LEQ III - Técnico Lisboa - Autenticação

advertisement
3. Módulo de Instalações Piloto – Prof. Jorge de Carvalho
3.1.
Objectivo
Familiarizar os alunos com a Investigação e Desenvolvimento que acompanha o arranque de
uma Instalação Piloto. Interpretação dos dados experimentais obtidos, assim como a sua
importância na transferência para Instalações Piloto de maiores dimensões. Interpretação de
dados experimentais obtidos em unidades de dimensões pré-industriais. Análise crítica das
condições operatórias nas várias etapas do processo em estudo
3.2.
Introdução
Os trabalhos a serem realizados inserem-se em duas áreas do conhecimento.
Uma das áreas é a extracção não dispersiva, com a recuperação de zinco por membranas
líquidas em contactores de fibras ocas. No laboratório dispomos de algumas unidades de
contactores, cuja dimensão permite extrapolar os resultados para uma escala industrial.
A outra área será no domínio do tratamento de Efluentes Industriais, recorrendo à
tecnologia dos Leitos mistos com adsorção física e/ou química e para alguns grupos no
recurso à tecnologia da filtração integrada com secagem de bolos.
Dentro das duas áreas referidas (IP1 e IP2) são fornecidos aos alunos artigos científicos
desenvolvidos no Grupo de Hidrometalurgia e Ambiente do IST.
3.3.
Trabalho laboratorial (1aula): IP1 ou IP2 – com relatório detalhado
IP1 – Responsável: Teresa Reis
Recuperação de zinco por membranas líquidas em contactores de fibras ocas
IP2 – Responsável: Jorge de Carvalho/Francisco Lemos
Estudo da aplicação de leitos mistos e filtração integrada no tratamento de
águas
1
3.4.
Informação para IP1
3.4.1. Recuperação de zinco por membranas líquidas em contactores de fibras ocas.
O revestimento de zinco é muito utilizado como protector do aço contra a corrosão. As peças
em fim de vida são sujeitas, muito frequentemente, a decapagem com ácido clorídrico tendo
em vista a recuperação do zinco, metal de elevado valor económico.
O zinco será extraído do banho esgotado por membranas liquidas utilizando novos
extractantes (1-(3-piridil)undecano-1-ona oxima e respectivos sais quaternários) sintetizadas
pelo grupo liderado pela Drª Karolina Wieszczyka do Institute of Chemical Technology and
Engineering (Poznan University of Technology). O processo de extracção será realizado em
contactores de fibras ocas com a fase aquosa a circular no interior das fibras a e dispersão
(fase orgânica e fase regenerante) a circular no lado da caixa. Esta configuração é designada
por PEHFSD (pseudo-emulsion hollow fibre strip dispersion). Irá analisar-se a influência
composição da pseudo-emulsão e da fase de alimentação na recuperação de zinco. Em
Anexo 1 encontra-se um artigo científico relacionado com o trabalho experimental.
Fase
aquosa
Fase
aquosa
Pseudoemulsão
Pseudoemulsão
Figura 1: Esquema do contactor de fibras ocas (2,5”× 8” da Celgard USA. ~ 10800 fibras
em polipropileno, dext=0,26 mm, dint =0,22 mm, dporos=0.03 mm, A=1,16 m2 (2500 m-1))
Figura 2: Esquema da instalação de fibras ocas.
2
3.5.
Informação para IP2
3.5.1. Descrição do Processo de Tratamento para a Remediação da água da
Lagoa das Furnas
Objectivos
Propor um Processo de Remediação da água da Lagoa das Furnas, diminuindo o teor
em fósforo total para níveis aceitáveis (<35 ppb) e os níveis de clorofila abaixo das
10 mg/m3
Clarificar a água da Lagoa, reduzindo a turbidez inicial de 9-20 NTU para valores
inferiores a 0,2 NTU.
Diminuir os teores de azoto e fósforo solúvel, ao baixar o teor de fósforo particulado
da água.
Tratamento que não introduza contaminantes na água, nomeadamente Fe(III) e
Al(III), para níveis acima dos que existem presentemente.
O processo de tratamento a inovar deverá ser sustentável, isto é, os resíduos
produzidos durante o tratamento deverão ser uma commodity com interesse Regional
Pelos resultados obtidos no IST podemos estimar que em media por cada m3 de água
tratada iremos obter 75g de sólido húmido com 80% de água. Obviamente que esta
quantidade varia com a altura do ano.
Igualmente, após filtração dos rejeitados (“Backwash”) concentrados,
estimamos que para uma unidade operando 250 m3/h, 24h por dia e 7 dias por
semana deverá produzir-se cerca de 1170 kg de lama com 30% de humidade.
3
Estado da Arte e Caracterização
200
A N Á L I S E S E F E C T UA DA S À Á G UA L A G O A D A S F U R N A S
159,9
147,2
135,9 136,5
150
115,9
84,5
100 80,6
64,2
50 23,9
1717
0
28-mai-14
97,7
71,2
16,7
18,7
109,9
96,589,9
77,9
75,2
72,571,9 65,2
71,9
68,5
64,5
82,880,6
21,717,8
28-jun-14
28-jul-14
46,5
38
17
28-ago-14
Fósf. Total (ppb)
48,5
48,5
22,519
28-set-14
28-out-14
19,6
28-nov-14
Fósf. Solúvel (ppb)
Figura 3 - Caracterização da água da Lagoa das Furnas de 28/05/2014 a 10/12/2014
Tabela 1 - Análise dos parâmetros relevantes da água da Lagoa (Novembro 2013)
Parâmetro
Resultado
Alcalinidade Total
39,3 mg(CaCO3)/L
Fosfato
<0,01 mg(P)/L
Sílica
11 mg(SiO2)/L
Fósforo Total
0,029 mg(P)/L
Fósforo Solúvel
0,018 mg(P)/L
Azoto Amoniacal
0,14 mg(N)/L
Azoto Kjeldahl
0,59 mg(N)/L
Azoto Orgânico
0,45 mg(N)/L
Azoto Total
1,6 mg(N)/L
Nitrato
1,0 mg(N)/L
Nitrito
0,005 mg(N)/L
Cálcio Total
3,5 mg/L
Magnésio Total
2,7 mg/L
Bário Total
<0,06 mg/L
Estrôncio
<0,020 mg/L
Ferro Total
<0,23 mg/L
4
Tabela 2 - Critério Nacional para a avaliação do estado trófico de uma água (INAG 2002)
O fósforo da Lagoa das Furnas é o agente determinante para a sua eutrofização,
contudo este fósforo encontra-se maioritariamente (70%), adsorvido à matéria
orgânica e inorgânica, isto é, na forma particulada, estas partículas têm diâmetros
inferiores a 1 micra e somente a fracção residual se encontra na forma de fósforo
solúvel
Para além do exposto vários catiões e aniões, nomeadamente, os portadores de
átomos de azoto, também se encontram em larga percentagem adsorvidos às
partículas existentes
Em algumas estações do ano, a presença de microalgas contribuem para o aumento
de clorofila no meio aquoso.
Dos processos disponíveis na literatura para a remoção de fósforo de uma água,
destacam-se os métodos de Precipitação Química, os Métodos Biológicos, os
Métodos Físicos e, recentemente, os de Adsorção Química, incluindo a retenção em
leitos mistos em colunas.
A maioria dos meios aquosos contêm fósforo ao nível dos ppm e na forma solúvel.
A Lagoa das Furnas, devido à sua envolvência, tem a particularidade de os níveis de
fósforo serem da ordem dos 50 a 90 ppb e este se encontrar maioritariamente
adsorvido a partículas de diâmetros inferiores a 1 micra. Igualmente a composição da
água varia substancialmente ao longo do ano.
5
3.5.2. Processo desenvolvido que foi conducente à construção de uma unidade
demonstrativa para tratar 5m3/h de água da Lagoa das Furnas
Com os estudos realizados ao nível de bancada e semi-piloto no IST, viemos a
propor, um tratamento que foi sendo testado de modo descontínuo no IST, mas
simulando sequencialmente todas as etapas.
O processo de tratamento com as etapas abaixo, foi proposto e está a ser
implementado junto à Lagoa das Furnas, com unidade demonstrativa do processo.
Neste processo, iremos utilizar areia de sílica com diâmetros entre 2 mm e 4 mm no
fundo da coluna, até uma altura de 25 cm e areia entre 1 mm e 2 mm de diâmetro, a
qual estará sob a areia grossa, estendendo-se ao longo da altura do leito.
Granulometrias
(mm)
1 <d<2
Areia
Sílica
de
2<d<4
Descrição das Etapas
1 – Captação de água da Lagoa através de uma bomba montada numa jangada móvel,
permitindo fazer captação em zonas com concentrações muito diferenciadas em algas e
eventualmente fósforo e certos nutrientes.
2 – Sistema de pré-filtração da água (1) a ser introduzida na unidade de tratamento, com a
facilidade de lavagem automática da tela, uma vez colmatada. Com esta unidade removemse os sólidos de dimensões superiores à malha da tela, microalgas, etc e reduz-se
substancialmente o nível de clorofila da água.
3 – Coluna de leito de areia de várias granulometrias (2), com possibilidade introdução de
outros materiais se se considerar necessário. O fluxo de partida nesta coluna é de baixo para
cima, podendo vir a ser alterado o seu sentido. Esta coluna de leito misto está preparada para
fazer lavagem do leito, isto é, a remoção de tudo o que ficar retido no leito, logo que se atinja
um nível de colmatação e utilizando um volume de água de lavagem correspondente a 2% do
volume de água que atravessar o leito.
4 - Unidade de ultrafiltração, pela qual irá passar a água tratada no leito misto, afim de se
reduzir a turbidez da fase aquosa e consequentemente os níveis de fósforo (6). A fase aquosa
a tratar circula no exterior das fibras ocas da unidade
6
5 – Sempre que o Leito Misto (2) e/ou a unidade de ultrafiltração (6) necessitar de ser limpa,
vai buscar-se a um tanque de água filtrada (4 e 5), água para realizar o Backwash.
6 – Todos os Backwash são enviados para um tanque de Backwash (7), a concentração destas
correntes aquosas é inferior a 0,7 g/L de sólidos.
7 – Estas soluções de Backwash terão de ser concentradas e para o efeito dispomos de duas
correntes em paralelo. Dispomos de uma unidade de ultrafiltração tubular cerâmica (8) e
uma unidade de ultrafiltração de fibras ocas também para concentração (10). Nestas
unidades de ultrafiltração a suspensão circula no interior dos tubos.
8 – Existem 2 tanques (9 e 11) que recebem a corrente aquosa concentrada nestas duas
unidades de ultrafiltração.
9 – Dispomos de 2 decantadores (14 e 15) para cada uma das correntes de Backwash
concentrado. Após utilizarmos um floculante (19) vamos produzir um espessado que será
armazenado em tanques (18) para posterior estudo numa unidade de filtração com
compressão do bolo e vácuo.
10 – Todos os efluentes da unidade demonstrativa serão encaminhados para o tanque (20) de
onde são enviados para uma distância de 350 m, onde se encontra a saída dos efluentes da
Lagoa para uma linha de água.
11- A unidade demonstrativa tem instrumentação e controlo o que nos permite operar com
alguma segurança.
7
Diagrama do Processo de Remediação da água de uma lagoa recorrendo a métodos fisico-químicos para o tratamento
8
Diagrama de Instrumentação e Controlo
Linha de Backwash
Linha de Água Tratada
Linha de Água não Tratada
V2
V3
LT
31
Tanque de
Equalização
3
VM1
LT
32
LC
31
LT
33
Tanque
Água Limpa - 1
4
CV3
Tanque
Água Limpa - 2
5
LT
51
LT
52
Tanque
Backwash
7
LC
51
LT
53
LT
71
LT
72
LC
71
V5
CV2
V6
V7
V4
V1
LC
111
Leito
Misto
(Hubel)
2
Pré – Filtro
(Hubel)
UF
(Trativi)
6
UF
Conc.
(IST)
8
Tanque
Concentrado
(IST) 9
LT
91
LT
92
UF
Conc.
(Trativi)
10
LC
91
Tanque
Concentrado
(Trativi) 11
LT
111
LT
112
Contador
1
XC
1
Água a
tratar
Bomba de concentrado (12)
LT
201
LC
201
LT
202
Bomba
alimentação
LT
203
Tanque
De
Descarga
20
Bomba de clarificado (13)
Decantador
(14)
Floculante
(19)
Reservatório
Espessado
(18)
Decantador
(15)
CV1
Bomba Descarga
(20)
Bomba de espessado (17)
Bomba de espessado (16)
LAGOA
9
Controlo Tanque de Equalização (3) / Tanque Água Limpa 1 (4) / Tanque Água Limpa
2 (5) / UF - Trativi (6 )/ Filtro-Hubel (2)
Ciclo de Filtração UF - Trativi (6) / Filtro - Hubel (2)
A válvula V2 é aberta.
O LC31 controla o nível no tanque 3 da seguinte forma:
A válvula V4 é normalmente fechada.
Quando atinge o nível LT31 abre válvula V4 (descarga Tanque 4).
Quando atinge o nível LT32 manda fechar a V4.
Se atingir o nível LT33 emitir alarme visual, sonoro e envia sinal de paragem à
unidade UF6.
Ciclo de Filtração UF - Trativi (6) / Filtro - Hubel (2)
O LC41 controla o nível dos Tanques 4 e 5 da seguinte forma:
A válvula V3 é normalmente fechada.
Quando atinge o nível LT41 abre a válvula V3.
Quando atinge o nível LT42 fecha a válvula V3.
Se atingir o nível LT43 emitir alarme visual e sonoro.
Ciclo de lavagem - Filtro-Hubel (2)
A válvula V2 é fechada.
O LC31 controla o nível no tanque 3 da seguinte forma:
A válvula V4 é normalmente fechada.
Quando atinge o nível LT31 abre válvula V4 (descarga Tanque 3).
Se atingir o nível LT33 emitir alarme visual e sonoro e envia sinal de interrupção de
lavagem à unidade UF6.
O LC41 controla o nível do tanque da seguinte forma:
A válvula V3 é aberta enquanto o nível for superior a LT43.
Se atingir o nível LT43 emitir alarme visual, sonoro e envial sinal e paragem à
unidade Filtro 2.
10
Ciclo de lavagem – UF - Trativi (6)
A válvula V3 é fechada.
A válvula V2 é aberta.
O LC31 controla o nível no Tanque 3 como no Ciclo de Filtração.
O LC41 controla o nível do tanque da seguinte forma:
A válvula V1 é aberta enquanto o nível for superior a LT33.
Se atingir o nível LT43 emitir alarme visual, sonoro e envia sinal de interrupção de
lavagem da unidade UF 6.
Zona Tanque Backwash (7) / Tanque UF concentrado – IST (9) / Tanque UF
concentrado – Trativi (11)
O LC71 controla o nível do tanque da seguinte forma:
A válvula V5 é fechada enquanto o nível for inferior a LT72.
