Power Quality & Harmonics
Monitoring produces a large volume of data; to be useful it must be reduced to a small number of representative indices — simple enough to apply consistently, yet detailed enough to keep the technical meaning. This guide sets out the recommended indices for harmonics, flicker, unbalance, dips and interruptions, separating the internal planning indices the operator uses to control emissions from the voltage-characteristic indices used to report supply quality.
The aim is not many complicated indices, but as few as possible without losing essential information. A good power-quality index should be:
so that it can be compared across sites, years and operating conditions, and used to report performance to customers, regulators and operators.
Section 1
Indices convert raw measurement data into values that can be compared with an objective — a planning level, voltage characteristic, contract, regulatory target or equipment criterion:
For steady-state disturbances — harmonics, flicker, unbalance — two categories are needed, with different purposes: planning indices and voltage-characteristic indices.
Section 2
Planning indices are the operator’s internal tool for controlling future disturbance — setting emission limits for large loads and evaluating the combined effect of many installations. They may be more detailed than reporting indices, because they must control both sustained levels and higher short-term emissions during start-up, bursts or process changes. They matter most when connecting disturbing installations:
| Disturbing Installation | Main Disturbance |
|---|---|
| Arc furnaces | Flicker, harmonics, unbalance |
| Large converters / rectifiers | Harmonics |
| Traction loads | Unbalance, harmonics, flicker |
| Large motor loads | Flicker, voltage dips |
| Wind and solar plant | Harmonics, flicker, voltage changes |
| Large single-phase loads | Unbalance |
Voltage-characteristic indices describe the quality customers can expect at their terminals under normal conditions. They are simpler — intended for customer, regulatory and management reporting and long-term tracking — and should not require excessive interpretation. The fundamental principle is that planning levels must leave margin below the characteristic:
because the total disturbance at a supply point is the sum of many contributions, and if every user reached the characteristic limit the combined level would exceed it:
Section 3
Indices are formed at two levels. A site index describes one location — a PCC, busbar or supply point. A system index describes a group of sites, a voltage level or a region:
The site index drives customer-specific assessment; the system index drives network-level reporting and long-term performance monitoring.
Section 4
Harmonic planning uses individual harmonic voltages and THD, based on a very short-time value \(U_{h,vs}\) (3-second RMS) and a short-time value \(U_{h,sh}\) (10-minute RMS). The recommended planning indices combine both, to control sustained levels and short bursts:
| Index | Meaning |
|---|---|
| \(U_{h,vs95,daily}\) | 95% daily value of the 3-second harmonic voltage |
| \(U_{h,sh99,weekly}\) | 99% weekly value of the 10-minute harmonic voltage |
| \(U_{h,vs99,weekly}\) | 99% weekly value of the 3-second harmonic voltage |
with the basic check:
The very short-time 99% value may be permitted to exceed the normal planning level by a factor set by the operator — recognising that brief emissions can be acceptable if they do not persist long enough to stress equipment or distort the system.
Section 5
For voltage characteristics the site index is simpler — the weekly 95% value of the 10-minute harmonic voltage, applied to each order and to THD:
The system index is the site-index value not exceeded by a high proportion of sites — calculated separately for each order, because the worst 5th-harmonic site is rarely the worst 7th- or 11th-harmonic site:
An alternative system index — the percentage of sites exceeding the objective — is simple and useful for regulatory reporting. In every case, harmonic accuracy depends on the whole measurement chain: at HV/EHV, capacitive voltage transformers can resonate within the harmonic range and distort the result, so the arrangement (VT type, frequency response, instrument class, aggregation, transient flagging) should be checked before the data is used, and maximum values treated with caution.
Section 6
For planning, the main index is \(P_{st}\), reported as \(P_{st95,weekly}\) and \(P_{st99,weekly}\), with the 95% value checked after allowing for transfer between voltage levels:
The operator may permit a higher 99% value by a set factor. \(P_{lt99,weekly}\) is not generally recommended for planning, because it can be skewed by voltage dips unless they are properly flagged and removed. For voltage characteristics the site index is the weekly 95% long-term value, which represents sustained exposure:
A routine flicker system index is usually not recommended, because flicker monitoring is concentrated at known problem sites and is not an unbiased sample:
And because emissions vary with process and network state, an important source — an arc furnace, say — may need several one-week surveys at different times of year to capture scrap composition, operating stage, fault level, compensation state and seasonal change. The rule: the measurement must capture the full operating behaviour of the disturbing process.
Section 7
Because unbalance, like harmonics, has thermal effects in machines, its planning indices follow the same philosophy, built on the negative-sequence factor:
The recommended planning indices are \(U_{\text{neg},vs95,daily}\), \(U_{\text{neg},sh99,weekly}\) and \(U_{\text{neg},vs99,weekly}\), checked as \(U_{\text{neg},\text{index}}\le U_{\text{neg},\text{planning level}}\); high or maximum values should be flagged for dips and swells. For voltage characteristics the site index is the weekly 95% value:
The system index is again the high-percentile site value, or the percentage of sites exceeding the objective. Unbalance measurement needs care: \(U_2\) is small relative to \(U_1\), so a small error in magnitude or phase angle — from VT inaccuracy, unequal burden or wiring — produces a large relative error. Use a fundamental-only, line-to-line measurement with accurate phase angle and a balanced burden.
