Advertisement

Chemistry and Technology of Fuels and Oils

, Volume 49, Issue 1, pp 59–64 | Cite as

Investigation of transformer oil by nuclear magnetic relaxation and Z-scanning

  • R. V. Arkhipov
  • B. I. Gizatullin
  • A. V. Shkalikov
  • O. E. Kurakina
  • O. A. Turanova
  • V. K. Kozlov
  • A. N. Turanov
Methods of Analysis

GK-grade transformer oil is investigated by methods of nuclear magnetic relaxation and Z-scanning. The size of the heterogeneous structures in the oil increases during its service. The solid-particle content in the liquid phase, however, remains essentially unchanged, and the excess particles precipitate. The feasibility of quantitative determination of the heterogeneous-structure/liquid ratio in oil samples investigated is discussed.

Key words

transformer oil nuclear magnetic relaxation Z-scanning 

Heterogeneous structures – colloidal, micelle-like, and solid particles – affect the properties of transformer oils (TO), and may substantially reduce the breakdown voltage, and alter the viscosity and other TO characteristics critical to power transformers, and, consequently the power systems on the whole [1, 2]. Owing to the physical and chemical processes that take place under the action of temperature, electric fields, and circulation of oil, the composition of the heterogeneous structures is altered over the life of the TO. These processes are currently being monitored by an obsolete method [3], which, in our opinion, is unreliable due to insufficient accuracy.

Investigations [4, 5, 6, 7] have indicated that heterogeneous structures always exist in TO. A method of determining the percentage of solid particles in oil products with use of nuclear magnetic (NM) relaxation is described in the literature [8]. We have been unable, however, to find information on the applicability of this method to TO. The Z-scanning method is sensitive to nonlinear optical absorption of delocalized electrons, occurring most frequently in the heterogeneous particles [9]. This study described results of TO investigation by NM-relaxation and Z-scanning methods.

Three TO samples were investigated. Sample I was a Grade-GK virgin transformer oil produced in 2007 (JSC Angarsk Petrochemical Co., Class PA, Technical Specification 38.101.1025–85). Sample II was a Grade GK oil produced in 1971 (GOST 972-69), and had been used in a TDN-16,000/110/6 transformer from 1972 though 2008. Sample III was a decantate of a Grade GK oil produced in 2002. The oil had been used in a VKM-10/31.5/1,000-U2 switch from 2004 through 2010 (a current cut-off with a force of 11 kA occurred on 27 February 2010).

Transverse relaxation (T 2 ) was investigated with use of an NMR relaxometer in hermetically sealed ampoules 6 mm in diameter with a sample height of 10 ± 1 mm by the Carr–Purcell–Meiboom–Gill method (CPMG, 90° x {– τ– 180° y τ – echo} n ) and solid-body echo (90° x τ – 90° y τ – echo) [10]. The resonance frequency of the protons ν r = 19.08 MHz, the duration of the 90-degree radio-frequency (RF) pulse P 90 = 2.3 μsec, the duration of the 180-degree pulse P 180 = 4.6 μsec, the delay prior to application of each pulse sequence to achieve equilibrium orientation of the protons was t s = 5 sec, the interval τ = 100-400 μsec for the CPMG method and 10 μsec for the solid-body echo, the number of accumulations of each response NS = 100 when the number of points n = 800, and the time required for parallelization of the receiving circuit after an RF pulse t p = 10 μsec.

The self-diffusion coefficients D were measured on an NMR diffusometer with use of a stimulated echo sequence (90° x t 1Gt 2 – 90° x τ – 90° x – t 2G – t 1 – echo) [11, 12], ν d = 64 MHz, P 90 = 15.1 μsec, t s = 5 sec, the interval between the first and second RF pulse t 1 + δ+ t 2 = 2 msec, the diffusion time Δ= 2t 2 + τ = 7 and 121 msec, NS = 10, t p = 8μ sec, n = 64, and the magnetic-field gradient G(t) = 0 – 30 T/m at a pulse duration δ = 0.12-0.55 μsec [13].

