Even minimal improvements in particle collection efficiency of electrostatic precipitators significantly reduce dust emission from fossil-fuelled power stations and reduce pollution. Yet current designs rely on the Deutsch collection theory, which was developed for tubular precipitators and has been applied to wire-plate precipitators on the assumption that the inter-electrode electric fields at the same discharge distance in both were similar. Differences in geometry and associated collector electric fields and current density non-uniformity have not been taken into account, although the collector electric field and current density of the wire-plate precipitator are not uniform. And observations show that precipitated dust patterns and the distribution of collector current density are interrelated. Investigations revealed a simple square law relationship between the collector electric field and the collector current density in the space charge dominated coronas. Applying this relationship to the Deutsch collection theory led to a current-density-based collection formula that takes into account the non-uniform collector current density distribution. The current-density-based collection formula is then used to assess the impact of collector current density on collection efficiency, the results closely following published measurements. Applying the current-density-based collection formula to estimate the dust accumulation shows that most of the dust accumulates at collector locations facing the corona wires. The effect of the non-uniform precipitated dust layer on collection performance is assessed using the distributed corona impedance - the ratio of the inter-electrode voltage and the non-uniform collector current. Re-distribution of the collector current profile as dust builds up is also compatible with published measurements. Finally this is applied to optimize the wire-plate precipitator collection performance. This shows that optimal collection performance is obtained with the wire-wire spacing less than the wire-plate distance, once again confirming published experimental results. This is the first analytical approach to show better collection performance can be achieved at the ratio of wire-wire spacing/wire-plate distance not equal to unity, which has been the standard industry practice since 1960.