Case M: validation

Contents
  1. Measurement locations
  2. Explore the points
  3. OpenFOAM 12
    1. Measured against modelled
    2. Field maps
    3. OpenFOAM 12, realizable k-epsilon (Eddy3D Outdoor), configuration sct03
    4. OpenFOAM 12, realizable k-epsilon (Eddy3D Outdoor), configuration sct05
    5. OpenFOAM 12, realizable k-epsilon (Eddy3D Outdoor), configuration sct07
  4. Not run
  5. Notes on every run
  6. Leaderboard

Engine status: OpenFOAM 12 scored · OpenLB pending · FluidX3D pending · ML surrogate pending

Nomenclature: symbols, columns and scores

Geometry, reference scales and release

  • \(H\), \(b\), \(D\) (m) — height of the building, cube, tower or tree belt; width of the building; edge of the cubes of Case C: the lengths the cases are drawn and normalised by
  • \(U_H\) (m/s) — reference speed: the mean approach-flow speed at the height \(H\) (each Data page gives the value used)
  • \(U_R\), \(H_R\), \(Z_R\) (m/s, m) — reference speed and height of a workbook: Case E, \(U_R\) at \(Z_R\) = 250 m full scale (1 m in the model); Case M, \(U_R\) and \(H_R\) at the reference point of the tunnel and of each field run
  • \(U_{\mathrm{ref}}\), \(L_{\mathrm{ref}}\) (m/s, m) — reference speed and length of a case’s normalisation: \(U_H\) and \(H\) (\(U_R\) and \(H_R\) in Case M); for the velocity ratios the approach speed at 15.9 m (Case E) or at 500 m (Case F)
  • \(U_b\) (m/s) — Case G: mean approach-flow speed at 9 m, the normalisation of its approach-flow table
  • \(q\), \(c_{\mathrm{gas}}\) (m³/s, ppm) — tracer release of the experiment: volume flow of the released gas, and its tracer content (\(10^6\) ppm when pure); \(Q\) is the source strength given to a solve
  • \(R_b\) (dimensionless) — bulk Richardson number of the heated-floor cases J and L (negative: unstable stratification)
  • \(\mathrm{Re_H}\), \(\mathrm{Re_D}\) (dimensionless) — Reynolds numbers \(U_H H/\nu\) and \(U_H D/\nu\) (\(\nu\): kinematic viscosity of air)
  • \(u_\ast\), \(z_0\), \(\kappa\) (m/s, mm, dimensionless) — friction velocity and roughness length of the log law fitted to the approach flow, and the von Kármán constant (0.41)
  • \(\mathrm{WD}\) (column wind_direction_deg; °) — wind direction (each frame says how it is counted); \(\mathrm{WD_R}\) is the mean direction of a field run at its reference point (Case M)
  • \(x\), \(y\), \(z\) (columns x, y, z; m) — coordinates in the case’s frame (\(+x\), \(+y\) and \(+z\) are the positive directions; each Data page says which way \(+x\) points), \(z\) up from the floor of the tunnel or from the ground
  • \(\mathrm{d}x\) (m) — lattice spacing of the OpenLB and FluidX3D runs, in metres or as a fraction of \(H\) (\(H/64\))
  • \(L_r\) (m) — reattachment length behind the building, read from a wake line at ground level; reported as \(L_r/H\)

Velocity and turbulence

  • \(U\), \(V\), \(W\) (columns U, V, W; m/s) — mean velocity along \(+x\), \(+y\), \(+z\); \(V\) is positive toward \(+y\) in every case
  • \(\lvert\mathbf{U}\rvert\) (m/s) — magnitude of the mean velocity vector, \(\sqrt{U^2+V^2+W^2}\)
  • \(V_{\mathrm{s}}\) (column Vs; m/s) — mean omnidirectional scalar speed of a thermistor (never a vector magnitude)
  • \(\overline{u'u'}\), \(\overline{v'v'}\), \(\overline{w'w'}\) (columns uu, vv, ww; m²/s²) — velocity variances (overbar: time average, prime: fluctuation about it)
  • \(\overline{u'v'}\), \(\overline{u'w'}\), \(\overline{v'w'}\) (columns uv, uw, vw; m²/s²) — Reynolds shear stresses
  • \(k = \tfrac{1}{2}\left(\overline{u'u'}+\overline{v'v'}+\overline{w'w'}\right)\) (column k; m²/s²) — turbulent kinetic energy
  • \(U/U_{\mathrm{ref}}\) (column velocity_ratio; dimensionless) — mean speed over the case’s reference speed (\(V_{\mathrm{s}}/U_{\mathrm{ref}}\) for a thermistor)
  • \(\sigma_{U/U_{\mathrm{ref}}}\) (column velocity_ratio_sd; dimensionless) — standard deviation of velocity_ratio over the observations behind a field value (Case F)
  • \(k/U_{\mathrm{ref}}^2\) (column k_ratio; dimensionless) — turbulent kinetic energy over the square of the reference speed, where only normalised values are published (Case G)

