Cast-resin transformers excel in their fire behavior - Siemens Power ...

Technical Paper ■ Author: Fritz Sorg, Rudolf Hanov, Heinz RaithelCast-resin transformersexcel in their fire behaviorPower Transmission and Distribution

Wherever distribution transformers are usedin locations very close to people, GEAFOL castresintransformers are the low-risk answer.Cast-resin transformers lack the limitations ofoil-insulated transformers but share their desirableproperties, such as operational reliabilityand a long service life. Today more than 80,000GEAFOL cast-resin transformers have alreadyproven themselves in supplying electric powerto subway tunnels, coal mines, airports, windpower systems and even on cruise ships and innuclear power plants. In many cases, the keyfactor prompting decision-makers to choose thecast-resin version was its superior fire behavior.The results of fire behavior tests at well-knowntest laboratories prove that GEAFOL cast-resintransformers – especially in comparison to oilinsulatedtransformers – are the best choicewhen it comes to fire behavior.Cast-resin transformersexcel in their fire behaviorFritz Sorg, Rudolf Hanov, Heinz Raithel*The GEAFOL cast-resin transformers which SiemensPower Transmission and Distribution (PTD) producesat its transformer plants are fire-retardant and selfextinguishing(Figure 1).If they are involved by an external fire, they do notsignificantly affect the course of the fire thanks totheir low fire load. Nor do the coils of the GEAFOLcast-resin transformer ignite from short-circuit arcscaused by internal defects. Moreover, no toxic componentsare produced, other than the type of flue gasescommon to fires. Siemens advises its customers forsafety reasons, especially with respect to fire behavior,to use GEAFOL cast-resin transformers in infrastructuralprojects such as department stores or officebuildings. What’s more, the operator benefits fromother advantages, such as superior economy andexcellent environmental compatibility.Dry-type transformers must always be specifiedaccording to their tested and proven Environmental,Climatic and Fire Category. GEAFOL cast-resin transformerscomply with Environmental Category E2,Climate Category C2 and Fire Category F1 – thehighest categories defined in IEC 60076-11, whichmeans that they deliver the highest level of operationalsafety.Figure 1: GEAFOL cast-resin transformer in the test bay at theSiemens transformer plant in Kirchheim/Teck: If GEAFOL castresintransformers are involved by an external fire, they don’tsignificantly affect the course of the fire thanks to their lowfire load. Nor do their coils ignite from short-circuit arcscaused by intrinsic flaws. Moreover, no toxic components areproduced, other than the type of flue gases common to fires.100%8863OilSiliconeoilAskarel17GJ1510.5Heat of combustionIf one compares the energy content liberated duringcomplete combustion of all insulating materials intransformers filled with mineral oil, silicone oil, orester liquid with that of cast-resin insulated distributiontransformers, the comparison invariably favorsthe cast-resin transformer. For example, the combustionenergy of a 630-kVA mineral oil transformer ismore than four times that of a GEAFOL cast-resintransformer (Figure 2).This comparison does not include characteristicssuch as flammability and combustion speed, butis shows that the fire load of cast-resin insulatedtransformers is substantially lower than that ofliquid-filled transformers.24GEAFOL4.000Figure 2: Combustion heat of the insulating materials ofcomparable 630-kVA transformers with different insulatingmaterials: The combustion energy of a 630-kVA mineral oiltransformer amounts to more than four times that of aGEAFOL cast-resin transformer.* Fritz Sorg, Dipl.-Ing., Manager of Research & Development,Rudolf Hanov, Dipl.-Ing., Project Manager of Research &Development and Heinz Raithel, Dipl.-Ing., Director Sales &Marketing of the Transformer Plant Kirchheim/Teck of theSiemens Group Power Transmission and Distribution (PTD).2

