to put the Layer 1 measurement in first, measure the actual count for six months, and only then decide on Layers 2 and 3.

Extra decisions to make for a plant in Thailand

When you design voltage monitoring in Thailand, there are several points where assumptions carried over from Japan do not hold.

Lightning in the rainy season is one of the main factors. Through the rainy season, broadly May to October, direct and induced lightning strikes on overhead distribution lines cause single-phase ground faults. Short Duration Voltage Sags can Cause Disruptions, cited above, states that the majority of power system faults are single-phase ground faults, and lists weather conditions, insulator contamination, animal contact, equipment failure and vehicle accidents as causes. In Thailand, the weather-related share of those is concentrated in the rainy season. So if you are going to start measuring, the measurement period has to span the rainy season, or you will not see the true annual picture. Measuring for three months of dry season and concluding “we do not get many sags” is the classic failure.

You also need to understand the supply side arrangements. Inside industrial estates, supply comes from PEA or MEA, and most plants take it at 22 kV or 33 kV. On tariffs, the Energy Regulatory Commission set the average rate for May to August 2026 at 3.95 baht per unit, of which the fuel adjustment charge was 16.23 satang and the base tariff 3.78 baht. That was an increase from 3.88 baht in the preceding period. Subsequently, according to reporting on 12 August 2026, the public lighting charge was removed from September onwards, bringing the average rate to about 3.89 baht and the base tariff to about 3.72 baht. Moving from 3.88 baht to 3.95 baht and then to about 3.89 baht, the level is essentially flat. Industrial supply carries separate demand charges and power factor charges so a simple comparison is not possible, but in a period where electricity costs are flat, an investment justified on the grounds of energy cost reduction becomes hard to get approved. Put the other way round, if you want to get a power-related investment through right now, stoppage loss is a far more persuasive basis than the electricity bill.

On the operational side, it matters to agree the decision criteria for night shifts and holidays when no expatriate manager is present. Is it acceptable to restart equipment immediately after a sag, or does a quality check come first? If that judgement is left to whoever is on shift, the spread in recovery time widens and the assumptions in the estimate above fall apart.

One brief word on comparison with neighbouring countries. In Vietnam, the nature of scheduled outages and grid instability differs from Thailand, and carrying the same specification across can leave you with something that does not work. We have set that out in Vietnam Equipment Monitoring 2026 — Why the Thailand Playbook Breaks Down.

How to roll it out

Working in 90-day blocks is realistic. Here is a guide to what belongs in each period.

In the first 30 days, dig out the stoppage records for the past 12 months and count how many have a cause recorded as “unknown” or “power abnormality”. At the same time, decide where the instruments at the point of supply and the mains will go, and confirm with your licensed electrical engineer whether CTs and VTs can be fitted. Build the calculation basis for loss per event during this stage.

In the second 30 days, install the instruments and put time synchronisation into every device. The goal is to reach a state where PLC stoppage logs and instrument event logs can be shown side by side on one screen. It is important not to rush into countermeasures here.

In the third 30 days, classify the events that have accumulated. Tabulate whether they originated on the grid or on site, how depth and duration are distributed, and at what percentage each machine drops out. Only once that table exists can you decide where Layer 2 goes and whether Layer 3 is needed at all.

After that, run measurement over six months including the rainy season, fix the actual annual count, and then make the Layer 3 decision. There is no need to hurry. If it is only the Layer 2 work, you can start at this point.

Frequently asked questions

What is a voltage monitoring system

It is a system that continuously measures RMS voltage on a half-cycle basis and records events outside the nominal range as pairs of retained voltage and duration. Its purpose and its measurement interval both differ from a power monitoring system, which measures energy. The common standard for the measurement method is IEC 61000-4-30, which has two accuracy levels, Class A and Class S.

How much does a voltage monitoring system cost

The model case in this article assumed one Class A instrument at the point of supply, three Class S instruments on internal branches, instrument transformers and panel work, plus logging and time synchronisation, for an initial 1,065,000 baht and 96,000 baht a year to run. It varies with the size of the plant and how much room there is to modify existing panels, but if measurement alone is the goal, configurations starting in the low hundreds of thousands of baht are possible.

Is a UPS enough as a countermeasure against voltage sags

On the information systems side, a UPS is effective. On the production equipment side, covering the whole load including motive power with a UPS is not realistic in terms of capacity or cost. The sequence that fits the cost-benefit picture is to first do the Layer 2 work of raising the immunity of control circuits and contactors, and then consider a sag compensation device only for the high-value processes that still stop.