Section 8
The recommended single-event indices are the retained voltage and the duration:
measured on all three phases (the dip may be unbalanced), with the start/end thresholds and reference voltage declared. Site indices are then built from all events over the monitoring period — for equipment compatibility, customer information, premium-power contracts and mitigation selection:
| Method | Use |
|---|---|
| Voltage-dip table | Events by retained voltage and duration |
| Contour chart | Event frequency on the magnitude-duration plane |
| SARFI index | Events below a selected voltage threshold |
| Curve-based count | Events more severe than ITIC, SEMI or a custom curve |
| Custom site index | Specific to a customer process or contract |
System indices aggregate site indices over a network — an average or a high-percentile (e.g. 95%) site value — optionally weighted by number of substations, customers, rated power, network category or voltage level. At HV/EHV, phase-to-phase measurement better represents what downstream equipment sees after transformer connections.
Section 9
Dips often cluster — an unsuccessful auto-reclose can produce several within seconds. Counting each separately overstates the total; aggregating too hard loses information. So closely related events are combined, for example taking the deepest retained voltage and the longest duration:
Future dip assessment may add phase-angle jump, point-on-wave, symmetrical-component voltages, dip type, origin (upstream or downstream) and cause classification — helping identify whether a dip came from a fault, motor start or transformer energisation, and improving prediction of equipment response, at the cost of more complex presentation.
Section 10
MV distribution uses customer-based indices — \(\mathrm{SAIFI}\), \(\mathrm{MAIFI}\), \(\mathrm{SAIDI}\) and \(\mathrm{CAIDI}\):
So SAIFI is sustained interruptions per customer, MAIFI momentary interruptions per customer, SAIDI the average annual interruption time per customer (minutes/customer/year), and CAIDI the average restoration time for customers who were interrupted.
Section 11
HV/EHV reporting is more complex; three philosophies are used, each answering a different question, and a utility should pick one and declare it:
| Philosophy | Main Question |
|---|---|
| Connection point performance | Did the transmission interface remain available? |
| End-customer performance | Did actual customers lose supply? |
| System energy performance | How much energy was not served? |
Connection-point indices — \(\mathrm{SAIFI}_{CPI}\), \(\mathrm{MAIFI}_{CPI}\), \(\mathrm{SAIDI}_{CPI}\), \(\mathrm{SAIRI}_{CPI}\) — describe the transmission interface, which may be interrupted even when the distribution network back-feeds the load. End-customer indices describe what customers actually experienced, but depend on downstream configuration and data. The two are not the same:
Section 12
The energy philosophy uses estimated energy not supplied, which represents severity better than a customer count when a single transmission customer is very large:
From this come the average interruption time, duration and frequency, and the energy availability:
System minutes express interruption severity relative to system size, and can be graded:
| Degree | System Minutes | Interpretation |
|---|---|---|
| 0 | Below 1 | Normally acceptable for some systems |
| 1 | 1 to 9 | Significant impact |
| 2 | 10 to 99 | Serious impact |
| 3 | 100 or more | Very serious impact |
Section 13
Interruption indices are only meaningful if the reporting basis is declared:
| Item | Why It Matters |
|---|---|
| Reporting philosophy | Connection point, end-customer or energy |
| Planned events included/excluded | Changes results significantly |
| Connection-point definition | Determines the event count |
| Momentary/sustained threshold | Defines MAIFI, SAIFI and SAIDI |
| Single-phase events included/excluded | Affects event frequency |
| Customer-caused events | Determines responsibility |
| Time aggregation method | Affects the event count |
| Load-shed events included/excluded | Affects ENS and system minutes |
As a rule, transmission reporting applies to transmission-caused events; customer- or distribution-caused events are excluded unless the transmission system failed to protect itself or interrupted third parties. Where customers hold reliability contracts, contracted interruptions may be excluded if only those customers are affected — but this must be declared, and for lower-than-standard contracts, dual reporting (including and excluding the contracted interruptions) avoids hiding genuine degradation.
Section 14
The recommended index set is compact and consistent: for harmonics, planning on \(U_{h,vs95,daily}\), \(U_{h,sh99,weekly}\), \(U_{h,vs99,weekly}\) and reporting on \(U_{h,sh95,weekly}\); for flicker, planning on \(P_{st95,weekly}\)/\(P_{st99,weekly}\) and reporting on \(P_{lt95,weekly}\); for unbalance, the same shape on \(U_{\text{neg}}\); for dips, the single-event \(U_{\text{retained}}\) and \(t_{\text{duration}}\) with SARFI / tables / contour charts; and for interruptions, \(\mathrm{SAIFI},\ \mathrm{MAIFI},\ \mathrm{SAIDI},\ \mathrm{CAIDI}\) at distribution and a declared philosophy at transmission.
Good indices reduce a flood of data to a few stable, comparable numbers — planning indices to control future emissions, voltage-characteristic indices to report supply quality, site indices for one point and system indices for the network. But an index is meaningful only when its definition, calculation basis and reporting context are declared. A robust reporting framework must always specify the index + measurement method + assessment period + statistical probability + site/system basis + included and excluded events + objective. Only then can power-quality performance be assessed consistently and compared fairly over time.
Eight-Part Technical Series
An eight-part guide to power-quality indices and assessment objectives — how harmonic voltage, voltage flicker, voltage unbalance, voltage dips and long interruptions are measured and indexed, how the measurement data is reviewed, and the recommended indices and objectives for planning and reporting.
The recommended indices for planning and reporting — harmonic, flicker, unbalance and dip indices, SAIFI/SAIDI/CAIDI, and planning levels vs voltage characteristics.
