Documenti di Didattica
Documenti di Professioni
Documenti di Cultura
INDUSTRIALES Y DE TELECOMUNICACIN
UNIVERSIDAD DE CANTABRIA
GRADUADO EN INGENIERA EN
TECNOLOGAS INDUSTRIALES
A mi familia.
Grado en Tecnologas
Industriales
ETSIIT Santander
1
Curso 2016/2017
Francisco Jos Iglesias Ruiz Anlisis numrico del comportamiento fluido-trmico-
magntico de un transformador monofsico
INDICE GENERAL
Grado en Tecnologas
Industriales
ETSIIT Santander
2
Curso 2016/2017
Francisco Jos Iglesias Ruiz Anlisis numrico del comportamiento fluido-trmico-
magntico de un transformador monofsico
ndice
1.INTRODUCCIN .................................................................................................... 9
2.ALCANCE ............................................................................................................. 11
3.ANTECEDENTES ................................................................................................. 12
4. TRANSFORMADORES, ASPECTOS CONSTRUCTIVOS ................................. 14
4.1 DEFINICIN ................................................................................................... 14
4.2 ESQUEMA DE UN TRANSFORMADOR ........................................................ 14
4.3 FUNCIONAMIENTO ....................................................................................... 16
4.4 PRDIDAS EN UN TRANSFORMADOR........................................................ 17
4.5 RENDIMIENTO ............................................................................................... 17
4.6 CALENTAMIENTO DE UN TRANSFORMADOR ........................................... 18
5. REFRIGERACIN DE TRANSFORMADORES .................................................. 20
5.1 INTRODUCCIN ............................................................................................ 20
5.2 NOMENCLATURA .......................................................................................... 21
5.3 MTODOS DE CONSERVACIN DEL LQUIDO DIELCTRICO ................. 22
6. ENSAYOS EN UN TRANSFORMADOR DE POTENCIA .................................... 25
6.1 ENSAYO DE VACO ....................................................................................... 25
6.2 ENSAYO DE CORTOCIRCUITO .................................................................... 26
7. MODELO FSICO DEL TRANSFORMADOR ...................................................... 28
7.1 AUTODESK INVENTOR ................................................................................. 28
7.2 DISEO DE ELEMENTOS ............................................................................. 28
7.3 CREACIN DE ENSAMBLAJES .................................................................... 30
7.4 PLANOS Y VISTAS DE PIEZAS Y ENSAMBLAJES ...................................... 31
8. MODELO DE SIMULACIN ................................................................................ 32
8.1 INTRODUCCIN ............................................................................................ 32
8.2 INTERPRETACION DEL MODELO ................................................................ 32
8.3 GEOMETRA DEL TRANSFORMADOR......................................................... 33
8.4 SIMPLIFICACIONES DEL MODELO GEOMETRICO .................................... 33
8.5 PRINCIPIOS FSICOS .................................................................................... 37
8.6 MODELO FSICO-MATEMATICO .................................................................. 38
8.7 PROPIEDADES FISICAS DE LOS MATERIALES ......................................... 40
8.7.1 Propiedades del aceite mineral................................................................. 40
Grado en Tecnologas
Industriales
ETSIIT Santander
3
Curso 2016/2017
Francisco Jos Iglesias Ruiz Anlisis numrico del comportamiento fluido-trmico-
magntico de un transformador monofsico
Grado en Tecnologas
Industriales
ETSIIT Santander
4
Curso 2016/2017
Francisco Jos Iglesias Ruiz Anlisis numrico del comportamiento fluido-trmico-
magntico de un transformador monofsico
Grado en Tecnologas
Industriales
ETSIIT Santander
5
Curso 2016/2017
Francisco Jos Iglesias Ruiz Anlisis numrico del comportamiento fluido-trmico-
magntico de un transformador monofsico
ndice de figuras:
Figura 6.1. Circuito equivalente de un transformador visto desde el lado del primario.
................................................................................................................................. 25
Figura 6.2. Esquema de ensayo de vaco. ............................................................... 26
Figura 6.3.Esquema de ensayo de cortocircuito. ..................................................... 27
Grado en Tecnologas
Industriales
ETSIIT Santander
6
Curso 2016/2017
Francisco Jos Iglesias Ruiz Anlisis numrico del comportamiento fluido-trmico-
magntico de un transformador monofsico
Grado en Tecnologas
Industriales
ETSIIT Santander
7
Curso 2016/2017
Francisco Jos Iglesias Ruiz Anlisis numrico del comportamiento fluido-trmico-
magntico de un transformador monofsico
Grado en Tecnologas
Industriales
ETSIIT Santander
8
Curso 2016/2017
Francisco Jos Iglesias Ruiz Anlisis numrico del comportamiento fluido-trmico-
magntico de un transformador monofsico
1.INTRODUCCIN
El punto de partida de este proyecto se basa en los resultados obtenido en un ensayo
de laboratorio. En este ensayo se ha diseado un transformador monofsico de
relacin de transformacin 110/220 y potencia nominal 0.8 KVA sumergido en aceite
mineral. El transformador es colocado en el interior de una cuba de acero inoxidable.
Con el fin de monitorizar el comportamiento de la mquina, cuatro termopares
ubicados en el fondo y la parte superior de la cuba, en el devanado externo y en el
ncleo. Estos termopares permiten ver la evolucin de las temperaturas tanto en el
aceite, como en la parte slida.
Las medidas captadas por estos sensores son capturadas cada 10 segundos por una
tarjeta de captura de datos y tratados informticamente. Dicho tratamiento consiste
en la representacin grfica de las temperaturas con el tiempo. Para poder obtener
informacin precisa y fiable es necesario estudiar el transformador en rgimen
permanente. Para ello se utilizar una carga resistiva de potencia igual a la nominal
del transformador (0.8 kW). Cuando ests temperaturas alcanzan un valor constante
(rgimen permanente en unas seis horas) se pueden tomar los valores obtenidos del
transformador baado en aceite mineral.
Grado en Tecnologas
Industriales
ETSIIT Santander
9
Curso 2016/2017
Francisco Jos Iglesias Ruiz Anlisis numrico del comportamiento fluido-trmico-
magntico de un transformador monofsico
Grado en Tecnologas
Industriales
ETSIIT Santander
10
Curso 2016/2017
Francisco Jos Iglesias Ruiz Anlisis numrico del comportamiento fluido-trmico-
magntico de un transformador monofsico
2.ALCANCE
Las conclusiones de este proyecto pretenden servir como punto de partida a la hora
de comparar la capacidad refrigerante de un aceite mineral (caso base) y un aceite
basado en esteres naturales (aceite alternativo). Adems, ser posible obtener
aproximaciones o datos para cualquier punto de trabajo del transformador, as como
para diferentes temperaturas del aceite.
Grado en Tecnologas
Industriales
ETSIIT Santander
11
Curso 2016/2017
Francisco Jos Iglesias Ruiz Anlisis numrico del comportamiento fluido-trmico-
magntico de un transformador monofsico
3.ANTECEDENTES
Este proyecto est basado, principalmente, en dos artculos publicados acerca de la
refrigeracin de un transformador. En ambos artculos se expone la idea de refrigerar
el transformador por medio de un fluido magntico el cual origina ciclos de conveccin
y mejora la capacidad de enfriamiento del refrigerante.
Los resultados obtenidos muestran unos ciclos de conveccin internos, los cuales son
mucho ms pronunciados significativos en el ensayo con ferro fluido. Esto permite
una mejor refrigeracin del transformador.
Grado en Tecnologas
Industriales
ETSIIT Santander
12
Curso 2016/2017
Francisco Jos Iglesias Ruiz Anlisis numrico del comportamiento fluido-trmico-
magntico de un transformador monofsico
Utilizando el refrigerante cargado con las partculas magnticas obtiene que la vida
til del transformador trabajando con sobrecarga en torno al 160% (necesario para
lograr temperaturas prximas a la temperatura de Curie) es 9 veces superior a las
expectativas de ese transformador refrigerado por ese mismo fluido sin las partculas
magnticas.
Grado en Tecnologas
Industriales
ETSIIT Santander
13
Curso 2016/2017
Francisco Jos Iglesias Ruiz Anlisis numrico del comportamiento fluido-trmico-
magntico de un transformador monofsico
4.1 DEFINICIN
En cuanto al ncleo del transformador, est constituido por chapas muy finas de hierro
o acero apiladas y recubiertas de una capa de aislante, denominada carlite y de
espesor nfimo, que reduce las perdidas magnticas (en el hierro). Estas chapas
suelen tener de forma de doble U, de manera que los devanados se montan en torno
a las columnas, dejando libre las secciones de unin entre columnas, denominadas
culatas.
Grado en Tecnologas
Industriales
ETSIIT Santander
14
Curso 2016/2017
Francisco Jos Iglesias Ruiz Anlisis numrico del comportamiento fluido-trmico-
magntico de un transformador monofsico
Los devanados forman el circuito elctrico del transformador. Estn formados de hilos
de cobre de seccin circular en transformadores pequeos y de seccin rectangular
en grandes transformadores de potencia. Estos conductores estn aislados, en el
caso del transformador a estudiar posteriormente por un barniz que evita que los
devanados de alta y baja tensin, entren en contacto entre s. En el caso de
transformadores acorazados, los devanados se colocan de forma concntrica en
torno al ncleo, quedando en la parte interna el devanado de baja tensin en el
interior, ya que es ms fcil de aislar que el devanado de alta tensin.
Grado en Tecnologas
Industriales
ETSIIT Santander
15
Curso 2016/2017
Francisco Jos Iglesias Ruiz Anlisis numrico del comportamiento fluido-trmico-
magntico de un transformador monofsico
4.3 FUNCIONAMIENTO
Grado en Tecnologas
Industriales
ETSIIT Santander
16
Curso 2016/2017
Francisco Jos Iglesias Ruiz Anlisis numrico del comportamiento fluido-trmico-
magntico de un transformador monofsico
1 2 1
= = =
2 1 2
4.5 RENDIMIENTO
Una vez analizadas las prdidas en cada zona del transformador, podemos hablar de
su rendimiento. Por definicin, el rendimiento es la relacin entre la potencia til o
Grado en Tecnologas
Industriales
ETSIIT Santander
17
Curso 2016/2017
Francisco Jos Iglesias Ruiz Anlisis numrico del comportamiento fluido-trmico-
magntico de un transformador monofsico
disponible para su uso entre la potencia de entrada o total utilizada. Tambin puede
expresarse como relacin de la potencia disponible en el secundario entre esa misma
potencia ms la suma de las prdidas totales en el transformador.
Una vez explicados los conceptos de las prdidas y rendimiento del transformador,
veremos cmo afectan al funcionamiento fsico de este.
Grado en Tecnologas
Industriales
ETSIIT Santander
18
Curso 2016/2017
Francisco Jos Iglesias Ruiz Anlisis numrico del comportamiento fluido-trmico-
magntico de un transformador monofsico
Llegamos de esta manera a unos valores de tensin e intensidad mximos, que fijaran
la capacidad de nuestro transformador.
Grado en Tecnologas
Industriales
ETSIIT Santander
19
Curso 2016/2017
Francisco Jos Iglesias Ruiz Anlisis numrico del comportamiento fluido-trmico-
magntico de un transformador monofsico
5. REFRIGERACIN DE TRANSFORMADORES
5.1 INTRODUCCIN
Grado en Tecnologas
Industriales
ETSIIT Santander
20
Curso 2016/2017
Francisco Jos Iglesias Ruiz Anlisis numrico del comportamiento fluido-trmico-
magntico de un transformador monofsico
5.2 NOMENCLATURA
LETRA SIGNIFICADO
LETRA SIGNIFICADO
Grado en Tecnologas
Industriales
ETSIIT Santander
21
Curso 2016/2017
Francisco Jos Iglesias Ruiz Anlisis numrico del comportamiento fluido-trmico-
magntico de un transformador monofsico
LETRA SIGNIFICADO
LETRA SIGNIFICADO
N Conveccin natural.
F Conveccin forzada.
