Вы находитесь на странице: 1из 8

4 Congreso Uruguayo de Mantenimiento, Gestin de Activos y Confiabilidad - URUMAN 2008

Del 02 al 04 de Septiembre

Comparison of the Different Available Models for Pressure Rise due to Internal Arcing and Evaluation of Fast Depressurization Systems for Tank Rupture and Fire Mitigation
G. Prigaud, G. de Bressy, S. Muller, R. Brady, P. Magnier

SERGI Holding Research Department


Abstract Transformer explosions are caused by electrical arcs occurring in the insulating oil. They generate dynamic pressure waves that propagate in the tank and interact with its structure leading to its explosion. This whole process was studied using experiments performed on large scale transformers. Moreover, to accurately study the phenomena exhibited during this experimental campaign and in order to minimize the costs and the risks of the testing, a computational tool was developed to simulate the transformer explosion. An unsteady compressible 2-phase flow approach was used and viscous flow, electromagnetic, thermal and gravity effects are taken into account. The equations are solved using a finite volume method on unstructured mesh, allowing computing complex 3D transformer geometries. Comparisons between experiments and simulations validate the model that can thus be used to study explosions of large transformer tanks and its prevention by fast-direct-tank-depressurization-based methods. Index Terms Pressure effects, propagation, protection, transformer, wave.

I. INTRODUCTION POWER transformers are one of the most dangerous electrical equipments because of the large quantity of oil they contain which is in direct contact with high voltage elements. Under such circumstances, low impedance faults that result in arcing can appear in transformer tanks once the oil loses its dielectric properties. Oil is then vaporized and the generated gas is pressurized because the liquid inertia prevents its expansion. The pressure difference between the gas bubbles and the surrounding liquid oil generates pressure waves, which propagate and interact with the tank. The average pressure rises and leads to the tank explosion and possible fire resulting in very expensive damages for electricity facilities. Despite all these risks, and contrarily to pressure vessels, no specific standards have been set to design and protect transformer tanks subjected to large dynamic overpressures. Realizing that a transformer explosion leads to huge financial losses, many studies have been performed to analyze the explosion process in order to establish strategies to prevent it [6] [7]. More recently a complete experimental study was performed by the CEPEL and the SERGI Holding Company. It consisted in arcing tests in industrial size oil-immersed transformers. Detailed results can be found in [11] and as it is the base of the model, the most important points are recalled in this paper. Since live tests performed on real scale transformers are expensive and dangerous, an alternative is to study transformer explosions using computational simulations. The first model [10] considered the pressure uniform in the tank (0D model) and the tank oil incompressible. The authors compute the amplitude of the pressure peak induced by an arc in a gas blanket located in the top of the tank considering a semi empirical energy conservation equation. The study detailed in [9] is based on similar hypotheses and considers that the tank expansion has to absorb the volume of the gas generated by the arc. Both works concluded that the best way to avoid an explosion is to give place to oil in order to absorb the oil expansion due to the electrical arc. A more elaborated method, detailed in [4] considers a 2D geometry and a potential flow model, using a source singularity to represent the arc. This incompressible and non-viscous model allows computing the deformation of the transformer tank considered as a thin-walled infinite cylinder shell. This model, completed with some empirical laws is also used in [3] to compute the overpressures generated by different arc characteristics for different tank geometry parameters. Computation capabilities and CFD modeling progress over the last years now allow simulating industrial type problems with efficient unsteady 3D CFD models in reasonable time. This paper thus presents: the development of a numerical tool able to describe all the complex physical phenomena occurring during an explosion; the description of arcing tests in industrial size transformers and the analysis of their results, with which the modeling is validated by comparisons with experimental data [11] in terms of pressure evolutions; the simulation results that evaluate a fast-direct-tank-depressurization-based method to prevent transformer explosions, which operation has been experimentally studied. -1-