Concentration and temperature

  • \(C\) (column C; ppm) — mean tracer-gas concentration as published (volume fraction \(\times 10^6\) of the released gas diluted in air)
  • \(c_{\mathrm{rms}}\) (column c_rms; ppm) — root mean square of the concentration fluctuation
  • \(c_0\) (ppm) — reference concentration, defined for each case on its Data page: \(c_0 = \dfrac{c_{\mathrm{gas}}\thinspace q}{U_H H^2}\) (\(U_R H_R^2\) in the denominator in Case M)
  • \(c^\ast\) (column concentration_ratio; dimensionless) — dimensionless concentration \(c^\ast = C/c_0\); the value of a solve is \(c\thinspace U_H H^2/Q\) for its mean concentration \(c\) and its source strength \(Q\)
  • \(\mathrm{Sc_t}\) (dimensionless) — turbulent Schmidt number of a solve (0.3, 0.5 and 0.7)
  • \(T\), \(T_{\mathrm{rms}}\) (columns T, T_rms; °C, K) — mean air temperature as published (Cases J and L) and the root mean square of its fluctuation
  • \(\overline{u'c'}\), \(\overline{v'c'}\), \(\overline{w'c'}\) (columns uc, vc, wc; ppm·m/s) — turbulent concentration fluxes
  • \(\overline{u'\theta'}\), \(\overline{v'\theta'}\), \(\overline{w'\theta'}\) (columns uT, vT, wT; K·m/s) — turbulent heat fluxes (\(\theta'\): the temperature fluctuation)
  • \(\theta = \dfrac{T_f - T}{T_f - T_H}\) (dimensionless) — floor-referenced temperature the solves of Cases J and L are scored on: 0 at the floor, 1 in the approach air at \(H\); \(T_f\) is the floor temperature and \(T_H\) the temperature of the approach flow at \(H\)

Scores

  • \(n\) — number of pairs of a measured value \(o_i\) and a modelled value \(p_i\) at one point
  • \(R\) — Pearson correlation coefficient of the pairs
  • \(R^2\) — square of \(R\): share of the measured variance a linear fit to the model explains
  • \(\mathrm{FAC2}\) — fraction of pairs with \(0.5 \le p_i/o_i \le 2\)
  • \(\mathrm{FB}\) — fractional bias \(2(\overline{o}-\overline{p})/(\overline{o}+\overline{p})\), positive when the model is low
  • \(\mathrm{HR}\) (hit rate) — fraction of pairs with \(\lvert p_i-o_i\rvert/\lvert o_i\rvert \le \delta_{\mathrm{rel}}\) or \(\lvert p_i-o_i\rvert \le \delta_{\mathrm{abs}}\), where \(\delta_{\mathrm{rel}} = 0.25\) and \(\delta_{\mathrm{abs}} = 0.05\thinspace U_H\) for a velocity component
  • \(\mathrm{NMSE} = \dfrac{\overline{(o-p)^2}}{\overline{o}\thinspace \overline{p}}\) — normalised mean square error
  • \(\mathrm{NAD} = \dfrac{\overline{\lvert o-p\rvert}}{\overline{o}+\overline{p}}\) — normalised absolute difference
  • \(\mathrm{MG} = \exp\overline{(\ln o-\ln p)}\) — geometric mean bias
  • \(\mathrm{VG} = \exp\overline{(\ln o-\ln p)^2}\) — geometric variance
  • The ratio of \(\mathrm{FAC2}\) keeps its sign, so a model of the other sign than the measurement is a miss; \(\mathrm{FB}\), \(\mathrm{NMSE}\) and \(\mathrm{NAD}\) divide by the means, so they are undefined (shown as —) for a signed component unless the measured and the modelled mean are both positive
  • VDI 3783-9 calls the hit rate \(q\) and the tolerances \(D\) and \(W\); on this site those letters keep their other meanings (\(q\) the tracer release rate, \(D\) the block size of Case C, \(W\) the vertical velocity)
  • The detection limit of a concentration is one value per case, \(10^{-3}\) of the largest measured value. It sets \(\delta_{\mathrm{abs}}\) of the hit rate and the both-below rule of \(\mathrm{FAC2}\), and floors the values in \(\mathrm{MG}\) and \(\mathrm{VG}\).
  • The scores \(\mathrm{FB}\), \(\mathrm{FAC2}\), \(\mathrm{NMSE}\), \(\mathrm{NAD}\), \(\mathrm{MG}\) and \(\mathrm{VG}\) follow Chang & Hanna (2004); the hit rate follows VDI 3783-9.

Quantities scored

  • \(U\), \(V\) and \(W\) (mean velocity components) over \(U_H\)
  • \(k\) (turbulent kinetic energy) over \(U_H^2\)
  • the speed \(\lvert\mathbf{U}\rvert\) and \(\sqrt{\lvert\mathbf{U}\rvert^2+2k}\), against a thermistor’s \(V_{\mathrm{s}}\)
  • \(c^\ast = C/c_0\) (dimensionless concentration) at turbulent Schmidt numbers \(\mathrm{Sc_t}\) of 0.3, 0.5 and 0.7
  • \(\theta\) (floor-referenced temperature)

Measurement locations

Measurement points, Case M
Figure 1. Case M, measurement points: A near field, within 100 m of the release point; B mid field, 100 to 300 m from the release point; C far field, beyond 300 m. Positions in metres, full scale. Star: release point; R: reference point; dotted circles: 100 m and 300 m from the release point.