Grado en Tecnologas
Industriales
ETSIIT Santander
22
Curso 2016/2017
Francisco Jos Iglesias Ruiz Anlisis numrico del comportamiento fluido-trmico-
magntico de un transformador monofsico
Para preservar las propiedades del lquido y de los aislantes que estn en contacto
con l hay varios mtodos a considerar. El ms utilizado es el diseo de la cuba y el
Rel Buchholz.
Grado en Tecnologas
Industriales
ETSIIT Santander
23
Curso 2016/2017
Francisco Jos Iglesias Ruiz Anlisis numrico del comportamiento fluido-trmico-
magntico de un transformador monofsico
Estos gases producidos ascienden por el conducto donde est situado el rel hasta
una caja con dos flotadores mviles, los gases ocupan la parte alta de la caja
desplazando al aceite hacia abajo. Si el nivel de aceite llega a un lmite fijado el
flotador cierra un circuito que produce la alarma.
Grado en Tecnologas
Industriales
ETSIIT Santander
24
Curso 2016/2017
Francisco Jos Iglesias Ruiz Anlisis numrico del comportamiento fluido-trmico-
magntico de un transformador monofsico
Figura 6.1. Circuito equivalente de un transformador visto desde el lado del primario.
Grado en Tecnologas
Industriales
ETSIIT Santander
25
Curso 2016/2017
Francisco Jos Iglesias Ruiz Anlisis numrico del comportamiento fluido-trmico-
magntico de un transformador monofsico
Con estas medidas podemos obtener el factor de potencia en vaco de acuerdo con
la ecuacin:
0 = 1 0 cos 0 =
Por otra parte, debido al pequeo valor de la cada de tensin primaria, se puede
considerar que el valor 1 coincide prcticamente con 1 pudiendo obtener las
componentes de 0 y, a partir de estas ltimas los valores y . Estas resistencias
y hacen referencia a las prdidas en el ncleo del transformador (la rama de
vaco).
= 0 cos 0 ; = 0 sin 0
1
= 1 ; =
Es decir, el ensayo de vaco nos permite determinar las prdidas en el hierro del
transformador y los parmetros en la rama en paralelo del circuito equivalente. De
este ensayo tambin podemos obtener la relacin de transformacin, ya que, la cada
de tensin en el cobre es prcticamente despreciable.
Grado en Tecnologas
Industriales
ETSIIT Santander
26
Curso 2016/2017
Francisco Jos Iglesias Ruiz Anlisis numrico del comportamiento fluido-trmico-
magntico de un transformador monofsico
carga por los devanados (en la prctica se realiza siempre en el lado de alta tensin).
La disposicin de los instrumentos de medida se realiza segn el siguiente esquema:
= 1 1 cos
En nuestro modelo a analizar realizaremos estos dos ensayos para obtener las
prdidas totales de transformador funcionando a plena carga y que desglosaremos
en: prdidas en el ncleo y en los devanados.
Estos datos sern uno de los puntos de partida a la hora de simular el modelo para
analizar su comportamiento.
Grado en Tecnologas
Industriales
ETSIIT Santander
27
Curso 2016/2017
Francisco Jos Iglesias Ruiz Anlisis numrico del comportamiento fluido-trmico-
magntico de un transformador monofsico
Para poder llevar a cabo la simulacin del transformador en las mismas condiciones
que el modelo real es necesario un despiece minucioso de este. En la seccin
correspondiente se adjuntan los planos de cada una de las piezas que componen el
transformador adems de la vista explosionada del montaje que se lleva a cabo. Este
modelo del transformador esta realizado con el programa de diseo mecnico
Autodesk Inventor.
Los bloques de construccin en Autodesk Inventor son las piezas. A partir de bocetos
simples, en dos dimensiones, se pueden crear modelos en el espacio a travs de
operacin bsicas en 3 dimensiones (extrusiones, barridos, revoluciones). Estos
archivos de piezas con extensin de trabajo .ipt pueden unirse entre ellas mediante
ensamblajes (.iam).
Grado en Tecnologas
Industriales
ETSIIT Santander
28
Curso 2016/2017
Francisco Jos Iglesias Ruiz Anlisis numrico del comportamiento fluido-trmico-
magntico de un transformador monofsico
Figura 7.1. Ejemplo de la creacin de una pieza a travs de su boceto en Autodesk Inventor
Figura 7.2. Ejemplo de operacin donde se consume el boceto original para crear la estructura 3D.
Grado en Tecnologas
Industriales
ETSIIT Santander
29
Curso 2016/2017
Francisco Jos Iglesias Ruiz Anlisis numrico del comportamiento fluido-trmico-
magntico de un transformador monofsico
Para definir la posicin relativa entre las piezas de un ensamblaje es necesario dotar
de una serie de restricciones a los componentes. Estas restricciones vienen dadas en
funcin de elementos o partes significativas de cada pieza, ya sean aristas, contornos
ms complejos, ejes o simetras.
Grado en Tecnologas
Industriales
ETSIIT Santander
30
Curso 2016/2017
Francisco Jos Iglesias Ruiz Anlisis numrico del comportamiento fluido-trmico-
magntico de un transformador monofsico
Por ltimo, Autodesk Inventor nos da la opcin de exportar esas piezas y ensamblajes
a formato .dwg donde seleccionamos las piezas requeridas y la vista que nos interese
para realizar los planos de esta. En este apartado tendremos todos los elementos de
acotacin y anotacin normalizados para acotar nuestras piezas y ensamblajes.
En los anexos a este proyecto se aportan los planos del transformador creados con
el programa Inventor con la extensin antes descrita. Se detalla las dimensiones
estructurales de cada pieza, con sus vistas de detalle en caso de necesitarla, del
transformador adems una vista en explosin del montaje realizado y la vista final de
la mquina.
Las medidas y dimensiones de los planos sern las utilizadas como base a la hora de
realizar las aproximaciones y simplificaciones pertinentes para lograr un anlisis del
comportamiento trmico y magntico tanto de la maquina en rgimen nominal, como
del lquido refrigerante que lo rodea.
Grado en Tecnologas
Industriales
ETSIIT Santander
31
Curso 2016/2017
Francisco Jos Iglesias Ruiz Anlisis numrico del comportamiento fluido-trmico-
magntico de un transformador monofsico
8. MODELO DE SIMULACIN
8.1 INTRODUCCIN
Una vez que tenemos el modelo fsico del transformador con sus caractersticas
definidas necesitamos realizar una serie de aproximaciones y simplificaciones para
poder convertir ese modelo real en uno equivalente y apto para realizar clculos y
estudios de manera efectiva y en simulaciones con tiempos de clculo factibles.
Grado en Tecnologas
Industriales
ETSIIT Santander
32
Curso 2016/2017
Francisco Jos Iglesias Ruiz Anlisis numrico del comportamiento fluido-trmico-
magntico de un transformador monofsico
Grado en Tecnologas
Industriales
ETSIIT Santander
33
Curso 2016/2017
Francisco Jos Iglesias Ruiz Anlisis numrico del comportamiento fluido-trmico-
magntico de un transformador monofsico
Las tapas que cubren el devanado por la parte superior e inferior se han
simplificado a dos lminas de material plstico con el fin de observar cmo
afectan a la transmisin de calor y a la circulacin del fluido en los puntos
cercanos a los arrollamientos. De igual modo procederemos con el soporte del
transformador, ser simplificada su geometra a una lmina de dimensiones
generales similares a la inicial.
Grado en Tecnologas
Industriales
ETSIIT Santander
34
Curso 2016/2017
Francisco Jos Iglesias Ruiz Anlisis numrico del comportamiento fluido-trmico-
magntico de un transformador monofsico
Figura 8.3. a) Ncleo formado por 120 lminas. b) Ncleo simplificado formado por 20 lminas.
El carlite utilizado como aislante entre lminas tiene un espesor despreciable (valores
de 0.001 mm.) por lo que no se considerar geomtricamente (solamente a la hora
de realizar los clculos en la simulacin). En la simulacin se introducir el carlite
como condicin de contorno interior del sistema.
Grado en Tecnologas
Industriales
ETSIIT Santander
35
Curso 2016/2017
Francisco Jos Iglesias Ruiz Anlisis numrico del comportamiento fluido-trmico-
magntico de un transformador monofsico
Figura 8.4. Comparacin de la geometra simplificada con el montaje realizado con Autodesk Inventor.
Grado en Tecnologas
Industriales
ETSIIT Santander
36
Curso 2016/2017
Francisco Jos Iglesias Ruiz Anlisis numrico del comportamiento fluido-trmico-
magntico de un transformador monofsico
En la figura anterior se muestran las entradas y salidas de fluido en las caras opuestas
del tanque de manera que tenemos 4 entradas de fluido refrigerante en la parte inferior
y 4 salidas en la parte superior.
Una vez obtenido este modelo podemos realizar una simplificacin debido a la
simetra presente respecto a los ejes YZ y XZ. Aplicando estas condiciones de
contorno de simetra podemos trabajar con un cuarto de transformador extrapolando
el comportamiento de ese sector al transformador completo. Esto es posible debido a
las condiciones de entrada. Estas son idnticas en los cuatro radiadores (orificios de
entrada en la imagen) y la generacin de calor uniformemente distribuida en los
devanados y ncleo.
Grado en Tecnologas
Industriales
ETSIIT Santander
37
Curso 2016/2017
Francisco Jos Iglesias Ruiz Anlisis numrico del comportamiento fluido-trmico-
magntico de un transformador monofsico
comportamiento del fluido refrigerante dentro del tanque. La ultima fsica hace
referencia a clculo de campos magnticos y flujos generados, en este caso, por la
presencia de los devanados y el ncleo del transformador. Como veremos ms
adelante, estas tres fsicas interactan entre mediante la variable temperatura, ya que
las propiedades de los materiales presentes en la geometra varan en funcin de los
valores de la temperatura en ese punto o instante. Es necesario acoplar
correctamente los clculos entre estas tres fsicas para obtener resultados lgicos y
comparables con los obtenidos en el ensayo real del transformador.
Para realizar los clculos, el programa Comsol se rige por un balance general de flujos
de energa que cumple la siguiente expresin:
(k T) = T u
Grado en Tecnologas
Industriales
ETSIIT Santander
38
Curso 2016/2017
Francisco Jos Iglesias Ruiz Anlisis numrico del comportamiento fluido-trmico-
magntico de un transformador monofsico
(U) + =0
E
B = 0 + 0 0
La ley de Gauss indica que las lneas de los campos magnticos deben ser cerradas.
En otras palabras, se dice que, sobre una superficie cerrada, sea cual sea sta, no
seremos capaces de encerrar una fuente o sumidero de campo, esto expresa la
inexistencia del monopolo magntico.
B
=0
Grado en Tecnologas
Industriales
ETSIIT Santander
39
Curso 2016/2017
Francisco Jos Iglesias Ruiz Anlisis numrico del comportamiento fluido-trmico-
magntico de un transformador monofsico
El programa Comsol Multiphysics cuenta con una librera interna de materiales con
sus propiedades predefinidas. As todo definiremos los parmetros de nuestro aceite
mineral de acuerdo a datos concreto de este aceite. Mediante interpolaciones
obtendremos una funcin sencilla que se aproxime de forma eficaz a los valores
obtenidos experimentalmente.
