阿尔伯塔大学-五年级语文教学计划

ARTICLE IN PRESS







Water Research 38 (2004) 3340-3348
SBR集成实时控制策略应用于养猪场废水脱氮处理中的研究
Ju-Hyun Kim
a,
*, Meixue Chen
b
, Naohiro Kishida
c
, Ryuichi Sudo
a
Center for Environmental Science in Saitama, 914, Kamitanadare, Kisai, Saitama 347-0115, Japan
State Key Laboratory of Environmental Aquatic Chemistry, Research Center for Eco-Environmental Sciences,
Chinese Academy of Sciences, Beijing 100085, P.O. Box 2871, China
c
Department of Environmental Resources Engineering, Waseda University, 3-4-1 Okubo, Shinjuku-ku, Tokyo 1698555, Japan
b
a
Received 7 May 2003; received in revised form 29 March 2004; accepted 11 May 2004




摘要

新 型的集成实时控制系统正在被设计和应用于水力负荷变化较大的养猪场废水处理。通过使用实时控制
技术 ，以ORP和pH分别作为厌氧段和好氧段的控制参数，从而实现外加碳源的自动添加控制，使得整个处理
系统正常运作。养猪场废水浓度变化幅度大、进水的有机物碳氮比率低是整个生物脱氮工艺的主要限制因素。< br>因此，必须补充足够的碳源才能保证脱氮过程的正常运行。许多研究者对以猪粪便作为外加碳源以保证生物
脱氮效果的可行性进行了探究。实时控制系统能够在进水负荷的循环变化过程中优化猪粪便的添加量。在 应
用了集成实时控制策略后，总碳和总氮的平均去除效率分别可达94%和96%之上。
r 2004 Elsevier Ltd. All rights reserved.

关键词：脱氮；外加碳源；ORP实时控制；SBR；养猪废水



1. 简介

养猪场废水是向环境排放的主要氮污染源之一。
传统生物脱氮 处理主要由一些系列的硝化阶段和反
硝化阶段所组成。养猪废水的浓度变化差别主要取决
于不同 的粪便处理方式，近年来以ORP和pH作为参数
(Lo et al., 1994; Plisson-Saune et al., 1996; Chapentier et
al., 1998; Fuerhacker et al., 2000)的实时控制系 统来
分别控制污水处理周期中好氧和缺氧阶段的SBR反
应器来处理养猪废水受到了关注(Ra et al., 1998, 1999;
Tilche et al., 2001)。但与传统的 处理过程不同的是，
使用ORP和pH作为控制参数进行实时控制的序批式
反应器能够针对不同 的处理情况如进水水力负荷和
处理状况等进行自动调整。因而每个处理周期的水力
停留时间是根 据不同情况而变化的(Ra et al., 2000)。
并且能够达到较高而稳定的氮去除效率(Ra et al.,
1998; Cheng et al., 2000)。
虽然基于ORP和pH的实时控制系统已经在不少

*Corresponding author. Tel.: +81-480-73-8369; fax:
+81-
480-70-2031.
E-mail address: a1098356@ (J.-H. Kim).
养猪废 水处理系统中得到应用，但是至今，这个系统
还难以称之为成功，由于研究者所得到的特定ORP和pH研究数据主要都是来源于完整的硝化和反硝化过
程并且集中于好氧阶段的控制(Ra et al., 1998; Cheng
et al., 2000)。但事实上，由这些特定的ORP和 pH数据
得出的控制点通常难以在使用驯化后的硝酸盐污泥
作为处理单元的系统中再现(Kim and Hao, 2001;
Kishida et al., 2003)。
生物脱氮过程只在异养细菌有可利用的碳源时
才会发生，因而若不补充充分的有机碳源，低碳氮比
废水将限制整体生物脱氮的效果。不少研究者都提出
可使用发酵养猪粪便或者活性污泥作为SBR反应 器
中脱氮过程的电子受体，研究结果也得出这样的外加
碳源对加强SBR的处理效果是具有可行 性的。但是常
由于过量添加反应过程中所需碳源而导致处理成本
的增加。因此外加碳源的添加量 必须与废水的水质水
量波动相适应。



