anotações - cerpch

Transcrição

anotações - cerpch

Comitê Diretor do CERPCH
Director Committee
CEMIG / FAPEPE / IEE-USP / FURNAS /
IME / ELETROBRAS / ANEEL / MME
Comitê Editorial
Editorial Committee
Presidente - President
Geraldo Lúcio Tiago Filho - CERPCH/UNIFEI
Editores Associados - Associated Publishers
Adair Matins - UNCOMA - Argentina
Alexander Gajic - University of Serbia
Alexandre Kepler Soares - UFMT
Ângelo Rezek - ISEE/UNIFEI
Antônio Brasil Jr. - UnB
Artur de Souza Moret - UNIR
Augusto Nelson Carvalho Viana - IRN/UNIFEI
Bernhard Pelikan - Bodenkultur Wien - Áustria
Carlos Barreira Martines - UFMG
Célio Bermann - IEE/USP
Edmar Luiz Fagundes de Almeira - UFRJ
Fernando Monteiro Figueiredo - UnB
Frederico Mauad - USP
Helder Queiroz Pinto Jr. - UFRJ
Jaime Espinoza - USM - Chile
José Carlos César Amorim - IME
Marcelo Marques - IPH/UFRGS
Marcos Aurélio V. de Freitas - COPPE/UFRJ
Maria Inês Nogueira Alvarenga - IRN/UNIFEI
Orlando Aníbal Audisio - UNCOMA - Argentina
Osvaldo Livio Soliano Pereira - UNIFACS
Regina Mambeli Barros - IRN/UNIFEI
Zulcy de Souza - LHPCH/UNIFEI
04
Editorial
Editorial
Mercado
Market
06
Cenário de energia mais cara
The most expensive energy scenario
Artigos Técnicos
Technical Articles
TECHNICAL COMMITTEE
Agenda 42
Prof. François AVELLAN, EPFL École Polytechnique Fédérale de Lausanne,
Switzerland, [email protected], Chair;
Prof. Eduardo EGUSQUIZA, UPC Barcelona, Spain, [email protected], Vice-Chair;
Dr. Richard K. FISHER, VOITH Hydro Inc., USA, [email protected], Past-Chair;
Mr. Fidel ARZOLA, EDELCA, Venezuela, [email protected];
Dr. Michel COUSTON, ALSTOM Hydro, France, [email protected];
Dr. Niklas DAHLBÄCK, VATENFALL, Sweden, [email protected];
Mr. Normand DESY, ANDRITZ Hydro Ltd., Canada, [email protected];
Prof. Chisachi KATO, University of Tokyo, Japan, [email protected];
Prof. Jun Matsui, Yokohama National University, [email protected];
Dr. Andrei LIPEJ, TURBOINSTITUT, Slovenija, [email protected];
Prof. Torbjørn NIELSEN, Norwegian University of Science and Technology, Norway,
[email protected];
Mr. Quing-Hua SHI, Dong Feng Electrical Machinery, P.R. China, [email protected];
Prof. Romeo SUSAN-RESIGA, “Politehnica” University Timisoara, Romania, [email protected];
Prof. Geraldo TIAGO F°, Universidade Federal de Itajubá, Brazil, [email protected].
Schedule
Ficha catalográfica elaborada pela Biblioteca Mauá –
Bibliotecária Margareth Ribeiro- CRB_6/1700
R454
Revista Hidro & Hydro – PCH Notícias & Ship News, UNIFEI/CERPCH,
v.1, 1998 -- Itajubá: CERPCH/IARH, 1998 – v.15, n. 59, out./dez. 2013.
Expediente
Editorial
Editor
Coord. Redação
Jornalista Resp.
Redação
Colaborador
Projeto Gráfico
Diagramação e Arte
Tradução
Revisão
Impressão
Geraldo Lúcio Tiago Filho
Camila Rocha Galhardo
Adriana Barbosa MTb-MG 05984
Adriana Barbosa
Camila Rocha Galhardo
Fabiana Gama Viana
Angelo Stano
Net Design
Lidiane Silva
Cidy Sampaio
Av. BPS, 1303 - Bairro Pinheirinho
Itajubá - MG - Brasil - CEP: 37500-903
e-mail: [email protected]
[email protected]
Fax/Tel: +55 (35)3629 1443
2
Trimestral.
Editor chefe: Geraldo Lúcio Tiago Filho.
Jornalista Responsável: Adriana Barbosa – MTb_MG 05984
ISSN 1676-0220
1. Energia renovável. 2. PCH. 3. Energia eólica e solar. 4. Usinas hi_
drelétricas. I. Universidade Federal de Itajubá. II. Centro Nacional de Re_
ferência em Pequenas Centrais Hidrelétricas. III. Título.
Joana Sawaya de Almeida
Patrícia Kelli Silva de Oliveira
Editora Acta Ltda
Hidro&Hydro - PCH Notícias & SHP News
é uma publicação trimestral do CERPCH
The Hidro&Hydro - PCH Notícias & SHP News
is a three-month period publication made by CERPCH
Tiragem/Edition: 6.700 exemplares/issues
contato comercial: [email protected] / site: www.cerpch.org.br
11
Universidade Federal de Itajubá
ISSN 1676-0220
ISSN 1676022-0
00059
9 771676 022009
HIDRO&HYDRO - PCH NOTÍCIAS & SHP NEWS | ISSN 1676-0220
EDITORIAL
Dear readers,
Prezado Leitor,
Nos últimos meses do ano o setor elétrico vem sofrendo com sucessivos problemas estruturais que refletiram no mercado de comercialização
de energia. Isso ocasionará um aumento das tarifas elétricas que impactará no bolso dos consumidores.
Diante desse cenário a revista Hidro&Hydro traz nessa edição uma
reportagem onde faz um panorama dos projetos desenvolvidos pelo setor,
além de mostrar seus entraves.
Segundo os especialistas ouvidos pela revista tanto o mercado regulado quanto o mercado livre estão instáveis, uma vez que as mudanças
setoriais que começaram em 2012 deram uma travada neste segmento.
Além da situação sobre o mercado de comercialização de energia abordado na revista, nesta edição o leitor pode conhecer pesquisas técnicas
realizadas por pesquisadores de universidades brasileiras e estrangeiras.
A revista aproveita para desejar à todos colaboradores e leitores boas
festas!
In the final months of the year, the electric sector has suffered successive structural problems, which were reflected in the energy trading
market. This leads to an increase in electricity tariffs that will impact the
wallets of consumers.
Considering this scenario, the magazine, Hidro&Hydro, features, in
this edition, an overview of the projects developed by the industry, aside
from showing their barriers.
According to the specialists interviewed by the magazine, the regulated market is as unstable as the free market, once the sectoral changes
that began in 2012, which held up this segment.
Besides the energy trading market situation highlighted in the
magazine, in this edition the reader can get to know technical research
achieved by researchers from Brazilian and foreign universities.
The magazine would like to take this opportunity to wish “Happy
Holidays!” to all of its collaborators and readers.
Apoio:
IAHR DIVISION I: HYDRAULICS
TECHNICAL COMMITTEE: HYDRAULIC MACHINERY AND SYSTEMS
4
MERCADO
HIDRO&HYDRO: PCH NOTÍCIAS & SHP NEWS, 59 (4), OUT,DEZ/2013
CENÁRIO DE ENERGIA MAIS CARA
Da Redação Translation: Romulo Vilas boas Chiaradia
Arquivo Furnas
Atrasos na entrada de projetos, falta de conexão para usinas
eólicas, leilões vazios, reservatórios em queda livre, despacho
máximo de térmicas, sucessivas cargas recordes, adiamento do
sistema de bandeiras, seguidos socorros emergenciais ao mercado. A combinação de fatores conjunturais e estruturais coloca o
preço da energia no país a ponto de uma explosão. Especialistas
do setor estimam um aumento tarifário nos próximos anos entre
10% e 25% com as seguidas mudanças setoriais e as seguidas
medidas de socorro ao mercado.
Para aliviar a situação, o governo decidiu repassar para as
concessionárias mais R$ 4 bilhões via Conta de Desenvolvimento
Energético (CDE) e autorizar a Câmara de Comercialização de
Energia Elétrica (CCEE) captar mais R$ 8 bilhões no mercado
financeiro para dar suporte às operações no mercado livre. Uma
ajuda de R$ 12 bilhões. Em 2013, o governo já tinha feito uma
repasse bilionário, que será pago pelos consumidores e contribuintes ao longo do tempo.
"A situação é delicada. Este recurso será suficiente para aliviar o caixa das empresas ao longo de 2014", observa Ricardo
Savoia, diretor de Regulação e Gestão Energética da Thymos Energia e Consultoria. Nas projeções da consultoria, o desequilíbrio
de caixa das distribuidoras pode chegar a R$ 35 bilhões.
Neste valor, computase R$ 12 bilhões referentes ao despacho térmico e R$ 23 bilhões da
exposição
involuntária
das distribuidoras, na
faixa de 3,2 mil MW médios devido ao insucesso
ou baixa contratação nos
dois leilões A-1 do ano
passado.
A descontratação só
deixa um caminho para as
distribuidoras: comprar
energia no mercado livre
que tem o preço balizado
pelo Preço de Liquidações
das Diferenças (PLD). Só
para se ter a ideia da
bomba relógio armada,
a CCEE estabe-leceu o
valor máximo de PLD, R$
822,83 por MWh, para
a contratação de energia em todos os quatro
submercados, na semana
de 22 a 28 de março.
Com os reservatórios em baixa, não há saída para o mercado
se não despachar as usinas térmicas a um custo mais elevado.
Os reservatórios do Sudeste, no dia 23 de março, operavam com
35,7% da capacidade, enquanto os do Nordeste, com 41,7%.
As vazões observadas estão muito longe da média histórica. Ou
seja, a tendência é de PLD elevado o ano inteiro. Ou seja, as
distribuidoras terão que comprar energia com um custo mais el-
Arquivo Furnas
Com um conjunto de problemas estruturais e conjunturais para serem resolvidos, especialistas
do setor elétrico estimam aumento de 10% a 25% no custo da energia para os próximos anos
6
As fotos que ilustram esta matéria/artigo foram registradas
durante o Fórum de Comercialização de Energia:Outlook 2014 e
gentilmente cedidas por Furnas e pela Blue Ocean Business Events,
responsáveis pela promoção e realização do evento.
The photos that illustrate this issue/article were recorded during
the Forum of Energy Trading: Outlook 2014, and courtesy of
Furnas and Blue Ocean Business Events, responsible for the
promotion and realization of the event.
evado para repassar ao consumidor somente na época de seu
reajuste tarifário.
Para Savoia, o valor médio do PLD deve ficar entre R$ 300 e
R$ 400/MWh, se o período chuvoso não for bom o suficiente para
recompor a energia acumulado nos reservatórios. Ou seja, um
valor que pode até mesmo atrapalhar a estratégia do governo
de realizar um leilão A-0, no final de abril, para tentar reduzir o
nível de exposição das distribuidoras, quadro desenhado pela
MP 579, convertida depois na lei 12.783, que definiu o modelo de
renovação das concessões.
Com 36.321 MW, ou seja 27% da capacidade instalada do
país, as térmicas são o principal seguro para momentos de crises
hidrológicas como o atual. O volume despachado influencia diretamente no custo da energia para o consumidor.
MARKET
THE MOST EXPENSIVE ENERGY SCENARIO
Translation: Joana Sawaya de Almeida
With a set of structural and cyclical problems to be solved, the electric industry specialists
estimate an increase of 10% to 25% in energy costs for the next year
Arquivo Furnas
lion. In 2013, the government had already been made billionaire, paid by consumers and taxpayers over time.
"The situation is delicate. This feature will be sufficient to
relieve the case of companies throughout 2014," says Ricardo
Savoia, director Thymos Regulation and Energy Management
and Energy Consulting. The consultancy’s projection is that the
imbalance of the distribution box may reach $35 billion.
This value computes to R$12 billion relative to thermal dispatch and £ 23 billion of involuntary exposure of distribution in
the range of 3200 MW due to failure or low recruitment in both
A-1 auctions last year.
The decontracting just leaves one choice for distributors:
buy energy on the open market that has the price marked by
Price Liquidations of Differences (PLD). Just to give you the idea
of a ticking time bomb, CCEE established the maximum PLD,
R$822.83 per MWh for energy contracting in all four submarkets
in the week of the 22nd to the 28th of March.
With reservoirs low, there is no exit for the market if they do
not dispatch the thermal plants at a higher cost. On 23 March,
the reservoirs in the Southeast had a 35.7% of operating capacity, while those in the Northeast had 41.7%. Flow rates are
observed far from the historical average. That is, the tendency
of PLD is high all year. In other words, the distributors have to
buy electricity at a higher cost to pass on to the consumer only
at the time of their tariff adjustment.
To Savoia, the average value of PLD should be between
R$300 and R$400/MWh, if the rainy season is not fruitful
enough to reconstruct the energy accumulated in the reservoirs.
In other words, a value
that can even hinder
the government's strategy to conduct an A-0
auction in late April to
try to reduce the level
of exposure of distributors, frame designed by
MP 579, converted after
the law 12.783, which
set the model for the renewal of concessions.
With 36,321 MW,
or 27% of the installed
capacity of the country, thermals are the
primary insurance for
moments of hydrological crises like the current one. The volume
shipped directly influences the cost of energy
to consumers.
In a recent media
interview, Edvaldo Santana, former director of
the National Electric En-
Delay in the entry of projects, lack of connection for wind
farms, empty auctions, reservoirs in free fall, maximum order
of thermal plants, successive record loads, deferring the flag
system, a series of emergency aids for the market. The combination of structural situational factors puts the price of energy
in the country on the verge of an explosion. Industry experts
estimate a tariff increase in the coming years between 10% and
25% with the sectoral shifts and continued relief measures to
rescue the market.
To alleviate the situation, the government decided to pass
the utilities an additional R$4 billion through the Energy Development Account (CDE) and authorize the Trading Chamber
(CCEE) to raise an additional U.S. $8 billion in the financial market to support open market operations. An aid of U.S. $12 bil-
7
MERCADO
Em entrevista recente à imprensa, Edvaldo Santana, ex-diretor
da Agência Nacional de Energia Elétrica (Aneel), apontou em 13 mil
MW de térmicas o limite operacional para evitar uma maior pressão
do preço da energia para o consumidor e o caixa do Tesouro. Entre
a última semana de janeiro e a primeira de fevereiro, por exemplo,
o governo autorizou despacho térmico de 16.300 MW.
Segundo ele, até esta faixa, o preço médio ficaria em cerca de
R$ 250/MWh, pois só seriam despachadas as usinas mais baratas, como as nucleares, as unidades a gás natural e aquelas a
carvão. Se as térmicas a óleo combustível forem utilizadas, o
preço ficará muito mais elevado, uma vez que o custo dessas
usinas podem chegar até R$ 1.700/MWh.
Mercado livre
Arquivo Furnas
Se no ambiente regulado o cenário é de instabilidade, inclusive
com risco de inadimplência na distribuição por conta do peso da
descontratação, no mercado livre a situação também exige cuidados. As mudanças setoriais que aconteceram a partir de 2012
deram uma boa trava a este segmento. Na verdade, a participação
8
do Ambiente de Contratação Regulado (ACR) está estável desde
2010, na faixa de 25% do consumo de energia do país.
"O ano de 2014 começou em 2012, com a edição de um conjunto
de medidas que afetaram o mercado livre. Uma delas é a portaria
455, que acaba com o registro ex-post dos contratos. Isso aumenta
o preço da energia em 5% para os consumidores livres especiais sem
trazer nenhum benefício", comenta Reginaldo Medeiros, presidente
da Associação Brasileira dos Consumidores de Energia (Abraceel).
Na avaliação de Medeiros, a MP 579 provocou uma grande
interferência no mercado em busca de reduzir o preço da energia para o consumidor. "Quanto menor a intervenção do governo
no mercado, melhor para a redução do preço. Os mecanismos
de mercado são mais eficazes para reduzir os preços", analisa o
executivo, temendo que as mudanças provocadas pela portaria
455, por exemplo, levem a uma corrida judicial.
Para Márcio Sant"Anna, sócio-diretor da Ecom Energia, a MP
579 provocou uma redução no número de migração para o mercado livre. "Também houve impacto nos preços, que ficaram mais
elevados", observa.
MARKET
NEWS
Arquivo Furnas
ergy Agency (Aneel), indicated 13000 MW as the operational
thermal limit to avoid further pressure from energy prices for
consumers and Treasury cash. Between the last week of January and February 1st, for example, the government authorized
thermal dispatch of 16,300 MW.
According to him, to this range, the average price would be
around R $ 250/MWh, because only the lowest plants would be
dispatched, such as nuclear, natural gas and coal units. If the
thermal fuel oil is used, the price will be much higher, since the
cost of these plants can reach £ 1.700/MWh.
In evaluating Medeiros, MP 579 caused considerable interference in the market in search of lower energy prices for consumers.
"The less government intervention in the market, better for the
price reduction. Market mechanisms are more effective to reduce
prices," says the executive, fearing that the changes caused by
Ordinance 455, for example, lead to a legal race.
To Márcio Santana, managing partner of Ecom Energy, MP
579 caused a reduction in migration to the free market. "There
was also an impact on prices, which were higher," he notes.
Free Market
If in the regulated environment is an instable scenario, including the risk of default in the distribution due to the weight
of decontracting, the open market situation also requires care.
The sectoral changes that happened in 2012 resulted in many
hang-ups in this sector. Indeed, the participation of the Regulated Contracting Environment (ACR) is stable since 2010, around
25% of energy consumption in the country.
"The year 2014 began in 2012 with the publication of a set
of measures that affected the free market. One is the ordinance
455, which ends with the ex-post record contract. This increases
the price of electricity by 5% for special free consumers without
bringing any benefits," said Reginaldo Medeiros, president of
the Association of Energy Consumers (Abraceel).
Curta nossa fan page:
www.facebook.com/cerpchoficial
@cerpch
9
AGENDA/SCHEDULE
EVENTOS EM FEVEREIRO DE 2014
Dia 18 e 19 – Conferência Waste-to-Energy 2014
Local: Quality Moema - Av. Rouxinol , 57 • Moema • São Paulo • SP Tel
direto do hotel: [11] 2197-7100
E-mail: [email protected]
Site: www.paginasustentavel.com.br/index.php?option=com_content&vie
w=article&id=1821:waste-to-energy-2014&catid=6:eventos&Itemid=4
Dia 21 a 23 – I
CRE 2014 - International Conference on Renewable
Energy 2014
Local: Pune – Índia
Site: www.saise.org/icre2014
Dia 24 e 25 – Fórum de Comercialização de Energia: Outlook 2014
Local: Auditório de Furnas - Rio de Janeiro – RJ
Site: www.blueoceanevents.com.br/conteudo/detalhe/nossos-eventos/
frum-de-comercializao-de-energia
Dia 26 e 27 – Mexico WindPower 2014
Local: Centro Banamex - Cidade do México – México
Site: www.mexicowindpower.com.mx/
Dia 27 e 28 – 4th Annual Smart Grids Smart Cities Forum
Local: Polonia
Site: http://energy.flemingeurope.com/smart-grids-smart-cities-forum
Dia 27 – ICE Nuclear 2014; Developing the UK's Industry
Local: Londres, UK
Site: http://www.ice-conferences.com/ice-nuclear-2014/
EVENTOS EM MARÇO
Dia 1 e 2 – 1
st International Conference on Energy, Environment
and Sustainable Economics
Local: Bangkok – Tailândia
Site: http://iceese.org/
Dia 5 e 7 – BioEnergy Italy 2014
Local: Itália
Site: www.bioenergyitaly.com
Dia 5 e 3 – 1
0th Energy Efficiency and Renewable Energy Congress and Exhibition for South-East Europe
Local: Sofia-Capital - Bulgária Site: www.eea.europa.eu/events/10th-energy-efficiency-renewableenergy
Dia 11 a 13 – 4