Explore the points

model lowmodel high
Loading the points… Hover a point to see where it sits in 3D and how the model differs from the measurement; click to pin it. This needs JavaScript; the figures on this page show the same scores.

OpenFOAM 12

Status: scored

The tracer release over the Atsugi campus of Tokyo Polytechnic University: ethylene in a 1:600 tunnel, with 18 field runs. An Eddy3D production run solved on 2026-10-18, realizable k-ε at numerics level 3, one steady solve on one mesh for the tunnel’s heading WD 180 only (1,394,053 cells, 3000 iterations). It stopped at the iteration limit with the SIMPLE residual control unmet; the initial residual of epsilon at the last iteration was \(1.9 \times 10^{-2}\).

  • Scored at WD 180 against the wind-tunnel run: \(U\), \(V\), \(W\) and \(k\) over the tunnel’s \(U_R\) = 4.08 m/s on the measurement and over the solved speed at the reference mast (4.686 m/s) on the model, because \(U_R\) is defined at the mast.
  • The concentration is the measured \(C/c_0\) (\(c_0\) = 132.5 ppm) against the model’s \(c^\ast = c\thinspace U_R H_R^2/Q\) with the same solved \(U_R\), for the turbulent Schmidt numbers 0.3, 0.5 and 0.7: three passive tracers on one flow, three runs (configuration sct03, sct05, sct07); the flow scores are the same in all three. One detection limit, \(10^{-3}\) of the largest measured value.
Data notes: Scores here and Eddy3D's own
  • The velocities and the concentration lose the CSV’s rounding (to 0.01 m/s and 0.1 ppm). \(U\) R is 0.7085 here and 0.7081 there, \(k\) R 0.4464 and 0.4499, and \(W\), whose values are of the order of the rounding, R 0.1814 and 0.1757 (FAC2 0.27 and 0.20). The field values divide by \(H_R\) = 41 m as the cleaned table states, Eddy3D’s by 40.8 m (a 1 % difference in every field \(c^\ast\)).
  • Eddy3D’s scores of \(V\) and \(W\) have a fractional bias (−6.15 and +2.15) that this repository leaves undefined for a signed component whose means are not both positive.

Measured against modelled

Case M, measured vs modelled velocity, k and concentration
Figure 2. Case M, WD 180: measured vs modelled \(U\), \(V\), \(W\), \(k\) and \(c^\ast\) at three turbulent Schmidt numbers (log axes show the positive pairs within eight decades). The last panel is the tunnel’s own WD 175 and WD 185 against its WD 180. The dots take the colour of their group in Figure 1.
Quantity R FAC2 FB Hit rate n
\(c^\ast\), \(\mathrm{Sc_t}\) 0.3 +0.412 0.200 +1.396 0.133 15
\(c^\ast\), \(\mathrm{Sc_t}\) 0.5 +0.251 0.200 +1.524 0.133 15
\(c^\ast\), \(\mathrm{Sc_t}\) 0.7 +0.198 0.133 +1.579 0.133 15
\(c^\ast\), \(\mathrm{Sc_t}\) 0.3, points turned 180° about the source (control) −0.050 0.071 +2.000 0.071 14
\(c^\ast\), \(\mathrm{Sc_t}\) 0.5, points turned 180° about the source (control) −0.058 0.071 +2.000 0.071 14
\(c^\ast\), \(\mathrm{Sc_t}\) 0.7, points turned 180° about the source (control) −0.058 0.071 +2.000 0.071 14
\(c^\ast\), \(\mathrm{Sc_t}\) 0.3, points mirrored across the plume axis (control) +0.675 0.467 +0.321 0.133 15
\(c^\ast\), \(\mathrm{Sc_t}\) 0.5, points mirrored across the plume axis (control) +0.655 0.400 +0.220 0.333 15
\(c^\ast\), \(\mathrm{Sc_t}\) 0.7, points mirrored across the plume axis (control) +0.650 0.400 +0.180 0.333 15
\(c^\ast\), tunnel WD 175 as a model of WD 180 (control) +0.932 0.733 +0.401 0.267 15
\(c^\ast\), tunnel WD 185 as a model of WD 180 (control) +0.976 0.800 −0.117 0.533 15
\(c^\ast\), mean of WD 175 and WD 185 as a model of WD 180 (control) +0.998 1.000 +0.111 0.867 15
\(c^\ast\), \(\mathrm{Sc_t}\) 0.3, against the field’s geometric mean +0.464 0.067 +1.012 0.000 15
\(c^\ast\), \(\mathrm{Sc_t}\) 0.5, against the field’s geometric mean +0.285 0.067 +1.202 0.000 15
\(c^\ast\), \(\mathrm{Sc_t}\) 0.7, against the field’s geometric mean +0.224 0.067 +1.287 0.000 15
\(c^\ast\), tunnel WD 180 as a model of the field’s geometric mean (control) +0.984 0.600 −0.593 0.000 15
\(U\) +0.708 0.267 +1.460 0.267 15
\(V\) +0.731 0.333 — 0.133 15
\(V\), sign flipped (control) −0.731 0.067 −0.616 0.133 15
\(W\) +0.181 0.267 — 0.733 15
\(\lvert\mathbf{U}\rvert\) +0.601 0.733 +0.185 0.400 15
\(k\) +0.446 0.133 +1.131 0.067 15

FB is positive when the model is low, and left out (—) for a signed component whose measured and modelled means are not both positive.