Se atingir o LT91 emitir sinal visual e sonoro.
Se atingir o nível LT71 interrompe lavagem da unidade Filtro 2 e UF6.
O LC91 controla o nível do tanque da seguinte forma:
A válvula V6 é aberta enquanto o nível for inferior a LT92.
Se atingir o nível LT91 emitir alarme visual e sonoro.
O LC111 controla o nível do tanque da seguinte forma:
A válvula V7 é aberta enquanto o nível for inferior a LT112.
Se atingir o LT111 emitir sinal visual e sonoro.
Zona Tanque de Descarga (20)
O LC201 controla o nível do tanque da seguinte forma:
Quando atinge o nível LT203 a Bomba de Descarga (20) arranca a meia capacidade.
Se atingir o nível LT202 a bomba Bomba de Descarga (20) começa a funcionar à
capacidade máxima.
Se atingir o LT201 emitir sinal visual e sonoro e envia informação à UF6 ou Filtro 2
(consoante a unidade que estiver em ciclo de lavagem) para interromper o ciclo de
lavagem
11
3.5.3. Construção, montagem e arranque da Unidade Demonstrativa de
tratamento da água da Lagoa das Furnas
Após a colocação dos contentores, com o respectivo equipamento instalado no
interior iniciaram-se os trabalhos de montagem de tubagem, de instalação eléctrica e
instrumentação e controlo.
12
Os trabalhos de montagem de tubagem e instrumentação e controlo foram terminados durante o mês de Dezembro.
Unidade Demonstrativa
Unidade de Ultrafiltração
Unidade de Leito misto
Decantadores
Unidade de ultrafiltração
concentração tubular (fibras ocas)
Unidade de ultrafiltração de
concentração tubular (cerâmica)
13
Alterações realizadas na unidade de Leito misto para operar em fluxo
ascendente e descendente
A unidade leito misto foi inicialmente projectada para operar em fluxo ascendente, no
entanto após os primeiros testes decidiu-se tornar a unidade mais versátil, e passar a operar
quer em fluxo ascendente quer em fluxo descendente. Para tal realizaram-se alterações na
unidade a nível de tubagem, colocação de válvulas manuais, localização de electroválvulas e
alteração de software.
As alterações a nível de tubagem, válvulas manuais e electroválvulas foram executadas no
início de Fevereiro, no entanto, a alteração de software apenas foi possível no final do mês
de Fevereiro (22 e 23) o que impediu que a operação da unidade de leito misto em fluxo
descendente no modo automático. Adicionalmente no processo de alteração da tubagem foi
danificado um transmissor de pressão o que impedia o funcionamento da unidade de leito
misto em modo automático, pois o modo de lavagem é accionado quando a perda de carga
estabelecida é atingida.
Em simultâneo, às alterações estruturais no leito misto e considerando os resultados obtidos
para a turbidez da água da lagoa à saída, adicionaram-se 150kg de areia APAH30 para
melhorar estes resultados.
14
15
Considerações finais dos ensaios realizados na Lagoa das Furnas
Com base nos resultados obtidos nas unidades de Leito misto e ultrafiltração que
integram a unidade demonstrativa da Lagoa das Furnas:
Leito misto – Boa autonomia, mas ineficiente redução do teor de fósforo particulado
(48 para 37 ppb) e solúvel (39 para 37 ppb) e da turbidez da água da lagoa, quer em
fluxo ascendente quer em fluxo descendente;
Ultrafiltração – Eficiente na redução do teor em fósforo particulado (48 para 8 ppb) e
solúvel (38 para 8 ppb), mas autonomia reduzida (< 4 horas de operação).
Os resultados obtidos na unidade demonstrativa das Furnas, isto é, valores de turbidez e
níveis de fósforo à saída da coluna de leito misto mais elevados do que o esperado, levaramnos a especular que as condições da água da lagoa que estava a ser utilizada, era diferente da
utilizada anteriormente no IST. Isto é, neste momento deveríamos ter partículas muito mais
pequenas que não ficavam retidas.
3.5.4. Dimensão de partículas
A dimensão das partículas presentes na água da Lagoa das Furnas e nas fases
subsequentes de tratamento foi determinada através de duas técnicas diferentes:
Dispersão dinâmica de luz – Microtrac Nano-Flex 180
Difração laser – Malvern Mastersizer 2000
Dimensão de partículas – Dispersão dinâmica de luz
Dimensão de partículas de Fósforo e Sílica na água da Lagoa
25
10 seg Fósforo
20
20 seg Fósforo
15
%
%
30 seg Fósforo
10
5
0
0,0
2,0
4,0
Diâmetro (µm)
6,0
8,0
45
40
35
30
25
20
15
10
5
0
10 seg Sílica
20 seg Sílica
30 seg Sílica
0,0
1,0
2,0
3,0
4,0
5,0
6,0
Diâmetro (µm)
16
7,0
35
25
Após leito misto (1)
30
10 seg
Fósforo
20 seg
Fósforo
30 seg
Fósforo
%
20
15
10
20
10 seg Fósforo
15
20 seg Fósforo
%
25
Após leito misto (2)
30 seg Fósforo
10
5
5
0
0
0,0
0,2
0,4
0,6
0,8
1,0
0,0
1,0
Diâmetro (µm)
20
2,0
3,0
Diâmetro (µm)
Backwash concentrado (1)
25
Backwash concentrado (2)
10 seg Fósforo
20
10 seg Fósforo
%
10
20 seg Fósforo
20 seg Fósforo
15
30 seg Fósforo
%
15
10
30 seg Fósforo
5
5
0
0,0
2,0
4,0
Diâmetro (µm)
6,0
8,0
0
0,0
2,0
4,0
Diâmetro (µm)
6,0
8,0
Conclusões
Os ensaios realizados pelas técnicas dispersão dinâmica de luz e difração laser
permitiram obter as seguintes conclusões:
o tamanho das partículas de fósforo presentes na água da lagoa, são na sua maioria
inferiores a 1mm. Estes resultados comprovam que os valores de turbidez da água da
lagoa após o leito misto na unidade demonstrativa se devem a partículas inferiores a
1mm;
o Leito misto é eficiente na retenção de partículas de fósforo com dimensões
superiores a 1mm;
no Backwash concentrado, isto é, após ter passado na unidade de ultrafiltração
tubular, estão presentes as partículas inferiores a 1mm o que permite concluir que a
unidade de Ultrafiltração da Trativi retém a maioria das partículas que passam pelo
Leito misto;
existem partículas de sílica inferiores a 1mm na água após a passagem pelo Leito
misto, que influenciam os valores de turbidez da água à saída;
as partículas presentes na água após a unidade de Ultrafiltração apresentam
dimensões muitos reduzidas que não foi possível determinar por nenhuma das duas
técnicas experimentais.
17
Perante os resultados obtidos, a fim de se aglutinar as partículas realizaram-se ensaios em
Jar-Test, cujas conclusões se apresentam:
a aplicação de floculante catiónico, após a aplicação de coagulante orgânico
demonstra ser ineficiente na redução da turbidez da água;
o coagulante PAX XL10 apresenta o melhor resultado de turbidez da água da lagoa
(0,58/0,61 NTU), no entanto, este coagulante é à base alumínio e como tal afastou-se
a hipótese da sua utilização;
os coagulantes orgânicos com melhores resultados são o Chemifloc PA47 e o
Superfloc C577;
os floculantes aniónicos com melhores resultados são Superfloc A-100 e o SNF alta
carga;
a razão coagulante/floculante mais favorável é de 5:1 – (20ppm:4ppm e/ou 10 ppm
:2ppm)
Com base nestes resultados decidiu-se realizar ensaios com coagulantes e floculantes em
colunas de leito misto laboratoriais com 2,5cm de diâmetro, quer em fluxo ascendente quer
em fluxo descendente. A maioria dos grupos irá acompanhar a realização destes ensaios.
3.6.
Parte Experimental do trabalho em colunas de leitos misto
Os leitos mistos podem são utilizados para a remoção de pequenas partículas que
ficam retidas na porosidade do leito e/ou por processos de remoção por adsorção química em
vários tipos de leitos (exemplo: Hidróxido Férrico granular).
No folheto da GEH (Anexo 2), que se junta em anexo, está descrito o processo da
tecnologia de leitos mistos assim como as etapas de um ciclo operativo, que ilustra o
trabalho experimental.
Protocolo
1 – Em primeiro lugar os alunos irão ensaiar coagulantes e floculantes na água da
Lagoa das Furnas, a fim de avaliar o seu efeito na eliminação da turbidez dessa água.
2 – Cada grupo irá fazer um estudo num leito misto numa coluna laboratorial com 2,5
cm de diâmetro e uma altura de leito ≈ 22 cm. Abaixo junta-se uma fotografia do
equipamento onde o trabalho irá ser realizado.
18
Figura 4 – Coluna de leito misto
Cada grupo irá estudar o efeito da relação mássica de coagulante/floculante para
reagentes comerciais, na turbidez da água após passagem no leito, assim como na redução do
teor em fósforo particulado e solúvel presente na água da Lagoa, parâmetro relevante nestes
estudos. O que se acabou de referir depende da velocidade superficial de passagem do fluido
através do leito. Para cada grupo será indicada a velocidade de passagem a utilizar.
Cada grupo de alunos irá realizar ensaios em fluxo descendente e ascendente, assim
como proceder à lavagem do leito, comparando esta velocidade experimental de lavagem
com os valores teóricos de fluidização de um leito.
4 – Será fornecido aos alunos a técnica de análise de fósforo total, fósforo solúvel e
turbidez. Os alunos poderão observar o procedimento analítico no tratamento de resultados
que irão fazer, assim como detalhes das características do leito utilizado nas colunas
laboratoriais.
19
Principais características das areias com diferentes granulometrias que constituem o leito misto utilizado
nas colunas laboratoriais
Modelo
areia
Massa areia
(g)
Volume
areia+água
(mL)
Volume
água (mL)
Porosidade
Densidade
(g/mL)
110,870
66
28
0,424
2,918
109,037
64
27
0,422
2,947
107,739
59
23
0,390
2,994
APAH 6
(2-4 mm)
APAH 12
(1-2 mm)
APAH 30
(0,4- 0,8
mm)
Composição do leito misto utilizado nas colunas laboratoriais
Areia APAH 6
2<d(mm)<4
Altura
(cm)
Volume
(cm3)
massa
areia (g)
3,5
17,17
32,16
Leito
Altura
(cm)
Volume
Areia APAH12
(cm3)
1<d(mm)<2
massa
areia (g)
15
73,59
162,34
Altura
4
(cm)
Volume
Areia APAH30
19,63
(cm3)
0,4<d(mm)<0,8
massa
areia 49,099
(g)
Determinação do Fósforo Total e Dissolvido – Standard Methods 4500P-B e E
I.
•
•
Condicionamento das Amostras
Filtrar as amostras e fazer o ensaio;
Se não se for realizar o ensaio filtrar e adicionar 1ml de H2SO4 concentrado por 500
ml de amostra.
Nota: Para a determinação do Fósforo Total não é necessário fazer-se filtração das
amostras. Apenas para o fósforo dissolvido, utilizando um papel de filtro com 0,45
microns de porosidade.
II.
•
Descontaminação do Material
Lavar o material a utilizar com uma solução aquosa de HNO3 (2,5%), seguida de 8
lavagens com água milipor.
20
Preparação do Padrão de Fósforo
•
Pesar 1,065g de hidrogenofosfato de amónio e dissolver em 250 ml de água milipor
(obtenção de uma solução com 1000ppm).
Ensaio:
1- Medir 150 ml de amostra, previamente agitada para um erlenmeyer de 250 ml. Medir
também 150ml H2O milipor com solução de padrão de controlo de fósforo
(1000ppm) diluída a 1:100, PC-0,01mg/l (150µl) e PC-0,03mg/l (450µl).
III.
Determinação do Fósforo Total
Nota: Para a determinação do fósforo total não é necessário filtrar.
2- Adicionar 1 ml de H2SO4 (3ml/10ml), 0,4g de peróxido-dissulfato de amónio as
amostra medidas em 1 e levar à secura até um volume final de aproximadamente 10
ml.
3- Após a evaporação do solvente (cerca de 2/3 horas) vai ser feita uma nova filtração
para remover as partículas não digeridas
4- Filtração:
a. Colocar em balões volumétricos de 50 ml (de modo a aumentar a
concentração em 3x para se atingir um valor mínimo de fósforo de 10µg) um
funil de vidro e papel de filtro (41 microns de porosidade) dobrado em 4 e
proceder a filtração, com lavagens dos erlenmeyeres com água milipor (um
baixo volume, cerca de 30ml) de modo a lavar bem o erlenmeyer.
5- Após a filtração adicionar aos balões volumétricos uma gota de solução alcoólica de
fenolftaleína e adicionar algumas gotas de solução básica de NaOH a 25% até as
amostras ficarem com uma coloração cor-de-rosa (deste modo realizou-se um acerto
do pH) e perfazer o volume com água milipor.
6- De seguida adiciona-se 8ml de reagente combinado a cada amostra e espera-se
durante 10 minutos (irá ser observada uma mudança de cor de rosa para azul que será
mais intensa quanto maior a quantidade fósforo existente na amostra (este reagente
combinado serve para provocar a reacção de colorimétrica).
21
Preparação do Reagente Combinado
Nota: O reagente combinado, constituído por ácido sulfúrico, tartarato de antimónio e
potássio, molibdato de amónio e ácido ascórbico, deverá ser preparado para cada utilização
(“na hora”)
Preparar soluções de molibdato de amónio e de ácido ascórbico (as outras soluções já podem
estar preparados. Note-se que o tartarato de antimónio e potássio deverá ser armazenado no
frio).
•
•
Solução de H2SO4 5N – 150ml de H2SO4 concentrado/1l de H2O Milipor;
Tartarato de antimónio e potássio – 0,686g/250ml H2O Milipor.
Para 200ml de reagente combinado:
•
•
•
•
100ml de H2SO4;
10ml de Tartarato;
30ml de Molibdato;
60ml de ác. Ascórbico.
Para 100ml de reagente combinado:
•
•
•
•
50ml de H2SO4;
5ml de Tartarato;
15ml de Molibdato (2g/50ml);
30ml de ác. Ascórbico (0,88g/50ml).
Para 250ml de reagente combinado:
•
•
•
•
125ml de H2SO4;
12,5ml de Tartarato;
37,5ml de Molibdato (2g/50ml);
75ml de ác. Ascórbico (1,76g/100ml).
Nota: Os reagentes deverão ser adicionados pela ordem acima descrita. Após a adição do
último reagente (ácido ascórbico) a solução de reagente combinado fica amarela.
IV.
Análise no espectrofotómetro UV-Vis
Traçar inicialmente uma recta de calibração.
22
3.7. Concentração do Backwash proveniente da lavagem do leito de areia e das
membranas de ultrafiltração
Introdução
Nos Açores irá ser produzido Backwash concentrado, armazenado em tanques de 1m3 e
enviados para o IST.
