A small voltage change may pass unnoticed if it happens slowly, yet a similar change can be very irritating if it repeats at a rate to which human vision is sensitive. Flicker assessment therefore needs a dedicated measurement method — not a simple voltage-deviation calculation.
Key idea
- Flicker = voltage fluctuation + lamp response + human perception, so it is measured with a flickermeter (IEC 61000-4-15).
- Two indices: \(P_{st}\) over 10 minutes and \(P_{lt}\) over 2 hours, combined statistically over a week.
- \(P=1.0\) is the conventional irritability threshold; planning levels sit a little below it.
- Severity follows the network: \(\Delta V \approx Z_s\,\Delta I\) — a weaker system gives more flicker.
Section 1
What flicker means in power-quality assessment
Voltage fluctuations are changes in RMS voltage magnitude; flicker is the visual effect those fluctuations create. It is best understood as a chain of effects rather than a single number:
\[ \Delta V(t)\;\rightarrow\;\text{lamp luminance variation}\;\rightarrow\;\text{human perception}\;\rightarrow\;\text{flicker severity} \]
The severity depends on the magnitude, repetition rate, waveform and duration of the fluctuation. A fluctuation that repeats several times per second can be far more disturbing than a slow variation of the same magnitude. The usual sources are loads whose current changes rapidly:
Table 1 — Typical flicker sources and their mechanism.
| Source | Flicker Mechanism |
|---|
| Arc furnaces | Rapid, random load-current variation |
| Large motor starts | Step voltage changes during starting |
| Welders | Repetitive current pulses |
| Crushers, mills and saws | Cyclic mechanical loading |
| Wind turbines | Power fluctuations and network interaction |
| Large fluctuating industrial loads | Repeated reactive-power changes |
| Weak networks | Larger voltage change for the same load variation |
Flicker is worse on a weak network because the same fluctuating current produces a larger voltage change:
\[ \Delta V \approx Z_s\,\Delta I \]
- \(\Delta V\)
- voltage fluctuation
- \(Z_s\)
- source (system) impedance at the connection point
- \(\Delta I\)
- fluctuating component of the load current
Weaker system → larger voltage fluctuation → higher flicker severity.
Section 2
Why a flickermeter is used
Human visual perception is not linearly related to voltage fluctuation, so flicker cannot be judged from voltage deviation alone. A flickermeter models the response of a reference lamp and the sensitivity of the eye and brain to changes in light. The standard method is defined in IEC 61000-4-15, and it converts the voltage waveform into an instantaneous flicker sensation and then into statistical severity indices:
\[ v(t)\;\rightarrow\;\text{fluctuation processing}\;\rightarrow\;\text{lamp-eye-brain model}\;\rightarrow\;\text{instantaneous flicker}\;\rightarrow\;P_{st},\,P_{lt} \]
The output is not in volts — it is a per-unit flicker severity. A severity of \(P=1.0\) corresponds approximately to the conventional irritability threshold for a significant proportion of observers under standard test conditions with a reference incandescent lamp.
Section 3
Short-term flicker severity, \(P_{st}\)
The short-term flicker severity \(P_{st}\) is evaluated over a 10-minute interval. Ten minutes is used because flicker is a perceptual disturbance: the duration and repetition of the fluctuation matter as much as the largest instantaneous change. A value of \(P_{st}=1.0\) sits at the conventional threshold where flicker becomes irritating to a significant number of people under the reference conditions of the standard.
Table 2 — General interpretation of \(P_{st}\) (guidance only).
| \(P_{st}\) value | General Interpretation |
|---|
| \(P_{st} < 0.7\) | Usually not disturbing |
| \(0.7 < P_{st} < 1.0\) | Noticeable but often acceptable |
| \(P_{st} \approx 1.0\) | Around the conventional irritability threshold |
| \(P_{st} > 1.0\) | More likely to be disturbing |
This interpretation should be applied with care: real perception depends on the lighting technology, background light, occupancy, time of day and the type of fluctuation.