865
860
855
848,7
850
845
839,2
840
835
270 280 290 300 310 320 330
TEMPERATURA (K)
Grado en Tecnologas
Industriales
ETSIIT Santander
40
Curso 2016/2017
Francisco Jos Iglesias Ruiz Anlisis numrico del comportamiento fluido-trmico-
magntico de un transformador monofsico
Conductividad trmica
0,132 0,13156
0,1315
0,131
CONDUCTIVIDAD TERMICA (W/M K)
0,1305
0,13 0,129678
0,1295
0,129
0,1285
0,128 0,12742
0,1275
0,127
290 300 310 320 330 340 350 360
TEMPERATURA (K)
Calor especfico
2150
2090,244
2100 2062,4
CALOR ESPECIFICO (J/KG K)
2050 2018,084
1984,04
2000
1945,924
1950 1910,8
1900 1873,764
1850
290 300 310 320 330 340 350 360
TEMPERATURA (K)
Viscosidad cinemtica
0,0000180,0000167
0,000016
0,000014
VISCOSIDAD CINEMATICA (M^2/S)
0,000012 0,0000104
0,00001
0,000008 0,0000064
0,000006 0,0000043
0,00000321
0,000003
0,000004
0,000002
0
290 310 330 350 370 390
TEMPERATURA (K)
Grado en Tecnologas
Industriales
ETSIIT Santander
42
Curso 2016/2017
Francisco Jos Iglesias Ruiz Anlisis numrico del comportamiento fluido-trmico-
magntico de un transformador monofsico
Figura 8.11. Propiedades predefinidas del ncleo del transformador (SILICON STEEL GO 3%)
Grado en Tecnologas
Industriales
ETSIIT Santander
43
Curso 2016/2017
Francisco Jos Iglesias Ruiz Anlisis numrico del comportamiento fluido-trmico-
magntico de un transformador monofsico
Grado en Tecnologas
Industriales
ETSIIT Santander
44
Curso 2016/2017
Francisco Jos Iglesias Ruiz Anlisis numrico del comportamiento fluido-trmico-
magntico de un transformador monofsico
9. COMSOL MULTIPHYSICS
9.1 INTRODUCCIN
Para poder resolver las ecuaciones planteadas en los apartados anteriores aplicadas
a la geometra y materiales descritos con una exactitud moderada necesitamos utilizar
una potente herramienta de clculo. Esta herramienta es un software informtico
capaz de realizar clculos ligados a resolucin de ecuaciones diferenciales referentes
a movimientos de partculas, cambios de temperaturas y flujos magnticos.
Comsol Multiphysics utiliza una tcnica numrica para la resolucin de las ecuaciones
basada en elementos finitos (MEF). A continuacin, explicaremos bsicamente en
que consiste este mtodo.
Con esta frmula se consigue pasar de un elemento continuo, con infinitos grados de
libertad, a un sistema con un numero de grados de libertad igual a la cantidad de
elementos en los que hemos dividido el elemento continuo inicial. Las ecuaciones
diferenciales a aplicar al sistema continuo pasaran a ser un conjunto de ecuaciones
(lineales o no lineales) proporcional a los elementos finitos creados.
Grado en Tecnologas
Industriales
ETSIIT Santander
45
Curso 2016/2017
Francisco Jos Iglesias Ruiz Anlisis numrico del comportamiento fluido-trmico-
magntico de un transformador monofsico
El mtodo de los elementos finitos lleva a cabo una discretizacin de cada dominio
continuo en elementos ms pequeos. Este mallado aplicado puede tener unas
caractersticas variables, dependiendo de la geometra local de cada zona del dominio
y de las fsicas a aplicar sobre l. Condiciona en gran medida la resolucin del
problema, ya que a cada nodo o elemento del mallado se le aplicaran las ecuaciones
elegidas para el modelo.
Este mtodo no deja de ser un mtodo aproximado a la solucin real del sistema
siendo mayor la precisin alcanzada cuanto ms fina es la malla, es decir, mayor
cantidad de elementos (y ecuaciones) y mayor el tiempo de computacin necesario
para su resolucin.
El programa Comsol nos permite trabajar con una geometra creada desde cero en el
mismo programa (a travs de herramientas de CAD) o importada desde ficheros
externos. A esta geometra se aadirn las condiciones de contorno pertinentes y las
ecuaciones (fsicas) que necesitamos en cada geometra. Aplicando un mallado a la
estructura procederemos a la simulacin (resolucin de las ecuaciones).
Comsol nos permite realizar estas acciones a travs de una interfaz simple e intuitiva.
Adems se podr trabajar en modelos multidimensional segn convenga (1D, 2D,
3D,..) e insertar gran variedad de fsicas a los modelos, siendo posible el acoplamiento
de estas cuando sea necesario.
Grado en Tecnologas
Industriales
ETSIIT Santander
46
Curso 2016/2017
Francisco Jos Iglesias Ruiz Anlisis numrico del comportamiento fluido-trmico-
magntico de un transformador monofsico
Mecnica de estructuras.
Modelizacin de componentes.
Matemtica aplicada.
Fsica, mecnica cuntica.
Reacciones qumicas, transferencia de calor, mecnica de fluidos.
Electromagnetismo.
A continuacin, se explicarn los pasos a seguir a la hora de trabajar con este tipo de
programas. Esta metodologa utilizada se puede dividir u ordenar en 3 etapas
generales: prepoceso, proceso y postproceso.
El prepoceso constituye todos los clculos previos a realizar antes de la solucin del
problema. Esto incluye la seleccin del modelo, el diseo de la geometra y la
introduccin y eleccin de las condiciones de contorno.
Una vez definidas estas condiciones debemos seleccionar las fsicas que trabajaran
en la geometra y sobre que partes de ella, siendo posible la asignacin de diferentes
fsicas en distintos subdominios de la geometra completa.
El ltimo paso del preproceso es definir el mallado, Comsol nos da la opcin de crear
un mallado por defecto en funcin del tipo de fsicas seleccionadas para el anlisis.
As todo el diseador puede variar y modificar el mallado creado segn crea
conveniente.
Grado en Tecnologas
Industriales
ETSIIT Santander
47
Curso 2016/2017
Francisco Jos Iglesias Ruiz Anlisis numrico del comportamiento fluido-trmico-
magntico de un transformador monofsico
Grado en Tecnologas
Industriales
ETSIIT Santander
48
Curso 2016/2017
Francisco Jos Iglesias Ruiz Anlisis numrico del comportamiento fluido-trmico-
magntico de un transformador monofsico
Los dominios generados con las operaciones de dibujo necesitan unas propiedades
y unos valores iniciales. Para ello seleccionamos Parmetros en la pestaa de
Definiciones globales y completaremos la tabla con los valores siguientes.
En la tabla se han aadido los valores conocidos del transformador como son: el
volumen tanto del ncleo como de los devanados (este volumen es la cuarta parte del
total ya que estamos aplicando dos planos de simetra), las prdidas obtenidas en los
ensayos de vaco y cortocircuito (tambin referentes a la cuarta parte del
transformador), las velocidades y temperaturas iniciales de las que partimos, as como
el ndice de sobrecarga del transformador. Este ndice de sobrecarga afectar a las
tensiones y corrientes soportadas por los devanados y el ncleo y, por tanto, a las
prdidas que generarn.
Una vez definidos los datos generales de partida es necesario agrupar los diferentes
dominios relativos a un mismo material. Distinguiremos entre el fluido que rodea el
transformador, el ncleo compuesto de las lminas, los devanados de cobre, las tapas
protectoras de estos, y las paredes de la cuba. Para definir estos dominios
seleccionamos la pestaa Componente y aadir Seleccin explicita. Sobre la
geometra se nos permitir seleccionar los diferentes subdominios o contornos de los
que se compone el dominio principal. Esto facilita y simplifica el trabajo a la hora de
Grado en Tecnologas
Industriales
ETSIIT Santander
49
Curso 2016/2017
Francisco Jos Iglesias Ruiz Anlisis numrico del comportamiento fluido-trmico-
magntico de un transformador monofsico
Como las propiedades del fluido las hemos definido en funcin de las tablas de datos
del captulo 10.7 es necesario insertar dichas funciones y definirlas como propiedad
perteneciente al dominio del fluido. En la pestaa Definiciones globales insertamos
las funciones necesarias y definimos sus parmetros. Aplicaremos estos pasos para
definir: densidad, capacidad calorfica, viscosidad cinemtica y conductividad trmica.
Grado en Tecnologas
Industriales
ETSIIT Santander
50
Curso 2016/2017
Francisco Jos Iglesias Ruiz Anlisis numrico del comportamiento fluido-trmico-
magntico de un transformador monofsico
Grado en Tecnologas
Industriales
ETSIIT Santander
51
Curso 2016/2017
Francisco Jos Iglesias Ruiz Anlisis numrico del comportamiento fluido-trmico-
magntico de un transformador monofsico
En esta fsica ser necesario aadir y definir los comportamientos del ncleo, los
devanados y el fluido respecto al campo magntico que se generar en la simulacin.
Grado en Tecnologas
Industriales
ETSIIT Santander
52
Curso 2016/2017
Francisco Jos Iglesias Ruiz Anlisis numrico del comportamiento fluido-trmico-
magntico de un transformador monofsico
Hemos realizado esta seleccin por separado ya que en ncleo del transformador
interviene la curva de magnetizacin del material. Esta curva registra el nivel de
saturacin del ncleo en funcin del campo magntico que circula a travs de l.
Comsol nos da la opcin de en vez de utilizar esta curva (mayor tiempo de
computacin en la simulacin), trabajar con otra curva aproximada a esta, llamada
Curva Efectiva HB. Por esta razn hemos definido por separado el ncleo del
transformador del fluido.
Grado en Tecnologas
Industriales
ETSIIT Santander
53
Curso 2016/2017
Francisco Jos Iglesias Ruiz Anlisis numrico del comportamiento fluido-trmico-
magntico de un transformador monofsico
En la imagen estn definidos los parmetros del devanado de baja tensin. Con el
devanado de alta tensin se repite el procedimiento respetando los valores propios
de este, (distinta tensin de alimentacin, rea correspondiente).
Por ltimo, se aadir una condicin para mejorar la estabilidad y conseguir una nica
solucin al resolver los sistemas de ecuaciones referentes al campo magntico.
Seleccionamos Fijar calibrado de campo A asignando todos los dominios al campo
y dando el valor de la unidad a la variable (A/m) que estar por definir.
Grado en Tecnologas
Industriales
ETSIIT Santander
54
Curso 2016/2017
Francisco Jos Iglesias Ruiz Anlisis numrico del comportamiento fluido-trmico-
magntico de un transformador monofsico
Con los dominios definidos necesitamos ajustar los contornos de la geometra. Estas
condiciones a definir son:
Grado en Tecnologas
Industriales
ETSIIT Santander
55
Curso 2016/2017
Francisco Jos Iglesias Ruiz Anlisis numrico del comportamiento fluido-trmico-
magntico de un transformador monofsico
Figura 9.9. Definicin del carlite en los contornos internos del ncleo.
Grado en Tecnologas
Industriales
ETSIIT Santander
56
Curso 2016/2017
Francisco Jos Iglesias Ruiz Anlisis numrico del comportamiento fluido-trmico-
magntico de un transformador monofsico
Figura 9.10. Asignacin de la bobina de baja tensin a generacin de calor debido a las prdidas.
Una vez definidos los dominios introducimos las siguientes condiciones en los
contornos.
Grado en Tecnologas
Industriales
ETSIIT Santander
57
Curso 2016/2017
Francisco Jos Iglesias Ruiz Anlisis numrico del comportamiento fluido-trmico-
magntico de un transformador monofsico
A la hora de definir el comportamiento del fluido ser necesario aadir las fuerzas que
soportan las partculas en el ciclo que realizan desde la entrada a la salida. Estas
fuerzas que experimentan son las fuerzas de flotabilidad y dependen de la gravedad
y la densidad del fluido en cuestin. Ser necesario introducir esta fuerza aadiendo
en la pestaa de condiciones de contorno de dominios la condicin de fuerza
volumtrica. Aparecer un campo de fuerzas a rellenar. En nuestro caso, la nica
fuerza presente es la descrita anteriormente cuya nica componente es en el eje z.
Grado en Tecnologas
Industriales
ETSIIT Santander
58
Curso 2016/2017
Francisco Jos Iglesias Ruiz Anlisis numrico del comportamiento fluido-trmico-
magntico de un transformador monofsico
Ya hemos definido las fsicas a utilizar, los parmetros iniciales, los materiales que
intervienen y las condiciones de contorno que afectan a las tres fsicas que aparecen.
El siguiente paso es acoplar las variables que intervengan en ms de una fsica. Es
decir, ligar los clculos de temperaturas y flujos magnticos con los flujos msicos
que circulan por el tanque.