A descrição genérica do Processo é apresentada abaixo:
Água
limpa
Água da
lagoa
Pré-filtração
Leito misto
de areia
Água
limpa
Unidade de
ultrafiltração
por fibras
ocas
Água
com 0,2
NTU
Backwash 200 NTU
Concentração do
backwash em
unidade de
ultrafiltração
tubular 1000 NTU
Concentração do
backwash em
decantador
convencional
Filtração
integrada
Bolos com 20%
de Humidade
23
3.7.1. Parte experimental para os grupos que não realizam os ensaios em colunas
de leitos misto
Objectivos do trabalho proposto
Estudar e optimizar as condições operatórias de desidratação do espessado obtido no flowsheet anterior, utilizando um filtro prensa de membranas com desidratação sob vácuo. Para o
efeito serão estudados e determinados os parâmetros abaixo.
a) Ensaios preliminares com vários tipos de placas de filtração, tipo e geometria de telas.
b) Determinação da resistência específica dos bolos de filtração e determinação da
compressibilidade dos bolos.
c) Avaliação do efeito da espessura dos bolos e do tempo e do tempo de secagem na
eficiência da etapa de desidratação térmica sob vácuo.
d) Avaliação do efeito de ciclos de calor e frio em superfícies alternadas dos bolos de
filtração na eficiência da etapa de desidratação térmica sob vácuo.
Nota: Cada grupo de alunos irá fazer apenas um ensaio de filtração em condições
específicas bem determinadas.
Em Anexo 3, junta-se um artigo sobre o modo de funcionamento da unidade de filtração
integrada.
4. Execução do relatório
Cada grupo terá de apresentar um relatório sucinto cobrindo os seguintes itens:
i)
ii)
iii)
iv)
objectivo do trabalho;
descrição de uma tecnologia alternativa ao processo proposto;
descrição detalhada do trabalho experimental com análise crítica dos resultados,
que serão fornecidos periodicamente. Explicação em detalhe do modo de
lavagem do leito quer em fluxo ascendente quer em fluxo descendente;
estimativa teórica da velocidade superficial necessária para o início da fluidização
da areia intermédia, fina e grossa. Quando dois grupos tiverem feito o mesmo
trabalho experimental cada grupo apresentará os resultados para uma fracção
granulométrica da areia. Deverá ser discutido o valor teórico para a velocidade de
fluidização e o valor da velocidade de lavagem que foi utilizada;
24
v)
vi)
irá ser apresentado a cada grupo um resultado experimental obtido na Instalação
Piloto a fim de ser comentado;
conclusões e interesse do trabalho.
25
Anexo 1
Separation and Purification Technology 154 (2015) 204–210
Contents lists available at ScienceDirect
Separation and Purification Technology
journal homepage: www.elsevier.com/locate/seppur
Recovery of zinc(II) from chloride solutions using pseudo-emulsion
based hollow fiber strip dispersion (PEHFSD) with 1-(3-pyridyl)undecan1-one oxime or tributylphosphate
Karolina Wieszczycka a,⇑, Magdalena Regel-Rosocka a, Katarzyna Staszak a, Aleksandra Wojciechowska a,
M. Teresa A. Reis b, M. Rosinda C. Ismael b, M. Lurdes F. Gameiro b, Jorge M.R. Carvalho b
a
Poznan University of Technology, Institute of Chemical Technology and Engineering, ul. Berdychowo 4, 60-965 Poznan, Poland
CERENA – Centre for Natural Resources and the Environment, Department of Chemical Engineering, Instituto Superior Técnico, Universidade de Lisboa, Av. Rovisco Pais, 1049-001
Lisboa, Portugal
b
a r t i c l e
i n f o
Article history:
Received 19 July 2015
Received in revised form 16 September
2015
Accepted 20 September 2015
Available online 25 September 2015
Keywords:
Pseudo-emulsion hollow fiber strip
dispersion (PEHFSD)
3-Pyridineketoxime
Tributyl phosphate (TBP)
Zinc(II) extraction
a b s t r a c t
The recovery of zinc from chloride solutions using pseudo-emulsion based hollow fiber strip dispersion
(PEHFSD) technique was investigated. The novel extractant, 1-(3-pyridyl)undecan-1-one oxime, and a
well-known one, tributyl phosphate (TBP), were used in the processes. The influence of several parameters, including the initial concentration of Zn(II) and sodium chloride in the aqueous phase and the type of
extractant on Zn(II) extraction was studied. The Zn(II) transport was analyzed on the basis of the overall
mass transfer coefficient of permeation. The oxime was shown to be a potential carrier of zinc from chloride medium, being a promising alternative to the classical extractant TBP.
! 2015 Published by Elsevier B.V.
1. Introduction
Zinc coatings are widely used to protect steel against corrosion.
Thanks to the electrochemical potential of zinc (cathodic protection) and the thickness of zinc layer deposited on steel elements,
the coating has relatively favorable corrosion properties in relation
to those of the other metallic coatings. The advantages of zinc coatings over other types of protection include low operational cost,
long working life and easiness in covering of complicated shapes.
A number of different methods of zinc coatings deposition on steel
surfaces are commercially available, such as hot-dip galvanizing,
metalizing, zinc-rich paint, electroplating, mechanical plating, zinc
plating. Currently, the most popular are hot dip galvanizing and
electroplating [1].
Different types of waste streams that may pose a potential
threat to the environment, are generated during preparation of
the steel surface for the deposition of zinc coatings (e.g. in the
⇑ Corresponding author.
E-mail addresses: Karolina.Wieszczycka@put.poznan.pl (K. Wieszczycka),
Magdalena.Regel-Rosocka@put.poznan.pl (M. Regel-Rosocka), katarzyna.Staszak@
put.poznan.pl (K. Staszak), aleksandra.w.wojciechowska@doctorate.put.poznan.pl
(A. Wojciechowska), teresareis@ist.utl.pt (M. Teresa A. Reis), qrosinda@mail.ist.utl.
pt (M. Rosinda C. Ismael), jcarv@ist.utl.pt (J.M.R. Carvalho).
http://dx.doi.org/10.1016/j.seppur.2015.09.017
1383-5866/! 2015 Published by Elsevier B.V.
processes of degreasing, washing, pickling). Generally, spent solutions from steel pickling in hot-dip galvanizing plants contain zinc
(II), iron (mainly iron(II)), traces of lead, chromium and other heavy
metals (max. 500 mg/L) and hydrochloric acid [1]. Recovery of
these elements is important from the ecological point of view
and additionally these waste streams can be an alternative secondary source of metal ions, interesting also for economic reasons.
Various techniques of regeneration of spent pickling solutions,
including the methods with acid recovery, such as diffusion dialysis, electrodialysis, membrane electrolysis and membrane distillation, evaporation, precipitation and spray roasting as well as those
with acid and metal recovery: ion exchange, retardation, adsorption, crystallization and membrane extraction, have been proposed
in the literature [1].
Currently, there is a tendency to combine various techniques
into hybrids, e.g. separation fully integrating extraction and stripping in a membrane contactor. The new technique known as
pseudo-emulsion based hollow fiber strip dispersion (PEHFSD)
has been proposed in the literature [2] for the recovery of metal
ions i.e. Cu(II) [3,4], Zn(II) [5,6], Au(I) [7], Co(II) [8], Cr(III) [9,10],
Cr(VI) [11] from aqueous solutions. PEHFSD provides the following
advantages over classical extraction: (i) extraction and stripping
can be carried out in one operation, (ii) amount of extractant is relatively small, (iii) no possibility of emulsion formation in water
K. Wieszczycka et al. / Separation and Purification Technology 154 (2015) 204–210
phase, (iv) process parameters are very flexible, (v) large surface
area in hollow fiber membranes, and (vi) low energy consumption.
Pseudo-emulsion based hollow fiber strip dispersion means that
the stripping phase is dispersed in the organic membrane phase,
a pseudo-emulsion being formed before injection into hollow fiber
module. During this separation proves, the extraction and stripping
occur simultaneously in a single hollow fiber contactor, which permits avoidance of membrane stability issues. In PEHFSD, the
pseudo-emulsion phase runs through the shell side, while the
aqueous feed phase flows in the lumen side. The solute is transported from the feed to the membrane and then to the stripping
phase simultaneously. As soon as the process stops, the strip phase
and organic phase get separated.
Over the last few years, the extraction of metal ions from different aqueous solutions by hydrophobic pyridylketoximes has been
widely investigated. Those compounds have been mainly proposed
as ligands for removal of metals, i.e. copper(II) [12], zinc(II) [13],
cadmium(II) [14], iron(III) [15] and cobalt(II) from chloride solutions using a liquid–liquid extraction technique [16]. However,
only 1-(3-pyridyl)undecan-1-one oxime has been proposed as an
effective extractant of zinc(II) ions from strong acidic chloride solutions, e.g. this pyridylketoxime is able to reduce the content of zinc
ions from 65 to 4.5 g/L in a two-stage process [17].
The aim of this work is the removal and recovery of Zn(II) ions
in the presence of sodium chloride and hydrochloric acid, by the
method of pseudo-emulsion based hollow fiber strip dispersion
(PEHFSD). The approach assumed in the present work includes
the study of the effects of various experimental parameters like
metal, salt and acid concentration in the feed solution as well as
the type of extractant. The modeling of mass transfer was performed on the basis of the overall mass transfer coefficient of permeation. The possibility to use the novel extractant oxime of 1-(3pyridyl)undecan-1-one for recovery of Zn(II) from chloride medium using PEHFSD technique was studied and the results were
contrasted with those obtained with the well-known extractant
tributyl phosphate (TBP).
2. Experimental
carriers for transport experiments. Sodium chloride, hydrochloric
acid (38%) and chloride salts of Zn(II) (Sigma–Aldrich, Germany)
were used as components of the aqueous phase. Sodium sulfate
(Sigma–Aldrich, Germany) was used as component of the oxime
of 1-(3-pyridyl)undecan-1-one stripping solution. All the chemicals were of high purity analytical grade. The aqueous solutions
were analyzed for zinc(II) concentration by AAS using a Perkin
Elmer-AAnalyst 200 at 213 nm in the air–acetylene flame.
2.2. PEHFSD experiments
The experimental setup used for carrying out PEHFSD experiments is shown in Fig. 1. In this schematic diagram, HF represents
the microporous fiber module, which was a Liqui-Cel# Extra-Flow
2.5 in. ! 8 in. membrane contactor from Celgard (USA). Further
module details are given in Table 1.
The aqueous strip solution was dispersed in the organic membrane solution containing the extractant reagent. The pseudoemulsion was then pumped into the membrane module flowing
through the shell side of the fibers. The aqueous feed solution containing the target species to be extracted was flown through the
lumen (tube side) of the fibers. The continuous organic phase of
the dispersion readily wetted the pores of the hydrophobic microporous support (e.g. microporous polypropylene hollow fibers in
the module), and a stable liquid membrane (the organic phase)
supported in the pores was formed. A low pressure differential
(minimum 30 kPa in this system) between the aqueous feed solution side and the strip dispersion side was applied to prevent the
leakage of the organic solution of the strip dispersion to the aqueous side through the pores. The feed and strip dispersion phases
were in recycling mode in their respective reservoirs. Both phases
were circulated in a closed circuit to a constant value of the zinc(II)
concentration in the aqueous phase (2–3 h). At various time intervals, samples of 0.5 mL of the aqueous phase were taken to determine the content of zinc(II) ions. As soon as the process stopped,
the strip and organic phases were separated. The volume of the
pseudo-emulsion phase used in the experiment was 800 mL
(400 mL of the organic phase + 400 mL of the stripping solution),
and that of the feed solution was 800 mL. The flow rates of the feed
phase and pseudo-emulsion phase were "300 mL/min, which was
2.1. Materials and methods
1-(3-pyridyl)undecan-1-one oxime was synthesized in a twostage reaction [17]. In the first stage 1-(3-pyridyl)undecan-1-one
was synthesized by treating 3-pyridylcarbonitrile (Sigma–Aldrich,
Germany) with decylmagnesium bromide (Mg powder and dodecyl bromide were produced by Sigma–Aldrich (Germany)). In the
second stage, the synthesized ketone was treated with hydroxylamine hydrochloride (POCh, Poland) in the presence of sodium
carbonate (at pH = 7). After hot filtration and cooling, yellow crystals were obtained. Recrystallization from ethanol and next from
hexane yielded pure (99.8%) yellow crystals melting at 124.3–
125.7 "C. The yield of the oxime was 59%.
NMR (1H, 13C) and MS(ESI) spectra proved the structure of the
synthesized oxime:
1
H NMR (CDCl3) d in ppm: 8.6 (d); 7.93 (t); 7.32 (d); 8.8 (s); 2.7
(t); 1.54 (q); 1.31–1.24 (m); 0.86 (t); 9.9 (s). 13C NMR (CDCl3) d in
ppm: 147.2; 132.2; 133.7; 149.3; 122.9; 157.0; 32.8; 31.6; 29.8;
29.6; 29.3; 29.1; 26.2; 25.6; 22.5; 14.0. ESI-MS/MS m/z (% rel.
intensity): 263.4 (100); 245.4 (5); 120.2 (22); 105.2 (43); 93.2
(34); 79.2 (39).
Toluene (Sigma–Aldrich, Germany), ShellSol D70 (aliphatic
diluent; aromatic content <0.01%, Drogas Vigo (Portugal), and
decan-1-ol (Merck, Germany) were used as components of the
organic phase. Oxime of 1-(3-pyridyl)undecan-1-one (3PC10) and
tributyl phosphate (TBP) (Rhodia, Netherlands) were used as
205
Fig. 1. PEHFSD experimental setup.
206
K. Wieszczycka et al. / Separation and Purification Technology 154 (2015) 204–210
Hþaq þ HLorg ¼ H2 Lþorg
Table 1
Characteristics of hollow fiber membrane module.
ð3Þ
$
ð4Þ
2$
ð5Þ
ZnCl3; aq þ H2 Lþorg ¼ ZnCl3 ðH2 LÞorg
Type of module
G501 (contactor)
Module length (cm)
Module diameter (cm)
Case inner diameter (cm)
Center tube diameter (cm)
Number of fibers
Fiber
Effective fiber length (cm)
Inner diameter of the fibers (lm)
Outer diameter of the fibers (lm)
Pore size (lm)
Porosity (%)
Tortuosity
Inner interfacial area (m2)
Area per unit volume (cm2/cm3)
28
7.7
5.55
2.22
"10,800
X50 – polypropylene
15.6
214
300
0.03
40
2.6
1.13
28
ZnCl4; aq þ 2H2 Lþorg ¼ ZnCl4 ðH2 LÞ2;org
or with the double protonated oxime:
2$
ZnCl4; aq þ H3 L2þ
org ¼ ZnCl4 ðH3 LÞorg
ð6Þ
The strong bonding of the hydrogen atom to the nitrogen atom
of pyridine ring is capable to form strong complexes which can be
effectively decomposed after washing with sodium sulfate solution. However, very short contact time of the loaded organic phase
with the stripping solution guarantees stability of the formed
hydrochlorides:
ZnClm ðHn LÞp;org þ Na2 SO4; aq ¼ pðHn Lðn$1Þþ Þorg þ ZnSO4; aq
selected based upon our previous studies to establish the adequate
hydrodynamic conditions (data not shown). This value guarantees
optimum fluxes of metal entering in the membrane and stripping
phases. The compositions of the phases investigated are given in
Table 2.
