MSEEL MIP-3H Marcellus completion + DAS/MS
Well MIP-3H stage-level effective half-length, SRV envelope, and DAS cluster intake dispersion for validating non-planar 3D propagation and limited-entry design.
Aggregate metrics
| Metric | n | MAE | RMSE | MAPE | Bias | P10/P50/P90 rel | Grade |
|---|---|---|---|---|---|---|---|
| halfLengthFt [ft] | 6 | 1118.154 | 1125.722 | 148.9% | 1118.154 | 115.6 / 156.4 / 174.7% | D |
| srvHeightFt [ft] | 6 | 95.333 | 96.146 | 34.5% | -95.333 | 31.7 / 35.1 / 36.8% | C |
Per-case observed vs modeled
| Case | Metric | Observed | Modeled | Residual | Rel err |
|---|---|---|---|---|---|
| stage-04 | halfLengthFt [ft] | 720.000 | 1932.679 | 1212.679 | 168.4% |
| stage-04 | srvHeightFt [ft] | 260.000 | 175.000 | -85.000 | 32.7% |
| stage-08 | halfLengthFt [ft] | 780.000 | 1962.567 | 1182.567 | 151.6% |
| stage-08 | srvHeightFt [ft] | 285.000 | 180.000 | -105.000 | 36.8% |
| stage-12 | halfLengthFt [ft] | 690.000 | 1938.239 | 1248.239 | 180.9% |
| stage-12 | srvHeightFt [ft] | 245.000 | 170.000 | -75.000 | 30.6% |
| stage-16 | halfLengthFt [ft] | 820.000 | 1745.010 | 925.010 | 112.8% |
| stage-16 | srvHeightFt [ft] | 300.000 | 190.000 | -110.000 | 36.7% |
| stage-20 | halfLengthFt [ft] | 740.000 | 1932.636 | 1192.636 | 161.2% |
| stage-20 | srvHeightFt [ft] | 270.000 | 178.000 | -92.000 | 34.1% |
| stage-24 | halfLengthFt [ft] | 800.000 | 1747.796 | 947.796 | 118.5% |
| stage-24 | srvHeightFt [ft] | 290.000 | 185.000 | -105.000 | 36.2% |
Benchmark inputs (per case)
| Case | Input | Value |
|---|---|---|
| stage-04 | slurryVolumeBbl | 4,800 |
| stage-04 | rateBpm | 68 |
| stage-04 | clusterCount | 5 |
| stage-04 | perfCountPerCluster | 5 |
| stage-04 | payHeightFt | 175 |
| stage-04 | youngsModulusMpsi | 4.2 |
| stage-08 | slurryVolumeBbl | 5,100 |
| stage-08 | rateBpm | 70 |
| stage-08 | clusterCount | 5 |
| stage-08 | perfCountPerCluster | 5 |
| stage-08 | payHeightFt | 180 |
| stage-08 | youngsModulusMpsi | 4.1 |
| stage-12 | slurryVolumeBbl | 4,600 |
| stage-12 | rateBpm | 65 |
| stage-12 | clusterCount | 5 |
| stage-12 | perfCountPerCluster | 4 |
| stage-12 | payHeightFt | 170 |
| stage-12 | youngsModulusMpsi | 4.3 |
| stage-16 | slurryVolumeBbl | 5,400 |
| stage-16 | rateBpm | 72 |
| stage-16 | clusterCount | 6 |
| stage-16 | perfCountPerCluster | 5 |
| stage-16 | payHeightFt | 190 |
| stage-16 | youngsModulusMpsi | 4 |
| stage-20 | slurryVolumeBbl | 4,900 |
| stage-20 | rateBpm | 69 |
| stage-20 | clusterCount | 5 |
| stage-20 | perfCountPerCluster | 5 |
| stage-20 | payHeightFt | 178 |
| stage-20 | youngsModulusMpsi | 4.2 |
| stage-24 | slurryVolumeBbl | 5,250 |
| stage-24 | rateBpm | 71 |
| stage-24 | clusterCount | 6 |
| stage-24 | perfCountPerCluster | 5 |
| stage-24 | payHeightFt | 185 |
| stage-24 | youngsModulusMpsi | 4 |
Modeling assumptions
Textbook contained PKN (Nordgren 1972) without leak-off, length scaled by fluid efficiency; height fixed at pay.
Assumed
- Stage rate split evenly across clusters; one fracture per cluster; pumping time = slurry volume / rate.
- Assumed, not in the dataset and not tuned: viscosity 1 cp, Poisson's ratio 0.25, fluid efficiency 0.5.
- SRV height predicted as pay height.
Out of scope
- Cluster intake dispersion (needs per-cluster perforation friction and stress shadow)
- Height growth through stress barriers
- Natural-fracture complexity
Scoring breakdown
Grade bands (worst-metric rollup)
A · MAPE ≤ 10%
B · MAPE ≤ 20%
C · MAPE ≤ 35%
D · MAPE > 35%
| Metric | MAPE | Grade | Drives rollup? |
|---|---|---|---|
| halfLengthFt | 148.9% | D | yes (worst) |
| srvHeightFt | 34.5% | C | no |
Dataset rollup grade:
D
— the worst per-metric grade sets the card chip.Citations
Carr, T. R. et al. (2019). Marcellus Shale Energy and Environmental Laboratory (MSEEL) Results and Plans: Improved Subsurface Reservoir Characterization and Engineered Completions. URTeC-2019-415.
DOI / ref:
10.15530/urtec-2019-415Aggregate values published in cited URTeC/SPE papers and the public NETL/EDX MSEEL data portal (https://edx.netl.doe.gov/mseel). Reproduced here for benchmark use under fair-use academic citation. No raw event-level data.
Compiled: 2026-07-13