Home / Guides / Draft Survey Complete Guide

Draft Survey: The Complete Guide

How to determine cargo weight by measuring how deep a ship sits in the water. The complete procedure from Archimedes' principle to AI-powered measurement. For marine surveyors, cargo officers, and commodity traders.

New to ship draft? Start with our Ship Draft Guide.

1. What Is a Draft Survey?

It's the most widely used cargo measurement method in the bulk shipping industry because it doesn't require specialized shore-side infrastructure. A surveyor with a small boat, a sounding tape, and a calculator can determine the weight of 200,000 tonnes of iron ore. That simplicity is both its strength and its weakness, accuracy depends entirely on the skill of the person taking the readings.

2. A 2,200-Year-Old Idea

Two developments turned the principle into a commercial tool. First, the Plimsoll Line (mandated by Britain's Merchant Shipping Act of 1876 after Samuel Plimsoll's campaign against overloaded "coffin ships") created a visible, legally enforced connection between draft and safe loading. Second, 19th-century naval architects in Europe built the mathematical framework, displacement tables, hydrostatic curves, trim and list corrections, that let anyone with the right tables convert a depth reading into a weight.

By the mid-20th century, draft surveys were the standard method for international bulk cargo trade. A 2025 Springer literature review described them as "an accepted and convenient means of establishing the weight of bulk cargoes for over a century, forming the basis for bills of lading, port fees, and trade settlement." The irony: we're still using Archimedes' bathtub insight to settle multimillion-dollar iron ore trades.

3. Equipment You Need

4. The Complete Procedure (10 Steps)

1 Read the Six Draft Marks

Read forward port & starboard, midship port & starboard, aft port & starboard. Get as close to the waterline as possible, parallax from looking down from deck level is the #1 source of reading error. In swell, take 20 rapid readings, discard the 2 highest and 2 lowest, average the remaining 16. Read to the nearest centimetre.

→ Detailed guide: How to Read Draft Marks

2 Calculate Mean Drafts & Apparent Trim

Positive AT = trim by stern. Negative AT = trim by head.

3 Correct Draft Marks to Perpendiculars

Draft marks are rarely at the perpendiculars. The offsets (Fd, Ad, Md) are documented in the vessel's stability booklet. If you skip this correction, assume at least 0.5–2 cm of error per metre of trim.

4 Apply Corrections & Calculate True Trim

5 Calculate Quarter Mean Draft

This is the key formula. The midship reading gets 6× weight to suppress the effect of hull bending (hog or sag):

For a vessel with significant hull deflection, a sagging ship shows midship draft higher than the fore/aft average; a hogging ship shows it lower. The 6× weighting corrects for this. Without it, a sagging ship would overstate cargo weight because the midship would pull the average draft deeper than the ship's actual mean immersion.

6 Enter Hydrostatic Tables

Using the Quarter Mean Draft (QM), linearly interpolate from the vessel's approved hydrostatic tables to obtain:

• Displacement (Δ₁), the vessel's total weight at this draft in saltwater
• TPC, Tonnes per Centimetre immersion (how much weight changes draft by 1 cm)
• LCF, Longitudinal Centre of Flotation (metres from midships, sign convention matters)
• MCTC, Moment to Change Trim 1 cm (tonne-metres)

7 First Trim Correction (LCF Correction)

Corrects for the fact that the ship trims about the LCF, not about midships:

Sign: if LCF is aft of midships and trim is by stern, the correction is negative. If you get the sign wrong, the error compounds through every subsequent step.

8 Second Trim Correction (Nemoto)

Corrects for the change in MCTC at different drafts. Always positive:

Displacement corrected for trim: Δ₂ = Δ₁ + FTc + STc

9 Density Correction

Hydrostatic tables assume saltwater at 1.025 t/m³. If the actual water density differs, the displacement must be corrected:

Take water samples at mid-depth, amidships. If surface and bottom samples differ by more than 0.001, sample at multiple depths and average. Use a certified glass hydrometer, electronic densitometers are convenient but calibrate them against a glass standard before each survey.

10 Subtract Deductibles → Cargo Weight

From Δ₃, subtract everything that isn't cargo:

• Ballast water (sounded and corrected for trim/list; this is the largest variable)
• Fuel oil, diesel oil, lube oil
• Fresh water (domestic + boiler)
• Bilge, sludge, other liquids
• Constant (unaccounted weight, paint, rust, stores, crew effects)
• Lightship weight (from the vessel's stability booklet)

→ Full walkthrough: How to Conduct a Draft Survey

5. The Correction Formulas (Quick Reference)

6. Seven Common Mistakes That Cost Money

2. Assuming the draft marks are at the perpendiculars. Many ships have marks offset by 2-5 metres. On a vessel with 3 metres of trim and a 2-metre forward offset, the uncorrected forward draft reading is off by about 3 cm. That's ~180 tonnes on a Panamax. Check the stability booklet, the offsets are documented there.