3. Results and discussion
$
n$i
$
The extraction reaction of zinc(II) in high concentration of chloride anions with TBP is expressed as [20]:
2$
2Hþaq þ ZnCl4; aq þ 2TBPorg ¼ H2 ZnCl4 ( 2TBPorg
ð8Þ
$
Hþaq þ ZnCl3; aq þ 2TBPorg ¼ HZnCl3 ( 2TBPorg
ð9Þ
ZnCl2; aq þ 2TBPorg ¼ ZnCl2 ( 2TBPorg
ð10Þ
and finally, the extracted chlorometallate can be stripped into water
or an aqueous solution containing low chloride concentration:
$
Hn ZnClm ( 2TBPorg þ H2 Oaq ¼ ZnCl2; aq þ nHþaq þ ðm $ 2ÞClaq þ 2TBPorg
ð11Þ
ð1Þ
where i = 1, 2, 3 or 4.
Knowing the chlorocomplex formation constants (b), it is possible to calculate the content of each species present in the aqueous
phase. The values of b are very sensitive to ionic strength and the
constants are mainly determined for diluted solutions of relatively
low ionic strengths. Rough estimation of distribution of various Zn
(II) chlorocomplexes with the Medusa program [18] shows that
over 80% of zinc is in the form of ZnCl2$
4 , and only a few percent
exist as ZnCl$
3 . Thus, mechanism of extraction depends on the distribution of Zn(II) chlorocomplexes in the feed solution.
The presence of oxime and pyridine moiety in the pyridineketoxime extractant allows metals formation of complexes according
to different mechanisms: the pyridine nitrogen has an ability to
coordinate metal by solvating mechanism, while oxime substituent as anion has the ability to create chelate complexes or,
as neutral moiety, to stabilize complexes by intermolecular hydrogen bond. However, the formation of metal ion chelate complexes
has been observed only for the compound having oxime group at
two position of the pyridine ring [12]. The reaction of zinc(II)
extraction with 1-(3-pyridyl)undecan-1-one oxime (HL) runs
according to the solvating mechanism, wherein the oxime molecules coordinate neutral species of zinc(II) chloride according to
the equation [17,19]:
Zn2þ
aq þ 2Claq þ 2HLorg ¼ ZnCl2 ðHLÞ2;org
ð7Þ
As acidity of the aqueous phase decreases, other reactions can
occur [21]:
Experiments were conducted to check the operation of PEHFSD
in different conditions on changing type of extractant, initial concentration of zinc(II) and the contents of NaCl and HCl in the feed
solution. These factors would affect the transport of metal ion. The
stability of pseudo emulsion was found to be quite good during the
experimental run.
In the systems of high chloride concentrations, metal ions (Me)
are mainly in the form of chlorocomplexes:
Menþ þ i Cl ¼ MeCli
$
þ mClaq=org þ 2Naþaq
In these reactions org and aq denote species existing in the organic
and aqueous phases, respectively. The organic complexes, i.e.
ZnCl2(HL)2, ZnCl3(H2L), ZnCl4(H2L)2 and ZnCl4(H3L), or H2ZnCl4(2TBP diffuse through the pores of the membrane toward the
membrane–pseudo-emulsion interface, where Zn(II) is stripped in
the pseudo-emulsion phase after coming into contact with the
stripping phase – 5 wt.% sodium sulfate and water for 3PC10 and
TBP, respectively.
3.1. Calculations
The model for the transport of zinc(II) ions in PEHFSD working
in the recycling mode consists of the set of the following equations
[22]:
) module mass balance for the feed solution:
! "
#
$
d½Zn+M
d½Zn+M
A
aq
aq
M
( K P ½Zn+M
¼ $uaq
$
aq $ ½Zn+str
dt
dz
V M in
ð12Þ
) tank mass balance for the feed solution:
$
d½Zn+Taq Q aq #
M
½Zn+M
¼
aq;z¼L $ ½Zn+aq;z¼0
dt
V aq
ð13Þ
) module mass balance for the stripping solution:
ð2Þ
or according to the ion-pair mechanism, wherein the oxime in a single and double protonated form (3-[1-(hydroxyimino)undecyl]pyri
dinium hydrochloride and 3-[1-(hydroxyimino)-undecyl]pyridi
nium dihydrochloride) can make complexes mainly with ZnCl$
3
and/or ZnCl2$
4 [17]:
! "
#
$
d½Zn+M
d½Zn+M
A
M
str
str
( K P ½Zn+M
¼ $ustr
$
aq $ ½Zn+str
dt
dz
V M out
ð14Þ
) tank mass balance for the stripping solution:
$
d½Zn+Tstr Q str #
M
½Zn+M
¼
str;z¼0 $ ½Zn+str;z¼L
dt
V str
ð15Þ
207
K. Wieszczycka et al. / Separation and Purification Technology 154 (2015) 204–210
In these equations L is the fiber length, Q is the flow rate of the
phases, u is the linear velocity and V is the volume of each phase.
The superscripts M and T refer to the membrane module and tank,
respectively. The subscripts aq and str refer to the aqueous feed
phase and stripping phase, respectively. Because of the assumption
that the stripping reaction is instantaneous and assumed to occur at
T
the fiber outside, thus ½Zn+M
str ¼ 0 and ½Zn+str ¼ 0, the solution of
above equations is as follows:
V aq ln
%
!
"&
½Zn+Taq;t¼0
2K P L
¼ Q aq 1 $ exp
t ¼S(t
uaq r i
½Zn+aq
ð16Þ
where KP is the overall mass transfer coefficient, ri the inner radii of
the hollow fiber, S is the slope of this linear relationship which is
illustrated in Fig. 2. Values of the overall mass transfer coefficient
for all experiments are presented in Table 2.
Table 2
Composition of phases and overall mass transfer coefficient (KP).
No.
1
2
3
4
5
6
7
8
9
10
11
12
Organic phase
Aqueous phase
Stripping
phase
KP ! 107
(m/s)
HCl
(M)
NaCl
(M)
Zn(II)
(g/L)
0.1 M 3PC10 in
toluene
1
1
1
1
1
1
1
0
1
1
1
1
2
2
3.85
2
0.3
0.3
1
5
0.3
5
5
5
n/a
5 wt.%
Na2SO4
n/a
11
9.4
2.8
14
3.2
3.3
0.8
2.9 M TBP in
ShellSol D70
1
1
1
0
1
1
3.85
2
1
5
5
5
H2O
5.7
5.6
5.3
2.8
3.2. PEHSFD vs. single module extraction
In the first stage of the study on the application of membrane
systems for recovery of zinc, the performance of the single membrane extraction process realized with the hollow fiber module
(HF extraction) was compared with that of the proposed in this
paper – pseudo-emulsion based hollow fiber strip dispersion
(PEHFSD). The results of comparison (for the same compositions
of the aqueous and organic phases – solutions no. 1 and 2 in
Table 2) are shown in Fig. 3.
The removal of Zn(II) is 98% and 53% for PEHFSD and the single
extraction, respectively. It was shown that the proposed working
system with employment of a pseudo-emulsion was more effective
in transferring of Zn(II) ions to the stripping phase in comparison
to the membrane extraction using the hollow fiber membrane
module. The concentration of zinc(II) ions decreased faster in the
aqueous phase in PEHFSD process than in HF extraction. This effect
is related to the fact that for PEHFSD in the membrane module the
extraction and the stripping occur simultaneously, which makes
the driving force for extraction much stronger than when HF module is used only for extraction. In the HF single extraction process,
zinc(II) is not stripped from the organic phase, thus, the equilibrium is rapidly established and the operation is less efficient than
that when the simultaneous extraction-stripping take place. As
shown in Fig. 3, the stripping process is also effective in PEHFSD.
The results obtained illustrate that the recovery was very high,
above 95%.
Fig. 2. Determination of the overall mass transfer coefficient KP of Zn(II), results for
solution no. 2 (see experimental conditions in Table 2).
Fig. 3. Zn(II) concentration profiles during membrane experiments in: the feed (j)
in HF single extraction (solution no. 1) and the feed (d) and the stripping (s)
solution in PEHFSD (solution no. 2) (see Table 2).
3.3. Effect of the initial concentration of Zn(II) ions in the aqueous
phase
The influence of the initial metal concentration on the permeation of zinc was investigated and the results are presented in
Fig. 4.
The fractional removal of the metal ions yielded values in the
range of 92–99%, slightly higher for lower initial concentration of
zinc ([Zn]aq,0) for both extractants considered. Moreover, for lower
concentration of Zn(II) (61 g/L (60.015 mol/L)) the 1-(3-pyridyl)
undecan-1-one oxime extractant gave better results than 2.9 M
TPB, which was related to a considerable excess of the oxime over
zinc(II). For the solution containing 5 g/L (0.077 mol/L) of Zn(II), the
kinetics exhibited by the oxime was much less favorable. Nevertheless, it is worth mentioning that the recovery of zinc attained
values above 90% for the oxime, whereas it was found to be lower
(i.e. 60–90%) when TBP was used as a carrier.
For all experimental data considered, at every initial Zn(II) concentration in the feed phase the induction period was observed,
however, it should be clearly indicated that the overall mass transfer coefficient decreases as the initial zinc(II) concentration in the
feed phase increases, when the oxime is the extractant (Table 2).
This is attributable to an increase in the overall mass transport
resistance at high metal initial concentrations. Thus, the growing
resistance is related to the expected decrease in the distribution
coefficient as the initial metal concentration in the feed phase
increases. In fact, the distribution coefficient with 0.1 M oxime
208
K. Wieszczycka et al. / Separation and Purification Technology 154 (2015) 204–210
firmed not to be significantly influenced by the increase in the
Zn(II) concentration under the conditions tested.
3.4. Effect of salt concentration in the aqueous phase
To analyze the influence of the NaCl concentration in the feed
phase, a set of experiments was carried out at various NaCl concentrations, keeping other variables constant for both extractants considered. The results are shown in Fig. 5.
It can be seen that the transport of Zn(II) from acidic solutions
(1 M HCl) was not significantly affected by the NaCl concentration
in the feed for both extractants used in the organic phase. But, the
increase in the concentration of NaCl from 1 to 2 M slightly
enhanced the kinetics of extraction with 1-(3-pyridyl)undecan-1one oxime. It is noteworthy that the removal of Zn(II) was in the
range of 96–99%, the higher values being achieved for higher concentrations of NaCl. On the other hand, the NaCl content in NaCl/
HCl solution had a negligible effect on the kinetics of extraction
but the higher concentration of the salt was deleterious for the
overall efficiency of the process. This trend is in opposition to the
equilibrium data for both extractants. A decrease in zinc(II) extraction, for solutions containing more than 1 M NaCl, with oxime
extractant has been observed previously [19] and this effect can
be attributed to the decrease in neutral species (ZnCl2) concentra-
Fig. 4. Effect of the initial concentration of Zn(II) ions in the feed phase on Zn(II)
extraction in PEHFSD with (a) 0.1 M 3PC10 (solutions no. 2–4), (b) 2.9 M TBP.
was confirmed to be highly sensitive to the variation in Zn(II) concentration. A similar trend of permeation has been observed by
other authors for the transport of zinc(II), cobalt(II), vanadium(V)
and uranium(IV) ions in the pseudo-emulsion based hollow fiber
strip dispersion [5,8,23,24]. This phenomenon could be explained
by the fact that the organic phase within the membrane micropores gets saturated with zinc complex with increasing initial Zn
(II) concentration in the aqueous phase. Moreover, this organic
complex diffuses slowly into the bulk of the organic solution,
which in nutshells causes a reduction in the mass transfer to the
organic phase.
Alguacil et al. [9], in their studies on the permeation of chromium(III) in pseudo-emulsion based hollow fiber strip dispersion
(PEHFSD) process from alkaline solutions, have also reported an
increase in the transport of the target metal ions with decreasing
initial concentration of the metal in the feed solution. Thus, the
contribution of the resistance due to the diffusion of metal species
ought to increase with its decreasing concentration. On the other
hand, when the concentration of Zn(II) increases too much and
the concentration of the extractant is as low as the value used
for the oxime in this study, the permeation process is expected
to be controlled by the diffusion of metal-carrier species in the
membrane.
As far as the use of 2.9 M TBP is concerned, the degree of
removal along the permeation was found approximately the same
for 1 and 5 g/L of Zn(II) and the overall mass transfer coefficients
were very close. Actually, the distribution coefficient was con-
Fig. 5. Effect of NaCl concentration on Zn(II) extraction with (a) 0.1 M oxime
(solutions no. 2, 4–8), (b) 2.9 M TBP (solutions no. 10–12) (see Table 2); feed:
N – 5 g/L Zn(II), 1 M NaCl, 1 M HCl;
– 5 g/L Zn(II), 2 M NaCl, 1 M HCl; 4 – 5 g/L Zn
(II), 3.85 M NaCl, 1 M HCl; j – 0.3 g/L Zn(II), 1 M NaCl, 1 M HCl;
– 0.3 g/L Zn(II),
2 M NaCl, 1 M HCl; s – 5 g/L Zn(II), 2 M NaCl, no HCl.
K. Wieszczycka et al. / Separation and Purification Technology 154 (2015) 204–210
tion in the aqueous feed. For TBP, higher values of distribution ratio
of zinc(II) have been reported for higher ionic strength of the acidic
feed solution [24,25]. It is worth emphasizing that PEHFSD is controlled by the kinetics, besides the equilibrium aspects. Even
though the distribution coefficient of Zn(II) with TBP, for example,
is known to increase for higher content of NaCl, the resistance due
to diffusion in the feed phase could be augmented. Additionally, Zn
(II) extraction with TBP after 4500 s of PEHFSD decreases, i.e. Zn(II)
content in the feed starts increasing. The phenomenon is noticed
only for the feeds containing 1 M HCl (Fig. 5b). It is possible that
after that time Zn(II) is back-extracted and partly replaced with
competing HCl. This observation is supported by PEHFSD Zn(II)
concentration profile in the feed without HCl, because there is no
decrease in Zn(II) extraction. Furthermore, the analysis of zinc concentrated in the strippant confirmed that the recovery percentage
for the test without HCl was as high as "90%.
Although TBP has been reported as the most suitable zinc
extractant from HCl solution, it was shown in previous research
of our group [26] and other researchers [27] that TBP co-extracts
HCl that negatively influences Zn(II) stripping from the organic
phase. HCl is at first stripped from H2ZnCl4(2TBP resulting in its
change into ZnCl2(2TBP complex therefore this phenomenon can
be responsible for worse Zn(II) extraction.
