Capabilities

Wellbore Genius capabilities

From fracture physics and live pad operations to CO₂ storage and geothermal design. Explore the complete engineering toolkit, with connected workflows that turn calculations into decisions.

Reviewed 2026-10-14 · Implemented workflows

What sets Wellbore Genius apart

Forecast before pumping. Score after.

Keep pre-stage predictions, actual outcomes and misses together instead of losing the forecast when the job changes.

Pre-stage half-length and ISIP from typed rock and fluid inputs
Forecast ledger with provenance and deletion
Pad history with pressure, rate, time and forecast-error views
Held-out forecast learning and automatic scoring when a stage completes

Model notes & assumptions

Not proprietary vendor physics or a field-calibrated fracture model. Missing rock/fluid values remain assumptions; scoring requires completed stages with actuals.

Pad history

See the next stage in the context of the pad.

Carry stage-order stress screening into the pressure table, forecast adjustment and fracture grid.

Shared interference settings across the planner, grid and forecasts
Stage-by-stage stress and pressure cells
v1/v2/v3 per-model scoring on the same forecast
Optional stress relaxation and 1-D poroelastic back-stress

Model notes & assumptions

One-way pressure-to-stress coupling and uniform rock, not a full coupled field-pressure simulation. Relaxation time and Biot/Poisson inputs are assumptions unless measured; unsupplied site properties are not measurements.

Pad fracture grid

Connect core evidence to a storage-site plan.

Take reservoir and seal entries through injection sizing, trapped tonnes, yearly plume movement and cell-level pressure.

Measured core-plug Land fit with source and condition caveats
Temperature/pressure-dependent CO₂ properties
Editable site plan, injection timeline, field grid and seal-pressure sweep
Account-synced CO₂ and reservoir inputs

Model notes & assumptions

Uniform-rock screening, not a heterogeneous compositional simulator or containment certification. Full-pad seal checks superpose every well's injection pressure and model the front through shut-in, but the Darcy front is conservative and never retreats.

CO₂ site plan

Ask for a goal. Get calculated candidates.

Rank geothermal designs against NPV or IRR goals and pump-power constraints using executed calculations, not invented answers.

Copilot goal parsing and a ranked feasible shortlist
Thermal decline, price escalation and pipe-loss-inclusive net power
Saved scenario comparison and live what-if controls

Model notes & assumptions

A search over the selected grid, not a guaranteed global optimum. Thermal decline and prices are planning assumptions, not measured production forecasts.

Geothermal designs

Explore more cases without changing the authority.

Use browser GPU previews for independent sensitivity cells while keeping a double-precision CPU reference and automatic fallback.

Geothermal NPV/design grids
CO₂ permeability × dip migration sweeps
Fracture-geometry ensembles

Model notes & assumptions

Hardware-dependent acceleration, not a claimed speedup. CPU calculations remain grading/export authority; the CO₂ field-grid map is a separate screening workflow.

Sensitivity workflows

Keep the evidence beside the answer.

Review model tiers, sources, assumptions and verification evidence before using a result in an engineering decision.

Versioned solver capability descriptors and tier-aware receipts
Supported features separated from proof levels
Measured PVT report follows the account
Shut-in interpretation with explicit sample/window refusals

Model notes & assumptions

Shipped does not mean field-validated. Partial coupling, ungraded references and missing measurements stay visible; no autonomous pump control is claimed.

Models and limits

The engineering essentials are here, too

Well testing, reservoir balance and well construction calculations using published engineering methods.

Well test Material balance Casing design Torque & drag Cement All tools

The full feature catalog

82 capabilities · 5 engineering disciplines

✓Full⚠Partial✗Missing—N/A or unverified

Explore every capability

82 capabilities across fracture physics, reservoir coupling, operations, diagnostics and reporting — 82 supported, 0 partial, 0 planned.

Verification & model evidence

Feature availability and validation level are tracked separately. Validation is graded per physics module on /trust: 1 of 30 modules field-validated, 6 benchmarked, 0 analytical, 23 research / not yet graded. Several capabilities share one module, so counted per capability that is 1 field-validated, 10 benchmarked, 0 analytical and 71 research.

