Milled peat used for energy is produced by scarifying the surface of a drained peat bog using a milling machine. The particulate layer produced dries atmospherically and is then collected, stored and ultimately burned for energy. In order to be valuable and to burn efficiently the product must be at the correct water content. The peat type, depth of milling and weather are the main factors influencing the water content of a milled peat layer. The ideal situation is to have an uneven layer depth but all of the same water content. Current milled peat production is analogous to farm crop production because in general, a uniform milling depth is used over a wide area, in the same way that uniform management is applied to a field on a farm. The idea of precision agriculture is to identify spatial variability within fields, and to use variable management in response. To do this, a means of mapping relevant variability is needed (usually via a sensor), in addition to a means of knowing location (commonly by the Global Positioning System - GPS) and a means of variable rate response (a mechanical method of changing action to that required for a given location). These concepts can be applied to industrial peat production to improve the efficiency of utilisation of the resource. The system is quite complex (1), theoretically requiring details of peat type, weather, initial water content, machine/bog interaction, amongst other things, but can be reduced to three basic components: mapping, location and variable rate action.

Location: A Global Positioning System (GPS) is ideal for location finding on the bog. Low-cost GPS units are only accurate to about 100 m, or when differentially corrected to 2 m. Adding differential correction adds to the cost of implementation. Due to the number of machines used on the bog, and the low margins involved, it was decided that if GPS could be used in non-differential mode it would be more appealing to the industry, but the accuracy would have to be better than 15 m. The structure of an industrial peat bog (parallel fields 15 m wide separated by drains) was used to implement a "pseudo-differential" correction (3). Once a tractor moves onto a given field it is limited as to where it can go by the drains either side. This knowledge means that if the location of the field is known (as is the case for surveyed industrial peat bogs) then it is possible to correct a basic GPS such that any point indicated by the GPS unit as lying outside the boundary of the current field can be "corrected". The method was tested using comparison with a survey grade GPS received accurate to a few millimetres.

Depth control - In order to be able to use the information about peat type at each location to optimise milling, it is necessary to have a means of controlling both the desired cut depth, and to accommodate for sinkage of the machine as bearing capacity changes with location. A sensor arm was developed by Condon (4) that could quantify the depth of the milling drum (and thus cut depth) relative to the bog surface at any given moment in time. This could be used as the input sensor to a control system that could set cut depth, and apply fine adjustments to account for sinkage. Complete mathematical and control models were developed by Condon to allow real-time depth control on the bog (4).

In an integrated system this output would act as an input to Condon's depth control system, thus delivering the three basic components required for system optimisation (viz. mapping, real-time location and variable rate action).

Fig. 1. Satellite derived density map of a bog (Landsat TM)

Fig. 2. Improvement in location fix by applying "pseudo-differential" correction to basic GPS data. A: basic data. B: Corrected data.

Fig. 3. Condon's control model for miller depth adjustment and maintenance (4).