How to Read a Power Quality Report | AHF & SVG

Most AHF and SVG quotations are built backwards. A buyer collects a short-circuit level, a transformer rating and a target THDi, then asks three suppliers for a price. The result is a device sized to a guess rather than to the disturbance. A power quality report already contains most of what is needed to size a harmonic filter or reactive compensator correctly, provided the reader knows which columns matter, where the data was captured, and which limits hide in the fine print.

The Measurement Problem Comes First

Before comparing numbers, establish how the data was captured. A portable three-phase analyzer, a permanently installed meter and an inverter's internal log will disagree, and the way they disagree matters more than the figures in the table.

First, where was the measurement at PCC taken? Data recorded at a feeder, inside a distribution board, or on the low-voltage side of another transformer describes a different circuit. Harmonic current divides between parallel paths, so the current at the point of common coupling can be lower or higher than at the terminals of a single drive.

Second, how were the current transformers installed and specified? CT placement decides whether the harmonic current of the whole installation is captured or only one branch. A CT with a narrow frequency response, or one already close to saturation, under-reports higher-order components. Zero-sequence current is invisible unless a neutral CT or four-wire residual connection exists, so a report showing no third-harmonic current may be missing the instrument rather than the phenomenon.

Third, what is the averaging interval, and how much of the load profile does the record cover? Ten-minute averages hide short severe bursts, while one shift of data says nothing about a night shift or a weekend. One full production cycle, with process states marked, is the minimum useful basis for selection.

Reading the Data Layer Without Over-Interpreting It

The measurement layer answers one question: what does the installation look like electrically during the recorded period? Work through it in a fixed order, because each layer constrains the next.

Start with RMS voltage and RMS current per phase. They confirm the operating point and show whether the record represents production load or a quiet period. Compare RMS current against nameplate and transformer ratings: a report captured at light load makes distortion ratios look alarming while the true harmonic current stays modest.

Next, check the power balance. If apparent power is noticeably larger than the vector combination of active and reactive power, distortion power is present. Read power factor carefully as well: displacement and true power factor are both legitimate, but they answer different questions, and a compensator is selected on the one the report actually contains.

Then read THDi and THDu together. THDi describes the current distortion produced by the load; THDu describes the voltage distortion the network delivers. High THDi with low THDu suggests a stiff supply and locally injected current, while high THDu with moderate THDi points to network impedance, background distortion or resonance.

Finally, open the individual harmonic orders. The distribution across orders drives equipment choice: fifth and seventh dominate with six-pulse loads, third and ninth point to single-phase and neutral paths, and a broad or shifting spectrum suggests mixed converter technologies. Where a neutral exists, zero-sequence current is the most informative single number, because it separates unbalance from distortion.

Turning Signals into an Engineering Interpretation

One reading can justify very different actions, so the interpretation should be written down before any device is chosen. The table below reflects the logic used in our own application reviews.

Classifying the Cause Before Choosing Hardware

Three causes cover most reports, and they coexist. The first is current distortion from converters: rectifiers, drives, chargers and UPS units draw non-sinusoidal current regardless of supply quality. The second is network behaviour: impedance, resonance with existing capacitor banks, and background distortion already on the busbar. The third is unbalance: unequal single-phase loading producing zero-sequence current in the neutral and negative-sequence current in machine windings.

A useful discipline is to ask which cause would remain if the other two disappeared. If converter current remains, the project needs harmonic filtering. If the answer is reactive demand that swings with the process, it needs a fast reactive compensator. If unbalance remains, no filter will fix it and neutral current will keep heating cables and transformers.

Matching the Cause to AHF, SVG, ALB, Passive or Hybrid

Selection becomes straightforward once the cause is classified and the measurement basis is trusted. An active harmonic filter (AHF) suits reports where harmonic current varies with load and the target is a defined distortion level across several orders; it cancels selected orders dynamically and follows a changing spectrum. A static var generator (SVG) belongs where reactive power and voltage move quickly, with reactive power swinging within seconds and power factor drifting from target, because it responds faster than switched steps without introducing resonance.

An active load balancer (ALB) answers reports where zero-sequence current and phase imbalance dominate: it transfers active power between phases, reducing neutral current and negative-sequence stress without generating reactive power. Passive solutions stay valid when the load profile is stable and the spectrum is dominated by one or two orders, since a tuned stage is efficient at a fixed frequency, but it must be detuned against network impedance to avoid resonance. A hybrid arrangement pairs a passive stage for the bulk of a stable reactive demand with an active stage for residual distortion, which often fits a large installation showing both a steady displacement component and a variable distortion component.

What to Submit So the Selection Can Be Checked

Sizing quality depends on the information supplied with the enquiry. The second table lists the project data that changes the outcome and the item to submit.

Limitations of the Report and of Any Single Device

A report is a snapshot of a system that keeps changing. It cannot show how a future extension, a new drive line or a relocated capacitor bank will alter the impedance at the busbar, and it cannot predict the interaction of equipment that is not installed yet. Where the record covers only part of the load profile, the missing hours are unknown, not an average.

Two cautions remain. Measurement errors are systematic: a wrong CT rating, reversed polarity or a missing neutral channel will mislead every conclusion that follows, so instrument details belong in the report. And no single device solves every disturbance: an active filter does not correct unbalance, a balancer does not remove distortion, and reactive compensation that improves power factor can worsen resonance when it is added without detuning.

FAQ: Reading a Power Quality Report

Should I trust THDi or THDu more?

Neither alone. THDi defines the filtering duty, while THDu reveals resonance and impedance problems. Read them as a pair: high THDi with low THDu is a filtering task, and high THDu with moderate THDi is a network investigation before any equipment decision.

Does CT placement change the result?

It changes the result completely. A CT on one branch measures one branch, and harmonic current divides among parallel paths, so branch data cannot be scaled to the point of common coupling without impedance information. Check the stated CT ratio, position and neutral provision, and treat any report that omits them as incomplete.

How is capacity actually selected?

Capacity follows measured RMS current and the harmonic current that must be cancelled, not the transformer rating. Filtering duty grows with load, so the sizing current is the highest sustained RMS current in the load profile, checked against the spectrum at that moment. Reactive capacity follows the reactive power range, and balancing capacity follows zero-sequence content.

What about maintenance and payback?

Active equipment needs periodic inspection of fans, filters and connections, and its lifetime cost depends on duty and ambient temperature, so state those conditions when quotations are compared. Payback cannot be promised from a report alone, because it depends on tariff structure, loss pattern and operating hours; the honest approach is to model those inputs on the measured load profile rather than assume a fixed saving.