Factory Solar Monitoring 2026 — 4 Blind Spots That Hide Falling PV Output

Very few factories that have put 500 kWp on the roof ever go back and check whether the 3,150,000 THB a year they assumed in the business case is actually arriving. A self-consumption system sells nothing, so the benefit only ever appears as electricity you did not have to buy, and that reduction disappears into Ft revisions and swings in production volume. This article rebuilds factory solar monitoring as what it really is, which is the job of putting the generation side, the consumption side and the billing side onto one shared timeline.

Self-consumption solar hides its own decline inside the electricity bill

An export system gets a meter reading every month. There is a buyer, there is an agreed rate per kWh, and money lands in the account. Generation is therefore reconciled against a third party every single month, and if the numbers drift, accounting notices.

Self-consumption has none of that machinery. The electricity produced on the factory roof is consumed inside the factory and that is the end of it. Nothing is exchanged with anyone outside, so no meter reading, no invoice and no line-item statement ever arrives. The benefit shows up only as less electricity purchased from PEA or MEA.

The trouble is that purchased electricity moves every month for reasons that have nothing to do with solar. It rises when production rises. It rises when you go from two shifts to three. It rises in the hot dry months when chillers and air conditioning run harder. It changes again whenever Ft is revised. The size of that noise is as large as the solar contribution, or larger. Which means that as long as you are looking at the bill, you cannot tell whether solar output is down 5% or down 10%.

The state you find in Japanese-owned plants across Thailand is usually this. The monitoring screen bundled with the inverters is genuinely running. There is one display in the substation room or in a corner of the office, showing today’s generation curve. But nobody opens it daily, and even when someone does, the conclusion is always the same, which is that it was cloudy today so of course it is low. There is no reference to compare against, so low cannot be distinguished from normal.

Let that go on for a few years and the plant degrades quietly. Dirt accumulates, one string drops out, and on days when production stops the system throttles itself even though the sun is shining. Each of these is individually small and each is buried in day-to-day variation. Added up over a year, in the model factory used in this article, they reach 26% of the expected saving.

The job of monitoring is to catch that 26% outside the bill, and that takes three things joined together. First, the generation side, meaning how many kWh actually came out and what percentage that is of what should have come out. Second, the consumption side, meaning whether the factory was able to use the electricity that was produced. Third, the billing side, meaning whether the drop in purchased kWh matches the calculation. When those three live on separate screens, you will never find out where the losses are.

The model factory used here — 500 kWp rooftop, fully self-consumed

Without concrete numbers this becomes an abstract argument, so the following model factory is used throughout. It is a Japanese-owned processing and assembly plant in a Thai industrial estate that retrofitted solar onto an existing roof.

The arithmetic is simple. 500 kWp × 1,400 kWh/kWp gives 700,000 kWh per year, and multiplying by 4.5 THB/kWh gives 3,150,000 THB per year. The figure that goes into the capital request is that 3,150,000 THB, and the investment decision is approved on the strength of it.

Two caveats belong with those assumptions. The first concerns the specific yield of 1,400 kWh/kWp per year. That is a design-stage placeholder, and on a real project you should pull PVOUT for the coordinates of your own site. Annual irradiation differs between northern Thailand and the eastern seaboard, and even on one site the result moves with roof tilt and orientation. The solar irradiation and PV potential dataset that the World Bank and ESMAP publish for Thailand is free and can be queried by coordinate.

The second caveat concerns the effective tariff of 4.5 THB/kWh. That is a modelling assumption for this article, built by adding Ft and VAT on top of the daytime TOU energy charge, and it is not a figure taken straight from a published tariff schedule. For reference, the official ranges posted by the Thailand Board of Investment, as of the 2023-06-30 update, are 4.10 to 4.33 THB/kWh for large-scale TOU on-peak and 2.58 to 2.64 THB/kWh for off-peak. On top of that sit a demand charge of 74.14 to 332.71 THB/kW and a power factor surcharge of 56.07 THB/kVAR applied when reactive power exceeds 61.97% of the active power demand. What rooftop solar displaces is essentially the daytime energy charge, so the effective rate settles at roughly the on-peak range plus Ft and VAT. You can derive your own effective rate by taking the daytime kWh and the corresponding baht from the last twelve months of invoices and dividing one by the other.

