Árbók VFÍ/TFÍ - 01.06.1998, Blaðsíða 314
312 Ritrýndar greinar
Introduction
Aquaculture is growing fast as an industry, developing from snrall and in many cases not
very sophisticated companies to bigger enterprises, who use more and more advanced tech-
nology. Competition is also increasing and profit margins are becoming smaller. In many
fish market areas, there are significant seasonal fluctuations in market prices for different
sizes of farmed fish. The fish farming companies are trying to target the price peaks by
accurately scheduling the harvest of the various pens, which store fish of different size distri-
butions. Also, they rnust keep in mind to supply the market with certain continuity through-
out the year. This is a complex decision problem, which requires management science meth-
ods.
The present measurement techniques, from for example the Icelandic company Vaki
Ltd., and the availability of detailed records of most salmon farm operations today, provide
the necessary data for a mathematical model of the aquaculture production process, includ-
ing a model for forecasting growth. The goal is to present a forecasting and optimization
system as a software-based management tool for use in the day-to-day management of
aquaculture enterprises. The industrial/economic objective of this would be increased proit
margins in the production of farmed fish. It is suggested that by optimizing one of the acti-
vities of aquaculture production planning, i.e. the harvest scheduling, revenue can be
increased considerably. Another benefit is providing indications of what information is
important to profitability and thus guidelines to the farmers for the effort to collect data.
It appears that there is no tool on the market that allows aquaculture companies to nrake
full advantage of data from their monitoring equipment in their decision making process.
There seems to be a need for developing better management tools for decision rnaking in
fish farming, i.e. for a software based management tool for the whole production process,
from on-growing and into post-harvesting (processing), giving a more holistic view to the
planning procedure.
To the knowledge of the author, none of the current software systems available for
aquaculture companies will offer the same opportunities for optimization as the model pro-
posed here. The existing management tools are mainly focused on record-keeping and feed
control, see El-Gayar (1997). If these systems provide optinrization, they frequently focus
on only one area, such as feeding. Such management tools may for example predict final
harvest weights, but do so only on the basis of average weight, contrary to our proposal,
which takes full account of the distribution of sizes of fish in the individual cages.
This study proposes optimization models that take into account the most important fact-
ors of fish farming. These factors includes the market prices and demand for different sizes
of fish, smolt quality, feeding and other costs, harvesting practice and the size distribution
of lish in each pen. By analyzing and applying such optimization models throughout the
production cycle, the producer should be able to target market requirements from an early
stage in the cycle.
The optimization models presented here are based on a Markov model to describe the size-
structured growth of fish, and keep track of the development of the size distributions in the
individual pens through the growing period. The models also take the possibility of partial,
selective harvesting from the pens into account. This is an extension of similar Markov