0043-1354$$ - see front matter r 2004 Elsevier Ltd. All rights reserved.
doi:.2004.05.006

ARTICLE IN PRESS
J.-H. Kim et al. Water Research 38 (2004) 3340-3348 3341

本研究的主要目 的就是建立养猪场废水的集成处
理系统和操作策略以适应不同负荷的变化。特别对于
氮碳比较低 的负荷周期，系统也能够优化外加碳源的
添加量以加强脱氮和去除废水中污染物的效果。因此，
作者研究了养猪粪便作为外加碳源的脱氮效果并决定
了其添加的脉冲模型。作为补充，作者也评价了以O RP
和pH作为实时控制参数的实际可行性。可对养猪场废
水连续处理，并具有实时控制和脉冲 输入控制集成策
略的SBR反应器被设计出来并进行实际运转。

2.试验方法

2.1.序批式反应器与操作策略

SBR反应器试验装置见图1。水温 维持在23±2℃。
反应器由树脂玻璃构成，工作容量为9L，内置机械搅
拌器。气体由通风装 置提供，通过置于反应器底部的
砂滤多孔石进入反应器。反应器有五个反应阶段：进
水阶段、缺 氧阶段、好氧阶段、污泥沉淀阶段以及出
水阶段。缺氧和好氧的时间由计算机根据过程变量控
制 ，而进水阶段、污泥沉淀阶段和撇水阶段分别设定
在5、55、5分钟。每个周期的废水进水量为0.3 L。
ORP，pH和DO传感器设置在SBR反应器内部。输
出信号直接由电脑接收。进水泵 、出水泵、曝气装置、
搅拌器以及粪便泵由电缆相连的继电器控制。本次试
验将由高碳氮比进水 负荷开始，该进水取于仅格栅处
理后的废水。实验中，在一周的运行之后，使用了浮
动进水持续 运行了8个月。一旦进水碳源不充分，猪场
粪便将作为外加碳源添加到SBR反应器中以保证脱氮
过程正常运行。
以简易自动控制为目的，猪场粪便的脉冲输入模
式被用于补充外加碳源。在 低碳氮负荷的进水周期中，
溶解的猪场粪便由脉冲计量泵以每次脉冲1g，脉冲间
隔10min 的运行方式抽入SBR反应器中。一旦添加的猪
繼電器匣(開 關)
场废物的量不足以维持 脱氮过程运行，则下一次添加
周期就会开始，而到达指示脱氮过程结束的硝酸盐膝
时，添加过程 就会结束。因此，利用脉冲输入模式，
能够较简便的达到猪场粪便添加量与废水波动性相适
应的 状态。

2.2. 养猪场废水、粪便与污泥

本研究使用的污水取自日 本埼玉县的一个当地农
场。而高碳氮比和低碳氮比的废水分别来自混凝处理
前后，并轮流在试验 中使用。原始废水的碳氮比由于
粪便和尿液的分别放置而显著的改变。废水中的总碳
总氮的比例 在0.45-1.53之间，要求在4℃左右储存。
猪场粪便也取自同一农场。在使用之前，使用了网 孔
直径为0.5mm的网筛以截留较大的固体颗粒，并用自
来水稀释，作为外加碳源。稀释后的 粪便特性见表1。
混合液体悬浮固体（MLSS）的平均浓度保持在
7000mgL左右。当 MLSS的浓度超过8000mgL时，将
抽出一部分污泥。在试验阶段，平均泥龄（SRT）为
32天。

表 1
稀释猪场粪便特性
参数
mg·L
-1
TOC
BOD
5
TN
TP
TSS
平均值
26,167
90,280
4529
2600
917
最小值-最大值
11,410-55,640
46,370-172,200
2418-6882
1500-3810
240-3950
标准偏差
(n ¼ 15)
17,010
31,850
1741
821
43,720