4 Congreso Uruguayo de Mantenimiento, Gestin de Activos y Confiabilidad - URUMAN 2008


Del 02 al 04 de Septiembre
II. NUMERICAL SIMULATIONS A. Model Requirement Under certain circumstances, the transformer oil loses its dielectric properties, which means an electrical arc occurs and vaporizes a part of the oil leading to the creation of a pressurized gas bubble; this high pressure then propagates in the oil and interacts with the tank eventually leading to its rupture. In this context, the pressure increase calculated when using an incompressible description of the fluids is induced by the direct and instantaneous competition between the gas bubble expansion and the small deformations allowed by the tank ([4],[10]): in this kind of modeling the pressure increase cannot appear if the liquid is not contained in a tank. On the other side, the compressible vision enables simulating pressure waves propagating at finite speed. They are induced by the electrical arc and are independent of the vessel. Delays in the pressure rise and a spatially progressive pressure wave/tank structure interaction can thus be accounted for. Moreover, in this kind of modeling the pressure rise is just due to local physical phenomena based on energy transfers between the oil and the arc, without assuming any specific topology for the problem. The much more general compressible approach is thus preferred to study the transformer explosions, and set an efficient transformer explosion computer simulation tool that deals with: Two phase flows to take into account gas and oil; Compressible flows to simulate the pressure waves propagation; Oil viscosity and stiffness; Electromagnetic, thermal and gravity effects; 3D transformer geometries, Computation of the mechanical loads in the bolts, welds and tank walls. B. Physical Modeling 1) Governing Equations The set of equations used to theoretically and numerically describe the phenomena is an Eulerian model for compressible two-phase flows. The hydrodynamic mathematical model is based on a set of Partial Differential Equations (PDE), which governs the hydrodynamic behaviour of mixtures. One of the major and most interesting models characteristics is its ability to accurately depict the pressure wave propagation inside liquids and gases. It is detailed in several references [1][13] and can be obtained from the general two-phase flow model described in [2][12] by an asymptotic analysis [8] and an assumption on the volume fraction evolution. Some physical effects are added in the modelling in order to be as closed as possible to reality in term of gravity, viscosity, and heat transfers. 2) Thermodynamic Closure For the model to be complete and consistent, a thermodynamic closure has to be added in order to link the mixture pressure to the mixture total energy. Each phase, gas and liquid, is thermodynamically described by its own Stiffened Gas Equation Of State (EOS). For each medium, this simple EOS leads to a theoretical sound speed in very good agreement with the experimental one. Then, using the pressure equilibrium assumption, and internal energy additivity property, a Stiffened Gas EOS is found for the mixture. This mixture EOS is very useful to handle the artificial mixture zone generated between liquids and gases by the model evolution. More details can be found in the literature [1][13] which show the ability of the method to describe the pressure wave propagation in two-phase flows. C. Numerical Methods 1) 3D Finite Volume Method A Finite Volume Method is chosen to numerically solve the PDEs system. A numerical flux has to be computed at each cell boundary. In order to calculate such a flux, the boundary physical state must be determined by solving a Riemann problem at each cell boundary.

-2-

4 Congreso Uruguayo de Mantenimiento, Gestin de Activos y Confiabilidad - URUMAN 2008


Del 02 al 04 de Septiembre
2) Godunov Schemes, Riemann Problems The model under consideration in the present study is based on a hydrodynamic hyperbolic model for compressible twophase flows. It is the most important part of the modelling because the pressure waves are depicted by this part of the equations. In a Finite Volume Method, there is a Riemann problem to solve at each cell boundary at each time step in order to compute the numerical fluxes. Details about Riemann problems, their adaptation, and their solving can be found in [5][14]. Following the rules detailed in [5] and [14] when presenting the Godunov solver for Euler equations we can also build an exact solver for our model. The various state expressions can be deduced from the ones in [5] and [14]. When dealing with a numerical tool this solver has to be associated with numerical schemes and boundary treatment. 3) Electromagnetic, Gravity, Viscous and Thermal Effects The gravity, viscous and thermal source terms of the equation as well as the arc energy transfer are taken into account by a time splitting scheme, including the contributions of the physical effects considered separately. The numerical time integration scheme is then based on a first order Runge-Kutta time scheme. III. APPLICATION TO TRANSFORMER EXPLOSION PREVENTION A. Experimental campaigns In 2004, 34 internal arcing tests were carried out by CEPEL, the Brazilian independent High Voltage Laboratory in transformer tank dielectric oil to: (1) study the vaporization process induced by an electrical arc and the resulting pressure wave propagation; (2) evaluate the efficiency of a fast-direct-tank-depressurisation-based method to prevent transformer tank explosions [11]. Three standard transformers were used for these tests. The maximum distance between an electrical arc and the protection system ranged up to 8.5 meters (28 ft). Various physical parameters such as pressure, gas temperature, applied current, arc voltage and tank acceleration were studied in details. This part presents the main conclusions of these tests performed on large transformer tanks 1) Experimental settings Each transformer was equipped with: (1) arcing generation devices to master the arc location and make the tests repeatable, (2) temperature probes, (3) accelerometers, and (4) pressure sensors at different locations in the transformer, in order to study the pressure wave propagation.

