Agricultural-economicsSheep stocking rate influences farm profit significantly. However, determining the optimal stocking rate can be difficult, particularly when climatic conditions are uncertain. Optimising stocking rate requires an understanding of the quantity and quality of feed available throughout the year, the optimal live weight profile throughout the year, the impact of seasonal variation, the impact of labour availability, the cost of alternative feeds, prices of livestock and livestock products, the risk preferences of the farmer and the interactions between the livestock and cropping enterprises in the farm system. To evaluate the stocking rate that maximises whole-farm profit thus requires detailed whole farm optimisation modelling. In this project we are developing and testing an improved whole farm optimisation model. We will then apply the model to determine the optimal stocking rate for farm businesses and identify key factors which may impact choice of stocking rate. The final results will likely be used by farmers to aid their business management.
Principal investigatorMichael Young firstname.lastname@example.org
Area of scienceAgricultural And Veterinary Sciences
Farm systems are very complex so representing this detail quickly leads to a large slow model. Initially our model (AFO) could easily be executed on our personal computers however, as we have developed and improved AFO over the last year, it has grown significantly in size. Now, depending on the level of detail included in our model it can take up to 1000 seconds to solve and requires up to 64GB of RAM. This would be fine if the model only needed to be executed a hand full of times however, once we start applying AFO and completing our analysis’s we will need to execute the model hundreds of times. This led us into a bit of a computational challenge, and we haven’t even finished devolving the model yet.
Simply put, the solution to our problem was/is to get a bigger computer. Cutting down the model is an option but that requires sacrificing detail which could potentially lead to inaccurate findings. After some simple research, we concluded that cloud computing could be useful. Using cloud computing we are able to access more computing power than we have available on our personal computers and we can run the model whilst still working on our computers.
Access to the nimbus cloud platform over the past six months has proved very helpful. The capacity to run AFO externally has maximised our productivity. More importantly, the extra computing power of the cloud has allowed us to continue developing AFO and improving its representation of the farming system. This is of high importance because it gives us more confidence in our advice to farmers. Overall, access to Nimbus cloud has been very beneficial for our project and will continue to be a key resource over the next few years