Engineering guide

    LV Radial Short-Circuit Network Analysis

    From MV upstream network to LV transformer, Main LV and downstream boards. Understand how each feeder changes fault current and why automatic cable data are only a starting point.

    KiloVolt Engineering Toolkit

    Faults become weaker along a feeder

    Every distribution-board location has its own prospective fault current. The upstream grid and transformer supply an initial driving voltage, but each downstream feeder adds resistance and reactance. A breaker at DB4 cannot simply use the Main LV short-circuit current as its complete fault assessment.

    1. Build the radial network

    Begin with Main LV → DB1. Add DB2 through DB5 as required. Each board contains one incoming feeder segment with independently editable length, conductor material, cross-section, cable arrangement, and parallel runs. Select any board to inspect its local maximum and minimum 3φ and 2φ results. The overview shows every board simultaneously.

    A radial feeder is one path from transformer to the selected fault; branching networks with parallel source paths are outside this initial implementation. The optional transformer-to-Main LV link may be a cable or a busduct with manufacturer-provided impedances.

    2. Upstream MV network and transformer

    Enter transformer kVA, MV and LV rated voltages, nameplate short-circuit impedance uk%, and R/X or rated load loss. The upstream network short-circuit level may be entered as MVA or kA on the MV side. A zero upstream value denotes an ideal source, not an actual measured network. The upstream network is converted to an impedance with the IEC 60909-0 voltage factor of the MV connection point (c = 1.10 for the maximum case and 1.00 for the minimum case) and referred to the LV side:

    ZQ=cQ UnQ2SkQ′′ (referred to LV: UnQ→ULV)Z_Q=\frac{c_Q\,U_{nQ}^2}{S''_{kQ}}\ \text{(referred to LV: }U_{nQ}\to U_{LV}\text{)}
    ∣ZT∣=uk100ULL2ST|Z_T|=\frac{u_k}{100}\frac{U_{LL}^2}{S_T}

    1. Maximum and minimum currents. The maximum current I″k,max decides what a breaker must be able to interrupt and what the cables must withstand. The minimum current I″k,min decides whether a protective device still trips fast enough at the far end of a circuit. They use different voltage factors and different assumptions about the network, so they are two separate calculations, not one result with a tolerance band.

    Ik,3ϕ′′=c ULL3 ∣Zk∣,Ik,2ϕ′′=c ULL2 ∣Zk∣,c=cmax or cminI''_{k,3\phi}=\frac{c\,U_{LL}}{\sqrt{3}\,|Z_k|},\qquad I''_{k,2\phi}=\frac{c\,U_{LL}}{2\,|Z_k|},\qquad c=c_{max}\ \text{or}\ c_{min}

    2. Why the LV voltage tolerance matters. The voltage factor c stands for how far the voltage at the fault location can sit above or below its nominal value. KiloVolt asks for the tolerance of the supply and sets both factors together. For the 2026-oriented profile it uses the pairs that DIgSILENT documents from IEC 60909-0:2026, section 5.3.1, Table 1:

    ±6 %: cmax=1.05, cmin=0.95±10 %: cmax=1.10, cmin=0.90\pm 6\,\%:\ c_{max}=1.05,\ c_{min}=0.95\qquad\qquad \pm 10\,\%:\ c_{max}=1.10,\ c_{min}=0.90

    The ±10 % pair is only the initial example on the screen. It does not mean that every Greek or other LV installation has a ±10 % tolerance. Any other combination is accepted as a user-selected what-if and is labelled SENSITIVITY CASE, including 1.10/0.95. That last pair is the one public sources print for ±10 % networks when they describe the 2016 edition (cmin = 0.95 for both tolerances), so the two editions are an open standards issue until the licensed 2026 text has been compared.

    3. How the factors change the result. The current is proportional to c. For the same impedance, going from the ±6 % to the ±10 % pair raises I″k,max by 1.10/1.05 and lowers I″k,min by 0.90/0.95. In the example network of this guide (1,600 kVA, 20/0.4 kV, 6 %, 250 MVA maximum and 100 MVA minimum upstream level) the maximum is 37.41 kA at ±6 % and 37.59 kA at ±10 %, and the minimum is 29.74 kA and 27.16 kA.

    4. From the nameplate to R and X. The nameplate gives the short-circuit impedance uk. With the rated load loss Pkr, the resistance follows from the losses; without it KiloVolt splits the impedance with the entered R/X.