The design of the windings and the castresinmixture determine the fire behaviorThe fire behavior of the cast-resin transformers whichSiemens manufactures, for example, in Kirchheim,Germany is dictated chiefly by the winding arrangementbecause this has a crucial influence on theproperties of the cast resin mixes that can be used.One of the secrets of good or bad fire behavior of acast-resin transformer is related to the mixing ratioof the insulation, which consists of an environmentallycompatible and recyclable epoxy resin/quartzpowder mixture in the GEAFOL transformer. TheKirchheim plant produces the insulating material forits cast-resin transformers from a mixture of two-thirdsquartz powder and one-third epoxy resin to createtransformers that are on the safe side with respectto their subsequent fire behavior.This is confirmed by prestigious testing laboratorieswhere Siemens has its cast-resin transformers testedfor fire safety. As a case in point, in 2005 a 1500-kVAtransformer was extensively and successfully testedin the testing laboratory of CESI (Centro ElettrotecnicoSperimentale Italiano) in Italy – including teststo validate Fire Category F1 according to IEC 600076– 11 Section 28.3 (Figure 3).In this test, a replica of a transformer leg wassimultaneously exposed to:■ Flames from a trough filled with burning spirits■ Radiant heat from a vertical radiant heater wall(24 kW, 750 °C)The following test criteria were used:■ The maximum temperature of emitted gases andits variation over time■ The generation of smoke development (the meanvalue of the degree of light transmission from the20th to the 60th minute after the start of the testmust not be less than 20 %).The good results of the fire behavior tests wereachieved without the addition of flame-retardantmaterials such as halogens. Nor had any additivessuch as alumina trihydrate been used that impairmechanical strength.These tests had been preceded by earlier fire behaviortests for flue gases in the Allianz Technology Centerand by tests of the pyrolysis products of moldedcast-resin material used in GEAFOL transformers.The test results demonstrated that the transformerscould not self-ignite due to short-circuit arcs associatedwith intrinsic flaws. If they become involved inan external fire, they do not significantly intensifythe course of the fire thanks to their low fire load.The tests also demonstrated that any impairmentsor hazards due to heat and gases from the fire arelargely the same as in ordinary fires.These tests were designed to study the behavior ofGEAFOL cast-resin transformers in critical operatingconditions, for instance under the influence ofhighly energetic short-circuit arcs, or when thetransformer becomes involved in external fires. Particularemphasis was placed on testing for environmentalimpact as well as for potential toxicity of thegases from the fire. Another priority was protectionof bodies of water. A cast-resin insulated, dry-typetransformer is fully compliant with the relevant standards,such as those of VDE 0101 concerning fireprotection and protection of bodies of water. Sucha transformer is therefore the product of choice forapplications in which meeting these conditions isparamount.The consequences of fire impact also had to be researchedcarefully and expertly in order to rule outabsolutely any hazards such as those associated withPCB-filled transformers (PCB = polychlorinatedbiphenyls 1) ). The main criterion was the question ofwhether and under what conditions a cast-resintransformer can burn and what combustion productscan occur. Combustion is an autonomous processonce the source of ignition or fire has been removed,in other words, the process is determined by theenergetic balance of the reaction heat and the heatdissipation. Heat input from the flame causes theplastic to be heated and decomposed. The gaseousdecomposition products burn. It is the very natureof autothermal processes that without sufficientheat input from the combustion itself no sustainablecombustion is possible. If cast-resin molded materialsand windings embedded therein, or synthetic liquidsand plastic insulating materials, are exposed to flames,the heat input can cause the temperature to reachthe ignition level. In the presence of sufficient