º INOVA FV - IV Workshop Inovação para o Estabelecimento do Setor de Energia Solar Fotovoltaica
no Brasil
Local: Auditório da Faculdade de Ciências Médicas/Unicamp/Campinas-SP
Site: www.iei-la.org/inovafv/index.php
Dia 13 – Fórum Agenda Setorial 2014: Regulação e Mercado
Local: Hotel Sofitel - Atlântica, 4240 - Copacabana - Rio de Janeiro – RJ
E-mail:
Site: www.ctee.com.br/agendasetorial/
Dia 14 a 31 – 1st International e-Conference on Energies
Local: On-line
Site: www.sciforum.net/conference/ece-1/
Dia 18 – Elétrica Segura – MG
Local: Minas Gerais – MG
Site: www.abracopel.org/
Dia 23 a 25 – V
I SMARS - Seminário Brasileiro de Meio Ambiente
e Responsabilidade Social do Setor Elétrico
Local: Brasília – DF
Site: www.smars.com.br
Dia 30 a 04/04 – Light+Building 2014
Local: Frankfurt – Deutschland
Site: http://light-building.messefrankfurt.com/frankfurt/en/besucher/
willkommen.html
10
Dia 31 a 03/04 – V CBENS - V Congresso Brasileiro de Energia
Solar
Data: 31/03/2014 a 03/04/2014
Local: Recife - Pernambuco – PE
Site: www.cbens2014.com.br/
EVENTOS EM ABRIL
Dia 2 e 3 – Energy Show 2014
Local: Florianópolis - SC
Site: www.energyshow.com.br/
Dia 8 a 9 – P
CH 2014 - 6º Encontro Nacional de Investidores em
Pequenas Centrais hidrelétricas
Local: São Paulo - SP
Site: http://viex-americas.com/
Dia 14 e 15 – Cenocon 2014
Local: Pestana São Paulo Hotel & Conference Center - São Paulo - SP
Site: www.rpmbrasil.com.br/index.aspx
Dia 29 e 30 – I
X Simpósio sobre pequenas e médias centrais
hidrelétricas
Local: FIEP - Curitiba – PR
Site: www.ixspmch.com.br/
Dia 29 e 30 – 5th Annual Smart Grids Summit 2014
Local: Malaga – Spain
Site: http://thesmartgridssummit.com/
EVENTOS EM MAIO
Dia 6 e 7 – Enase 2014
Local: Hotel Sofitel - Av. Atlântica, 4240/Copacabana/Rio de Janeiro–RJ
Dia 6 a 8 – Conferência Internacional REGSA 2014
Local: Auditório da Assembléia Legislativa - Sessões - Unidade Florianópolis (Trajano e Dib Mussi) - Florianópolis – SC
Site: www.unisul.br/wps/portal/home/pesquisa-e-inovacao/seminariosde-pesquisa/conferencia-internacional-regsa
Dia 14 e 15 – 5º Seminário Internacional de Energia Nuclear
Local: Centro Empresarial Rio - Ed. Argentina - Praia de Botafogo, 228 2º andar - Auditório - Rio de Janeiro - RJ
Email: [email protected]
Site: http://planejabrasil.wordpress.com/
Dia 18 a 21 – X
III SEPOPE – Simpósio de Especialistas em Planejamento da Operação e Expansão Elétrica
Local: Bourbon Cataratas Convention & Spa Resorts - Rodovia das Cataratas, km 2,5 - CEP 85853-000 - Foz do Iguaçu – PR
Site: www.sepope.com.br/
Dia 27 e 28 – I
II Engeo - Encontro Nacional de Geoprocessamento do Setor Elétrico
Local: Usina Hidrelétrica de Itaipu - Foz do Iguaçu – PR
Site: www.engeo2014.com.br/
EVENTOS EM JUNHO
Dia 2 a 5 – InterSolar
Local: München – Deutschland
Site: http://conference.intersolar.de
Dia 3 a 5 – 1
0ª Edição - Redes subterrâneas de energia elétrica
2014
Local: Centro de Convenções Frei Caneca - São Paulo – SP
Site: www.rpmbrasil.com.br/
Dia 5 e 6 – S
EMEAR: Seminário de Meio Ambiente e Recursos
Energéticos
Local: Universidade Federal de Itajubá – UNIFEI/EXCEN
Site: www.cerpch.unifei.edu.br/semear
NOISE INDUCED BY KARMAN VORTICES IN A 7 MW FRANCIS TURBINE:ANALYSIS OF CAUSES AND SOLUTION .............................12
Pierre-Yves Lowys, Marcelo Aquino, Ricardo Andrade, Joaquim Eduardo Pereira
PRELIMINARY ASSESSMENT OF UNCERTAINTIES OF METHODOLOGIES FOR MAXIMUM FLOW RATES
DETERMINATION FOR SHPS AND ΜCHS PROJECTS IN THE BRAZILIAN CONTEXT........................................................................16
Regina Mambeli Barros, Geraldo Lúcio Tiago Filho, Marcelo daige Prado Leite, Ivan Felipe Silva dos Santos,
Fernando das Graças Braga da Silva, Jéssica dos Santos
IMPORTANCE OF DRAFT TUBE MODELING IN NUMERICAL SIMULATIONS OF HYDRAULIC TURBINES............................................. 24
Importância da modelagem do Tubo de Sucção em simulações numéricas de turbinashidráulicas
Mauricio Formaggio, Thi C. Vu, Christophe Devals, Ying Zhang, Bernd Nennemann, François Guibault
ACTIONS AND INNOVATIONS IN DESIGN OF HPP RETIRO BAIXO ............................................................................................31
Thiago Villela Torquato, Gabriel Villela Torquato, Deborah Montenegro C.F. Albuquerque, Ana Alice Cesario
TECHNICAL ARTICLES
Technical Articles Seccion
ARTIGOS TÉCNICOS
STUDY OF PROCESS OF AMENDMENTS VULCANIZATION NITRILE RUBBER FOR SEALING
VALVES USED IN BUTTERFLIES CONDUITS SHPS ..................................................................................................................35
Moisés Toigo, João Henrique Bagetti, Sergio Luis Marquezi
IAHR DIVISION I: HYDRAULICS
TECHNICAL COMMITTEE: HYDRAULIC MACHINERY AND SYSTEMS
Classificação Qualis/Capes
B5
B4
ENGENHARIAS I; III e IV
Biodiversidade
Interdisciplinar
Áreas de: Recursos Hídricos
Meio Ambiente
Energias Renováveis
e não Renováveis
A revista está indexada no DOI sob o prefixo 10.14268
11
11
NOISE INDUCED BY KARMAN VORTICES IN A 7 MW
FRANCIS TURBINE: ANALYSIS OF CAUSES AND SOLUTION
FRANCIS TURBINE:ANALYSIS OF CAUSES AND SOLUTION
Pierre-Yves Lowys
2
Marcelo Aquino
3
Ricardo Andrade
4
Joaquim Eduardo Pereira
1
ABSTRACT
The main structural parts of hydraulic turbines are designed according to specific criteria in order to avoid resonance of natural modes
with possible excitation sources. For major parts such as runner blades or stay vanes, more accurate approaches could complete these
design rules, like numerical fluid-structure calculation or experimental tests. For some phenomena like vortex shedding, determining
the real dynamic load and the response of the structure is not an easy task as the physics is complex and still difficult to model precisely.
To detail this kind of phenomena and the recent advances made with analysis tools, this paper focuses on the case of so-called Karman vortices known as one possible cause of resonance in hydraulic parts. It is illustrated by a case study of unexpected and noisy
resonance observed on a Francis turbine. The “Special Measurement Team” developed by Alstom Brazil in Taubaté was involved to find
the root cause of the problem by means of site tests on the prototype unit. The diagnostic of resonance phenomena was confirmed by
numerical simulation of the part. It helped to define an efficient way to solve the problem without major production losses and without
affecting the turbine performance levels. The feedback from this experience has been included in design rules to improve future projects of this nature that Alstom may need to research.
KEYWORDS: Francis turbine, hydraulic, analysis tools
1. INTRODUCTION
2 horizontal Francis turbines with output of 8 MW each under 90 m of water head compose the power plant. It was commissioned successfully in 2005, but an abnormal high frequency
acoustic noise was registered in the vicinity of both units. First
investigations have allowed eliminating some basic hypothesis
for the origin of the noise, but no easy way was found to locate
its origin and remove it. The main turbine components have been
periodically checked; thus even if it was not presenting any risk
for the structural integrity, the noise was recognized as disturbing
for the comfort of operation and it was important to resolve it for
the global quality of the product.
At this step, the only reliable fact was the hydraulic origin
of the noise: It was for example observed that the noise disappeared for few hours just after introducing accidentally some
“long” grass in the intake pipe of the units. After shutting down
and restarting the unit, the noise came back as the machine was
cleaned. This event allowed excluding non-hydraulic sources, like
an electrical phenomenon from the electrical generator or a mechanical vibration of any auxiliary devices.
The Picture 1 shows a typical spectrum analysis of the acoustic noise recorded near the turbine. It may be observed a principal frequency close to 3700 Hz at low load, and another one close
to 2440 Hz at highest load. At partial load, both frequencies exist
simultaneously. This behavior is most probably linked to the resonance of any structural part excited by the hydraulic flow passing
through it. The variation of the hydraulic condition (speed, direction) and / or the mechanical properties of the structure (position, rigidity…) in relation to the load of the turbine could explain
the modification of the frequencies versus the load.
This preliminary diagnostic allowed focusing the investigations on the hydraulic parts of the turbine to find the origin of
the phenomenon and the incriminated components. The objec-
tive was obviously to avoid any unnecessary modification of the
turbine without being certain of a positive result.
10%
50%
100%
Fig. 1: Spectrum FFT of the acoustic noise with increasing outputs of the
turbine
2. PRELIMINARY DIAGNOSTIC
From experience, such kind of high frequency noise may be
observed on hydraulic turbines in case of so called Karman resonance. Examples of occurrences are widely described in literature
e.g. in [Ref. 2].
The formation of periodic Karman vortices could occur in the
whole range of operation at the trailing edge of profiled components (valve body, stay vanes, guide vanes, runner blades, aeration tube…). We expect that the frequency fK of this phenomenon
increases linearly with the flow velocity W according to:
Alstom
Alstom
Alstom
4
Alstom
Hydro France, 82 Léon Blum Avenue – PO Box 75 – Cedex 9, ZIP Code 38041 – Grenoble – France
Brasil Energia e Transporte Ltda, Charles Schnneider – Pq Senhor do Bonfim, ZIP Code 12040-001 – Taubaté – SP – Brazil
12
HIDRO&HYDRO: PCH NOTÍCIAS & SHP NEWS, 59 (4), OUT,NOV,DEZ/2013, DA PÁG. 12-15
1
2
3
TECHNICAL ARTICLES
Fk = St .
W
e
[eq. 1]
W is the flow velocity at the trailing edge, e is the wake thickness and St the so-called Strouhal number.
Experimental studies have shown that the Strouhal dimensionless parameter is in the range of 0,22 to 0,25 for our kind of
hydraulic shapes, in any case lower than 0,28 [Ref. 1].
Beyond the formation of Karman vortices that generally occurs in all the operation range of any turbine, it is known that
significant amplitude of vibration will be only reach in case of
resonance. Otherwise, the vortex shedding all along the profile
doesn’t have a global coherency and won’t bring enough energy
to the structure. But as soon as the displacement amplitude becomes sufficiently large, the structural displacement will control
the fluid excitation leading to the so-called “lock-in” phenomenon. The vortex shedding frequency is therefore “locked” onto
the structural eigen frequency over some velocity range [Ref.
4]. In other words, the vortex shedding frequency will not any
more increase with the flow velocity as per [eq. 1], but will stay
constant and equal to the involved natural frequency of the component.
These theoretical aspects suggest comparing both excitation
and natural frequencies in order to find out possible resonance.
The Karman excitation frequency is estimated as peer [eq. 1].
For our purpose of diagnostic, we use the value of 0.24 as a realistic estimation of the Strouhal number (at design stage, one
usually adopt the majoring value of 0.28 in order to keep Karman
excitation frequencies safe below the first natural frequency).
The result of the calculation of Karman frequencies is given in
the Table 1. They are compared with estimated natural frequencies for each main component ( estimation s are in water through
analytical formulas or finite elements calculation).
and the surrounding water simulated by adding fluid elements.
We obtained the confirmation that the 2 eigenmodes shown on
Picture 2 may match with the noise frequency heard. For these
mode shapes, the effect of added mass of water is a lowering
of eigenfrequencies of about 12 % compared to the vibration in
air. It is interesting to observe that both of them have the main
displacement amplitude near to the trailing edge of the profile.
In fact, this is a necessary condition for a natural mode to have a
significant coupling with the Karman excitation.
Fig. 2: mode shapes of the guide vane submerged in water
3. HYPOTHESIS CONFIRMED BY FIELD TEST
3.1.First indication: test of air injection.
As far as the noise appears to be related to any hydraulic phenomenon, a simple test of air injection was first performed. It is
known that a slight airflow could have a significant action on this
hydraulic vibration.
A temporary 12 bar compressor was used to
3
inject air in any high-pressure area of
2 the turbine.
1
Table 1: Estimated Excitation and natural frequencies of main
structures.
Stay vane
Guide vane
4
Runner
blades
Velocity
(at nominal output) [m/s]
9
19
9
Thickness
[mm]
4
2
2 to 4
[Hz]
540
2300
540
to 1080 (*)
Karman frequency
Natural frequencies [Hz]
1565 (bending)
1710 (torsion)
260 (global rotation)
500 to 1300
1075 to 3510 (local
(First 4)
first 6)
(*) A Donaldson type profile is used at the outlet of the blades and is
known to cancel risk of Karman resonance. For this, the frequencies are
given for information only.
From this analytical approach, we suspect principally the
guide vanes to be at the origin of the noise. At the contrary, the
stay vanes appear with very low risk of resonance. The runner
blades are designed with a Donaldson type profile at the outlet,
which is from experience free of Karman resonance. In addition,
the expected frequency if any (500 to 1000 Hz) should be much
lower than the actual recorded noise (2440 and 3700 Hz). For
these reasons, the blades were unlikely considered to be at the
origin of the phenomenon. Thus the guide vanes were the most
suspected source of the noise.
To precise this hypothesis, a detailed calculation of natural
frequencies has been done for the guide vanes using finite elements analysis. The guide vane was made of structural elements,
1:
2:
3:
4:
intake pipe (high pressure tapes)
through the spiral case (winter-kennedy tapes)
through the head cover between stay-vanes and guide vanes
through lower cover behind the runner band (low pressure side)
Fig. 3: points of air injection
The air was injected successively in 4 different locations
shown on the Picture 3, in order to try to locate more precisely
the structure involved in the noise generation. Unfortunately, the
noise was clearly affected only by air injection in the intake pipe
(at point 1) far upstream from the turbine. Injection in other locations did not show significant effect.
It was supposed that the injection point in the spiral case
(point 2) or in the head cover (point 3) was too close to the structure to achieve a good mix of the air in the water. In other term,
13
the air stream coming from these locations could stay too close to
the walls to affect significantly the stay vanes or the guide vanes.
Only the runner blades may be excluded as reliable noise sources
from this test.
3.2.Second indication: the range of apparition
The amplitude of the main peaks of vibration at 2440 Hz and
3700 Hz is plotted against the turbine output on the Picture 4
(left). It can be seen that the range of apparition of the main
phenomenon at 2440 Hz is wider than in the case of a common
Karman resonance (Picture 4 on the right). Even if the resonance
of this mode is forced to stay constant for a wider range of flow
velocity through the “lock-in” phenomenon described above, we
can’t explain that the noise amplitude stays almost constant
above 30 % and up to 100 % of the load. At least if we suppose a
proportional increase of the velocity with the load as it is the case
around the stay vanes or the runner blades.
In fact, if we consider the guide vanes, the flow velocity near
the trailing edge does not significantly increase with the load, but
is almost constant in a wide range of output. It is due to the link
between the turbine discharge and the distributor opening. As a
simplified approach, the increase of the passage section between
two guide vanes compensates for the increase of the flow rate,
keeping the velocity almost constant.
left: as recorded on site (vibration
level vs. turbine load)
right: example of amplitude curve vs.
upstream flow velocity. Source: [4]
Fig 4: amplitude of the noise
This observation let us to believe that the guide vanes were
most probably at the origin of the phenomena because the constant flow velocity was compatible with the observed wide range
of apparition of the noise. On the contrary, the stay vanes or the
runner blades were unlikely concerned because the variation of
the flow velocity with the output should have produced a more
restricted resonance.
3.3.Last confirmation: temporary modification of the
structure
Fig. 5: temporary modification of the guide vane
As far as the air injection test couldn’t fully confirm which
structure was involved, one took advantage of a low production
period of the power plant to perform a final test. Since the guide
vanes were most probably at the origin of the noise, a temporary
modification of the profile was carried out on them using a thin
rubber sheet fixed by metallic braces as shown on the Picture
5. This device was intended to modify the wake thickness of the
guide vane, while preventing any water leakage between during
pressurization of the spiral case when the unit started up.
Such modification has led to a complete vanishing of the noise,
proving definitively that the guide vanes were at the origin of it.
4. FINAL SOLUTION
As far as the Karman vortices are generated at the outlet
edge of the profile, we studied different possible solutions to
modify this part of the guide vanes. A comparative summary of
main alternatives is given in the Table 2.
A common solution to solve Karman vibration is to sharpen
the outlet edge of the profile (see for example [Ref. 2]). The
reduction of the thickness changes the frequency of the Karman
vortices as per [eq. 1], but also reduces the amplitude of the
excitation force [Ref. 3]. Due to some of the specificities of this
project, the guide vanes outlet thickness was already lower than
usual designs. Considering this fact and the impossibility to reduce further the outlet thickness, the proposed alternatives 1,
2 and 3 were based on increasing the trailing edge thickness,
to decrease the excitation frequency below the first natural frequency of the profile according to equation [eq. 1]. Solution 4,
almost not changing the thickness, is known to reduce the excitation amplitude of the vortices by smoothing the sharp angle at
the outlet (similar to the Donaldson profile used on runner blade
outlet). The efficiency of the solution 5 to eliminate the noise was
proven during the preliminary diagnostic test with air injection,
but would have needed maintenance after implementation.