PC輸入輸出卡

實時監測pH和ORP
的動態變化參數值



糞便
出水
進水


ORP傳感器
pH 傳感器
DO傳感器



空氣
曝氣器






SBR
砂滤多孔石
出水桶
Fig. 1. 具有實時控制策略的SBR反應器工作原理圖

ARTICLE IN PRESS
3342 J.-H. Kim et al. Water Research 38 (2004) 3340-3348
2.3.取样及分析方法

常规化验参 数包括TOC、BOD
5
、总氮(TN)、
NH
4
-N、NO
3
-N、NO
2
-N、总磷(TP)、PO
4
-P、MLSS、< br>MLVSS以及总悬浮颗粒(TSS)。覆盖全过程的径迹分
析主要在高负荷和低负荷段. 混合样品取自径迹分析
期间。对 NH
4
-N、NO
3
-N 以及 NO
2
-N的分析分别取自
每一次的径迹分析。 BOD
5
、 TSS、MLSS以及MLVSS
的分析标准采用美国公共卫生协会标准, 1995。The NH
4
-N、NO
3
-N、NO
2
-N以及PO
4
-P 由离子色谱仪分
析检测(Yokogawa IC 7000)。总碳由Shimadzu 总碳分
析仪测定(TOC 5000)。TN 和 TP由总氮和总磷分析仪
测定 (TN-30, TP-30, Mitsu-bishi Chemical Corp)。

3.结果及讨论

3.1高碳氮比负荷周期实时控制点

在高碳氮比负荷周期，采用以ORP以及pH 作为缺
氧阶段和好氧阶段控制参数的实时控制技术，在处理
过程无外加碳源的情况下脱氮段仍有 效运转。在高碳
氮比负荷运转初期，出水稳定并且处理效果优良。在
图2中的A点是设置进水点 ，5分钟之后缺氧段开始。
从污染物数据中可得出，在75分钟内NO
3
-N被还原为
NO
2
-N并且最终完全被还原为氮气 。B点是ORP曲线
上的硝酸盐膝， 代表了硝酸盐已被完全去除。据报道，
在脱氮完成后硫酸盐的量开始减少，并且因此导致了
OR P的突然下(Poisson - Sauna et al., 1996)。C点标志
了好氧阶段的 开始。pH曲线的初始上升阶段是由于系
统在缺氧阶段开始释放二氧化碳和消耗挥发性脂肪
酸( Ra et al., 1998)。在好氧的条件下，NH4-N随时间
减少。氨由于硝化反应而转化 ，硝酸盐浓度则随时间
上升。由于系统中的氨被除去pH的下降。E点标志了
硝化过程的结束即 氨低谷。在硝化过程中，NH
4
-N 转
化为NO
3
-N，如方程式(1) and (2) (EPA, 1975)所示。







在硝酸铵氧化过程中需要一定的碱度（1mg的氨
氮需要7.14mg碳酸钙碱）。在硝化过程中碱性 物质的
减少和酸性物质的产生降低了pH。当氨完全被除去时
标志着废水中碱性物质消耗的结束 。
3.2在低碳氮负荷周期中的实时控制点

在低碳氮负荷周期中控制点的选定对 于集成控
制策略来说非常重要。低碳氮负荷进水负荷的径迹分
析见图3。A点是缺氧阶段的开始 。由污染物数据可得，
来自于好氧阶段硝化反应的NO3-N利用进水提供的
碳源进行缓慢的脱 氮反应。在2小时后，由于进水并
没有提供充分的碳源，反硝化过程并没有反应充分。
S点，开 始添加粪便，在脉冲式添加粪便后，反硝化
反应继续进行，其速率显著上升。在第一次添加粪便
后的十分钟，NO3-N的浓度由11.7 mg·L-1降低到 9.6
mg ·L-1，但并没有 完全反硝化，因此10min后又进行
第二次添加，并不断循环直到达到反硝化完全为止。




