V-HC-04 — two-layer slab, k-ratio 10.0:1

Tags: heat_conduction, laminate, harmonic_mean, steady

References:

  • series thermal-resistance network (exact)
  • FDS Verification Guide insulated_steel_plate / ht_pile

Problem statement

FDS insulated_steel_plate / ht_pile analog. Two inert solids layered via composition step ICs; Dirichlet T at both faces; steady exact solution is the series-resistance network:

q = ΔT / (d₁/k₁ + d₂/k₂), Tinterface = Tbot + q·d₁/k₁, piecewise linear

What is verified: the distance-weighted harmonic-mean face conductivity is NODALLY EXACT for piecewise-constant k with the jump on a face — the steady cell-center temperatures and the boundary fluxes must match the resistance network to integrator precision, for k-ratios 10:1 and 100:1.

Off-face variant (c): composition step ICs are SAMPLED AT CELL CENTERS, so a prescribed interface that is not on a face quantizes to the nearest face — there are no mixed cells. Measured 2026-07-06: the numerics match the QUANTIZED-interface network to 1e-11 (i.e. the scheme is still nodally exact), and the entire user-facing error is the true-vs-quantized network gap (1.64% in flux, 1.47 K in profile at Δz = 0.5 mm) — pinned here against its a-priori Δz/2 sensitivity bound.

No convergence spec: (a)/(b) are nodally exact (error is integrator-, not mesh-limited); (c) is quantization-dominated and non-monotone in n.

V-HC-04a — two-layer slab, k-ratio 10.0:1

Quantities of interest (n = 40)

QoIvalueexacterrortolerancewithin tolprovenance
L∞ steady profile vs (quantized) network1.518e-111.518e-111e-08yesharmonic-mean face k is NODALLY EXACT for face-aligned jumps; k-ratio 10.0:1
steady bottom-face flux vs network909.1909.13.336e-111e-08yesboth faces must carry the exact series flux
steady top-face flux vs network909.1909.16.595e-131e-08yesboth faces must carry the exact series flux
steadiness (last half-interval drift)8.219e-108.219e-101e-06yestend = 10·L²/αmin

V-HC-04a T_profile_steady

V-HC-04a T_profile_steady_error

Solver configuration

settingvalue
integratorKenCarp4 (default)
abstol1.0e-10
reltol1.0e-8

V-HC-04b — two-layer slab, k-ratio 100.0:1

Quantities of interest (n = 40)

QoIvalueexacterrortolerancewithin tolprovenance
L∞ steady profile vs (quantized) network3.024e-113.024e-111e-08yesharmonic-mean face k is NODALLY EXACT for face-aligned jumps; k-ratio 100.0:1
steady bottom-face flux vs network99.0199.014.082e-101e-08yesboth faces must carry the exact series flux
steady top-face flux vs network99.0199.019.774e-141e-08yesboth faces must carry the exact series flux
steadiness (last half-interval drift)3.112e-093.112e-091e-06yestend = 10·L²/αmin

V-HC-04b T_profile_steady

V-HC-04b T_profile_steady_error

Solver configuration

settingvalue
integratorKenCarp4 (default)
abstol1.0e-10
reltol1.0e-8

V-HC-04c — two-layer slab, k-ratio 10.0:1 (off-face interface)

Quantities of interest (n = 40)

QoIvalueexacterrortolerancewithin tolprovenance
L∞ steady profile vs (quantized) network1.518e-111.518e-111e-08yesharmonic-mean face k is NODALLY EXACT for face-aligned jumps; k-ratio 10.0:1
steady bottom-face flux vs network909.1909.13.336e-111e-08yesboth faces must carry the exact series flux
steady top-face flux vs network909.1909.16.595e-131e-08yesboth faces must carry the exact series flux
interface-quantization smearing (documented)0.0163600.016360.02079yesTHE laminate-off-mesh error users see: flux off by 1.64% at Δz = 0.0005 m
steadiness (last half-interval drift)8.219e-108.219e-101e-06yestend = 10·L²/αmin

V-HC-04c T_profile_steady

V-HC-04c T_profile_steady_error

Solver configuration

settingvalue
integratorKenCarp4 (default)
abstol1.0e-10
reltol1.0e-8