oxygeninput, gaseous products then ignite.Two conditions must therefore be met in orderto maintain a stable, self-sustaining combustion:First, the temperature of the material must riseto the ignition point, which is above 450 °C in theGEAFOL cast-resin molded material. Second, thecombustion must generate a sufficient quantityof heat to sustain itself.Figure 3 a + b: Very goodresistance to direct flameaction was certified:1,500-kVA GEAFOL castresintransformer under testat the CESI Test Institute inJuly 2005 before (above)and after (below) testingto verify compliance withIEC 600076 fire category F1.1) Note: Production of PCB-filled transformers is banned today but there are still some in operation subject to strict regulations.3

Realistic tests of the fire behaviorof cast-resin transformers[a][b]The experimental tests of the fire behavior of castresindry-type transformers were conducted underrealistic conditions on complete transformers in twodirections:■ The effect of high-power arcs on the interior andon the surface of the transformer■ The effect of adjacent fires on the transformerand an analysis of the gases from the fireStandard versions of 800-kVA transformers wereselected for these tests.The low fire load of GEAFOL cast-resin transformersstems from the fact that over 90 percent of theirweight is contributed by metallic materials such aselectric sheet, aluminum and steel, and less than tenpercent by insulating materials. Moreover, only abouthalf of the mass of the insulating materials is combustible,while two-thirds of the cast-resin insulationconsists of silicon dioxide (SiO 2 ) – in other words, ofvery fine quarz sand. The bottom line: Less than fivepercent of the total weight of a GEAFOL cast-resintransformer is contributed by combustible materials.Figure 4: GEAFOL cast-resin transformerin the short circuit arc test:a) Before the testb) During the testc) After the test[c]Fire behavior testsin the presence of arcsInternal transformer defects in cast-resin transformers– such as shorts between turns, coil shorts, rarelyflashovers (phase-phase or phase-ground) or disruptivedischarges from the higher-voltage winding tothe lower-voltage winding – can cause short-circuitarcs with temperatures of up to 10,000 °C that affectthe insulating materials. The duration of such arcs islimited by fusegear that is located upstream to thetransformer and typically has a break time of lessthan 0.3 seconds.To evaluate the behavior of the cast-resin transformerin the presence of arcs, arc tests were conducted onan 800-kVA GEAFOL transformer at the high-powertest station of Forschungsgemeinschaft für Hochspannungs-und Hochstromtechnik e.V., a testinglaboratory in Mannheim, Germany. The test conditionswere more demanding than conditions encounteredin actual practice. Tests 1 and 2 posed thefollowing conditions:■ Three-phase HV terminal short circuit,triggered by an igniter wire■ Short circuits of 0.5 second and2 seconds duration■ Transformers at operating temperature (windingsin the short circuit heated to nearly 100 °C)■ Short circuit power 150 MVA.Figure 4 shows the transformer with the igniter wirepoised directly beneath the delta connecting block,as well as the arc during the short circuit and thetransformer after the two-second short circuit.Exposed to the heat of the short-circuit arc, some ofthe resin components of the cast-resin molded materialwere consumed by fire in a thin surface layer.What remained in place there was a quartz powder4

LV windingwith axialduct[a]structure and traces of carbon black, with the quartzpowder layers created by the fire providing a protectiveeffect for the deeper resin layers. At the basepoints of the arc on the metallic terminals, the conductormaterial partially melted and vaporized. Oncethe arc collapsed, no residual burning of insulatingmaterials was visible. This was also confirmed by filmsmade with high-speed cameras. Despite the visiblesurface damage, the transformer remained functional.HV6451327231456Following the two arc test, the transformer was subjectedto two additional tests: Holes were drilled intothe HV windings of all three phases, then