Figure 1 : Life Tests Transformer Principle Drawings An explosion prevention system, which details can be found in Figure 1 and appendix, was also installed on each transformer to study the ability to mitigate tank rupture by a tank depressurization method based on the fast and direct passive mechanical response of a depressurization set to the pressure wave. 2) Experiments To study in detail the pressure wave propagation influence, the electrical arcs were ignited at three different locations, as shown in Figure 1: on the top cover close to the Decompression Set location (position A), on the top cover opposite the Depressurization Set location (position B), and in the lower part of the tank opposite the Depressurization Set location (position C). The position D shown in Figure 1 is the location where the depressurization set was installed. Most of the tests were carried out with electrical arcs with currents ranging from 5 to 15 kA, and fed during 83 milliseconds. This duration corresponds to the average response time of an old circuit breaker and was chosen to maximize the generated gas volume.

-3-

4 Congreso Uruguayo de Mantenimiento, Gestin de Activos y Confiabilidad - URUMAN 2008


Del 02 al 04 de Septiembre
B. Transformer Explosion Prevention: Test Results Analysis 1) Generated gas

Figure 2 : Generated gas volume v. arc energy (P=1 bar, T=300 K) During the CEPEL test campaign, the electrical arc produced from 1 to 2.3 m3 (35 to 88 ft3) of gas. This volume is plotted as a function of the arc energy in Figure 2. For the tested energy range, the gas volume generated during an electrical arc is a logarithmic function of the arc energy, which seems in accordance with the vaporization process and especially with the saturation of the vaporization for high energy arcs: the arc remains in the generated gas volume using its energy to crack the oil vapour rather than continuing directly vaporizing the oil, which results in a smoother vaporization process. 2) Pressure Wave Propagation In Figure 3, experimental pressure profiles are displayed. Each curve shows what happens near each sensor located in positions A, B and C (see Figure 1). The arc is generated in C and the shock wave propagation can be step by step followed because of the pressure peaks displacement from C to A. The other pressure peaks (smaller than the main peak) are due to wave reflections off the walls. The pressure does not rise spatially uniformly in the tank. The experiments show the pressure waves propagate in the oil at a finite speed.
Pressure Signal : Close to the arc (C) At the tank cover (B) Close to the TP (A)

Maximum static withstand limit +1.2 bar (+ 17.4 Psi) relative pressure

Figure 3: Experimental Pressure Measurements

Figure 4 : Maximum relative Pressure close to the Arc v. Arc energy (reference pressure: atmospheric)

3) Pressure Peaks Only one main pressure peak has been noticed for each test. The pressure profiles show variations after that main peak but their magnitude remains low compared to the first pressure peak level. They are due to waves reflections. The pressure peaks amplitude is determined by the created arc. This peak ranges from +1.5 to +13 bar (+21.75 to +188.55 psi) for arc energies from 0.01 MJ to more than 2.4 MJ as shown in Figure 4. The maximum pressure seems to strongly increase with the arc energy while the energy remains in the low range, this dependence weakening as the energy increases. The pressure rise is indeed the result of the strong oil vaporization that takes place in the arc very first moments, the energy transferred after while having less impact on the pressure build-up.

