Within an organisation calibration kits are used every day to ensure that the machines are ready for production whether that’s after a preventative maintenance or a qualification activity post a machine down event. The Calibration kit process allows engineers to check critical process attributes that effect production. In many organisations calibration kits can outnumber the quantity of production units in process by up to 3 times. This is largely due to factory management only being concerned with calibration kit management when a production line is stopped due to waiting for a calibration kit. In recent years there has been significant work completed on the Calibration kit process from a demand and supply side however the key components of the Calibration kit process and its inherent variability make the management of the Calibration kit process extremely difficult. Breaking down the Calibration kit process to its most basic of functions show that it can be defined as a reusable article within a closed loop supply chain. The management issues that affect RA’s (Reusable articles) within a closed loop supply chain such as Calibration kits include fleet size definition, control and improvement of return rate and control and improvement of cycle time. To date ‘pool managers’ have struggled with this aspect of RA management given the variability that exists in the system when it comes to cycle time, quality and fleet shrinkage. To date the methodologies for determining fleet size within an RA process have ranged from ‘rules of thumb’ to the development of optimised simulation models. However the issues with these methodologies to date range from inappropriate assumptions in the analytical space to a time consuming overly complex process in the area of optimised simulation modelling. The research presented in this thesis, investigates and tests if it possible to find a balance between the basic rules of thumbs which are easy to interpret/ apply and the area of optimised simulation modelling which is at the upper echelons of advanced analytics but is sometimes out of reach of a fleet size manager due to lack of time, data and expertise. The results of this work established that it is possible to determine a generalizable analytical model for fleet sizing that would adequately replicate the results of a simulation based optimisation approach. The model, although showing positive results from an accuracy and robustness perspective, is limited by the maximum and minimum of fleet size requirements borne from the data on which it has being trained and therefore is not generalizable to problems where fleet sizes larger than 45 may be required. But it should be possible to extend the analytical model for such problem domains.