Work Sampling in Shipyards and the Naval Sector
Plan an activity study in shipyards with attention to phases, safety, coverage and limits of comparison.
Updated:
Shipyards combine trades, equipment, access constraints and phases that change during a project. Sampling can estimate time allocation for the defined resource and period; it does not require every worker to repeat the same task or the whole yard to be stationary. Comparing periods or extrapolating to a project requires explicit treatment of mix and coverage changes.
1. Define the population
Specify vessel/project, phase, zones, shifts, trades and observation unit. Welding, fitting, piping, electrical work, painting and insulation have different constraints. Position, access and methods can prevent direct task comparisons; they do not mean usable historical information never exists.
Size the study using actual phases, trades and included working periods. Record indoor/outdoor conditions, permissions and weather to interpret changes in mix. Scheduling must respect safe access rather than treating hazardous conditions as observation opportunities.
2. Stratification and sample size
Stratify by trade, block, shift or phase according to the question. A separate precision target is not mandatory for every possible combination. Distinguish a weighted overall estimate from a cell-specific estimate.
$$n_0=\left\lceil\frac{z^2p_0(1-p_0)}{e^2}\right\rceil$$
With p₀=0.5 and absolute e=0.05, z=1.96 gives 385 independent visits; z≈2 gives 400. With factor 4 and p₀=0.3, the result is 336. These are not interchangeable with an allocation table as though each cell had equal precision.
This simulated table retains a 3,750-record allocation proposal, not a statistical derivation:
| Trade | Small block | Medium | Large | Total |
|---|---|---|---|---|
| Welding | 220 | 350 | 480 | 1,050 |
| Fitting | 180 | 290 | 380 | 850 |
| Piping | 140 | 220 | 310 | 670 |
| Painting/blasting | 100 | 160 | 240 | 500 |
| Electrical | 80 | 120 | 180 | 380 |
| Other | 60 | 90 | 150 | 300 |
| Total | 780 | 1,230 | 1,740 | 3,750 |
There are 18 cells. Requiring 400 records in each would require 7,200, not 3,000–4,000. Stratum weights and within-round correlation determine analysis. A pilot helps clarify categories and variability, but there is no universal 200–300-record minimum per trade.
See NIST planning and Penn State stratified sampling.
3. Schedule and safe observation
Distribute times across relevant periods and record actual timestamps. Do not select visits only when a block is accessible or producing. Record inability to observe with a reason. A visit covering twenty people is not automatically twenty independent trials.
Four or six weeks is an option, not guaranteed representation of future phases. “Covering lunar phases” has no statistical rationale without an explicit, verifiable operating hypothesis. Safety and access procedures take priority over the sampling schedule.
4. A proposed taxonomy, not a naval standard
| Group | Examples | Interpretation limit |
|---|---|---|
| Direct transformation | Welding, fitting, permanent installation | Define task, unit and quality requirement |
| Necessary support | Drawings, setup, scaffolds, movement, NDT, tests | Do not remove controls or protection merely because they do not transform product |
| Waiting to investigate | Trade interference, material, access, weather | Observing a wait does not establish avoidability |
| Rework | Correcting an operation | Not the fraction of defective units |
| Break/personal | A defined working-time state | Respect agreements, privacy and denominators, rather than assigning blame |
Define the classification with people who understand the process and test its boundaries with observable situations. Do not use another yard’s value-added share as a target without checking definitions, phases, load and comparison conditions.
5. Reproducible simulated case
This is not a customer implementation. Assume an 85-ton block with 14 welders, four fitters and two pipefitters. A plan schedules 2,730 records across 21 working days (130/day) over five weeks. The report must identify unit and trade records; the context names three trades, not four.
One illustrative cell contains 600 hypothetical independent records:
| State | Records | % |
|---|---|---|
| Welding | 288 | 48% |
| Preparation | 96 | 16% |
| Movement | 48 | 8% |
| Trade-interference wait | 54 | 9% |
| Weather wait | 24 | 4% |
| Missing material | 30 | 5% |
| Rework | 36 | 6% |
| Defined break | 24 | 4% |
| Total | 600 | 100% |
For 48%, an approximate Wald95 margin is 1.96√(0.48×0.52/600)≈4.00 percentage points, before dependence adjustments. Do not attach these intervals to other cells without their counts and n.
The last five categories total 28%; they are not entirely removable waste. Over 1,200 hypothetical person-hours matching that cell, they represent 336 associated hours. At assumed, not market-benchmark, costs of €38–45/hour, their capacity valuation is €12,768–15,120—not demonstrated cash savings. Do not extrapolate the cell to 40–60 blocks or a whole vessel without a supporting design.
6. Actions and tools
Investigate trade sequence, preparation, materials and rework causes with additional evidence. Weather safety and NDT controls are not negotiable based on a time share. Before/after comparisons need mix, phase and condition checks.
WorkSamp supports random planning, capture—including local offline storage—and documented reports. Before collection, check how phases and trades will be represented, which export the analysis needs, and how pending synchronization and permissions will be handled.
Cronometras supports direct time studies. Induly can provide working-time records but is not the exclusive way to obtain a valid hours denominator. ASETEMYT is a directory where professional experience must be checked; ProdCont is an industrial-organization contact option. No ISO certification, per-vessel savings, fixed learning period or mandatory quarterly schedule is promised.
7. Start with a pilot
Choose a bounded question, test definitions and coverage, and size follow-up using actual precision and resources. More blocks and phases require new weights and assumptions, not merely repeating a generic schedule. Open WorkSamp or request information to assess suitability.