-4-

4 Congreso Uruguayo de Mantenimiento, Gestin de Activos y Confiabilidad - URUMAN 2008


Del 02 al 04 de Septiembre
The pressure peaks values do not strongly depend on the arc energy since in Figure 4, when comparing tests for which pressure peaks respectively equal +8 bar (+116 psi) and +8.8 bar (127 psi), the maximum pressure only varies in 0.8 bar (11.6 psi) while the corresponding arc energies vary within on order 10 of magnitude (0.1 MJ and 1 MJ respectively).: 4) Tank Withstand To static pressure: To check the mechanical properties of the transformers, static tests were performed before applying any low impedance fault. The withstand limit was found to be +0.7 bar (+10.15 psi) for the biggest transformer, T3. This limit (+0.7bar,+10.15 psi) was used in the analysis as a threshold for the tank depressurization during the dynamic tests. To dynamic pressure: Even if the local pressure measured during the dynamic tests is on average 6 or 10 times higher than the static withstand limit (Figure 6), no tank damage and no tank permanent deformation occurs because the pressure peaks are very short. In fact, the structure can locally withstand high dynamic pressure increases because of the walls elasticity and the prevention method small inertia to operate. C. Computer Simulation Tool: Application to Transformer Tank Rupture Mitigation In order to validate the mathematical method presented in this paper, numerical tests have been performed and compared to the experimental data. Simulations have been run with and without the depressurization set to evaluate the method efficiency to prevent tank rupture. 1) Experimental Tests for Comparison The transformer is 4.6 m long and 3.8 m high. The 0.2m-diameter depressurization set is located 3.6m above the ground. The experimental test used for comparison with computer simulation results is test 31, in which the transformer is subjected to an 83ms arc occurring in position B with a maximal current peak of 30kA (nominal value 14kA) and a maximum voltage of 1kV. It considers a depressurization set with a calibrated burst pressure of +1.5 bar (+21.75 psi) and with an outer reference absolute pressure of 0.1 bar (1.45 psi).

Experiment (Test 31) Pressure close to the arc Simulation Pressure close to the arc Figure 5: Geometry influence on pressure profiles 2) Geometry, Initial and Boundary Conditions The simulated geometry is that of the test 31 (see section C.1). The boundaries are walls except for the depressurization set. In the initial state of the simulations the gas bubble has already been created by the arc and is already under pressure (P=4.3bar, 62.4psi; =4.3 kg/m3, 0.27 lb/ft3). Bubbles initial overpressure is determined according to the arc energy. 3) Experiment/Simulation Comparison Experimental and numerical results regarding the pressure time evolution are in good accordance (Figure 5): Qualitatively, (1) the pressure wave propagation is well described and especially the delay in the pressure rise, and (2) the three same phases can be observed in both cases: (a) a very sharp pressure rise following the arc ignition, (b) a pressure drop, and (c) a phase where the pressure alternatively rises and decreases because of the complex wave dynamics due to the wave reflections off the transformer walls. It can be checked that in both cases the pressure returns to the initial reference pressure. Quantitatively, (1) the numerical and experimental pressure maxima are very close (5.45 bar absolute pressure, 79 psi, for test 31; 5.5 bar, 80 psi, for computation) (2) and numerical and experimental depressurization times (time for which all pressure profiles below +0.7bar) are agreement (73. 4 ms for test 31, and 59 ms for computation).

-5-

4 Congreso Uruguayo de Mantenimiento, Gestin de Activos y Confiabilidad - URUMAN 2008


Del 02 al 04 de Septiembre
4) Without prevention system In Figure 6, the results from computer simulations show the pressure evolutions for the same arcing conditions as those applied during CEPEL test 31 (see section C.1) but without the protection: after the arc feeding, the average pressure remains close to an equilibrium state equal to 7 bar (100 psi), much higher than the static withstand limit pressure. In these conditions, the transformer would have exploded as soon as the tank wall elasticity limits were over, i.e. as soon as the tank walls could not store any more mechanical energy due to the pressure increase. The simulation tool thus allows computing and accurately studying physical configurations which cannot be tested experimentally, and especially dangerous situations such as the case of transformer explosions. Figure 6 : Pressure when the tank is not protected

IV. NUMERICAL SIMULATION RESULTS: EXPLOSION PREVENTION ON A LARGE TRANSFORMER (400MVA) The simulations give results in accordance with the experimental data, for a relatively low cost and without any safety issue. They are thus used to compute the effects on a fast drain on the pressure distribution inside the tank.
Depressurisation set Depressurisation set activation