Managing the demand charge itself is a separate discipline from solar monitoring. How to hold down the 15-minute maximum demand is covered in the guide to factory power monitoring systems, so please refer to that. This article stays inside the narrower question of measuring the electricity you generate yourself.

Four blind spots that pull generation below plan

The reasons output falls short of plan sort into four categories. What they share is that none of them is individually large enough to stand out from daily variation, which is exactly why they are never found by feel.

Each is examined in turn below. The last one, curtailment, is both the largest hole and the one least likely to be noticed.

Blind spot 1 — soiling works slowly, so nobody feels it

Through the Thai dry season, from November to April, dust settles on industrial estate roofs. If there is a foundry or a powder-handling plant nearby, it accumulates faster. Bird droppings cover a small area, but fully shading a single cell drags down the current of the entire string that cell belongs to.

The awkward property of soiling is that it never costs you 20% overnight. It works at a pace of 0.2% in a week and 1% in a month, quietly. Human beings cannot perceive change at that rate, and the people who look at the system every day are the least likely to spot it.

Assume an annual average loss of 4% for the model factory and you get 700,000 kWh × 0.04 = 28,000 kWh, which at 4.5 THB/kWh is 126,000 THB. That is plainly more than the cost of calling in a cleaning contractor once. Even so, most factories schedule cleaning by the calendar, typically twice a year. Run it by the calendar and you routinely get the perverse outcome of washing panels that are clean just after the rainy season and leaving them alone at the end of the dry season when the dust is thickest.

Blind spot 2 — one dead string looks like only 4% of the array

The model factory is built as 25 strings × 20 kWp. When one string stops generating because of a blown fuse, a bad connector or a failed input channel on the inverter, total array output falls by 4%.

4% is far smaller than normal day-to-day weather variation. Mix in some cloud and generation moves 30% without anyone blinking. So monitoring that watches only the inverter total cannot detect a dead string, and this is the important part. It is not that the loss is hard to notice. Polling the aggregate means the information does not exist at all.

Left unattended for six months, the loss is 20 kWp × 1,400 kWh/kWp × 0.5 years = 14,000 kWh, which at 4.5 THB/kWh is 63,000 THB. Most of the time the root cause is one fuse or one connector, and the part costs a few hundred baht. What produced the 63,000 THB loss was not the component. It was the six months of not being able to see.

Blind spot 3 — temperature derating is not a fault, so build it into the expected PR

On a Thai roof, module temperature reaches 55 to 65 degrees Celsius. The maximum power temperature coefficient of a crystalline silicon module is typically around -0.35%/degree C, so a rise of 35 degrees above the 25 degree standard test condition costs a little over 10% of output in principle. Rooftop mounting gives worse rear ventilation than ground mounting, so the effect shows up more strongly.

This is not a failure. It is a physical phenomenon known at design time, and it belongs on the expected-value side of the equation. Evaluate PR without accounting for it and you will misdiagnose a healthy system as underperforming. Conversely, if you let this temperature loss explain everything, you will miss genuine faults.

There is one practical conclusion. Fix the expected PR for Thai climate conditions at the outset, and for a new system 75 to 80% is the working benchmark. Import a design value from a plant in Japan and the expectation will be too high, every day will look like an anomaly, and nobody will read the alerts any more.

Blind spot 4 — zero-export curtailment, throwing away energy from a healthy system

This is the biggest hole. A fully self-consuming, zero-export system is not permitted to push power back to the grid. When factory load falls below generation, the inverter throttles output. As a means of preventing reverse flow this is correct behavior and correct protection. It nevertheless remains true that electricity you could have generated is being thrown away.