If we already have a power monitoring system, do we still need voltage monitoring

Yes, you still need it. Power monitoring systems typically measure on a 15 to 30 minute interval, and a 0.2 second sag is averaged away into invisibility. You need either to add a voltage-side instrument in the same panel, or to confirm whether your power monitoring hardware supports half-cycle recording.

How far should monitoring of the incoming substation go

The point of supply alone only tells you that a sag arrived. Only when you also place a measurement point at the panel next to the equipment that stops, and cross-reference the two with synchronised time, can you separate grid-origin from site-origin events. There is no need to fit every machine. Narrowing it down to the equipment with a history of stoppages is enough.

Can we claim compensation from the utility

There is no scheme that compensates automatically. However, with a Class A compliant record you have the material to demonstrate that the event originated on the grid side, and then to open a discussion about reviewing the supply route or protection coordination. The practical reality is that without records, an approach to the utility does not start a conversation at all.

Summary

A voltage monitoring system is not a mechanism for lowering the electricity bill. It is a mechanism for making unexplained stoppages countable. The gap in frequency between one or two outages a year and more than 20 sags a year is the starting point for this investment.

Equipment stops not because the power goes off, but because contactors drop out and inverters issue undervoltage trips. The level equipment should ride through is already defined, in the form of SEMI F47 and IEC 61000-4-34, and it takes one line in the purchase specification for new equipment. For existing equipment, holding to the three-layer sequence of measure, ride through, and compensate only the processes where that is still not enough, is what determines the cost.

In the model case, payback with Layer 2 was 1.6 years and 4.3 years with Layer 3 added, because the saving only rises by a factor of 1.5 while the investment rises by a factor of 3.6. For most plants the answer is to stop at Layer 2, and Layer 3 is justified only where you run continuous processes in which a single stoppage costs an order of magnitude more.

And the first step needed to make that judgement is real measurement over a period that spans the rainy season. How many times a year, at what depth, and which machines dropped out. Without that table, comparing quotations only lets you discuss which number is bigger.

You do not need a settled internal position on voltage monitoring or sag countermeasures to talk to us. We are happy to start by counting together how many of your stoppage records from the past year are marked “cause unknown”, and to work through where measurement points should go and where Layer 2 effort should be concentrated, in the order the work actually happens. An equipment list and a single-line diagram of your incoming supply let us get more specific. Please get in touch through our contact page.

References

• IEC 61000-4-30:2025 Electromagnetic compatibility (EMC) – Part 4-30 Power quality measurement methods, IEC – Source for the fourth edition publication date, the Class A and Class S distinction, and the parameters covered
• Power Quality Bulletin No. 3 Voltage Sag Immunity Standards – SEMI F47 and F42, Pacific Gas and Electric Company, July 2018 – Source for the SEMI F47 ride-through table, the EPRI DPQ figure of 12 events a year, and the nine ways to raise immunity
• Short Duration Voltage Sags can Cause Disruptions, Pacific Gas and Electric Company – Source for the comparison of one to two outages a year against more than 20 sags a year, consecutive sags from reclosing, and the description of equipment sensitivity
• Motoki et al., Cost of Industrial Process Shutdowns Due to Voltage Sag and Short Interruption, Energies 2021, 14, 2874, MDPI – Source for the 33-company survey average of 7,364.75 US dollars per event, 6.72 US dollars per kilowatt, the losses by sector, and the correlation coefficient of 0.221 with contracted demand
• IEC 61000-4-34:2005 EMC – Part 4-34 Voltage dips, short interruptions and voltage variations immunity tests for equipment with mains current more than 16 A per phase, IEC – Scope of the immunity test standard applying to equipment above 16 A
• IEC 61000-4-34:2005/AMD2:2025, IEC – Amendment 2, issued 7 August 2025
• IEC 61000-4-11:2020 EMC – Part 4-11 Voltage dips, short interruptions and voltage variations immunity tests for equipment with input current up to 16 A per phase, IEC – Scope boundary against equipment of 16 A or less
• ERC sets power tariff at 3.95 baht per unit for May-August, Nation Thailand, 1 April 2026 – Figures for the average tariff, fuel adjustment charge and base tariff for May to August 2026
• Thailand cuts September power tariff, caps first 200 units at 3 baht, Nation Thailand, 12 August 2026 – Change to an average of 3.89 baht and a base tariff of about 3.72 baht from September 2026
• SEMI F47 – Specification for Semiconductor Processing Equipment Voltage Sag Immunity, SEMI – Confirmation of the standard title and scope of SEMI F47
• IEEE 1159-2019 – IEEE Recommended Practice for Monitoring Electric Power Quality, IEEE – Source for the classification of sags, with durations from half a cycle to one minute expressed as retained voltage percentage