The concentration across the three Schmidt numbers:

Run \(c^\ast\) over the imposed speed at the mast: FB over 4.08 m/s: FB mean \(c^\ast\) tunnel mean \(c^\ast\) model peak tunnel peak model top five shared near three, tunnel model far two, tunnel model
\(\mathrm{Sc_t}\) 0.3 +1.480 +1.464 0.998 0.177 3.04 1.27 3 of 5 2.52 0.26 0.21 0.11
\(\mathrm{Sc_t}\) 0.5 +1.592 +1.579 0.998 0.135 3.04 1.37 3 of 5 2.52 0.11 0.21 0.10
\(\mathrm{Sc_t}\) 0.7 +1.640 +1.628 0.998 0.117 3.04 1.38 3 of 5 2.52 0.05 0.21 0.09
  • Tracer. R is 0.41, 0.25 and 0.20 and FAC2 0.20, 0.20 and 0.13 from \(\mathrm{Sc_t}\) 0.3 to 0.7; the mean \(c^\ast\) is 0.18, 0.13 and 0.12 against the tunnel’s 1.00 (FB +1.40 to +1.58), the peak 1.27 to 1.38 against 3.04, and three of the five highest points are shared. Over the imposed speed at the mast’s height (3.934 m/s) FB is +1.48 and over the tunnel’s 4.08 m/s +1.46 at \(\mathrm{Sc_t}\) 0.3 (the solved speed at the mast is 4.686 m/s, 1.19 of the imposed one).
  • Near and far points. The three points nearest the source carry \(c^\ast\) 2.52 in the tunnel and 0.26, 0.11 and 0.05 in the model; the two far downwind ones 0.21 in the tunnel and 0.11, 0.10 and 0.09 in the model. The tunnel’s near points exceed its far ones by a factor of about twelve; the model’s by 2.3 and 1.1 at \(\mathrm{Sc_t}\) 0.3 and 0.5, and at \(\mathrm{Sc_t}\) 0.7 they carry less (0.055 against 0.087).
  • Controls on the plume. The points turned 180° about the source (14 of the 15 lie in the fluid) score R −0.05 to −0.06 (the model’s mean over them is \(7 \times 10^{-6}\) against 0.19 at the real points, \(\mathrm{Sc_t}\) 0.3). The points mirrored across the plume axis score R +0.65 to +0.68, FAC2 0.40 to 0.47, against R +0.20 to +0.41 at the real points.
  • Neighbouring headings. The tunnel’s own WD 175 and WD 185 scored as models of its WD 180 reach FAC2 0.73 and 0.80 (R 0.93 and 0.98), and the mean of the two 1.00 (R 0.998). 8 of the 15 points change by more than a factor of two between WD 175 and WD 185 (up to 11 at point 14). The model’s FAC2 is 0.13 to 0.20.
  • Field campaign. The 10 of 18 runs within 10° of 180 at 3 m/s or more, each with its own \(U_R\) and tracer flow, give a geometric-mean \(c^\ast\) at each point; the spread of the field at point 1 is a factor of 4.5. The model against that mean scores R 0.46, 0.29 and 0.22, FAC2 0.07 and FB +1.01 to +1.29, and lies inside the field’s minimum–maximum range at 2 of the 15 points. The tunnel’s WD 180 against the same mean scores R 0.98, FAC2 0.60 and FB −0.59: the tunnel is higher than the field’s geometric mean at the same points (mean \(c^\ast\) 1.00 against 0.54).
  • Velocity. \(\lvert\mathbf{U}\rvert\): R 0.60, FAC2 0.73, FB +0.19, means 0.206 modelled against 0.247 measured. \(U\): R 0.71, FAC2 0.27, FB +1.46, the mean \(U/U_R\) 0.026 modelled against 0.164 measured. \(V\): R 0.73; flipping its sign gives −0.73. \(W\): R 0.18. \(k\): R 0.45, FAC2 0.13 and FB +1.13 (mean \(k/U_R^2\) 0.008 against 0.029).
Interpretation: Case M tracer, plume and velocity
  • The tracer is under-predicted at every Schmidt number and the pattern is weakly correlated; the choice of normalising speed does not change this. A higher Schmidt number lowers the modelled \(c^\ast\) further.
  • The tracer goes downwind. The mirrored points score better than the real points, so the modelled plume is displaced sideways against the tunnel’s.
  • The model’s FAC2 of 0.13 to 0.20 is below the 0.73 and 0.80 that the tunnel’s neighbouring headings score against its WD 180, and a steady solve at one heading is not a mean over headings.
  • The magnitude of the velocity is close and the streamwise component is not: \(V\) correlates and its sign agrees with the tunnel’s, \(W\) does not.
Case M, the tracer on the plan of the campus
Figure 3. Case M, WD 180: \(\log_{10}\) of \(c^\ast\) at the 15 points on the plan of the campus, the tunnel and the three Schmidt numbers on one scale (open: below \(10^{-3}\) or zero). The star is the gas source and the square the reference mast; the arrow is the wind.