When does it happen? On days when production stops. Weekends, Thai public holidays, the long break around Songkran, and planned shutdowns for stocktaking or annual maintenance. For the model factory, the energy that could have been generated across all those stop days is put at 210,000 kWh, exactly three tenths of the 700,000 kWh annual generation. The real ratio depends on your operating calendar, so substitute your own number of working days.

Factory electricity does not fall to zero on a stop day. Chillers, the server room, security lighting, dehumidification in the die store and idling air conditioning all remain as base load. In the model factory that base load absorbs 70,000 kWh. So of the 210,000 kWh only 70,000 kWh can be used, and the difference of 140,000 kWh is never generated at all. In money terms that is 140,000 × 4.5 = 630,000 THB.

Valuing that number needs a caveat. Most stop days are weekends and holidays, which under TOU fall entirely into off-peak. What the wasted electricity is actually worth therefore depends on whether the day is on-peak or off-peak. This article standardizes on 4.5 THB/kWh across the whole year so that the blind spots can be compared on one ruler, and valuing curtailment at the off-peak rate would make the figure smaller. It would still be the largest of the three losses.

The nasty part is that this phenomenon is caught by no inspection at all. The panels are not dirty, every string is alive, the inverter is throwing no errors and insulation resistance is fine. Call in the maintenance contractor and they will go home reporting no abnormality, because genuinely nothing is wrong on the equipment side. The cause sits on the operating-plan side, and it only becomes visible when generation and consumption are lined up on the same timeline.

Add up the three that carry a number and 26% of the expected saving disappears

Adding the three losses that were given a value gives 126,000 + 63,000 + 630,000 = 819,000 THB per year. Against the expected saving of 3,150,000 THB, that is 26%. Blind spot 3, high temperature, is excluded from the total because it belongs in the design assumption. The three losses overlap in time to some extent, so strictly speaking there is a double count worth a few tens of thousands of baht, but not enough to change the order of magnitude.

Look closely at the weighting. Of the 819,000 THB, 630,000 THB, which is more than three quarters, is curtailment. Cleaning the panels will not return one baht of that 630,000 THB, and neither will more frequent inspections. What cleaning returns is the 126,000 THB. Told that a solar array is generating less than expected, most people reach first for cleaning and inspection, but ranked by money the priority runs the other way.

What to measure — PR, specific yield and self-consumption rate

Separating the four blind spots takes three metrics. One is not enough.

PR is the health check of a solar array, and it normalizes the effect of weather in the denominator. That is what makes the excuse about it being cloudy stop working. PR is however a composite, so its value alone will not tell you whether the drop is soiling, a dead string, high temperature or curtailment. Isolating the cause takes string-level currents, module temperature, and the self-consumption rate described next.

Specific yield is the realistic substitute for an existing system that has no pyranometer. Given nothing more than the inverter logs, you can produce it today. Its weakness is that it cannot absorb differences in weather, so an 8% drop against the same month last year could be a difference in dry-season rainfall or it could be degradation, and you cannot tell which. Treat specific yield as a metric for reading trends rather than for hunting anomalies.

The self-consumption rate has a completely different character. It is not a generation-side metric. It only exists once the generation side and the consumption side are reconciled against each other. No matter how precisely you collect inverter data, this number will never emerge from it. And it is the only metric that detects the curtailment of blind spot 4. It runs at 95% or above on working days and drops sharply on stop days, and the size of the drop is the volume of electricity being thrown away.

Do not get the denominator wrong here. In a zero-export system, curtailed energy is never recorded as generation in the first place. Put actual generation in the denominator and the self-consumption rate will sit near 100% even on heavily curtailed days, detecting nothing. Use expected generation derived from irradiance as the denominator. The reason this article keeps insisting that the pyranometer is the fork in the road of monitoring design is not only PR. It is also that the pyranometer is what builds this denominator.

