A harmonic, flicker or unbalance level can usually be assessed over a week. Voltage dips are different: they are caused by faults, protection operation, motor starts, transformer energisation and switching — events that do not occur regularly — so dip performance normally needs a much longer window, often at least a year, and is described by counts of events grouped by retained voltage and duration.
The dip question is “how often do dips occur, and how severe are they?” — not “what is the average power-quality level?”
Key idea
- Dips are event-based — assess over ≈1 year, not one week.
- Each dip is described by retained voltage and duration, from half-cycle RMS \(U_{\text{rms}(1/2)}\).
- Reported by SARFI, magnitude-duration tables, coordination charts, dip energy and dip severity.
- Impact depends on depth and duration together, and on equipment ride-through.
Section 1
What a voltage dip is
A voltage dip is a temporary reduction in RMS voltage below a defined threshold, after which the voltage recovers. The chain from cause to consequence is:
\[ \text{disturbance}\;\rightarrow\;\text{RMS voltage reduction}\;\rightarrow\;\text{retained voltage and duration}\;\rightarrow\;\text{equipment impact} \]
Table 1 — Common causes of voltage dips.
| Cause | Typical Effect |
|---|
| Short-circuit faults | Deep dips until protection clears the fault |
| Motor starting | Shallower dips of predictable duration |
| Transformer energisation | Voltage depression and possible transient distortion |
| Faults on adjacent feeders | Dips seen by nearby customers |
| Network switching | Step changes or short voltage reductions |
| Large fluctuating loads | Repeated dips or voltage changes |
A dip whose retained voltage falls below the interruption threshold — typically \(10\%\) of declared voltage, though some utilities differ — is classified as a short interruption.
Section 2
RMS measurement for voltage dips
Dips are measured with a half-cycle refreshed RMS value, \(U_{\text{rms}(1/2)}\): the RMS is computed over one cycle and refreshed every half cycle. That gives:
\[ \text{50 Hz:}\quad 1\ \text{cycle}=20\ \text{ms},\;\; \tfrac{1}{2}\ \text{cycle}=10\ \text{ms} \qquad \text{60 Hz:}\quad 1\ \text{cycle}=16.67\ \text{ms},\;\; \tfrac{1}{2}\ \text{cycle}=8.33\ \text{ms} \]
which is enough time resolution to capture the start, end and shape of a dip. The measurement sequence is:
\[ v(t)\;\rightarrow\;U_{\text{rms}(1/2)}(t)\;\rightarrow\;\text{threshold crossing}\;\rightarrow\;\text{retained voltage and duration} \]
Section 3
Retained voltage, dip depth and duration
A dip is characterised mainly by its retained voltage (the lowest RMS reached during the event, as a percentage of the reference) and its duration. The depth is the shortfall, and the duration is the threshold-to-threshold time:
\[ D = 100\% - U_{\text{retained}} \qquad\qquad t_{\text{dip}} = t_{\text{end}} - t_{\text{start}} \]
- \(U_{\text{retained}}\)
- lowest RMS voltage during the dip (per cent of reference)
- \(D\)
- dip depth
- \(t_{\text{start}},t_{\text{end}}\)
- times the RMS crosses the start and recovery thresholds
For example, \(U_{\text{retained}}=55\%\) gives \(D=45\%\). The dip starts when the RMS falls below the dip threshold (commonly \(90\%\)) and ends when it rises above a slightly higher recovery threshold (commonly \(91\%\)); the small hysteresis avoids repeated start-stop counting when the voltage hovers near the threshold.
Section 4
Threshold and reference selection
The threshold strongly affects both counting and duration: an event is only counted if the voltage falls below it, and because the dip envelope is rarely rectangular, the same event can have a different measured duration at the \(90\%\) and \(80\%\) thresholds. The reference voltage matters too:
Table 2 — Reference voltage options.
| Reference Voltage | Meaning |
|---|
| Nominal voltage | Standard system voltage |
| Declared voltage | Voltage agreed or declared for the supply |
| Sliding reference | Pre-event reference that follows slow voltage variation |
A complete dip description = reference voltage + dip threshold + retained voltage + duration.
Section 5
Multi-phase voltage dips
A dip may affect one, two or three phases, and a single-phase fault produces a different pattern on line-to-neutral and line-to-line voltages. To avoid counting one disturbance as three events, overlapping per-phase dips are treated as a single event, and the retained voltage is the lowest \(U_{\text{rms}(1/2)}\) on any channel during the event.