Because of its high simplicity, the retained solution was to perform a local modification of the outlet edge by smoothing the sharp
angle on one side of each guide vane (solution 4). A good access to
the distributor was achieved by dismantling the runner. Around one
gram of metal per guide vane was removed by a fine manual grinding
(see Picture 6). This operation had the main advantages to avoid dismantling of the distributor, to present no risk of structural deformation, and last but not least to have no effect on the global hydraulic
performances of the turbines (efficiency, output).
Guide vane
with rubber
Stay vane
Access from the runner side
Detail of the grinding
Fig. 6: detail of the guide vane modification
After manual grinding of the first unit, a slight noise at 2440
Hz was still audible over a reduced range around 60 % output.
In order to avoid a new dismantling of the runner, a vibration
measurement was performed on each guide vane trunnion with
an acceleration sensor and allowed to locate precisely guide vane
#11 still vibrating (see Picture 7). Thanks to this finding, an im-
14
Table 2: possible solutions to remove the noise from the guide vanes.
Type of modification
Scheme
1- N
ew set of guide vanes with
increased outlet thickness
Advantages
Reference solution
2- C
hange completely the outlet
edge (cutting and welding)
Risks and/or disadvantages
-T
otal cost and delay
-S
top time of the unit (dismantling the distributor)
-D
eformation of the profile after welding
- Stop time of the unit (dismantling the distributor)
Lower delay
3- L
ocal modification of the outlet
edge (welded or bolted piece)
Very short delay,
Lower stop time
4- L
ocal modification of the outlet
edge (grinding)
- Uncertain result
Simplicity,
Very short delay,
Lower stop time
5- Air injection in the penstock
No stop time,
Proven result
proved manual grinding was performed from the inner side of
the spiral case through the stay vanes, without dismantling the
runner and in a very short time. The complete vanishing of the
vibration was proven thanks to a final measurement (Picture 8).
e
frequ
ency
-D
eformation of the guide vane after welding
- Modification of hydraulic performances
- Not suitable for long term operation
e
id
gu
n
va
After first round of grinding
After final grinding on guide vane #11
Fig. 7: vibration level on guide vane trunnions after first and second round
of grinding (amplitude versus frequency and guide vane number)
- Cost of operation and maintenance (compressor)
way to find out the origin of the phenomenon and to implement
a solution.
The slight modification performed on the outlet edges of the
guide vanes led to completely eliminate both the vibration and
the noise generated at the turbine. In this case, the sharp angle
at the outlet edge associated with the low thickness generated
vortices of high frequency with enough energy to induce resonance with a natural modes of the structure.
The complete and reliable diagnostic was achieved thanks to
complementary approaches involving both theoretical and numerical calculation, and specialized site measurements. This is a
typical advantage of the Special Measurement Team created at
Alstom Brazilian unit in Taubate. Such team offers a large range
of measurements and diagnostic capabilities, associated with a
tight proximity to the local market. This proximity allows Alstom
to understand the customer necessities related to any particular
machine, to measure and to analyze several operation conditions,
to compare both experimental and theoretical approaches and at
the end to provide the more appropriate and accurate technical solution. All of this generates a high efficiency to perform quick and
efficient troubleshooting, to reduce the time of non-availability and
finally to increase the reliability of the generating units.
6. REFERENCES
frequ
ency
Before modification
ut
tp
ou
After modification
Fig. 8: noise level before and after final modification of the guide vanes
outlet (amplitude versus frequency and turbine output)
5. CONCLUSION
In most cases the guide vanes are not concerned by such
risk of resonance, contrary to the stay vanes or the runner
blades where similar facts were related and addressed through
specific studies. This case study has shown step by step the
• [1] JL. Deniau, 1996, “Study of stay vane vibration by hydroelastic model”, 18th. IAHR Symposium on Hydraulic Machinery and Cavitation, Valencia, Spain.
• [2] Papillon, Brooks, Deniau, Sabourin, 2006, “Solving the
guide vane vibration problem at Shasta”, Hydrovision 2006,
Portland Oregon, USA.
• [3] Mazzouji, Segoufin, Lowys, Deniau, 2006, “Investigation
of unsteadyness in hydraulic turbines”, 23th. IAHR Symposium
on Hydraulic Machinery and Cavitation, Yokohama, Japan.
• [4] Ausoni, Farhat, Escaler, Avellan, 2006, “Cavitation in Karman Vortices and flow induced vibration”, 6th. International
Symposium on cavitation CAV2006, Wageningen, Netherlands.
15
ARTIGOS TÉCNICOS
PRELIMINARY ASSESSMENT OF UNCERTAINTIES OF METHODOLOGIES
FOR MAXIMUM FLOW RATES DETERMINATION FOR SHPS AND
ΜCHS PROJECTS IN THE BRAZILIAN CONTEXT
Regina Mambeli Barros
3
Marcelo daige Prado Leite
4
Ivan Felipe Silva dos Santos
5
Fernando das Graças Braga da Silva
6
Jéssica dos Santos
1
2
ABSTRACT
In a hydropower plant (CH) design, as is the case of Small Hydro Power (SHP) and micro-hydro (μCHs), the accurate calculation of extreme flows, based on a series of data provides a better assessment of the available energy in the CH project. The present study aimed to
evaluate the uncertainties in the prediction of flood and minimum flows, as well as about the theoretical probability distributions commonly used for the calculations of these extreme flows. For this purpose, the statistical analysis were performed by using the Computational
Software System for Hydrological Analysis, Siscah® 1.0, developed by the Research Group of Water Resources, of Federal University of
Viçosa (GPRH / UFV), for two stations fluviometric: with small averages values of discharges (Fazenda da Guarda; Code 61.25 million)
and large average values of discharges (Paraíba do Sul RN; Code 58380001). Regarding to the minimum flows, it is necessary to subtract
from the design flow, the value of environmental or health flow. In some Brazilian states, this flow for maintenance of biota is defined
based on number of minimum flows, like the q7, 10. For flood flows, we used the Log-Normal distributions II, log-Pearson III, Gumbel and
Pearson III. For this latest, two scenarios were considered for obtaining the partial series, as proposed by Chaudhry [4]. Based on these
results, it was conclude about the hydrological uncertainties mentioned by Serinald (2009), namely: natural or inherent uncertainty due
to the stochastic representation of natural conditions with the randomness and complexity inherent to them; model uncertainties, and
uncertainties statistics on the estimated parameters. It was confirmed as according to described by Serinald (2009), the statistical uncertainty, related to the estimated parameter and which can be minimized by increasing the sample size. There was even a reversal from
the 10-years returning time, for which the values of the Gumbel distribution for the partial series (Scenario 2 and Scenario 4), began to
be less than for the total series (Scenario 1 and scenario 3).
KEYWORDS: Small Hydro Power, Minimum Flow, Floods stream flows, Instalated Power.
1. INTRODUCTION
There is a great resurgence in interest at world level about
the development of small hydroelectric systems. Increasing is
primarily driven by the belief that such systems, which include
mini, micro and pica systems, are a source of clean energy with
little or no adverse impact on the environment [1, 12]. Regarding to the Small Hydropower (SHP), their dimensioning is a very
critical point, since it affects not only just the return of investment, but also the maximum exploitation of hydro potential and
performance obtained by hydroelectric plant [8, 13].
In the projects for sanitation or hydraulic designs, there are
more restrictive expectations regarding the security than those
flood or drought observed, requiring inferences beyond observation. In other words, extrapolation has been required. The
best way to extrapolate the empirical probabilities is by using
the model probabilities, appropriate to this phenomenon under
discussion [4].
The design of the SHP capacity is closely linked to the discharge availability and is based on the flow duration curve analysis, which is constructed from records of the flow historical series
or is processed by probabilistic methods or prediction [4, 7 and 8].
A flow duration curve is one of the most informative for displaying the full range of river discharges, for events from reduced
flows to flooding. It is a relationship between a given flow value
and the percentage of time that this discharge is equaled or exceeded, or in other words, the relationship between the magnitude and frequency of discharges [10].
The flow range available onsite, Q, is included between the
Maximum Flow which flows for at least 1 day per year (Qmax)
and the Minimum Flow, which flows on site (Qmin), according to
Equation (1) according to Santolini et al. [8].
Qmin < Q < Qmax
(1)
1.1 Minimum discharges
The discharge provided by Equation (1) cannot be fully exploited, since it is established by Brazilian law that a Sanitary
or Environmental Discharge should be downstream released for
ecosystem maintenance purposes, regarding the existing conditions before the SHP construction. Therefore, the discharge
exploitable variation is effectively obtained from the discharge
duration curve after being subtracted this Sanitary or Environmental Discharge [9, 11, 12].
The average of annual average-minimum discharge series of
7-days at least, is known as Dry Weather Discharge. The 7-day
period, which is covered by the 7-day Average Annual Minimum
Discharge, eliminates the daily variations in the component artificial river discharge. In addition, an analysis based on a time
series of average discharges of 7 days is less sensitive to mea-
1
Civil Engineer, Phd. and Masters from PPG-SHS/EESC/USP, Phd. Professor - IRN/ UNIFEI/ National Reference Center in Small Hydro Power, Av.BPS, 1303, Itajubá-MG, CEP:
37500-903, tel.:(35) 36291224, [email protected]
2
Mechanical Engineer, Phd. in Hydraulic Systems from USP and Masters in Mechanical Engineering in Flow Machines from UNIFEI, Director and Phd. Professor - IRN/ UNIFEI/
National Reference Center in Small Hydro Power, Av. BPS, 1303, Itajubá-MG, CEP: 37500-903, tel.: (35) 36291454, fax: (35) 36291265, [email protected]
3
Master student in Energy Engineering and Mechanical Engineering from Federal University of Itajubá, National Reference Center in Small Hydro Power, [email protected]
Student in Hydraulic Engineering from Federal University of Itajubá, [email protected]
4
Civil Engineer, Phd. and Masters from PPG-SHS/EESC/USP, Phd. Professor - IRN/ UNIFEI/ National Reference Center in Small Hydro Power, Av.BPS, 1303, Itajubá-MG, CEP:
37500-903, tel.:(35) 36291485, [email protected]
5
Student in Hydraulic Engineering from Federal University of Itajubá, [email protected]
16
FOR MAXIMUM FLOW RATES DETERMINATION FOR SHPS AND ΜCHS
PROJECTS IN THE BRAZILIAN CONTEXT
TECHNICAL ARTICLES
surement errors. At the same time, in most cases there is no
great difference between 1-day and 7-day minimum discharges.
Frequency analyses of minimum discharges are part of the frequency analysis of extreme events and, as such, are covered in
many books of classical statistical [10].
In Brazil, in a number of States, Sanitary or Environmental
Discharge is obtained from the lowest minimum average discharge
calculated for 7-days of permanence and 10-years of recurrence
period (or Returning Time), e.g., the q7, 10 [11, 12]. Therefore,
q7, 10 is obtained by processing probabilistic methods for the series of data discharge, such as the Weibull distribution [4, 7].
Hence, the design turbinable discharge, Q, should be adequately selected within the range of available discharge-site
(Equation 2) [7].
(2)
Where: E is the energy resulting from P (t), which is the
power obtained by the turbine; ρg is the water density; Q(t) is
the discharge; h is the effective head (fall) (disregarding head
losses); ηhid is the hydraulic efficiency (dependent on the type
of turbine and discharge); ηmec and ηele are the mechanical and
electrical efficiency, respectively, both assumed to be constant
and equal to 0.96 and 0.94, also, respectively.
1.2.Maximum discharges
The flooding design (or design event) is the discharge value
to be attributed and is corresponding to a high probability of nonexceedance, generally expressed in terms of returning time, TR
[5, 6 and 7]. The maximum flooding discharges are calculated for
the design of civil works, such as the spillways and cofferdams
for Small Hydro Power (SHP), and the data series used for its
determination is constituted by the annual maximum discharges
series (Souza et al. [11]). When taking into account the criteria
of Eletrobras [3], in order to design the deviation works of dams
are recommended: embankment dam (TR, 100 years), rock-fill
dam (TR, 50 years), and concrete dam (TR, 25 years).
Estimation of discharge design based on fluviometric stations
data requires the selection and parameterization of a proper
probabilistic model. Various probability distributions have been
considered in studies of a number of authors [5, 9, 12, 13 and
14], in different situations, for this purpose. Typical examples
include the Gumbel distribution, as well as a Gamma distribution,
and many others, less commonly used. Souza et al. [11] mentioned that Gumbell and log-Pearson III distributions are widely
used, since it is available a track record of over 10 years of extreme discharge to be evaluated.
Regarding to the use of normal distribution, it is mentioned
that this transformation requires a data series (normalization).
In addition, these transformations are not always able to ensure
that the transformed series follow a normal distribution (Jain and
Singh, 1986 apud Yue et al. [14]). In practice, extreme events
such as peaks and volumes of flooding can often be approximated and represented by a Gumbel distribution (Gumbel, 1958
cited Yue et al [14].)
Basically, there are two different paths to a flooding analysis
problem. One of them corresponds to series of annual discharge
(streamflow annual flood series, AFS), and another series of partial discharge (streamflow partial duration series, PDS). Todorovic [13] evaluated three stochastic models based on of flooding
discharges PDS. Each model depends on certain assumptions
regarding to properties of exceedance of a base level x0. Todorovic [13] obtained good agreement between the theoretical and
observed distributions, showing that assumptions about the exceedances are not too restrictive. Concerning to the series partial
flow, Righetto [6] emphasized that these are very useful for the
characterization of the floods. A partial series is obtained from
observed discharges, considering only those that exceed a given
reference value (Rasmussen and Rosbjerg, 1989 apud Righetto
[6]). Therefore, after a value being fixed for the flooding discharge reference, Qr, it must be taken only the discharges observed with values greater than Qr.
In hydrology and other fields of study, the parameter of interest is often a given quantil, namely: for example, the 0.990 quantil of the annual flooding distribution, with a value corresponding
to 100-years returning time (Stedinger, 1983 apud Serinald [9]).
It is always necessary to assess the type of uncertainty considered in a given study. In statistical analysis, the uncertainty
sources are normally grouped into three main categories: [9]:
i.natural or inherent uncertainty, which is the randomness
and complexity of the natural process, which cannot be
reduced in any way;
ii.
statistical uncertainty, which is related to the estimated
parameter and can be minimized by increasing the sample
size; and
iii.model uncertainty, which depends on the selection of the
statistical or physical model. It cannot be reduced by the
addition of information (eg, sample size), but only by increasing the knowledge of the process, and the adoption
of more complex models.
Righetto [8] warned that regarding the magnitude of flooding, despite its assessment with reliability to be very important,
the probable errors of judgment compromise far fewer the calculations to determine the distribution of peak discharge, due to the
small variability of discharges for large returning time. However,
errors of judgment to the date of occurrence can seriously compromise the reliability of the frequency distribution of flooding.
2. METHODOLOGY
In order to evaluate the number of daily-discharges averages
along N-years of observation were set the theoretical probability
distributions described as follows, by using the use of Computational System for Hydrological Analysis (Sistema Computacional
para Análises Hidrológicas, in Portuguese), Siscah® 1.0 software, from Group for Research of Water Resources of the Federal
University of Viçosa (GPRH/UFV) [6]. The following stations were
used: Fazenda da Guarda (Code 61250000) and Paraíba do Sul
RN (Code 5838001). The hydrological year started in October,
as recommended by Souza et al. [11]. Due to the failures presence, the following data were discarded from the series (Station Code 61250000): between September/1934 to June September/1935, August to December/1965, May/December 1989,
January and February/2000, February 2001, April to May/2003,
and finally, from July to December/2003. Regarding the Station
Code 58380001, were discarded periods between January and
October/1972, April 1987, and finally in December/1987.
2.1. Minimum discharges
With regard to the minimum discharges and for comparative purposes, theoretical probability distributions were adjusted,
namely: Weibull, Log-Pearson III and Pearson III distributions.
17
ARTIGOS TÉCNICOS
r = 100 . [1– (1 – P)N ] (%)
2.2. Maximum discharges
In order to estimate the maximum discharges, the type II,
log-Pearson III, Gumbel and Pearson III Log-Normal distributions
were used, by using the Siscah® 1.0 software (GPRH/UFV [7]),
based on the aforementioned methodology, as well as for stationary analysis. Subsequently, for the Gumbel distribution two
scenarios for obtaining the partial series were considered, by developing in a Microsoft® Excell® spreadsheet based on equation
as proposed by Chaudhry [4]:
• Station Code 61250000: Qr = 5.04 m3/s e Qr = 10.06 m3/s; and
• Station Code 58380001: Qr = 348.6 m3/s e Qr = 516.0 m3/s;
2.3. Gumbel distribution