The transmission coefficient of the TO was investigated by the standard method of Z-scanning with an open diaphragm [9] and use of a laser with a wavelength λ = 532 nm. The duration of the pulses was t pul = 9 nsec, the pulse-recurrence frequency ν pul = 12 Hz, the average “pumping-excitation” power P pum = 52 MW, and the peak power P pul = 100 kW. An oil-filled quartz cuvette with a thickness l = 1 mm was placed with a fine-adjustment microscrew along the optical axis of a collecting lens with a focal distance L = 50 mm. The intensity of the laser pulses that had passed through the lens and cuvette was measured at the outlet by an FD-24k photodiode, beyond which an opaque partition was positioned with an opening 5 mm in diameter. When z = 0, the cuvette resided within the focus of the collecting lens, and the size of the construction in focus was less than 100 μm. All measurements were conducted at 25 ± 1°C.

Figure 1 shows the transverse magnetization droops (TMD) of the samples, which were obtained by the CPMG method. Application of a solid-body echo sequence made it possible to study the components with short (10-100 μsec) relaxation times (Fig. 2). The TMD of the three samples investigated was approximated by the formula:
$$ A(t)={A_0}\left( {{p_a}\cdot exp\left\{ {-\frac{t}{{{T_{2a }}}}} \right\}+{p_b}\cdot exp\left\{ {-\frac{t}{{{T_{2b }}}}} \right\}+{p_c}\cdot exp\left\{ {-\frac{t}{{{T_{2c }}}}} \right\}} \right) $$
where A 0 is the initial amplitude of the signal, T 2a , T 2b , and T 2c are the times of spin-spin relaxation of the components with populations p a , p b , and p c , respectively (p a + p b + p c = 1).
Fig. 1

TMD of samples (see figures on curves), obtained by CPMG method (for ease of visualization, the plots for samples II and III are shifted along the vertical).

Fig. 2

TMD of samples (see figures on curves), obtained by solid-body echo method.

Let us stress that the component c is not visible in the CPMG, and the values of T 2a and T 2b in the solid-body echo are appreciably distorted due to nonuniformity of the magnetic field. The T 2i and p i values of the TMD-decomposition components of the samples investigated, which were determined with a correlation coefficient |R 2| > 0.95, are presented in Table 1.

Table 1

Sample

T 2 a , msec

p a

T 2b , msec

p b

T 2c , μsec

I

181

0.62

61

0.36

22

II

174

0.62

57

0.36

21

III

149

0.70

46

0.28

19

The presence of two Lorentz components with relaxation times T 2a = 164 ± 17 msec and T 2b = 53 ± 8 msec in the TMD of the TO samples correlates with results of these same samples by high-resolution 1H NMR spectroscopy [14], in the spectra of which the widths at the half heights of the two most intense lines are matched up as 1:3. The signals of these components pertain to the profiles of the liquid-phase molecules. The third component – Gaussian form – has T 2c = 23 ± 5 μsec, and as a result, its width is greater by three orders, and is not observed in the high-resolution method. It pertains to molecules with vigorous intermolecular dipole-dipole interaction [10], which enters into the composition of solid, colloidal, micelle-like structures. The percentage of component c in the TO samples is low (p c = 2%).

A(G 2, Δ) curves of diffusion damping (DD) of the signal of a stimulated spin echo of the samples investigated were obtained in a series of experiments with constant time characteristics (t 1, t 2, τ, δ). The shape of the curves is monoexponential, and they are described with a correlation coefficient |R 2| > 0.99 by the formula
$$ A\left( {{G^2},\varDelta } \right)=A\left( {{t_1},{t_2},\tau, \delta } \right)\cdot exp\left\{ {-{\gamma^2}{G^2}{\delta^2}\varDelta D} \right\} $$
where γ is the gyromagnetic ratio for the protons.

For all three samples investigated, D = (5.7 ± 0.3)⋅10–11 m2/sec when Δ = 7 and 121 msec. These values of the self-diffusion coefficient are characteristic of viscous liquids [15]. It follows from the independence of D on Δ that self-diffusion of the liquid phase experiences virtually no constraints due to heterogeneous structures, since their percentages and dimensions are small.

