Why batch solvers fail under real operational decision-making
Our main challenge with other solutions was not accuracy at convergence, but usefulness during decision-making. Once execution began, meaningful changes required restarts or rebuilds. This was the first system we used where physics continued resolving while decisions were being tested, which fundamentally changed how we explored scenarios.
Executive Introduction
For decades, high-fidelity simulation has been constrained by a batch-solve paradigm: configure a model, run the solver, wait for convergence, then analyze results. This approach shaped CFD workflows, training simulators, and digital twins alike. While adequate for offline analysis, it imposes a fundamental limitation in operational and training environments where engineers, instructors, and operators must interact with a system as it evolves.
This case demonstrates a materially different outcome: a simulation platform built on runtime physics, where the system continues solving continuously while conditions, decisions, and configurations are modified live. Rather than optimizing batch solvers, Endeavor eliminated the batch paradigm entirely — redefining what simulation means for operations and training.
Organizational Context
This case involved technically mature organizationsresponsible for high-consequence drilling, well control, and managed pressure drilling operations. These teams routinely used CFD tools, conventionalsimulators, and digital twins, but accepted a persistent friction: simulationwas something you ran, not something you worked inside.
Operationally, this created several constraints:
As a result, simulation supported explanation and review,but rarely decision-making or judgment development under realistic conditions.
How the System Was Used
Endeavor’s platform was deployed as a continuously executing simulation environment rather than a job-based solver. Engineers, instructors, and operators interacted with the system while physics continued resolving in real time.
Parameters such as pressure, flow, boundary conditions, equipment state, and operational decisions were modified live — without stopping, restarting, or rebuilding the simulation environment. Flowpaths remained dynamic. Consequences propagated immediately.
The system never entered a “solve → observe → reconfigure” loop. It remained physically active at all times.
This allowed users to explore scenarios interactively, observe second- and third-order effects as they emerged, and test edge cases that would be impractical or impossible in batch-based tools.
Characterization of the Structural Change
Traditional simulation architectures — including FEA and CFD — are batch-oriented by design. Once a solver begins, inputs are effectively locked. Any change requires stopping the run, modifying parameters, and restarting computation. This architecture forces users to think sequentially and conservatively, often working around the software rather than with it.
Endeavor rejected this model entirely.
By maintaining a continuously solved world state, the platform allowed physics to remain active while the system was manipulated. This enabled behaviors that batch solvers structurally cannot represent:
This was not a performance optimization. It was an architectural inversion: simulation became a live environment rather than a queued computation.
“We had tools that solved physics—just not when decisions mattered.”
Value Captured & Realized
Knowledge and Insight
Engineering and training teams developed intuition through interaction rather than post-hoc analysis. Behaviors that were previously hidden behind solver latency or simplified models became observable as they formed. This fundamentally changed how users understood system dynamics, especially under abnormal or degraded conditions.
Operational Impact
Iteration cycles collapsed. Scenarios that previously required hours or days to evaluate could be explored in minutes. Engineers tested more cases because the cost of exploration dropped to near zero. Training environments shifted from scripted success/failure outcomes to judgment under consequence.
Across deployments, iteration efficiency improved by an estimated 70–90% compared to batch-based workflows.
Cost and Risk Implication
Reducing solver wait time and restart overhead recovered hundreds of engineering hours per project — conservatively $50,000 to $200,000 USD per study. More importantly, improved understanding of dynamic behavior reduced reliance on approximation and late-stage correction, lowering exposure to high-cost operational surprises.
Established Outcome
High-fidelity simulation does not require waiting. It requires architecture that allows physics to remain alive while humans interact with the system. Endeavor established runtime physics as the new baseline for operational and training simulation — not an enhancement, but a replacement for batch-solve paradigms.
Closing Perspective
When simulation forces users to stop thinking while it computes, insight is lost. By removing that pause entirely, Endeavor transformed simulation from a task into an environment. This case establishes that in high-consequence operations, realism is not about resolution or visuals — it is about whether physics is allowed to continue when decisions are made.
Back to Case Studies