six-millimeter-thicknails were inserted and connected toigniter wire, with the object of creating direct shortcircuits at individual turns and windings. The resultwas the same: Although on phase U a portion of aHV coil was actually blown off by the extremely highforces of the short circuit, no ignition or subsequentburning of the cast-resin molded material or otherinsulation portions was induced, even by these rigorousmeasures.Fire behavior tests with woodand propane fueled fireAs early as in 1983, fire behavior tests had beenconducted with an 800-kVA GEAFOL cast-resin transformerin the fire test building at the Allianz Centerof Technology in Ismaning, near Munich. To studythe fire behavior, the test objects were exposed totwo types of fire: external fires fueled by wood andexternal fires fueled by propane gas. These directexposures to different external fires were designedto emulate the exposure of transformers to differentpatterns of flames. The wood-fueled fire – set up onthe floor of the fire test building – exposed the testtransformer on a relatively broad front, but its impacton the insulating materials of the windings was partiallyimpeded by local shielding provided by the ironcore and portions of the frame. On the other hand,the flames of the propane gas burner exposed thelower area of the windings directly and unimpeded.The second test setup exposed the transformer tothe most severe test. Though the overall extent andscope of the fire was greater with the wood fire thanin the propane fire, the fire damage was clearlygreater with the propane gas. The transformer wasfitted at the core, LV- and HV-windings with a totalof 16 high-temperature resistant nickel chrome-nickelthermocouples, in a symmetrical arrangement ofeight thermocouples each for both tests. The firewoodused for the external wood fire consisted of ten kilogramsof untreated fir wood, stacked bonfire-fashiondirectly under the W transformer leg and ignitedwith the aid of excelsior. The calorific value of theapplied amount of fir wood amounted to about188 MJ and temperature of the flames ranged upto 1000 °C.In the setup with the propane gas fire, eight propanegas burners with a broad flame pattern were arrangedequidistantly around the base so as to evenly flamethe underside of the HV winding of the U leg. The560°C480400320240160800720°C640560480400320240160▲Limb UGas external fireExposure to flameExposure to flameLimb V[b]Figure 5: Temperature-time relation at the test transformer during the course of the fire:a) Location of the thermocouples at the transformerb) Test 1 with external wood firec) Test 2 with external propane gas firedistance of the burner from the lower edge of thecoil was about 60 millimeters, the flame orientationabout 45°, and the flame temperature ranged up to1200 °C. The external flame was maintained for30 minutes. The measured gas consumption duringthat time amounted 2,400 grams. At a specificcalorific value of 46,340 kJ/kg this corresponds toa total calorific value of about 111 MJ. The variationof the temperature over time at the windings of thetransformer reflects the course of the fire with thewood setup and the propane gas setup, respectively(Figure 5).▲Limb WWood external fire54 Measuring13 points6 Limb W2787 Room temperature, ceiling – 8 Room temperature, wall[c]Flame extinguishesafter approx.50 min.807 Limb V8 Roomtemperature,0ceiling0 5 10 15 20 25 30 35 40 45 50 min 60t541362MeasuringpointsLimb U5

[a]Under the prolonged effect of the high temperaturesof the external fire, the insulating materials of theLV and HV windings ignited, and due to the drafteffect of the axial channel in the LV winding andof the leakage channel, more intense combustionwith a visible flame occurred above the transformer(Figure 6). Noteworthy was the fact that the outersurface of the HV winding was set afire only in theimmediate area of the external fire, and that the firespread only minimally to the adjacent transformer leg.[b][c][d][e]Figure 6: Combustion sequence during exposureto flames of the external wood fire:The outer surface of the HV winding was ignitedonly in the immediate area of exposure to theexternal fire, and hardly any fire spread to theadjacent leg.[f]6

