-
Leon Zhang Sales Manager
-
Email: sale@relayprotect.com
-
Phone/WhatsApp: +86 13655813266

Radial, Ring and Network Power Distribution Topologies
Power distribution runs from the substation feeder to the meter: topologies, voltage levels, transformer impedance, earthing and grading.
What does power distribution cover and which topology fits?
Power distribution runs from a substation’s outgoing feeder to the customer’s meter: feeders at 6 to 35 kV, the distribution transformer and the low-voltage network after it. The boundary is a single breaker. Upstream is transmission, built for bulk transfer; downstream is distribution, built to deliver to many small loads. That shift changes the goal from stability to selectivity and from redundancy to restore-by-switching, because a feeder has hundreds of customers along its length. IEC 60038:2009 fixes 11 kV, 22 kV and 33 kV on most overhead feeders and 400/230 V secondary in Europe, Asia and Africa, against 208/120 V or 480/277 V in North America. Short-circuit levels follow IEC 60909-0:2016; device numbers 50/51, 51N, 67 and 79 follow IEEE C37.2. Networks are radial, ring or a mixture. A radial feeder has one source-to-load path, is cheap to protect, and drops everything downstream until the fault is found. Setting review and quotation within 12 hours, MOQ 1, 7-15 working days.
Power distribution is the part of the network between a substation’s outgoing feeder and the customer’s meter: primary feeders at 6 to 35 kV, the distribution transformer, and the low-voltage network that follows it. IEC 60038:2009 fixes the nominal voltages on both sides of that transformer, and most design arguments on a distribution system come back to it.
At a glance
- Distribution starts where transmission ends, at the substation’s outgoing feeder breaker, not at the property line.
- Two voltage bands matter: primary at 6 to 35 kV and secondary at 400/230 V or 208/120 V.
- Radial, ring and network are the only three topologies in general use. Everything else is a variant.
- A distribution transformer’s impedance percentage fixes the fault current at its terminals. It is a protection setting, not a catalog detail.
- Earthing arrangement decides whether a single earth fault trips anything at all.
- Grading margin between a fuse, a recloser and a feeder relay is what stops one fault taking out a whole feeder.
- Distributed generation changes the voltage profile along a feeder. It does not change the voltage level.
- Most distribution failures are joint failures, not equipment failures. The joint is where the money goes.
What power distribution covers, and where it stops
The boundary is a single breaker. Everything upstream of the substation’s outgoing feeder breaker is transmission or sub-transmission, designed for bulk transfer between nodes. Everything downstream is distribution, designed to deliver to many small loads spread over geography.
The boundary explains the design.
That difference in purpose drives every difference in design. A transmission line has one customer at each end and a fault is a system event. A distribution feeder has hundreds of customers along its length and a fault is a local inconvenience, so the design goal shifts from stability to selectivity, and from redundancy to restore-by-switching.

If you are looking for the equipment inside the fence, that is a different article on substation design and protection. This one picks up at the feeder exit and follows the circuit to the meter.
Selecting a relay for a live project?
Send the system voltage, CT ratio and tripping scheme. Our engineers confirm the model and reply within 12 hours.
Voltage levels: where distribution begins and ends
IEC 60038:2009 sets the preferred nominal values, and the practical bands fall out of them. The power distribution band sits between 6 kV and 35 kV, with 11 kV, 22 kV and 33 kV carrying most of the world’s overhead feeders. Secondary distribution is 400/230 V in most of Europe, Asia and Africa, and 208/120 V or 480/277 V in North America.
| Band | Nominal value | Typical fault level | What it feeds |
|---|---|---|---|
| Sub-transmission | 66 to 132 kV | 20 to 40 kA | Distribution substations |
| Primary distribution | 11, 22, 33 kV | 10 to 25 kA | Distribution transformers, MV customers |
| Secondary distribution | 400/230 V | 20 to 50 kA | LV boards, domestic and commercial loads |
| North American secondary | 208/120 V, 480/277 V | 10 to 65 kA | Same duty, different convention |
Table 1. The four bands a distribution engineer actually works in. Fault levels are typical, not limits; the transformer and the source impedance set them.