It was developed a Microsoft® Excell® spreadsheet based on
equation as proposed by Chaudhry [4] and according to is given
by Equations (25) to (27)
(25)
(26)
(27)
Ou –ln[–ln(1 – P)] = b
Where: X is the annual maximum flooding average and, σ is
the standard deviation of annual maximum flooding.
A comparison between the theoretical line and empirical values of probability versus discharge demonstrates the model adequacy and also the convenience of using the theoretical line for
obtaining the flooding corresponding to the returning time for the
project [4].
(30)
Where: r: risk of occurs in the N-years coming, at least once
Also, in possession of the Gumbel distributions values for scenarios 1 to 4, it was possible to correlate the returning times (TR)
with respective discharges (Qr) for all scenarios, including the
discharges of interest for SHP works, e.g., the risk values in all
scenarios (1 to 4): for 2-, 50-, 100- and 1000- N (Souza et al.
[11]), as follows:
• A
nalysis of the N = 100 and TR = 10000 years, for jumpable
permanent works in CGH;
• Analysis of the N = 50 and TR = 500 years, for jumpable
works as permanent concrete dam, and
• Analysis of the N = 50 and TR = 1000 years for no permanent
jumpable works as embankment dam.
The series of average maximum discharges, Qmax,m (m3/s),
were obtained from historical data available on the web site of
HIDROWEB® the Brazilian National Water Agency (Agência Nacional de Águas, ANA; in Portuguese).
3. RESULTS AND ANALYSIS
3.1 Distributions and confidence intervals
The values of maximum discharges for distributions studied,
as well as the values of their parameters, standard errors, confidence intervals, variance and asymmetry coefficient are presented in Tables 1 and 2 and Figures 1 and 2. The calculated value
of long-term average flow, QMLT was 3.7246 m³/s and 155.1709
m³/s, respectively for stations 61250000 and 58380001.
2.3. Sensitivity analysis of the partial series
As previously mentioned, a series was obtained from the partial discharges observed by considering only those that exceeded
a certain reference value (Rosbjerg and Rasmussen, 1989 apud
Righetto [7]). Thereby, were fixed two values for discharge, respectively for two studied scenarios (Scenario 1 and Scenario
2) for the flooding reference discharge, Qr, and only the flows
observed with values greater than Qr (Equation 28) were taken.
Reference discharges were taken by observing what was described by Righetto [7], wherein these must be sufficiently high
so that the flooding events may be considered independent. Such
a comparison has already been proposed in Chaudhry [4].
ξ(s) = max[0;Q(s)-Qr)], if [0,t] FIG. 1: Maximum Flow Estimative for 10-years Returning Time for Station
Code 61250000
(28)
• Station Code 61250000: Q
r = 5.04 m3/s (Scenario 1) e
Qr = 10.06 m3/s (Scenario 2); e
• Sation Code 58380001: Q
r = 348.6 m3/s (Scenario 3) e
Qr = 516.0 m3/s (Scenario 4);
2.4. Calculation of discharges with typical returning times
of hydraulic structures (SHP)
Probability (Equation 29) and risk (Equation 30) were calculated, according to the methodology proposed by Souza et al.
[11].
(29)
Where: P: the flooding maximum (or minimum) discharge occurrence probability, at least once, in a future period equal to the
returning time TR passed;
18
FIG. 2: Maximum Flow Estimative for 10-years Returning Time for Station
Code 58380001
TECHNICAL ARTICLES
TABLE 1: Values of interest from the statistical analysis and regarding to maximum discharges for Station Code 61250000
Distribution
Sup.
Inf.
conficonfistanEvent
dence
dence
dard
(m³/s)
intervals
intervals errors
(95%)
(95%)
Alfa
Beta
Gama
Avarege
Variance asymmetry
Number
of events
Standard
deviation
Amplitude
of the
confiError
dence
interval
Gumbel
39.6
34.98
30.4
2.34
0.15
18.47
0.0
22.4
76.99
0.29
8.77
9.15
0.0
Pearson 3
37.2
33.93
30.6
1.68
1.50
34.55
-29.2
22.4
76.99
0.34
8.77
6.57
0.0
-0.19
5.342
4.0
Logpearson 3
44.5
37.15
29.8
3.75
3.02
0.20
0.87
0.44
14.68
0.0
Lognormal 2
38.3
33.87
29.4
2.28
22.4
76.99
0.29
69
8.77
8.94
0.0
Lognormal 3
37.1
33.88
30.7
1.64
22.4
76.99
0.29
8.77
6.42
0.0
TABLE 2: Values of interest from the statistical analysis and regarding to maximum discharges for Station Code 58380001
Distribution
Sup.
Inf.
conficonfistanEvent
dence
dence
dard
(m³/s)
intervals
intervals errors
(95%)
(95%)
Alfa
Beta
Gumbel
1511
1291.0
1071
112
0.0
736.7
Pearson 3
1380
1230.3
1081
76
40.7
47.5
-0.1
7.3
asymmetry
Standard
deviation
Amplitude
of the
confidence
interval
Gama
0.0
863.1
78780.8
0.2
280.7
440.7
0
-1071.7 863.1
78780.8
0.3
280.7
297.8
0
7.7
Variance
Number
of
events
Avarege
Distribution
Logpearson 3
1585
1302.7
1021
144
6.7
0.1
0.7
0.4
564.2
0
Lognormal 2
1424
1232.3
1041
987
863.1
78780.8
0.2
280.7
383.1
0
Lognormal 3
1372
1228.3
1084
73
863.1
78780.8
0.2
280.7
287.8
0
It is observed from Tables 1 and 2 that the values of the
sampling asymmetries were resulted less than 1.5 in all distributions, as recommended in BRAZILIAN CENTRAL ELECTRIC S.A.
[4]. It was also found that the largest values of peak discharges
for flooding-theoretical-probability values, resulted for Gumbel
and LogPearson III distributions. For values of 10-years returning
period, TR, (Station Code 61250000), Gumbel and Log-Pearson
III distributions values were respectively 34.98 m3/s and 33.93
m3/s. Regarding the station code 58380001, the largest values
obtained for peak discharges were also obtained by the Gumbel
and Log-Pearson III distributions, which resulted in respectively
1,291.0 m3/s and 1,302.7m3/s. These discrepancies corroborate
the recommended by Serinald [9] concerning the model uncertainty. Such uncertainties are due to the choice of the statistical
model or physical. It cannot be recovered by the addition of information (such as the sample size), but only by increasing the
knowledge of the process, and the adoption of more complex
models.
3.2. Gumbel distribution and sensitivity analysis of partial
series
The graphs of Figures 3 and 4 show the values obtained from
analysis of the Gumbel distribution for both scenarios as follows:
scenario 1 (Qr = 5 . 4 m3/s) and Scenario 2 (Qr = 10 . 6 m3/s).
They present the estimated events by Gumbel distribution relating to returning time.
Concerning the series of partial discharge, as can be observed
in the graphs of Figures 3 and 4, values of the frequency curve
by the method of Gumbel range especially for larger values of
returning time in these distributions.
Furthermore, as shown in Table 3 for a 10-years TR and the
risk of 19.00%, scenario 1 has presented a value of 33.9m3/s
and scenario 2 has shown a discharge value of 34.0 m3/s. This
difference accounted for 0:55% between these two scenarios (2
33
Error
28481
and 1). For deviation works in SHP, this difference would be not
significant. For a 500-years TR and a risk of 9:53%, scenario 1
has presented a discharge value of 61.0 m3/s and scenario 2
of 58.9 m3/s. This difference has represented -3.42% between
these two scenarios 2 and 1. For jumpable permanent works as
concrete dams, the differences were becoming a slight larger. For
a 1000-years TR and a risk of 4.88%, scenario 1 has presented
a discharge value of 65.7 m3/s and scenario 2 of 63.2m3/s. This
difference has represented -3.78% between these two scenarios (2 and 1). Finally, for 10,000-years TR and a risk of 1.00%,
scenario 1 has presented a discharge value of 81.5 m3/s and
scenario 2 of 77.7m3/s. This difference has represented -4.67%
between these two scenarios (2 and 1).
In addition, according to the presented in Table 4 - also for
10-years TR and the risk of 19.00% -, scenario 3 has showed a
discharge value of 1,229.2 m3/s and scenario 4 of 1,232.2 m3/s.
This difference accounted for 0:24% between 4 and 3 scenarios.
For deviation works in SHP, this difference could be considered as
not significant. For a 500-years TR and risk of 9:53%, scenario
3 has presented a discharge value of 2,096,6 m3/s and scenario
4 of 2,031.7 m3/s. This difference has represented -3.09% between 4 and 3 scenarios. In this case also, for jumpable permanent works as concrete dams, the differences were becoming a
slight larger. For a 1000-years TR and risk of 4.88%, scenario 3
has presented a discharge value of 2,248.4 m3/s and scenario
4 of 2,171.6 m3/s. This difference has represented -3.41% between 4 and 3 scenarios. Finally, for 10,000-years TR and risk of
1.00%, scenario 3 has presented a discharge value of 2,752.4
m3/s and scenario 4 of 2,636.2 m3/s. This difference has represented -4.22% between 4 and 3 scenarios.
These values corroborate what was described by Serinald [9]
regarding to the statistical uncertainty, especially those related to
the estimated parameter being able to be minimized by increasing the sample size.
19
FIG. 3: Gumbel distribution for Scenarios 1 and 2, regarding to the returning time (Station Code 61250000)
FIG. 4: Gumbel distribution for Scenarios 3 and 4, regarding to the returning time (Station Code 58380001)
TABLE 3: Interesting values from Gumbel distribution for Station Code 61250000
T
QT
(Scenario 1)
QT
(Scenario 2) /
years
m3/s
m3/s
QTScen2-Scen1/
QTScen1
Risk
(N=2)
%
Risk
(N=50)
%
Risk
(N=100)
%
Risk
(N=1000)
%
1.01
8.0
10.4
29.39
99.99
100.00
100.00
100.00
1.6
18.6
20.1
7.87
85.94
100.00
100.00
100.00
1.8
19.9
21.3
6.81
80.25
100.00
100.00
100.00
2.0
21.0
22.2
6.03
75.00
100.00
100.00
100.00
2.6
23.4
24.5
4.53
62.13
100.00
100.00
100.00
2.8
24.1
25.1
4.19
58.67
100.00
100.00
100.00
3.0
24.6
25.6
3.89
55.56
100.00
100.00
100.00
4.0
27.0
27.8
2.82
43.75
100.00
100.00
100.00
5.0
28.7
29.3
2.14
36.00
100.00
100.00
100.00
6.0
30.1
30.6
1.66
30.56
99.99
100.00
100.00
7.0
31.3
31.7
1.29
26.53
99.96
100.00
100.00
8.0
32.2
32.6
0.99
23.44
99.87
100.00
100.00
9.0
33.1
33.4
0.75
20.99
99.72
100.00
100.00
10.0
33.9
34.0
0.55
19.00
99.48
100.00
100.00
25
40.4
40.0
-0.89
7.84
87.01
98.31
100.00
50
45.2
44.4
-1.69
3.96
63.58
86.74
100.00
100
49.9
48.8
-2.33
1.99
39.50
63.40
100.00
250
56.2
54.5
-3.01
0.80
18.16
33.02
98.18
500
61.0
58.9
-3.42
0.40
9.53
18.14
86.49
1000
65.7
63.2
-3.78
0.20
4.88
9.52
63.23
10.000
81.5
77.7
-4.67
0.02
0.50
1.00
9.52
Risk
(N=100)
Risk
(N=1000)
T
QT
(Scenario 1)
QT
(Scenario 2) /
years
m3/s
m3/s
1.01
20
402.1
469.7
QTScen2-Scen1/
QTScen1
Risk
(N=2)
%
16.84
99.99
Risk
(N=50)
%
%
100.00
100.00
%
100.00
1.6
741.0
782.2
5.56
85.94
100.00
100.00
100.00
1.8
782.6
820.5
4.85
80.25
100.00
100.00
100.00
2.0
816.9
852.2
4.31
75.00
100.00
100.00
100.00
2.6
894.9
924.0
3.26
62.13
100.00
100.00
100.00
2.8
915.5
943.0
3.01
58.67
100.00
100.00
100.00
3.0
934.3
960.3
2.79
55.56
100.00
100.00
100.00
(Scenario 1)
QT
QT
(Scenario 2) /
m3/s
m3/s
4.0
1,009.4
1,029.6
2.00
43.75
100.00
100.00
100.00
5.0
1,065.0
1,080.8
1.49
36.00
100.00
100.00
100.00
6.0
1,109.2
1,121.6
1.12
30.56
99.99
100.00
100.00
7.0
1,145.9
1,155.4
0.83
26.53
99.96
100.00
100.00
8.0
1,177.4
1,184.4
0.60
23.44
99.87
100.00
100.00
9.0
1,204.8
1,209.7
0.41
20.99
99.72
100.00
100.00
10.0
1,229.2
1,232.2
0.24
19.00
99.48
100.00
100.00
25
1,436.7
1,423.5
-0.92
7.84
87.01
98.31
100.00
50
1,590.7
1,565.4
-1.59
3.96
63.58
86.74
100.00
100
1,743.5
1,706.2
-2.14
1.99
39.50
63.40
100.00
250
1,944.6
1,891.6
-2.73
0.80
18.16
33.02
98.18
500
2,096.6
2,031.7
-3.09
0.40
9.53
18.14
86.49
1000
2,248.4
2,171.6
-3.41
0.20
4.88
9.52
63.23
10000
2,752.4
2,636.2
-4.22
0.02
0.50
1.00
9.52
T
years
QTScen2-Scen1/
QTScen1
Risk
(N=2)
Risk
(N=50)
Risk
(N=100)
Risk
(N=1000)
%
%
%
%
TABLE 5: Parameters of the different statistical distributions for q7,10 at the Station Code 61250000
Distribution
Sup.
Inf.
conficonfistanEvent
dence
dence
dard
(m³/s)
intervals
intervals errors
(95%)
(95%)
Alfa
Beta
Gama
Avarege
Variance
asymmetry
Weibull
1.017
0.868
0.718
0.076
5.258
1.595
-0.494
1.430
0.177
-0.296
Pearson 3
1.032
0.874
0.715
0.081
0.072
34.726
-1.068
1.430
0.180
-0.339
Number
of events
Standard
deviation
Amplitude
of the
confiError
dence
interval
8.77
9.15
0.0
0.424
0.32
0
69
Logpearson 3
1.006
0.933
0.859
0.037
Lognormal 2
1.050
0.945
0.840
0.054
-0.231
2.319
0.839
0.304
0.124
1.313
0.352
0.15
0
1.430
0.180
-0.289
0.424
0.21
0
Note: With respect to the Lognormal III distribution, the asymmetry coefficient resulted in less than zero, therefore no possible solution by using the
method of moments.
TABLE 6: Parameters of the different statistical distributions for q7,10 at the Station Code 58380001
Distribution
Sup.
Inf.
conficonfistanEvent
dence
dence
dard
(m³/s)
intervals
intervals errors
(95%)
(95%)
Alfa
Beta
Gama
Avarege
Variance
asymmetry
Weibull
49.8
38.3
26.9
5.8
5.5
76.5
-37.5
67.7
483.9
-0.3
Pearson 3
50.8
38.3
25.8
6.4
4.8
21.3
-35.4
67.7
499.1
-0.4
Number
of events
Standard
deviation
Amplitude
of the
confiError
dence
interval
22.0
22.9
0
22.3
25.0
0
33
Logpearson 3
47.1
41.7
36.3
2.8
Lognormal 2
50.6
42.6
34.7
4.1
-0.3
1.8
4.7
4.1
0.2
1.5
0.4
10.8
0
67.7
499.1
-0.3
22.3
15.9
0
Note: With respect to the Lognormal III distribution, the asymmetry coefficient resulted in less than zero, therefore no possible solution by using the
method of moments.
21
3.3. Minimum discharges
The simulation results by using Siscah® 1.0 (GPRH / UFV)
[6], the minimum discharges and in particular, the q7, 10, to
the stations under study (Code 6125000 and 58380001) consist
of the statistical distribution shown in the graph of Figure 5 and
Table 5, as well as Figure 6 and Table 6.
m3/s. Finally, with respect to the log-normal distribution, this
value resulted in 42.6 m3/s. Long-term average discharge, Qmlt,
resulted in 155.1709 m3/s for this station (Code 58380001).
Therefore, after subtracting these percentages (70% times q7,10
) from Qmlt, would result in 26.842 m3/s, 26.809 m3/s, 29.194
m3/s e 29.836 m3/s values for turbinable discharge, taking into
account respectively the Weibull, LogPearson III, Pearson III e
LogNormal II distributions. These sanitary discharge values deducted from long-term average discharge, by applying Equation
2 and assuming a net head of 10m, would result in Powers values
as 11.349MW, 11.352MW, 11.141MW and 11.084kW.
Consequently, for the case study of lower discharges, it is
possible that such differences may be more significant.
4. FINAL CONSIDERATIONS
Fig. 5: Statistical distributions obtained from simulation of minimum discharges for Station Code 6125000
Fig. 6: Statistical distributions obtained from simulation of minimum discharges for Station Code 58380001
It can be observed from the graphs of Figures 4 and 5, as
well as from Tables 5 and 6 that different statistical distributions
provide different minimum-7-day discharges values from distributions, especially when considering the magnitude of the confidence intervals.
This may be noted in Table 5, regarding to q7,10, for which
the Weibull distribution has resulted in 0.868 m3/s. LogPearson
III and Pearson III distributions resulted respectively 0.874 m3/s
and 0.933 m3/s. Finally, with respect to the Lognormal distribution, this value resulted in 0.945 m3/s. For this station (Code
58380001), which long-term average discharge, Qmlt, has resulted in 3.7246 m3/s. For example, in Minas Gerais, sanitary
discharge is 70% of q7, 10; therefore by subtracting of that percentage of Qmlt, would result values of 3.117 m3/s, 3.113 m3/s,
3.072 m3/s and 3.063 m3/s taking into account respectively the
Weibull, LogPearson III, Pearson III and LogNormal II distributions. Such sanitary discharge values deducted from long-term
average discharge, by applying Equation 2 and assuming 10m
as net head, 275.67kW, 275.31kW, 271.64kW and 570.89kW of
Powers values would result.
Furthermore, in Table 6, the q7,10 value for Weibull distribution has resulted in 38.3m3/s. LogPearson III and Pearson III
distributions have resulted respectively in 38.3 m3/s and 41.7
22
In the present study, different statistical distributions of a
data series of extreme discharges for two stations case studies (Code 61250000 and Code 58380001) in Brazil were evaluated, taking into account average discharges whether of lower
and higher values: Qmlt of 3.7246 m³/s and 155.1709 m³/s, respectively for stations code 61250000 and 58380001. There were
some distinctions about the values of minimum discharges, which
could in this case study, change an SHP Installed Capacity and
especially considering a μCH. However, it should be considered
that uncertainties are implicit in statistical analyzes and such uncertainties in a confidence interval of 95% (upper and lower) may
guide to a greater range of simulated event, which could affect
the calculation of the power in a plant design in some particular
cases, especially for those with small discharge, sometimes making it financial unfeasible, or large, with considerable losses in a
discharge of design.
Some distinctions were observed concerning the accurate
values of minimum discharges, which could in this case study,
change an Installed Capacity for SHP and especially for a μCH
project. However, it should be considered that uncertainties are
comprehended in statistical analyzes as well as such uncertainties in a confidence interval of 95% (upper and lower) can lead
to a greater range of simulated event, which could affect the
calculation of the power plant in some particular cases, especially
for those with small discharge, even making it financial unfeasible, or in case of large discharges, with considerable losses of
instalated power.
The present study also aimed to evaluate, by using the statistical analysis of these two aforementioned fluviometric stations, as well as the inherent uncertainties in forecasting flooding discharges and theoretical probability distributions which are
commonly used. With the results obtained, it was concluded and
corroborated what was recommended by Serinald (2009) concerning the hydrological uncertainties.
The first uncertainty, based on the analysis of the natural factors of river basin, i.e., local climatic conditions and the regional
characteristics, and which have been aimed to be stochastically
represented, consists on inherent or natural uncertainty, which
represents the randomness and complexity of the natural process and cannot be reduced in any way. The uncertainties of the
model, i.e., the discrepancies obtained from the results analysis
of distributions corroborated what was recommended by Serinald
[9] regarding the model uncertainty. Such uncertainty resides in
the statistical model choice, which is not capable of be reduced
by the information expansion, but only by increasing the knowledge of the process, as well as the adoption of more complex
models. Concerning the statistical uncertainty which is relating
the parameter estimates - despite being subject to reduction by
