Material dependency of measuring cells (Copper & Gold)

In this experiment the new measuring cells (Cu & Au) are used. This experiment is meant to study if the material choice of the measuring cells can affect the results. Two measuring cells with a distance equal to previous experiments, 2 mm, using the old measuring cell are simultaneously placed in a container filled with transformer oil. One measuring cell is made of gold, the other of copper. The geometric capacitance of both cells are measured and IDA’s settings are adjusted correspondingly for each measuring cell. Then the resistivity is measured. Firstly, degassed mineral oil is used. The oil is warmer than room temperature after degassing, because of the heating plate, figure6.4. As such the test is done three times, alternating between gold and copper, while the oil cools down in the climate chamber. This gives three set of resistivity measurements that can be compared.

Figure 8.19: Measured resistivity using Cu and Au cells. Distance between plates = 2.0 mm The results are a significant difference in measured resistivity. This result is unexpected and a cause for more experiments being performed. Before this result the hypothesis was that there would be no difference between the materials as IDA calculates the resistivity only using the geometric capacitance and the real part of the impedance.

The geometric capacitance of the cell should depend on the distance between electrodes and their area, given by equation 7.1. As the geometric design measurements are equal for both the gold and copper electrodes this parameter is thought to be constant. Measurements of the geometric capacitance confirms this as the different cells differ at the top end (when the distance between electrodes is smallest) around ∼ 5pF. This difference is within acceptable limits as the distance between the electrodes vary a bit depending on the quality of the washers and how tightly the electrodes are screwed together.

The geometric capacitance value is therefor set as a constant for each corresponding distance. These values are shown in table8.4.

Distance between electrodes [mm] Geometric capacitance [pF]

1.3 80

2.0 50

3.3 30

Table 8.4: Geometric capacitance increases with shorter distance between electrodes

The impedance parameter depends on the measured voltage and current by the electrodes. Gold and copper are both good conductors, the small difference in conduction relative to the insulating properties of the transformer oil shouldn’t affect the measured results as significantly as the results show.

The cell is cleaned and rebuilt using a longer distance (∼ 3.3 mm). Another set of three, alternating, measurements are done using the same batch of oil as the previous measurement of 2 mm distance.

Figure 8.20: Measured resistivity using Cu and Au cells. Distance between plates = 3.3 mm Less difference in resistivity is shown at the new distance. This information is interesting and needs to be confirmed. Therefore two tests, one using isoparaffin oil A another using mineral oil, are done with altering distance between the electrodes of 1.3 mm and 3.3 mm. The upcoming experiments are performed more carefully compared to the previous two.

A set of electrodes, either 1.3 mm or 3.3 mm, are placed in the same container. The container is placed in the climate chamber and left for at least 30 minutes. This to consider the temperature as a constant. The resistivity is measured for a pair (copper and gold with equal distance between electrodes) and then the pair is replaced with the other pair. In between the experiment using isoparaffin A and the mineral oil. The setup (cell + container) are cleaned according to the cleaning procedures.

The new measured resistivity using the isoparaffin oil A with different distances and electrode material are as follows.

Figure 8.21: Measured resistivity using Cu and Au cells. Distance between plates = 1.3 mm (left) and 3.3 mm (right)

Isoparaffin A shows the same behaviour as longer distance yields less difference in result. The measured resistivity is higher for gold than copper for both distances. This corresponds to the results from the previous measurement at 3.3 mm, figure8.20, and 2 mm, figure8.19.

The new measured resistivity using the mineral oil with different distances and electrode material are as follows.

Figure 8.22: Measured resistivity using Cu and Au cells. Distance between plates = 1.3 mm (left) and 3.3 mm (right)

The measured Au values for the 1.3 mm distance is taken from the mean value of 6 measurements. Generally, there is a noticeable difference when it comes to both distance and choice of material of the electrodes. With gold showing higher resistivity than copper and larger distance between electrodes yielding greater difference in measured resistivity.

The percentage difference between Au and Cu for the different measurements throughout this section are displayed below. Since the initial measurement of 2 mm, using mineral oil, consists of measurements of varying temperature, the result corresponding to the lowest temperature, closest to room temperature thus closest in temperature to the other measurements, is considered. Regarding the first measurement of 3.3 mm, using mineral oil, a mean of the three measured results is considered. As the percentage difference is a relative term between two measurements. Although resistivity cannot be fairly compared for each set of experiments as the environment changes. For a relative comparison of results, this should not be an issue.