The step between the two distribution bands is where the design language changes. Above the transformer, clearances and insulation coordination dominate. Below it, the fault current is set almost entirely by the transformer, and the protective devices are chosen to survive it rather than to interrupt a system fault.
Radial, ring and network: three topologies
Every distribution network in service is one of three shapes, or a mixture of them on the same system.
Three shapes, one trade.
A radial feeder has one path from the source to each load. It is cheap, it is simple to protect, and a fault anywhere on it drops everything downstream until someone finds the fault. Most rural overhead feeders are radial, and most industrial spurs are too.
A ring main runs the feeder out and back, with a normally open point in the loop. A fault is isolated by opening the two switches either side of it, and the healthy sections are restored from the other end. The customers between those switches stay off until the cable is repaired; everyone else is back within minutes.

A network, in the strict sense, has two or more sources permanently connected in parallel through a mesh. Nothing needs switching when a cable fails, because the fault is cleared by the nearest protective devices and the rest of the mesh keeps working. City centre networks and some hospital and data centre systems work this way. The price is directional protection at every node and a fault study that has to be redone whenever the mesh changes.
| Topology | Paths to a load | Restore time after a fault | Protection demand | Where it is used |
|---|---|---|---|---|
| Radial | one | repair time | overcurrent only | rural overhead, spurs, most LV |
| Ring, open point | two, one open | switching time, minutes | overcurrent plus a switch plan | urban MV cable, industrial estates |
| Ring, closed | two, both closed | none for healthy sections | directional or differential | dense urban, campus |
| Network | two or more | none | directional, network protectors | city centres, critical facilities |
Table 2. Closed rings are rare outside dense city networks because the circulating current has to be managed, not just the fault current.
Primary distribution: feeders, reclosers and sectionalizers
An overhead primary feeder is protected by a recloser at the substation exit and by one or more sectionalizers further out. The recloser opens on a fault, waits, and closes again. Most overhead faults, perhaps 70 to 80 per cent of them, are transient: a branch, a bird, a flashover that clears itself. One or two reclose attempts restore those without a crew.
Sectionalizers count the fault current passages and open after a set number, so that a permanent fault is isolated to the smallest possible section while the recloser keeps the rest alive. The coordination between the two is a counting exercise, and it exists in a narrow window. Set the counts wrong and either the sectionalizer never operates or it opens on a transient and adds an outage that did not need to happen.
Underground cable feeders get no reclosing.
A cable fault is permanent by nature, and reclosing into it does damage at the fault point that turns a repair into a joint replacement. Cable feeders are switched, tested and then restored manually.
Secondary distribution: the low-voltage network
Below the distribution transformer, the design problem changes completely. The distances are short, the currents are large, and the fault level is high enough that a loose termination is a fire risk rather than a nuisance.
Voltage drop, not current-carrying capacity, usually sets the cable size on a long LV run. A 100 A load at the end of 150 m of 35 mm² copper loses roughly 8 V on the phase and neutral together, which is 3.5 per cent before anything else is connected. Add the transformer’s own regulation and the last customer on the run is outside the ±10 per cent the supply standard allows.
This is where contactors and solid-state switching start to matter, because a large single-phase load switching on a weak LV network produces a voltage step that other customers see as flicker. The flicker limit, not the thermal limit, decides whether a welding set can be connected at the end of a rural spur.
Distribution transformer sizing: kVA, vector group and losses
A distribution transformer is chosen on four numbers: kVA rating, voltage ratio, impedance percentage and vector group. The nameplate carries all four, and the one most often ignored in a tender is the impedance.
Rating follows the load, with a diversity factor applied to the connected load rather than the sum of it. A 630 kVA unit typically serves 150 to 250 domestic customers. Oversizing to 1,000 kVA to leave headroom raises the no-load loss for the life of the unit and raises the fault level on the LV board, which then has to be rated for it.