Capability	Status	Proof (per /trust)	Justification	Module
Frac physics
	Supported	Benchmarked	Planar 3D + comparator across PKN / KGD / radial	Open →
	Supported	Research	DDM + curvature + out-of-plane tilt + reconvergence weld	Open →
	Supported	Research	AT-1/AT-2 damage + He-Hutchinson kink/pierce	Open →
	Supported	Research	Two-criterion Mohr-Coulomb classifier + play presets	Open →
	Supported	Benchmarked	Geertsma + 3D Mindlin + Δσxy/xz/yz nucleus-of-strain	Open →
	Supported	Benchmarked	Per-bench (ν,α), depth-attenuated	Open →
	Supported	Benchmarked	Phase 9 segregated implicit T → P → σh kernel	Open →
	Supported	Research	DFN graph + RP/GW dispatcher over DDM tips	Open →
	Supported	Research	Inside a project workspace: Per-factor Spearman tornadoes	Open →
	Supported	Benchmarked	Barenblatt-style cohesive tip w/ Zod-validated inputs + public REST endpoint /api/public/cohesive-zone/compute + stable serialized error payload	Open →
	Supported	Research	XFEM + peridynamics tracks compared with published UXFEM 2023 and Fu 2013 results; evidence link pending	Open →
	Supported	Research	DP-cap plasticity module for compaction & pore-collapse regimes	Open →
Reservoir & coupling
	Supported	Research	Phase 14 IMPES + tracers + fluid regions	Open →
	Supported	Research	Peng-Robinson energized-CO₂ gas-FVF bridge	Open →
	Supported	Research	Per-cell, per-species upstream-flux on top of IMPES	Open →
	Supported	Research	Terzaghi + monotonic loading + hysteresis-aware unloading	Open →
	Supported	Research	Full saturation-function library, drop-in for 3-phase IMPES	Open →
	Supported	Research	nx0 / 2nx0 / 4nx0 × upwind/TVD with scheme recommender	Open →
	Supported	Research	v2.0.1 EPC bundle, validator surfaces dangling refs	Open →
	Supported	Research	Native GRDECL writer w/ property cubes	Open →
	Supported	Research	PH/PS flash EoS + hydrate formation curve wired into choke and geothermal-injector advisors	Open →
	Supported	Research	Round-trippable per-layer rock-property CSV (E, ν, UCS, φ, k) with template + validation	Open →
	Supported	Research	Analytical timestep loop `runFdiBidirectionalLoop` couples Geertsma Δσ_h into per-step half-length growth and inverts an optional SWPM residual into a rate-delta advisory. Sprint 1b.	Open →
	Supported	Research	EDFM connectivity index + Warren-Root λ/ω dual-porosity transfer w/ CSV round-trip; wired into reservoir coupling backlog	Open →
	Supported	Research	One-click regional PVT presets tied to the compositional-3P and gas-lift IPR paths	Open →
	Supported	Research	Separated-mode neural operator: 10M+ node pressure / stress fields without cubic memory growth	Open →
	Supported	Research	Stage-parallel relaxation against a predicted interference field; serial pump-order run is the asserted fixed point, equivalence badge on every run	Open →
	Supported	Research	Signed Δσ per stage from stored parent wells on other pads; state + physics hashes stamped into every receipt	Open →
Wellbore & ops
	Supported	Benchmarked	Phase 8 segregated implicit pipe w/ two-way coupling + α frontier	Open →
	Supported	Benchmarked	Zuber-Findlay + DAK + Sutton + Wichert-Aziz sour-gas	Open →
	Supported	Research	4-limit window w/ severity ladder	Open →
	Supported	Research	5-method screening + pad-shared compressor allocator	Open →
	Supported	Research	Lubinski sinusoidal + Dawson-Paslay helical buckling	Open →
	Supported	Research	Per-stage plug + perf-gun + diversion scheduler	Open →
	Supported	Research	8 grades / 13 tubing / 10 casing / 9 packers / 6 plugs / 5 perf guns	Open →
	Supported	Research	autoSelectTubing heuristic w/ hard safety-factor filters	Open →
	Supported	Benchmarked	Beggs-Brill multiphase pressure-drop solver alongside drift-flux and single-phase paths	Open →
Diagnostics
	Supported	Research	All 4 + auto-detect + ISIP auto-pick + live HM	Open →
	Supported	Research	5-way transport picker + DAS→cluster ε→geometry inversion	Open →
	Supported	Research	Vavryčuk ISO/DC/CLVD + Brune source radius + DFN bridge	Open →
	Supported	Benchmarked	3-method classifier + closure picker + Nolte falloff	Open →
	Supported	Field-validated	Swept-ellipse capture + Σ ≤ 1 mass conservation + tip-deflection	Open →
	Supported	Research	DFIT · Nolte-Smith · DeclineAnalysis · StepDownAuto · IsipAutoPick · microseismic · CompletionQualityIndex	Open →
	Supported	Research	WITSML 2.0 log ingest wired into live-pumping transport picker	Open →
	Supported	Research	WebGPU volume ray-march + marching-cubes iso-surface + async probe-at-cursor with epoch guard + Timeline4D scrubber + shareable playback URL	Open →
	Supported	Research	Three cards mounted on /live-pumping (EntryFrictionHistogramCard · PreClosureInterpreterCard · FluidFrontPlayerCard) + free public /tools/dfit-interpreter with pre-closure regime classifier and step-down mini-tool. Sprint 1a.	Open →