Figure 7 shows the fire behavior of the transformerin a temporal sequence during the test with the externalpropane gas fire. The fire and the visible flamewere more intense than in the wood fire. Nevertheless,the flame action diminished immediatelyafter the propane gas burner was shut off, and selfextinguishedcompletely in about 20 minutes.Burning portions of the insulation self-extinguishedafter removal of the external energy input. The hardpapercomponent of the support blocks continuedto burn with small flames. Surfaces exposed directlyto the flames were severely affected, while adjacentportions of the insulating material remained largelyundamaged. Even in the leakage channel, the firedid not spread horizontally – despite the extremelyabundant input of oxygen (Figure 9).In Figure 9 is it is clearly evident that the aluminumof the LV winding of the U leg has been heated to alevel exceeding the melting temperature. In addition,the overall view of the HV windings shows that thefires at the outer legs did not spread to the centerleg in substantial degree (Figure 8).Figure 8: Condition of the test object after the firebehavior test: The overall view of the HV windingsalso shows that the fires at the outer legs did notspread to the center leg in substantial degree.[a][b]Figure 9: LV windings after the fire behavior test:Surfaces exposed directly to the flames wereseverely affected, while adjacent portions of theinsulating material remained largely undamaged.Even in the leakage channel the fire was did notspread horizontally – despite the extremelyabundant input of oxygen.[c][d]Figure 7:Combustion sequence during exposure by the external propane gas fire.a) After six minutes b) After eight minutesc) After 20 minutes d) After 33 minutes7

Flue gas:sample collection and analysis resultsThe fire room where the tests were conducted hasa floor space of four by four meters and a height of4.1 meters. It was ventilated by fresh air blownthrough one firewall, with the flue gas exhaustedthrough opposite firewall. This is where the 3-cmthickpipe for the collection of the flue gas sampleswas located. The pipe ran up vertically the entireheight of that wall and was equipped at 50-cm intervalswith apertures about 10 square centimeters insize to admit the flue gas. Located in these apertureswere storage filters. The flue gases were extracted bya membrane pump, and pumped at a rate of 30 litersper minute directly to the gas chromatography massspectrometer. The line was about 25 meters long.During the measurement, the line and the membranepump were heated to 200 °C. Ten microlitersof flue gas were collected at 20-second intervalsthrough a split system and a gas metering device foronline measurement. A capillary gas chromatographinstalled here was used only as an interface for themass spectrometer, not for the preliminary separationof the combustion gases. At approximatelyone-second intervals, mass spectra in the rangefrom m/e = 12 to m/e = 650 were obtained andstored (m/e: mass-charge ratio).Concurrently with the online registration of the massspectra, gas samples were withdrawn for the quantitativeanalysis of the flue gas components (gas collectingtubes). Table 1 summarizes the mass-spectrometricresults of the combustion gas analyses andshows the qualitative composition of the combustiongases from the online analysis as a relative functionof the maximum peak CO 2 intensity = 100 %.m/e Molecular or fragment ion Isotope ratio Maximal peak intensity 2 ) %Wood fireGas fire22 CO2+2 Carbon dioxide, double charged 0.27 0.3425 C 2 H + Fragment ion of azetylene 0.025 0.0326 C 2 H+2 Azetylene 0.18 0.1744 CO+2 Carbon dioxide12C 10016O 100 100 1004513CO+2 Carbon dioxide isotope13C 1.12 1.81 2.17C 2 H 5 O +Fragment of Ethanol4612C 16 O 18 O Carbon dioxide isotope18O 0.204 0.65 0.75C 2 H 6 O +Ethanol50 C 4 H+2 Aromatic fragment 0.02 0.0278 C 6 H+6 Benzene 0.06 0.0691 C 7 H+7 Toluene fragment 0.05 0.0492 1) C 7 H+8 Toluene 0.03 0.02Table 1: Mass spectroscopy of the combustion products of a cast-resin transformer1) For m/e larger than 92, no data that meet the specified premises2) CO 2 peak set = 100 %8

