Line parameters
A computation of line parameters is composed of a problem, which defines what is computed, and a formulation, which selects the formulas. The computation is executed by a backend, and all computations are invoked by the same call, compute(problem, formulation; kw...).
Problem definition
The problem LineParametersProblem(system, earth; temperature, frequencies) is defined by the cables of a CableSystem, the EarthModel below the air, the temperature of all conductors and dielectrics in °C, equal to 20 by default, and the frequencies of interest in Hz. The frequencies must be positive, finite and sorted in strictly increasing order. All numbers of the system, earth, temperature and frequencies are promoted together with Float64, so that the number type T of the problem is at least Float64.
Coaxial formulation
The CoaxialFormulation selects one formula for each physical contribution. Each slot takes a formula of its family, and the corresponding keyword accepts :id, :id => (key = value, ...) with the options of the formula, a formula object, or :default, which selects the default formula of the slot. The table below gives the default formula of each slot and the other formulas that the slot admits.
| Keyword | Family | Default | Also admits | Contribution |
|---|---|---|---|---|
internal_impedance | InternalImpedance | :schelkunoff1934 | :wedepohl1973 | surface impedances of the conductors |
insulation_impedance | InsulationImpedance | :ametani1980 | magnetic series term of a dielectric interval | |
insulation_admittance | InsulationAdmittance | :lossless | :lossy | admittivity of insulator layers |
semicon_admittance | SemiconAdmittance | :lossless | :lossy | admittivity of semicon layers |
earth_impedance | EarthReturn | :martins2026 | :gary1976, :wedepohl1973, :saad1996, :lucca1994, :ametani2009, :carson1926, :pollaczek1926, :papadopoulos2010, :wise1934, :xue2018 | earth-return impedance |
earth_admittance | EarthReturn | :martins2026 | :ideal, :papadopoulos2010, :wise1948, :xue2018 | earth-return potential coefficients |
shunt | Shunt | :series | potential coefficients inside the cables | |
resistivity | Resistivity | :linear | resistivity of the conductors at the temperature | |
earth_properties | EarthProperties | :constant | :alipio2014, :cigre2019, :datsios2019, :longmire1975, :messier1985, :portela1999, :scott1967, :visacro1987, :visacro2012 | properties of each earth layer |
equivalent_earth | EarthProperties | :bottommost | :martins2020, :xue2021 | equivalent earth of the layers below the air |
An earth slot records one formula per domain of a pair of cables:
air: both cables lie above the interface.earth: both cables lie below the interface.mixed: one cable lies above the interface and the other below it.
One selection fills the three domains. With earth_impedance = (air = :martins2026,), each domain not given takes the default of the slot. A formula that treats one domain goes in that place, as earth_impedance = (air = :gary1976, earth = :saad1996, mixed = :lucca1994). Formulas of one identifier in the earth slots must be equal. :martins2026 obtains the impedance and the potential coefficients from the same circumferential currents.
The earth-return formulas treat these domains and give these quantities:
| Formula | Domains | Quantities |
|---|---|---|
:martins2026 | air, earth, mixed | Zg, Pg |
:ideal | air, earth, mixed | Pg, zero for a pair in the earth or across the interface |
:carson1926, :gary1976, :wise1934 | air | Zg |
:wise1948 | air | Pg |
:pollaczek1926 | earth, mixed | Zg |
:saad1996, :wedepohl1973 | earth | Zg |
:papadopoulos2010, :xue2018 | earth | Zg, Pg |
:ametani2009, :lucca1994 | mixed | Zg |
A computation refuses a formula placed in a domain that it does not treat when the system has a pair in that domain. So earth_impedance = :carson1926 computes an overhead system, and a buried system refuses it. The remaining keywords are the options of the formulation:
| Option | Default | Meaning |
|---|---|---|
Γ | 0 | longitudinal wavenumber in 1/m, one value or one per frequency |
bundle | true | merge the conductors that share a positive connection index |
kron | true | remove the conductors of connection index zero by eliminate! |
transposition | false | apply the ideal cyclic transposition |
A nonzero Γ needs a formula that reads it in each place of both earth slots, and the constructor refuses any other.