	Supported	Research	Inverts DAS intake fractions to ε_i and feeds PKN L_i/L_mean=(V/V̄)^(2/3)·ε^(1/3); mounted under the DAS section	Open →
	Supported	Research	Dipole strain kernel reproducing LF-DAS rabbit ears, DTS warmback, seeded MC bands, CSV-estimated measurement noise, HDF5/SEG-Y raw vendor ingest	Open →
Workflow & reporting
	Supported	Research	T1-T5 sweep + per-algorithm penalty catalog + Pareto frontier	Open →
	Supported	Research	Crew / pump / sand pools + zipper-stagger + surrogate-tuned	Open →
	Supported	Research	OFAT sweep + diagnostic tornado + Find-Optimum + cohort rollup	Open →
	Supported	Research	Wellbore-sensitivity route · Quick 5-factor tornado + Custom factor-picker · per-workspace depth persistence	Open →
	Supported	Research	Calibration PDF + 17-section FracPro-2021 parity + Subpart RR	Open →
	Supported	Research	Git-style fork tree + per-leaf author/version attribution	Open →
	Supported	Research	⌘K palette + Gemini-backed copilot picks above lexical matches	Open →
	Supported	Research	Hands-free command intake from any route	Open →
	Supported	Research	Dieterich rate response + OK_CC / KS_CC / TX_RRC presets + audit log	Open →
	Supported	Research	Hash-chained ledger + Subpart RR PDF + field cross-check	Open →
	Supported	Research	Find optimum (NM+PSO) · Pareto sweep · per-algorithm penalties · surrogate auto-tuner	Open →
	Supported	Research	parent-child poroelastic + multi-pad scheduler + artificial-lift allocator	Open →
	Supported	Research	Fleet-prior tightening + Gemini copilot picks + surrogate response-surface OFAT	Open →
	Supported	Benchmarked	Four peer-reviewed HF benchmarks with numeric residuals, citations, and PDF export	Open →
	Supported	Research	In-app HSE incident logging, hazard checklists, and daily HSE report export	Open →
	Supported	Research	Node CLI + Python SDK: workflow JSON runner, importers (LAS3, GRDECL, RESQML, WITSML2, ProdML, Pipesim, Eclipse RSM), exporters (RESQML, VTU, FracPro, CMG IMEX, Aries)	Open →
	Supported	Research	Five 2024-2026 mandates rolled into the existing MRV/TLP infra with one-click exports	Open →
	Supported	Research	SKIP LOCKED FIFO claiming, per-team concurrency lanes, queue-health card, retry/backoff, p90 SLOs with root-cause labelling; unlimited compute by default	Open →
	Supported	Research	Suggested studies ranked from containment / tip-regime / screen-out / offset-interference signals; engineer approves before anything queues	Open →
	Supported	Research	Agentic sweep designer emits a reviewable batch spec; no run is queued without approval	Open →
	Supported	Research	Drift watcher proposes clamped parameter updates with RMSE-before/after, bias and observation counts; accepted updates are recorded with a lineage digest	Open →
	Supported	Research	Deterministic markdown notice: evidence, materiality band, per-programme relevance across the five regulatory-pack-v2 modules, verifiable lineage digest	Open →
	Supported	Research	Lexical + semantic recall over prior batches, designs and outcomes with citations back to the run	Open →
	Supported	Research	Every agent turn carries a digest over its inputs, a replay diff against the stored trace, and an exportable audit trail	Open →
	Supported	Research	Ten typed MCP tools incl. search_prior_studies, propose_sweep and interpret_batch, each team-authorised and rate-limited	Open →
	Supported	Research	Cinematic render tiers, figure-mode export, versioned scene links, panel field-data CSV/JSON, OOM-safe chunked animation capture	Open →
	Supported	Research	Settings-signature batching, Kahn dependency waves, retry/backoff with blocked-dependent cascade, locked schedule CSV	Open →
	Supported	Research	Every advertised capability is checked against the code path that delivers it, exported as a CI snapshot	Open →
	Supported	Research	Monotonicity / scaling / symmetry / conservation relations with per-family tolerance bands and evidence CSV	Open →
	Supported	Research	PDE/BC/IC residual training against transient pressure diffusion, gated promotion, persisted run log + CSV	Open →
	Supported	Research	`.wbg.depletionState` envelopes, input diff with psi attribution, selective sync with undo/conflict detection, cross-pad consistency + auto-align, sync audit trail	Open →
	Supported	Research	Runs the three lanes together, what-if knob panel, one combined CSV receipt, posteriors persisted to the pad/acreage model	Open →
	Supported	Research	Mass-conservation, monotonicity, scaling and dual-porosity-limit invariants across a case matrix; schema-versioned, provenance-stamped, downloadable receipt	Open →
	Supported	Research	Multi-turn tool calls against the app's own solvers, mandatory citations back to stored posterior records, fallback audit log, streaming answers	Open →