The practical value of running all three metrics together is that the corrective actions are opposites. Because PR falls under curtailment too, you must never look at a PR drop alone and go running to the equipment. The order is fixed. Look at the self-consumption rate first. If it is holding at 95% or above on working days while PR is falling, the problem is on the equipment side and you proceed to cleaning, inspection and parts replacement. If instead the PR drop is concentrated on stop days, and on those days the factory load itself has vanished, the equipment is healthy and neither cleaning nor inspection will change anything. What to pursue then is how load is shaped on stop days, whether storage is justified, or the production schedule itself. Faced with the same symptom of generation being down, acting without separating the causes will reliably send you to the wrong place.

IEC 61724-1 governs more than instrument accuracy

There is an international standard for monitoring the performance of photovoltaic systems. IEC 61724-1, formally titled “Photovoltaic system performance – Part 1: Monitoring”, is currently in Edition 2.0, published on 2021-07-21.

The points a practitioner needs are few. First, the 2021 edition withdrew Class C, leaving two classes, Class A and Class B. If an old document or a vendor proposal still refers to Class C, it is citing the 2017 edition or earlier.

Second, accuracy requirements are specified separately for the pyranometer, the module temperature sensor and the energy meter. So far this is a hardware conversation, but hardware is not all the standard demands. The class is determined together with calibration intervals, inspection procedures, the handling of missing data and the methods for checking data quality. Buying one high-accuracy pyranometer does not make a system Class A. This matters when you are comparing quotations. A proposal that lists sensor model numbers but says nothing about how calibration and inspection will be operated is meeting only half the standard.

Third, the 2021 edition also specifies how to measure the soiling ratio and how to monitor bifacial modules. If you want to move from calendar-driven cleaning to cleaning driven by measured soiling, this is the part of the standard you build the practice on.

For factory practice the largest fork in the road is whether to install a pyranometer at all. Without one you cannot build the denominator of PR, so PR cannot be produced at all. And without PR, every drop in generation can be explained away by the weather. Put another way, from the moment one pyranometer set is installed, the weather explanation stops working. That is where monitoring design divides.

How to build the monitoring architecture for factory solar

The following points cannot be skipped when settling the architecture. They hold regardless of the size of the system or the brand of equipment.

• Take string-level currents from the inverters over Modbus TCP or SunSpec. An architecture that polls only inverter totals is structurally blind to the string failure of blind spot 2. Most industrial inverters hold string-level currents internally, but many do not expose them on the standard monitoring screen. Get the register map before you commit and confirm that string currents can be read.
• Fit at least one plane-of-array pyranometer and one rear-of-module temperature sensor. This is the only way to build the denominator of PR. The pyranometer is meaningless unless it is mounted at the same tilt and the same orientation as the panels. Calculate PR from a horizontally mounted pyranometer and you carry a systematic seasonal error.
• Put the incomer meter and the main feeder meters on the same time base. The self-consumption rate can be produced nowhere else. If the clocks drift, the period when curtailment occurred will not line up with the period when load disappeared, and causality becomes unreadable. Synchronizing over NTP is the minimum condition.
• If a power monitoring system already exists, do not stand up a separate solar screen. Fold it into the same platform. Split the display in two and nobody will be left reconciling the generation side against the consumption side. How to build that platform is set out in the guide to factory energy monitoring systems, so if you already have an architecture, build upward from there.
• Store data at 15-minute granularity or finer. Zero-export curtailment acts over minutes. Keep nothing but daily totals and curtailment dissolves into the average and vanishes. Where possible, retain raw values at 1-minute granularity and build the aggregates downstream.

Before fixing the architecture, take an inventory of what is already installed. Check the inverter model numbers and whether communication ports exist, whether string currents can be read, the location and communication specification of any existing meters, and how many years of inverter logs are retained. If a year of logs exists, you can start today, without adding a single sensor, on lining up monthly specific yield against the previous year.

If you intend to use the generated energy in CO2 reduction reporting, decide the emission factor methodology and the definition of the aggregation period first. That subject is handled in the article on CO2 emission visualization. The raw monitoring data may be identical, but adding a reporting purpose changes the granularity and retention period you need.

