Weight Grading Fish: How Graders Sort by Weight Class

What a weight grader does

A weight grader weighs every piece of fish individually on a moving belt and sends it to the outlet assigned to its weight class, so that each box, tray or tote downstream contains fish of a uniform size. Grading by weight replaces visual size sorting with a measured value, which makes size categories repeatable from one haul, shift and crew to the next.

In practice a grader does two related jobs. Grading assigns each piece to a weight class (for example 0.3–1 kg, 1–2 kg, 2–4 kg). Batching uses the same weight data to build packs that reach a target weight with as little overweight as possible. Batching is explained in detail in Reducing Giveaway: Batching Algorithms Explained; this article concentrates on the grading machine itself.

In weighing terminology, an automatic instrument that weighs discrete loads is a catchweigher. The Measuring Instruments Directive defines an automatic checkweigher as a catchweigher that subdivides articles into two or more subgroups according to the difference between their mass and a nominal set-point (Directive 2014/32/EU, Annex VIII). Whether a particular grading application falls under legal metrology depends on how the result is used commercially; see the legal metrology hub.

Infeed, weigh and sort belts: how the machine is built

A belt grader has three functional sections: an infeed belt that singulates and spaces the fish, a short weigh belt mounted on a load cell, and a sorting belt with outlets that divert each piece to its class.

Infeed belt

The infeed belt turns a pile of fish into a single file with gaps. The weigh belt can only produce a clean result if exactly one piece is on it during the measuring window, so spacing on the infeed determines the maximum usable speed of the whole machine. Operators or a stepped belt arrangement (a faster belt after a slower one) create the gap.

Weigh belt

The weigh belt is a short conveyor supported by one or more load cells. The indicator subtracts the dynamic tare of the belt, filters the signal while the piece is fully on the belt, and assigns a weight. Because the belt runs continuously, the indicator has only a fraction of a second per piece, which is why the geometry discussed below matters.

Sorting belt and outlets

The sorting belt carries the weighed piece past a row of outlets. The controller tracks each piece by belt position (encoder or timing) and fires the correct gate, flap or pusher when the piece arrives. Pneumatic actuators are common; the outlet feeds a box, tray, chute or a batching station.

Throughput versus accuracy

Throughput and accuracy pull in opposite directions: the faster the belt, the shorter the time a piece spends fully on the weigh belt, and the less time the indicator has to average out vibration. The limits follow from simple geometry.

A piece is completely on the weigh belt only for the time the belt needs to travel the difference between the weigh belt length and the fish length. A 500 mm weigh belt carrying a 250 mm fish at 1 m/s therefore gives a measuring window of a quarter of a second. To keep a second piece off the belt during that window, the distance from the front of one piece to the front of the next must be at least the length of the weigh belt. The table shows what this means for a hypothetical 500 mm weigh belt; the figures are worked examples, not the specification of any machine.

Three practical conclusions follow:

• Long fish reduce capacity. A 350 mm fish at 150 pieces per minute leaves roughly a tenth of a second for the measurement. That is why graders quote throughput as dependent on product size.
• Spacing matters as much as speed. Two pieces touching on the weigh belt produce one wrong result and usually a rejected piece.
• Class width should match achievable accuracy. If the dynamic result scatters by ±10 g, pieces within 10 g of a class boundary will sometimes land in the neighbouring class. Narrow classes on a fast line produce more misgrades than wide classes on a slow one.

Motion on a working deck

At sea the vessel adds a second disturbance. A load cell measures force, and vertical acceleration of the deck changes that force: a vertical acceleration of 0.3 m/s² changes the reading by about 3 %. Heave and pitch periods are several seconds long, far longer than a measuring window of 0.1–0.5 s, so simple averaging on the belt cannot remove the effect. How static marine scales deal with this is explained in How Motion Compensation Works and on the weighing at sea hub. For any belt grader intended for use on board, ask the supplier explicitly how vessel motion is handled and what accuracy is achieved under sea conditions.

Weight classes: how they are defined

Weight classes are defined either by regulation, by buyer specification or by the vessel's own commercial strategy, and a grader stores them as recipes (presets) per species and product form.

In the EU, Council Regulation (EC) No 2406/96 lays down common marketing standards and states that products are sized by weight or by number per kilogram (Article 7). Its Annex II sets size categories per species. A selection:

Source: Regulation (EC) No 2406/96, Annex II. Always check the consolidated version and any buyer specification before programming classes.

Commercial classes are frequently narrower than the regulatory categories. A buyer may ask for 2–3 kg and 3–4 kg cod instead of one 2–4 kg category, or for fillet portions in 20 g steps. Each extra class needs its own outlet and its own packing station, so the number of classes is limited by deck space as much as by the grader.

Setting class boundaries

1. Start from the size categories or buyer specification that the product is sold against.
2. Check the actual size distribution of recent catches; a class that receives 2 % of the fish rarely justifies an outlet.
3. Decide how boundary pieces are handled: lower-inclusive (1.000 kg goes to 1–2 kg) is a common convention, but whichever rule is used should be stated in the recipe and agreed with the buyer.
4. Compare class width with the dynamic accuracy of the grader at the intended speed.
5. Store the result as a named recipe so the same classes are used on every trip.

Fresh versus frozen fish on a grader

Fresh and frozen products behave differently on belts and on a load cell, and a grader recipe should be set up separately for each.

The glaze rule comes from Regulation (EU) No 1169/2011, Annex IX point 5; the freezer vessel temperature requirement from Regulation (EC) No 853/2004, Annex III, Section VIII. In both cases, automatic zero tracking between pieces is essential: the grader should re-zero the weigh belt whenever it is empty, so that slime, water or ice does not accumulate into the reading. Material and wash-down requirements are covered in the hygiene and materials hub.

What to check when choosing a grader

Choosing a grader means matching its weighing range, speed, construction and data handling to the product, the classes and the place where it will run, and then verifying the result on your own fish.

Before buying, ask for a trial with your own product at the intended speed and check-weigh a sample of graded pieces on a static scale, so that misgrades per class are counted rather than estimated. Wash-down ratings are explained in IP ratings for marine scales, the metrology rules for automatic instruments in OIML R51 vs R76, and the data fields worth exporting in Weighing Data Logging at Sea.

Pieces that are too large or heavy for a belt grader, or single boxes that need a check weight, are usually weighed on a static bench scale such as the M2 Series Marine Scale. Grading results become more valuable once they are recorded per lot; see Catch Registration and Traceability from Haul to Harbour and the onboard processing hub.