Results of Z-scanning of the sample investigated – normalized for the highest transmission coefficient τ /τ max – are presented in Fig. 3 as a function of the coordinates of the cuvette. The thermal-lens effect in these experiments can be neglected, since the duration of the pulse (9 nsec) is significantly shorter, during which a stationary density distribution of the medium is established (~ 50 nsec) [16]. Thermal-lens formation due to the accumulation effect can be avoided as a result of low pulse-recurrence frequency. Moreover, lack of gas-bubble formation, and, consequently, absence of scattering from them should be noted in these experiments. These processes are possible, however, in the case when the oil is broken down due to partial absorption of light energy during use of a high-power laser [17].
Fig. 3

Transmission coefficient of specimens normalized for highest value (see figures on curves) as function of position of cuvette (enlarged segment of this plot is shown in inset).

The samples investigated demonstrate nonlinear optical absorption in the visible region of the spectrum, despite the absence of a thermal lens and gas bubbles. Here, a significant amplification of the τ /τ max -z curve is observed in sample III as compared with samples I and II. The difference between the results for the virgin TO (sample I) and used TO (sample II) falls within the range of experimental error, although the values of τ/τ max (z = 0) are clearly smaller for sample II. To obtain more graphic results, and also study the dependence of nonlinear optical absorption on P pul , it is necessary to employ pulses of shorter duration (of the order of several femtoseconds); this will be the subject of our continuing investigations.

Interaction between radiation and delocalized electrons is the most probable mechanism of nonlinear absorption in similar systems [9]. Let us discuss three possible variants:
  • existence of metalloorganic compounds;

  • content of aromatic hydrocarbons in the liquid phase; and,

  • existence of heterogeneous (solid, colloidal, and micelle-like) structures.

The method of electron paramagnetic resonance spectroscopy [14] has demonstrated that metalloorganic Cu2+, Fe3+, Fe2+, Fe+, Er3+, Yb3+, and Ti3+ compounds exist in the samples under consideration. Their concentration (<1019/cm–3), however, is too low to have a pronounced effect on the properties of the samples under investigation; this is also confirmed by results of the NMR-relaxation.

The second variant is, in part, acceptable only for sample I. The one-order intensification of the effect of nonlinear optical absorption is inexplicable in sample III, since the content of aromatic hydrocarbons in the liquid phase of these oils does not differ so significantly [14].

Delocalization of electrons may take place in carbide particles under the action of an electrical arc (sample III); in colloidal formations – “bundled” supramolecular structures of aromatic rings that are formed in liquid oil products [18, 19]; and, in micelles, in the formation of which systems with conjugated π-electron bonds play a key role. Shkalikov et al. [7] describe the size distribution of particles in TO that they have investigated: particles ranging in size from 2 to 20 nm are contained in all samples, particles ranging in size from 100 to 200 nm are also present in sample II, while in sample III, particles ranging in size from 75 to 100 nm are agglomerated into coarse structures with a size of up to 1,000 nm. It is precisely the size of the heterogeneities that is a key factor affecting nonlinear optical absorption in TO. The third version fully explains the difference in the τ/τ max -z curves of the samples investigated.

Based on experimental results of the Z-scanned TO, it is therefore possible to conclude that the percentage of coarse particles, some of which remain in a suspended state and do not precipitate, increases substantially over the service life of the oil. The in-service content of solid particles in the liquid phase of the oil, however, remains essentially unchanged (H ≈ 2%), and the excess particles precipitate.