Five cost layers of monitoring, and the payback period

Cost cannot be judged when it arrives as a single lump-sum quotation. Break it into layers and it becomes clear what stops being visible when you cut each one.

Layer 2 has a wide range because pyranometer price more than doubles across classes. Whether Class B grade is enough or Class A grade is required is decided by the purpose of the monitoring, and if the purpose is internal operational improvement, Class B grade functions perfectly well. Layer 3 falls to 0 when a power monitoring system already exists and the incomer energy is already on the same platform. It swings up to 400,000 THB when meters have to be added from scratch, with panel modifications, additional CTs and work that requires a shutdown.

Summing layers 1 to 4 gives an initial cost range of 340,000 to 1,400,000 THB. The lower bound assumes layer 3 is 0 and the upper bound assumes every layer at full price. Layer 5 is annual, so it is excluded from the initial cost.

Take 900,000 THB as a mid-range configuration. What you can recover is not the whole 819,000 THB derived earlier. Part of the curtailment is locked in by the production schedule and cannot be moved, and soiling can never be driven to 0%. Assume you recover 60% and that is roughly 490,000 THB per year, which gives a simple payback of 900,000 divided by 490,000 = 1.8 years.

Do not get the order of cuts wrong. The first thing people cut is the layer 2 pyranometer, but cutting it means PR cannot be produced, which removes the reference against which every other layer of data would have been evaluated. The next thing people cut is layer 3, but cutting that means the self-consumption rate cannot be produced, which hides the curtailment that accounts for three quarters of the money. What can safely be trimmed is the depth of build in layer 4, meaning the number of dashboard screens and how much report generation is automated.

A configuration with layer 5 set to 0 is not recommended. Sensors and dashboards can all be in place, but if nobody has decided who looks at what and what they then do, data merely accumulates. Set the thresholds, fix the format of the daily report, and connect the cleaning decision to the soiling ratio. Spending 60,000 to 180,000 THB a year on that operational design is plainly worth it, as the breakdown of the 819,000 THB makes clear.

Thai regulation — since December 2024, rooftop solar outside industrial estates no longer needs รง.4

Permitting for putting solar on a factory roof in Thailand changed substantially at the end of 2024.

The Ministerial Regulation on the Designation of Types, Categories and Sizes of Factories (No. 3) B.E. 2567 was issued on 2024-12-27 and took effect the following day, on 2024-12-28. Under it, rooftop solar located outside industrial estates no longer requires รง.4, the factory licence, regardless of capacity. Previously it was required for systems above 1,000 kW, so for a plant considering an expansion from the 500 kWp class toward the 1 MW class, one procedural step has been removed.

That does not mean every filing has gone away. The main obligations that remain in practice are as follows. First, the ERC energy business licence, or the notification that accompanies an exemption. A total below 1,000 kVA is exempt from the generation licence, but even when exempt, the notification to the ERC is still required. Exempt and not required are not the same thing. Second, building control. Approval covering roof loading and structural modification remains a separate matter, and for a retrofit onto an existing roof, structural calculation is unavoidable. Third, grid interconnection approval from PEA or MEA. Even when the system operates in zero export, prior consultation is required, including how reverse power flow will be positively prevented. Other procedures can remain depending on project conditions, so do not treat this list as exhaustive.

Inside an industrial estate the rules are separate. Where the site falls under IEAT jurisdiction, land use permission and notification of commencement of operations continue to apply. IEAT has previously signaled an exemption policy for rooftop systems dedicated to self-consumption, so confirm the latest notification before you start. Because most Japanese-owned factories in Thailand sit inside industrial estates, do not apply the ministerial regulation above to your own site without checking. The starting point is confirming whether your location is inside or outside an estate.

