M.Ing. / The dissertation is an investigation on the implementation of new technologies (five off) in a coal pulverising with main aim to optimise mill maintenance interventions. The technologies in question are: • Stationary air throat replaced with a rotating throat assembly. • Hydro-pneumatic mill loading cylinders replaced with airbags. • Classifier cone modification. • Introduction of triton material for the mill spider guide plates. • High chrome mill grinding balls. Every maintenance intervention, even if planned, negatively affects a plant’s availability and reliability. A Babcock and Wilcox (B&W) at Kriel power station (ESKOM) was used for the testing of the mentioned technologies. The mill model/size is a B&W 10.8E mill. The aim of the introduction of new technology on a mill is to optimise the period between required maintenance activities. A higher availability will assist in achieving good plant maintenance performance indicators. It needs to be noted that the dissertation focussed on the financial and technical parameters of a specific modification. This in an effort to increase uptime and reduce costs as part of a business drive for bigger profit margins. The new technologies tested were thus evaluated from a technical and financial point of view. Each technology was implemented at different time periods and nowhere was any tests performed in parallel on a single mill. To get approval from an investment committee for release of money for tests/modifications, technical and financial assumptions need to be made regarding the performance parameters of the modification/change. Once a modification is being tested, actual plant data can be used as inputs into the execution phase of the modification as assumptions can be replaced with test data. A financial model was developed to “test” the financial feasibility of the proposed changes/modifications. With new technology successful implemented in the plant the current maintenance strategies for maintenance interventions can be re-evaluated as the proposed modifications removed historic barriers that determined the current used based maintenance intervals. An example is where a mill is taken from service to replace or repair the stationary air throat (typically every 5 000 operating hours). With the newly applied technology (rotating throat assemblies), there is no need for maintenance interventions every 5 000hrs as a rotating throat assembly can run without major interventions for 60 000hrs.

In power generation from solid fuel such as coal-fired power plants, combustion efficiency can be monitored by the loss on ignition (LOI) of the pulverised fuel. It is the role of the pulveriser-classifier combination to ensure pulverised fuel delivered to the burners is within the specified limits of fineness and mass flow deviation required to keep the LOI at an acceptable level. However, government imposed limits on emissions have spurred the conversion of many coal fired power plants to convert to the use of Low NOx Burners. To maintain good LOI or combustion efficiency, the limits of fineness and mass flow deviation or inter-outlet fuel distribution have become narrower. A lot of existing pulveriser units cannot operate efficiently within these limits hence retrofits of short term solutions such as orifice balancing and classifier maintenance has been applied. The work performed in this thesis relates to an investigation into coal classifier devices that function to control fineness and inter pipe balancing upstream of the burner and downstream of the pulverisers. A cold flow model of a static classifier was developed to investigate the flow characteristics so that design optimisations can be made. Dynamic similarity was achieved by designing a 1/3 scale model with air as the continuous phase and glass cenospheres of a similar size distribution as pulverised fuel, to simulate the coal dust. The rig was operated in positive pressure with air at room temperature and discharge to atmosphere. The Stokes number similarity (0.11-prototype vs. 0.08-model) was the most important dimensionless parameter to conserve as Reynolds number becomes independent of separation efficiency and pressure drop at high industrial values such as 2 x 10 4 Hoffman, 2008). Air-fuel ratio was also compromised and an assumption of dilute flow was made to qualify this. However, the effect of air fuel ratio was ascertained by its inclusion as an experimental variable. Experiments were conducted at air flow rates of 1.41-1.71kg/s and air fuel ratios of 4.8-10 with classifier vane angle adjustment (30°- 60°) and inlet swirl umbers (S) of 0.49 – 1. Radial profiles of tangential, axial and radial velocity were obtained at several cross sections to determine the airflow pattern and establish links with the separation performance and outlet flow balance. Results show a proportional relationship between cone vane angle and cut size or particle fineness. Models can be derived from the data so that reliable predictions of fineness and outlet fuel balance can be used in power stations and replace simplistic and process simulator models that fail to correctly predict performance. It was found that swirl intensity is more significant a parameter in obtaining balanced flow at the classifier outlets than uniform air flow distribution in the mill. However the latter is important in obtaining high grade efficiencies and cut size. The study concludes that the static classifier can be further improved and retrofit-able solutions can be applied to problems of outlet flow imbalance and poor fineness at the mill outlets.