A noticeable change in the structure of the Spanish pork sector has been observed in recent years, Pig farms have become more and more specialized and the size of their operations has been increasing. Moreover modern pig farms have tended to integrate and coordinate their operations into Pork Supply Chain (PSC). The overall aim of this thesis was to formulate a set of models to support sow herd management and piglet production in a pork supply chain context, giving practical answers to relevant questions often asked by decision makers. Hence, the main strategic and tactical decisions regarding sow herds and piglet production management in a pork supply chain context were considered. Basically four models under uncertainty were developed. The first one was a linear programming formulation of a semi-Markov model to design pig facilities. It showed that herd distribution based on physiological states and movements between facilities were useful to calculate the room needs for each sow facility. The formulation considers recent EU regulations regarding animal welfare and the impact on economic cost of housing facilities. Continuing with animal welfare issues, a framework for the integration of clinical signs into a sow replacement model was developed. The sow replacement model used in the framework was a multi-level hierarchical Markov process using Bayesian updating. The results showed how the incorporation of clinical signs in sow replacement models led to better culling policies through more efficient detection of the weakest sows in the herd. The next two models formulated under finite time horizon showed the herd distribution or structure moving to the steady state. Temporary shocks in parameters or transitory perturbations were better represented by finite time horizon models where scenarios collected part of the uncertainty of the system. The use of Linear Programming led to incorporate herd constraints more easily than in a Markov Decision Process. Finally the two-stage stochastic programming model with recourse showed to be a suitable tool to deal with the uncertainty of the system through scenarios. Additional benefits for practical purpose were the scheduling of purchasing of gilts, planning piglet production and replacement policy, all under a rolling time horizon scheme.
It is unpractical to build one single model for answering all questions of the PSC, however it is more reasonable to propose different models capable to solve more accurately partial problems for improving the management of the PSC and integrate all together in some information system. Hence the models presented in this thesis are suitable tools to deal with main strategic and tactical decisions in sow herds producing piglets in a general PSC context.