Vector group decides whether the transformer can be paralleled with a neighbour and how the zero-sequence current behaves. Dyn11 is the default for distribution units in most of the world, because the delta primary blocks triplen harmonics from the LV side and the earthed star secondary gives a neutral for single-phase loads.
Distribution transformers: what the impedance percentage really fixes
Impedance percentage is the voltage that has to be applied to the primary, at rated current on short circuit, expressed as a percentage of rated voltage. It is measured, it is stamped on the nameplate, and it is the single number that sets the fault current at the secondary terminals.
The arithmetic is short enough to do in your head. Rated secondary current is kVA times 1,000 divided by root three times the line voltage. Short-circuit current is that figure multiplied by 100 and divided by the impedance percentage.

| Rating | LV current at 400 V | Fault current at 4 % | Fault current at 6 % | Typical duty |
|---|---|---|---|---|
| 315 kVA | 455 A | 11.4 kA | 7.6 kA | small commercial, one street |
| 630 kVA | 909 A | 22.7 kA | 15.2 kA | residential block |
| 1,000 kVA | 1,443 A | 36.1 kA | 24.1 kA | mixed commercial |
| 1,600 kVA | 2,309 A | 57.7 kA | 38.5 kA | industrial estate |
| 2,000 kVA | 2,887 A | 72.2 kA | 48.1 kA | heavy industrial, dedicated |
Table 3. Fault current at the secondary terminals, before cable impedance is added. Real boards see less, sometimes 30 per cent less at the far end of a long run.
The choice is a trade, and it is made early. A 4 per cent unit gives better voltage regulation and a lower loss of life under motor starting. It also puts 36 kA on a board that might have been specified at 25 kA, and every breaker, fuse and busbar in the LV assembly has to be rated for it. Specifying 6 per cent to keep the fault level down is normal practice on a board that feeds mostly motor load.
Power distribution equipment: what sits at each node
The equipment list is short and it repeats at every voltage step. What changes is the rating and the duty.
| Node | Power distribution equipment | Function | Failure that hurts most |
|---|---|---|---|
| Substation exit | Feeder breaker, CTs, VTs, relay | Fault clearance, measurement, reclosing | Relay settings drift after a network change |
| Along the feeder | Reclosers, sectionalizers, load-break switches | Sectionalising, isolation for work | Switch not operated for years seizes |
| Transformer point | RMU or switch-fuse, transformer, LV board | Step down, protect, distribute | Fuse chosen on rating, not on coordination |
| LV distribution | ACBs, MCCBs, contactors, busbars | Distribution and motor control | Thermal damage at a loose joint |
| Customer point | Meter, main switch, consumer unit | Metering and final protection | Neutral and earth swapped at the terminal |
Table 4. Five nodes, five duties. A ring main unit’s interlocks belong in this list too: they are the reason a cable can be worked on safely.
Two deserve a note. A ring main unit’s interlock scheme prevents an earth switch being closed onto a live cable, and it is tested at commissioning and then rarely again. A recloser’s settings live in a small PLC-class controller nobody logs into between faults, which is how a feeder ends up reclosing into a permanent cable fault three times before anyone notices.
Voltage drop and losses: the arithmetic that decides cable size
Voltage drop on a distribution circuit is the current times the impedance, and on a low-voltage run the resistance dominates. Three-phase drop is root three times current times the per-phase impedance times length. Single-phase drop is twice that per unit length, because the return current also sees the neutral.
Worked once, it stops being abstract. A 200 A three-phase load at 400 V, 180 m from the board, on 95 mm² copper at 0.193 Ω/km: drop is 1.732 × 200 × 0.193 × 0.18, about 12 V, or 3 per cent. Change to 70 mm² at 0.268 Ω/km and the drop becomes 16.7 V, or 4.2 per cent, which is still inside the limit but leaves nothing for the transformer’s own regulation.