Table 3 — Phase measurement configuration.
| Measurement | Relevance |
|---|
| Line-to-neutral voltage | Useful for LV and four-wire systems |
| Line-to-line voltage | Often used for MV/HV and customer compatibility |
| All three phases | Needed to identify the unbalanced dip type |
| Lowest phase value | Common basis for retained-voltage reporting |
Section 6
Additional dip characteristics
Retained voltage and duration are the headline quantities, but two dips with identical magnitude and duration can still affect equipment differently. Other characteristics matter:
Table 4 — Additional dip characteristics.
| Characteristic | Why It Matters |
|---|
| Phase-angle jump | Can disturb converters, drives and synchronous equipment |
| Point-on-wave | Important for some electronic and magnetic equipment |
| Number of affected phases | Determines the equipment response |
| Pre-event voltage | Affects the actual remaining voltage margin |
| Voltage recovery shape | Slow recovery may affect contactors and drives |
| Distortion during the dip | Faults and transformer saturation may distort the waveform |
| Dip unbalance | Sensitive three-phase loads respond differently |
Magnitude-duration reporting is therefore practical, but it is not a complete description of every equipment effect.
Section 7
Dip energy and dip severity
Two single-event indices reduce a dip to one number. Dip energy accumulates the voltage shortfall over the event:
\[ E_{VS}=\int_{0}^{T}\left(1-V(t)^2\right)dt \]
- \(V(t)\)
- RMS voltage during the event, in per unit
- \(T\)
- duration of the event below the dip threshold
A deeper or longer dip gives higher energy; the index is a single number but can be dominated by one long event and does not always track equipment behaviour. Dip severity instead compares the event against a reference tolerance curve such as the SEMI curve:
\[ \text{dip severity}=\frac{\text{actual dip severity}}{\text{reference tolerance severity}} \]
A value greater than 1 means the event is more severe than the reference curve. It combines magnitude and duration into one figure, but depends strongly on the chosen curve — one suited to a given type of equipment may not suit another.
Section 8
SARFI indices
SARFI — the System Average RMS Variation Frequency Index — counts the events per year that exceed a chosen severity criterion. The common voltage-threshold form is:
\[ SARFI_X = \text{events per year with } U_{\text{retained}} < X\% \]
- \(SARFI_X\)
- events per year more severe than the threshold \(X\)
- \(X\)
- retained-voltage threshold (per cent)
- \(U_{\text{retained}}\)
- lowest RMS voltage reached during the dip
so \(SARFI_{70}\) is the number of events per year with retained voltage below \(70\%\). A curve-based form, \(SARFI_{\text{SEMI}}\), instead counts events more severe than a predefined curve such as SEMI or ITIC. SARFI is simple and lets sites, systems and years be compared easily, but a pure voltage threshold loses duration information — a \(60\%\) dip lasting 3 cycles and one lasting 500 ms both count in \(SARFI_{70}\), though their equipment impact may differ greatly.
Section 9
Magnitude-duration tables, charts and areas
A magnitude-duration table groups events by retained-voltage rows and duration columns, each cell holding a count, \(N(U_{\text{retained}},\,t_{\text{duration}})\):
\[ \text{rows: }90\%\text{–}80\%,\;80\%\text{–}70\%,\;70\%\text{–}60\%,\ldots \qquad \text{columns: }0.1\text{–}0.25\,\text{s},\;0.25\text{–}0.5\,\text{s},\ldots \]
Tables preserve far more information than a single number and compare readily with equipment tolerance curves, though many cells can make site-to-site comparison harder. A voltage-sag coordination chart goes further, plotting supply performance (events per year) against equipment tolerance on the retained-voltage / duration plane — answering “how often will dips exceed my equipment’s tolerance?”:
Table 5 — Reading a coordination chart, and dip performance areas.
| Region | Interpretation |
|---|
| Dip above the equipment curve | Equipment should normally ride through |
| Dip below the equipment curve | Equipment may trip or malfunction |
| Shallow short dips | Common system events |
| Deep short dips | Close-up faults |
| Long-duration dips | Delayed protection or abnormal recovery |
| Very deep, long dips | Possible protection or system problem |
Grouping dips into performance areas links categories to likely causes: a short dip retaining \(70\%\)–\(80\%\) is expected in normal operation, while a long dip below \(40\%\) may indicate a close fault or delayed clearance.
Section 10
Equipment sensitivity
Dips matter because much equipment cannot ride through them:
Table 6 — Typical equipment response to dips.
| Equipment | Possible Dip Response |
|---|
| Contactors and relays | Drop out |
| Variable-speed drives | DC-link undervoltage trip |
| PLCs and control systems | Reset or malfunction |
| Computers and IT equipment | Shutdown or reboot |
| Process equipment | Production interruption |
| Motor loads | Torque reduction and stalling risk |
| Power-electronic converters | Current limit or trip |
Dip impact depends on retained voltage and duration — a \(50\%\) dip for 20 ms may be fine, while a \(70\%\) dip for 1 s may trip.
Section 11
Why dip objectives are difficult, and the one-year period
Unlike harmonics or flicker, dips resist universal limits because they depend on the network and its environment:
Table 7 — What dip performance depends on.
| Factor | Influence |
|---|
| Network topology | Fault locations and voltage propagation |
| Fault rate | Event frequency |
| Protection clearing time | Dip duration |
| Earthing arrangement | Phase voltages during faults |
| Overhead vs underground | Fault rate and dip frequency |
| Weather and environment | Fault occurrence |
| Equipment sensitivity | The practical impact |
So standards tend to define measurement and reporting methods rather than strict universal limits. And because events are random, dips must be assessed over a long period — typically at least a year. One quiet week may show no dips at all; one stormy week may overstate the rate. The meaningful index is the annual frequency — the number of dips per year, ideally in each magnitude-duration category — supported where needed by fault statistics and simulation.
