Electricity Today38

1 INTRODUCTIONThe study of the “Power Plants

Transformer Explosion and FirePrevention” started on request of a largewestern U.S. utility which had experi-enced a severe hydro power plant trans-former explosion. This incident occurredunderground, destroying one transformerand damaging another nearby. This nowcommon event lead to a four-month totalpower plant shutdown and 10 monthsunit unavailability.

Several on-line monitoring systemswere installed on the transformers to tryand anticipate another failure. The majorconcern was that this monitoring mightnot prevent an explosion, so SERGI wascontacted to avoid a second catastrophicincident, should another transformer fail.

SERGI specializes in transformervessel explosion and fire prevention. TheDesign Department has conductedresearch since 1995 to understand andquantify the energetic transfer phenome-non that takes place within a vessel dur-ing a short circuit, which has led to theMagneto-Thermo-Hydrodynamic model(MTH) development.

SERGI decided to study the preven-tion of power plant transformer explo-sions, because of the harsh arcing thatcan occur. During short-circuit, trans-former faults are fed for seconds due tothe generator inertia. Therefore, design-ing a system able to prevent a powerplant transformer explosion fully satis-fies the explosion prevention of any otherkind of incident for transmission or dis-tribution transformers.

The purpose of these studies was todetermine an efficient way to depressur-ize the transformer vessel during andafter these short-circuit conditions inorder to avoid explosion and the resultingoil fire. SERGI extensively investigated

the event, simulating causes, calculatingthe transformer vessel pressure rise, andproposed a suitable solution. The aimwas therefore to design a system capableof preventing the worst-case scenario inorder to satisfy the prevention of anyother kind of incident.

As in every case already studied bySERGI, the results brought to pressurerise with slopes from 60 to 900 bars persecond.

The fault was fed both by the gener-ator and the network. As a consequenceof the ongoing exponential gas produc-tion due to the continuous transformerenergization from the generator side, cal-culations have shown an oil plug phe-nomenon which appeared in theDepressurization Chamber after systemoperation, leading to another pressurerise. Therefore, more calculations wereconducted in order to size and shape aDepressurization Chamber that wouldeliminate this occurrence. The pressureinside the transformer will, therefore, bekept below the vessel maximum toleratedpressure during transformer short-circuit.

Other aspects were also examined,

such as the oil-gas mix exhaust speed andflow through the DepressurizationChamber, as well as the volume of explo-sive gas produced and the volume of oilto expel, in order to safely depressurizethe transformer.

The resulting DepressurizationChamber diameter required to avoid ves-sel explosion was determined to be 12inches. It was calculated that the SERGITransformer Protector Rupture Disk usedfor depressurization fully opened in 0.6millisecond for a 236 kA short-circuit.

With other researches already pub-lished by SERGI, along with the compa-ny experience in Transformer Explosionand Fire Prevention, the best-adaptedsolution has been designed for powerplants. In addition to the vessel, theTransformer Protector is also conceivedto avoid explosion and fire of all trans-former oil capacities such as Bushing,Oil Bushing Cables Boxes and On LoadTap Changers.

The study presented here, which isthe result of twenty-five months of

FALL TRANSFORMER REPORT

STUDY AND DESIGN OF POWER PLANTTRANSFORMER EXPLOSION AND

FIRE PREVENTION - PART IBy David Scheurer, Research Department Engineer; Alexis Nesa, Research Department Engineer;Sylvain Prigent, Research Department Engineer; Philippe Magnier, C.E.O, SERGI; Moumèn Darcherif,Research Manager, EPMI

Figure 1: Transformer winding geometry

Continued on Page 40

ET October 9/26/07 9:40 AM Page 38

ET October 9/26/07 9:40 AM Page 39

research, includes the fault calculation, energetic transfer in theoil, the vessel pressure increase, the amount of gas produced,the fluids evacuation and the associated SERGI TransformerProtector design.

2 SINGLE-PHASE TRANSFORMER CHARACTERISTICS ANDSIMPLIFIED GEOMETRY DESCRIPTION

The studied transformer is single-phase, step up, locateddownstream of a 450 MVA generator.• Nominal Power: 150 MVA• Primary voltage: 18 kV delta• Secondary voltage: 133 kV L-N• Impedance: 15.5 %• Cooling: Forced Oil, Water cooled• Dimension: 121” diameter, 156” high

The physical and electromagnetic material parameters, HVand LV windings were taken into account as well as the influ-ence of paper, paperboard insulation, thermal and hydraulicconcerns. The heat transfer between the transformer and theoutside or the magnetic circuit was imposed, as conditions tothe limit.

