Solid Material Properties

Material catalogue

The SOLID_MATERIALS registry (src/core/solidProperties.ts) provides named cp(T) and k(T) curves. Every curve is used only inside its published validity range. Outside that range the value is clamped to the nearest end of the range (constant extrapolation), never extrapolated. The Property Panel material mode shows each material's source, validity range, stated accuracy, and clamping behaviour inline.

Registry key	Material	Range (K)	Sources	Caveats
ofhc-copper	OFHC copper (UNS C10100/C10200)	4–300	NIST Cryogenic Material Properties Database, rev. 02/03/2010	k assumes RRR = 100; k is strongly RRR-dependent below ~100 K (factor ~2–5 across RRR 50–500)
grcop-84	GRCop-84 (Cu-8 at.% Cr-4 at.% Nb)	296–1173	Ellis, NASA/CR-2000-210055 (NTRS 20000064095)	High-temperature chamber-liner copper, not C18150 CuCrZr and not OFHC; k is the all-data regression mean, not the source's lower 95 % CI; clamped outside 296–1173 K
aluminum-6061-t6	Aluminum 6061-T6 (UNS A96061)	4–300	NIST Cryogenic Material Properties Database	Cryogenic range only; clamped outside
stainless-steel-304	Stainless steel 304 (UNS S30400)	4–1600	NIST (4–300 K) + ANL-75-55 304L correlations (≥300 K)	Composite (see splice note below); ANL states 304/304L property differences are negligible
stainless-steel-316	Stainless steel 316 (UNS S31600)	4–1600	NIST (4–300 K) + ANL-75-55 316L correlations (≥300 K)	Composite (see splice note below); above ~1200 K the ANL 316L correlations are ANL's own extrapolation toward the melting range
inconel-718	Inconel 718 (UNS N07718)	298–1375	Agazhanov, Samoshkin & Kozlovskii, J. Phys.: Conf. Ser. 1382 (2019) 012175 (CC-BY 3.0)	High-temperature range only (clamped below 298 K); the γ″/δ transformation intervals 900–1070 K (cp) and 800–1173 K (k) are bridged linearly (no single-phase correlation exists there)
ptfe	PTFE (Teflon)	4–300	NIST Cryogenic Material Properties Database	Cryogenic range only; clamped outside
g10-cr-normal	G-10 CR fiberglass epoxy, normal (through-thickness) k	10–300 (k); cp 4–300	NIST Cryogenic Material Properties Database	Anisotropic: k is the through-thickness direction
g10-cr-warp	G-10 CR fiberglass epoxy, warp (in-plane) k	12–300 (k); cp 4–300	NIST Cryogenic Material Properties Database	Anisotropic: k is the in-plane direction

Fit forms and stated accuracies are recorded per material in the registry provenance block (shown in the UI) and in the source-header comments of solidProperties.ts.

NIST cryogenic fits (OFHC copper, 6061-T6, 304, 316, PTFE, G-10)

Primary source: NIST Cryogenic Technologies Group, Cryogenic Material Properties Database (index page: https://trc.nist.gov/cryogenics/materials/materialproperties.htm). Underlying program: Marquardt, Le & Radebaugh, "Cryogenic Material Properties Database", 11th International Cryocooler Conference, 2000, https://trc.nist.gov/cryogenics/Papers/Material_Properties/2000-Cryogenic_Material_Properties_Database.pdf (data references for the copper curves: see the database's references page and NIST Monograph 177, Simon, Drexler & Reed, "Properties of Copper and Copper Alloys at Cryogenic Temperatures", 1992, https://nvlpubs.nist.gov/nistpubs/Legacy/MONO/nistmonograph177.pdf).

All NIST curves here use the database's log-log polynomial log10(y) = Σ_j a_j·(log10 T)^j, except OFHC-copper k, which uses the RRR-dependent rational polynomial log10 k = (a + c·√T + e·T + g·T^1.5 + i·T²)/(1 + b·√T + d·T + f·T^1.5 + h·T²). RRR = 100 is adopted, which is NIST's "average sample" curve; the RRR spread is a documented material uncertainty. The NIST 316 cp is two fits (4–50 K and 50–300 K) that join at 50 K with a 0.15 % step, inside their 2 % stated accuracy. NIST publishes the same k coefficients for 316 as for 304. G-10 k is anisotropic: the two registry entries carry the two published directions. The NIST k equation ranges start at 10 K (normal) and 12 K (warp), which sets those entries' stated validity range.