在第三次添加之后，硝酸盐的含量降低到零。在
ORP曲线斜 坡阶段，B点突然的变化标志着在厌氧反
硝化中的氮氧化物已经完全反应完毕，系统停止自动
添 加粪便。带有脉冲输入的实时控制集成策略在控制
养猪废水的过程中，主要依赖于在ORP时间轴数据中
的硝酸盐突变点，从而优化粪便添加的过程。C点标
志着好氧阶段的开始。由于低碳氮负荷进水 中的缺氧
阶段二氧化碳以及挥发性脂肪酸产生不足，因此D点
不明显。在好氧情况下，氨氮随时 间坚守。由于氨在
硝化过程被转化，因此硝酸盐浓度随时间增加。而pH
的下降主要是由于系统 中与废水碱度密切相关的氨含
量减少。E点代表了硝化过程的结束，称之为氨低谷。
氨的完全去 除标志着废水中的碱性物质消耗和pH降
低过程的结束。pH在E点上升可能是由于二氧化碳的
释放引起的(Chen et al., 2002)。

3.3集成实时控制策略的确定

猪场粪便的实时控制集成策略以及脉冲输入控制
的设计方案见图4。在缺氧阶段，持续时间 r 被作为鉴
定外加碳源添加必要性的选择参数。一旦进水所提供
的碳源超过这个时间仍无法进行彻 底的反硝化作用，
则系统则会自动添加养猪场粪便（S点）。ORP和pH
数据控制分别应用于 缺氧段和好氧段。dORPdt 和
dpHdt的值被用于监测实时控制点。ORP和pH的值每
1s监测一次并且每五分钟内进行一次平均值。在两个
相邻的平均值之间计算出dORPdt 和dp Hdt。实际上，
由于传感器的不稳定性以及3min间隔内无法实施实
时控制等原因，5mi n的间隔对于计算实时控制策略中
的dORPdt 和dpHdt是比较合适的。
高碳氮负荷和低碳氮负荷进水的dORPdt 和
dpHdt数据可见图5、6。虽然pH感应 器不够稳定并且
B点也不是非常明确，dpHdt还是能够用于硝化段控制
参数，因为氨低谷（ E点）的数值立即由负数转为整
数。dpHdt的数值变化比dORPdt在E点的变化更为明
显；因此，使用pH做为硝化段控制参数更合适。在低
碳氮比负荷周期中，猪场粪便作为补充氮源以便充 分
进行脱氮反应，并且使得ORP数据中的控制点更加清
晰。在靠近B点的位置，由程序计算得 出的dORPdt
的数值显著的降低，并且在该点之后，dORPdt的数值
持续性的降低。该 点可作为反硝化结束点并且小于2
mV·min -1的数值可作为反应系统脱氮情况的实时控
制点。为避免错误的程序控制，在B点设置了5 mV·min
-1作为控制值，同时集成实时控制策略被设计为可逐
步自检差错直到指定的控制点出现。

3.4集成实时控制系统性能

通常一天内，带有自动添加猪场粪便的集 成实时
控制系统策略都会运行很多个周期。该系统能为硝化
过程后的连续性反硝化过程提供最优 化的条件。污染
物的总去除效率可见表2。通过使用实时控制集成策略
和粪便的脉冲输入控制， 在极端波动的进水条件下，
仍能够保持相对稳定的出水水质。平均TOC和总氮去
除效率分别超 过94% 和96%。

