Measurement of soil in sugar cane using non-destructive techniques.
The soil being delivered with sugar cane consignments. from the cane fields to the factories, has been a recent cause for concern in the South African sugar industry. The soil impurities increase the wear of processing machinery reduce extraction efficiency and represents an unnecessary transport of material. The cost due to soil was estimated at R63 million (about US $8 million) over the 1996/97-season. The need to reduce costs, due to the unwanted soil component, has been given a high priority. Ashing is currently used by the sugar industry to estimate the amount of soil in cane. Although simple to implement, the method is destructive, requires long processing times and limited to small sample sizes. In fecent times, non-destructive techniques have become more prominent in industry. Hence, the decision to apply such techniques to the soil in cane-problem. This dissertation describes an experimental investigation into Dual-Energy Transmission (DuET) and X-ray lmaging for quantifying the amount of soil in cane. DuET can determine the relative concentrations of the components of a binary mixture by measuring the transmission of low- and high-energy gamma photons through the mixture. The principle of DuET was successfully demonstrated with aqueous solutions of ferric chloride. Experimentally-determined mass attenuation coefficients of water and ferric chloride were compared to theoretical values. DuET was then applied to dried, shredded sugar cane spiked with various amounts of soil. Results showed large variations in the predicted soil concentrations. These variations were attributed to radiation scatter and incomplete volume sampling by the radioactive source. However, new experimental arrangements are expected to improve the technique: initial test results are given of a sample holder that continuously rotates a sample up and down through the source-detector axis. An alternative approach to processmg DuET-spectra, using the discrete wavelet transform coupled with an artificial neural network, is also introduced. X-ray Imaging was the second technique investigated. A literature survey revealed that this technique had not previously been applied to the soil in cane-problem. The present work constituted an initial investigation to determine the feasibility of applying X-ray imaging to measure the amount of soil in cane. The soil/cane-samples, that were used for DuET, were imaged us ing a commercial mammography unit, and the resulting radiographs were analysed using image processing techniques. Although the results are promising, a more comprehensive investigation is foreseen.