Stainless-steel high-temperature extension and the 300 K splice

Above 300 K the two stainless steels use ANL-75-55 (C. S. Kim, "Thermophysical Properties of Stainless Steels", Argonne National Laboratory, September 1975, DOE OSTI 4152287, https://www.osti.gov/biblio/4152287) recommended solid-region correlations (SI-converted):

• 304L: cp = 0.1122 + 3.222×10⁻⁵·T cal/(g·K); k = 0.08116 + 1.618×10⁻⁴·T W/(cm·K)
• 316L: cp = 0.1097 + 3.174×10⁻⁵·T cal/(g·K); k = 0.09248 + 1.571×10⁻⁴·T W/(cm·K)

The NIST cryogenic curves and the ANL correlations do not agree exactly at 300 K (the offsets are ~8 % for 304 cp, ~2 % for 316 cp, ~15 % for 304 k, and ~9 % for 316 k). The catalogue does not jump. The composite keeps the NIST value exactly at 300 K and removes the offset with a documented linear blend over 300–500 K. For T ≥ 500 K, the curve is the pure ANL-75-55 correlation. The composites are capped at 1600 K. The ANL correlations underlie data to 1620/1170 K for cp and 1600/1200 K for k for 304L/316L and are ANL's smoothed solid-region recommendations, while melting begins at 1670–1730 K.

Inconel 718 (298–1375 K)

Source: Agazhanov, Samoshkin & Kozlovskii, "Thermophysical properties of Inconel 718 alloy", J. Phys.: Conf. Ser. 1382 (2019) 012175 (open access, CC-BY 3.0, https://doi.org/10.1088/1742-6596/1382/1/012175), using piecewise correlations (the paper's eqs. (1)–(3) for cp, (9)–(10) for k) with stated errors of 2–3 % (cp) and 3–5 % (k). No credible low-temperature cp correlation is published for this catalogue, so the curve starts at 298 K and clamps below. The γ″/δ phase-transformation intervals (900–1070 K for cp, 800–1173 K for k), where the source reports no single-phase correlation, are bridged by a straight line between the branch endpoints (a documented approximation, not measured data).

GRCop-84 (296–1173 K)

Source: D. L. Ellis, "Thermophysical Properties of GRCop-84", NASA/CR-2000-210055 (2000), https://ntrs.nasa.gov/citations/20000064095: DSC cubic for cp (eq. 12, 296–1173 K) and the all-data unweighted ln(T) regression for k (eq. 17). Companion liner context is found in Ellis, NASA/TM-2005-213566, where MCC liner k is quoted as 305–320 W/m·K. GRCop-84 is Cu-8 at.% Cr-4 at.% Nb, developed at NASA GRC for regeneratively cooled chamber liners. It is not C18150 CuCrZr and not OFHC (the ofhc-copper NIST curve stops at 300 K). The catalogue stores the k regression mean, not the source's lower 95 % confidence interval (mean − 1.860×Sy.x). Published k data start near 80 K. The catalogue range is the intersection with cp (296–1173 K), and clamps outside.

Design

• Schema (src/core/schema.ts): SolidPropertySpec = number | { table: [[T, v], …] } | { material: '<name>' }. SolidNode.cp and conduction-Conductor.k both accept it. A constant number is also permitted. The solver keeps a separate, untouched constant code path.
• Canonical T-dependent form: piecewise-linear table in T (K), clamped at the end knots (value = endpoint, slope = 0 outside; enthalpy extends linearly with the clamped value). It was chosen because then (a) the property value, (b) its enthalpy integral H(T) = ∫ cp dT (piecewise quadratic), and (c) its slope are ALL exact for the represented curve. The solver's residual and Jacobian are exact for the discrete model with no quadrature error anywhere.
• Preset registry (src/core/solidProperties.ts): each SOLID_MATERIALS entry stores the source fit coefficients + provenance (source, URL, fit form, validity range, stated accuracy, notes; assumed RRR for OFHC copper). It samples them ONCE at module load via adaptive midpoint refinement to ≤ 1e-3 relative interpolation error (see Validation margins below). Composite curves (the stainless steels, Inconel 718) are sampled segment-by-segment (sampleComposite) so knots land exactly on the splice/branch temperatures. getSolidMaterialTable(name, 'cp'|'k') returns a copy of the knots for calibration-style probes (e.g. scaling the whole curve).
• Solver coupling (solveThermalSubsystem, src/core/solver.ts): solid storage is the enthalpy form m·(H(T_new) − H(T_old))/dt, NOT m·cp(T̂)·ΔT/dt at a representative T. Justification: (1) backward Euler in enthalpy is exactly energy-conserving per step for the represented curve. The per-step enthalpy release is m·ΔH regardless of how sharply cp collapses across the step (on a cryogenic line one step can move a wall through cp 385→150). A representative-cp form would mis-weight that integral by O(10 %) per step at the quench front. (2) The Jacobian term m·cp(T_new)/dt is the EXACT derivative of the enthalpy form (dH/dT = cp), so the existing exact-analytic-Jacobian Newton structure is preserved (verified entry-by-entry against central finite differences in src/core/__tests__/solidThermalTransient.test.ts). (3) It telescopes exactly in the constant-extraction lumped case.
• Conduction k(T): k is evaluated at the endpoint mean Tm = (Tf+Tt)/2 (standard mean-property link). The Jacobian gets the exact extra term G'(Tm)·(Tf−Tt)/2 on both endpoint derivatives (G' is piecewise-constant from the table). For thin-wall transfer lines, axial conduction is orders of magnitude weaker than wall-to-fluid convection, so k(T) matters mainly for conduction-dominated hardware (thick walls, short links, MLI standoffs).
• Jacobian probe hook: probeThermalSubsystem(ctx, state, {dt, prevState}, hMap?, Toverride?). This is the thermal analogue of the fluid-side probeJacobians, used by the FD guard. It shares one assembly function (assembleThermalSubsystem) with the Newton loop so probe and production can never drift.
• ParaHydrogen is included in SUPPORTED_REAL_FLUIDS (now the curated favorites subset of the generated 124-fluid HEOS catalogue; see docs/fluid-catalogue.md). A test pins both its inclusion and OrthoHydrogen's exclusion from the favorites. The latter has no transport model in this coolprop-wasm build: it stays discoverable in the picker but validation rejects it with a zero-transport error.
• Validation (validate.ts): table specs need ≥2 points, strictly-increasing positive T, positive values. Material names are checked against the registry. Constant-cp error messages are untouched.