Section 4
Long-term flicker severity, \(P_{lt}\)
The long-term flicker severity \(P_{lt}\) is formed from twelve consecutive \(P_{st}\) values, covering a 2-hour period, using a cubic average:
\[ P_{lt}=\left(\frac{1}{12}\sum_{i=1}^{12}P_{st,i}^{\,3}\right)^{1/3} \]
- \(P_{st,i}\)
- short-term severity for each of the twelve 10-minute intervals
- \(P_{lt}\)
- long-term severity over the 2-hour window
Because the summation is cubic, the higher \(P_{st}\) values dominate \(P_{lt}\) — which reflects the fact that occasional severe flicker is more disturbing than many small fluctuations. The 2-hour value represents sustained flicker exposure over a longer period.
Section 5
The relationship between \(P_{st}\) and \(P_{lt}\)
The two indices are usually correlated but they are not the same thing: \(P_{st}\) shows short-term severity over 10 minutes, while \(P_{lt}\) shows longer-term severity over 2 hours. For a regular, stable fluctuating process, \(P_{lt}\) can be close to \(P_{st}\). For an intermittent process, \(P_{lt}\) is usually lower than the worst \(P_{st}\), because the disturbance is not continuous across the full 2 hours. Site-specific relationships are sometimes observed — for example, around arc furnaces:
\[ P_{lt,95\%}\approx 0.84\,P_{st,95\%} \]
- \(P_{st,95\%}\)
- 95th-percentile weekly short-term flicker severity
- \(P_{lt,95\%}\)
- 95th-percentile weekly long-term flicker severity
but a ratio like this is particular to the site and must not be treated as universal. The practical distinction is simple:
\(P_{st}\) → short-term irritation · \(P_{lt}\) → sustained flicker exposure.
Section 6
Statistical flicker indices
Flicker varies with load operation and network condition, so the indices are assessed statistically over a defined period — commonly one week. A weekly 99% \(P_{st}\) value means 99% of the 10-minute \(P_{st}\) values that week were below it; a weekly 95% \(P_{lt}\) value means 95% of the 2-hour \(P_{lt}\) values were below it. The comparison then takes the form:
\[ P_{st,99\%}\;\le\;P_{st,\text{objective}} \qquad\qquad P_{lt,95\%}\;\le\;P_{lt,\text{objective}} \]
- \(P_{st,99\%}\)
- 99th-percentile weekly short-term severity
- \(P_{lt,95\%}\)
- 95th-percentile weekly long-term severity
- \(P_{\cdot,\text{objective}}\)
- applicable flicker objective or limit
Whether the 95% or 99% value is used depends on the applicable standard, guideline or contract.
Section 7
Assessment intervals and period
The flicker assessment runs down a fixed chain of time bases, from the instantaneous flickermeter output to a weekly statistic:
\[ \text{instantaneous flicker}\;\rightarrow\;P_{st}\,(10\,\text{min})\;\rightarrow\;P_{lt}\,(2\,\text{h})\;\rightarrow\;\text{weekly statistical value} \]
Table 3 — Flicker quantities and their time basis.
| Quantity | Time Basis | Purpose |
|---|
| Instantaneous flicker | Continuous | Internal flickermeter response |
| \(P_{st}\) | 10 minutes | Short-term flicker severity |
| \(P_{lt}\) | 2 hours | Long-term flicker severity |
| Weekly 95% / 99% value | One week | Compliance or planning assessment |
One week is the usual minimum because it captures daily load variation, working-day and weekend operation, and different operating modes. For industrial processes with long operating cycles, the period must be long enough to capture the full cycle of the load.
Section 8
Voltage dips, swells and flagged data
Voltage dips, swells and interruptions can drive large changes in the flickermeter output, yet they are not part of the normal flicker caused by repetitive fluctuations. Modern power-quality practice therefore flags flicker values affected by such events so they can be reviewed or removed from the statistics where appropriate. This matters because high percentile values — especially 99% or weekly maxima — can be inflated by isolated dips or switching events.
