The Deckhand Chronicles

One result, then the work behind it

The sloshing CFD study reproduced the tank's first-mode frequency within about 0.30 % of linear theory across the tested fill levels. Read the study. It also includes a SPHERIC Test 10 impact-pressure check; frequency agreement alone does not establish accurate impact pressures. The result applies to the published two-dimensional tank cases. The analysis is reviewed against available public literature. A different tank, fill level or forcing will need its own inputs and checks. The report keeps the convergence evidence beside the result so the reader can inspect the basis.

If there are engineering needs, I am here to help.

In the coming issues, we will go into greater detail on engineering workflows and their automation, along with the use of AI agents.

Deckhand is the name of this newsletter and the AI assistant setup behind it.

What began as a feasibility study is now a working demo on a synthetic sea state, built toward a real-time monitoring use case.

Demo: short-horizon motion forecast with a go / caution / no-go call

Ask big, then check the work

Give your AI assistant an outcome, a bounded scope and a clear definition of done: define the must-have result. Start with one case. Check the inputs, units, first calculation and saved output before expanding the work. For a longer task, agree on a checkpoint and look for completed cases, recorded exceptions and results read back from the actual output. Check results against a hand calculation or a published benchmark, which gives a stronger quality assessment. Keep an engineer responsible for the assumptions and the decision. Ask for missing evidence when a claim cannot be checked. Fixed units, clear criteria, repeatable checks and an independent review keep the engineering precise. The aim is useful work that can be checked and repeated.

Engineering areas

• Hydrodynamics and motions — tank-sloshing frequency and convergence example.
• Structural and fatigue — unstiffened plate-buckling worked example (screening estimates).

More areas in coming issues.

The examples show their inputs, methods and limits; they provide a starting point for defining a study and help make operational decisions.

Make repeat work easier

Run new calculations from one input file format, faster. Measure the gain in elapsed time, rework and review findings. The repeatability and workflow automation benefits are most evident when there is a reanalysis or a parametric study.

Help with the next operation

AceEngineer provides marine engineering analysis and calculation workflows for an operator or contractor; reply with the operation, available inputs and required decision to agree on the study scope.

The next two issues

October 27 — Cathodic protection: anode design.

November 10 — Asset integrity.

The date book

ADIPEC: November 2–5, 2026, Abu Dhabi.

OTC 2027: Monday–Wednesday, May 3–5, 2027, Houston.

Dates checked October 20, 2026; confirm with the organizer before travel.