Two further points about scope. Everything above concerns rooftop installations only, and ground-mounted systems are treated differently. Do not mix this with the debate about the residential purchase scheme, the so-called net billing arrangement, either. What this article addresses is factory systems that do not sell the electricity they generate. Since nothing is sold, revisions to purchase prices are irrelevant, and the economics are determined solely by the effective tariff of the electricity displaced and by the share that was actually self-consumed.

A 90-day roadmap for standing up monitoring

Attempt everything at once and you will stall on pyranometer procurement lead times and switchboard modification schedules. Cut the three months into three stages.

What days 0 to 30 are for is data extraction, not procurement. If a year of inverter logs exists, this stage answers the question of what percentage of the assumed 700,000 kWh is actually being produced. If it turns out you are well below plan, every subsequent investment decision becomes far easier.

When the pyranometer goes in during days 31 to 60, do not treat time synchronization lightly. If the pyranometer, the inverters and the incomer meter run on separate clocks, the self-consumption rate stops meaning anything on a period-by-period basis. Writing confirmation of time synchronization into the completion criteria for the installation work will save rework later.

Days 61 to 90 are operational design. Four things get decided here, which are the format of the daily report separating working days from stop days, the thresholds, who looks when a threshold is crossed, and who acts. As a workable threshold example, PR falling 5% below the baseline on two consecutive days is easy to live with. A one-day dip happens routinely when cloud passes over, so requiring two consecutive days keeps the alert count at a realistic level.

There is one operational practice that must change at this stage. Set the cleaning interval by soiling ratio rather than by the calendar. Once the trend of PR against irradiation makes the progress of soiling visible, you can make judgments with numbers, such as every six weeks in the dry season and not at all after the rains. Simply dropping the twice-yearly cleaning schedule in favor of soiling-driven cleaning will visibly shrink the 126,000 THB loss.

Five common failures in solar monitoring

• Settling for the free monitoring screen bundled with the inverters. It shows only totals, so string failures and curtailment are structurally invisible. Being free and containing the information you need are two different things.
• Skipping the pyranometer. PR can no longer be produced, and every drop can be explained by the weather. Being explainable in that way is the same as being unverifiable.
• Standing up the generation screen and the power monitoring screen separately. Reconciling them becomes nobody’s job, and no one ever steps up to produce the self-consumption rate. Six months later one of the two screens is no longer opened.
• Throwing curtailment at the maintenance contractor as a fault. The equipment is healthy so nothing comes back. The report says no abnormality found, the cause stays unidentified, and next year you throw away the same volume again.
• Running cleaning off the calendar. You wash when the panels are clean and leave them when they are dirty. It is the least rewarding state possible, paying for cleaning while losses stay flat.

What these five share is that none of them is a case of not monitoring. There is a screen, there is a contractor, and cleaning does happen. The reason 819,000 THB still disappears is that what is being measured and what is being lost are not the same thing.

Frequently asked questions

Is the screen bundled with the inverters not enough for a solar monitoring system?

It depends on the purpose. If all you want is to watch today’s generation, it is enough. But most bundled screens are designed to display inverter totals and do not expose string-level currents. Many have no pyranometer input either, in which case PR cannot be calculated at all. They also cannot take in data from the factory consumption side, so the self-consumption rate is structurally unavailable. Of the four blind spots listed in this article, what a bundled screen realistically reveals is only a major hardware failure. In money terms, only a small fraction of the 819,000 THB is within its reach.

When factory solar output drops, what should I look at first?

There is an order. Look at the self-consumption rate first. If that has fallen, the equipment is healthy and the cause is curtailment, and proceeding to cleaning or inspection will produce nothing. If the self-consumption rate has not fallen, look at PR next. If PR has fallen, the problem is on the equipment side, so lay out the string-level currents. One string low means a string failure, and all strings low together means soiling or high temperature. Checking module temperature separates those two. Following this order eliminates pointless cleaning and pointless inspection call-outs.

Below what performance ratio should I treat the system as abnormal?