The macro @formulation writes a formulation as a block, one keyword per line, and the tags @air, @earth and @mixed write a selection per domain. The macro lowers to the constructor and does not add a capability. The refusals above apply to it. With a type first, it builds that formulation:
using LineCableModels
formulation = @formulation begin
earth_impedance = @air :gary1976 @earth :saad1996 @mixed :lucca1994
earth_admittance = :ideal
transposition = true
end
fem = @formulation FEM begin
earth_properties = :alipio2014
endA layered earth takes a law in earth_properties and a model in equivalent_earth, which Earth return describes. Here the air lies over a layer 5 m thick and a semi-infinite layer:
earth = EarthModel((EarthLayer(Inf, 1.0, 1.0), EarthLayer(100.0, 10.0, 1.0, 5.0),
EarthLayer(1000.0, 10.0, 1.0)))
layered = CoaxialFormulation(earth_properties = :alipio2014,
equivalent_earth = :martins2020)The formulas are described in the theory pages Matrix formulation, Conductors, Dielectrics and Earth return.
Computation
The call compute(problem, formulation; kw...) runs the default backend of the formulation type, given by the trait backend: LCMBackend, the analytical computation of this package, for CoaxialFormulation, and GmshBackend for FEM. Another backend is selected by passing it as the first argument, compute(backend, problem, formulation; kw...). The keywords are the computation options of the backend, and LCMBackend has one option, trace, equal to false by default.
At each frequency f, with s = j2πf, the computation of LCMBackend consists of the following steps:
- Evaluation of the material laws, which give the resistivity of each conductor at the operating temperature, the admittivity of each dielectric layer and the properties of the air and the earth.
- Computation of the surface impedances of the conductors and of the magnetic terms of the dielectric intervals, which are assembled by the loop recursion into the impedance matrix inside the cables,
Zin. - Computation of the potential coefficients inside the cables,
Pin, by the series combination of their layers. - Computation of the earth-return impedance
Zgand potential coefficientsPgof the cables, from their outer surfaces. - Addition of the entries of
ZgandPgfor each pair of cables, including each cable with itself, to every pair of their conductors. - Bundle and Kron reductions of the impedance and potential matrices.
- Computation of the admittance
Y = s P⁻¹from the reduced potential matrixP. - Ideal transposition of
ZandY, whentranspositionistrue.
The same steps are given in the docstring of solve!, in the Matrix formulation page.
Results
The result of the computation is a LineParameters object, with the following fields:
| Field | Meaning | Unit |
|---|---|---|
Z | series impedance by row, column and frequency | Ω/m |
Y | shunt admittance by row, column and frequency | S/m |
frequencies | frequencies of interest | Hz |
terminals | label of each row | |
line_length | length of the line | m |
formulation | formulation used in the computation | |
trace | matrices from which the backend assembles Z and Y |
Each row is labeled as cable:<i>:<terminal>, where i is the index of the cable in the system and terminal is the name of the terminal. The label of a bundle joins the labels of its terminals with +. With trace = true, the trace of LCMBackend contains Zin in Ω/m and Pin in m/F per conductor, and Zg in Ω/m and Pg in m/F per cable, each with one slice per frequency. Otherwise, the four arrays are empty.
Options
The options of each owner are declared by defaults, where the owner is a formula or a formulation type for FormulationOptions, and a backend type for ComputationOptions. The record of options is built once, from the defaults and the keywords given. An unknown key throws an ArgumentError, each value is converted to the type of its default, and the record is checked by validate. The defaults of the coaxial formulation are:
LineCableModels.defaults(CoaxialFormulation)(Γ = 0.0 + 0.0im, bundle = true, kron = true, transposition = false)Reference systems
The test suite includes six reference systems, whose line parameters computed by the analytical and finite-element formulations are stored in test/references/:
two_bare_wires: two copper wires with a radius of 42.5 mm, buried at 1 m depth with a horizontal spacing of 1 m, in a homogeneous earth of 0.1 Ω·m.two_insulated_wires: the same wires with an insulation 1 mm thick.coaxial_circular_strands: one coaxial cable with a core of circular strands, buried at 1 m depth in a homogeneous earth of 100 Ω·m.coaxial_rectangular_strands: the same cable with a core of rectangular strands.overhead_bare_wires: the two bare wires at a height of 1 m above an earth of 0.1 Ω·m.mixed_bare_wires: the two bare wires with one placed above the earth and the other buried.
The validation consists of comparing the analytical and finite-element formulations on these systems. The analytical results of the package agree with the reference values of Z and Y within 1e-9 of the largest diagonal entry at each frequency, with Zin and Pin within 1e-12 and Zg and Pg within 1e-9. The finite-element results are compared with the analytical ones by means of the relative errors defined in Finite element solver, at an accuracy level of 1 %. For each earth-return formula, the README of test/references/ gives its frozen results on the systems that it treats and their deviation from those of :martins2026.