Size follows drop, not current.
Technical losses in a distribution network run at 4 to 8 per cent of the energy delivered in most systems, and roughly half of that is in the LV network and the transformers rather than the MV feeders. Loss is proportional to current squared, so a feeder running at half load has a quarter of the loss, and reconfiguring a ring’s open point to balance the two legs often saves more than replacing cable.
Earthing arrangements: TN, TT and IT
The earthing arrangement decides what happens on the first earth fault, and it is the item most often copied from the last project without checking.
| System | Neutral at source | Exposed metal parts | Earth fault current path | Typical use |
|---|---|---|---|---|
| TN-S | earthed | to a separate PE run throughout | metallic, high, trips fast | campus and industrial |
| TN-C-S | earthed | to a PEN split at the origin | metallic, high, trips fast | most European domestic supply |
| TT | earthed | to a local electrode only | through soil, low, needs RCD | rural overhead, farms |
| IT | isolated or high impedance | to a local electrode | very low, first fault alarms only | hospitals, process plant |
Table 5. The four arrangements. A TT system without an RCD will not clear an earth fault at all, which is the single most common earthing defect we see on rural supplies.
IT systems are the exception.
The first earth fault does not trip; it raises an alarm and the plant keeps running. That is the point in a hospital theatre or a continuous process, and it only works if someone investigates before a second fault arrives.
Protection on a power distribution feeder
Feeder protection is graded, and the grading is what most commissioning time is spent on. Three layers usually sit in series: the substation feeder relay, a recloser part way out, and a fuse at the transformer.
The device numbers follow IEEE C37.2-2022, and only a few of them appear on a distribution feeder: 50 for instantaneous overcurrent, 51 for time overcurrent, 51N for the neutral element, 67 for directional, 79 for reclosing, and 27 and 59 for the voltage elements that supervise them. A relay that carries 50 and 51 with a reclose function is enough for a simple radial feeder.

The margin is usually quoted as 0.2 to 0.3 s between adjacent curves at the maximum fault current they share. That gap covers the breaker’s own operating time, the relay’s overshoot and a tolerance for the current transformers. Below about 0.2 s the curves overlap in practice even when they look separated on the drawing, and the device that trips first is decided by tolerances nobody wrote down.
Transformer inrush is the trap. Energising a distribution transformer draws 8 to 12 times rated current for around 0.1 s, and a 51 element set fast enough to grade with a downstream fuse will see that as a fault and trip the feeder. The usual fixes are a harmonic restraint, a short definite-time block on the second harmonic, or simply a slower curve with a higher pickup than the load requires.
At the transformer itself, a fuse is often the whole protection. That works when the fuse is selected against the transformer’s inrush and its through-fault capability, not just its kVA. Pick a fuse on kVA alone and it will either blow on energisation or fail to protect the winding.
Larger or critical transformers get differential protection (87T) instead, which is a different discipline and not usually applied below a few MVA on a distribution system. Our notes on breaker ratings and trip curves cover the device side of this.
Fault levels and the grading margin
Fault level on a power distribution feeder is a property of the source, not of the load, and it falls as you move away from the transformer. Calculating it properly means IEC 60909-0:2016 and the source impedance, the transformer impedance and the cable impedance in series.
For most distribution work a simpler statement is enough. At the LV terminals the transformer dominates, so the fault level is roughly the number in Table 3. At the far end of a long feeder, cable impedance can cut that by 30 to 50 per cent, and the grading you calculated at the board no longer holds at the last pillar.
That is why grading studies are done at both maximum and minimum fault level: a setting that looks comfortable at the substation can fail at the remote end, and the minimum case decides whether the fuse operates at all.
Distributed generation on a power distribution feeder
A feeder designed for one-way flow does not automatically work with generation at the far end. IEEE 1547-2018 sets the interconnection requirements, and the distribution engineer’s problems start where the standard stops.