Oil is injected at the vessel bottom at 20° Celsius, 68°Fahrenheit and extracted at the top.

3 FAULT SCHEME AND CALCULATION3.1 Calculation

The method used to calculate the fault current is theSymmetrical Component Method, based on the fragmentationof the problem in two major parts, each subdivided as follows:

1. Direct system study• Thevenin for the direct fault calculation;• Reverse scheme;• Homopolar impedance calculation.2. Fault equation settling and resolution

3.2 Fault SchemeThe problem resolution is detailed in reference. The fault

current was calculated with the symmetrical componentmethod and gave the value of 11.8kA during the incident.

However, the utility has required SERGI to simulate the casewith 34.5kA, 118kA and 236kA faults in order to oversize theresulting transformer protection.

4 SIMULATION4.1 Method

The simulation purpose is to formulate a coupling of elec-tromagnetic, thermal and hydrodynamic phenomena. It wastherefore necessary to:

• Determine the magnetic field created by the inductanceand/or arc in the surrounding field versus the injected currentper phase;

• Calculate the induced currents and the Joule and Eddycurrent local dissipated power;

• Calculate the temperature by using the resulting abovevalues as heat sources.

The calculation is done through four sub models. The firstequation to solve is magneto-dynamic:

It derives from the Maxwell equations, which rule theoverall electromagnetic phenomena.

The magnetic field is expressed through the potential vec-tor:

The thermal sub model resolves the partial derivative equa-tions in a Cartesian geometry and is the following:

Where:

Electricity Today40

Transformer StudyContinued from Page 38

Figure 2: Short circuit scheme

ET October 9/26/07 9:40 AM Page 40

The hydrodynamic sub model resolves Navier-Stokesequations and mass conservation equations as:

The kinetic submodel is ruled by the Arrhenius Law:

Where: A : Pre-exponential factorEa : Activation energyR : Perfect gas constantT : Temperaturek : Constant expressing the quickness

of the reactionThe pre-exponential factor of the expressed Arrhenius Law

is determined experimentally. The data are gathered accordingto each type of gas and associated activation energy:

• The oil-gas composition is identified for every tempera-ture;

• The different element concentration versus time and tem-perature in mineral oil environment is known; this factor is partof the software coded by SERGI.

4.2 Transformer coolingCalculations were conducted with the hypothesis of forced

and directed oil circulation. The magnetic core was made freein temperature.

4.3 Steady stateBefore simulating the disruptive failure in the transformer,

the steady state was simulated in order to obtain a basis for therest of the calcula-tion.

Steady state cal-culation enables toobserve the trans-former behaviourunder nominal elec-trical load. It corre-sponds to the normaltransformer operationat normal tempera-ture and pressure lev-els. Figure 3 showsthe heat dissipation inthe right side of thetransformer winding.

4.4 Transient stateThe calculated conditions can be worsened in the case of

maneuver shocks, power surges, etc. In this case, the transientstate simulations were conducted under three different faults:

• Maneuver shock at 35.4kA• Lightning surge at 118kA• Lightning shock at 236kAWe made the hypothesis that no generator circuit breaker

would operate during the fault. The fault is therefore simulatedall through the depressurization process.

The electrical arc is simulated in the transient state byinserting a copper wire fed by the above given currents. Oncethe wire reaches the temperature of electrical arcing, oil crack-ing process, gas production and pressure rise increase severely.

5 VESSEL PRESSURE RISE DURING SHORT-CIRCUITS5.1 Results

Along with the calculation of the evolution of temperature,an associated pressure is calculated for each type of fault.Figures 4 to Figure 6 below show the calculated pressure risecurves versus time. Yellow circles on the figures indicate arcignition. Red lines represent the vessel tolerance to pressure.The lap time before arcing is due to the copper wire heating,creating the favourable arc plasma.

41October 2007

Figure 3: Vessel temper-ature variation in steadystate, in Celsius andFahrenheit

Figure 4: Transformer vessel pressure evolution after a 35.4 kAshort-circuit

Figure 5: Transformer vessel pressure evolution after a 118 kAshort-circuit

Figure 6: Transformer vessel pressure evolution after a 236 kAshort-circuit

Look in the November/December issue of Electricity Today forthe conclusion of this article.

ET October 9/26/07 9:19 AM Page 41