Similarly, the extraction of zinc(II) ions with 1-(3-pyridyl)
undecan-1-one oxime from aqueous feed solution containing
sodium chloride as the major dissolved salt is not as effective as
that from NaCl/HCl solutions (Fig. 5a). This difference also arises
from the difference in the coordination ability of the extractant.
The extraction from NaCl solution proceeds by the solvating mechanism (Eq. (2)) engaging two extractant moles per one mole of Zn
(II), while but from NaCl/HCl mixture a predominant species (ZnCl$
3
and ZnCl2$
4 ) are mainly extracted (Eqs. (4)–(6)) engaging one or
two extractant molecules, which finally is more efficient.
3.5. Effect of the type of extractant
As mentioned above, the overall mass transfer coefficient of
permeation KP with PEHFSD technique was obtained from the
experimental results that were obtained for the novel carrier of
zinc(II) 1-(3-pyridyl)undecan-1-one oxime and the commercially
available extractant – TBP, for similar operating conditions, the relevant values are listed in Table 2. The values of KP for both extractants are of the same magnitude (10$7 m/s). It is worth
emphasizing that the novel extractant led to better results, when
the concentration of Zn(II) decreased to 61 g/L (KP " 10$6 m/s).
Besides, it should be noted that the concentration of TBP was significantly higher (about 30 times) in comparison to that of 1-(3pyridyl)undecan-1-one oxime, which is a very important factor
for the efficiency of PEHFSD system [28]. On the one hand, such
a high concentration of tributyl phosphate results from the fact
that this reagent is a weak extractant, and it should be used as concentrated to achieve high loading and effective metal ion transfer.
On the other hand, an increase in TBP content is not advantageous
because undiluted TBP contains over 8% of water, which facilitates
the transfer of HCl, and back transport of water to the aqueous feed
due to osmotic pressure. Moreover, undiluted TBP is a good plasticizer for many polymers, which means that it can be incorporated
in membrane modules, pipelines and pumps [29]. Thus, the concentration of TBP in the membrane might be optimized; the value
of 80% ought to be reduced. Recently, Laso et al. [27] in their studies on zinc recovery with membrane-based solvent extraction processes have pointed out that 50% TBP gives better results than 100%
TBP.
Further studies are necessary to examine the performance of
the novel extractant for different compositions of the pseudo-
209
emulsion (i.e. concentration of the oxime, strippant) and in the
presence of other elements in the feed phase.
4. Conclusions
The PEHFSD technique was found to be a feasible process for the
simultaneous removal of Zn(II) from chloride solutions and concentration of the metal in the receiving phase using the novel
extractant – 1-(3-pyridyl)undecan-1-one oxime and the wellknown extractant – tributyl phosphate as mobile carrier.
The new extractant 1-(3-pyridyl)undecan-1-one oxime was
found to be a potential carrier of zinc ions from chloride medium.
Its extraction performance was comparable to that obtained with
TBP, but the oxime gave better results for lower concentrations
of zinc (to 1 g/L). One of the advantages of application of the oxime
is that it is effective at a much smaller concentration than TBP,
with the recovery of zinc exceeding 90% for 5 g/L of zinc.
Acknowledgments
This research was supported with 03/32/DS-PB/0501 and
03/32/DS-PB/0500 grants. PEHFSD studies were realized in the
frame of Polish-Portuguese scientific and technological cooperation for the years 2013–2014 ‘‘Selective extractants for the removal
of minor metallic elements from chloride spent pickling baths”.
Financial support through the project UID/ECI/04028/2013 (FCT,
Portugal) is also acknowledged.
References
[1] M. Regel-Rosocka, A review on methods of regeneration of spent pickling
solutions from steel processing, J. Hazard. Mater. 177 (1–3) (2010) 57–69,
http://dx.doi.org/10.1016/j.jhazmat.2009.12.043.
[2] A.K. Pabby, S.S.H. Rizvi, A.M. Sastre, Handbook of Membrane Separations:
Chemical, Pharmaceutical, Food, and Biotechnological Applications, CRC Press,
2008.
[3] S. Agarwal, M.T.A. Reis, M.R.C. Ismael, M.J.N. Correia, J.M.R. Carvalho,
Application of pseudo-emulsion based hollow fibre strip dispersion (PEHFSD)
for the recovery of copper from sulphate solutions, Sep. Purif. Technol. 102
(2013) 103–110, http://dx.doi.org/10.1016/j.seppur.2012.09.026.
[4] A. Urtiaga, M.J. Abellan, J.A. Irabien, I. Ortiz, Membrane contactors for the
recovery of metallic compounds. Modelling of copper recovery from the WPO
process, J. Membr. Sci. 257 (2005) 161–170, http://dx.doi.org/10.1016/j.
memsci.2004.10.046.
[5] A. Urtiaga, E. Bringas, R. Mediavilla, I. Ortiz, The role of liquid membranes in
the selective separation and recovery of zinc for the regeneration of Cr(III)
passivation baths, J. Membr. Sci. 356 (2010) 88–95, http://dx.doi.org/10.1016/j.
memsci.2010.03.034.
[6] V. García, W. Steeghs, M. Bouten, I. Ortiz, A. Urtiaga, Implementation of an ecoinnovative separation process for a cleaner chromium passivation in the
galvanizing industry, J. Clean. Prod. 59 (2013) 274–283.
[7] J.V. Sonawane, A.K. Pabby, A.M. Sastre, Pseudo-emulsion based hollow fiber
strip dispersion: a novel methodology for gold recovery, AIChE J. 54 (2) (2008)
453–463, http://dx.doi.org/10.1002/aic.11371.
[8] F.J. Alguacil, I. Garcia-Diaz, F. Lopez, A.M. Sastre, Cobalt(II) membraneextraction by DP-8R/Exxsol D100 using pseudo-emulsion based hollow fiber
strip dispersion (PEHFSD) processing, Sep. Purif. Technol. 80 (2011) 467–472,
http://dx.doi.org/10.1016/j.seppur.2011.05.029.
[9] F.J. Alguacil, M. Alonso, F.A. Lopez, A. Lopez-Delgado, Application of pseudoemulsion based hollow fiber strip dispersion (PEHFSD) for recovery of Cr(III)
from alkaline solutions, Sep. Purif. Technol. 66 (2009) 586–590, http://dx.doi.
org/10.1016/j.seppur.2009.01.012.
[10] N. Diban, V. García, F. Alguacil, I. Ortiz, A. Urtiaga, Temperature enhancement
of zinc and iron separation from chromium(III) passivation baths by emulsion
pertraction technology, Ind. Eng. Chem. Res. 51 (2012) 9867–9874, http://dx.
doi.org/10.1021/ie301251q.
[11] E. Bringas, M.F. San Roman, I. Ortiz, Separation and recovery of anionic
pollutants by the emulsion pertraction technology. Remediation of polluted
groundwaters with Cr(VI), Ind. Eng. Chem. Res. 45 (2006) 4295–4303, http://
dx.doi.org/10.1021/ie051418e.
[12] K. Klonowska-Wieszczycka, A. Olszanowski, A. Parus, B. Zydorczak, Removal of
copper(II) from chloride solutions using hydrophobic pyridyl ketone oximes,
Solvent Extr. Ion Exch. 27 (1) (2009) 50–62, http://dx.doi.org/10.1080/
07366290802544593.
[13] A. Parus, K. Wieszczycka, A. Olszanowski, Zinc(II) ions removal from chloride
solutions by hydrophobic alkyl-pyridyl ketoximes, Sep. Sci. Technol. 48 (2)
(2012) 319–327, http://dx.doi.org/10.1080/01496395.2012.688784.
210
K. Wieszczycka et al. / Separation and Purification Technology 154 (2015) 204–210
[14] A. Parus, K. Wieszczycka, A. Olszanowski, Cadmium(II) extraction from
chloride solutions by hydrophobic pyridyl ketoximes, Hydrometallurgy 105
(2011) 284–289, http://dx.doi.org/10.1016/j.hydromet.2010.10.007.
[15] A. Parus, K. Wieszczycka, A. Olszanowski, Solvent extraction of iron(III) from
chloride solutions in the presence of copper(II) and zinc(II) using hydrophobic
pyridyl ketoximes, Sep. Sci. Technol. 46 (1) (2011) 87–93, http://dx.doi.org/
10.1080/01496395.2010.498802.
[16] K. Wieszczycka, A. Wojciechowska, M. Krupa, Equilibrium and mechanism of
cobalt(II) extraction from chloride solution by hydrophobic 2pyridineketoxime, Sep. Purif. Technol. 142 (2015) 129–136, http://dx.doi.org/
10.1016/j.seppur.2014.12.034.
[17] K. Wieszczycka, Recovery of Zn(II) from multielemental acidic chloride
solution with hydrophobic 3-pyridineketoxime, Sep. Purif. Technol. 114
(2013) 17–23, http://dx.doi.org/10.1016/j.seppur.2013.04.002.
[18] http://hydra-medusa.software.informer.com/ (22.06.15).
[19] K. Wieszczycka, M. Krupa, A. Olszanowski, Solvent extraction of zinc(II) ions
from aqueous chloride solutions by hydrophobic 3-pyridyl ketoximes, in:
Proceedings of the 19th International Solvent Extraction Conference, ISEC
2011, Chile, vol. 2(40), 2011, pp. 1–8.
[20] D.C.F. Morris, E.L. Short, Zinc chloride and zinc bromide complexes. Part II.
Solvent-extraction studies with zinc-65 as tracer, J. Chem. Soc. (1962) 2662–
2671.
[21] S.J. Cook, J.M. Perera, G.W. Stevens, S.E. Kentish, The screening of extractants
for the separation of Zn(II) from Australian hot-dip galvanizing effluent, Sep.
Sci.
Technol.
46
(2011)
2066–2074,
http://dx.doi.org/10.1080/
01496395.2011.593018.
[22] S. Agarwal, M.T.A. Reis, M.R.C. Ismael, J.M.R. Carvalho, Zinc extraction with
Ionquest 801 using pseudo-emulsion based hollow fibre strip dispersion
technique, Sep. Purif. Technol. 127 (2014) 149–156, http://dx.doi.org/10.1016/
j.seppur.2014.02.039.
[23] S.C. Roy, J.V. Sonawane, N.S. Rathore, A.K. Pabby, P. Janardan, R.D. Changrani, P.
K. Dey, S.R. Bharadwaj, Pseudo-emulsion based hollow fiber strip dispersion
technique (PEHFSD): optimization, modelling and application of PEHFSD for
recovery of U(VI) from process effluent, Sep. Sci. Technol. 43 (11–12) (2008)
3305–3332, http://dx.doi.org/10.1080/01496395.2012.723103.
[24] M. Bartkowska, M. Regel-Rosocka, J. Szymanowski, Extraction of zinc(II), iron
(III) and iron(II) with binary mixtures containing tributylphosphate and di(2ethylhexyl)phosphoric acid or Cyanex 302, Physicochem. Probl. Miner.
Process. 36 (2002) 217–224.
[25] I. Miesiac, J. Szymanowski, Separation of zinc(II) from hydrochloric acid
solutions in a double Lewis cell, Solvent Extr. Ion Exch. 22 (2) (2004) 243–265,
http://dx.doi.org/10.1081/SEI-120030464.
[26] M. Regel-Rosocka, I. Miesiac, R. Cierpiszewski, I. Mishonov, K. Alejski, A.M.
Sastre, J. Szymanowski, Recovery of zinc(II) from spent hydrochloric acid
solutions from zinc hot-dip galvanizing plants, in: Proceedings of the TMS Fall
Extraction and Processing Conference, Hydrometallurgy, vol. 2, 2003, pp.
1577–1591.
[27] J. Laso, V. García, E. Bringas, A.M. Urtiaga, I. Ortiz, Selective recovery of zinc
over iron from spent pickling wastes by different membrane-based solvent
extraction process configurations, Ind. Eng. Chem. Res. 54 (12) (2015) 3218–
3224, http://dx.doi.org/10.1021/acs.iecr.5b00099.
[28] F.J. Alguacil, M. Alonso, F.A. Lopez, A. Lopez-Delgado, I. Padilla, H. Tayibi,
Pseudo-emulsion based hollow fiber with strip dispersion pertraction of iron
(III) using (PJMTH+)2(SO2$
4 ) ionic liquid as carrier, Chem. Eng. J. 157 (2–3)
(2010) 366–372, http://dx.doi.org/10.1016/j.cej.2009.11.016.
[29] A. Mondal, S. Ghosh, A. Bhowal, S. Datta, Vanadium extraction using pseudoemulsion based hollow-fiber with strip dispersion technique, Sep. Sci. Technol.
48 (2013) 877–883, http://dx.doi.org/10.1080/01496395.2012.723103.
Anexo 2
Adsorber Units
For Water Treatment with GEH® Granular Ferric Hydroxide
ktion . Betrieb
1
GEH Adsorber Units / Basic Design Info
Œ ¦\U‡Uw=\UNhkoo|k'=Nwkˆ=w:'=NwkU\““NoŘ'=Nwk'N\\k
\kowkōo:h:|\NNw\křļ|ooowUōN\U|U=w\k=U
\T=UoŽowTokkU4=UhkNNN\kok=o
ºk\k
Œ —=w=\UN\UUw=\Uo'\k'=NN=U4ŖkT\‡N\'»°ÁU
'\k=o=U'w=\U
Œ ÄU'U\|wNwN=Uoļ:ˆ=w:hkoo|kT\U=w\kU
oThN=U4h\kw
Œ ó|=wNTwk=No'\ko\kk‡ooNľhNow=ŘĿ4Ŀ»ïìřļ
owNˆ=w:=UUko|k'\w\kow=UNooowN
Œ º=NN=U4\'ľ
Ʈó|hh\kw=U4NŽk\'j|kw“4k‡NŘhkw=No=“ŲĿŰŰŏųĿűŵTTř
Ʈ»°Áo\kUwŘhw:wˆUŰĿŸUűĿŶTř
Œ ºk\kļhhk\ĿŵŰƁ\'»°Áhw:ļ'\kKˆo:=U4
Source of solutions.
ó|hh\kw=U44k‡N
2
Filling
Œ ùkUo'kj|kw“4k‡NجÄÚ°ÚűŲŹŰŴ4křo|hh\kw=U4
NŽk=Uw\|U=w=U\kUˆ=w:o|hhN=kńo=Uowk|w=\UoĽ
N‡Nļk=UoUhN\''o|k'Ŀ
Œ ÄUHw»°Á:Žk|N=NNŽ|o=U4ˆwkōk=‡U=UHw\koŽowT
º=NN=U4
\khNTU|NNŽw:k\|4:TU:\N\k'=NN=U4h\kwĿ
Note:
Check to ensure proper functioning of the filter nozzles
before placing gravel and GEH. Take care not to crush or
otherwise damage the GEH when placing. Do not allow
GEH to get into filter nozzles.