The first is voltage rise. A feeder with 20 per cent of its capacity in rooftop generation can push the voltage at the far end above the statutory limit at midday, because the current that used to drop voltage now raises it. Tap changers on the transformer respond to the wrong signal, and the answer is usually a voltage regulator or an active network management scheme rather than more copper.
The second is protection. A fault on the feeder is now fed from both directions. A recloser that was correctly graded for a radial feeder sees current it was never set for, and an anti-islanding function may disconnect generation that the network actually needed. Directional elements (67) and a reviewed grading study are the minimum response.
The third is that the fault level is no longer a constant. Inverter-based generation contributes current that is limited by its controls, not by physics, so the fault level at a given point depends on how many inverters are online. Protection settings that assume a fixed fault level become a range.
Power quality on the low-voltage network
Three limits do most of the work on a low-voltage power distribution network: voltage magnitude, voltage distortion and flicker. EN 50160:2019 sets the first and the third for European public supply, and IEEE 519-2022 sets the harmonic limits at the point of common coupling.
IEEE 519-2022 is worth reading for one structural point. Voltage distortion limits are absolute, 8 per cent total harmonic distortion below 1 kV, while current limits scale with the ratio of short-circuit current to load current. A weak feeder therefore gets a tighter current allowance for the same load, which is the opposite of what a customer usually assumes.
The practical LV consequence is the neutral. Triplen harmonics from single-phase electronic loads add in the neutral rather than cancelling, so a neutral sized for balanced fundamental current can carry more than its phase conductors. That is a design error that only appears years after commissioning, when the load mix has changed.
A surge protective device at the LV board is the other half of power quality, and it is a different subject: clamping is covered in our guide to surge protection and where each type belongs. The varistor in that device degrades with every operation, which is why the status window is worth reading during the annual inspection.
How distribution networks fail
Failure statistics on power distribution systems are remarkably stable across countries, and they point somewhere other than the equipment catalog.
| Failure | Mechanism | Symptom | Correction |
|---|---|---|---|
| Joint failure | Moisture ingress or a poor crimp | Intermittent earth fault, then a permanent one | Remake the joint, test before re-energising |
| Cable fault | Tree root, dig-in, water treeing in aged XLPE | Earth fault that does not clear on reclose | Locate, cut out, joint or replace the section |
| Transformer failure | Overload in summer, or through-fault duty | Gas relay trip, oil discolouration | Load check and through-fault review, not just a swap |
| Fuse fatigue | Repeated transient currents near the rating | Nuisance blows in hot weather | Review the fuse curve, not the fuse size |
| Voltage complaint | Load growth at the end of a long LV run | Lights dim at the last house only | Reconductor, or move the open point |
Table 6. Five faults that account for the majority of unplanned interruptions on a distribution network.
The first row is the one that gets underestimated. A joint that has taken in water does not fail cleanly. It arcs, clears, arcs again, and each episode leaves a little more carbon in the insulation until the fault is permanent. By then the cable has usually been dug up twice looking for it.
When the textbook answer is wrong
Five assumptions survive in design offices long after they stop being true.
- A ring is not a duplicate supply while the open point is open. Between the fault and the open point, the customers are off until the cable is repaired.
- Impedance percentage is not a loss figure. A low-impedance transformer is more efficient at full load but it raises the fault duty on everything downstream, and that cost lands on the LV board.
- Headroom is not free. Oversizing a transformer for future load increases no-load loss every hour of its life and can force a higher LV fault rating today.
- A higher fault level is not automatically safer. More fault current means faster clearance, but it also means the equipment has to survive the energy, and an under-rated board fails explosively rather than gracefully.
- An earthing arrangement cannot be copied between sites. TT without an RCD does not clear a fault, and TN-C-S on a supply with a long PEN depends on a bond you do not control.
Each of those is a scope decision, and each is cheaper to settle at the design stage than after the board has been built and the setting sheets issued.
Frequently asked questions
What is power distribution in simple terms?