Field maps

The solved velocity on the measured planes, every point drawn on it as a dot coloured by its measured value on the same scale: a dot that stands out from the colour around it is a point the model misses.

Solved field and measured points, Case M, plan
Figure 4. Case M, the pedestrian level of the campus (\(z = 35.5\thinspace\mathrm{m}\) above the tunnel floor), tunnel WD 180: (a) \(u\) over \(U_R\); (b) the same with iso-lines every 0.1 and the edge of the reversed flow, \(u = 0\), dashed; (c) the in-plane speed with streamlines.

OpenFOAM 12, realizable k-epsilon (Eddy3D Outdoor), configuration sct03

Quantity: \(U\), \(V\), \(W\) and \(k\) at the 15 points of the tunnel's WD 180 run over the tunnel's \(U_R\) = 4.08 m/s (measured) and over the solved speed at the reference mast (modelled; \(U_R\) is defined at the mast), and the dimensionless concentration \(c^\ast = C/c_0\) (\(c_0 = \dfrac{q}{U_R H_R^2}\) = 132.5 ppm) against the model's \(c\thinspace U_R H_R^2/Q\) for turbulent Schmidt number 0.3; the 18-run field campaign is compared as well.

All 16 headline scores
Score R \(R^2\) FAC2 FB Hit rate n
\(c^\ast\) +0.412 0.170 0.200 +1.396 0.133 15
\(c^\ast\), over the imposed speed at the mast's height (3.934 m/s) +0.412 0.170 0.200 +1.480 0.067 15
\(c^\ast\), over the tunnel's \(U_R\) = 4.08 m/s +0.412 0.170 0.200 +1.464 0.067 15
\(c^\ast\), points turned 180° about the source (control) -0.050 0.002 0.071 +2.000 0.071 14
\(c^\ast\), points mirrored across the plume axis (control) +0.675 0.456 0.467 +0.321 0.133 15
\(c^\ast\), tunnel WD 175 as a model of its WD 180 (control) +0.932 0.868 0.733 +0.401 0.267 15
\(c^\ast\), tunnel WD 185 as a model of its WD 180 (control) +0.976 0.952 0.800 -0.117 0.533 15
\(c^\ast\), mean of the tunnel's WD 175 and WD 185 as a model of its WD 180 (control) +0.998 0.995 1.000 +0.111 0.867 15
\(c^\ast\) against the field's geometric mean (10 runs) +0.464 0.215 0.067 +1.012 0.000 15
\(c^\ast\), tunnel WD 180 as a model of the field's geometric mean (control) +0.984 0.969 0.600 -0.593 0.000 15
\(U\) +0.708 0.502 0.267 +1.460 0.267 15
\(V\) +0.731 0.534 0.333 — 0.133 15
\(V\), sign flipped (control) -0.731 0.534 0.067 -0.616 0.133 15
\(W\) +0.181 0.033 0.267 — 0.733 15
\(\lvert\mathbf{U}\rvert\) +0.601 0.362 0.733 +0.185 0.400 15
\(k\) +0.446 0.199 0.133 +1.131 0.067 15
  • Normalisation: \(U\), \(V\), \(W\) and \(k\) at the 15 points over the tunnel's \(U_R\) = 4.08 m/s on the measurement and over the solved speed at the reference mast on the model (\(U_R\) is defined at the mast), \(k\) over \(U_R^2\); \(c^\ast = C/c_0\) (\(c_0 = \dfrac{q}{U_R H_R^2}\) = 132.5 ppm) against the model's \(c\thinspace U_R H_R^2/Q\) (\(H_R\) = 0.068 m x 600, \(Q = 1\thinspace\mathrm{kg/s}\)) for turbulent Schmidt number 0.3, with one detection limit of \(10^{-3}\) of the largest measured value; hit rate \(\delta_{\mathrm{rel}} = 0.25\), \(\delta_{\mathrm{abs}} = 0.05\) in those units; the Schmidt numbers are three runs on one flow
  • Reproduce: Eddy3D fixture Eddy3DCli.Tests/Validation/TestAijCaseMContainerE2E (WD 180, the default) (TPU Atsugi campus, 1:600, tunnel WD 180, turbulent Schmidt number 0.3; OpenFOAM 12 realizable k-ε, numerics level 3, 3000 iterations, 1,394,053 cells, solved 2026-10-18); figures validation/plot_cases_fm_openfoam.py

Notes:

  • \(U_R\) is the solved speed at the reference mast, 4.686 m/s, which is 1.19 of the tabulated inlet's speed at the mast's elevation (3.934 m/s) and 1.15 of the tunnel's 4.08 m/s. Over the imposed speed \(c^\ast\) scores FB +1.48 and over 4.08 m/s FB +1.46, against +1.40 over the solved speed.
  • Controls on the same solution: the points turned 180° about the source score R -0.05, FAC2 0.07, FB +2.00 (14 of the 15 lie in the fluid; the model's mean over those points is 7.3e-06 against 0.186 at the real ones), and the points mirrored across the plume axis score R +0.68, FAC2 0.47, FB +0.32 (the real points: R +0.41, FAC2 0.20, FB +1.40).
  • Field campaign (18 runs): the 10 runs within 10° of 180 at 3 m/s or more, each with its own \(U_R\) and q and \(H_R\) = 41 m as the cleaned table states (the scores divide by 40.8 m, a 1 % difference). The model against the field's geometric mean scores R +0.46, FAC2 0.07, FB +1.01, and lies inside the field's minimum-maximum range at 2 of 15 points; the tunnel's WD 180 against the same mean scores R +0.98, FAC2 0.60, FB -0.59; the spread of the field at point 1 is a factor of 4.5.