adding the sample size -, it was confirmed what was recommended by the Serinald (2009), since it was observed that higher values for the returning time were related larger differences for the
two considered scenarios, namely: Scenario 1 (Qr=5:04 m3/s)
and Scenario 2 (Qr=10:06 m3/s) regarding to the fluviometric
station code 61250000. The same behavior was observed for the
other two considered scenarios, namely: Scenario 3 (Qr=348.6
m3/s) and Scenario 4 (Qr=516.0 m3/s) regarding to the fluviometric station code 58380001. There was even a reversal, from
the 10-years returning time for which the values from the Gumbel distribution for the partial series (Scenario 2) began to be
less than those observed values from the total series (Scenario
1), as well as could be observed when was considered Scenario
4 related to Scenario 3.
•
•
•
•
ACKNOWLEDGEMENTS
We are thankful to the Coordination of Improvement of Higher
Education Personnel (Coordenação de Aperfeiçoamento de Pessoal de Nível Superior, CAPES; in Portuguese) for their support.
The Research Support Foundation of Minas Gerais (Fundação de
Amparo a Pesquisa de Minas Gerais, FAPEMIG; in Portuguese),
for the support given by individual participation in event into the
country. We are thankful to the National Reference Centre for
Small Hydropower (Centro Nacional de Referência em Pequenas
Centrais Hidrelétricas, CERPCH; in Portuguese) of Federal University of Itajubá (UNIFEI). We are enormously thankful to Professor
Fazal Hussain Chaudhry from School of Engineering of São Carlos, University of São Paulo (Escola de Engenharia de São Carlos
da Universidade de São Paulo, EESC/USP; in Portuguese) for his
valuable teachings in Hydrology.
5. REFERENCES
•
•
•
•
•
• [1] ABBASI, T.; ABBASI, S.A. (2011). Small hydro and the environmental implications of its extensive utilization. Renewable and Sustainable Energy Reviews, Vol. 15, p. 2134-2143.
• [2] ANAGNOSTOPOULOS, J.S.; PAPANTONIS, D.E. (2007).
Optimal sizing of a run-of-river small hydropower plant. Energy Conversion and Management, vol. 48, p. 2663–2670
• [3] BRAZILIAN CENTRAL ELECTRIC S.A. - ELETROBRÁS. MINISTRY OF MINES AND ENERGY – MME (CENTRAIS ELÉTRICAS
BRASILEIRAS S.A. – ELETROBRÁS. MINISTÉRIO DE MINAS E
ENERGIA – MME). (2000). Guidelines for Small Hydroelectric
•
•
Powerplants Studies and Projects (Diretrizes para Estudos e
Projetos de Pequenas Centrais Hidrelétricas). Eletrobrás.
[4] CHAUDHRY, F. (2001). Hydrology: quantitative aspects
(Hidrologia: aspectos quantitativos). Lecture notes. São Carlos-SP, Brazil.
[5] Di BALDASSARRE, G., LAIO, F.; MONTANARI, A. (2009).
Design flood estimation using model selection criteria. Physics and Chemistry of the Earth, Parts A/B/C, v. 34, n. 10-12,
p. 606-611
[6] RESEARCH GROUP IN WATER RESOURCES OF THE FEDERAL UNIVERSITY OF VIÇOSA - GPRH / UFV (GRUPO DE
PESQUISAS EM RECURSOS HÍDRICOS DA UNIVERSIDADE
FEDERAL DE VIÇOSA – GPRH/UFV). (2009). Computational
System for Hydrologic Analysis (Sistema Computacional para
Análises Hidrológicas). Version 1.0.
[7] RIGHETTO, A. M. (1998). Hydrology and Water Resources
(Hidrologia e Recursos Hídricos). São Carlos-SP, Brazil: EESC/
USP. 840 p.
[8] SANTOLIN, A.; CAVAZZINI, G., PAVESI , G.; ARDIZZON,
ROSSETTI, G.A. (2011). Techno-economical method for the
capacity sizing of a small hydropower plant. Energy Conversion and Management, Vol. 52, n. 7, p. 2533–2541.
[9] SERINALD, F. (2009). Assessing the applicability of fractional order statistics for computing confidence intervals for
extreme quantiles. Journal of Hydrology, v. 376, n. 3-4, p.
528-541
[10] SMAKHTIN V. U. (2001). Low flow hydrology: a review.
Journal of Hydrology, Vol. 240, n. 3–4, p. 147–186
[11] SOUZA, Z.; SANTOS, A. H. M.; BORTONI, E. (2009).
Hydroelectric Power Plants: implementation and commissioning (Centrais Hidrelétricas: implantação e comissionamento).
2 ed. Rio de Janeiro-RJ, Brazil: Ed Interciência. 520 p.
[12] TIAGO FILHO, G.L.; STANO JÚNIOR; BRASIL JÚNIOR.,
A.; LEMOS, H.; FERRARI, J.T; LEMOS, H.; NUNES, C.F. et al.
(2008). Small hydroelectric projects (Pequenos aproveitamentos hidrelétricos), Ministry of Mines and Energy (Ministério de Minas e Energia): Brasília-DF, Brazil. 216 p.
[13] TODOROVIC, P. (1978). Stochastic models of floods. Water Resources Research, v. 14, n. 2, p. 345-356
[14] YUE, S.; OUARDA, T. B. M. J.; B. BOBÉE, B.; P. LEGENDRE, P.; BRUNEAU, P. (1999). The Gumbel mixed model for
flood frequency analysis. Journal of Hydrology, v. 226, n. 1-2,
p. 88-100
ANOTAÇÕES
23
IMPORTANCE OF DRAFT TUBE MODELING IN
NUMERICAL SIMULATIONS OF HYDRAULIC TURBINES
Mauricio Formaggio
2
Thi C. Vu
3
Christophe Devals
3
Ying Zhang
2
Bernd Nennemann
3
François Guibault
1
ABSTRACT
Low head power plant projects requires the installation of high specific speed turbines where the impact of the draft tube on the overall
machine behavior is extremely important. The concept behind hydraulic development of draft tube is based on energy recovery between the inlet and outlet section, done by converting kinetic energy that leaves the runner in potential energy at the end of the draft
tube diffuser. The energy of the flow which leaves the runner has to feed properly the inlet section of the draft tube cone in order to
achieve excellent levels of performance. On the other hand, a poor draft tube design can be responsible for considerable problems of
sudden efficiency loss and instability at full load. In cases of refurbishment projects, the technical risk associated is highly dependent
on the quality of the existing components, especially those who will not be modified, as, in most cases, the draft tube.
KEYWORDS: numerical simulations, hydraulic turbines, flow, efficiency, energy recovery.
IMPORTÂNCIA DA MODELAGEM DO TUBO DE SUCÇÃO
EM SIMULAÇÕES NUMÉRICAS DE TURBINASHIDRÁULICAS
RESUMO
Centrais hidrelétricas de baixa queda exigem a instalação de turbinas com alta rotação especifica onde o impacto do tubo de sucção
no comportamento geral da turbina é e extremamente importante. O conceito atrás desenvolvimento hidráulico do tubo de sucção
está baseado na recuperação de energia do escoamento entre as seções de entrada e saída do mesmo, feita através da conversão da
energia cinética que deixa o rotor em energia potencial no final do difusor do tubo de sucção. A energia presente no escoamento após
o rotor é utilizada para alimentar a região de entrada do cone da sucção a fim do tubo alcançar um excelente nível de desempenho.
Por outro lado, um tubo de sucção deficiente pode ser responsável por problemas consideráveis de súbita perda de rendimento e instabilidade em altas cargas. Em casos de reformas, o risco técnico é altamente dependente da qualidade dos componentes existentes
da turbina, especialmente aqueles que não serão modificados, como, na maioria das vezes, o tubo de sucção.
PALAVRAS-CHAVE: simulações numéricas, turbinas hidráulicas, escoamento, eficiência, recuperação de energia.
1. INTRODUCTION
Steady state computations are routinely used by design engineers to evaluate and compare losses in hydraulic components.
In the case of the draft tube diffuser, however, experiments have
shown that while a significant number of operating conditions
can adequately be evaluated using steady state computations, a
few operating conditions require unsteady simulations to accurately evaluate losses. This paper presents a study that assesses
the predictive capacity of a combination of steady and unsteady
RANS numerical computations to predict draft tube losses over
the complete range of operation of a Francis turbine. For the prediction of the draft tube performance using k-ε turbulence model,
a methodology has been proposed to average global performance
indicators of steady flow computations such as the pressure recovery factor over an adequate number of periods to obtain correct results. The methodology will be validated using a systematic comparison with experimental results.
In an industrial setting, accurate numerical prediction of the
performance of a draft tube over its complete range of operation must be performed as efficiently as possible. The choice of
mesh density, turbulence model and time accuracy must therefore be balanced carefully against the need of obtaining results in
a timely fashion. This paper aims to validate standardized RANS
CFD simulations for the global performance prediction of a draft
tube through comparison with experiments carried out at the
LMH laboratory in Switzerland [1].
Recently, several studies have been devoted to the accurate
prediction of unsteady pressure fluctuations inside the draft tube,
associated with the precession of the vortex rope. To reach an
adequate level of accuracy, these calculations require advanced
turbulence modelling andunsteady flow simulations on meshes
comprising several million nodes [2, 3, 4]. In design mode however, global performance characteristics must be obtained rapidly and these types of detailed simulations cannot be performed
routinely. Efforts must therefore be devoted to the validation of
computational schemes and simplifying approaches that allow
reaching adequate levels of precision in a reasonable time and
using relatively modest computational resources. The present
study is thus aimed at providing design engineers with faster,
more reliable analysis tools that can be readily integrated into
their design process.
2. MATERIAL AND METHODS
Steady state flow simulations are often totally adequate to
predict global performance of well-behaved draft tubes operated
at or close to the best efficiency point of the turbine.
Andritz Hydro Inepar do Brasil SA, Rod. Manoel de Abreu km 4.5, Araraquara, Brazil
Andritz-Hydro Ltd., 6100 TransCanada highway, Pointe Claire, QC, H9R 1B9, Canada
3
Dept of Computer and Software Engineering, École Polytechnique de Montréal, Montréal, QC, H3C 3A7, Canada
1
2
24
However, when a badly designed draft tube is present, or
when operating conditions become less favorable, convergence
problems are observed, which often translate into an oscillatory
behavior of the individual residual history of the main variables of
the governing equations.
These high residuals prevent the solution process from reaching global convergence, and global quantities, such as the pressure recovery factor, from reaching a stable value. Typical convergence histories for a stable and a periodic convergence are
illustrated in Fig. 1.
Acuumulated Time Step
Accumulated Time Step
Wall static pressure measurements near the inlet section and at
the draft tube outlet allow us to calculate the draft tube pressure
recovery factor which is defined as the static pressure difference
between inlet and outlet divided by the kinetic energy at the inlet. Six operating points having the same speed coefficient ψ and
various flow coefficients f, covering full load to part load conditions, were selected for the study. LDV measurements of velocity
profile and turbulent kinetic energy profile at the draft tube inlet
are used as inlet boundary conditions for the CFD simulation.
The geometry of the draft tube was imported into XMD, an
Andritz design tool for draft tube geometry. A single multi-bloc
structured mesh combining an O-type block structure near all
solid walls and a H-type for the inside flow domain was generated
using the in-house automatic draft tube mesh generator DTmesh
[7]. The generated mesh with 675K nodes as shown in Fig. 2 was
exported in the CGNS format [8]. Figure 2 shows also positions
of monitoring points used during the computation, which match
positions of pressure taps at the draft tube inlet (4 points) and
outlet (8 points) during the experimental investigations.
Fig. 1: CFX steady state convergence history for stable (left) and unstable
solution (right)
As can be observed, for some operating conditions, residuals cannot decrease fully to a prescribed convergence criterion.
Analysts generally associate such a convergence behavior to the
probable evidence of unsteady flow phenomena in the solution,
but a question then remains as to whether the steady state solution, when averaged over a number of periods, constitutes an
adequate estimation of the unsteady solution. In order to verify
the hypothesis that steady state computations do in fact reach
the correct limit value, two computational approaches have been
considered to compute the pressure recovery coefficient in cases
where the convergence of the steady state flow simulations were
oscillatory. Results are then compared with experimental results.
These approaches are 1) to average the periodic pressure recovery factor values obtained using a steady state simulation over
a number of periods and 2) to compute the time average of the
fluctuating pressure response of an unsteady flow simulation.
During the process of CFD validation for the draft tube, several parameters have been considered for the study, such as mesh
density [5], turbulence modeling, type of inlet boundary condition, turbulence condition at inlet, etc. The inlet flow condition for
the draft tube can be simply specified as an axi-symmetrical flow
profile or can be obtained directly from the runner by performing stage simulation with coupled runner-draft tube components.
The turbulence conditions such as the turbulence kinetic energy and the turbulence kinetic dissipation must be also specified
as part of the inlet flow conditions for the k-ε turbulence
model. Such information is not well known and the user has
to make a well educated guess for the inlet turbulence flow condition. During the course of the study, we have found that this type
of information could influence greatly the convergence behavior
of the numerical solution.
2.1. Test case
As a test case, we have chosen the draft tube of the FLINDT
project, investigated in a Francis turbine model of high specific
speed, nq = 88. It is a symmetrical elbow draft tube with one pier.
The geometry of the draft tube was carefully selected in order to
obtain the desired efficiency drop toward the full load condition.
Details on the FLINDT project are described in Refs. [1] and [6].
The draft tube flow behavior and pressure recovery factor have
been investigated at several operating conditions of the runner.
Fig. 2: Hexa mesh and monitoring points of draft tube geometry
As mentioned above, the axi-symmetrical velocity profiles
and the turbulence kinetic energy profile are obtained from LDV
measurements. Since no measurements were available to specify the value of the eddy dissipation, the inlet turbulence eddy
dissipation profile was calculated using a mixing length scale LT
model. According to CFX documentation [9], the eddy dissipation
rate is related to the turbulent kinetic energy through the following relation: e = k3/2⁄LT. The mixing length scale therefore constitutes a parameter that must be correctly calibrated in order for
the simulations to match experiments. Three values have been
used in this study: 1%, 0.5% and 0.25% of the runner throat
diameter (Dth).
3. RESULTS AND DISCUSSION
3.1. Numerical model
The study uses the ANSYS-CFX-12.1 commercial software.
Standard RANS and URANS models using a two equation k-ε turbulence closure model were used to perform all computations.
CFX is a code based on the finite volume method which implements several discretization schemes. All computations were
performed using the high-resolution scheme for the momentum
equations, and the first order upwind scheme for the turbulent
advection equations. A non-slip condition was imposed for all
solid surfaces. At the outlet, an average static pressure over the
whole outlet was specified. Six operating conditions with 3 different inlet turbulence length scales were computed and compared
to the pressure recovery factor which is defined as:
25
ΔPstat is the average pressure difference between inlet and
outlet planes of measurement, ρ the fluid density, and Q the
mass flow rate through reference area Aref.
For standard steady state computations, the convergence criterion is usually set to 1 x 10-5 on the root mean square (RMS)
residuals for all main variables. For monitoring purposes, the
convergence criterion was instead set to a very strict value at 1
x 10-12 RMS and a maximum of 2000 iterations was imposed for
all computations. The CFX time step option for steady state cases
was set to auto timescale and the timescale factor set to 1.0.
Convergence history was also verified by the pressure monitoring
points and this allowed computing the pressure recovery coefficient at each iteration.
3.2. Results with steady flow simulations
Figure 3 illustrates the convergence behavior and comparison
with experiment of the pressure recovery factor value for steady
state computations. The numerical pressure recovery factor is
obtained by averaging the periodic pressure recovery factor values obtained using steady state simulation over a number of periods. The mean value has a high and low bound corresponding to
the monitored fluctuation values illustrated as error bars in Fig. 3.
tuation of the pressure recovery factor for the operating point
f = 0.38 and a moderate fluctuation for f = 0.39. When inlet
turbulent viscosity is increased, it is expected that the amplitude
of fluctuations of the pressure recovery factor be reduced, but
this is not the case as observed with the results obtained for
LT=1%Dth. For this inlet condition, the amplitude of fluctuations
of the pressure recovery factor at f = 0.38 remains the same
and we get another point (f = 0.36) having a non-convergent
periodic behavior. For LT=0.25%Dth, three operating conditions
show a non-convergent periodic behavior: 0.368, 0.38 and 0.39
and the pressure recovery factor of the BEP (f = 0.368) is not
well predicted.
Figures 4 shows the evolution of the turbulence viscosity
in the draft tube cone at f = 0.368 for 3 values of LT = 1.0%
Dth, 0.5% Dth and 0.25% Dth as boundary condition at the draft
tube inlet.
The illustration shows the variation of the relative turbulent
viscosity from 0 to 1000. The relative turbulence viscosity is defined as μt/μ, where μt and μ are the turbulent viscosity and the
dynamic viscosity respectively. The turbulence viscosity in the
draft tube cone is dictated mainly by the prescribed turbulence
viscosity at the draft tube inlet but increases rapidly at the draft
tube cone center and further in the downstream region. The red
areas indicate locations where turbulence viscosity ratio is higher
than 1000. The general pattern remains the same for the 3 turbulent inlet conditions.
Fig. 4: Viscosity ratio at BEP f=0.368 – Lt = 1% Dth, Lt = 0.5% Dth and
Lt = 0.25% Dth
Fig. 3 Pressure recovery factor prediction for three levels of turbulent energy dissipation
As can be observed in Fig. 3, turbulence boundary conditions
at the inlet have a significant influence on the simulation results.