Do not confuse flicker with voltage-dip events — flicker assessment targets repeated luminance variation, not isolated disturbances.
Section 9
Objectives, and the three reference levels
Flicker objectives are anchored to the conventional perception threshold. Under reference conditions, \(P=1.0\) is approximately the level considered irritating to a significant proportion of observers, so common objectives sit close to \(P_{st}=1.0\) and \(P_{lt}=0.8\). Planning levels are usually a little lower — for example \(P_{st}=0.9\) and \(P_{lt}=0.7\) — because the operator must manage the combined effect of many fluctuating loads. If every customer reached the compatibility level individually, the combined system flicker could become excessive.
As with harmonics, three terms must be kept distinct:
Compatibility level = the disturbance level equipment and users should generally tolerate (for flicker, near the perceptibility threshold).
Voltage characteristic = the supply-voltage quality a user can expect at the terminals (often a weekly \(P_{lt}\) value on public networks).
Planning level = the operator’s internal target used to allocate flicker emission among fluctuating loads.
Planning levels carry a margin for existing and future users, so they sit below the others:
Planning level < compatibility level.
Table 4 — The three flicker reference levels.
| Reference Level | Role for Flicker |
|---|
| Compatibility level | Coordinate system disturbance with equipment immunity |
| Voltage characteristic | Describe expected supply quality at the customer terminal |
| Planning level | Allocate emission among fluctuating loads (connection studies) |
Planning levels are especially important when connecting large fluctuating loads — arc furnaces, large motors, welders, wind farms or industrial plant.
Section 10
Flicker across LV, MV, HV and EHV
At LV and MV, flicker objectives relate directly to customer experience, because lighting and end-use equipment connect at those levels. At HV and EHV, customers do not normally connect lamps directly, so the objectives are mainly about coordination — how flicker transfers through transformers and downstream networks to MV and LV customers. The transfer is captured by a coefficient:
\[ T_{P}=\frac{P_{\text{downstream}}}{P_{\text{upstream}}} \]
- \(P_{\text{upstream}}\)
- flicker severity at the higher voltage level
- \(P_{\text{downstream}}\)
- flicker severity at the lower voltage level
- \(T_P\)
- flicker transfer coefficient
If \(T_P < 1\) the flicker is attenuated as it transfers downstream; if \(T_P > 1\) it is amplified. In many systems, motor loads and network impedance attenuate the transfer from HV/EHV to lower levels — one reason higher planning levels can sometimes be accepted at transmission voltages, provided the downstream levels remain acceptable.
Section 11
The voltage-change relationship and flicker risk
For loads with rapidly varying active or reactive power, the voltage change at the connection point can be approximated from the source resistance and reactance:
\[ \Delta V \approx R\,\Delta P + X\,\Delta Q \qquad\Longrightarrow\qquad \Delta V_{pu} \approx R_{pu}\,\Delta P_{pu} + X_{pu}\,\Delta Q_{pu} \]
- \(\Delta P\)
- active-power variation of the load
- \(\Delta Q\)
- reactive-power variation of the load
- \(R,\,X\)
- source resistance and reactance at the connection point
In most MV and HV systems \(X\) is much larger than \(R\), so reactive-power fluctuations are often the dominant cause of voltage fluctuation — which is exactly why arc furnaces, motor starts and rapidly varying reactive loads produce flicker. The risk increases under the following conditions:
Table 5 — Conditions that increase flicker risk.