    ∣ZT∣=uk100ULL2ST,RT=Pkr ULL2ST2,XT=∣ZT∣2−RT2|Z_T|=\frac{u_k}{100}\frac{U_{LL}^2}{S_T},\qquad R_T=\frac{P_{kr}\,U_{LL}^2}{S_T^2},\qquad X_T=\sqrt{|Z_T|^2-R_T^2}
    xT=XT STULL2(relative reactance, not simply uk)x_T=\frac{X_T\,S_T}{U_{LL}^2}\quad\text{(relative reactance, not simply }u_k\text{)}

    DIgSILENT words this term "short-circuit voltage of the transformer in p.u. (uk)", pandapower "transformer impedance relative to the rated values". KiloVolt uses the relative reactance. For a transformer with 6 % impedance and R/X = 0.15 the two differ by about 1 %, which matters only at the fourth decimal of KT; the wording difference is recorded as an open point.

    5. The transformer correction KT. It corrects the transformer impedance for the voltage-source method. cmax appears in it because the factor is defined for the maximum voltage case; DIgSILENT states that the factor formed with cmax is used even in minimum calculations. The sources do not give a physical reason, and KiloVolt does not claim one.

    KT=0.95 cmax1+0.6 xT,ZTK=KT ZTK_T=\frac{0.95\,c_{max}}{1+0.6\,x_T},\qquad Z_{TK}=K_T\,Z_T

    For minimum currents KiloVolt now applies KT,min = KT (formed with cmax), the treatment DIgSILENT documents for IEC 60909-0:2026, section 6.3.3. Public implementations disagree: pandapower 3.5.6, run during this work, applies no KT in its minimum case, and the v0.1.66 release of this tool used KT,min = max(1, KT). The two alternatives remain available as labelled sensitivity cases, and the screen shows the minimum current under all three. This is an open standards issue: no public source settles the normative text. For cmax = 1.05 the factor is always below 1, so applying it lowers the transformer impedance and raises the minimum current (29.74 kA against 28.87 kA without it in the example). A higher minimum current is the less conservative reading; use it knowingly.

    KT,min⁡=KT (default),KT,min⁡=max⁡(1,KT) (legacy),KT,min⁡=1 (no correction)K_{T,\min}=K_T\ \text{(default)},\qquad K_{T,\min}=\max(1,K_T)\ \text{(legacy)},\qquad K_{T,\min}=1\ \text{(no correction)}

    6. The upstream network. The network is replaced by a source impedance from its fault level S″k at the MV connection point, using the MV voltage factor of that point (1.10 for the maximum case, 1.00 for the minimum case) and referred to the LV side:

    ZQ=cQ UnQ2SkQ′′ (referred to LV: UnQ→ULV)Z_Q=\frac{c_Q\,U_{nQ}^2}{S''_{kQ}}\ \text{(referred to LV: }U_{nQ}\to U_{LV}\text{)}

    The minimum fault level can be much lower than the maximum (fewer generators or lines in service, other switching states), and the R/X of the network can differ too; pandapower keeps separate maximum and minimum values for both, and so does KiloVolt. If you enter only the maximum level, the minimum case reuses it, but with the MV factor 1.00 instead of 1.10 the source impedance is about 9 % lower, so the minimum current can be overstated. Entering a minimum level means source data provided; it does not verify the calculation model. A single R/X for both cases is a modelling assumption and is flagged.

    7. Why protection needs verified source data. A minimum current that is too high makes a protective device look more sensitive than it is. For a protection-sensitivity assessment the minimum source level, its R/X and the transformer treatment must come from the utility and from a standard-based study, not from a screening tool. KiloVolt marks every minimum without a supplied minimum level as NOT VERIFIED.

    8. Cables, temperature and missing sources. The built-in R20 values are maximum permitted resistances; real cables with lower resistance give a higher maximum current. For the minimum case the conductor is hotter and its resistance higher. Motors, generators and inverters add fault current that this screen does not include, so the maximum current can be understated for a real installation.