OpenFOAM 12, realizable k-epsilon (Eddy3D Outdoor), configuration sct05

Quantity: \(U\), \(V\), \(W\) and \(k\) at the 15 points of the tunnel's WD 180 run over the tunnel's \(U_R\) = 4.08 m/s (measured) and over the solved speed at the reference mast (modelled; \(U_R\) is defined at the mast), and the dimensionless concentration \(c^\ast = C/c_0\) (\(c_0 = \dfrac{q}{U_R H_R^2}\) = 132.5 ppm) against the model's \(c\thinspace U_R H_R^2/Q\) for turbulent Schmidt number 0.5; the 18-run field campaign is compared as well.

All 16 headline scores
Score R \(R^2\) FAC2 FB Hit rate n
\(c^\ast\) +0.251 0.063 0.200 +1.524 0.133 15
\(c^\ast\), over the imposed speed at the mast's height (3.934 m/s) +0.251 0.063 0.133 +1.592 0.067 15
\(c^\ast\), over the tunnel's \(U_R\) = 4.08 m/s +0.251 0.063 0.200 +1.579 0.133 15
\(c^\ast\), points turned 180° about the source (control) -0.058 0.003 0.071 +2.000 0.071 14
\(c^\ast\), points mirrored across the plume axis (control) +0.655 0.429 0.400 +0.220 0.333 15
\(c^\ast\), tunnel WD 175 as a model of its WD 180 (control) +0.932 0.868 0.733 +0.401 0.267 15
\(c^\ast\), tunnel WD 185 as a model of its WD 180 (control) +0.976 0.952 0.800 -0.117 0.533 15
\(c^\ast\), mean of the tunnel's WD 175 and WD 185 as a model of its WD 180 (control) +0.998 0.995 1.000 +0.111 0.867 15
\(c^\ast\) against the field's geometric mean (10 runs) +0.285 0.081 0.067 +1.202 0.000 15
\(c^\ast\), tunnel WD 180 as a model of the field's geometric mean (control) +0.984 0.969 0.600 -0.593 0.000 15
\(U\) +0.708 0.502 0.267 +1.460 0.267 15
\(V\) +0.731 0.534 0.333 — 0.133 15
\(V\), sign flipped (control) -0.731 0.534 0.067 -0.616 0.133 15
\(W\) +0.181 0.033 0.267 — 0.733 15
\(\lvert\mathbf{U}\rvert\) +0.601 0.362 0.733 +0.185 0.400 15
\(k\) +0.446 0.199 0.133 +1.131 0.067 15
  • Normalisation: \(U\), \(V\), \(W\) and \(k\) at the 15 points over the tunnel's \(U_R\) = 4.08 m/s on the measurement and over the solved speed at the reference mast on the model (\(U_R\) is defined at the mast), \(k\) over \(U_R^2\); \(c^\ast = C/c_0\) (\(c_0 = \dfrac{q}{U_R H_R^2}\) = 132.5 ppm) against the model's \(c\thinspace U_R H_R^2/Q\) (\(H_R\) = 0.068 m x 600, \(Q = 1\thinspace\mathrm{kg/s}\)) for turbulent Schmidt number 0.5, with one detection limit of \(10^{-3}\) of the largest measured value; hit rate \(\delta_{\mathrm{rel}} = 0.25\), \(\delta_{\mathrm{abs}} = 0.05\) in those units; the Schmidt numbers are three runs on one flow
  • Reproduce: Eddy3D fixture Eddy3DCli.Tests/Validation/TestAijCaseMContainerE2E (WD 180, the default) (TPU Atsugi campus, 1:600, tunnel WD 180, turbulent Schmidt number 0.5; OpenFOAM 12 realizable k-ε, numerics level 3, 3000 iterations, 1,394,053 cells, solved 2026-10-18); figures validation/plot_cases_fm_openfoam.py

Notes:

  • \(U_R\) is the solved speed at the reference mast, 4.686 m/s, which is 1.19 of the tabulated inlet's speed at the mast's elevation (3.934 m/s) and 1.15 of the tunnel's 4.08 m/s. Over the imposed speed \(c^\ast\) scores FB +1.59 and over 4.08 m/s FB +1.58, against +1.52 over the solved speed.
  • Controls on the same solution: the points turned 180° about the source score R -0.06, FAC2 0.07, FB +2.00 (14 of the 15 lie in the fluid; the model's mean over those points is 9.2e-06 against 0.143 at the real ones), and the points mirrored across the plume axis score R +0.65, FAC2 0.40, FB +0.22 (the real points: R +0.25, FAC2 0.20, FB +1.52).
  • Field campaign (18 runs): the 10 runs within 10° of 180 at 3 m/s or more, each with its own \(U_R\) and q and \(H_R\) = 41 m as the cleaned table states (the scores divide by 40.8 m, a 1 % difference). The model against the field's geometric mean scores R +0.28, FAC2 0.07, FB +1.20, and lies inside the field's minimum-maximum range at 2 of 15 points; the tunnel's WD 180 against the same mean scores R +0.98, FAC2 0.60, FB -0.59; the spread of the field at point 1 is a factor of 4.5.

