Nuthall, Peter L.2011-02-072011-02-071977https://hdl.handle.net/10182/3208Farmers make decisions in a planning environment in which planning information is frequently imperfect and outcomes uncertain. Efficient planning systems must therefore be dynamic in nature. However, available systems which are capable of being directly applied to individual farms on a whole farm basis do not reflect the true nature of the planning environment. This study explores the nature of these problems through developing a dynamic planning system for pig fattening. The study considers the nature of the pig fattening problem and specifies the important factors that planning models must allow for representing the realistic planning environment. The features of currently available models are reviewed leading to a statement of the improvements required. A realistic model of pig fattening is developed and possible solution algorithms are considered. As the models require detailed pig growth information a simulation model designed to predict response from alternative feeding patterns was developed. The output from this model together with the output from a model which calculates growth distributions and posterior probabilities on potential growth provide some of the input data for a stochastic multi-period linear programming model of the problem. The remainder, feed cost information, is obtained from least cost models formulated to represent the features of the dynamic planning problem. With imperfect information, planning involves continual re-planning as new observations are made. Consequently only the solution to the first period of the model is likely to be implemented. To ensure that the first period decision set is optimal it is shown that a minimum number of periods must be included in the planning model. The minimum number is referred to as the planning horizon. Methods of determining a planning horizon are reviewed and it is concluded that available methods do not provide a general method for all planning situations. A method is developed for the pig fattening problem. Finally, a number of planning experiments were carried out to demonstrate the value of the planning models and systems developed. The results indicate the potential value of applying sophisticated planning methods to individual farms and provide a means to examine the condition under which detailed individual farm dynamic planning can be worthwhile.enhttps://researcharchive.lincoln.ac.nz/pages/rightspig industrypig fatteningdynamic planningplanning systemssimulation modelfarm managementplanning horizonThesisANZSRC::070106 Farm Management, Rural Management and AgribusinessANZSRC::070105 Agricultural Systems Analysis and ModellingANZSRC::010406 Stochastic Analysis and ModellingQ112839381