Electrical arc

V(m/s) P(bar)

Figure 7: Chronology of the Prevention Technology Operation up to 50 ms A 400 MVA transformer (7.8 m long and 4 m high) is considered with an electrical arc (11 MJ-arc ie generating about 3.3m3 of gas) occurring near a bushing, generating a 11 bar (160 psi) gas bubble. The picture sequence (Figures 7 and 8) allows following the pressure propagation inside the tank and the drain operation as soon as the first pressure peak has activated the depressurization set (4 ms after the arc occurrence, Figure 7). The drained oil velocity is represented by vectors which color accounts for the velocity magnitude, V(m/s).

-6-

4 Congreso Uruguayo de Mantenimiento, Gestin de Activos y Confiabilidad - URUMAN 2008


Del 02 al 04 de Septiembre

Figure 8: Inner Tank Pressure Evolutions a) with and b) without protection Protection Efficiency Illustration Following the idea of Kawamura [9], the drain gives place to the pressurized fluids so that after 120 ms, the pressure is back to safe levels (see Figure 8.a)). Otherwise, when the tank is not equipped with any protection system, and if it is subjected to a similar low impedance fault, the tank is still exposed to very dangerous pressure levels (up to 15 bars, 217 psi) after 120 ms (Figure 8.b)): without the tank protection, the static pressure stabilizes around 7.5 bars and the transformer explodes (about t=90ms). A technology based on a fast tank drain has thus a very positive effect on the tank protection. V. CONCLUSION A complete modeling for unsteady compressible two-phase flows has been adapted and presented in this paper to study the phenomena leading to the transformer tank rupture and to evaluate explosion prevention methods. This models main advantage consists in describing accurately the pressure wave propagation and its progressive interaction with the structure, physically highlighted by internal arcing experiments in oil-filled power transformers. These tests also showed experimentally the efficiency of an explosion prevention method. This one is based on the fast tank depressurization induced by the quick oil drainage out of the transformer. The oil drainage is triggered by the direct and passive mechanical response of a depressurization set to the pressure wave. The tests results were used to validate the computer simulation tool by comparison with experiments. Computer simulations were performed to check the prevention ability to efficiently depressurize larger transformer tanks (400MVA) when subjected to internal arcing of high energy level. VI. APPENDIX: TANK RUPTURE MITIGATION METHOD The explosion prevention technology studied in this paper is a passive mechanical technology that absorbs the high overpressures generated by the electrical arc, thus preventing the tank rupture and the subsequent fire. This fast-direct-tankdepressurization-based method activates as soon as the high pressure peak of the pressure wave reaches it. As shown in Figure 9, the main transformer tank, the On Load Tap Changers (OLTC) and the Oil Cable Boxes (OCB) can be protected.
(1) (2) (3)

(5) (4)

Figure 9: Transformer equipped with fast direct tank depressurization based method (TP) Indeed, the electrical fault generates a dynamic pressure peak, which travels at the speed of the sound inside the transformer oil, 1,200 Meter per second (4,000 feet per second). This first shock wave makes the rupture disc (located in 1) burst. Oil and gas are then quickly expelled out of the transformer tank through the decompression chamber (located in 2) to an oil gas separation tank (item 3) and the explosive gases are then channeled away to a remote and safe area. Then, nitrogen is