It is the last stage of the electricity supply chain: the feeders leaving a substation, the transformers that step the voltage down, and the low-voltage cables that reach individual customers. Transmission moves bulk power between regions; distribution delivers it to many small loads across a service area.
What is the difference between transmission and distribution?
Voltage and purpose, mainly. Transmission runs at 110 kV and above to move power between nodes and has few customers at each end. Distribution runs at 6 to 35 kV and then at 400/230 V, has hundreds of customers along each feeder, and is designed to be restored by switching rather than to stay stable as a system.
Why is 11 kV so common for distribution?
It is the lowest standard value that still keeps current and voltage drop manageable over a few kilometres. Going to 33 kV halves the current for the same power and extends the reach, but it needs more clearance and more expensive switchgear. Rural systems go higher, dense urban systems often stay at 11 kV and shorten the runs instead.
Does a higher impedance transformer reduce fault current?
Yes, and that is often the reason for choosing it. A 1,000 kVA unit at 6 per cent passes about 24 kA at its 400 V terminals, against 36 kA at 4 per cent. The trade is slightly worse voltage regulation and slightly higher load loss, which is usually acceptable on a board that feeds motors.
Which earthing system should a distribution network use?
TN-C-S is the usual choice where the supply authority provides a reliable PEN, because it clears earth faults with overcurrent devices alone. TT is used where the soil and the supply make a metallic return unreliable, but it needs an RCD to clear faults at all. IT is reserved for installations that must keep running after a first earth fault.
What are typical losses in a distribution network?
Four to eight per cent of the energy delivered, split roughly evenly between the MV feeders and the transformers plus LV network. Loss scales with current squared, so balancing load between feeder sections usually saves more than reconductoring the same sections would.
Can a distribution feeder accept rooftop solar?
Usually, up to a point. The limit is voltage rise at the far end and the effect on protection grading, not the capacity of the cable. Above roughly 15 to 20 per cent of feeder capacity in inverter-based generation, most networks need a voltage regulator, a reviewed grading study or active management.
Conclusion
Power distribution is a chain of trades, each made at a different voltage. The topology decides how long an outage lasts, the transformer impedance decides how hard the fault is, the earthing arrangement decides whether the fault trips anything, and the grading decides how much of the feeder goes dark with it.
Get those four right and the equipment catalog becomes a detail. Get any one of them wrong and no amount of good switchgear will cover for it, which is why the decisions that matter are all made before the first cable is pulled.
Related reading
- Substation design and protection, for the equipment upstream of the feeder exit.
- Circuit breakers and their ratings, for trip curves, frames and interrupting duty.
- Surge protection, for what happens at the LV board when a transient arrives.
- Arc fault breakers, for the failure mode that starts at a loose termination.
- Diesel generator protection, for the generation side of a standby arrangement.
Sources and standards
| Source | What it covers | Type | Edition |
|---|---|---|---|
| IEC 60038:2009 | Standard voltages for supply systems and equipment design | Standard | 2009, AMD1 2021 |
| IEC 60076-1:2011 | Power transformers, general requirements and rating | Standard | 2011 |
| IEC 60909-0:2016 | Short-circuit current calculation in three-phase AC systems | Standard | 2016 |
| IEEE C37.2-2022 | Device function numbers used in the protection sections above | Standard | 2022 |
| IEEE 1547-2018 | Interconnection of distributed energy resources | Standard | 2018 |
| IEEE 519-2022 | Harmonic control at the point of common coupling | Standard | 2022 |
| EN 50160:2019 | Voltage characteristics of public distribution systems | Standard | 2019 |
Table 7. Standards referenced above, with the edition used for the figures in this article.
Loss and voltage-drop figures come from our own commissioning records on 11 kV and 22 kV feeders and 400 V boards between 2023 and 2026. Fault levels in Table 3 are calculated from nameplate impedance with no cable contribution, so field measurements will read lower.
Related guides: Contactors · Annunciator · Generator interlock
Send the nameplate data if you want the settings checked before the panel is built.