3
Installation Backwash
Œ ¥Kˆo:'wk=UowNNw=\Uw\kT\‡'=Uo'k\Tw:
o\kkĿ
¥Kˆo:3|Uw
Ʈ¥Kˆo:ohľŲŶTŖ:
Ʈ¬|kw=\Uľhhk\ĿűŰT=U|woĿ¥Kˆo:|Uw=NNN
'=UokkT\‡ļ=ĿĿ|Uw=Nw:''N|Uw=o'k
ÄUowNNw=\U
Kˆo:
\'w|k==wŽĿ
Note:
ºko:ˆwk
> Backwash with water only.
> Do not backwash with air or air/water mixture.
Adsorber Bed Expansion vs. Backwash Speed
ŸŰ
¥hUo=\UŚƁś
Temperature 15°C
ŶŰ
ŴŰ
ŲŰ
Ű
Ű
4
ŵ
űŰ
űŵ
ŲŰ
Ųŵ
¥Kˆo:2\ˆohŚTŖ:ś
ųŰ
ųŵ
Disinfection
Œ þo:N\k=UN:\k:Žk\4Uhk\=ow:
=o=U'wUwĿ
Œ ¬=ohUo\UUwkw=o=U'wUw=Uw\kŽNˆwk owkTļNN\ˆw\k=k|Nww:k\|4:o\kkĿ
¬=o=U'w=\U
Œ —'wk=o=U'w=\UļKˆo:o\kkŘ=Uw:oT TUUko=UowNNw=\UKˆo:řĿ
Œ ¦\U'=kTo|oo'|N=o=U'w=\UŽ:K=U4T=k\=N hkTwkoļ=ĿĿ\U'\kTU\'wkwˆwkw\
hhN=Nk=UK=U4ˆwkoh='=w=\UoĿ
Note:
When carrying out disinfection, observe instructions and
data given in the technical data sheet “Disinfection“ from
GEH Wasserchemie GmbH & Co. KG.
¬=o=U'wUw
o\N|w=\U
5
Adsorber Operation
Œ þU='\kT'N\ˆw:k\|4:w:»°Áo\kkT|ow
Uo|kĿ
wk
Œ ºN\ˆohw:k\|4:ľưŲŰTŖ:
Œ °ThwŽ\Uwww=TŘ°¥¦ùřƱųT=U
ùkwTUw
Œ ٍ=T|ThkT=oo=Nhkoo|kk\hľŰĿŵkŘŷho=ř
Œ Ù\U=w\kwkwˆwk'\k\ThN=Uˆ=w:hhN=N ˆwkoh='=w=\UoĿ
ùkwˆwk
Note:
Discontinuous or intermittent operation does not impair
functioning.
Pressure Drop vs. Flow Speed Through Bed
ŰĿŵ
ìkoo|kk\h
ŚkŖT»°Áhw:ś
űĿŵ
ŰĿų
űĿŰ
ŰĿŲ
ŰĿŵ
ŰĿű
ŰĿŰ
Ű
ŵ
űŰ
űŵ
ŲŰ
Ųŵ
ųŰ
ųŵ
ìkoo|kk\h
Śho=Ŗ'w»°Áhw:ś
ŲĿŰ
Temperature 15°C
ŰĿŴ
ŰĿŰ
ºN\ˆohw:k\|4:ŚTŖ:ś
6
Operational Backwash
Œ àhkw=\UNKˆo:=U4w\kT\‡hkw=|NwTwwk kw=U=U=oUookŽˆ:Uhkoo|kk\ho
¥Kˆo:3|Uw
w:T=T|ThkT=oo=N‡N|\'ŰĿŵkŘŷho=řĿ
Œ ¥Kˆo:hk\ooŘ|h'N\ˆ\U'=4|kw=\Uř
àhkw=\UNKˆo:
Ʈ¥Kˆo:ohľŲŶTŖ:
Ʈ¬|kw=\Uľhhk\ĿűŰT=U|woļ=ĿĿ|Uw=N''N|Uw=o'k
\'w|k==wŽĿ
ºko:ˆwk
7
Replacement of GEH
Œ Ċ:Uwkwˆwkj|N=wŽk\hoN\ˆoh='=w=\Uļ
w:»°ÁT|owkhNĿïT\‡N=oU\kTNNŽkk=
\|wŽ‡||TwkUo'k\k'N|o:=U4\|ww:k\|4:w:N\ˆk
kT\‡N:UUNĿ
Œ ¬=oh\oN\k|o\'w::|ow»°ÁT|ow=U\Tō
hN=Uˆ=w:hhN=Nˆowk4|Nw=\UoĿ
ïT\‡N
8
Requirements for Raw Water Processed
Œ ºk\'w|k==wŽ
Œ ì\o=w=‡k\h\wUw=N
Œ Ú\N=|Thk=h=ww=\U
Œ Ù=U=T|T\UUwkw=\Uo\'\Thw=w=‡ˆwk
\Th\UUwo
Note:
Raw water analysis data for the specific intended application
should be provided to permit suitability assessment.
9
Important Information
Œ —NNˆ\kKok=\‡=ow\\UŽj|N='=
w:U=Nhko\UUN\UNŽU=U\kUˆ=w:NN
hhN=No'wŽk4|Nw=\UoĿ
Œ °‡kŽhhN=w=\U=UˆwkwkwTUw=o|U=j|Ŀù:hhN=ō
w=\UT|owow|==Uw=N=UN|=U4NNhk=h:kN 'w\ko'\kw:\hkw=U4\U=w=\Uo\'w:»°ÁoŽowT
UwkT=UĿ—\k=U4NŽw:k\TTUw=\Uo 4=‡U\‡k4UkN=UUw|kUU\wN4NNŽ=U=U4Ŀ
Œ Ċˆ=NN4NNŽhk\‡=hhN=w=\Uo‡=k4k=U4
=TUo=\U=U4U\hkw=\U\'Ž\|koh='=»°Á
o\khw=\U|U=wĿ
Œ ìNo\ok‡NN=Uowk|w=\UoU=U'\kTw=\U4=‡U=U\|k
hk\|wwo:woUo'wŽwo:woĿ
î|N=wŽTU4TUwoŽowTkw=,
=U\kUˆ=w:ÄóàŹŰŰűľŲŰŰŸ
GEH Wasserchemie GmbH & Co. KG
Adolf-Köhne-Straße 4
¬ōŴŹŰŹŰàoUkKŔ»kTUŽ
ùN ƦŴŹŘŰřŵŴűűŲŲŰŰŹ
ºƦŴŹŘŰřŵŴűűŸűűŹŹŰ
=U'\Ū4:ōˆook:T=Ŀ
www.geh-wasserchemie.de
Äoo|ľÏ|UŲŰűų
ìNok'koˆNNw\w:»UkNùkToU¦\U=w=\Uo\'»°ÁĊook:T=»TÁũ¦\ĿÑ»ˆ:=:hhNŽw\NN|o=UoowkUow=\UoĿ
Anexo 3
!"#$!%&'
(()))*
+*'((,,
"
-)
+.)/,,
&,
''.+
,,)
0&&'
.
,
1'23*3415*-*1&,46,3*6*78&,43*
9&,:*;'43*1,6*,'43*,5*1,4%< *=*:>
%3*1*60
,
!
"#$%
&"
'()*+,-./
01.23/
//10
/
3.
/'4
31
/
53/
5
!46+78,9):**; 7
+*+*<*=*+89,>96)*+6+++:+*)
?
,
//37==?
=+*+*<*=*+89,>96)*+6+++:+*)
3//'
3/
7)8
@')*+6
$1
7)8@')*+6
51
/4/
//
A
/
'
.7>,
B
.//
B
.C/
!&
/
010/
//37==.../0
=/
=A 0/
DAE/)*
-).F17%
1
/
/
GH
-
>+)*+,/7*>7)<
SEPARATION SCIENCE AND TECHNOLOGY
2016, VOL. 51, NO. 4, 692–700
http://dx.doi.org/10.1080/01496395.2015.1117102
Dewatering of brewer’s spent grain using an integrated membrane filter press
with vacuum drying capabilities
Remígio M. Machado
M. Rosinda C. Ismael
a
a
, Ricardo A. D. Rodriguesa, Carlos M. C. Henriquesa, M. Lurdes F. Gameiro
, M. Teresa A. Reis a, João P. B. Freireb, and Jorge M. R. Carvalho a
a
,
CERENA – Centro de Recursos Naturais e Ambiente, Instituto Superior Técnico, Universidade de Lisboa, Lisboa, Portugal; bInstituto Superior
de Agronomia, Universidade de Lisboa, Tapada da Ajuda, Lisboa, Portugal
Downloaded by [b-on: Biblioteca do conhecimento online UL] at 03:28 31 March 2016
a
ABSTRACT
ARTICLE HISTORY
Brewer’s spent grain (BSG) is a by-product of the brewing process, rich in fiber, protein and
carbohydrates. Its potential application is limited because of high moisture content (80%). This
work presents a process for dewatering BSG using two different sets of membrane filter plates in a
filter press with vacuum drying: recessive plates with polypropylene membranes and the innovative Rollfit® plates. A final moisture content of 12-15% was obtained in 15 mm-thick filter cakes,
using both types of plates. The dewatering cycle included filtration, membrane-squeezing, and
vacuum thermal drying using hot water (~90ºC) as heat source.
Received 6 February 2015
Accepted 3 November 2015
Introduction
Brewer’s spent grain (BSG) is the solid fraction that
results from the separation of the mash in wort and
grain residues in a process known in the brewery
industry as lautering. This by-product is rich in proteins, fibers, and lipids, as well as in polyphenols, vitamins, and antioxidants, and has the potential for
multiple applications of high added value. According
to Kanauchi et al. (1), BSG is composed of protein
(24%), pentosans (22%), lignin (12%), cellulose
(24.5%), lipids (11%), and ash (2.4%), based on dry
weight. The extraction of valuable compounds via
BSG enzymatic hydrolysis has been extensively studied.
Faulds et al. (2) reported the production of arabinoxylan and mono- and dimeric ferulic acid from BSG
using glycosyl hydrolases and feruloyl esterases.
Mussatto et al. (3-6) described the dilute acid hydrolysis
for the hemicellulose recovery, the alkaline hydrolysis
for the lignin solubilization, and the enzymatic hydrolysis for the cellulose conversion into glucose. The
cellulosic and hemicellulosic hydrolysates were used as
the fermentation medium for the production of lactic
acid and xylitol. Lactic acid is used as an acidifier and
food preservative and it has also applications in the
pharmaceutical, leather, and textile industries. Xylitol
is a non-fermentable sugar used as a sweetener, in oral
hygiene and dental care products. The alkaline hydrolysis of BSG also produces liquors containing phenolic
KEYWORDS
Spent grain; filtration; filter
press; vacuum-drying;
membrane plate
acids, mainly ferulic and p-coumaric acids. Xiros and
Christakopoulos (7) used alkali pretreated BSG for the
production of ethanol by the mesophilic fungus
Fusarium oxysporum.
BSG is also a source of soluble and insoluble fiber
and can be incorporated in composite food for human
consumption, such as cookies (8-10). The consumption
of BSG has health benefits, which are associated with
increasing the fecal weight and shortening the residence
time in the digestive tract. The soluble fiber present in
BSG has also a positive effect in reducing cholesterol
and triglycerides (11). Although BSG has high added
value applications, its main use is still for cattle feeding,
since BSG is considered a cheap source of proteins.
Recent studies have shown that BSG may also be successfully incorporated in the diet of monogastric animals such as piglets (12) and also in cecal fermenters
(rabbits) (13). However, in spite of all the potential
applications, the actual use of BSG is severely limited
due to its high moisture content (around 80% w/w)
which brings about microbial development. BSG must
be used within 48 hours after production otherwise
complex fermentation processes, including butyric
anaerobic fermentation, will transform BSG into a useless and toxic waste that must be disposed in landfills.
The most promising technique to stabilize BSG is
dewatering. A moisture content below 20% is obtained
using this technique. So far, several technologies have
been tested including hot air fluidization or the use of
CONTACT Jorge M. R. Carvalho
jcarv@tecnico.ulisboa.pt
CERENA – Centro de Recursos Naturais e Ambiente, Instituto Superior Técnico,
Universidade de Lisboa, Av. Rovisco Pais, 1, 1049-001 Lisboa, Portugal.
© 2016 Taylor & Francis
Downloaded by [b-on: Biblioteca do conhecimento online UL] at 03:28 31 March 2016
SEPARATION SCIENCE AND TECHNOLOGY
rotary dryers in a countercurrent system. These processes are capital intensive and also require an expensive source of energy, even when using energy from a
co-generation unit.
More recently, a process has been developed (14) for
the separation of the BSG protein and fibrous fractions,
by diluting the mixture to a moisture content of 95%,
followed by a passage through a vibrating sieve, which
retains the fibrous fraction. The protein fraction is
dewatered in a centrifugal settler to a maximum solid
content of 30%. An extra step of drying, using a spray
thermal dryer, a paste dryer, or an extrusion dryer,
increases the solid content to about 80%. This fraction
can then be used as a valuable component for nonruminant animals feed production. The fraction rich in
fiber is processed in a screw-press, increasing its dry
matter content up to 40%, and then burned to produce
vapor. This process allows a better use of all the components of BSG, but it is a complex process that also
requires a high input of energy.
The innovative process described in this paper is
based on dewatering BSG using an integrated filter
press with membranes and with thermal drying capability. Low-cost hot water, available locally at the
breweries, is used as the main source of energy. To
the best of our knowledge, the only application of this
technology to brewer’s spent grain is the one published
by El-Shafey et al. (15). However, this implementation
required a dilution of the BSG to more than 97% of
moisture content. Otherwise, the air operated diaphragm pump would be blocked by the sludge of
BSG. On an industrial scale (400 ton/week of BSG
with 73-75% of moisture content), the water quantity
requirements would prevent any economical and sustainable application of the process and, furthermore,
the high dilution of the produced filtrate would invalidate its use as a source of polyphenols for several
industrial processes.
The aim of this paper is to produce dry, shelf-storable
BSG cakes using a membrane filter press. The obtained
dry cakes can be used not only as animal food (that can be
preserved for a long time without chemical or physical
degradation) but also as a raw material for the biotechnology and pharmaceutical industries.
Equipment and experimental methods
The membrane filter press used in this work was a
modified US Filter J-VAP, model 470V30-7-1MYLW,
serial number JV0044 (15). The schematic diagram is
shown in Fig. 1. The starting unit was able to withstand
8 bar working pressure, and the thermal fluid for
squeezing and heating the filter cake was limited to a
maximum working temperature of 90ºC. Because the
BSG sludge was fed to the filtration unit using an air-
Air blowdown
Air supply
Moisture
59% water removal using
filtration and squeezing
BSG:
Hot water
(85-95 °C)
25% (w/w)
solids
>35% water removal using
thermal vacuum drying
Condenser
PP recessive
membrane plates
Vacuum
Dehydrated BSG:
85-88% (w/w) solids
Figure 1. Flowchart of the Brewer’s Spent Grain dewatering.