A lower value of the dissipation length scale leads to a smaller
turbulent to molecular viscosity ratio, leading to less energy dissipation and a generally more unsteady behavior for some operating conditions.
In the present case, a value of LT=0.5%Dth appears to strike
an adequate balance between stability and sensitivity to flow
features. For this set of computations, there is a very large fluc-
26
In Figure 5 we can observe the evolution of the turbulence
viscosity at the two extreme operating conditions, f = 0.34 (part
load), f = 0.41 (full load) and also for f = 0.38, for the same
value of LT = 0.5% Dth. For operating condition f = 0.34 at part
load where highly swirling flow is taking place at the draft tube
cone, the turbulent viscosity develops rapidly and the region with
turbulent viscosity ration higher than 1000 occupies a large portion of the draft tube cone and the elbow. This is the reason why
the numerical solution is very stable at this operating condition as
shown in Fig. 3. This high turbulent viscosity region decreases as
the swirl intensity decreases in the draft tube cone as observed
for f = 0.368 (Fig. 5) and f = 0.38.
For the operating condition f = 0.41 at full load, the high
turbulence viscosity is not present in the draft tube cone but increases rapidly in the elbow region where a flow recirculation is
taking place at the draft tube elbow ceiling.
Fig. 5 Viscosity ratio at operating point f = 0.34, f = 0.38 and
f = 0.41 – Lt = 0.5% Dth
Fig. 6 Position of the experimental planes
Figure 6 shows the position of the experimental measurement plans in the FLINDT
draft tube. Section 1.75 stands at the end of the draft tube cone and Section 15.5 is
upstream of the pier nose. Comparison of numerical result with experimental data at
the two measurement plans has been carried out for all operating conditions. As example, comparison plots for pressure, velocity (normal, radial and tangential components) and energy E are shown in Fig. 7 for two operating conditions f = 0.368 (BEP)
and f = 0.38. In general, the numerical results match well with the experimental data
for all cases.
Fig. 7: Comparison between
numerical and experimental
data
27
3.3. Results with unsteady flow simulations
A second set of computations was performed for a subset
of the operating conditions, LT = 0.5% Dth, this time using an
unsteady problem formulation. Each unsteady computation was
initialized using the previously computed steady solution for the
same operating point.
We have chosen the operating point f = 0.38, which exhibited
the largest fluctuation amplitude for the steady flow computation,
as an example. Also this is the operating point where the turbine
efficiency has a sudden dramatic drop. The time steps were set
to correspond respectively to increments of 6º, 1º and 0.5º of the
runner rotation. Figure 8a) illustrates the behavior of the pressure recovery factor as a function of iteration number, and the
transition between steady and unsteady computations that occurred at iteration 2000 (time 117.2s), 5255 (time 1021.4s) and
6300 (time 1069.8s). As can be observed, the amplitude of the
recovery factor fluctuation is affected by the change in time step.
However, the average value of the recovery factor computed over
several periods of the steady computations is extremely close to
the temporal average computed in unsteady mode. This observation is confirmed numerically in Table 1 for which both averages
correspond quite precisely with the experimental value.
TABLE 1: Comparison of steady and unsteady averages of the
recovery factor for f = 0.38 and Lt = 0.5% Dth
Steady
state
Transient
state
Dt = 0.27s
(6º)
Transient
state
Dt = 0.046s
(1º)
Transient
state
Dt = 0.023s
(5º)
Min
0.5780
0.5739
0.5389
0.5340
Average
0.6557
0.6547
0.6803
0.6634
Max
0.7313
0.7382
0.7556
0.7309
Experiment
0.667
Figures 8b), 8c) and 8d) show the recovery factor as a function of time at the different transition from steady state to 6°
time step, from 6° time step to 1° time step, and from 1° time
step to 0.5° time step respectively. As observed in Fig 8a), the
amplitude of the recovery factor changes but the period remains
constant.
explains a large drop on the turbine efficiency at this operating
point.
4. CONCLUSIONS
This paper has presented a validated numerical simulation approach to evaluate global draft tube performance. This approach,
based on steady-state flow simulations using the k-ε turbulence
model and a moderately refined mesh, offers a highly effective
methodology that can reliably be used by designers to compare
relative global draft tube performance of nearby design operating
points. This study demonstrates the importance of the choice of
turbulent inlet boundary conditions even close to the best efficiency operating condition. The influence of these fluctuations,
which are not related to the presence of a large vortex rope at
the draft tube inlet, can correctly be averaged by steady-state
simulations.
The ANSYS-CFX flow solver using the high-resolution scheme
is very sensitive to inlet turbulence profile effects. Indeed, since
the numerical scheme shows little diffusion, unsteady fluctuations are well detected. It is therefore very important to average global performance indicators such as the pressure recovery
factor over an adequate number of periods to obtain correct results. In particular, for this test case, unsteady phenomena were
observed for the flow condition f = 0.38, for all eddy dissipation
length scales considered. Detailed unsteady flow analyses have
shown however that the averaged steady flow result computed
using LT = 0.5% Dth was very close to time averages of unsteady
simulations.
Another observation that can be made following the present
validation case study is the fact that as the operating conditions
move away from the best efficiency point, the predictive performance of the k-ε turbulence model tends to deteriorate. This
is particularly the case for part load operating conditions (f =
0.34), for which estimated performance are systematically overestimated. This can probably be attributed to unsteady phenomena inside the cone section of the draft tube, that are damped by
the turbulence model used, even in unsteady simulation mode.
Further investigation with more advanced turbulence models is
required.
ACKNOWLEDGMENTS
The authors would like to acknowledge the National Science
and Engineering Council of Canada (NSERC) for its support to the
project CRD#386829-09, in partnership with Andritz Hydro Ltd.
The Flindt project was supported by Swiss Federal Institute of
Technology, Électricité de France , Alstom, Andritz Hydro (former
General Electric Canada, Sulzer Hydro and VaTech Voest Alpine
MCE), Voith Hydro, Swiss Federal Commission for Technology and
Innovation (PSEL) and the German Ministry of Science and Technology (BMBF).
Fig. 8 Fluctuation of the pressure recovery factor versus (a) iteration number and versus (b-c-d) time – f=0.38 - Lt = 0.5% Dth
For this particular point of operation, the unsteadiness of the
numerical solution is due to an intermittent flow detachment
from the pier wall in the left draft tube channel, as shown in
Fig. 9, which represents snapshots of the evolution of the velocity
at different time steps during one period. The figures on the left
hand side show the velocity contours at different vertical plans
in the draft tube, while the figures on the right show the velocity
contours at a horizontal plan near the draft tube ceiling. Due to
the intermittent flow detachment behavior, the flow rate distribution between the two draft tube channels keeps changing and this
28
NOMENCLATURE
Aref
Ref. section area [0.17538 m2]
Cμ
k-ε turbulence model cte
[0.09 ]
RANS
Reynolds Average NavierStokes
Dth
Runner throat diameter
[0.4 m]
RMS
Root Mean Square
k
Turbulent kinetic energy
[m2s-2]
URANS
LT
Turbulent length scale [m]
H
Net head (m)
BEP
χ
Best Efficiency Point
Unsteady Reynolds Average
Navier-Stokes
Pressure recovery factor
Fig. 9: Unsteady flow behavior at operating point f = 0.38 - Lt = 0.5% Dth
On the contrary, Fig. 10 represents a stable flow behavior at the BEP f=0.368.
Fig. 10: Flow behavior at operating point f = 0.368 (BEP) - Lt = 0.5% Dth
29
N
Rotational speed [1000
rpm]
ε
Eddy dissipation [m2s-3]
nq
Unit specific speed N*Q0.5/
H0.75
φ
Flow coefficient = Q/(pwRth3)
Q
Flow rate [m3s-1]
μ
Dynamic viscosity [8.899.
10-4Pa.s]
Pstat
Static pressure [Pa]
ρ
Fluid Density [997 kg.m-3]
Rin
Radius of the DT
[0.212806m]
ω
Angular velocity [104.66 Hz]
Rth
Runner throat radius [0.2 m]
ψ
Speed coefficient = 2gH/
(w2Rth2)
Δt
Time step [s]
5. REFERENCES
• [1] Avellan, F., “Flow Investigation in a Francis Draft Tube:
the Flindt Project,” Proceedings of the 20th IAHR Symposium
on Hydraulic Machinery and Systems, 2000, Charlotte, North
Carolina, USA.
• [2] Nilsson, H., “Evaluation of OpenFOAM for CFD of Turbulent Flow in Water Turbines,” Proceedings of the 23rd IAHR
Symposium on Hydraulic Machinery and Systems, 2006, Yokohama, Japan.
• [3] Ruprecht, A., T. Helmrich, T. Aschenbrenner, and T. Scherer., “Simulation of vortexrope in a turbine draft tube,” Proceeding of the 21st IAHR Symposium on Hydraulic Machinery
and Systems, 2002, Lausanne, Switzerland.
• [4] Stein, P. et al, “Numerical simulation of the cavitation
draft tube vortex in a Francis turbine,” Proceeding of the 23rd
IAHR Symposium on Hydraulic Machinery and Systems, 2006,
Yokohama, Japan
• [5] Vu, T.C., F. Guibault, J. Dompierre, P. Labbé, and R. Camarero., “Computation of Fluid Flow in a Model Draft Tube Using Mesh
Adaptive Techniques,” Proceedings ofthe 20th Hydraulic Machinery and Systems,2000, Charlotte, North Carolina, USA.
• [6] Mauri, S., J.L. Kueny, and F. Avellan., “Numerical prediction of the flow in a turbine draft tube,” Influence of the
boundary conditions, FEDSM00, ASME Fluids Engineering, DivisionSummer meeting, 2000, Boston, Massachusetts, USA.
• [7] Guibault, F., Y. Zhang, J. Dompierre, and T.C. Vu., “Robust
and Automatic CAD-based Structured Mesh Generation for
Hydraulic Turbine Component Optimization,” Proceedings of
the 23rd IAHR Symposium. 2006, Yokohama, Japan.
• [8] Allmaras, S. and D. McCarthy., CGNS CFD Standard Interface Data Structures - Version 2.3.8. Cgns Project Group 2004
Available at: http://www.grc.nasa.gov/WWW/cgns/sids.
• [9] ANSYS CFX - User Manuel, ANSYS CFX Solver, Release
12.1: Theory.
ANOTAÇÕES
30
ACTIONS AND INNOVATIONS IN DESIGN OF HPP RETIRO BAIXO
Thiago Villela Torquato
Gabriel Villela Torquato
2
Deborah Montenegro C.F. Albuquerque
2
Ana Alice Cesario
1
1
ABSTRACT
This article aims to present innovations in the deployment project HPP Retiro Baixo. Due to proximity to large urban centers, the importance of Paraopeba River and environmental requirements, it was decided to develop a set of actions in order to minimize the impact of
its implementation. Among these actions highlight those relating to constructive arrangements to improve access to rescue fish within
the facility that are presented below:
a) Installation of suction access ports with dimensions of 1200 x 800 mm to facilitate removal of fish;
b) Replacement of screws hatch access suction stainless steel screws for easy opening work;
c) Installation of openings dimensions 2000 x 2000 mm in slabs;
d) Installation of bars downstream of gate to prevent the entry of pods within the draft tube;
e) Installation of air injection points located in the gallery access for suction, with the goal of improving water quality during machine
stoppages;
f) Installation of points for injection of water to maintain / renew the water level and the air in the draft tube;
g) Diversion of water cooling system for Fish Transfer Mechanism;
h) Study of leasing a Fish Transfer System with the aid of physical modeling [1].
These innovations make HPP Retiro Baixo a place of experimentation where it is possible to verify if the changes suggested and
implemented were effective and allowed improvement of work to minimize the environmental impacts caused by the construction and
operation of a hydroelectric power plant.
KEYWORDS: power gereration, fish transfer system, hydro power plant
1. INTRODUCTION
It is known that the implementation of a hydroelectric project
causes environmental and social impacts. The impact occurs on
the flora, fauna, the local population and especially on the river
and fish fauna. Fishes are affected by the damming of a river,
which breaks the migration route, decreases feeding territory
and prevents access to the reproduction sites, nurseries, usually
in smaller tributaries. It is also common to concentration shoals
downstream of the plant, which during operative maneuvers can
be impacted with pressure differences, variation in volume and
water velocity.
2. MATERIALS AND METHODS
The first inventory to use the Paraopeba River in power generation dates from 1966, when the Central Electric of Minas
Gerais - Cemig commissioned a consortium of Brazilian, American and Canadian companies (Canambra), a study of the San
Francisco River basin and several other basins in the state for
use of hydropower. In 1985, Cemig updated these studies in light
of new technical and environmental conditions. In 2001, a new
optimization study concluded that the change in the construction
of the dam from Km 58 to Km 62, from the original encounter of
Paraopeba River with the São Francisco River, would reduce the
flooded area of 52 km² to 22,58 km², decreasing construction
costs and making the project environmentally viable. That was
the beginning of HPP Retiro Baixo.
In the mean while the environmental team identified some
points to be changed in the plant design that would reduce the
impact of this project on the Paraopeba River. This Power Plant is
located in São Francisco River Basin, down the stretch Paraopeba
River near the meeting of the river with the HPP Três Marias Reservoir at coordinates 18°52'47,62"S , 44°46'34.08"W. Between
the cities of Pompeu and Curvelo in Minas Gerais State. It has a
maximum installed generation capacity of 82 Mw and 38.5 MW of
firm energy generation capacity. The generation is done through
two Kaplan turbines vertical axis with unit capacity of 41 MW, the
maximum water discharge is 247.57 m³/s. The spillway capacity
of this plant has to vent 3,945 m³/s.
3. DEVELOPMENT
During maneuvers in operating hydroelectric, normally the
water volume that pass through the turbine decreases, with this
the fish that accumulate in the tailrace unable to enter the turbine are attracted by the low flow to the draft tube. If the turbine starts operating again in a short time the water flow expels
the fish away, but if there is a need to close the draft tube and
drain the unit, can occur fish death, to avoid problems like this
was proposed some changes in the design phase of the plant.
The study demonstrated the abundance of fish in the region
and the risk of accidents during construction, commissioning
and operation of HPP Retiro Baixo. Thus, most of the implemented actions aimed to mitigate the impacts on fish.
Since the early studies, all employees and service providers
of environmental and social areas, to be hired, received training
on the operation of a hydroelectric plant, facilitating dialogue
between the engineering and environmental area. With that
there was an integration of environmental teams, engineering,
design, entrepreneurs and operation, facilitating the development of an integrated work between engineering and environment.
3.1 T
he first change was the general arrangement of the
plant, the first proposed design showed the power house on the left bank of the river and the spillway on the right bank. Between the two structures
is form a stretch of river about 400 meters long, this
RUMO AMBIENTAL
RETIRO BAIXO ENERGÉTICA SA
1
2
31
would only receive water during floods when the spillway is opened, after closing the floodgates large puddles would be formed with risk of keeping fish trapped.
To avoid this problem, all the hydraulic circuit was transferred to the river’s left bank avoiding the formation of
these puddles, and also concentrating the work in one
place. For the stretch between the hydraulic circuit and
earth dam has been proposed a landfill to avoid that this
area gets flooded by the waters downstream during major
floods.
Fig. 3: Project the grids anti-shoal (monitoring fish set the mesh size of
the grid anti-shoal)
Fig. 1: General arrangement of the plant - Original Basic Design
3.3 A
common problem in hydropower plants is, trapping fish
inside the draft tube, even with the installation of grid
anti-shoal it’s expected that some fish are able to enter before the full closure of the draft tube. To facilitate
the rescue of the fish it was proposed an exchange in
the hatch door’s dimensions for 1200 x 800 mm, the access to the draft tube is through gallery located on the
first floor of the power house. The fish are caught in the
draft tube with nets and placed in buckets of 60 liters to
be lifted to the gallery above. In this gallery the fish are
transferred to a larger bucket with a capacity of 1500
liters. From this gallery the bucket with fish is lifted by
a crane through a shaft of dimensions 2000 x 2000 mm
thought for this purpose. These openings in slabs, accelerate the rescue process, both in the draft tube, as in
the penstock. Preventing carrying the buckets on stairs
or long galleries, thus minimizing the risk of death of the
fish. After transported by crane, the bucket containing the
fish is placed in a truck that transports it to upstream fish
release site.
Fig. 2 General arrangement of the plant - Basic design after insertion Environmental
3.2 T
o prevent the entry of fish shoals inside the draft tube
during stops units, grids were installed downstream of the
gates. During the projec was left a groove in the concrete
where the guides for the grids were installed, separated
from the guides of the gates, forming independent systems. Without using the crane and with automatic activation of the grids, firing a system of electric winches
that descends in three minutes, when the unit stops.
the grids are positioned above the exit of the draft tube.
The grid was designed using the information from the
monitoring of fish populations, which showed that in this
stretch of the Paraopeba River most fish were juveniles
starting the migration, so the grid mesh was set to 2 cm
wide.
32
Fig. 4: General arrangement of the plant
3.4 D
uring a fish rescue, agility is necessary and one of the
points of greatest obstacle is the opening of the hatches, screws are usually made of iron and with time and
moisture they end up rusting, making it very difficult and
time consuming to open access to the draft tube, the local to rescue fish. In this power plant proposed a change
