Geomagnetically induced currents in the New Zealand power system

This thesis focuses on the use of magnetotelluric (MT) data from both the North Island and South Island of New Zealand to model Geomagnetically Induced Currents (GIC) in the New Zealand power network. The model results have been compared with those from a previously used thin-sheet (TS) conductance model and with measured GIC.

Initially, a single station modelling approach using a uniform conductivity Earth model is used to model the measured GIC in a transformer at Islington (ISL). This model is further improved by separately modelling low and high frequency components of GIC and then combining these to give full GIC. The model reproduces most of the GIC variations and the correlation coefficient is >70% for major magnetic storms from 2002-2015. As the model reproduces an average response of the network towards geoelectric fields it underestimates the most of extreme GIC. The analysis of GIC from other substations suggests that measured GIC depend on local geoelectric fields and the substation configuration within the network which cannot be captured using a single station approach. These limitations of single station model are addressed using more realistic geoelectric fields based on magnetotelluric data and consideration of the full network.

To compute geoelectric fields in the whole network the gaps between MT sites are filled using a Nearest Neighbor interpolation technique. As the northern part of the North Island has no MT data an equivalent circuit approach is followed to model GIC for only the lower part of the network. The MT model GIC are in the period range of 2-30 minutes, based on the available MT data period range. Both the MT and TS techniques are used to compute geoelectric fields and to model GIC for the St. Patrick’s Day storm of 2015 and a 20 November 2003 magnetic storm. Both the MT and TS methods show the same transformers as experiencing large GIC during both storms. The primary difference between the models is that amplitudes of high frequency components of the TS model are significantly smaller than for the MT model. In particular they do not produce large GIC during the sudden storm commencement (SSC) of the St. Patrick’s Day magnetic storm. For the 20 November 2003 storm the TS model effectively reproduces the low frequency components and extreme GIC. The model results show that the North Island power network could be at risk during adverse space weather conditions.

Although the South Island has sparser MT data the same technique is used to model SI GIC during both the St. Patrick’s Day and 2003 magnetic storms. Results are compared with measured data from ISL, South Dunedin (SDN) and Halfway Bush (HWB) transformers. The MT model effectively reproduces the measured GIC variations particularly during SSC during the St. Patrick’s Day storm. The TS model gives a very small GIC magnitude during the SSC. During the 20 November 2003 storm both the MT and TS models reproduce strong amplitudes of low frequency components seen in the ISL measured data.

Both the MT and TS models show a substantial scale difference between measured and model GIC both for ISL and HWB transformers that needs to be further explored either in terms of better geoelectric interpolation or power network parameters. Overall, the MT model appears much more promising for future GIC modelling, particularly during a sudden storm commencement and for abrupt GIC variations.