Grado en Tecnologas
Industriales
ETSIIT Santander
59
Curso 2016/2017
Francisco Jos Iglesias Ruiz Anlisis numrico del comportamiento fluido-trmico-
magntico de un transformador monofsico
Flujo no isotrmico, indica que la temperatura del fluido est ligada a la calculada en
el intercambio de calor. Basta con definir el dominio del fluido, los dos campos que
necesitas acoplar, en este caso Flujo laminar (spf) y Transferencia de calor en
solidos (ht).
Grado en Tecnologas
Industriales
ETSIIT Santander
60
Curso 2016/2017
Francisco Jos Iglesias Ruiz Anlisis numrico del comportamiento fluido-trmico-
magntico de un transformador monofsico
Estas cuatro sondas sern colocadas en los puntos descritos utilizando el comando
sonda que nos proporciona Comsol. Dentro de la funcin sondas, podemos colocar
sondas de dominio, de contorno y puntuales. Estas ltimas son las que utilizaremos
para comparar la simulacin con los datos obtenidos del ensayo fsico.
Adems de las sondas puntual, se colocarn sondas que recogern los valores de
temperatura media y mxima en los arrollamientos y el ncleo del transformador.
Tambin obtendremos datos de la temperatura media del aceite, la tapa de la cuba,
la potencia disipada por el slido (ncleo) y la velocidad promedio de salida del fluido
de la cuba.
Grado en Tecnologas
Industriales
ETSIIT Santander
61
Curso 2016/2017
Francisco Jos Iglesias Ruiz Anlisis numrico del comportamiento fluido-trmico-
magntico de un transformador monofsico
Figura 9.14. Sondas puntuales colocadas para replicar las sondas del ensayo.
Grado en Tecnologas
Industriales
ETSIIT Santander
62
Curso 2016/2017
Francisco Jos Iglesias Ruiz Anlisis numrico del comportamiento fluido-trmico-
magntico de un transformador monofsico
9.9 MALLADO
La forma de los elementos del mallado puede ser muy variada, siendo diferente el
resultado de elegir un tipo de elemento u otro. Un mayor nmero de elementos no
tiene porqu implicar un mejor resultado si los elementos elegidos no son los ms
adecuados. En nuestro modelo utilizaremos elementos tetradricos, ya que estos se
adaptan mejor a elementos irregulares, adems de ser la forma predefinida por
Comsol.
Grado en Tecnologas
Industriales
ETSIIT Santander
63
Curso 2016/2017
Francisco Jos Iglesias Ruiz Anlisis numrico del comportamiento fluido-trmico-
magntico de un transformador monofsico
A la hora de valorar la calidad del mallado utilizado Comsol valora la distorsin que
sufre cada elemento de malla respecto al triangulo ideal (equiltero) al adaptarse a la
geometra correspondiente. Por lo tanto, un mallado fino ajustar mucho mejor y
sufrir menor distorsin de sus elementos y su calidad ser mayor. La calidad del
mallado utilizado se muestra a continuacin.
Grado en Tecnologas
Industriales
ETSIIT Santander
64
Curso 2016/2017
Francisco Jos Iglesias Ruiz Anlisis numrico del comportamiento fluido-trmico-
magntico de un transformador monofsico
En este apartado se indican tanto el tipo de estudio realizado como los resolvedores
utilizados para obtener los resultados finales. Para realizar el anlisis se emplea un
barrido paramtrico en el cual se altera la temperatura inicial de los elementos y se
mantiene fija la velocidad de entrada del fluido refrigerante por el conducto inferior.
Se realizar un estudio en rgimen permanente del comportamiento del transformador
bajo estas condiciones iniciales con un ndice de carga del 150%. Las sondas
colocadas en los puntos de medida correspondientes al ensayo fsico deben de
Grado en Tecnologas
Industriales
ETSIIT Santander
65
Curso 2016/2017
Francisco Jos Iglesias Ruiz Anlisis numrico del comportamiento fluido-trmico-
magntico de un transformador monofsico
Grado en Tecnologas
Industriales
ETSIIT Santander
66
Curso 2016/2017
Francisco Jos Iglesias Ruiz Anlisis numrico del comportamiento fluido-trmico-
magntico de un transformador monofsico
Estos dos estudios mostrados trabajarn de forma conjunta con el barrido paramtrico
aplicado a las temperaturas iniciales de toda la geometra de manera que se
resolvern las ecuaciones para cada valor de temperatura inicial introducido en el
barrido paramtrico.
313.15 K: 40 C
318.15 K: 45 C
323.15 K: 50 C
328.15 K: 55 C
Con estos datos ya estara definido todo lo necesario para resolver los clculos
necesarios en nuestra geometra. En un primer paso se compilarn las ecuaciones
de la geometra de cada bobinado seguido del estudio frecuencia-estacionario.
Grado en Tecnologas
Industriales
ETSIIT Santander
67
Curso 2016/2017
Francisco Jos Iglesias Ruiz Anlisis numrico del comportamiento fluido-trmico-
magntico de un transformador monofsico
10. RESULTADOS
Se puede observar el efecto produce el carlite utilizado como aislante entre las chapas
del ncleo. Estas laminas consiguen una distribucin ms uniforme del flujo que
Grado en Tecnologas
Industriales
ETSIIT Santander
68
Curso 2016/2017
Francisco Jos Iglesias Ruiz Anlisis numrico del comportamiento fluido-trmico-
magntico de un transformador monofsico
circula por el ncleo, siendo ms importante conforme nos acercamos a los bordes.
Esto es debido a la presencia de los arrollamientos, si hubiramos considerado el
ncleo como una nica lamina gruesa y maciza nos hubiramos encontrado con que
la gran mayora de lneas de flujo se encontrarn en los bordes del ncleo.
Grado en Tecnologas
Industriales
ETSIIT Santander
69
Curso 2016/2017
Francisco Jos Iglesias Ruiz Anlisis numrico del comportamiento fluido-trmico-
magntico de un transformador monofsico
Se observa claramente como el fluido se calienta a medida que entra en contacto con
el soporte del transformador y los devanados.
Grado en Tecnologas
Industriales
ETSIIT Santander
70
Curso 2016/2017
Francisco Jos Iglesias Ruiz Anlisis numrico del comportamiento fluido-trmico-
magntico de un transformador monofsico
En cuanto al comportamiento del fluido, se muestra una captura del movimiento del
aceite a lo largo del tanque. Las lneas mostradas corresponden a los diferentes
desplazamientos que realiza el aceite desde su entrada por el conducto inferior, a su
salida por el superior.
Grado en Tecnologas
Industriales
ETSIIT Santander
71
Curso 2016/2017
Francisco Jos Iglesias Ruiz Anlisis numrico del comportamiento fluido-trmico-
magntico de un transformador monofsico
El comportamiento del fluido es muy similar para las cuatro temperaturas de entrada
simuladas. El fluido se divide en dos caminos principales. El primer flujo choca con el
transformador absorbiendo parte del calor generado por este y asciende describiendo
una espiral. Esto ocurre de esta manera debido a que el fluido es incapaz de progresar
hacia la zona central del transformador ya que es la zona intermedia a dos flujos
iguales y de sentidos contrarios (simetra aplicada).
El segundo flujo se desplaza hacia la zona del borde del tanque, refrigerando en
menor medida y ascendiendo a medida que aumenta su temperatura.
Grado en Tecnologas
Industriales
ETSIIT Santander
72
Curso 2016/2017
Francisco Jos Iglesias Ruiz Anlisis numrico del comportamiento fluido-trmico-
magntico de un transformador monofsico
Las zonas con velocidades ms altas son ambos conductos, siendo la huella de
entrada ms pronunciada en la entrada de fluido, pero dndose mayores velocidades
en la salida del fluido.
En este apartado analizaremos y compararemos los valores obtenidos con las sondas
reales (en el ensayo fsico) con los datos almacenados en la simulacin. Dividiremos
los resultados en dos grupos: el primero haciendo referencia las temperaturas medias
del fluido, devanados y dems componentes, y el segundo comparando las sondas
colocadas en los mismos puntos que las colocadas en el ensayo de laboratorio.
Grado en Tecnologas
Industriales
ETSIIT Santander
73
Curso 2016/2017
Francisco Jos Iglesias Ruiz Anlisis numrico del comportamiento fluido-trmico-
magntico de un transformador monofsico
*Esta sonda marca la temperatura media de todo el aceite en contacto con la tapa del
tanque.
T_in (degC) T. Max. Devanados T. media Ncleo T. Max. Ncleo Potencia total disipada (W)
Grado en Tecnologas
Industriales
ETSIIT Santander
74
Curso 2016/2017
Francisco Jos Iglesias Ruiz Anlisis numrico del comportamiento fluido-trmico-
magntico de un transformador monofsico
DATOS EXPERIMENTALES
En la tabla se muestran los datos obtenidos con las sondas en el barrido parametrico.
Debajo del valor de temperatura se muestra el porcentaje de error en comparacion
con los datos experimentales.
DATOS DE LA SIMULACION
Grado en Tecnologas
Industriales
ETSIIT Santander
75
Curso 2016/2017
Francisco Jos Iglesias Ruiz Anlisis numrico del comportamiento fluido-trmico-
magntico de un transformador monofsico
Se han considerado aceptables los datos de las sondas con errores por debajo del
5% del valor experimental. Observamos que con temperaturas iniciales mayores
logramos ajustarnos bastante a los valores de temperatura del fluido pero las
temperaturas en el cobre y el hierro son mucho mayores de lo que debieran ser.
Grado en Tecnologas
Industriales
ETSIIT Santander
76
Curso 2016/2017
Francisco Jos Iglesias Ruiz Anlisis numrico del comportamiento fluido-trmico-
magntico de un transformador monofsico
11.1 CONCLUSIONES
Ciclos de conveccin internos del aceite con similitudes respecto a los artculos
de partida.
Grado en Tecnologas
Industriales
ETSIIT Santander
77
Curso 2016/2017
Francisco Jos Iglesias Ruiz Anlisis numrico del comportamiento fluido-trmico-
magntico de un transformador monofsico
12.BIBLIOGRAFIA
Jaykumar Patel, Kinnari Parekh, R.V. Upadhyay (2016). Prevention of hot spot
temperature in a distribution transformer using magnetic fluid as a coolant.
Jess Fraile Mora. (2003). Mquinas elctricas (5 ed.). Madrid: Mc Graw Hill.
Grado en Tecnologas
Industriales
ETSIIT Santander
78
Curso 2016/2017
Francisco Jos Iglesias Ruiz Anlisis numrico del comportamiento fluido-
trmico-magntico de un transformador monofsico
1
Francisco Jos Iglesias Ruiz Anlisis numrico del comportamiento fluido-
trmico-magntico de un transformador monofsico
ANEXO N1.
Prototyping a Ferrofluid-Cooled Transformer. Lucian PslaruDanescu,
Alexandru M. Morega, Mihaela Morega. (2013).
ANEXO N2.
Prevention of hot spot temperature in a distribution transformer using
magnetic fluid as a coolant. Jaykumar Patel, Kinnari Parekh, R.V. Upadhyay
(2016).