3. Not accounting for list. If the vessel has even a 1-degree list, port and starboard drafts at each position differ by several centimetres. Averaging them eliminates the list effect, but only if you actually take both readings. Taking one side and doubling is a shortcut that produces garbage results.

4. Ballast water: trusting the gauges. Tank gauges fail, stick, or read incorrectly when the vessel is trimmed. Always sound tanks manually with a calibrated tape and water-finding paste. Unpumpable ballast, the ~50-200 tonnes that suction can't remove, must be estimated from the sounding tables, not guessed.

5. Water density: sampling from the surface only. In estuaries and rivers, freshwater can layer on top of saltwater. A surface sample gives a density of 1.005 when the actual mean is 1.018. For a vessel displacing 80,000 tonnes, that's a 1,000-tonne error. Sample at mid-depth, amidships. If uncertain, sample bow and stern, surface and bottom, average all readings.

6. Forgetting that ballast and fuel change during the survey. If you're deballasting or bunkering during the draft survey, your initial and final conditions aren't comparable. All tank operations must be suspended before the initial survey begins and not resume until the final survey is complete.

7. Treating the constant as constant. The ship's constant (unaccounted weight from sludge, stores, crew effects, paint accumulation) drifts over time. A well-maintained vessel might have a stable constant of 150-300 tonnes. A vessel that hasn't had its constant verified in years could be off by 500 tonnes or more. Compare constants across multiple surveys to spot anomalies, a sudden jump often indicates an unrecorded repair, modification, or accumulation of unpumpable ballast.

→ Glossary: Draft Survey | → Comparison: AI vs Traditional Draft Survey Accuracy

7. AI vs. Manual Draft Survey

• Parallax error: reading from deck height introduces 1-3 cm of systematic error per mark
• Subjectivity: two surveyors reading the same mark often disagree by 1-2 cm
• Safety: small-boat operations near large vessels in ports are high-risk
• Wave interference: in swell over 0.5 m, manual readings degrade rapidly
• No audit trail: there's no objective record of what the surveyor actually saw, only the numbers they wrote down

AI-powered draft survey systems address all five. A 4K camera at water level (or on a pan-tilt mount ashore) captures the draft marks. Semantic segmentation neural networks separate the waterline from the scale digits. Anti-wave algorithms (curve fitting across multiple video frames) stabilize the reading. The system outputs a time-stamped, traceable digital record. The surveyor becomes a verifier, not a reader, reviewing AI-generated readings rather than squinting at rusted hull markings from a bouncing inflatable boat.

A 2025 paper in the journal Sensors documented a YOLOv8-based computer vision system deployed at the Port of Santos, Brazil. Across 34,000+ images over 6.5 days, it achieved mAP50-95 mask scores of 0.980 for bow hull segmentation and 0.965 for stern. The system operates continuously, 24/7 monitoring of hull height, independent of human attention span or fatigue.

→ Full comparison: AI vs Traditional Draft Survey | → GOTEC AI algorithms technical details

8. Regulations & Standards

9. Frequently Asked Questions

What accuracy can I expect from a properly conducted draft survey?

How long does a draft survey take?

Traditional: 2-4 hours (30-45 min reading marks by boat, 30-60 min sounding tanks, 60-90 min calculation). AI-powered: 20-30 minutes (camera deployment + automated reading, simultaneous tank sounding, automated calculation).

When should I use a draft survey vs. belt weighing?

Draft survey: when loading/discharging at anchor, at ports without calibrated belt weighers, or when belt weigher readings are disputed. A draft survey is also the standard fallback when two belt weighers disagree. Belt weighing: when available and legally certified (typically ±0.25% accuracy vs. draft survey's ±0.5%). The two methods serve as cross-checks, if they disagree by more than 1%, investigate before signing off.

Does ballast water really matter that much?

Yes. Ballast water is typically the largest variable in the deductible calculation. On a Capesize bulk carrier, ballast capacity runs 50,000-80,000 tonnes. A 1% error in ballast measurement, misreading one tank by 10 cm on the sounding tape, is a 500-800 tonne error in the cargo figure. At $100/tonne for iron ore, that's a $50,000-80,000 mistake. This is the single most important thing to get right after the draft readings themselves.

10. Further Reading & Related Content