Finds the behaviour of the total weight of a (typical) cylindrical packet of biscuits, subject to lower and upper specifications ('specs') in packet weight and height, i.e., weight (W) in (L_W, U_W) and height (H) in (L_H, U_H).

The individual item weight is assumed to have: truncated Gaussian weight, w, with parameters μ_W, σ_W (of the original Gaussian), truncated at a, b; and height, h, such that h = w.d, with d (cm⁄g) Gaussian (μ_D, σ_D).

The filling procedure leads to a random (dependent) number (occupancy) of items in the packet.

A graph is produced for the simulated distribution ('pdf' and 'cdf') of the weight. The "fraction defective" [out of (L, U)] is important in industry.

The base data refer to a commercial packet of biscuits (figures) with nominal weight (mass) of 200 g (the legal "lower spec", L). The producer attempts to obey this limit by using a target weight of μ > L and avoiding too large giveaway. (For the base data, the out-of-spec fraction is #1.8 %.)

#Suggested other data: σ = 0.9 g (with smaller σ leading to better quality, but typically difficult or expensive to get, hence the need for technico-economic optimization) or σ = 2 g (possibly with weights both 0 g).