2
IEEE TRANSACTIONS ON INDUSTRY APPLICATIONS, VOL. 49, NO. 3, MAY/JUNE 2013 1289
AbstractThis paper presents the work conducted on pro- within the ferrofluids may be controlled by adequately adjusting
totyping a step-up/step-down, single-phased, low-power, and the incident magnetic field [3], [4].
medium-voltage electrical transformer cooled by a fluid with col- This paper is concerned with the utilization of a magnetic
loidal magnetic nanoparticles. The magnetic and fluid dynamic
properties and the heat capacities of the ferrofluid (magnetic nanofluid type of ferrofluid as coolant and insulating medium in
nanofluid) and that of the regular coolant (UTR-40 transformer a prototype electrical transformer. The magnetic fluid (MF) is
oil) were experimentally determined and comparatively evaluated. a colloidal dispersion of magnetite nanoparticles in transformer
Mathematical models for the electromagnetic field and the heat oil UTR-40, with oleic acid as surfactant. The idea of using
transfer were defined and numerically solved to assess the capacity a magnetic nanofluid based on transformer oil (NFM/UTR) as
of the transformer to sustain the working conditions. The simu-
lation results were utilized to improve the design of a prototype, an insulating media, with same or better dielectric properties
where the UTR-40 regular coolant is replaced by ferrofluid. The and better thermal conductivity (in an external magnetic field)
numerical simulation results and the experiments evidence the against pure oil [5], [6], contributes to the solution of these
superior performance of the prototype. problems, thus decreasing the construction expenses and the
Index TermsElectromagnetic analysis, magnetic liquids, operation costs of transformers.
numerical simulation, power transformers prototype, thermal The expected results resided in the preparation of the
analysis. NFM/UTR magnetic nanofluid based on transformer oil with
satisfactory dielectric and magnetic properties in prototyping
I. I NTRODUCTION a small power transformer filled with NFM/UTR with bet-
ter operational condition against its counterpart filled with
R ECENTLY, a growing interest has been noticed in the
usage of high-power devices for electrical engineering,
particularly in transformers with colloidal ferrofluids [1], which
pure transformer oil. The monodomain magnetic nanoparticles
(115 nm) are ultrastably dispersed in the nonpolar basis en-
adds to their important but more classical industrial applications vironment [4], [5]. Each magnetic nanoparticle is coated with a
(e.g., sensors, sealing systems, and different types of contacts monomolecular layer of chemisorbed organic stabilizer, usually
[2]). The ferrofluids possess heat transfer and dielectric prop- oleic acid.
erties that are superior compared to their classical counterparts Ferrofluids also increase the transformers capacity to resist
(e.g., transformer oils) and may be utilized in improving the overvoltages and exhibit better resistance to degradation in
heat flow within the aggregate active part, thus contributing time because of humidity in comparison with classic oils [6].
to increasing their withstanding capacity to faults such as High-efficiency and smaller size transformers may thus be
electromagnetic impulses. The ferrofluids may be driven by built, with longer lifetime and higher loading capacity. These
external magnetic fields, and the magnetization body forces beneficial effects of the magnetic nanofluids are now exploited
by the realization of the prototype transformer TMOf2-36-40.
Other detailed information is available by numerical simulation
Manuscript received November 1, 2011; revised March 26, 2012 and (e.g., the heat transfer in the spaces between the windings and
August 7, 2012; accepted August 13, 2012. Date of publication March 14, between the windings and the core).
2013; date of current version May 15, 2013. Paper 2011-EMC-581.R2, pre-
sented at the 2010 International Conference on Optimization of Electrical
and Electronic Equipment, Brasov, Romania, May 2022, and approved for II. M AGNETIC AND H EAT T RANSFER P ROPERTIES
publication in the IEEE TRANSACTIONS ON INDUSTRY APPLICATIONS by
OF THE F ERROFLUID
the Electric Machines Committee of the IEEE Industry Applications Society.
This work was supported by the National Research Authority under Grant
21-043/2007.
A. Magnetic Properties
L. PslaruDanescu and V. Stoica are with the National Institute for Research
and Development of Electrical Engineering (INCDIE ICPE-CA), 030138
The magnetization curve (Fig. 1) of the UTR-40-based MF
Bucharest, Romania (e-mail: paslaru@icpe-ca.ro; vstoica@icpe-ca.ro). was drawn by using a vibrating sample magnetometer (VSM
A. M. Morega and M. Morega are with the University Politehnica of 880, DMS/ADE Technologies, USA). The experiments were
Bucharest, 060042 Bucharest, Romania (e-mail: amm@iem.pub.ro; mihaela@
iem.pub.ro).
conducted at room temperature (25 C). The magnetization
O. M. Marinica is with the Romanian AcademyTimisoara Division, curve provides information on the magnetic saturation (Ms =
300223 Timisoara, Romania (e-mail: marinica.oana@gmail.com). 3.979 kA/m). A more accurate value (Ms = 4.130 kA/m)
F. Nouras and N. Paduraru are with the Research, Development and Test-
ing National Institute for Electrical Engineering, 200746 Craiova, Romania may be obtained from the linear region of the magnetization
(e-mail: micro@icmet.ro). curve for high values of magnetic field using the magnetiza-
I. Borbth and T. Borbth are with Roseal Corporation S.A., 535600 tion vs. magnetic field strength representation M = M (1/H).
Odorheiu Secuiesc, Romania (e-mail: office@roseal.topnet.ro; borbath.tunde@
gmail.com). The initial susceptivity of the sample was determined out
Digital Object Identifier 10.1109/TIA.2013.2252872 of the slope of the linear region of the magnetization curve
Fig. 1. Magnetization curve for the UTR-40-based MF. The curve of the first
Fig. 2. Temperature dependence of the heat capacity for the UTR-40 () and
magnetization is practically undistinguishable.
for the UTR-40-based magnetic nanofluid () samples.
(I = 0.08) for lower values of magnetic field. Fig. 1 presents
one of the fundamental properties of MFs, namely, the lack of
hysteresisa property specific to superparamagnetic materials.
Considering these experimental results and conveniently
modeling the magnetization of the magnetic nanofluid in the
numeric simulations reported later, the magnetization is ap-
proximated by the analytic formula
M = arctan(H) (1)
B = 0 (H + M) (2)
B. Heat Transfer Properties Fig. 3. Sketch of the simplified model for the experimental apparatus [8].
The heat capacity of the MF sample was measured using a of the magnetic nanofluid. Fig. 3 shows a sketch that outlines
laboratory calorimeter, equipped with a data acquisition system the main features of the apparatus. Here, B (in tesla) is the
(PCMCIA Card DAQ NI 6062E and signal conditioner NI SC magnetic flux density, H (in amperes per meter) is the magnetic
2345). The calorimeter constant was determined using distilled field strength, M (in amperes per meter) is the magnetiza-
water. Several tests of heating and cooling to a prescribed tion, and Q (in joules) is the heat. This basic experimental
temperature (40 to 80 C) for both the carrier (transformer oil) arrangement was numerically simulated to validate the finite-
and the MF sample were performed. The heat capacity of the element method (FEM) implementation of the mathematical
sample was determined from the heat flux balance between the model, later used for the heat transfer analysis. The mathe-
heater and the fluid inside the calorimeter. matical model, simulation results, details, and conclusion are
The measured densities of the transformer oil and of the given in [8].
MF sample at 24 C were 866 and 963 kg/m3 , respectively.
The specific heat capacities of the two fluids are compared in
III. P ROTOTYPE OF THE P OWER T RANSFORMER
Fig. 2, where the error bars are within 3% error limit. These
results show that the heat capacity of the MF is lower than that The single-phased, small-power (40 kVA), medium-voltage
of the pure transformer oil. It increases with temperature, in (36/0.4 kV), and step-up or step-down transformer built and
agreement with the theoretical prediction of the mixture heat analyzed in our study (TMOf2-36-40) is internally cooled by
capacity equation [3]. the natural circulation of either the transformer oil or the
magnetic nanofluid; the exterior cooling is provided through the
natural circulation of air.
C. Modeling the Experimental Apparatus
The design of the transformer prototype (Fig. 4) relies on
We first modeled an experimental apparatus previously used constructive solutions that comply with the aimed purpose.
to characterize the MF dynamics and heat transfer properties TMOf2-36-40 presents the advantage of reduced volume and
PSLARUDANESCU et al.: PROTOTYPING A FERROFLUID-COOLED TRANSFORMER 1291
TABLE I
PHYSICAL QUANTITIES USED IN THE MATHEMATICAL MODELS
cp (u )T = k2 T + E Je . (4)
(u )u = p + 2 u + fmg + fT . (5)
The magnetic body forces are [10] B. Numerical Model and Solution Procedure
The magnetic nanofluid flow is mainly of a thermal nature;
fmg = 0 (M )H. (7) therefore, it is characterized by small velocity. Hence, we as-
sume that this motion does not influence the incident magnetic
Design constraints, materials, and external conditions lead field. Consequently, the electromagnetic field problem may be
to temperature gradients that drive heat fluxes throughout the decoupled and solved in the first place. The resulting resistive
system, such that the mass density (a temperature-dependent power and body magnetic forces are inputted next, through
intensive property) is nonuniform throughout the working fluid. their rms values, to the heat transfer and flow problems that are
Consequently, buoyancy (body) forces may emerge, entrain- solved together because of the nonlinear coupling by buoyancy.
ing a convection motion aimed at restoring the local thermal The mathematical model (3)(7) is solved through Galerkin
equilibrium. The coolant then conveys the heat produced in FEM, as implemented by Comsol Multiphysics [12].
the active parts of the transformer (iron core and windings)
to the case, where it is transferred to the environment. Heat
is carried by convection, in the working magnetic nanofluid, V. N UMERICAL S IMULATION R ESULTS AND D ISCUSSION
and by conduction, in the solid parts. As indicated by (5), the The orientation of the magnetic forces versus the thermal
magnetic body forces add to buoyancy, such thatunlike the forces is an important factor in providing an optimal design
standard transformer oilthe convective flow is the result of [9], [10]; simpler 2-D models were used to investigate this
their combined action. effect. Here, the computational models are only some relevant
Equation (7) shows that the magnetic body forces are im- slices of the 3-D problem, and the heat sources are specified
portant in the regions where the magnetic field gradients are per unit length. Meshes of roughly 30 00060 000 quadratic
large, e.g., by the end parts of the windings. Simulation results vector elements (for the magnetic field problem) and quadratic
reported next are consistent with this observation. Lagrange elements (for the heat transfer problem) provide
The thermal properties of the model are assumed linear. For accurate numerical solutions.
the natural gravitational flow of the coolant, we assume the
Boussinesq model [11], where the mass density dependence on
temperature occurs in the body force term only (5). A. Simplified 2-D Model
The boundary conditions that close the model are the
1) Horizontal Configuration: Fig. 7 shows the numerical
following.
simulation results for 1) oil and 2) magnetic nanofluid. In this
1) The magnetic insulation for the transformer case. plane, the buoyancy and magnetic forces act concurrently, and
2) The stand on which the transformer is mounted is isother- their combined effect is more important in the end regions.
mal (a thermostat), at the ambient temperature (T = As seen, heat is transported from the active parts to the case
300 K); all other parts of the case are cooled by mod- in one step. When magnetic nanofluid is used [Fig. 7(b)], a
erate natural convection (the heat transfer coefficient by larger number and more localized recirculation cells enhance
convection is h = 2 W/m2 K). the transfer efficiency.