Percentage difference between two numbers is given by

Dif f erence[%] = 100 × |x1− x2|

x1+x2

2

(8.3)

Figure 8.23: Percentage difference in measured resistivity using Cu and Au cells with varying distance and transformer oil. Green = Isoparaffin oil A, Blue = Mineral oil.

9. Discussion

In this chapter the results from previous experiments and theory are discussed. It starts with observations about the ion drift model, then the triangular method is compared to IDA. A correction algorithm is suggested to the triangular method and pros/cons for the two measurement techniques are discussed. Afterwards the moisture and dielectric properties of the various transformer oils used in this study are gone into with observations regarding the moisture of certain oils that the author finds to be especially interesting. The electrode material of Cu and Au yielding different resistivity measurements is, as the writing of this report, still an unexpected result. Some ideas on to what may cause this phenomenon are argued for/against.

Lastly, general ideas about the technological advancements in the field of transformer oils are brought up with arguments based of results obtained in this thesis together with information about the environmental consequences related to the different transformer oils.

9.1 The Ion Drift Model

One of the advantages of the ion drift model is that it only requires four parameters, which are measured at low voltages; mobility of positive and negative ions, resistivity and relative permittivity, to simulate the dielectric behaviour of the transformer oil at very high voltages.

In this study it is found that the relative permittivity remains fairly constant for each of the four oils. With ester oil showing the highest relative permittivity of ∼ 3 and the remaining three oils spanning from 2.0-2.2, shown in figure8.16and8.18. This constant behaviour of the relative permittivity could allow for a reduction of parameters required in the model from 4 to 3. Giving the relative permittivity a constant value. This constant behaviour of the relative permittivity is especially noticeable for transformer oil which originates from crude oil (isoparaffin & mineral oil). As mentioned earlier, the triangular method can only measure resistivity. By knowing the constant behaviour of the relative permittivity, it would remove the necessity to use the more complex IDA diagnostic tool. As both the mobility and resistivity can be measured using existing Labview software at ABB Power Grids Västerås.

As mentioned in the ion drift model description, the ions are generated from dissociation of ionic pairs. This process relates to the applied electric field as given by the “b” term related to the field dependent equation 3.7. Plugging in the constants and a temperature of 300 Kelvin. It’s clear that the required applied voltage to affect the dissociation of ionic pairs is large.

b ≈p

24.5 × 10⁻⁶|E| (9.1)

Considering this, the chosen voltage for the experiments, up to 4 V peak to peak for the triangular method, should not affect the ions in the transformer oil in any substantial way.

By a quantitative comparison between the relaxation time to the transit time by help of the κ-term.

κ = τtransit

τrelax

= d²

0rρµtV (9.2)

The chosen, adjustable, parameters such as voltage and distance between the electrodes can be evaluated.

Using a voltage of 2 V and a distance of 2 mm (the originally recommended settings, used in most exper-iments), taking the measured resistivity, relative permittivity and mobility (provided by the other master thesis) for each oil into consideration. The desired criteria of a κ-term greater than one, thus having the ions in a relaxed state, is fulfilled for each oil. However, for mineral and isoparaffin B the margin is not great.

Transformer oil κ, rounded to closest integer

Mineral 10

Ester 1882

Isoparaffin A 109

Isoparaffin B 4

Table 9.1: κ values for each transformer oil

Larger distance between the electrodes or lower applied voltage would yield a higher κ value. Another prob-lem when using a short distance is expanded on in the section regarding the different electrode material.

Larger distance would mean more oil between the electrodes thus increasing the total resistance. This would make it harder to measure the current as it already is in the pA range when using the recommended settings (2 mm & 2 V). This is especially interesting for the higher resistive oils (Mineral & Isoparaffin B, more about this in the dielectric parameters section of the discussion) as it could require a longer distance between the electrodes to ensure the relaxed state of the ions. Something that would require a higher applied voltage to ensure a measurable current.

Looking at the expression of κ, equation9.2. We see that the term increases with the square of the distance and decreases linearly with applied voltage. A trade-off between higher applied voltage, thus pushing the system further away from thermodynamic equilibrium, and having the ions in a relaxed state needs to be made. The recommendation from the author of this thesis would be to keep the voltage at 2 V for IDA or 4 V for the triangular method and slowly increase the distance until any noise is noticeable. When noise is noticeable, double the voltage and do the same procedure again. Voltages that are in the range of tens of volts shouldn’t affect the ions in the oil too much as mentioned in the previous part regarding the "b"

term, equation 9.1. Noise in the triangular method is easily distinguished from the oscilloscope readings and knowledge of the theoretical current response, figure 8.1. IDA has a safety mechanism that stops a measurement if the current is below a certain threshold.