to expedite the opening and the option was to use screws
made of stainless steel, because this material does not
rust. It was also proposed to use a pneumatic machine to
remove the screws.
3.5 After closing the draft tube with the gates, the water exchange between the enclosure and the river stops happening and depending on the amount of fish the water
quality drops quickly. In order to maintain good water
quality for a longer period has been proposed to install
points of oxygenation in the bottom of draft tube. To this
was installed a battery of injectors along the draft tube,
some injecting air, and some others with water. The proposal is that, during drainage and the hatch opening, air
is injected to maintaining the oxygen level inside the enclosure, if the rescue takes too long to happen is possible
to exchange the water using the water injection nozzles.
Fig. 6: Construction of the reduced model
Fig. 5: Detail of the distribution of the nozzles in the design and detail of
the injection nozzle
3.6 The turbine cooling water is usually released downstream
of the power house, next to the structure of the power plant, water is usually released in height forming a
waterfall, which attracts fish leaping trying to go up
the river, usually, this waterfall is near the structure,
the fish ends up hitting the concrete and suffering injuries. Another problem is that this water is usually
released with a temperature above the river’s water,
which also attracts fish by the temperature difference.
Some power plants built most recently, have tested
ways to prevent the fish from getting hurt hitting the
concrete. In Retiro Baixo power plant the cooling water
was channeled and released along with the water attraction, of the Fish Transfer System, thus helping in attracting it into the system.
3.7 T
o define the location of Fish Transfer System, was proposed a research in scale model, taking into account the
hydraulic variables, the topography of the region the dynamics of fish, etc. For it, was built in the Center of Hydraulic Research of UFMG, by a multidisciplinary team,
the model in 1:43 scale in Styrofoam and covered with
fiberglass, it was placed specimens of Mandi Amarelo
(Pimelodus maculatus), fish that migrate short distances
and are abundant in the region. These fish were observed
for a long period, to identify the behavior of dislocation
within the system. It was observed that the shoals swam
preferentially by the left bank of the tailrace. So the installation of the Fish Transfer System was made at the
left margin [1].
Fig. 7: Observation of fish behavior in the reduced model
During operation of the HPP Retiro Baixo the technical innovations implemented in the design were tested and the conclusion
was that there was an improvement environmental performance
without prejudice to the operation of the plant. The rescues were
faster and more efficient. There is the possibility of keeping fish
alive into the draft tube for a short time thanks to the injection
of air and the water exchange. The diversion of the cooling water
system for the fish transposition system is working perfectly.
The interest in building this power plant, taking into account
the innovations proposed by the environmental area, showed
that it is possible to make a venture environmentally better,
without prejudice to power generation and implementation costs
negligible, because they occurred in design and were adjusted
throughout the construction. In some cases the environmental
improvements also help the power plant's operation, for example, a fish rescue system efficiently reduces downtime and the
turbine back up and running in less time.
5. REFERENCES
• [1] Martinez, C.B.; Viana E.M.F., Faria M.T.C. (2010) Metodologia para Identificar a Locação de Cardumes de Peixes à jusante
de UHE. The 8th Latin-American Congress on Electricity Generation and Transmission – CLAGTEE, Ubatuba, 2009.
33
• BARBIERI, G.; SANTOS, M.V.; SANTOS, J.M.. Época de reprodução e relação peso/comprimento de duas espécies de
Astyanax (Pisces, Characidae). Pesquisa Agropecuária Brasileira, v.17, n.7, p.1057-1065, 1982.
• BENEDITO-CECILIO, E.; AGOSTINHO, A.A. Estrutura das populações de peixes do reservatório de Segredo, p.113-139. In:
AGOSTINHO, A.A.; GOMES, L.C. (Org.). Reservatório de Segredo: bases ecológicas para o manejo. Maringá: EDUEM, 1997.
387p.
• BRITSKI, H.A.; SATO, Y.; ROSA, A.B.S. Manual de identifi-
cação de peixes da região de Três Marias: com chaves de
identificação para os peixes da bacia do São Francisco. 3a ed.
Brasília: Câmara dos Deputados/Codevasf, 1988. 115p.
• DAJOZ, R. Ecologia geral. Petrópolis: Editora Vozes, 1973.
471p. ECOUTIN, J.M.; ALBARET, J.J.;TRAPE, S. Length-weight
relationships for fish populations of a relatively undisturbed
tropical estuary: The Gambia.
• Fisheries Research, v.72, p.347-351, 2005. HEMMERT, H.
Ecologia. São Paulo: EPU/Springer, Ed. Universidade de São
Paulo, 1982. 335p.
ANOTAÇÕES
34
STUDY OF PROCESS OF AMENDMENTS VULCANIZATION NITRILE
RUBBER FOR SEALING VALVES USED IN BUTTERFLIES CONDUITS SHPS
STUDY OF PROCESS OF AMENDMENTS VULCANIZATION NITRILE RUBBER FOR
SEALING VALVES USED IN BUTTERFLIES CONDUITS SHPS
1
Moisés Toigo
João Henrique Bagetti
2
Sergio Luis Marquezi
1
ABSTRACT
Due to the growth of Brazilian economy in recent years there has been a considerable increase in demanding electricity power. Supplying all this demand, Brazil has been increasing the use of its potential Hydride, with the hydropower building. Therefore, many
companies have emerged and have specialized in building turbines for power plants as well as other components of these small plants
such as butterfly valves, conduits and gates. Such situation proof sets old methods and technologies used by these industries, being
required a constant update in processes and created products with the intention of obtaining safest, efficient and reliable equipment.
The purpose of this work was studying the vulcanization joint process nitrile rubber seals for the use in butterfly valves in small power
plants to discover more efficient and reliable methods for achievements of joints. Based in the regulation ASTM D412-06aε2 were performed trials to verify the endurance of vulcanized joint in a universal trial machine. During the work, it could be observed that the
factors which most affected the vulcanized redressed endurance were the temperature and the time. The results led to the conclusion
that the temperature used was too high, being 60 °C above the average, among other factors that contributed to the joint weakening.
KEYWORDS: Vulcanization. Nitrile rubber. Vulcanized joint. Butterfly Valve.
1. INTRODUCTION
Butterfly Valves are used in large-scale SHP (Small Hydro
Power). They are positioned in the powerhouse, near the turbine
inlet tube to block the way of the penstock for water turbine, in
case it is needed to stop the machine.
For this reason, it is essential that the type of gasket used in
these valves provide excellent sealing, so that there is a stoppage
of the machine (due to a trigger rotor or even scheduled maintenance), as well as to avoid possible flooding in the machine’s house.
For economic reasons, the seals are usually not acquired with
the specific perimeter to each valve, in other words, they are
bought in larger quantities, being helpful the cut in the perimeter desired in the factory. Because of this, it’s necessary that it
has been accomplished the process of vulcanizing in the rubber
seam, which is often made in a traditional manner in manufacturing of valves, which can cause premature leaks.
In partnership with a manufacturer of hydraulic turbines and
the University of the West of Santa Catarina, the authors accomplished a survey formulating solutions and applying new processes in attempt to solve problems related to resistance to rupture
and leakage of these amendments.
2. THEORETICAL
2.1. Nitrile Rubber
As Gomes (2012), nitrile rubber (NBR) provides a good balance between resistance to cold temperature (among -10 °C and
-50 °C), oil, gasoline and solvent, resistance that in Acrylonitrile
content function. These characteristics combined with good resistance to high temperature and abrasion resistance become NBR
recommended for a variety of applications. It also shows good
resistance to dynamic fatigue, low gas permeability and is easily
mixed with polar materials such as PVC.
There is specific types of NBR, which are called degree. Each
degree contains linked to the polymer chain, antioxidants which
become less volatile, so that these NBR’s degrees are less soluble
in gasoline and oil as well as increasing heat resistance. There
are still several other special degrees of NBR to not only be used
as advantage in the processes of transfer by vulcanization or injection, but also to the need for cleaning of the mold is lower,
since it reduces the occurrence of the phenomenon known as
"mold fouling".
The combined with NBR rubber reinforcing fillers, carbon
black or silica, allows to obtain vulcanizates with excellent physical properties. The mechanical properties depend on the temperature of vulcanization.
The resistance to compression deflection depends primarily
on the content of acrylonitrile (ACN) type and NBR used in the
vulcanization system chosen, achieving more excellent values for
this property.
The hardness of the vulcanized NBR with low and medium ACN
content, remains constant over a wide range of temperatures
(70 °C to 130 °C) while the tensile strength decreases significantly with increasing temperature.
2.2. VULCANIZATION
The vulcanization can be described as an exchange of physical properties of rubber of a predominantly plastic state into a
predominantly elastic state by temperature, pressure and time.
The discovery of vulcanization is attributed to Charles Goodyear in the United States, and Thomas Hancock in England.
Both patents developed in 1840. The vulcanization of the rubber caused a pronounced improvement in chemical and physical
properties in relation to the material not vulcanized.
According to Costa et all, (2003), the most important step in
relation to chemical vulcanization occurred with the discovery of
organic accelerators in 1900. In addition to increasing the speed
of vulcanization, these additives have brought many other advantages. The use of accelerators allowed the use of lower temperatures and smaller curing times. Consequently, there was no
longer the need to undergo drastic conditions for the rubber and,
thereby, the possibility of thermal and oxidative degradation was
minimized. Furthermore, the level of sulfur could be reduced,
without harm to the physical properties of the vulcanizated.
In vulcanizing the rubber is heated in the presence of sulfur
and accelerators and activators agents. The vulcanization con-
Universidade do Oeste de Santa Catarina, Campus de Joaçaba; [email protected]
Universidade do Oeste de Santa Catarina, Campus Joaçaba; [email protected]
3
Universidade do Oeste de Santa Catarina, Campus Joaçaba; [email protected]
1
2
35
sists in the crosslinking of the polymer molecules in the individual responsible for the development of a rigid three-dimensional
structure quantity proportional to the resistance of these connections.
The exact determination of the method and conditions of vulcanization (time, temperature and pressure), should be done not
only in view of the composition employed, but also as the dimensions of the artifact to be manufactured and its application. The
state of vulcanization affects various physical properties of the
artifact.
These are the basic parameters to achieve the most different types and qualities of vulcanized therefore be calculated to
achieve optimum vulcanization within its cycle. "Optimum vulcanization" is usually the conditions required to obtain 90% of the
maximum tensile strength (T90), compared to the original force
supported by seamless rubber and vulcanized patches.
As the vulcanized patches, due to the orientation of the molecules of the rubber, it is allowed that some angles set out better
effect at the moment of the connection, giving to the mass a better integrity and strength. According recommendation SAMPLA of
Brazil (2012), hot vulcanized seams must have a length equal to
the width of the rubber to be amended, and amendments in formats such as, "V", "serrated" and "simple to the 65°" have best
effects when used at 65° angle.
Source: Viero M. (2010)-Reestudo do comportamento
do sistema de vedação da válvula borboleta.
Fig. 2: Profile cutting rubber for gluing.
Step 3: Place the gasket on the template with the format of
your profile. Close the template, providing the grip necessary for
fixing joint (Picture 1).
Step 4: Place the template with the joint be vulcanized in the
greenhouse. Performing the heating cycle, increasing the maximum temperature of 180 °C and then cooled in ambient conditions
at a temperature of 80 °C. Repeat the process three times.
3. METHODOLOGY
3.1. Material Researched
The research material was Nitrile Rubber (NBR) with hardness 75/80 Shore A. This rubber was chosen after a job already
done by Viero M. (2010) according to the ASTM D2000 (2010),
for obtaining the best possible resistance to water and warping,
inasmuch as the material is in direct contact with high water
pressure. The format of the rubber should be defined by the engineering company, to enable better sealing and durability under
the conditions of use. Figure 1 shows a profile rubber seal of a
butterfly valve.
Source: Viero M. (2010)-Reestudo do comportamento
do sistema de vedação da válvula borboleta.
Picture 1: Template for vulcanizing rubber and glue.
3.2.2.Processes proposed
Source: M. Viero (2010)-restudy ofsystem behavior sealing butterfly valve.
Fig. 1: Profile rubber sealing butterfly valve.
3.2.1 Current Process
The rubber used in seals are acquired in the form of meter
strings, following the profile shown in Figure 1. Therefore, it is
necessary that the cutting is performed rubber exact size of the
perimeter channel in the rubber plug accommodation. The process for vulcanizing the splice is divided into four stages, namely:
Step 1: Cutting the two ends of the rubber with 45° angle
(Figure 2).
Step 2: Accomplishing the bonding of the two cut edges, applying to a layer of glue for hot vulcanization at both ends, adding
a rubber strip connecting the cold junction to occur as parts of
Figure 2.
Based on research material for vulcanization as well as the
process described in Section 3.2.1, proposed a test procedure to
test vulcanization as described below.
Perform test with two types of amendments. In the first type,
the cutting edges was achieved with the cutting angle 60° (Drawing 3 (A)), so increasing the link connection. In the second kind,
the cut was made in "V" (Figure 3 (b)), increasing the contact
area and pressure. The two types as defined from the theoretical
framework explained in Section 2.2 (VULCANIZATION), which
indicates that the links of the chains of polymers have greatest
effect with certain angles and with a larger contact area.
Fig. 3: Angles of the proposed amendments.
36
The work in question followed the following assumptions:
• The gluing process followed the same procedure described
in item 3.2.1 as well as the template used for pressing and
heating the joint to be vulcanized was the same as described
in the same item.
• The temperature range for the vulcanization of nitrile rubber
is 130 ºC to 150 ºC, because this temperature range links the
rubber would not be broken and therefore does not weaken
it. Therefore, it was proposed to use as the maximum temperature in the thermal cycle 130 °C.
• Use only one heat cycle heating and cooling instead of three,
as shown in step 4 of the item 3.2.1. Looking to thereby prevent damage to the rubber structure of the increased exposure due to its rubber heating temperature limit. This thermal
cycle was carried out for both samples with seam at an angle
of 60°, as for samples with seam "type V".
3.3. Trials
To test the new amendment models proposed, as well as the
current model and unvulcanized rubber itself, we opted for the
traction test, where we could observe the maximum loads of rupture and form of fracture.
The traction tests were performed on a universal traction machine, the EMIC DL30000F model, with test method Tensile Cylinder with cell of 30 tf, which is the laboratory materials Unoesc
- Joaçaba.
The tensile properties depend on both the material and the
test conditions (extension rate, temperature, humidity, sample
geometry, pre-conditioning test), so the materials should be
compared only when tested under the same conditions.
As the objective is to verify the effectiveness of the vulcanization process of rubber used on butterfly valves, it was opted to
carry out tests on the joints vulcanized profile format used in the
valves. The preparation of the samples for the traction test was
as follows:
• According to ASTM D412-06aε2, the rubber should have a reduced section in the center of the sample to allow rupture
at this point and not at the base, as would be expected for
a sample of constant section. But because the samples used
have already a set format, opted to use the rubber with constant section, to approximate the actual conditions which are
subjected in the field. For this purpose, preliminary rupture
tests were effective, with the broken rubber in areas pretty
far from the coupling zone of the machine, thus demonstrating the feasibility of using such samples.
• Also according to the standard ASTM D412-06aε2, traction machine must be programmed for constant spacing of
500 ± 50 mm / min. However, the machine available for testing has a maximum spacing of 100 mm / min.
• For preparation of the samples was taken into account the
vulcanization process and the way to fixing the universal testing machine. Considering these two points, the samples were
initially made with 450 mm for later adapted to 270 mm and
meet the demands of the universal testing machine. Using 75
mm on each side for mounting in the machine, leaving the
center free 120 mm. As illustrated in the Figure 4.
Fig. 4: (a) the length of the sample, (b) section of the sample.
To keep the comparison results, the test parameters were
kept, so all samples were tested under the same conditions of
temperature, humidity, loading rate and offset. All samples were
prepared according to the procedures described. The period of
the tests was the 6:40 p.m. to 7:40 p.m. of the day 10/04/2012.
For easier analysis of the results of samples submitted to vulcanization procedures described in paragraphs 3.2.1 and 3.2.2.,
they were named as described in Table 1.