    3. Automatic R/X for each cable segment

    Select Cu or Al, single-core or multicore, section, length and identical parallel runs. Built-in R20 values are based on maximum Class 2 plain-conductor DC resistance at 20°C from IEC 60228:2023. This tabulated MAXIMUM conductor resistance is NOT a verified lower bound for the maximum-fault case; actual lower R20 produces higher possible Ik,max. A separate minimum R20 sensitivity input is available, but the complete fault-duty upper bound still requires verification. The reactance is an indicative geometry-based assumption: 0.08 Ω/km multicore, 0.09 Ω/km single-core touching, or 0.12 Ω/km single-core flat. Those X values are NOT standardized universal cable parameters.

    RT=R20 [1+α(T−20)]R_T=R_{20}\,[1+\alpha(T-20)]
    Zcable=L1000Nparallel(RT+jX)Z_{cable}=\frac{L}{1000N_{parallel}}(R_T+\mathrm j X)

    The parallel-run expression assumes equal effective impedance, equivalent installation, equal lengths and current sharing. Manufacturer AC resistance/reactance or measured data override the internal reference when available. A maximum prospective fault calculation cannot be certified by simply substituting IEC 60228's maximum DC resistance; the breaker screen never issues an unqualified PASS. For busducts, both R/X entries are mandatory because there is no universal busduct reference.

    4. Three-phase, phase-to-phase, maximum and minimum

    For every board the engine aggregates the source, corrected transformer and all preceding feeder impedances. Cold cable resistance and cmax give a maximum-fault screening case; hotter resistance and cmin give a minimum-fault screening case. The initial two-phase fault expression assumes the positive/negative sequence impedances are equal for the stated simplified radial model.

    ZΣ,k=ZQ+KTZT+∑j=1kZL,jZ_{\Sigma,k}=Z_Q+K_T Z_T+\sum_{j=1}^{k}Z_{L,j}
    Ik,3ϕ′′=cULL3∣ZΣ∣I''_{k,3\phi}=\frac{cU_{LL}}{\sqrt3\lvert Z_\Sigma\rvert}
    Ik,2ϕ′′≈cULL2∣ZΣ∣I''_{k,2\phi}\approx\frac{cU_{LL}}{2\lvert Z_\Sigma\rvert}

    Neither minimum result is the line-neutral or line-earth fault current. A phase-to-PE estimate for TN-S systems is given in section 7 and in the calculator; line-neutral faults are not modelled.

    5. Compare breaker interrupting and making capacity

    Enter Icu (ultimate breaking capacity, kA RMS), Ics (service breaking capacity, kA RMS) and Icm (rated short-circuit making capacity, kA peak) for the selected board. The designer compares them to local maximum RMS and indicative peak current. Icm is compared to the peak current, never the RMS kA. Ratings are only valid at their rated operational voltage Ue: without Ue the comparison is reported as NOT ASSESSED, and a Ue below the system voltage is a FAIL. The screen never issues a PASS: the modelled maximum fault is not a verified upper bound (the default cable data are upper limits), so a comparison in which no rating is exceeded is reported as NOT ASSESSED, with each comparison (Icu against Ik3,max, Icm against ip, Ue) listed. An Ics lower than the maximum fault current is listed as an advisory shortfall by default; tick the project-specification option to make it a FAIL when your client or specification requires Ics ≥ Ik. Otherwise an Ics shortfall is only advisory: it reduces service capability but is not a stated IEC 60364 requirement, whereas Icu or Icm below the fault level is a FAIL. Icu here stands for the rated breaking capacity (Icn for IEC 60898-1 devices); a manufacturer-verified coordinated backup arrangement can allow a lower standalone rating, which this screen does not assess. The board table and the selected-board panel show a badge: Fail when a rating is exceeded, otherwise Not assessed. Not assessed is never a pass; see how to read the results.

    ip=2 κ Ik′′,κ=1.02+0.98e−3R/Xi_p=\sqrt2\,\kappa\,I''_k,\quad\kappa=1.02+0.98e^{-3R/X}

    The calculator shows Ib = I″k for the modelled sources only (network feeder and transformer behave as far-from-generator sources); motor, generator and inverter contributions are not modelled, so this is not a verified breaking current. Trip time, time-current curves, I²t, protection coordination, selectivity, and other equipment fault contributions are NOT assessed in this simplified model. Compare with actual device manufacturer documentation before installation.