Drawing the line on an absolute value is not recommended, because the reasonable level shifts with installation conditions. For a new rooftop system in Thailand, 75 to 80% is the benchmark, but roof tilt, orientation, ventilation and surrounding shade will move it by several points. In practice, record the measured PR of your own system in its first year as a baseline and judge by relative decline from there. A condition such as 5% below baseline on two consecutive days generates fewer false alarms than an absolute threshold and is far more likely to survive in daily operation. Year over year, the baseline itself has to be lowered to account for degradation.

Do I really need a pyranometer, and is there a way to keep the cost down?

If you want PR, yes. PR cannot be defined without one. There are, however, ways to keep the cost down. One is class selection, and if the purpose is internal operational improvement, Class B grade is sufficient rather than Class A grade. Where there is no need to assert performance guarantees to a third party, that difference does no practical harm. The other is quantity, and a system with consistent roof tilt and orientation can start with a single set. If the budget still cannot be found, a realistic sequence is to begin operating on year-on-year specific yield and defer the pyranometer to the following fiscal year. Just be clear that specific yield alone cannot absorb differences in weather, so treat it as an interim measure.

How can I tell whether generation is being thrown away on days when production stops?

Overlay generation and factory power consumption on the same time axis at 15-minute granularity or finer. If curtailment is occurring, the top of the generation curve will be flattened off on a clear day, and the height of that flat section will match the factory consumption line almost exactly. A weather-driven drop makes the curve wander irregularly, so the two are distinguishable by eye. To pin it down numerically, compare the self-consumption rate on stop days against working days. If working days hold at 95% or above while stop days fall sharply, that gap is the volume being thrown away. Some inverter models record a power-limitation status flag in their logs, so the register map is worth checking.

Summary

Self-consumption solar sells nothing, so its results are buried inside the electricity bill. What gets lost while buried, for the model factory in this article, is 819,000 THB a year, or 26% of the expected 3,150,000 THB saving. The breakdown is 126,000 THB of soiling, 63,000 THB of string failure and 630,000 THB of zero-export curtailment. The largest hole, curtailment, is not an equipment fault, so no amount of additional cleaning or inspection will bring it back.

What is needed is PR on the generation side, the self-consumption rate on the consumption side and purchased energy on the billing side, all carried on one shared timeline. Producing PR takes a pyranometer, and producing the self-consumption rate takes an incomer meter and time synchronization. A mid-range configuration covering both comes to 900,000 THB against 490,000 THB per year recovered, for a simple payback of 1.8 years. Three months is enough to get from an inventory of existing inverter logs to a daily report in live operation.

Even if you are only at the stage of confirming whether your existing solar array is generating what it was supposed to, TOMAS TECH is happy to help. Given the inverter logs and the last twelve months of electricity invoices, we can produce year-on-year specific yield without adding a single sensor, and help you judge how far to go in adding pyranometers and meters from there. Please get in touch through the contact page.

References

• Tilleke & Gibbins, “Thailand Removes Factory License Requirement for Solar Rooftop Installations” — dated 2025-01-14. Explains the scope of the รง.4 exemption under Ministerial Regulation No. 3 and the ERC notification that remains
• Hunton Andrews Kurth, “Thailand Eases Regulations for Solar Rooftop Installations” — a practitioner-oriented summary of the same amendment, including the issue date and the effective date
• The National Law Review, “Thailand Eases Regulations for Solar Rooftop Installations” — a commentary article covering the same ground
• IEC, “IEC 61724-1:2021 Photovoltaic system performance – Part 1: Monitoring” — Edition 2.0, published 2021-07-21. Requirements for Class A and Class B, and the method for measuring soiling ratio
• Thailand Board of Investment, “Utility Costs” — updated 2023-06-30. Official ranges for large-scale TOU energy charges, demand charges and the power factor surcharge
• World Bank / ESMAP, “Thailand – Solar irradiation and PV power potential maps” — Global Solar Atlas. Primary data for replacing the specific yield assumption with PVOUT at your own site coordinates