3) The no-slip (zero contact velocity) boundary conditions This actually means a lower hot-spot temperatureby ap-
are set for the coolant flow at the solid parts (core, proximately 10 in this caseand a more uniform tempera-
windings, and case). ture distribution. These results suggest that the vertical design
PSLARUDANESCU et al.: PROTOTYPING A FERROFLUID-COOLED TRANSFORMER 1293
of the transformer may be advisable, and this indication is The two configurations were investigated by using notional
evaluated next. 3-D models. Numerical simulation is now much more demand-
2) Vertical Configuration: Fig. 8 shows the numerical sim- ing in software and hardware resources.
ulation results for the vertical design. A more efficient heat 1) Horizontal Configuration, 3-D Model: Symmetry was
transfer is observable when the magnetic nanofluid is used. This used to simplify the problem, and Fig. 9(a) shows the computa-
design is superior and leads to a more uniform temperature tional domain.
distribution; the usage of ferrofluid rather than the transformer As before, the electromagnetic field is solved first (harmonic
oil leads to a decrease of 2030 of the hot-spot temperature. quasi-steady working conditions and balanced primary and sec-
The 2-D models show that, for transformer oil, the vertical ondary amperturns); the resistive (active) power in the windings
configuration is cooler than the horizontal configuration, and is the heat source, and the magnetic field is the source for the
there is no significant improvement when MF is used instead (magnetization) body forces.
of oil. However, when in the vertical configuration, MF is The magnetic field model leads to a nonlinear algebraic
used instead of transformer oil, and the hot-spot temperature system of equations that was solved by using a parallel solver;
is significantly lower. the heat transfer model was solved using a segregated solver
Under nominal working conditions (compensated amper- approach, combining a parallel solver for the energy equation
turns), the magnetic field is mainly the leakage fieldthe and a direct solver for the hydrodynamic problem. The parallel
channel between the high-voltage and low-voltage windings solver greatly reduces the CPU time. Fig. 10 presents the flow
concentrates the field. Numerical simulation results indicate fields by numerical simulation for the oil and ferrofluid.
that the upper bounds of the magnetic flux density are 15.1 mT The flow in Fig. 10(a) is purely thermal and evidences the
for transformer oil and 15.4 mT for magnetic nanofluid, re- regions of higher temperature gradients. Large convection cells,
spectively. These values are estimated for a primary winding almost orthogonal to the Oy direction, are localized mainly in
equivalent current density of 2 106 A/m2 , and it is recorded the end part. The coolant conveys the heat from the windings
in the channel by its mid-height. to the case wall, where it is discarded to the ambient by natural
This small discrepancy between the two cases suggests that convection. The bottom wall contacts a thermostat.
the electrical parameters of the transformer do not significantly The flow in Fig. 10(b) is more complex, localized in the
change when the transformer oil is replaced with the magnetic regions of higher magnetic field gradient.
nanofluid (a diluted formula) used in our study. This fact is The two types of flows are different, but the efficiency of
confirmed also by experimental measurements performed in the using MF rather than transformer oil is seen here as even more
laboratory,1 e.g., the short-circuit voltage (< 5%) and the open- arguable than in the 2-D simulations since the difference in hot-
circuit current (< 3%) remain unchanged. spot temperatures is less than 10 . In both cases, the end parts
It is mentioned again that, in view of (5), when magnetization (yoke regions) play an important role, and it is clear that the
body forces occur, although the amplitude of the magnetic 3-D effects are more complex than evidenced in the 2-D mod-
els. As indicated by the 2-D analysis, the usage of ferrofluid
in this design does not improve significantly the heat transfer.
1 Technical Bulletin FT348482, ICMET-Craiova, LCD-AEMEE, 2009. More details are given in [10].
1294 IEEE TRANSACTIONS ON INDUSTRY APPLICATIONS, VOL. 49, NO. 3, MAY/JUNE 2013
Next, the temperatures of the windings were calculated using Fig. 13. Prototype transformer, TMOf2-36-40, in the test rig.
the method of the resistance variation. The diagram of the test TABLE II
circuit for the heating experiment is shown in Fig. 12, where MEASURED AND CALCULATED HEATING OF THE
G is the mains; K1, K2, and K3 are switches; Tc is a current TRANSFORMER WINDINGS
transformer; mV and V are voltmeters; I is a switch; A1 and A2
are ampermeters; and S1 and S2 are dc power supplies.
The winding temperatures are determined at steady
state. The measurement starts after disconnecting power
supply G (Fig. 12) through switch I and switching the winding
connections from the ac supply to the electronic stabilized
supply S2 for the low-voltage winding and S1 for the high- TABLE III
HEATING MEASURED ON DIFFERENT PARTS OF THE TRANSFORMER
voltage winding. The measurement of the electric resistance
of the low-voltage and high-voltage windings starts as soon as
the dc current measurement is stabilized, after connecting to
the measurement kit. The sequence of indirect measurement
of the winding temperatures is as follows:
1) reconnection of the secondary winding to generator G
through switch I (Fig. 12);
2) reconnecting the primary winding to the medium-voltage The heating of the windings is = t a , where a is the
source through switch K1; ambient temperature and 0 is the winding average temperature,
3) the measurement of the low-voltage winding resistance measured in cold state (initial), after a long conservation (over
using the ampermeter and voltmeter method, by connect- 48 h) in the transformer testing room. The ambient temperature
ing switch K3; in the initial stage a is 0 due to the specific conditions at
4) the measurement of the medium-voltage winding resis- the test location and due to the relatively constant temperature
tance using the ampermeter and voltmeter method, by outside the room.
connecting switch K2.
The measurements are performed in two stages: at the be- B. Results and Discussion
ginning of the heating tests and after 3 h of continuous func-
The comparative results of measurements and tests (Fig. 13)
tioning. For monitoring of the ambient temperature, we used
were presented in the test reports issued by the laboratories
thermocouples placed at 1 m around the transformer.
where they were performed. The results for the two cases are
The average temperature of the windings t measured in the
presented in Tables II and III.
warm state of the transformer was calculated according to SR
These experimental data provide lumped thermal circuit re-
EN 60076-2:2002 standard
sults. The numerical simulation (e.g., Fig. 8) provides the tem-
t = (Rt /R0 )(T + 0 )T (8) perature distribution and the convection flowan information
that is useful, for instance, in evidencing the hot-spot regions
where, for copper, T = 235 K. The other quantities in (8) and the internal heat transfer mechanisms.
are as follows: 0 the average temperature of the winding, Fig. 14 presents the evolution of the temperature at the finned
measured in the cold state of the transformer; Rt the winding tank registered with the Ti 20 thermographic camera.
resistance, measured in the warm state of the transformer; and Fig. 15 shows the temperature variation as recorded during
R0 the winding resistance, measured in the cold state of the the experiments, at the top part of the case, in two different
transformer. locations. The (thermal) steady-state regime is attained faster at
1296 IEEE TRANSACTIONS ON INDUSTRY APPLICATIONS, VOL. 49, NO. 3, MAY/JUNE 2013
VII. C ONCLUSION
The main conclusions drawn in this paper are as follows.
1) The magnetization curve allows the determination of an
important physical property of MFs, namely, the satura-
tion magnetization (Ms = 3.979 kA/m), which reflects
the volume concentration, low in this case, of the mag-
netic nanoparticles in MFs ( = 0.012).
2) The investigation methods of microstructural and flow
properties, i.e., vibrating sample magnetometry and
Fig. 14. Temperature in the insulated crossing and low-voltage terminal zone, rheometry, indicate a high colloidal stability of MF sam-
recorded with the thermographic camera Ti20, using magnetic nanofluid.
ple based on UTR-40, with Fe3 O4 particles and oleic
acid as monolayer surfactant, and the existence of a small
fraction of agglomerates.
3) The stable MFs exhibit a Newtonian behavior for normal
shear stresses (less than 103 s1 ).
4) The usage of an MF as coolant reduces the weight and
volumes of the transformers following the intensification
of the cooling effect in the presence of the magnetic
nanofluid.
5) The design of the magnetic circuit and the metallic
construction (tank, bottom, and lid) leads to a reduced
consumption of the main materials: copper and magnetic
steel sheet, including the magnetic nanofluid.
6) The numerical simulations of the magnetic field and the
heat transfer within the transformer show that the direc-
tions of the magnetic and gravitational thermal forces are
an important element in providing optimal cooling.
7) The several 2-D models aimed at conveniently evaluat-
ing the electromagnetic field, the heat transfer, and the
convection flow mechanisms reveal that vertical config-
uration provides better usage of the ferrofluid. The 3-D
models confirm that vertical positioning ensures better
cooling.
8) The usage of a ferrofluid as coolant provides magneti-
zation body forces that add to the thermal gravitational
forces. However, in natural convection, magnetic body
forces do not act always to significantly improve the
heat removal provided by the thermal body forces. In
the vertical configuration of the transformer, these forces
Fig. 15. Temperature dynamics at two locations, for the transformer with
oil and MF. Heat transfer steady state is faster attained at location T120. act concurrently with the thermal flow, and the overall
(a) Temperature recorded at the top lid. (b) Temperature profile on the case, effect is the enhancement of the heat transferred from
at the LV insulator passage. the aggregate active parts (core and windings) to the
ambient.
location labeled T120. This discrepancy in the duration of the 9) The simulation results provided recommendations for the
thermal transients at different locations is neither unexpected optimization of the design of the single-phased trans-
nor surprising since the transformer is a complex structure that former, which is the object of this project. Mainly, in the
does not behave as a lumped thermal system. current constructive version, the transformer is built in
The faster temperature growth when oil is the cooling agent vertical construction (the high and low voltage have the
is explained by different loading conditions; in that case, the turns disposed horizontally).
PSLARUDANESCU et al.: PROTOTYPING A FERROFLUID-COOLED TRANSFORMER 1297
Nicolae Paduraru was born in Chisinau, Tnde Borbth was born in Odorheiu Secuiesc,
Romania, in 1934. He received the B.S. degree Romania, in 1981. She received the Engineer
from the Electrical Engineering Institute of Craiova, diploma in computer science and engineering from
Craiova, Romania, in 1957. the Faculty of Automation and Computers, Univer-
Since 1957, he has been a Designer Engineer sity Politehnica University of Timisoara, Timisoara,
with Electroputere, Craiova, where he was the Senior Romania, in 2005. She has been working toward
Designer for measuring transformers from 1958 to the Ph.D. degree in the Power Engineering Fac-
1964 and the Head of the design group in 1964. In ulty, University Politehnica of Bucharest, Bucharest,
19681998, he was the Head of the design group Romania, since 2009, working on technical applica-
for measuring transformers, Electroputere Research tions of the magnetic nanofluids, especially magnetic
Institute. Since 1980, he has been a Senior Specialist nanofluid seals.
with Craiova University, Craiova. Since 1999, he has been a Design Engineer She has been a Research Engineer in automation with Roseal Corporation,
with the Research, Development and Testing National Institute for Electrical Odorheiu Secuiesc, Romania, since 2007. She is involved in research projects
Engineering (ICMET), Craiova. He has outstanding experience as a research from the magnetic fluid domain and their technical applications.
engineer in electrical fields. He has published over 50 papers in technical jour-
nals and conference proceedings. His research interests include unconventional
insulation systems for high-voltage equipment and unconventional measuring,
currents and voltages, and sensors.
a r t i c l e i n f o a b s t r a c t
Article history: A MneZn ferrite magnetic uid (TCF-56) having 5.6 mT uid magnetization and high pyromagnetic co-
Received 12 August 2015 efcient, (vM/vT 177 A/m K) has been investigated as a coolant in a 3 kW prototype transformer for
Received in revised form overloading condition (167%). The winding temperature of a transformer submerged in magnetic uid
19 December 2015
reaches at 396.8 K after 3 h of overloading, which is 20 K lower, when the same experiment was carried
Accepted 21 December 2015
Available online 16 January 2016
out with pure transformer oil. Similarly, core and top oil temperature also decrease by 14 and 21 K,
respectively, when TCF-56 is used. This cooling performance of TCF-56 attributes to the thermo-magnetic
convection, which sets up due to the signicant change in magnetization of the uid with increasing
Keywords:
Heat transfer
temperature. This can be explained using the Rayleigh number. The normal life of a transformer under
Thermo-electrical energy 167% overloading condition is calculated for pure oil and TCF-56. The result shows nine times increase in
Magnetic uid normal life expectancy in TCF-56 uid compared to use of pure oil. The study leads to the conclusion that
MneZn ferrite magnetic uid (TCF-56) used in a transformer can deliver more power than its nameplate
rating with an improved normal life.
2015 Elsevier Masson SAS. All rights reserved.
1. Introduction gradient across the oil reservoir is observed, which results into
creation of a localized region of intense temperature between the
A transformer is a static electrical device that operates using the core and windings and between the windings, known as the hot
principle of mutual induction. During the power transfer some spot. The hot spot causes degradation of the insulation of the
amount of the input power is dissipated as losses, such as core loss, winding as well as the conductive components of a transformer,
hysteresis loss, eddy current loss and stray loss. These losses are of which nally results in a continuous sparking. Because of this, the
the order of 1% of its full rated load and can be minimized by (i) insulating oil decomposes, forms an oil-carbon and decreases ash
using superior magnetic material for the core, (ii) tuning the point of the oil. All these eventually lead to the failure of a
thickness of the steel lamination, (iii) by closing the leakage ux transformer.
lines, etc. With the increasing power rating of the transformer, The hot spot temperature (HST) determines the normal lifespan
these losses also increase, results in the temperature rise inside the of a transformer on loading. The permitted value of HST at rated
transformer. The upper limit of temperature rise in the individual load is specied by ANSI (American Standards Institute). This is a
parts of a transformer must meet the criteria as dened in the function of resistive temperature rise in the winding [2]. To forestall
relevant standards [1,2]. To conform to these specications, one and/or trim down the hot spot winding temperature, an auxiliary
needs cooling mechanisms to disperse the high temperature cooling mechanism with ns or pumping devices to spread the oil
induced by the losses. The typical cooling mechanism used for this inside the transformer tank is often employed. However, addition
is a natural air ow or insulating oil ow. of ns requires more space and increases the overall weight of the
In an oil cooled distribution transformer, the heat transfer transformer. On the other hand, oil pumps are cumbersome,
mechanism is grounded on the Archimedes law. But this is rela- consume power and need regular maintenance. The alternate way
tively poor and less efcient, due to which a large temperature to enhance the heat dissipation is to modify the transformer oil
cooling properties.