TABLE 1: Description of the samples.
Rubber Current Process
Rubber splice used to date by the
partner company.
Rubber Process 130 °C 60°
Rubber with proposed amendment
described in section 3.2.2.
Rubber Process V
Rubber splicing also described in
item 3.2.2 of work
Original Rubber
Seamless rubber, used as a
benchmark for others.
The traction tests performed at Unoesc - Joaçaba were evaluated according to their maximum strength and fracture occurred.
4.1. ANALYSIS OF DATA MAXIMUM STRENGTH
To analyze the data obtained in tests of the samples, we used
the values of maximum force achieved before rupture, and from
these it also uses the standard deviation values found among the
five samples of each process. The standard deviation is used to
demonstrate the level of reliability of the process, since it is a
manual process, errors may occur during their execution, as well
as a process of low quality can cause different effects in each
sample. Therefore, when comparing the samples to the current
process, the standard deviation is used to measure the strength
with maximum traction, the quality of the splice.
The maximum strength supported by the sample is exposed
in Figure 1. These values are the averages of maximum forces
encountered for each sample type. In Table 2, the data have been
obtained with their respective standard deviations.
Observing Table 2 , it can be noted that the proposed process
in which the temperature is 130 ° C with seam 60º, reaches
values not much higher (2%) than those obtained in the current
process, but to compensate the deviation standard was lower
(39%), increasing the reliability of the results.
Allied to this, there has to take into account the economic
factor, since it reduced the vulcanization cycle, both at room temperature (28% smaller) and in time (65% less) while maintaining
the values of tensile strength stable when compared to the current process. Therefore, even without increasing expected values
of resistance, which would be ideally 90% of the seamless rubber
as described in item 2.2, it becomes feasible to execute the proposed process, with regard to economic factor.
As regards the case in "V", it got surprisingly tensile strength
values well below the expected 23% less than the current process, and this process was expected the best results. This fact
had not found its explanation in this work, even as it is a scientific
research and own dates and deadlines, but it is emphasized that
the research is still in progress and new results will emerge over
time to explain this situation in specific.
As for the process standard deviation "V", this achieved the
best results of standard deviation among the samples, being
37
54% lower than the current process. As previously explained,
this increases process reliability.
It can also be seen in Figure 5, none of the samples whose
seams are achieved optimal minimum tensile strength compared
to similar samples seamless, reaching a maximum T66, a factor which should be T90, as referenced in Item 2.2. This situation demonstrates that the processes proposed here are far from
"perfect", because there are many factors that can influence your
getting.
It may be noted also that the parameter of T90, which is
rooted in the Ringtread Manual Recovery System, is a parameter
for recovery tires, but features very close to the vulcanization
process of amendments in nitrile rubber. This fact does not make
it completely reliable application of the amendment process, by
having different formats and features as the application, but
rather a viable and reliability equivalent, if not superior.
All data obtained show that there are still unknown aspects
in this process, requiring testing different types of seam, glue,
accelerator, enabling find best results as regards the maximum
force of rupture resistance.
Photo 2: (a) Amendment current process. (b) Detail of seam detached with
waste breakdown, the current process.
Subsequently, the samples Rubber Process 130 °C 60° gave
results very similar to the current process, or detachment of the
joint, but it was observed in most cases a higher detachment
without disruption of adjacent material (Photo 3). This situation
shows that the nitrile adjacent splice suffered no or little change
due to the process temperature. This detachment of standardized
samples directly reflected in the results of standard deviation,
this value decreased considerably (38%).
Photo 3 - Details of the amendment detached of samples Rubber Case
130 ° C 60 °.
Already on the seams of the rubber samples V process, the
ruptures occurred mostly in the material unaffected by the chemical process of splicing (Photo 4) and not in the amendment itself,
not detachment occurring in most cases, such a situation can be
explained by the fact that the amendment has increased contact
area, thereby increasing the vulcanized area.
Fig. 5: Maximum Force average for each type of sample
TABLE 2: Description of the samples.
Maximum tensile strength in N
Type of
Process
S1
S2
S3
S4
S5
Average
Standard
Deviation
Rubber
Current
Process
1423
1702
1165
1314
1453
1423
197,31
Rubber
Process 130
°C 60°
1473
1443
1633
1453
1294
1453
120.40
966
1085
1195
1085
955
1085
90
2601
2290
2797
2742
2539
2593
200
Rubber
Process V
Original
Rubber
4.2 FRACTURES
Complementing the analysis of resistance has made the analysis of the fracture of samples. In samples of the current process,
there has been a detachment in the seam and a small break from
adjacent material (Photo 2), the rupture of adjacent material demonstrates that the high temperatures used in this process, that is,
at a temperature of 180 ºC and 45º, causes a degradation of the
adjacent material. This is due to the fact that the fluxing material is
much less resistant than the rubber itself, assuming that rupture occurs in this area and not in the nitrile rubber. In some cases a small
break in the final moments of separation was relevant in the results
of a standard deviation higher.
38
Photo 4: Amendment V half broken, half detached.
We also observed that the seams where the detachment occurred, it was only partial (Photo 4). In all cases, after reaching
half the seam, they suffer a breakdown of rubber not affected,
and even in those cases, the rubber broke section highest to
lowest.
It is noteworthy that these amendments have the best reliability indices, keeping the standard deviation below 100 N.
Photo 5: Amendment V ruptured.
5. CONCLUSIONS
With this work, could be evaluated the mechanical properties
of nitrile rubber vulcanized seams to seal butterfly valves used in
conduits SHPs, using processes as usual and proposed.
With respect to parameters studied, observing the maximum
force, it follows that the samples with just a process of heating
up to 130 °C show a higher average strength compared to the
samples of the current process (three processes of heating to
180 ° C and amendment to 45 º).
Comparing samples by type of fracture, it is observed that the
samples with seam "V" have a fracture with greater homogeneity,
and a "tear" of rubber in areas not affected by the amendment
(chemically affected), and not a simple detachment as observed in
the other. Also admits that the crack is in the Startup section was
larger and spreads to the smaller section of the rubber. This allows
us to affirm that one has greater security for this process, since in
normal usage situations, the part with the rubber is thinner that is
in contact with the water and subject to premature wear.
Also comparing the standard deviations, amendments vulcanized "V" were more reliable than the others, having standard
deviation value much lower, on the order of 54% lower compared
to the value obtained for the current process.
Regarding the economic factor has been that the proposed processes, have considerable savings in time and machine operator, 65%
lower due to temperature change limit and process heating / cooling.
This work now serves as a reference for future research,
which already began as a continuation of this, seeking to explain
and quantify the other factors in the case and were not explained
here. It is noteworthy that all data and thorough research, are
available to access various Internet platforms.
It is argued that the processes proposed here can be used in
a way equivalent to the current process, or even more so, getting
good results with their proper safety and reliability.
6. REFERENCES
• Gomes, Manuel Morato. Nitrile Rubber (NBR), available at:
• <http://www.rubberpedia.com/borrachas/borracha-nitrilica.
php>. Accessed on 03/08/2012.
• Helson M. da Costa, Leila LY Visconte, Regina Nunes CR. HISTORICAL ASPECTS OF VULCANIZATION. Polymers: Science
and Technology, 13 (2), pp 125-129, 2003.
• Sampla Brazil's Industry and Trade Ltda belts. MANUAL
AMENDMENTS, available at: <http://www.sampla.com.br/
Manualemendas.pdf>. Last accessed on 09/27/2013
• Viero M. Re-study of the system behavior sealing butterfly
valve. Internal publication, Unoesc - Joaçaba, 2010.
• ASTM D412-06 Standard Test Methods for Vulcanized Rubber
and Thermoplastic Elastomers - Tension. 2012. Digital version for Universities, available at: <http://www.astm.org/
Standards/D412.htm>.
• Ringtread System. Recovery Manual Section 8, Vulcanization,
available at: <http://www.steffenpneus.com.br/manualmarangoni/manualmarangoni08.pdf>. Accessed on 03/08/2012.
ANOTAÇÕES
39
ANOTAÇÕES
ANOTAÇÕES
40
HIDRO&HYDRO: PCH NOTÍCIAS & SHP NEWS, 59 (4), OUT,NOV,DEZ/2013, ANOTAÇÕES
ANOTAÇÕES
ANOTAÇÕES
HIDRO&HYDRO: PCH NOTÍCIAS & SHP NEWS, 59 (4), OUT,NOV,DEZ/2013, ANOTAÇÕES
41
ARTIGOS TÉCNICOS
TECHNICAL ARTICLES
INSTRUÇÕES AOS AUTORES
INSTRUCTIONS FOR AUTHORS
Forma e preparação de manuscrito
Form and preparation of manuscripts
Primeira Etapa (exigida para submissão do artigo)
First Step (required for submition)
O texto deverá apresentar as seguintes características: espaçamento 1,5; papel A4 (210 x 297 mm), com margens superior,
inferior, esquerda e direita de 2,5 cm; fonte Times New Roman 12;
e conter no máximo 16 laudas, incluindo quadros e figuras.
Na primeira página deverá conter o título do trabalho, o resumo e as Palavras-chave. Os quadros e as figuras deverão ser
numerados com algarismos arábicos consecutivos, indicados no
texto e anexados no final do artigo. Os títulos das figuras deverão
aparecer na sua parte inferior antecedidos da palavra Figura mais
o seu número de ordem. Os títulos dos quadros deverão aparecer
na parte superior e antecedidos da palavra Quadro seguida do
seu número de ordem. Na figura, a fonte (Fonte:) vem sobre a
legenda, à direta e sem ponto final; no quadro, na parte inferior
e com ponto final.
O artigo em PORTUGUÊS deverá seguir a seguinte sequência: TÍTULO em português, RESUMO (seguido de Palavras-chave), TÍTULO DO ARTIGO em inglês, ABSTRACT (seguido de
keywords); 1. INTRODUÇÃO (incluindo revisão de literatura);
2. MATERIAL E MÉTODOS; 3. RESULTADOS E DISCUSSÃO; 4.
CONCLUSÃO (se a lista de conclusões for relativamente curta, a
ponto de dispensar um capítulo específico, ela poderá finalizar
o capítulo anterior); 5. AGRADECIMENTOS (se for o caso); e 6.
REFERÊNCIAS, alinhadas à esquerda.
O artigo em INGLÊS deverá seguir a seguinte sequência:
TÍTULO em inglês; ABSTRACT (seguido de Keywords); TÍTULO
DO ARTIGO em português; RESUMO (seguido de Palavras-chave); 1. INTRODUCTION (incluindo revisão de literatura); 2.
MATERIALAND METHODS; 3. RESULTS AND DISCUSSION; 4.
CONCLUSIONS (se a lista de conclusões for relativamente curta,
a ponto de dispensar um capítulo específico, ela poderá finalizar
o capítulo anterior); 5. ACKNOWLEDGEMENTS (se for o caso);
e 6. REFERENCES.
O artigo em ESPANHOL deverá seguir a seguinte sequência:
TÍTULO em espanhol; RESUMEN (seguido de Palabra llave), TÍTULO do artigo em português, RESUMO em português (seguido
de palavras-chave); 1. INTRODUCCTIÓN (incluindo revisão de
literatura); 2. MATERIALES Y METODOS; 3. RESULTADOS Y DISCUSIÓNES; 4. CONCLUSIONES (se a lista de conclusões for relativamente curta, a ponto de dispensar um capítulo específico,
ela poderá finalizar o capítulo anterior); 5. RECONOCIMIENTO
(se for o caso); e 6. REFERENCIAS BIBLIOGRÁFICAS.
Os subtítulos, quando se fizerem necessários, serão escritos com letras iniciais maiúsculas, antecedidos de dois números
arábicos colocados em posição de início de parágrafo.
No texto, a citação de referências bibliográficas deverá ser
feita da seguinte forma: colocar o sobrenome do autor citado
com apenas a primeira letra maiúscula, seguido do ano entre
parênteses, quando o autor fizer parte do texto. Quando o autor
não fizer parte do texto, colocar, entre parênteses, o sobrenome,
em maiúsculas, seguido do ano separado por vírgula.
O resumo deverá ser do tipo indicativo, expondo os pontos
relevantes do texto relacionados com os objetivos, a metodologia,
os resultados e as conclusões, devendo ser compostos de uma
sequência corrente de frases e conter, no máximo, 250 palavras.
Para submeter um artigo para a Revista PCH Notícias & SHP
News o(os) autor(es) deverão entrar no site www.cerpch.unifei.
edu.br/submeterartigo.
Serão aceitos artigos em português, inglês e espanhol. No
caso das línguas estrangeiras, será necessária a declaração de
revisão linguística de um especialista.
35-39
Segunda Etapa (exigida para publicação)
The manuscript should be submitted with following format:
should be typed in Times New Roman; 12 font size; 1.5 spaced
lines; standard A4 paper (210 x 297 mm), side margins 2.5 cm
wide; and not exceed 16 pages, including tables and figures.
In the first page should contain the title of paper, Abstract
and Keywords. The tables and figures should be numbered consecutively in Arabic numerals, which should be indicated in the
text and annexed at the end of the paper. Figure legends should
be written immediately below each figure preceded by the word
Figure and numbered consecutively. The table titles should be
written above each table and preceded by the word Table followed by their consecutive number. Figures should present the
data source (Source) above the legend, on the right side and no
full stop; and tables, below with full stop.
The manuscript in PORTUGUESE should be assembled in the
following order: TÍTULO in Portuguese, RESUMO (followed by
Palavras-chave), TITLE in English; ABSTRACT in English (followed by keywords); 1. INTRODUÇÃO (including references);
2. MATERIAL E MÉTODOS; 3. RESULTADOS E DISCUSSÃO; 4.
CONCLUSÃO (if the list of conclusions is relatively short, to the
point of not requiring a specific chapter, it can end the previous
chapter); 5. AGRADECIMENTOS (if it is the case); and 6. REFERÊNCIAS, aligned to the left.
The article in ENGLISH should be assembled in the following order: TITLE in English; ABSTRACT in English (followed by
keywords); TITLE in Portuguese; ABSTRACT in Portuguese (followed by keywords); 1. INTRODUCTION (including references);
2. MATERIAL AND METHODS; 3. RESULTS AND DISCUSSION;
4. CONCLUSIONS (if the list of conclusions is relatively short,
to the point of not requiring a specific chapter, it can end the
previous chapter); 5. ACKNOWLEDGEMENTS (if it is the case);
and 6. REFERENCES.
The article in SPANISH should be assembled in the following order: TÍTULO in Spanish; RESUMEN (following by Palabrallave), TITLE of the article in Portuguese, ABSTRACT in Portuguese (followed by keywords); 1. INTRODUCCTIÓN (including
references); 2. MATERIALES Y MÉTODOS; 3. RESULTADOS Y
DISCUSIÓNES; 4. CONCLUSIONES (if the list of conclusions is
relatively short, to the point of not requiring a specific chapter, it
can end the previous chapter); 5.RECONOCIMIENTO (if it is the
case); and 6. REFERENCIAS BIBLIOGRÁFICAS.
The section headings, when necessary, should be written
with the first letter capitalized, preceded of two Arabic numerals
placed at the beginning of the paragraph.
References cited in the text should include the author’s
last name, only with the first letter capitalized, and the year
in parentheses, when the author is part of the text. When the
author is not part of the text, include the last name in capital
letters followed by the year separated by comma, all in parentheses.
Abstracts should be concise and informative, presenting the
key points of the text related with the objectives, methodology,
results and conclusions; it should be written in a sequence of
sentences and must not exceed 250 words.
For paper submission, the author(s) should access the online
submission Web site www.cerpch.unifei.edu.br/submeterartigo
(submit paper).
The Magazine PCH Notícias & SHP News accepts papers in Portuguese, En-glish and Spanish. Papers in foreign languages will be
requested a declaration of a specialist in language revision.
Second Step (required for publication)
O artigo depois de analisado pelos editores, poderá ser devolvido ao(s) autor(es) para adequações às normas da Revista
ou simplesmente negado por falta de mérito ou perfil. Quando
aprovado pelos editores, o artigo será encaminhado para três
revisores, que emitirão seu parecer científico. Caberá ao(s)
autor(es) atender às sugestões e recomendações dos revisores;
caso não possa(m) atender na sua totalidade, deverá(ão) justificar ao Comitê Editorial da Revista.
After the manuscript has been reviewed by the editors, it is
either returned to the author(s) for adaptations to the Journal
guidelines, or rejected because of the lack of scientific merit and
suitability for the journal. If it is judged as acceptable by the
editors, the paper will be directed to three reviewers to state
their scientific opinion. Author(s) are requested to meet the reviewers, suggestions and recommendations; if this is not totally
possible, they are requested to justify it to the Editorial Board.
Obs.: Os artigos que não se enquadram nas normas acima
descritas, na sua totalidade ou em parte, serão devolvidos e
perderão a prioridade da ordem sequencial de apresentação.
Obs.: Papers that fail to meet totally or partially the guidelines above described will be returned and lose the priority of the
sequential order of presentation.
42

anotações - cerpch

Transcrição

Documentos relacionados

- cerpch

ig igfunny facebook namesdog knott vidieo of massage in europe

ig igan acrostic poem for nathaniel hawthorneintegumentary aol

Farming system changes in South-West China: Impacts of rubber

CNPJ/MF 02 - Tractebel Energia

neste documento

CNPJ/MF 02 - Tractebel Energia

The Brazilian Wind Energy Market

João Flávio Nogueira, MD Bologna, Italy – 1700`s

audio shop heat

Net Zero Energy Buildings, towards innovation and development

Apresentação Oportunidades de Financiamento DS3

Alternative Fuel Products

Effectiveness of the endotracheal tube cuff on the trachea: physical

Dry Rolling Friction and Wear of Elastomer Systems and

Toshima kida