Underground Duct Bank Calculation
Adjust amperage or choose a guided comparison. Follow the heat and compare the response. A transparent, steady-state model that runs entirely in your browser.
Try a comparison
Amperage is always editable. Choose one other input to adjust. Switching inputs restores the previous input to baseline. Both cases use the same amperage.
The thermal field
Updates automaticallyNOT FOR DESIGN
Field: soil and concrete, solved together. Gray outline: concrete encasement. Circles: equivalent conduit heat sources. Three cable symbols are schematic.
Conduit temperatures
| CONDUIT | DEPTH | CONDUCTOR | HEAT |
|---|
How this learning model works
Changing amperage or selecting a comparison automatically recalculates the results. Use the guided buttons to explore soil, concrete, burial depth, conductor size, and material. Watch the temperatures and isotherms respond.
The drawing shows the actual modeled concrete boundary. Width and height are derived from conduit size, number of rows and columns, center spacing and concrete edge cover. Burial depth is measured from grade to the top of the concrete. Depths in the table are measured to conduit centers.
System voltage
The system is set to 34.5 kV line-to-line. Voltage is recorded as a scenario property; dielectric heat is a preset W/m-per-cable value. Voltage alone does not specify dielectric loss: cable capacitance, frequency and dielectric loss tangent are also required. This change does not automatically alter heat generation.
Two materials, one heat balance
A two-dimensional, cell-centered finite-volume calculation solves steady heat conduction through separate soil and concrete regions. Conductivity is 100 divided by thermal resistivity entered in °C·cm/W. Face conductances use the two adjacent half-cell thermal resistances in series. This preserves heat-flow continuity across material interfaces.
k = 100 / resistivity
q_conduit = 3 I² R₂₀ k_AC [1 + α(T_conductor − 20)] (1 + λ) + 3 Wd
The grid is fine near the bank and expands toward remote boundaries. Grade and far boundaries are held at ambient temperature; far boundaries lie at least six times the total bank-bottom depth away from the bank. Numerical unit-source solutions are combined into a thermal coupling matrix. Temperature-dependent losses are then solved simultaneously, and the current-at-limit value is found by bisection. The ground temperature field uses those same numerical solutions.
Simplifications to keep in mind
This is a teaching prototype, not the validated Cable View engine or a complete Neher-McGrath/IEC rating implementation. Each conduit is an equivalent circular heat-source region with a fixed illustrative conductivity of 10 W/(m·K). Conduit sizes are idealized as 6-inch or 8-inch modeled diameters, not manufacturer-specific outside diameters. Cable symbols show three equally loaded single-conductor cables; individual insulation layers and positions are not resolved.
The former internal-resistance control has been removed. For continuity with the earlier experiment, a fixed illustrative conductor-to-conduit thermal resistance of 0.7 K·m/W per conductor remains in the model. It is not a calculated cable-library value. Dielectric and metallic losses are lumped. Model results are sensitive to these assumptions.
Conductor resistance uses nominal metallic area: 1 kcmil = 0.5067075 mm². Copper resistivity at 20 °C: 1.7241 × 10⁻⁸ Ω·m, temperature coefficient 0.00393/K. Aluminum: 2.8264 × 10⁻⁸ Ω·m and 0.00403/K. AC and metallic-loss factors are fixed illustrative assumptions for these lessons.
No thermal backfill, soil dry-out, moisture migration, load cycling, harmonics, unequal circuit loading, cable crossings, or longitudinal heat flow are modeled. Numerical convergence is checked, but mesh and domain convergence for each installation have not been established. Use the tool to learn trends, not to approve an installation or select a design ampacity.
Method background: NIST: finite-volume discretization. This prototype uses its own JavaScript solver; it does not run FiPy.