OpenFOAM 12, realizable k-epsilon (Eddy3D Outdoor), configuration sct07

Quantity: \(U\), \(V\), \(W\) and \(k\) at the 15 points of the tunnel's WD 180 run over the tunnel's \(U_R\) = 4.08 m/s (measured) and over the solved speed at the reference mast (modelled; \(U_R\) is defined at the mast), and the dimensionless concentration \(c^\ast = C/c_0\) (\(c_0 = \dfrac{q}{U_R H_R^2}\) = 132.5 ppm) against the model's \(c\thinspace U_R H_R^2/Q\) for turbulent Schmidt number 0.7; the 18-run field campaign is compared as well.

All 16 headline scores
Score R \(R^2\) FAC2 FB Hit rate n
\(c^\ast\) +0.198 0.039 0.133 +1.579 0.133 15
\(c^\ast\), over the imposed speed at the mast's height (3.934 m/s) +0.198 0.039 0.133 +1.640 0.067 15
\(c^\ast\), over the tunnel's \(U_R\) = 4.08 m/s +0.198 0.039 0.133 +1.628 0.133 15
\(c^\ast\), points turned 180° about the source (control) -0.058 0.003 0.071 +2.000 0.071 14
\(c^\ast\), points mirrored across the plume axis (control) +0.650 0.423 0.400 +0.180 0.333 15
\(c^\ast\), tunnel WD 175 as a model of its WD 180 (control) +0.932 0.868 0.733 +0.401 0.267 15
\(c^\ast\), tunnel WD 185 as a model of its WD 180 (control) +0.976 0.952 0.800 -0.117 0.533 15
\(c^\ast\), mean of the tunnel's WD 175 and WD 185 as a model of its WD 180 (control) +0.998 0.995 1.000 +0.111 0.867 15
\(c^\ast\) against the field's geometric mean (10 runs) +0.224 0.050 0.067 +1.287 0.000 15
\(c^\ast\), tunnel WD 180 as a model of the field's geometric mean (control) +0.984 0.969 0.600 -0.593 0.000 15
\(U\) +0.708 0.502 0.267 +1.460 0.267 15
\(V\) +0.731 0.534 0.333 — 0.133 15
\(V\), sign flipped (control) -0.731 0.534 0.067 -0.616 0.133 15
\(W\) +0.181 0.033 0.267 — 0.733 15
\(\lvert\mathbf{U}\rvert\) +0.601 0.362 0.733 +0.185 0.400 15
\(k\) +0.446 0.199 0.133 +1.131 0.067 15
  • Normalisation: \(U\), \(V\), \(W\) and \(k\) at the 15 points over the tunnel's \(U_R\) = 4.08 m/s on the measurement and over the solved speed at the reference mast on the model (\(U_R\) is defined at the mast), \(k\) over \(U_R^2\); \(c^\ast = C/c_0\) (\(c_0 = \dfrac{q}{U_R H_R^2}\) = 132.5 ppm) against the model's \(c\thinspace U_R H_R^2/Q\) (\(H_R\) = 0.068 m x 600, \(Q = 1\thinspace\mathrm{kg/s}\)) for turbulent Schmidt number 0.7, with one detection limit of \(10^{-3}\) of the largest measured value; hit rate \(\delta_{\mathrm{rel}} = 0.25\), \(\delta_{\mathrm{abs}} = 0.05\) in those units; the Schmidt numbers are three runs on one flow
  • Reproduce: Eddy3D fixture Eddy3DCli.Tests/Validation/TestAijCaseMContainerE2E (WD 180, the default) (TPU Atsugi campus, 1:600, tunnel WD 180, turbulent Schmidt number 0.7; OpenFOAM 12 realizable k-ε, numerics level 3, 3000 iterations, 1,394,053 cells, solved 2026-10-18); figures validation/plot_cases_fm_openfoam.py

Notes:

  • \(U_R\) is the solved speed at the reference mast, 4.686 m/s, which is 1.19 of the tabulated inlet's speed at the mast's elevation (3.934 m/s) and 1.15 of the tunnel's 4.08 m/s. Over the imposed speed \(c^\ast\) scores FB +1.64 and over 4.08 m/s FB +1.63, against +1.58 over the solved speed.
  • Controls on the same solution: the points turned 180° about the source score R -0.06, FAC2 0.07, FB +2.00 (14 of the 15 lie in the fluid; the model's mean over those points is 1.0e-05 against 0.125 at the real ones), and the points mirrored across the plume axis score R +0.65, FAC2 0.40, FB +0.18 (the real points: R +0.20, FAC2 0.13, FB +1.58).
  • Field campaign (18 runs): the 10 runs within 10° of 180 at 3 m/s or more, each with its own \(U_R\) and q and \(H_R\) = 41 m as the cleaned table states (the scores divide by 40.8 m, a 1 % difference). The model against the field's geometric mean scores R +0.22, FAC2 0.07, FB +1.29, and lies inside the field's minimum-maximum range at 2 of 15 points; the tunnel's WD 180 against the same mean scores R +0.98, FAC2 0.60, FB -0.59; the spread of the field at point 1 is a factor of 4.5.