The course of the fire can be tracked by the intensityof the molecular ions of carbon dioxide CO 2+(masschargeratio m/e = 44; Figures 10 and 11). Carbondioxide is therefore used as the target gas for theevaluation of the course of the fire.0.30%0.1500.10%0.0500.20%0.100100%500CO 22+m/e = 22C 2 H +m/e = 25C 2 H 2+•m/e = 26CO 2+•m/e = 440.30%0.1500.10%0.0500.20%0.100100%500CO 22+m/e = 22C 2 H +m/e = 25C 2 H 2+•m/e = 26CO 2+•m/e = 441.80%0.9000.6%0.3013CO 2+•+C 2 H 5 O +m/e = 45[C 18 O 16 O] + •C 2 H 5 OH +•m/e = 461.80%0.9000.6%0.3013CO 2+•+C 2 H 5 O +m/e = 45[C 18 O 16 O] + •C 2 H 5 OH +•m/e = 460.10%0.0537CI + m/e = 370.10%0.0537CI + m/e = 37000.10%0.05C 4 H 2+m/e = 500.10%0.05C 4 H 2+m/e = 50000.10%0.05C 6 H 6+•m/e = 780.10%0.05C 6 H 6+•m/e = 7800Peak intensity0.10%0.0500.10%0.050C 7 H 7+m/e = 91C 7 H 8+•m/e = 92500 1000 1500 2000scanPeak intensity0.10%0.0500.10%0.050C 7 H 7+m/e = 91C 7 H 8+•m/e = 92500 1000 1500 2000scan08.45 17.30 26.15 min 35.00t08.45 17.30 26.15 min 35.00tFigure 10: Mass chromatogram during exposure to the external wood fire(Scan 1 TO 2278. During this test 2,278 gas samples were obtained)Figure 11: Mass chromatogram during exposure to the externalpropane gas fire (Scan 1 TO 3534. During this test 3,534 gas sampleswere obtained)9

An exposure of the transformer to external flamescan only generate the kind of molecular ions andfragment ions that exhibit behavior analogous tothat of CO 2 in the mass chromatogram, in otherwords, ions whose concentration increases afterignition of the combustion. On the other hand,the 37 CL + ion for example appears even in the initialmass spectra: Its concentration is independent ofthe course of the fire. That leads to the conclusionthat chlorine ions do not originate from combustiongas components but from chlorine-containing contaminantsin the air.Table 2 shows the results of the quantitative analysisof the gas samples from the gas collecting tubes. Asexpected, during the exposure to the external propaneflames an increased proportion of carbon dioxidewas observed, and otherwise only small amounts ofazetylene, besides the components of the air. Thecombustion gases during the exposure to the externalwood fire, on the other hand, also contained ethylene,benzene, toluene and ethyl benzene. However, atleast benzene and toluene could not be caused bythe wood fire alone, since both these componentswere also shown to be present in the online tests ofthe combustion gases during exposure to the externalpropane flames. The carbon monoxide content wasdetermined through gas chromatography. The highestcarbon monoxide content found in the tested gassample from a gas collecting tube was 0.044 milliliterper 100 milliliters.ComponentWood fireGas collecting tube no.Propane gas fire1 3 6 8 9 11 14 16Water 0.17 – – – – – 0.13 –Azetylene – 0.003 0.002 0.002 0.003 0.005 0.004 –Nitrogen 82.6 78.9 78.9 75.0 79.4 79.6 79.3 79.7Carbon monoxide 0.003 0.01 0.012 0.08 – 0.012 0.013 0.003Ethylene 0.005 – 0.005 – – – – –Oxygen 15.4 18.8 18.9 23.4 19.4 18.1 17.9 19.2Argon 1.29 0.95 0.99 1.19 0.92 0.90 0.92 0.92Propene 0.001 – – – – – – –Carbon dioxide 0.56 1.40 1.29 0.43 0.29 1.36 1.67 0.15Butene 0.0002 – – – – – – –Benzene 0.002 0.001 0.003 – – – – –Toluene 0.001 0.001 0.002 – – – – –Ethylbenzene 1) 0.0009 – – – – – – –Table 2: Concentration of combustion gas components in the gas collecting tubes (evacuated glass vessels), stated in ml per 100 ml1) No larger molecules than ethyl benzene were detected.10