Validation margins (property layer)

• OFHC copper: sampled table vs NIST fit equation is ≤ 0.097 % (cp) / 0.099 % (k) over 4–300 K. This is ~50× tighter than the fits' stated accuracy.
• OFHC NIST cp fit vs canonical handbook anchors (consistent with NIST Monograph 177 §7-1): 10 K: 0.8566 vs ≈0.86 (−0.4 %); 20 K: 7.506 vs ≈7.5 (+0.1 %); 77 K: 195.92 vs ≈193.5 (+1.3 %); 150 K: 324.11 vs ≈324 (+0.0 %); 300 K: 389.40 vs 385 (+1.1 %). All within NIST's stated fit accuracy (10 % below 15 K; 5 % above).
• OFHC k fit (RRR=100): 396.3 W/m·K at 300 K; RRR spread at 20 K is a factor >2.5 (documented material uncertainty).
• All catalogue tables reproduce their source curves to ≤ 0.2 % over the full validity range (enforced by tests; most are ≤ 0.1 %).
• Stainless composites: table value at 300 K equals the NIST fit (splice anchor, no jump). For T ≥ 500 K the tables match the ANL-75-55 correlations, including the report's own printed anchors at 1000 K (304L: cp 0.1444 cal/(g·K), k 0.2429 W/(cm·K); 316L: cp 0.1414 cal/(g·K), k 0.2496 W/(cm·K)).
• Inconel 718: table matches the source paper's Table 2 recommended values at 298–800 K and 1100–1300 K to ≤ 0.1 % (the paper's 4-digit print rounding).
• GRCop-84: sampled tables vs Ellis NASA/CR-2000-210055 eqs. 12 and 17 to ≤ 0.2 % over 296–1173 K. k(550 K) ≈ 305 W/m·K (low end of the TM-2005 MCC band).

Test inventory

• src/core/__tests__/solidProperties.test.ts: OFHC fit/table validation vs published values at 10/20/77/150/300 K with the stated tolerances. This test covers table mechanics (interp, clamp, exact integral, dH/dT consistency), spec validation, and registry provenance. It creates deterministic fit-value snapshots for every material. It includes catalogue table-vs-source checks (≤0.2 %) over each material's full validity range. It covers stainless 300 K splice-continuity and ANL 1000 K anchor tests. For Inconel 718, it includes source-table anchors, branch-continuity and transformation-gap bridge tests. It provides GRCop-84 Ellis fit snapshots, MCC-band k check, and clamp/provenance tests. It covers G-10 anisotropy and per-direction range tests, as well as clamp behaviour at both ends of every range. It asserts registry-wide provenance/validity-range invariants and checks OFHC non-regression.
• src/core/__tests__/solidThermalTransient.test.ts: thermal-subsystem Jacobian vs central FD (mixed T-dep + constant-cp network incl. clamp-region nodes); lumped-mass constant-extraction telescoping (exact) + 200 K crossing vs an independent in-test march on the raw NIST fit; Newton-cooling with cp(T) vs fit-reference and constant-cp vs exact BE recurrence; golden constant-cp bit-identity (transient traces + steady state).
• src/ui/tests/solidPropertyField.test.tsx: material labels for the whole catalogue. The material mode renders each material's source, validity range and clamping note. specValueAt honours per-material clamp ranges.
• src/core/__tests__/realFluid.test.ts: ParaHydrogen is supported (Tsat(74.97 psia) = 27.292 K = the para curve, matching the NBS-published saturation temperatures). OrthoHydrogen remains excluded.

Deferred candidates

• Ti-6Al-4V: the NIST cryogenic page publishes only k (20–300 K). The public NASA handbook (NASA-CR-123775, "Materials data handbook. Titanium 6Al-4V", 1972) is a scanned document whose tabulated cp(T)/k(T) could not be extracted with the accuracy this catalogue requires. It is deferred rather than fabricated; the NIST k-only curve is insufficient for the joint cp+k requirement.

Reproduce

Use scripts/chilldown-baseline.ts and the configurations in src/ui/tests/chilldownTwoPhase.test.ts as the maintained starting point. getSolidMaterialTable supports calibration-style probes (e.g. scaling the whole cp/k curve). Golden bit-identity values live in src/core/__tests__/solidThermalTransient.test.ts.