Potential to replace the insulating oil by magnetic uid to
* Corresponding author. eliminate hot spot in the transformer core and its winding is
E-mail address: kinnariparekh.rnd@charusat.ac.in (K. Parekh).
http://dx.doi.org/10.1016/j.ijthermalsci.2015.12.012
1290-0729/ 2015 Elsevier Masson SAS. All rights reserved.
36 J. Patel et al. / International Journal of Thermal Sciences 103 (2016) 35e40
observed by many researchers [3e21]. The idea is to use a magnetic conductivity with temperature for TCF-56 and pure oil TASHOIL-50.
uid (ferrouids) whose Curie temperature is close to the device's The value of l for TCF-56 is 2.5% higher compared to that of the base
temperature (393 K) so that convection due to magnetic buoyancy oil at all the temperatures under investigation (303e338 K). In
is achieved [3]. The resultant convective ow transfers the heat, addition, under the inuence of parallel magnetic eld, thermal
becomes colder and regains its magnetization to complete the conductivity of TCF-56 shows a further enhancement of 38% at
regenerative cycle [22]. This offers opportunities in the passive 0.226 T eld (Fig. 1d). The specic heat capacity of uid, (cp), (Fig. 1e)
cooling of electrical circuits, machinery, and processes [23]. It is to at 303 K (derived using the value of thermal conductivity and
be added here that, Bahiraei et al. [24e26] have studied hydro- density) is lower than that of TASHOIL-50. This agrees with earlier
thermal characteristics of water based MneZn ferrite nanouid observation by Anne et al. for 13 different nanoparticles-base uids
within annulus, square cavity and double-pipe heat exchanger in compared to the base uid [30]. The temperature dependent vis-
the presence of a magnetic eld. According to this report, heat cosity (h) of the uid is measured using Anton Paar rheometer
transport increases between uid and wall. (model Rheolab QC) in the range of 293e338 K (Fig. 1f). Both uids
Numerous theoretical calculations for 2D [20] and 3D [21] exhibit Arrhenius behavior and value of activation energy is
model transformer working under nominal condition indicates 6.18 0.02 and 6.08 0.02 kcal/mol, respectively for pure oil and
hot spot temperature lowers by 10e30 K when magnetic uid is magnetic uid which remains nearly same. This indicates that at this
used. Nevertheless, the experimental results [16e20] reported by V. concentration, ow rate will not be hindered by viscosity change.
Segal [17] have shown 3.6 K reductions in winding temperature. The electrical resistivity is measured as per the IS 6262/6103
Similarly, Stoian [18] has observed 3.4 K reduction inside the coil standard using I.R Tester and Oil dissipation factor meter purchased
core temperature powered by a 50 Hz AC power. The magnetic uid from Power Electronical, Nashik, India. The resistivity of TASHOIL-
used in the experiments was magnetite dispersed uid having a 50 is of the order of 8 1011 U m and that of TCF-56 is
high Curie temperature 858 K [27]. Hence, to attain to the theo- 7.4 108 U m. The value of resistivity decreases by three orders of
retical value of temperature decrement, one requires a magnetic magnitudes. The decrease in resistivity of the uid is because of the
uid whose Curie temperature is near to the device temperature lower value of electrical resistivity of the particles (~101 U m) [31]
and should also have high pyromagnetic coefcients in the oper- compared to that of the insulating oil (~1011 U m). Similarly,
ating range of temperature. This is a motivation behind this study. breakdown voltage (BDV) (model PE-OBDV-6, Power Electronical,
In the present work, we have designed a MneZn magnetic uid Nashik, India) is measured for both samples as per IS 6792 using the
having Curie temperature (380 K) near to the device temperature. gold plated spherical electrodes xed at 2.5 mm gap. The average of
The uid has a high pyro-magnetic co-efcient, (vM/vT 177 A/ ve measurements was taken. The average value of the BDV for
m K) this is nearly ve times higher than magnetite (vM/ TCF-56 (69.5 kV) shows 115% enhancement compared to TASHOIL-
vT 29.8 A/m K), in the working range of temperature 50 (32.3 kV). This superior electrical breakdown performance
(373e393 K). To check the efciency of the synthesized magnetic compared to that of pure oil is due to the electron charging of the
uid as a coolant, the experiment was conducted on a custom nanoparticles to convert fast electrons from eld ionization to slow
designed single phase 3 kW prototype transformer for the over- negatively charged nanoparticles charge carriers with effective
loading conditions. The normal life expectancy under overloading mobility reduction by a factor of about 1 105 [32]. The parameters
condition is estimated using the results. obtained at 303 K for TASHOIL-50 and TCF-56 are presented in
Table 1.
2. Experimental
2.2. Design and fabrication of prototype transformer
2.1. Synthesis and characterization of magnetic uid
A 3 kW (230 V, 13 A) prototype transformer was designed and
A chemical co-precipitation method is used to prepare magnetic fabricated to conduct the experiment. The core of the transformer
nanoparticles of MneZn ferrites. The particles were coated with was made up of cold rolled Non Grain Oriented steel laminations
oleic acid and dispersed in transformer oil (TASHOIL-50, Tashkent (CRNO) material with core dimension of 2.58 103 square meter
transformer oil, Baroda, India) conforming to IS: 335/1993 [15]. The and 7 number core sizes. 18 SWG wire was used for winding the
magnetic property of the transformer coolant uid (TCF-56) is primary and secondary coil. A mild steel (MS) cabinet of 6 mm
measured using vibrating sample magnetometer (model LakeShore thickness was made with volume capacity of 3.5 L of the trans-
VSM 7404). Fig. 1a shows the room temperature magnetic response former oil. The output of the transformer was connected to 3
of the uid. The experimental data were tted using standard heating coils, each of 2.5 kW power rating. Fig. 2a shows the
Langevin's theory with log-normal particle size distribution. The schematic representation of a model transformer whereas Fig. 2b
value of saturation magnetization (Ms) derived from the t is shows the photograph of prototype transformer used for the
5.6 mT [15]. The temperature dependent magnetic properties are experiment. The temperature distribution in a prototype trans-
measured using modied Quincke's experiment [28,29]. Fig. 1b former under overloading condition (167%) is measured by sub-
shows the plot of the change in entropy of the uid as a function of merging the transformer in the TASHOIL-50 and magnetic uid
temperature under a constant magnetic eld of 0.1 T. The Curie (TCF-56), separately. The temperature at core (TCore), winding
temperature, thus obtained by tting the data with the equation (TWinding), top of the oil (TTop oil) and bottom of the oil (TBottom oil)
mentioned in our earlier paper [29] is 380 1 K. The pyromagnetic were monitored using four thermocouples (LM35, precision tem-
co-efcient (vM/vT) obtained using the ratio of the difference of perature sensor, 218e423 K, accuracy 0.5 K, calibrated at 294 K)
magnetization at 303 K and at the Curie temperature to the tem- positioned at these parts as shown in Fig. 2a.
perature difference. The value of vM/vT is 177 A/m K. This is nearly
ve times higher than magnetite based magnetic uid (29.8 A/m K). 3. Results & discussion
The same tube is used to measure the thermal expansion co-
efcient (b0) of the oil as well as magnetic uid. 3.1. Temperature prole of prototype transformer
The thermal conductivity (l) as a function of temperature is
measured using the principle of transient hot wire technique (Flu- The temperature prole for TASHOIL-50 and TCF-56 obtained in
con LAMBDA system). Fig. 1c shows the variation of thermal prototype transformer is shown in Fig. 3. From Fig. 3aec it is clear
J. Patel et al. / International Journal of Thermal Sciences 103 (2016) 35e40 37
1.0 0.145
0.010
Exp TASHOIL-50
0.8 Thy TCF-56
0.008
0.140
(W/m-K)
0.6
S (J/K-m )
3
0.006
M/Ms 0.4 Exp
0.135
Thy 0.004
0.2 0.130
0.002
0.0 a b c
0.000 0.125
0.0 0.2 0.4 0.6 0.8 1.0 300 310 320 330 340 350 300 310 320 330 340
H (T) T (K) T (K)
ln()
1.98 TASHOIL-50
2.0
TCF-56
0.16
1.95 1.6
0.14 d 1.92 e
1.2 f
2.9 3.0 3.1 3.2 3.3
0.00 0.05 0.10 0.15 0.20 300 310 320 330 340 -3 -1
H (T) T (K) 1/T (10 K )
Fig. 1. (a) Normalized magnetic response of the uid as a function of magnetic eld. Symbols are experimental and line is a t to Langevin's theory incorporating log-normal size
distribution, (b) variation of entropy with temperature for TCF-56 uid. Symbols are experimental and line is a t [29], (c) variation of thermal conductivity for both uids as a
function of temperature, (d) magnetic eld dependent thermal conductivity for TCF-56 uid, (e) Specic heat capacity as a function of temperature and (f) viscosity as a function of
temperature. Line is a t to Arrhenius behavior.
Table 1
Properties of the TASHOIL-50 and TCF-56 at 303 K temperature. that the use of magnetic uid does alter the temperature prole,
Properties TASHOIL-50 TCF-56 and reduction in the temperature is around 14e21 K. The experi-
ments were repeated three times, and value reported in Table 2 is
r (kg/m3) @ 303 K 817 870
4 (%) 0.00 1.27
an average of three experiments. It is to be mentioned here that, the
Ms (mT) 0.0 5.6 variation obtained in each experiment is less than 1%. The reduc-
vM/vT (A/m K) 0.00 177 tion in maximum temperature at the core and winding indicates
b0 (1/K) 0.00084 0.0008 the accelerated heat dissipation with TCF-56 uid.
l (W/m K) @ 303 K 0.139 0.142
The reduced temperature in TCF-56 compared to that of
l (W/m K) @ 338 K 0.126 0.129
cp (kJ/kg K) @ 303 K 1.920 1.996 TASHOIL-50 can be explained by considering the thermo-
h (mPa s) @ 303 K 9 22 magnetic convective instability in magnetic uid. This convec-
rele (U m) @ 363 K 8 1011 7.4 108 tive instability arises from heat conduction through a magnetized
BDV (kV) 32.3 69.5
uid layer having a temperature dependent magnetization
Fig. 2. (a) Internal schematic of the transformer, indicating spacing between transformer to the surrounding tank and placement of thermocouples at various positions
(TC1 thermocouple at the core, TC2 thermocouple at winding, TC3 thermocouple at top-oil portion, TC4 thermocouple at bottom-oil portion) and (b) model prototype
transformer.
38 J. Patel et al. / International Journal of Thermal Sciences 103 (2016) 35e40
TWinding (K)
360 360
TCore (K)
340 340
Capacity: 3 kW
320 Load : ~ 5 kW 320
300 a 300 b
0 1 2 3 0 1 2 3
Time (hours) Time (hours)
400 400
380 380
360 360
340 340
320 320
c d
300 300
0 1 2 3 0 1 2 3
Time (hours) Time (hours)
Fig. 3. Winding, core, top oil and bottom oil temperature of the transformer as a function of time when submerged in TASHOIL-50 (open symbols) and TCF-56 (lled symbols).