Not run

  • A second mesh or iteration count.
  • WD 175 and WD 185, which the tunnel also measured. Model set-up: a cylinder on the terrain sheet, none of the tunnel’s side walls and ceiling, a smooth floor under the tabulated inflow. Its source is a 4 m cube of 20 cells where the tunnel’s outlet is 6 mm (3.6 m at 1:600).
  • OpenLB, FluidX3D, UMF and the ML surrogate.
Interpretation: Limits of this run
  • With one mesh and the residual control unmet, none of these numbers is mesh- or iteration-independent.
  • A finer source or a finer surface mesh was not tried, so the source size’s effect on the tracer scores is unknown.

Notes on every run

Notes:

  • One heading and one steady solve: WD 180, the only heading with measured velocities. The tunnel's WD 175 scored as a model of its WD 180 gives R +0.93, FAC2 0.73, FB +0.40, its WD 185 R +0.98, FAC2 0.80, FB -0.12, and the mean of the two R +1.00, FAC2 1.00, FB +0.11; 8 of the 15 points change by more than a factor of two between WD 175 and WD 185 (computed from the pairs).
  • The turbulent Schmidt number does not change the flow: the three runs of this case are one solve with three passive tracers, so \(U\), \(V\), \(W\) and \(k\) are the same in all three and only \(c^\ast\) differs.
  • Model set-up: a cylinder of 540 m radius and 240 m height on the terrain sheet; the tabulated approach profile over a smooth floor; no counterpart of the tunnel's side walls (at ±360 m) and 600 m ceiling (blockage 5 %). The 6 mm outlet is 3.6 m at 1:600; the source here is a 4 m cube on the ground (a zone of 20 cells, 2 m surface cells), 1 kg/s as a volume source with no exit momentum. The 15 points are 3 m above the ground.
  • Mean \(c^\ast\) over the three points nearest the source: 2.52 (tunnel), 0.26 / 0.11 / 0.05 (model, \(\mathrm{Sc_t}\) 0.3 / 0.5 / 0.7); over the two far downwind points: 0.21 (tunnel), 0.11 / 0.10 / 0.09 (model). Peak \(c^\ast\): 3.04 (tunnel), 1.27 / 1.37 / 1.38 (model); 3 / 3 / 3 of the five highest points are shared with the tunnel.
  • The velocity at the 15 points: the mean \(U/U_R\) is 0.026 modelled against 0.164 measured. FB is not defined for \(V\) and \(W\) (their means are not both positive); \(V\) correlates at R +0.73 and flipping its sign gives -0.73.
  • The solve ran to its 3000-iteration limit with the residualControl unmet (no 'solution converged' line); the initial residuals at the last iteration were \(U_x\) 6.1e-04, \(U_z\) 6.9e-04, \(p\) 1.0e-04, \(\varepsilon\) 1.9e-02, \(k\) 1.6e-04 and the tracers 8.1e-06 to 1.1e-05. checkMesh: Mesh OK. 1,394,053 cells. The tracer probe prints seven significant digits; a probe at six decimals moved no \(c^\ast\) value by more than 0.003.
Interpretation: Case M near-source ordering

The tunnel's three points nearest the source carry about 12 times its two far downwind ones; the modelled plume carries less, and \(\mathrm{Sc_t}\) 0.7 reverses the order.

Interpretation: Case M controls

The tracer goes downwind: the model's mean over the points turned 180° about the source is far below its mean at the real points. The mirrored points score above the real points because the modelled plume is displaced sideways against the tunnel's. The tunnel's neighbouring headings WD 175 and WD 185 are the noise floor of a single heading.

How the scores are computed
  • Scores recomputed from the pairs by validation/dispersion_metrics.py; the SHA-256 of every file the snapshot was read from is in its results file, under sources.

Leaderboard

Rank Run Source Hit rate, \(U\) Hit rate, \(V\) Hit rate, \(W\) Hit rate, \(\lvert\mathbf{U}\rvert\) Hit rate, \(k\) Hit rate, \(c^\ast\) Hit rate FAC2 R \(R^2\)
1 OpenFOAM 12 · \(\mathrm{Sc_t}\) 0.3 Eddy3D 0.267 0.133 0.733 0.400 0.067 0.133 0.289 0.322 +0.513 0.300
2 OpenFOAM 12 · \(\mathrm{Sc_t}\) 0.5 Eddy3D 0.267 0.133 0.733 0.400 0.067 0.133 0.289 0.322 +0.487 0.282
3 OpenFOAM 12 · \(\mathrm{Sc_t}\) 0.7 Eddy3D 0.267 0.133 0.733 0.400 0.067 0.133 0.289 0.311 +0.478 0.278
4 OpenFOAM simpleFoam (Eddy3D Outdoor case writer) (eddy3d-of12-nutcap-lifted) community 0.133 0.133 0.667 0.533 0.133 0.067 0.278 0.311 +0.467 0.255

All runs scored by validation/score_submissions.py. Ranked by mean hit rate (VDI 3783) over the tasks; bold: best per column. Not ranked: the ML surrogate (no point ids) and negative controls. Community rows are self-reported (Validate your own solver).