Diabetes mellitus is a disease characterised by abnormally high glucose concentration, insulin dysfunction and resistance which may lead to health problems such as cardiovascular disease. This paper presents a mechanistic pancreas model of insulin dynamics which incorporates experimental physiological data. This model will provide an efficient and accurate way to determine the specifics of a metabolic problem in the pancreas. We use Intravenous Glucose Tolerance Test (IVGTT) data from the literature to identify the model parameters by implementing a deterministic optimisation method called DIRECT (Dividing RECTangles). Different data sets are used for optimisation and validation.