    6. Worked transformer example

    A 1,600 kVA, 400 V transformer with uk = 6% has:

    ∣ZT∣=40021,600,000×0.06=0.006 Ω|Z_T|=\frac{400^2}{1,600,000}\times0.06=0.006\ \Omega

    With R/X = 0.15, R≈0.000890 Ω and X≈0.005933 Ω before transformer correction. Adding a 100 m Cu 150 mm² cable provides a significant additional resistance compared with the transformer alone. The designer repeats the complex impedance sum for Main LV, DB1, DB2 and every subsequent panel.

    7. Phase-to-PE fault and the instantaneous trip

    For protection against indirect contact in a TN-S system the question is whether a fault between a phase conductor and the PE conductor is large enough to operate the protective device quickly. The calculator adds the PE conductor of every feeder to the minimum-fault impedance it already knows:

    Ik1,min=cmin⁡ U0∣Zloop∣,Zloop=ZQ+KTZT+∑j(Zph,j+ZPE,j)I_{k1,\mathrm{min}}=\frac{c_{\min}\,U_0}{|Z_{\mathrm{loop}}|},\qquad Z_{\mathrm{loop}}=Z_Q+K_TZ_T+\sum_{j}\left(Z_{ph,j}+Z_{PE,j}\right)

    U0 = ULL/√3. Without the optional transformer-to-Main LV feeder there is no PE conductor yet at the Main LV bus, and with source/transformer zero-sequence impedance assumed equal to positive-sequence impedance, this simplified model makes the phase-to-PE estimate equal the three-phase minimum at Main LV. This is a modeling result, not a general physical equality. If a cable feeder is configured between the transformer and Main LV, its PE conductor is part of the loop and the Main LV result differs from the three-phase minimum; if that feeder is a busduct, the PE return path is incomplete and no device verdict is given. Further down, the PE return path changes the result. The PE section is chosen automatically from IEC 60364-5-54 Table 54.2 only when the PE and phase materials match (same material: S up to 16 mm², 16 mm² for 16 to 35 mm², half of S above that, rounded up to a standard size) or entered by hand. Its resistance is taken at the stated minimum-fault temperature; its reactance is approximated as equal to the phase-conductor reactance. That approximation isnot guaranteed conservative for every return path or cable geometry. If the PE material differs, the same-material sizing shortcut cannot be used: the engineer must enter an independently verified PE section.

    PE routing. By default the PE runs in the same cable as the phase conductors. The assumption under the section heading can be switched to a separate PE conductor. Its loop reactance then follows from the geometry of two round conductors, with r′ = 0.7788 r the geometric mean radius of each conductor and an assumed centre-to-centre spacing d = 0.10 m that you do not enter:

    Xloop′=2πf μ02π[ln⁡dr1′+ln⁡dr2′],r′=0.7788 r,r=S/πX'_{\mathrm{loop}}=2\pi f\,\frac{\mu_0}{2\pi}\left[\ln\frac{d}{r_1'}+\ln\frac{d}{r_2'}\right],\qquad r'=0.7788\,r,\quad r=\sqrt{S/\pi}

    It is never taken below the same-cable value. The effect is large, and it depends on the spacing: for a 150 mm² feeder the minimum phase-to-PE current is about 8 % lower than with the PE in the same cable at 3 cm, 22 % lower at 10 cm and 32 % lower at 30 cm; for a 35 mm² feeder about 3 %, 6 % and 9 %. Steel supports, trays, armour and a PE on a different route are not modelled, and the device verdict stays NOT ASSESSED. The 10 cm spacing is a screening assumption, not a value taken from a standard.

    The return path, source and transformer zero-sequence network, and installation conditions must be independently established for an automatic-disconnection check. KiloVolt marks the device verdict NOT ASSESSED even if the indicative current reaches a selected magnetic-trip threshold.

    Instantaneous trip of an MCB

    The magnetic release holds below a lower multiple of the rated current In and must trip at an upper multiple. To check that a device will operate, the calculator compares its estimated current with the upper limit, similar to the threshold used in manufacturers' maximum-loop-impedance tables. This is an indicative comparison, not a verified trip PASS:

    CurveHolds up toTrips at least atMax. loop impedance, 16 A, 230 V
    B3 × In5 × In2.9 Ω
    C5 × In10 × In1.4 Ω
    D10 × In20 × In0.7 Ω
    K10 × In14 × In1.0 Ω

    Source: ABB, MCB technical document 9AKK108468A3912 (IEC/EN 60898-1 for B, C and D; IEC/EN 60947-2 for K); the last column is U0/(k·In), which reproduces that document's table. Manufacturers differ: in the same ABB document the S800 range under IEC 60947-2 lists D and K identically (10.4 to 15.6 × In), so D and K are not always different. For an MCCB, or any product with its own data, choose “Other” and enter the multiple.