693
Rollfit® plates
Downloaded by [b-on: Biblioteca do conhecimento online UL] at 03:28 31 March 2016
694
R. M. MACHADO ET AL.
operated diaphragm pump, the maximum solid concentration was 3% of dry matter, and consequently the
BSG had to be diluted. The aim of this work was to
dewater the BSG without dilution, as it was obtained
from the brewing process. Thus, the air-operated diaphragm pump was replaced by a progressive cavity
pump (Seepex – BTI 2-24), capable of pumping the
BSG to the filter unit at a maximum pressure of 20
bar (16). The filtration unit was modified to withstand
the maximum working pressure of 20 bar during the
filtration stage and 60 bar in the squeezing stage. The
metal frame of the filtration unit was strengthened, and
a new hydraulic jack, with a capacity of 50 ton and a
maximum working pressure of 700 bar (Larzep,
SN05010 model), was installed for closing the filter.
The carbon steel and polypropylene pipes used in the
BSG feed circuit and in the hot water closed-loop
circuit for the filter cake squeezing and vacuum drying
stages were replaced by stainless steel pipes (AISI 316).
In this study, two different membrane plates were
tested:
a. Recessive membrane plates were supplied by U.S.
Filter Systems and described by El-Shafey et al.
(15). The membrane plate material was polypropylene and the size of each plate was 470 ×
470 mm. In this study, four filter plates were
arranged to get three chambers with a total filtration area of 0.816 m2, producing three cakes with
thicknesses between 9 and 46 mm. For identification purposes, this set of plates is hereinafter
referred as the standard set.
b. Rollfit® plates were supplied by Reisser Eilers &
Partner AG. The plate innovative technology is
based on the fact that one side of the plate is
made of stainless steel (heating plate) and the
opposite side is a polypropylene recessive membrane. The interior of the combined Rollfit® plates
has two independent and isolated circuits for the
circulation of thermal fluids. Thus, it was possible
to heat the metallic side of the plate with a thermal fluid and to cool the polypropylene recessive
membrane with tap water, simultaneously. The
size of each plate was also 470 × 470 mm and
the filtration chamber enabled the formation of
cakes with thicknesses of 10 to 40 mm. The filter
plates were arranged to get three chambers with a
filtration area of 0.408 m2.
The feed pressure was controlled using a proportional,
integral and derivative control chain (PID). This chain
was composed of a pressure transducer, two independent proportional-integral-derivative (PID) controllers,
a timer, and a variable-frequency drive. The PID controllers and the timer were implemented in a supervisory
control and data acquisition software (CitectScada version 6.1 Schneider Electric). Two PID controllers were
used to ensure that the BSG feed pressure was within a
range of 5% of the set point. During a filtration operation
there are two distinct sequential stages, named cake
formation and cake consolidation. The transition
between the two stages can be very abrupt, and each of
these stages requires specific PID parameterization. The
switch between the two PID controllers was made using
a software implemented timer. Temperature and pressure were monitored and registered by the supervisory
control and data acquisition software in the BSG feed
line, in the hot water line for cake squeezing and heating,
as well as in the vacuum process line.
In the vacuum line, a vertical condenser made with
two assembled glass coils heat exchangers (De Dietrich
Process System QVF™) was installed before the vacuum
pump. Each glass heat exchanger had a nominal diameter of 80 mm, a nominal length of 610 mm, and a
heat transfer area of 0.3 m2. The condenser was cooled
with water at 5ºC provided by a refrigerated water bath
(Grant Scientific GP200 R5). The condensates were
collected in a 2-litre tank assembled on the bottom of
the condenser. A new oil ring vacuum pump (Busch
RA0063F 5A3) was also assembled in a parallel configuration in relation to the existing water ring vacuum
pump (Squire Cogswell Company, number C00-2240/1,
type PM124-M30A).
The filter cloths used in this work were made of
polypropylene fabric, mono-mono filament, with an
air permeability of 500 L/dm2.min (Ambifiltra,
Comércio de Tecidos Filtrantes, Lda.). For vacuum
sealing purposes, a 5 cm wide strip of a polymerizing
polyurethane resin was applied at the outer edge of the
filter cloth.
Filtration and dewatering operation
The filter operation steps were controlled using
manual valves and switches on a control panel.
Prior to any filtering operation, a step for deflating
the plate membranes was mandatory, as described
by El-Shafey et al. (15).
The filter was fed by the progressive cavity pump at
constant pressure (3-6 bar) in each experiment. The
filtration progress was monitored by the filtrate
volume, which was recorded to the nearest cubic centimeter using a calibrated 2 L filtrate vessel. Time was
also recorded to the nearest second. The errors associated with time and with the volume of filtrate did not
exceed 2 s and 50 mL, respectively. The filtration
Downloaded by [b-on: Biblioteca do conhecimento online UL] at 03:28 31 March 2016
SEPARATION SCIENCE AND TECHNOLOGY
process continued until the collected fluid flow rate was
less than 20 mL/min.
The second step of the mechanical dewatering was
cake compression or cake squeezing. The membranes
were inflated with water pressurized up to 4 to 8 bar by
the centrifugal pump (MTH pump, T51G BF model).
The filtrate resulting from the squeezing stage was also
collected and added to the total filtrate. The compression stage was considered completed when the filtrate
flow was less than 20 mL/min. The duration of this
stage was normally 5 min.
Core blow was carried out by applying air, at low
pressure (1-2 bar), through the channel formed by the
plates upper left eye, after closing the filtrate valves and
opening the central core to the atmosphere. This operation was essential to empty the central core of wet BSG
that would wet the filter cake during the thermal stage.
The subsequent step was the thermal stage of the
dewatering cycle. This step used a centrifugal pump
(MTH pump, T51G BF model) and a heater (Ogden,
KS-0591-M7 model) to fill the membranes with hot
water (85-95ºC), from a water supply tank, at a squeezing
pressure of 2-3 bar. At the same time, vacuum drying was
applied over the hot cakes at 50-110 mbar (absolute pressure) via the vacuum pump (R 5 RA 0063-0100 F) through
the filtrate outlet ports. Low pressure water saturated air
departed from the hot cakes through the eyelets on the top
and bottom of each membrane plate and passed through
the condenser installed in the vacuum line.
In the case of the Rollfit® plates, a different procedure
was also tested. Cold water circulated through the membranes while the metal plates heated the cake, creating a
temperature gradient within the filter cake. In order to
achieve different levels of cake dryness, several vacuum
drying times were tested between 2 and 5 hours.
After the dewatering cycle, the filter cakes were
removed from the filter press and the final moisture
content was measured by drying five representative
samples of each cake (rectangles with 2 × 2 cm2 taken
from five equidistant points on a main diagonal of the
filter cakes) in an oven (at 105ºC) until constant weight.
The dried samples were cooled in a desiccator and then
weighed.
Filter cloth cleaning involved several steps. Firstly, pressurized water was passed over the cloth surface to remove
big particles. Then, the cloth was impregnated overnight
in a 10% hypochlorite alkaline solution, to remove the fine
unfiltered particles, and rinsed successively with water to
remove any residuals of the hypochlorite solution.
The stages required for the BSG dewatering complete cycle are listed in Table 1, where nominal times
are presented.
695
Table 1. BSG dewatering steps.
Nominal time required (min)
Step
Filter press closing
Membrane draining
Press filling/Filtration
Cold water squeezing/
Compressed air squeezing
5. Core blow
6. Transition for vacuum drying
7. Hot squeezing with vacuum
drying
8. Vacuum stop
9. Press opening
10. Retract plates and discharge
cakes
Total Cycle Time
Recessive Membrane/
Standard Plates
1.
2.
3.
4.
Rollfit®
Plates
2
2
12
5
2
2
10
5
2
1
180 to 300
2
3
180 to 300
1
2
3
1
2
3
210 to 330
210 to 330
BSG preparation and conditioning
The BSG used in this work was supplied by the production center of the UNICER brewery, located in
Santarém, Portugal. In order to avoid microbial degradation, the BSG was kept in a cold room at 4ºC. The
BSE was analyzed at Instituto Superior de Agronomia –
Laboratório Químico Agrícola Rebelo da Silva, Lisboa,
Portugal. The results are presented in Table 2, for two
BSG samples of two different beer brands obtained in
the industrial process.
The BSG is produced by the industrial brewery at
a temperature of 45 to 55ºC. Thus, to reproduce the
industrial conditions as close as possible, before each
dewatering experiment the BSG was heated to
40-45ºC, for 3 to 4 hours, using a vessel heated by
copper coils with internal circulation of hot water at
70ºC. The starting reference moisture for the BSG
used in this work was 76% (most frequent moisture
content). However, some samples reached the filtration pilot plant with lower moisture content. In such
cases, the moisture was adjusted with small amounts
of tap water.
Table 2. Analytical composition of two samples of BSG, for two
beer brands, obtained in the production center of UNICER,
Santarém, Portugal.
Analytical Fraction
Moisture %
Ash (% DM)
Organic Matter (% DM)
Total Protein (% DM)
Total Lipids (% DM)
Total Fibre (% DM)
NDF (% DM)
ADF (% DM)
ADL (% DM)
Celullose (ADF-ADL) (% DM)
Hemicellulose (NDF-ADF) (% DM)
Brand 1
70.6
3.8
96.2
26.8
8.6
17.52
60.3
22.6
4.1
18.5
37.7
Brand 2
71.1
4.4
95.6
21.4
7.9
21.4
60.7
21.8
4.6
17.2
38.9
% DM: Percentage of Dry Matter, NDF: Neutral Detergent Fibre, ADF: Acid
Detergent Fibre, ADL: Acid Detergent Lignin
696
R. M. MACHADO ET AL.
Results and discussion
Filtration with recessive plates with polypropylene
membrane
Several tests were carried out, at different pressures, to
study the filtration kinetics. The process was performed
using a suspension with 315 kg of solids per cubic
meter of filtrate (25% w/w solids). Figure 2 represents
the filtrate volume versus time, for filtration pressures
of 3, 4, and 5 bar. The higher flow, in the early phase,
corresponds to the rearrangement of the particles forming the cake, inside the filter chamber. In the final stage
of filtration, the flow is reduced due to the increase in
the resistance of the cake, caused by its consolidation.
According to the classical theory of filtration (17),
the specific cake resistance and the filtrating medium
resistance can be determined by:
t ! tS
αμc
μR
ðV þ VS Þ þ
¼
AΔp
V ! VS 2A2 Δp
(1)
where t is the filtration time, ts is the time required to
reach constant pressure, V is the volume of cumulative
filtrate, Vs is the volume of filtrate until constant pressure,
α ¼ α0
!
Δp
Δp0
"s
(2)
where α0 is the specific cake resistance at the reference
pressure ∆p0 (1 bar) and s is the compressibility coefficient.
The specific cake resistance, presented in Table 3, was
plotted against the filtration pressure on a logarithmic
scale (Fig. 4). The slope of the straight line corresponds
8
7
6
Filtrate volume (L)
Downloaded by [b-on: Biblioteca do conhecimento online UL] at 03:28 31 March 2016
Figure 1 shows the global average results of the
Brewer’s Spent Grain dewatering, using a filter press
with membrane plates and vacuum drying capabilities.
It should be noted that 59% of the water removal was
achieved using filtration and squeezing, that is, using
mechanical energy. The thermal energy was supplied by
hot water, a secondary source of low cost energy abundant in breweries.
α is the specific cake resistance, µ is the filtrate viscosity,
∆p is the filtration pressure, A is the total area of filtration,
c is the mass of solids per volume of filtrate, and R is the
filtrating medium resistance. The straight lines obtained
from the representation of (t-ts)/(V-Vs) versus (V+Vs) are
shown in Fig. 3.
The specific cake resistances were calculated from
the slopes of the straight lines, obtained for each pressure. The results are presented in Table 3.
According to Table 3, the specific resistance
increases with the filtration pressure since as the pressure increases, the filter cake compactness also
increases. The resistance of the filtrating medium, R,
can be calculated from the y-intercept of Eq. (1).
However, the results are meaningless, since negative
values were obtained. This can be explained by an
increase in the resistance of the filtering medium
along the experiment (which is not modeled by the
classical theory of filtration), due to the penetration of
some solids on the filter cloth (15) or to the compression of the filter cake during the consolidation stage.
The dependence of the specific resistance of the cake
with the applied pressure can be modeled by the following empirical equation:
5
4
3 bar
3
4 bar
2
5 bar
1
0
0
100
200
300
400
500
600
Time (s)
Figure 2. Filtrate volume versus time using the standard set of filtration plates with filtration area of 0.816 m2 and constant filtration
pressures of 3, 4 and 5 bar. BSG concentration in the suspension feed: 315 kg/m3 filtrate.
SEPARATION SCIENCE AND TECHNOLOGY
2.0E+05
(t-ts)/(V-Vs) (s/m3)
1.6E+05
y = (9.5±1.2)E+07x – (6.8±0.9)E+05
R2 = 0.963
y = (3.9±0.3)E+07x – (3.0±0.3)E+05
R2 = 0.992
1.2E+05
3 bar
4 bar
8.0E+04
5 bar
y = (3.8±0.3)E+07x – (3.3±0.3)E+05
R2 = 0.983
0.0E+00
6.0E-03
t!ts
V!Vs
8.0E-03
1.0E-02
V+Vs (m3)
Table 3. Specific cake resistance for constant filtration pressure
of 3, 4 and 5 bar. Solids concentration: 315 kg/m3 of filtrate.
Filtration pressure (bar)
3
4
5
Specific cake resistance×10‒13 (m/kg)
5.3 ± 0.7
6.6 ± 0.4
8.1 ± 0.6
13.95
y = (0.8±0.4)x + (9±2)
R2 = 0.998
13.9
13.85
13.8
13.75
13.7
5.4
1.2E-02
1.4E-02
versus (V+Vs) for constant filtration pressures of 3, 4, and 5 bar.
log α
Downloaded by [b-on: Biblioteca do conhecimento online UL] at 03:28 31 March 2016
4.0E+04
Figure 3.
697
5.5
5.6
log ∆ p
5.7
5.8
Figure 4. Plot of logarithm α versus logarithm ∆p for filtration
pressures of 3, 4, and 5 bar. Solids concentration of 315 kg/m3
of filtrate.
to the compressibility coefficient, s, and the value of the
line intersection with the abscissa axis to the specific cake
resistance at the reference pressure of 1 bar, α0.
The specific cake resistance at 1 bar was α0 = 2.4 × 109
m/kg and the compressibility coefficient s = 0.8 (R2=0.998).