Check it yourself — pencil-and-paper cases

6 / 6 match

Closed-form textbook relations with every number substituted, beside what this app's solver returns for the same inputs. Retype the arithmetic and you should land on the same answer. Fracture geometry is not here — that is graded against measured field data in the pad-scale verification suite.

Casing burst rating — Barlow

±0.05% allowed · Δ 0.000%

MATCHES

API 5C3 / Barlow thin-wall form

Equation: p_burst = 2 · Y · t / OD
Inputs: OD 5.500 in · wall 0.361 in · yield 80,000 psi
By hand: 2 × 80,000 × 0.361 / 5.500 = 57,760 / 5.5 = 10,502 psi
This app: 10,502 psi

Exact closed form — the only difference is display rounding.

Pipe body cross-section

±0.05% allowed · Δ 0.004%

MATCHES

Geometry

Equation: A = (pi/4) · (OD² − ID²), ID = OD − 2t
Inputs: OD 5.500 in · wall 0.361 in ⇒ ID 4.778 in
By hand: (pi/4) × (5.5² − 4.778²) = 0.7854 × (30.25 − 22.829) = 5.828 in²
This app: 5.8282 in²

Exact closed form.

Pipe body axial yield

±0.05% allowed · Δ 0.000%

MATCHES

API 5C3

Equation: F_yield = A · Y
Inputs: A 5.828 in² · yield 80,000 psi
By hand: 5.828 × 80,000 = 466,257 lbf ≈ 466 kips
This app: 466,257 lbf

Exact closed form.

Solution gas-oil ratio — Standing

±0.2% allowed · Δ -0.003%

MATCHES

Standing (1947)

Equation: Rs = γg · [ (p/18.2 + 1.4) · 10^(−a) ]^(1/0.83), a = 0.00091·T − 0.0125·API
Inputs: p 3,000 psi · T 200 °F · 35 °API · γg 0.75
By hand: a = 0.00091×200 − 0.0125×35 = 0.182 − 0.4375 = −0.2555; Rs = 0.75 × [(164.84 + 1.4) × 10^0.2555]^(1/0.83) = 722 scf/STB
This app: 721.9 scf/STB

Correlation, not a law: Standing's own scatter against his data set is roughly ±15 %. The check here only confirms the arithmetic, not the fit.

Oil formation volume factor — Standing

±0.2% allowed · Δ 0.003%

MATCHES

Standing (1947)

Equation: Bo = 0.9759 + 0.00012 · [ Rs·sqrt(γg/γo) + 1.25·T ]^1.2
Inputs: Rs 722 scf/STB · T 200 °F · 35 °API (γo 0.8498) · γg 0.75
By hand: term = 722 × sqrt(0.75/0.8498) + 1.25×200 = 678.3 + 250 = 928.3; Bo = 0.9759 + 0.00012 × 928.3^1.2 = 1.413 bbl/STB
This app: 1.4127 bbl/STB

Correlation; Standing quotes about ±5 % on Bo for the fluids he fitted.

Induced-seismicity magnitude bound — McGarr / Kanamori

±0.05% allowed · Δ -0.008%

MATCHES

McGarr (2014), Kanamori (1977)

Equation: M0_max = G · ΔV, Mw = (2/3)·(log₁₀ M0 − 9.1)
Inputs: ΔV 238,481 m³ (1.5 MMbbl) · G 2.0 × 10¹⁰ Pa
By hand: M0 = 2.0e10 × 238,481 = 4.77 × 10¹⁵ N·m; Mw = (2/3)×(15.6786 − 9.1) = 4.386
This app: 4.3857 Mw

Upper bound only — it assumes every injected barrel drives slip on one fault. Real events are smaller.

Methodology

Each capability carries a status: ✓ full, ⚠ partial (known limitations or beta), and — not yet available. Every ✓ links to the page that delivers it, so you can check it yourself.

Spot something wrong? Email a correction.