Pinus radiata logs currently make up a large proportion of New Zealand’s forest exports, which in turn are a major revenue earner accounting for approximately $3.8bn per annum. If shipped untreated the logs may be infested with parasites that can do significant harm to foreign forestry and agricultural industries. The current pre-shipment treatment method, known as phytosan- itation, for many green logs is based on fumigation with methyl bromide. However, methyl bromide has been identified as an ozone depleting agent. To mitigate the harm from methyl bromide, New Zealand has regulated that the release of the chemical into the atmosphere must cease by April 2021. Hence there is a need to develop an alternative form of treatment.

Joule heating is currently being explored as an alternative phytosanitary treatment. A key issue involves the formation of hot spots during the heating process. Hot spot formations are capable of causing catastrophic damage to the timber before the treatment process is complete. Due to the sharp difference in conductivity between clear sapwood and knots, these hot spots have been identified as most likely forming surrounding knots in the timber. Therefore a non-invasive way of identifying these hot spots during Joule heating is required.

A prototype scanning unit has been produced that contains an array of capacitive sensors and search coils for measuring electromagnetic fields for use in this research. Based on testing of the scanning unit the search coils were identified as having a sensitivity of 1.25 mV/nT with a 4 nT change in flux density being repeatably discernible. Based on the operating range of the measured signal this results in a signal to noise ratio (SNR) range of 35.6 - 54.0 dB. The capacitive sensors were identified as having a sensitivity of 80 µV/V m⁻¹ with a 10 Vm⁻¹ change in field strength being repeatably discernible.

A steepest descent imaging system, with iterations constrained by Kirchoff’s current law has been developed and is compared to a number of other techniques. This method has been demonstrated to produce more accurate images in the presence of noisy signals than other methods considered.

Using simulations, developed by the author using MATLAB software, appropriate stopping criteria for the imaging methods and appropriate resolution-to-scanning-distance ratios were evaluated. The stopping criteria were found to be dependent on the noise level and the scanner distance, with an average change of 0.1 mA in the current field being generally suitable through- out the testing conducted. The distance ratio also has a significant effect. However it was found, through simulations, that at a SNR of 35 dB, the search coils should not be at a distance greater than twice that of the desired resolution.

An experimental setting was constructed to verify the imaging method in the simulations. While the scanning unit would have benefited from having additional search coils aligned in an alternate orientation, it was still able to produce satisfactory results to a resolution of 20 mm at all distances tested.

The collection of evidence submitted in this work suggests that the imaging method developed would be capable of imaging the outer-most current distribution in timber during Joule heating. Within expected signal ranges the method is expected to be capable of identifying knots in the timber as small as 20 mm diameter, assuming the search coils can be positioned within 70 mm of the centre of the imaged current. This is a finer resolution than the specifications require. Hence, at a greater range this method should be adequate for identifying knots of 40 mm diameter as was originally required. If used in conjunction with other works this is expected to be sufficient to determine a suitable heating strategy for a given log.