References

  1. 1.
    A. C. M. Wilson, Insulating Liquids: Their Uses, Manufacture, and Properties, Peter Peretrinus LTD, London-New York, (1980).Google Scholar
  2. 2.
    R. A. Lipshtein and M. I. Shakhnovich, Transformer Oil [in Russian], Energoatomizdat, Moscow (1983).Google Scholar
  3. 3.
    Effect of particles on transformer dielectric strength, CIGRE Technical Brochure No. 157 (2000).Google Scholar
  4. 4.
    Methods and Means of Diagnosing Insulating Oils [in Russian], edited by V. K. Kozlov, OOO ITs Energoprogress, Kazan’ (2003).Google Scholar
  5. 5.
    Power Transformers [in Russian], edited by S. D. Lizunov and A. K. Lokhanin, Énergoizdat, Moscow (2004).Google Scholar
  6. 6.
    L. G. Gafiyatullin, O. A. Turanova, V. K. Koslov et al., Optika Spektrosk.,109, No. 1, 102 (2010).Google Scholar
  7. 7.
    A. V. Shkalikov, E. A. Yushkova, L. G. Gafiyatullin et al., Ibid, 110, No. 5, 750 (2011).Google Scholar
  8. 8.
    N. V. Shkalikov and V. D. Skirda, Uche, Zapiski Kazan. Gosudar. Univ. Ser. “Fiz.-Matem. Nauki,” 151, Book 3, 41 (2009).Google Scholar
  9. 9.
    A. A. Vashman and I. S. Pronin, Nuclear Magnetic Relaxation Spectroscopy [in Russian], Energoatomizdat, Moscow (1986).Google Scholar
  10. 10.
    W. S. Price, Conc. Magn. Reson., 9, No. 4, 299 (1997).CrossRefGoogle Scholar
  11. 11.
    W. S. Price, Ibid, 10, No. 4, 197 (1998).CrossRefGoogle Scholar
  12. 12.
    G. I. Vasil’ev, V. D. Skirda, D. Sh. Idiyatullin et al., In: Petrochemistry-96: Fourth Scientific-Practical Conference on Intensification of Petrochemical Processes. Handbook of Theses [in Russian], Nizhnekamsk (1996).Google Scholar
  13. 13.
    E. W. Van Stryland and M. Sheik-Bahae, In: Characterization Techniques and Tabulation for Organic Nonlinear Materials, Marcel Dekker Inc. (1998).Google Scholar
  14. 14.
    O. A. Turanova, A. R. Sharipova, A. A. Sukhanov et al., Heftekhim., 50, No. 6, 472 (2010).Google Scholar
  15. 15.
    M. W. Deem, J. Phys. Chem., 98, No. 3, 1002 (1994).CrossRefGoogle Scholar
  16. 16.
    R. A. Ganeev, A. I. Ryasnyanskii, M. K. Kodirov et al., Zh. Tekhn. Fiz., 72, No. 8, 58 (2002).Google Scholar
  17. 17.
    G. M. Mikheev, D. L. Bulatov, T. N. Mogileva et al., Pis’ma v Zh. Tekhn. Fiz. 33, No. 6, 41 (2007).Google Scholar
  18. 18.
    J. P. Dickie and T. F. Yen, Anal. Chem., 39, No. 14, 1847 (1967).CrossRefGoogle Scholar
  19. 19.
    N. Nasnaoui C. Achard, and M. Rogalski, Revue de L’institut Franzais du Petrole, 53, No. 1, 41 (1998).Google Scholar

Copyright information

© Springer Science+Business Media New York 2013

Authors and Affiliations

  • R. V. Arkhipov
    • 1
    • 2
    • 3
    • 4
  • B. I. Gizatullin
    • 1
    • 2
    • 3
    • 4
  • A. V. Shkalikov
    • 1
    • 2
    • 3
    • 4
  • O. E. Kurakina
    • 1
    • 2
    • 3
    • 4
  • O. A. Turanova
    • 1
    • 2
    • 3
    • 4
  • V. K. Kozlov
    • 1
    • 2
    • 3
    • 4
  • A. N. Turanov
    • 1
    • 2
    • 3
    • 4
  1. 1.Kazan’ (along the Volga) Federal UniversityKazan’Russia
  2. 2.E. K. Zavoiskii Kazan’ Physico-Technical InstituteKazan’Russia
  3. 3.Kazan’ Science CenterRussian Academy of SciencesKazan’Russia
  4. 4.Kazan’ State University of EnergeticsKazan’Russia

Personalised recommendations