    Why the instantaneous-trip verdict is NOT ASSESSED

    An indicative current above the curve threshold is not enough to confirm disconnection. This model assumes source/transformer zero-sequence impedances and the PE reactance. The breaker's internal resistance and arc resistance are also neglected. The actual protective device, fault loop and required clearing time (IEC 60364-4-41 and the applicable national rules) must be verified independently. The model covers TN-S screening only. If the main feeder is a busduct, its return path is not modeled either, making the estimate even less complete.

    8. Boundaries and common mistakes

    This is an IEC 60909-oriented radial LV engineering screen, not a complete accredited short-circuit study. It does not include motor, generator or inverter fault contributions, parallel supplies, line-neutral faults (the phase-to-earth estimate is limited to TN-S systems, section 7), detailed protective-device clearing times or a certified IEC 60909 breaking-current model. Nominal source and reactance assumptions can materially affect results.

    Calculation basis. The equations and factors follow the structure of IEC 60909-0:2016 (the transformer correction KT, the voltage factors c and the peak factor κ). IEC 60909-0:2026 has since been published. The voltage factor pairs and the KT treatment of the minimum case follow what DIgSILENT documents for the 2026 edition (sources are listed in docs/REFERENCE_TRACEABILITY_v0.1.80.md). KiloVolt has not been reconciled with the 2026 edition: that needs a licensed copy of the standard, so numerical equivalence with it is not certified, and the equation numbers quoted in the code are those of the earlier edition. The reconciliation status is recorded in docs/IEC_60909_2026_REVIEW_REQUIRED.md.

    Reference and sensitivity cases. A case is the documented reference only for a COMPLETE pair: 1.05/0.95 (±6 %) or 1.10/0.90 (±10 %). Any other combination, such as 1.10/0.95, 1.05/0.90 or 1.05/1.00, is a user-selected what-if and the screen says SENSITIVITY CASE. A cmin above the documented value raises the minimum current and can make it higher than the maximum: the screen then shows INCONSISTENT BOUNDS and a marker at the board, and the two columns must not be used as a range. If a documented pair still gives an inverted range, the screen says that the combination requires technical investigation. Without a minimum source fault level the minimum case reuses the entered level with the MV factor 1.00 instead of 1.10, which lowers the source impedance by about 9 % and can overstate the minimum current: enter the utility's minimum level for a meaningful minimum.

    KT = 1 comparison. Where leaving KT out would give a lower minimum current (KT below 1, always the case at ±6 %), the screen shows that lower figure beside the primary one, labelled SENSITIVITY COMPARISON. The primary result keeps KT formed with cmax. The lower figure is the more conservative one for protection sensitivity, so use it as a check; it is never presented as the 2026-oriented result. No protection-sensitivity PASS is ever issued from a minimum current whose source assumptions or methodology are unverified.

    Tolerance. The default ±10 % pair (cmax 1.10, cmin 0.90) is the initial design assumption, not a mandatory selection: choose ±6 % (1.05 and 0.95) or the advanced mode when your supply requires it.

    Earth-fault estimate. The phase-to-PE current is built on the minimum network, so it inherits cmin, KT,min and the minimum source level: changing any of them changes it directly. It treats the transformer and the upstream network as Z0 = Z1 in a single loop (the loop counts the upstream impedance in all three sequence terms, so it is on the safe side compared with a Dyn symmetrical-components calculation, by up to about 7 % at a 100 MVA source), is limited to TN-S and includes no earth-electrode, soil or TT/IT impedance.

    On screen, fault currents are marked Estimate and shown rounded to 0.1 kA with a leading ≈ (for example ≈ 21.1 kA), so they are not read as more precise than the model allows; the mathematical steps keep full precision. The breaker comparison is labelled indicative for the same reason.

    Never confuse an approximate maximum three-phase value with the minimum earth-fault current required to verify automatic disconnection, and never assume a manufacturer's nominal breaker kA is interchangeable with Icu, Ics or Icm.

    Primary references: IEC 60909-0, IEC 60228:2023 and IEC 60947-2, alongside actual manufacturer tables and national regulations.

    Open LV Radial Short-Circuit Designer →