Transient thermal, casing and tubing design, and trajectory with anti-collision: three validated engines behind one API and MCP layer, callable by engineers, operator systems and AI agents.
Every client reaches the same engines through the same gateway, and every run leaves the same deterministic record. The web UI, an operator's data platform and an AI agent get identical numbers.
Production, injection, drilling and shut-in cases. Annular fluid expansion and pressure build-up on the same model.
Thermal engine → C · ENGINE 02Load cases, triaxial checks and API TR 5C3 / ISO 10400 ratings, with thermal-coupled loads.
Tubulars engine → A · ENGINE 03Planning profiles, minimum curvature, ISCWSA error models and separation factor against every offset.
Trajectory engine →No desktop installs and no file exports by email. A design is a shared, versioned object the whole team works on.
Every calculation is an endpoint. Data platforms and digital twins connect through documented APIs, not custom one-off integrations.
Deterministic, auditable physics an agent can discover and invoke directly over MCP, with the same answer on every run.
Take one engine or all three. Seats for engineers, metered runs for machines, and nothing held back behind a perpetual module licence.
Transient heat transfer through tubing, annuli, cement and formation for every operation in the well's life, with trapped-annulus pressure solved on the same model.
Production, injection, circulation and shut-in, chained as nested operations with restart from any point.
Trapped-annulus pressure build-up with bleed-off, compressible spacers and vacuum-insulated tubing evaluated against each string's limits.
Growth and load at the wellhead from multi-string thermal and pressure history.
Temperature and pressure history for every string and annulus.
Trapped-fluid pressure build-up from the same temperature field.
Thermal and APB terms added to every load case, per string.
VME and API ratings against each design factor, with the governing depth.
Drilling, production and contingency loads evaluated for every string, with thermal-coupled loads taken directly from the thermal engine.
Piston, ballooning, buckling and temperature terms at every packer, with tubing-to-casing load transfer carried into the casing design.
Survey-driven dogleg, side force and buckling limits taken from the trajectory engine, so bending follows the real wellpath.
Load points checked against the connection's rated envelope as well as the pipe body.
Rated performance against factored loads: burst, collapse, axial and triaxial checks with connection ratings applied per string.
Ductile and tension burst limit states with measured wall and yield distributions, including the brittle fracture limit for hard-line grades.
Probabilistic strength and load distributions returning a failure probability per limit state, for wells where a single factor is too blunt.
Planning profiles, minimum curvature surveys and ISCWSA error models, with separation factor computed against every offset on the pad at every station.
Most integrations stop at moving data. NtLkT can also be told to compute, then hand the result to the next system, so it takes part in the workflow.
Pull surveys, pipe catalogs, pore pressure and temperature data from the systems that already hold them.
Send designs, envelopes and safety factors to data platforms, digital twins and reporting.
Compute on command: design this string, generate a hundred trajectories from A to B, rerun the pad after a new survey. Then trigger the next machine.
The engines exposed as tools an agent discovers and calls at runtime. The agent reads the tool definitions and chooses which to use, with no glue code per model.
Typed functions over the REST API, handling auth, retries and data models. For teams building their own agent and keeping full control in code.
Drive every engine from a shell. Shell-based agents and scripts run it the same way an engineer would, step by step.
Reads the request, plans the steps, generates candidate designs and decides which engines to call.
Deterministic, physics-based checks against the design factors. A verdict for every candidate, never an opinion.
Only designs that pass leave the loop, with a run record that is auditable, repeatable and permissioned like an engineer's work.
The AI never invents the engineering. It runs validated engines, faster.Every agent action is permissioned, validated and logged. Engineers explore hundreds of design options in parallel instead of one at a time.
| POST /v1/thermal/transient | Chained operations, per-annulus temperature and pressure |
| POST /v1/apb | Annular pressure build-up with mitigation options |
| POST /v1/strings | Define a string: sizes, grades, connections, depths |
| POST /v1/analyses | Burst, collapse, axial and triaxial checks |
| GET /v1/analyses/{id}/envelope | Design-limit envelope and loading points |
| POST /v1/trajectories | Plan a profile to one or many targets |
| POST /v1/anticollision/scan | Separation factor against every offset |
| run_thermal_case | Transient profile for an operation sequence |
| compute_apb | Trapped-annulus pressure and required bleed |
| evaluate_triaxial | Safety factor and governing depth for a load case |
| check_connection | Load point against the connection envelope |
| plan_trajectory | Profile, dogleg and survey stations |
| scan_anticollision | Minimum separation factor and closest approach |
| explain_limit_state | Cited derivation of the limit that governed |
Post a legacy load case to the API and get a value-by-value comparison against your own spreadsheet. Every run is reproducible from its inputs, so a design review can be replayed years later and land on the same numbers.
| Check | Legacy sheet | NtLkT | Δ |
|---|---|---|---|
| Burst (SF) | 1.71 | 1.71 | 0.0% |
| Collapse (SF) | 1.33 | 1.33 | 0.0% |
| Tension (SF) | 2.04 | 2.04 | 0.0% |
| Triaxial VME (SF) | 1.36 | 1.38 | 1.5% |
| Wellhead growth (in) | 2.11 | 2.12 | 0.5% |
| Min separation factor | 1.62 | 1.61 | 0.6% |
Named seats with module bundles for thermal, tubulars and trajectory / anti-collision. Annual, and add seats as the team grows.
Metered per run. A developer tier for integration and pilots, and volume tiers for service companies and agent platforms.
Unlimited seats within scope, dedicated hosting and SLA, with validation support and training included.
Bring a set of live wells, run them through all three engines, and compare the results with your current tools.