Section 12
Objectives in practice, and comparing indices
Because strict universal limits are hard to set, objectives are usually expressed as expected performance or a reporting method:
Table 8 — Practical dip objective types.
| Objective Type | Meaning |
|---|
| Dips per year | General supply performance |
| Magnitude-duration table | Number of dips in each severity range |
| SARFI values | Events below selected voltage thresholds |
| Curve-based index | Events below SEMI, ITIC or a customer curve |
| Contractual reporting threshold | Utility reports dips below a stated voltage and duration |
| Site-specific target | Based on customer equipment immunity |
A contract might, for example, report only events with \(U_{\text{retained}} < 70\%\) and \(t > 600\ \text{ms}\) — not because shallower dips do not occur, but because that threshold defines contractual attention. No single index is perfect:
Table 9 — Strengths and limits of the main indices.
| Method | Advantage | Limitation |
|---|
| SARFI | Simple site/year comparison | Duration may be lost |
| Magnitude-duration table | Clear and widely understood | Can become too detailed |
| Coordination chart | Direct link to equipment tolerance | More complex to use |
| Dip energy | One number, depth and duration | May be dominated by one event |
| Dip severity | Combines magnitude and duration | Depends on the chosen curve |
For supply comparison, SARFI or magnitude-duration tables work well; for industrial customers, tolerance curves and coordination charts are more meaningful; for contracts, a selected voltage and duration threshold is used.
Section 13
Site indices and system indices
Dip data is processed at rising levels — from a single event, to one site, to a whole network:
\[ \text{sampled voltage}\;\rightarrow\;\text{single-event index}\;\rightarrow\;\text{site index}\;\rightarrow\;\text{system index} \]
Table 10 — Index levels with examples.
| Level | Example |
|---|
| Single-event index | Retained voltage and duration |
| Site index | Number of dips per year at one site |
| System index | Average SARFI across multiple sites |
| Equipment index | Dips below an equipment tolerance curve |
The distinction matters because a good site can be hidden inside a poor system average, and a healthy system average can hide a poor individual site.
Section 14
Mitigation of voltage dips
Dips cannot be eliminated, because faults and switching are part of normal operation — but their frequency, duration or impact can be reduced, from the network side or the equipment side:
Table 11 — Voltage dip mitigation measures.
| Mitigation Method | Purpose |
|---|
| Faster protection clearing | Reduces dip duration |
| Network automation | Reduces exposure time |
| Fault-reduction / undergrounding | Reduces event frequency and fault rate |
| Dedicated feeder | Reduces exposure to other customers’ faults |
| Stronger supply point | Reduces dip depth for some events |
| Dynamic voltage restorer | Protects sensitive loads |
| UPS systems | Maintains supply to critical equipment |
| Drive ride-through / contactor devices | Reduces nuisance trips and dropout |
| Process immunity improvement | Reduces production loss |
Network mitigation reduces the dips; equipment immunity reduces the consequences — for industry, improving ride-through is often the most cost-effective.
Section 15
Interpretation and key message
The practical assessment runs straight down a chain, from the sampled waveform to a site or system index:
\[ v(t)\;\rightarrow\;U_{\text{rms}(1/2)}\;\rightarrow\;\text{event detection}\;\rightarrow\;U_{\text{retained}},t\;\rightarrow\;\text{classification}\;\rightarrow\;\text{site / system index} \]
A useful single-event statement gives retained voltage, duration, location, reference and affected phases — for example \(U_{\text{retained}}=48\%,\ t=180\ \text{ms}\), line-to-line at the PCC — which says far more than “a voltage dip of 52%”. For annual performance, a statement such as \(SARFI_{70}=6\) events/year, or “4 events/year below the SEMI curve”, conveys the supply performance directly.
Voltage dip performance = event frequency + retained voltage + duration + equipment immunity.
Key message
Voltage dips are event-based reductions in RMS voltage, characterised by retained voltage \(U_{\text{retained}}=\min\!\big(U_{\text{rms}(1/2)}\big)\) and duration \(t_{\text{dip}}=t_{\text{end}}-t_{\text{start}}\), with depth \(D=100\%-U_{\text{retained}}\). They are counted below a threshold (commonly \(90\%\)), with interruptions below about \(10\%\), and — unlike harmonics, flicker and unbalance — must be assessed over roughly a year because they are driven by faults and protection. Report them with the index that fits the purpose: \(SARFI_X\), magnitude-duration tables, coordination charts, dip energy or dip severity. A robust assessment must state the measurement method + thresholds + reference voltage + assessment period + reporting index + assessment location + equipment tolerance — only then can dip performance be compared with process requirements or network expectations.