-7-

4 Congreso Uruguayo de Mantenimiento, Gestin de Activos y Confiabilidad - URUMAN 2008


Del 02 al 04 de Septiembre
injected (item 4) to have the whole transformer safe, cool and ready for repairs. Note that additional Rupture Discs can be placed in order to protect the OLTC or the OCB (item 5). VII. REFERENCES
[1] [2] [3] [4] [5] [6] [7] [8] [9] [10] [11] [12] [13] [14] G. ALLAIRE, S. CLERC & S. KOKH, A Five Equation Model for the Simulation of Interfaces between Compressible Fluids, Journal of Computational Physics, vol. 181, 2, pp. 577-616, 2002. M.R. BAER & J.W. NUNZIATO, A Two-phase Mixture Theory for the Deflagration to Detonation Transition in Reactive Granular Materials, Intern. J. of Multiphase Flow, vol. 12, 6, pp. 861-889, 1986. J.-B. DASTOUS, M. FOATA, A. HAMEL, Estimating Overpressures in Pole-Type Distribution Transformers Part II: Prediction Tools, IEEE Transactions on Power Delivery, Vol. 18, No. 1, 2003. M. FOATA, M. IORDANESCU, C. HARDY, Computational Methods for the Analysis of Explosions in Oil-Insulated Electrical Equipments, IEEE Transactions on Power Systems, Vol. 3, No. 1, 1988. S.K. GODUNOV, A Finite Difference Method for Numerical Computation of Discontinuous Solutions of the Equations of Fluid Dynamics, Math Sb., vol. 47, pp. 357-393, 1959. O.R. HANSEN, SEBK Project: Results from Phase 2 Laboratory Scale Experiments, Gexcon, May 2001. O.R. HANSEN, A. WIIK, B. WILKINS, SEBK project phase 4 & 5, Transformer explosion and suppression: Large scale hybrid hydrogen and transformer oil explosions with & without suppression, 2001. A.K. KAPILA, R. MENIKOFF & D.S. STEWART, Two-phase Modelling of Deflagration to Detonation Transition in Granular Materials: Reduced Equations, Physics of Fluids, vol. 113, 10, pp. 3002-3024, 2001. T. KAWAMURA, M. UEDA, K. ANDO, T. MAEDA, Y. ABIRU, M. WATANABE, K. MORITSU, Prevention of Tank Rupture Due to Internal Fault of Oil Filled Transformers, CIGRE, 12-02, 1988 W.R. MAHIEU, Prevention of High Fault Ruptures of Pole-Type Distribution Transformers, IEEE Transactions on Power Apparatus Systems, Vol Pas-94, 1975. G. PERIGAUD, H. CUNY, S. PRIGENT, P. MAGNIER, Transformer Explosion and Fire Prevention, Live Tests on Large Transformers: Analysis and Simulations, Australia, TechCon 2006. R. SAUREL, R. ABGRALL, A multi-phase Godunov Method for Compressible Multi-fluid and Multi-phase flows, Journal of Computational Physics, vol. 150, pp. 425-467, 1999 R. SAUREL, R. ABGRALL, A Simple Method for Compressible Multi-fluid Flows, SIAM J. of Scien. Comput., vol. 21, 3, pp. 1115-1145, 1999. E. TORO, Riemann Solvers and Numerical Methods for Fluid Dynamics, Press, ed. Springer Verlag, Berlin, 1997. Guillaume Perigaud (M06) holds a Diploma of Mechanical Engineering and a MSc degree in Fluid Mechanics and Transfers (Ecole Centrale Nantes, Nantes, France, 2000). He holds a PhD in Mechanics and Heat Transfers (Universit de Marseille I, France, 2003). As a IEEE member he is actively involved in the IEEE PES Power Transformer Subcommittee Task Force regarding Transformer Tank Rupture & Mitigation.

Sbastien Muller is a Transformer Protector Corporation (TPC) researcher. He holds a Diploma of Mechanical Engineering, and a MSc in Fluid Mechanics (ENSMA, Poitiers, France, 2002). He is Doctor in Fluid Mechanics (Universit dOrlans, France, 2007). He is an NFPA member, currently working on the development of a simulation tool that models the vaporization process due to the arc interaction with the liquid oil.

Gal de Bressy is a Transformer Protector Corporation (TPC) researcher. He holds a Diploma of Mechanical Engineering and a MSc in Mechanics (IFMA, Clermont Ferrand, France, 2005). He is currently working on the development of a 3D CFD tool in the field of transformer explosion and manages the Intellectual Property issues.

Ryan Brady (M07) is a Transformer Protector Corporation (TPC) researcher. He holds a BS in physics (SFASU, Nacogdoches, TX, 2002) and a MSc in physics (UNT, Denton, TX, 2005). He is currently working on the parallelization of 3D CFD tools in the field of transformer explosion

Philippe Magnier is the Transformer Protector Corporation (TPC) chairman. He is Doctor in Nuclear Physics (Universit Paris Orsay, 1974) and holds a M.B.A. (CPA, Paris, 1988).

-8-

Вам также может понравиться