Table 2
Temperature value of transformer after 180 min with TASHOIL-50 and TCF-56. 8 b0 rg a
>
>
>
> vM dH
Sample code Temperature (1 K) after 3 h >
> m
> 0 b
TCore TWinding TTop oil TBottom oil c p A0 d4 < vT H dz
NRa (1)
hl > > 2
TASHOIL-50 114.5 124 113 80 >
> vM A
TCF-56 100.3 103 97 84 >
> m0 0 c
>
: vT H vM
1
vH T
3.2. Normal life expectancy calculation The work has been carried out under the DST/TSG/161-G Project
sponsored by DST, Govt. of India, New Delhi.
Aging or deterioration of insulation is a function of temperature,
moisture, and oxygen content. With modern oil preservation sys- References
tems, the moisture and oxygen (contributes to deterioration of
insulation life) can be minimized, leaving the insulation tempera- [1] IEEE Std C57.91, IEEE Guide for Loading Mineral-Oil-Immersed Trans-
ture as the controlling parameter. In aging studies, general ten- formers and Step-Voltage Regulators, IEEE Power & Energy Society, USA,
2011.
dency is to consider the aging effects produced by the highest [2] Facilities Instructions, Standards, and Techniques, Permissible Loading of Oil-
temperature (hot spot), as that portion will ordinarily undergo Immersed Transformers and Regulators, vol. 1e5, April 1991.
higher deterioration. Normal life expectancy in a distribution [3] K. Raj, R. Moskowitz, Ferrouid-Cooled Electromagnetic Device and Improved
Cooling Method, US patent US5,462,685, 1995.
transformer at a continuous hot spot temperature of 383 K is re-
[4] V. Segal, D. Nattrass, K. Raj, D. Leonard, Accelerated thermal aging of
ported as 20.55 years, equivalent to 1,80,000 h [1]. Experimental petroleum-based ferrouids, J. Magn. Magn. Mater. 201 (1999) 70e72.
evidence indicates that the insulation deterioration with time and [5] V. Segal, A. Rabinovich, D. Nattrass, K. Raj, A. Nunes, Experimental study of
temperature follows a theory of Arrhenius reaction rate. This theory magnetic colloidal uids behavior in power transformers, J. Magn. Magn.
Mater. 215e216 (2000) 513e515.
relates the accelerated aging factor, FAA to a hot spot temperature [6] V. Segal, K. Raj, An investigation of power transformer cooling with magnetic
given as [1,39,40]. uids 5 (1998) 416e422.
40 J. Patel et al. / International Journal of Thermal Sciences 103 (2016) 35e40
[7] V. Segal, A. Hjortsberg, A. Rabinovich, D. Nattrass, K. Raj, AC (60 Hz) and im- [21] L. PslarueDa nescu, A.M. Morega, M. Morega, V. Stoica, O.M. Marinica ,
pulse breakdown strength of a colloidal uid based on transformer oil and F. Noura, N. P aduraru, I. Borb th, Prototyping a ferrouid-cooled
ath, T. Borba
magnetite nanoparticles, in: Conference Record of the 1998 IEEE International transformer, IEEE Trans. Ind. Appl. 49 (3) (2013) 1289e1298.
Symposium on Electrical Insulation, Arlington, Virginia, USA, June 7e10, 1998, [22] R.E. Rosensweig, Magnetic uids, Ann. Rev. Fluid Mech. 19 (1987) 437e453.
pp. 619e622. [23] R.A. Curtis, Flows and wave propagation in ferrouids, Phys. Fluids 14 (10)
[8] J.C. Lee, W.Y. Kim, Experimental study on the dielectric breakdown voltage of (1971) 2096e2102.
the insulating oil mixed with magnetic nanoparticles, Phys. Procedia 32 [24] M. Bahiraei, S.M. Hosseinalipour, M. Hangi, Numerical study and optimiza-
(2012) 327e334. tion of hydrothermal characteristics of MneZn ferrite nanouid within
[9] B. Berkovsky, V. Bashtovoi (Eds.), Magnetic Fluids and Applications Handbook, annulus in the presence of magnetic eld, J. Supercond. Nov. Magn. 27
Begell House, New York, 1996, p. 831. (2014) 527e534.
[10] D. Bica, Preparation of magnetic uid for various applications, Rom. Rep. Phys. [25] M. Bahiraei, M. Hangi, Natural convection of magnetic nanouid in a cavity
47 (1995) 265e272. under non-uniform magnetic eld: a novel application, J. Supercond. Nov.
[11] M. Chiesa, S.K. Das, Experimental investigation of the dielectric and cooling Magn. 27 (2014) 587e594.
performance of colloidal suspensions in insulating media, Colloids Surfaces A [26] M. Bahiraei, M. Hangi, Investigating the efcacy of magnetic nanouid as a
Physicochem. Eng. Aspects 335 (2009) 88e97. coolant in double-pipe heat exchanger in the presence of magnetic eld,
[12] I. Nkurikiyimfura, Y. Wang, Z. Pan, Heat transfer enhancement by mag- Energy Convers. Manag. 76 (2013) 1125e1133.
netic nanouidsda review, Renew. Sustain. Energy Rev. 21 (2013) [27] R.E. Rosensweig, Ferrohydrodynamics, Dover Publications, 1997.
548e561. [28] K. Parekh, R.V. Upadhyay, Temperature dependence of magnetization of
[13] M. Timko, P. Kopcansky, M. Molcan, L. Tomco, K. Marton, S. Molokac, magnetic uid using Quincke's method, Indian J. Pure Appl. Phys. 35 (1997)
P. Rybar, F. Stoian, S. Holotescu, A. Taculescu, Magnetodielectric proper- 523e528.
ties of transformer oil based magnetic uids, Acta Phys. Pol. A 121 (2012) [29] K. Parekh, R.V. Upadhyay, R.V. Mehta, Magnetocaloric effect in temperature-
5e6. sensitive magnetic uids, Bull. Mater. Sci. 23 (2000) 91e95.
[14] G.Y. Jeong, S.P. Jang, H.Y. Lee, J.C. Lee, S. Choi, S.H. Lee, Magnetic-thermal- [30] A.K. Starace, J.C. Gomez, J. Wang, S. Pradhan, G.C. Glatzmaier, Nanouid heat
uidic analysis for cooling performance of magnetic nanouids comparing capacities, J. Appl. Phys. 110 (2011), 124323-1-5.
with transformer oil and air by using fully coupled nite element method, [31] S. Chikazum, Physics of Ferromagnetism, Clarendon Press, Oxford, 1997, p.
IEEE Trans. Magn. 49 (5) (2013) 1865e1868. 605.
[15] K. Parekh, R.V. Upadhyay, Characterization of transformer oil based magnetic [32] J.G. Hwang, M. Zahn, F.M. O'Sullivan, L.A.A. Pettersson, O. Hjortstam, R. Liu,
uid, Indian J. Eng. Mater. Sci. 11 (2004) 262e266. Effects of nanoparticle charging on streamer development in transformer oil-
[16] F.D. Stoian, S. Holotescu, A. Taculescu, O. Marinica, D. Resiga, M. Timko, based nanouids, J. Appl. Phys. 107 (2010), 014310-1-17.
P. Kopcansky, M. Rajnak, Characteristic properties of a magnetic nanouid [33] M.I. Shliomis, Magnetic uids, Sov. Phys. Usp. 17 (2) (1974) 153e159.
used as cooling and insulating medium in a power transformer, in: The 8th [34] B.M. Berkovsky, V. Bashtovoi, Instabilities of magnetic uids leading to a
International Symposium on Advanced Topics in Electrical Engineering, 2013, rupture of continuity, IEEE Trans. Magn. 16 (2) (1980) 288e297.
ISBN 978-1-4673-5980-1. [35] E.L. Resler Jr., R.E. Rosensweig, Magnetocaloric power, AIAA J. 2 (8) (1964)
[17] V. Segal, Colloidal Insulating and Cooling Fluid, US patent 5,863,455, 1999. 1418e1422.
[18] F.D. Stoian, Experimental study of cooling enhancement using a Fe3O4 mag- [36] H. Matsuki, K. Yamasawa, K. Murakami, Experimental considerations on a
netic nanouid in an applied magnetic eld, J. Phys. Conf. Ser. 547 (2014) new automatic cooling device using temperature-sensitive magnetic uid,
012044. IEEE Trans. Magn. 13 (5) (1977) 1143e1145.
[19] T. Cader, S. Bernstein, C. Crowe Pullman, Magnetic Fluid Cooler Transformer, [37] B.A. Finlayson, Convective instability of ferromagnetic uids, J. Fluid Mech. 40
US patent 5,898,353, 1999. (4) (1970) 753e767.
[20] A.M. Morega, M. Morega, G. Mihoc, C. Iacob, L. Pslaru-Da nescu, V. Stoica, [38] R.E. Rosensweig, Fluid dynamics and science of magnetic liquids, Adv. Elec-
F. Nouras, F.D. Stoian, A novel, ferrouid-cooled transformer electromag- tron. Electron Phys. 48 (1979) 103e199.
netic eld and heat transfer by numerical simulation, in: 12th Interna- [39] PN-71/E-81000 transformers loading of oil-immersed transformers.
tional Conference on Optimization of Electrical and Electronic Equipment, [40] N. Rashid, Short-Time Overloading of Power Transformers (Master thesis),
2010. 2011. Stockholm, Sweden.
Francisco Jos Iglesias Ruiz Anlisis numrico del comportamiento fluido-
trmico-magntico de un transformador monofsico
DOCUMENTO N. PLANOS
1
Francisco Jos Iglesias Ruiz Anlisis numrico del comportamiento fluido-
trmico-magntico de un transformador monofsico
2
1 8
2
LISTA DE COMPONENTES
3 ELEMENTO CANTIDAD COMPONENTE
1 120 Chapa_Nucleo_Superior
2 120 Chapa_Nucleo_Inferior
3 2 Aislante_Lateral
4 4 1 Soporte
5 2 Clemas
8 1 Tapa_Superior_Devanados
7 1 Tapa_Inferior_Devanados
6 1 Devanado_Baja_Tensi n
9 1 Devanado_Alta_Tensi n
Tipo de documento Creado por:
PLANOS TRANSFORMADOR Francisco Jos Iglesias Ruiz
T tulo. T tulo suplementario Aprobado por Rev.
E.T.S.I Industriales T.
Explosi n del transformador Referencia t cnica Idioma
ES
Fecha N de Plano (Titulaci n) Hoja
Escala 1:4 01
ALZADO PERFIL 66,00
159,50
66,70
158,00
PLANTA
114,40
ES
Fecha N de Plano (Titulaci n) Hoja
Escala 1:2 02
1
4
0
7,8
D
A
0,55
8,80 8,80
A (5: 1)
114,40
25,00
58,58
66,00
2
B 0,55
103,00
D(1:1) E(1:1)
4,20
0
,0
16,30
R2
132,40 25,00 5,60
R6,0 2,34
0
5,39
R6,0
16,49
150.00 0
R6,00
4,68
0
3 4,21
,5
C ( 5:1 )
R3 4,00
C
78,00
0,20
00
0,70 1,40
,
R2
5,70
Escala ( 1:1 ) 1,20 7,97
5 F-F( 1 : 1 )
0,80
10,00
6,00
10,00
F G
10,00 22,00
10,00
12,00
R2
,00
36,00
F 4,00 20,00 I ( 4: 1 )
38,00 Escala ( 1:2 )
R2
6 0
,0
,00
R8
7,00
I
H(2:1)
80,00
6,50 0,90 R1,00
J J
28,00
0,70
H 64,00
5,00 K(4:1)
J-J ( 1 : 2 )
1,00
K
12,35
0
00 ,0
R5, 2
3,00 R2
00
92,00
,
O
3,14
R1
76,00
123,70 L
0,65 20,89
00
,
R8
P M M
108,00
0,20
6,20
P(1:1)
1,00
,00
100,00
50,00
R12
N(4:1)
23,20
1,00
Q( 1 : 1 ) 2,00 6,00
8 1,00 M-M ( 1 : 2 )
5,65 N
66,00
27,95
R5,00
00
R1,
66,00
23,85 76,00 Q
0,65 3,14
R12,00
R(1:1)
3,20
100,00
50,00
R
1,00
7,20
R2
,0