| Condition | Effect |
|---|
| Low system short-circuit level | Larger voltage change for the same current |
| Large load power variation | Larger voltage fluctuation |
| Repetition in the sensitive frequency range | Higher perceived flicker |
| Large reactive-power variation | Larger voltage modulation |
| Frequently operating load | Higher statistical flicker severity |
| Existing background flicker | Less margin for the new load |
Section 12
Assessing a new fluctuating load, and mitigation
When a new fluctuating load is connected, the assessment must consider both its emission and the existing background:
\[ P_{\text{existing}} \;+\; P_{\text{new load}}\;\longrightarrow\;P_{\text{total}} \]
Because flicker indices are statistical and perceptual, simple arithmetic addition is not always appropriate — planning studies use summation laws or allocation methods set by the applicable standard. A practical connection assessment runs through nine steps:
Table 6 — A practical flicker connection assessment.
| Step | Purpose |
|---|
| 1 | Identify the fluctuating load and its operating cycle |
| 2 | Determine the point of common coupling |
| 3 | Obtain the short-circuit level and impedance angle |
| 4 | Estimate the voltage fluctuations caused by the load |
| 5 | Calculate or measure \(P_{st}\) and \(P_{lt}\) |
| 6 | Consider transfer to lower voltage levels |
| 7 | Compare with planning levels or contractual limits |
| 8 | Check the cumulative effect with existing flicker |
| 9 | Identify mitigation if required |
Flicker is reduced either by reducing the voltage fluctuation or by limiting its transfer to sensitive customers. For arc furnaces and similar loads, fast dynamic compensation (SVC or STATCOM) is the usual choice:
Table 7 — Common flicker mitigation methods.
| Mitigation Method | Purpose |
|---|
| Increase fault level at the connection point | Less voltage change for the same load variation |
| Connect at a higher voltage level | Stronger system connection |
| Dynamic reactive compensation | Reduces reactive-power fluctuation |
| SVC or STATCOM | Fast voltage and reactive-power control |
| Soft starters or variable-speed drives | Reduce motor-start voltage dips |
| Controlled switching | Reduce sudden voltage steps |
| Network reinforcement / dedicated feeder | Reduce source impedance and impact on others |
Reduce the fluctuation and the severity follows: \(\Delta V(t)\downarrow\;\Rightarrow\;P_{st},\,P_{lt}\downarrow\).
Section 13
The five points, and the key message
The reference threshold \(P_{st}=1\) is based on laboratory conditions and an incandescent lamp, so it should not be read too mechanically — LED lighting, daylight, mixed sources, occupancy and the type of fluctuation all change real perception. Engineering judgement is needed where measured values slightly exceed the threshold but no complaints arise. For planning and connection studies, however, the applicable objective should still be followed unless the network operator agrees otherwise. Above all, a flicker statement must pin down five things:
- Source of the fluctuation
- Location of assessment (PCC, customer bus, equipment terminals)
- Measurement method (flickermeter, simulation)
- Statistical index (\(P_{st}\), \(P_{lt}\), 95%, 99%)
- Comparison objective (planning level, voltage characteristic, compatibility level, contract)
This is why \(P_{st}=0.95\) or \(P_{lt}=0.8\) on its own is incomplete; a defensible statement looks like:
\(P_{lt,\,95\%,\,\text{weekly}} = 0.8\) at the PCC — index, statistic, period and location, all stated.
Key message
Flicker is the visual effect of voltage fluctuation, measured with a standard flickermeter because perception depends on magnitude, frequency, duration, lamp response and eye-brain sensitivity. The two indices are \(P_{st}\) (10-minute) and \(P_{lt}\) (2-hour, the cubic average of twelve \(P_{st}\) values), assessed statistically over at least a week. The engineering behind it is \(\Delta V \approx Z_s\,\Delta I\) — equivalently \(\Delta V_{pu}\approx R_{pu}\Delta P_{pu}+X_{pu}\Delta Q_{pu}\) — so flicker depends on both the fluctuating load and the strength of the network. A robust assessment must state the measurement method + assessment period + statistical index + objective + assessment location + treatment of flagged events. Only then can two flicker figures be compared fairly and technically.