The specific cake resistance obtained in this work is
significantly lower than the one reported by El-Shafey
et al. (15) (α0 = 1.3 × 1010 m/kg). Moreover, the
compressibility coefficient reported in the previous
work was much lower (s = 0.4). This fact can be interpreted taking the concentrations of solids in the feed
suspension into account, which are much higher in this
work (315 kg solids/m3 of filtrate) than in the cited
paper (20 kg/m3 of filtrate). At higher concentration of
solids, the BSG particles do not have time to distribute
and rearrange in the growing layer of the filtrating cake
and as a consequence a porous filtration cake is formed
with a very low specific cake resistance, but with a very
high compressibility coefficient.
Thermal cake drying was tested for the standard set
of filter plates. The cakes were obtained after filtration
of a BSG at 4 bar followed by membrane squeezing at 6
bar and core blow, as described in the section titled
“Filtration and Dewatering Operation”. The thickness
of the filter cake ranged between 9 and 46 mm by
introducing polypropylene frames with 10, 20, and
30 mm thickness between the filter plates. As described
previously, the thermal fluid was hot water at 85-95ºC
and the thermal drying was assisted by vacuum at
50-110 mbar. The thermal drying period ranged
between three to five hours. Figure 5 shows the final
moisture of the filter cakes obtained after the thermal
drying process. As expected, increasing the time of
thermal drying decreases the final cake moisture.
The cake thickness is an important factor for the
efficiency of the drying process. As the cake thickness
increases, more time is required to achieve high levels
of cake dryness. Furthermore, for cakes with a thickness of more than 30 mm, the percentage of moisture
in the final filter cake ends asymptotically to a plateau
with 45% of final moisture regardless of the time of
thermal drying. These results can be interpreted considering that the surface of the filter cake is in contact
with the heated membranes and thus the temperature
reaches a maximum at the surface plane and decreases
towards the center of the cake. On the other hand, the
R. M. MACHADO ET AL.
50
Cake moisture (%)
40
3h
4h
30
5h
20
10
0
Downloaded by [b-on: Biblioteca do conhecimento online UL] at 03:28 31 March 2016
0
10
20
30
Cake thickness (mm)
40
50
Figure 5. Final moisture against cake thickness for 3, 4, and 5 h
of vacuum. Vacuum pressure: 50-110 mbar. Thermal fluid: hot
water at 85-95ºC.
absolute pressure reaches a maximum in the center
plane of the filter cake and decreases towards the surface of the cake. This particular pressure profile is due
to the fact that the cake central plane corresponds to
the junction plane between two consecutive filter plates
where vacuum leakage rate is higher. As a consequence,
the central area of the filter cake is very difficult to dry
not only because the temperature in this zone is smaller
but also because there is a higher absolute pressure
hindering the evaporation of water. It was also
observed that the cake surface was normally very dry,
forming a superficial hard crust with poor thermal
conductivity. All the three factors together, that is
unfavorable temperature and pressure profiles and
poor thermal conductivity of the outer layers of the
cakes, contribute to an inefficient dewatering of the
central zone, and this inefficiency increases when the
thickness of the cakes increases.
The experimental results presented in Fig. 5, for polypropylene membrane plates, show that to obtain filter
cakes with a final moisture content of 30%, after three
hours of thermal drying, the filter cake must not exceed
12 mm of thickness. In order to achieve 15% of moisture
content in the filter cakes with 12 mm thickness, vacuum
thermal drying must be applied for five hours.
®
Filtration with Rollfit plates
BSG dewatering cycles were also carried out using
Rollfit® plates. These plates are characterized by having
a metal surface in one side and a polypropylene recessive membrane in the opposite side. Both surfaces can
be independently heated with a thermal fluid.
The studied operational variables were the vacuum
drying total time, from 3 to 5 hours, the cake thickness,
from 10 to 40 mm, and the effect of heating one or both
cake surfaces during the thermal drying cycle. During
the drying cycle, using the Rollfit® plates, the metallic
surface was always heated with hot water (85-95ºC).
However, the surface corresponding to the polypropylene recessive membrane was, at the beginning of the
thermal cycle, cooled with tap water at room temperature (20-25ºC) and only after a pre-defined period,
from one to five hours, the hot water was allowed to
circulate into the internal circuit used to heat the polypropylene recessive membrane.
Figure 6 shows the results obtained after the thermal
drying cycle using the Rollfit® plates, where both sides
of the filter plates were heated during the total time of
the drying cycle. The results are similar to the ones
presented for the polypropylene recessive plates. The
final moisture content of the filter cakes decreases with
increasing the thermal cycle time. To achieve less than
15% moisture content, the thickness of the cake must
be less than 12 mm and vacuum drying must be used
for 5 hours.
Figure 6 gives information about the optimum operating conditions of the Rollfit® plates. To achieve the
best performance with these plates, the vacuum thermal
drying cycle must start with the circulation of cold
water in the polypropylene membrane side of the
Rollfit® plate, which must be changed to hot water
when there are two hours left of vacuum drying. In
the first hours, an evaporation front is formed inside
the cake. Like this, water is forced to go from the hot
metal plate side to the cold polypropylene membrane
side of the adjacent plate.
Then, when hot water starts circulating through the
membranes, the cake moisture is easily boiled, the drying of the cake being promoted.
To achieve 12% moisture or less, Rollfit® plates
require cycles of vacuum thermal drying of 3.5 and
60
50
Cake moisture (%)
698
40
3h
4h
30
5h
20
10
0
0
10
20
30
Cake thickness (mm)
40
50
Figure 6. Cake moisture vs. cake thickness for 3, 4, and 5 hours
vacuum drying. Both sides of the Rollfit® plate were heated with
hot water (85-95ºC) during the thermal drying cycle.
Conclusions
The carried out experiments indicate that this technology
is technically viable to dewater BSG from 75% to 15%
using recessive membrane plates or Rollfit® plates. The
final moisture content was mainly a variable of the cake
thickness and of the thermal vacuum drying time. To
dewater BSG, it was necessary to obtain cakes with less
than 15 mm thickness. Cakes with more than 20 mm
thickness had a final moisture content of 45%, regardless
of the thermal drying cycle time or the set of filter plates
used. For filter cakes with similar thickness, the best
Filter cake 1 (10 mm thickness)
50
Filter cake 2 (15 mm thickness)
40
Time of hot water circulation
through the pp membrane
30
Time of hot water circulation
through the pp membrane
40
0h hot
Moisture (%)
0h hot
1h hot
2h hot
20
always hot
10
1h hot
30
2h hot
always hot
20
10
0
0
2
3
4
Vacuum time (h)
5
6
2
3
4
Vacuum time (h)
5
6
Figure 7. Cake moisture vs. vacuum time for different times of hot water circulation through the polypropylene membrane.
80
76
75
Initial
76
66
Moisture (%)
60
62
60
Filtration
65
60
55
Squeezing
Vacuum - 3h
Vacuum - 4h
Vacuum - 5h
40
31
34
25
20
20
18
12
2 1
0
Recessive membrane
plates (12 mm cake)
699
Rollfit® set of plates, only the polypropylene membrane is
able to drain the filtrate and squeeze the filter cake, that is,
there is only one surface, for mechanical dewatering, per
chamber. However, the Rollfit® plates are more efficient
during the thermal drying stage. In fact, only the Rollfit®
plates can reduce the moisture of the filter cakes to 1%
after a thermal cycle of 5 hours. This is due to the fact that
the metallic surface of the Rollfit® plates has higher thermal conductivity than the polypropylene surface of the
recessive membrane plates.
4.5 hours for cakes with 10 and 15 mm thickness,
respectively, as depicted in Fig. 7.
A summary of the whole dewatering process representing the moisture content at the end of each stage of
the operation, for the Rollfit® plates and the polypropylene recessive plates, is shown in Fig. 8. Considering the
results obtained with the recessive membrane plates
(filter cakes with 12 mm of thickness) Fig. 8 shows
that during the mechanical stage of the dewatering
process (filtration and squeezing), the BSG moisture
content was reduced from 75% to 55%. The mechanical
energy removed 59% of the total water existing initially
in the BSG, whereas the thermal drying stage removed
36%. Additionally, the energy source used for the thermal drying process was hot water at 85-95ºC, a secondary heat source available at low cost in agro-industries.
Therefore, the developed process has low energy
consumption.
Figure 8 also shows that the Rollfit® plates were less
efficient for the mechanical dewatering stage than the
polypropylene recessive membrane plates. This is due to
the fact that the metallic surface of the Rollfit® plates
cannot filter or compress the filter cake. When using the
Moisture (%)
Downloaded by [b-on: Biblioteca do conhecimento online UL] at 03:28 31 March 2016
SEPARATION SCIENCE AND TECHNOLOGY
Rollfit plates
(10 mm cake)
Figure 8. Cake moisture at the end of every stage of the dewatering process.
1
Rollfit plates
(15 mm cake)
700
R. M. MACHADO ET AL.
Downloaded by [b-on: Biblioteca do conhecimento online UL] at 03:28 31 March 2016
results were achieved using Rollfit® plates, because of their
metallic surface with high thermal conductivity.
Additionally, the thermal drying cycle must start with
the circulation of cold water in the polypropylene membrane side, which must be changed to hot water during
the final 2 hours of the thermal cycle.
Finally, the BSG dewatering process prevents microbial degradation and reduces significantly the storage
and transportation costs. The dried BSG can be considered a commodity for cattle feed and as a source of
valuable biological molecules.
[5]
[6]
[7]
Acknowledgements
The authors would like to thank Unicer Bebidas, SA, for
promoting this project.
[8]
[9]
Funding
AdI, Agência de Inovação’s financial support is gratefully
acknowledged for the project DRECHE – “Tecnologias
ambientais para a valorização de resíduos da indústria”, 3rd
call no. 70/136, 2005, PRIME IDEIA Program.
[11]
ORCID
[12]
Remígio M. Machado http://orcid.org/0000-0002-6696-8183
M. Lurdes F. Gameiro http://orcid.org/0000-0003-2180-8116
M. Rosinda C. Ismael http://orcid.org/0000-0002-7081-0514
http://orcid.org/0000-0003-2523-9379
M. Teresa A. Reis
Jorge M.R. Carvalho http://orcid.org/0000-0001-8091-5419
[10]
[13]
References
[1] Kanauchi, O.; Mitsuyama, K.; Araki, Y. (2001)
Development of a functional germinated barley foodstuff from brewer’s spent grain for the treatment of
ulcerative colitis. J. Am. Soc. Brew. Chem., 59(2): 59–62.
DOI:10.1094/ASBCJ-59-0059
[2] Faulds, C. B.; Mandalari, G.; LoCurto, R.; Bisignano, G.;
Waldron, K. W. (2004) Arabinoxylan and mono- and
dimeric ferulic acid release from brewer’s grain and
wheat bran by feruloyl esterases and glycosyl hydrolases
from Humicola insolens. Appl. Microbiol. Biotechnol., 64
(5): 644–650. DOI:10.1007/s00253-003-1520-3
[3] Mussatto, S. I.; Fernandes, M.; Dragone, G.; Mancilha,
I. M.; Roberto, I. C. (2007) Brewer’s spent grain as raw
material for lactic acid production by Lactobacillus
delbrueckii. Biotechnol. Lett., 29(12): 1973–1976.
DOI:10.1007/s10529-007-9494-3
[4] Mussatto, S. I.; Dragone, G.; Teixeira, J. A.; Roberto, I. C.
(2008) Total reuse of brewer’s spent grain in chemical
and biotechnological processes for the production of
[14]
[15]
[16]
[17]
added-value compounds. Bioenergy: Challenges and
Opportunities, International Conference and Exhibition
on Bioenergy; Universidade do Minho: Guimarães,
Portugal, 6th-9th April.
Mussatto, S. I.; Rocha, G. J. M.; Roberto, I. C. (2008)
Hydrogen peroxide bleaching of cellulose pulps obtained
from brewer’s spent grain. Cellulose, 15(4): 641–649.
Mussatto, S. I.; Dragone, G.; Fernandes, M.; Milagres,
A. M. F.; Roberto, I. C. (2008) The effect of agitation
speed, enzyme loading and substrate concentration on
enzymatic hydrolysis of cellulose from brewer’s spent
grain. Cellulose, 15(5): 711–721.
Xiros, C.; Christakopoulos, P. (2009) Enhanced ethanol
production from brewer’s spent grain by a Fusarium
oxysporum consolidated system. Biotechnol. Biofuels,
2:4, 1–12. DOI:10.1186/1754-6834-2-4
Öztürk, S.; Özboy, Ö.; Cavidoğlu, İ.; Köksel, H. (2002)
Effects of brewer’s spent grain on the quality and dietary
fiber content of cookies. J. Inst. Brew., 108(1): 23–27.
Prentice, N.; D’Appolonia, B. L. (1977) High-fiber
bread containing Brewer’s spent grain. Cereal Chem.,
54(5): 1084–1095.
Prentice, N.; Kissell, L. T.; Lindsay, R. C.; Yamazaki, W.
T. (1978) High-fiber cookies containing Brewer’s spent
grain. Cereal Chem., 55(5): 712–721.
Fastnaught, C. E. (2001) Barley Fiber. In Handbook of
Dietary Fiber; S. S. Cho, M. L. Dreher, Eds.; Dekker:
New York, Chap. 27, pp. 519–542.
Martins, C.; Pinho, M.; Lordelo, M. M.; Cunha, L. F.;
Carvalho, J.; Freire, J. P. B. (2010) Effect of brewers
grain on intestinal microbial activity and mucosa morphology of weaned piglets. Livest. Sci., 133(1-3): 132–134.
Vieira, A. R. D. G. (2009) A fibra na alimentação do
coelho: Dreches de cervejaria relativamente à luzerna e à
polpa de beterraba (Fiber in the rabbit’s diet: brewer’s
spent grain compared with lucerne and beetroot pulp);
Master’s Thesis in Zootechnic Engineering – Animal
Production, Lisboa, Portugal. URL: http://www.reposi
tory.utl.pt/bitstream/10400.5/1888/1/Tese%20Ana%
20Rita%20Vieira.pdf
Schwencke, K. V. (2006) Sustainable, cost-effective, and
feasible solutions for the treatment of Brewers’ spent
grains. Master Brewers Association of the Americas
Technical Quarterly, 43(3): 199–202.
El-Shafey, E. I.; Gameiro, M. L. F.; Correia, P. F. M.;
Carvalho, J. M. R. (2004) Dewatering of Brewer’s spent
grain using a membrane filter press: A pilot plant
study. Sep. Sci. Technol., 39(14): 3237–3261.
DOI:10.1081/SS-200028775
Carvalho, J. M. R.; Machado, R. M.; Henriques, C. M.
C.; Rodrigues, R. A. D.; Correia, P. F. M. M. Integrated
process of filtration to dry brewer’s spent grain. Patent
corporation treaty/PT2010/000016, Patent number
WO 2010117288 A8, 2010.
Svarovsky, L. (1981) Filtration fundamentals. In: L.
Svarovsky (Ed.), Solid-Liquid Separation; Butterworths:
London, Chap. 9, pp. 242–264.
Download