M.S.

South Africa contributes approximately 0.5 percent to the total world milk demand and is the third largest producer of fresh cow milk in Africa after Sudan and Kenya. In comparison to any other enterprise, the cost of milk production is influenced by numerous factors, that in turn affect the profitability of the farm enterprise; however one of such factors is high electricity cost. In this regard, there is need for efficient operation of the milk processing plant at all stages and at the same time maximising on product quality and minimising on the cost of production including energy. At the dairy farm, milk handling mainly commences as the milk leaves the cow udder at 35°C–37°C and must be cooled rapidly to a storage temperature of 4°C in a bid to stop microbial activity. The cooling of the milk can be done directly by the bulk milk cooler (BMC) from 37°C to the required storage temperature of 4°C or it can be done successively through pre-cooling. The process of pre-cooling involves the use of a heat exchanger where in most instances the plate heat exchanger (PHE) is used as the pre-cooler (PC) thereby leading to energy savings in a dairy facility. Cooling of milk involves significant amount of energy and it could account for about 20 percent of the total energy consumed on a farm. The aim of the research was to develop mathematical models that could be used to predict the electrical energy performance and capture the cooling saving of an on-farm direct expansion bulk milk cooler (DXBMC) during the milk cooling process. Accordingly, data acquisition system (DAS) was designed and built to accurately measure the power consumption of the BMC, temperature of raw milk, room temperature, temperature of cold water, relative humidity and ambient temperature. The volume of milk produced per day was extracted from the daily records on the farm. In addition, the temperature sensors were connected to a four channel HOBO data loggers which were configured to log at every five-minutes interval. The results were analysed and the mathematical models were developed using MATLAB. The statistical Toolbox in MATLAB was used to rank the predictors according to their weight of importance using the ReliefF Algorithm test. The results showed that on average, the daily electrical energy consumed by the BMC at the two milking times was higher during the peak period (127.82 kWh and 93.86 kWh) than the off-peak period (48.31 kWh and 43.23 kWh). On average, the electricity used for cooling of milk on the dairy farm was 17.06 kWh/m3 of milk. The average monthly electricity used per cow on the farm was 8.03 kWh/cow which translated to an average of 0.26 kWh/cow/day The average specific energy consumption of the cooling system per litre of milk cooled was 0.02 kWh/L and was almost constant throughout the whole period of monitoring. Furthermore, the BMC was able to cool 57.33 L/kWh during the off-peak period which increased by 7.7 percent to 62.13 L/kWh during the peak period. Furthermore, mathematical models represented as multiple linear regression (MLR) models were built and developed using the experimental data. The developed mathematical models had good agreement with the experimental data as evidenced by the correlation coefficients of 0.922 and 0.8995 along with 0.935 and 0.930. The ReliefF Algorithm test revealed that the volume of milk was the principal contributor to the energy consumption of the BMC for both the morning (AM) and afternoon (PM) milking period. The Relative Prediction Error (RPE) was used to evaluate the suitability of the developed models. In this light, the AM off-peak model had RPE of 18.54 percent while the PM off-peak model had 14.42 percent. In addition, the AM peak and PM peak models had RPE of 19.23 percent and 18.95 percent respectively. This suggested that the MLR models for the off-peak and peak milking periods (both AM and PM) had acceptable prediction accuracy since the RPE values were between 10 percent and 20 percent. The findings from the experimental study showed that the coefficient of performance (COP) of the AM milking period was higher (2.20) than that of the PM milking period of the BMC (1.93). Increase in the milk volume led to an increase in the COP such that the peak period with higher milk volumes recorded a high COP increase of 12.61 percent and 19.81 percent for the AM and PM milking periods respectively.

No description available.