Combustion products“under the looking glass”The measurements showed that the GEAFOL castresinmolded material contains very small quantitiesof chorine-containing compounds as impurities thatresult from the production of the epoxy resin by theconversion of bisphenol A with epichlorohydrin.To make absolutely certain that the combustionproducts do not contain the toxic substances2,3,7,8-Tetrachlorodibenzodioxin (TCDD) and2,3,7,8-Tetrachlorodibenzofuran (TCDF) which canoccur during pyrolysis of askarel, the resin moldingcompound was burnt in the Bayer-ICI-Shell apparatus(pyrolysis temperature 600 °C) and the combustionproducts examined for presence of these two substances.This was done by capturing the combustion productsin a filter and in methanol as an absorption solution.The extract from the filter was combined with theabsorption solution, known quantities of 2,3,7,8-TCDD and 2,3,7,8-TCDF were added to part of thissolution which was examined like the second halfof the test solution using gas chromatography. Thechromatograms are shown in Figure 12. They showthat 2,3,7,8-TCDD and 2,3,7,8-TCDF are not containedin the combustion products. The detection limit is0.05 µg/g.The only relevant toxic component generated by theburning of the cast-resin transformer appears to becarbon monoxide. But carbon monoxide is a gas thatis commonplace in fires. A laboratory test demonstrated,on the other hand, that the two highly toxicsubstances, namely 2,3,7,8-Tetrachlorodibenzodioxinand 2,3,7,8-Tetrachlorodibenzofuran, are not formedin the combustion of a cast-resin transformer.The results of these extensive tests, then, supportthe conclusion that the operation of GEAFOL castresintransformers in electrical systems does notcreate any risks of significantly intensifying a fireor risks of toxicity that exceed the normal risk inresidential or industrial fires.What’s more, the excellent fire behavior of GEAFOLcast-resin transformers is not only validated by tests:GEAFOL transformers, which Siemens manufacturesin the range from 50 kVA to 40 MVA, have proven bythe thousands, in many years of worldwide use, thatthey work safely, economically and reliably, evenunder difficult conditions.a)Comments on the resultsThe tests with short-circuit arcs induced by an externalsource of ignition on the outer surface of theGEAFOL transformers as well as within demonstratedthat the cast-resin transformer cannot be caused toignite even by extreme exposure to highly energeticarcs. However, an external fire that spreads to thetransformer can ignite the insulating materials. Butwhen the ignition source is removed, the resultingflames self-extinguish without significant horizontalpropagation of the combustion. It can be concludedthat the cast-resin, dry-type transformer cannot selfignitedue to highly energetic arcs, and that the courseof the fire, when the transformer becomes involvedin an external fire, does not intensify significantly,i.e. no more than due to the low fire load. In thecombustion gases formed by the burning of this transformeronly benzene, toluene and methylbenzenewere demonstrated besides carbon monoxide –but no typical, specific components that could beattributed to the cast-resin mixture.b)Retention time of 2,3,7,8 t.c.d.fRetention time of 2,3,7,8 t.c.d.d2,3,7,8 t.c.d.f (1.4 ppm added)2,3,7,8 t.c.d.d (0,22 ppm added)The toluene and methylbenzene contents amountedto ten percent of the then applicable maximum workplaceconcentration (MAK value) – in the safe rangeif one considers that in this case the MAK valuesare being compared to short-time effects during thecourse of the fire. Even four times the technicalguideline concentration (TRK value) of benzene isnot hazardous in this context. During the new testsin 2005, the measured emission values were farbelow the then permissible workplace limits (whichreplaced MAK in that year).0 10 20 30 40 50 min. 60tFigure 12: Gas chromatograms of the products of combustion from the resin mouldingcompound (Bayer-ICI-Shell apparatus; pyrolysis temperature 600 °Ca) from cast resin (aliquot par after sample evaluation)b) from cast resin (aliquot par after sample evaluation, with addition of 1.4 ppm of2,3,7,8 tetrachlorodibenzofuran and 0.22 ppm of 2,3,7,8 tetrachlorodibenzodioxinbefore sample evaluation).11

Siemens AGPower Transmission and DistributionTransformers DivisionHegelstr. 2073230 Kirchheim/TeckGermanyOrder No. E50001-U413-A105-X-7600Printed in GermanyDispo 19201TH 101-060836 102060 WS 04070.5www.siemens.com/energyThe information in this document contains general descriptions of the technical options available, which do not always have to be present in individual cases.The required features should therefore be specified in each individual case at the time of closing the contract.