4.结论

这一控制策略能够为细菌的生长和性 能提供最佳
条件。基于本研究所取得的结果，养猪场废水处理脱
氮反应实时控制的实际重要性可 总结为以下几点：

3343 J.-H. Kim et al. Water Research 38 (2004) 3340-3348

1. 在养猪场废水SBR处理中，传统实时控制主要有反
供 了条件，HRT的操作具有弹性，基于生物内部活动
硝化不充分和硝酸盐积累两个缺陷。本研究再一次肯
和进水特性，因而保证了相对彻底的污染物除去效果。
定了使用猪场粪便作为反硝化过程的电子 受体的可行
这一系统完全可在浮动进水荷载下用于养猪场废水处
性。并且可基于在ORP时间表 中的硝酸盐突破点，来
理的工程实际。
优化由集成实时控制策略决定的养猪粪便添加量。
4. 虽然ORP和pH能作为彻底反硝化的控制参数，但对
2. 系统取得了较高的氮去除效 率。使用这一控制策略
后，即使在进水负荷极不稳定的情况下，也能取得稳
于硝化阶段来说，p H时间表的控制点尚不够明确，而
定的出水效果。
dpHdt的数值变化却更为显著。因此作者建议ORP和
3. 通过使用集成控制策略，实时控制的优势也能够发
pH应当分别作为反硝化和硝化阶段的控制参数。
挥出来。比如，由于优化的外加碳源为充分反硝化提

进水 (Agitator on)
缺氧阶段控制
停留 5 min

打开粪便添加系统
读取
dORPdt



dORPdt>-5
停留 10 sec


否
否
是
读取
dORPdt
总时间 < r

总时间 < w
关闭粪便添加系统，停留 10 min
是
否




是
读取
dORPdt
否
否
dORPdt<-5


dORPdt<-5

是
停留 30 min
是
r: 120 min
w: 90 min




开曝气器 好氧阶段控制
读取 dpHdt

dpHdt < 0

否


是
读取 dpHdt
否
dpHdt >0


是
停留 20 min
关曝气器、搅拌器

沉淀及出水


Fig. 4.实时控制策略

ARTICLE IN PRESS
3346

进水
0.035

0.030
0.025
0.020

0.015
0.010

0.005
缺氧阶段 好氧阶段

好氧阶段

缺氧阶段
好氧阶段

好氧阶段

J.-H. Kim et al. Water Research 38 (2004) 3340-3348
进水 进水 进水 进水

0.000
e
e
e
e
-0.005
-0.010
25
20
15
10
5
0
-5
-10
-15
-20
B B B B
0 250 500 750 1000 1250 1500 1750
Time (min)
Fig. 5. 高碳氮比负荷控制策略的确定(进水TOCTN比: 1.22-1.53)

进水
ARTICLE IN PRESS
J.-H. Kim et al. Water Research 38 (2004) 3340-3348


进水 进水




好氧阶段

缺氧阶段 好氧阶段

3347
进水
0.040
0.035
0.030
0.025
0.020
0.015
0.010
0.005
0.000
-0.005
-0.010
20

15

10

5

0

-5
缺氧阶段
缺氧阶段 好氧阶段






e










B B





500 750
e e

-10

-15

-20
0 250
B





1500 1000 1250
Time (min)
Fig. 6. 低碳氮比负荷控制策略的确定(进水TOCTN比: 0.45-0.79).


Table 2
实时控制策略去除效果
参数 mg L
1
进水 (n ¼ 52)
Mean
TOC
BOD
5
TN
NH
4
-N
NO
3
-N
TP
PO
4
-P
TSS
—: 检测不出.

864
3206
722
589
—
46
18
917
Min-Max
250-3288
1257-5588
408-1138
385-971
—
13-93
9-13
240-3950
Std. dev.
573
1544
173
137
—
40.4
6.8
1902
出水 (n ¼ 52)
Mean
40
15
26
o0.1
18
23
19
16
Min-max
32-48
8-23
15-39
o0.1
5-26
11-32
7-24
2-27
Std. dev.
4
6
6
5
4
5
7
Removal rate (%)
94.7
99.6
96.2
50.0
98.9

ARTICLE IN PRESS
Water Research 38 (2004) 3340-3348
Integrated real-time control strategy for nitrogen removal in
swine wastewater treatment using sequencing batch reactors
Ju- Hyun Kim
a,
*, Meixue Chen
b
, Naohiro Kishida
c
, Ryuichi Sudo
a
Center for Environmental Science in Saitama, 914, Kamitanadare, Kisai, Saitama 347-0115, Japan
State Key Laboratory of Environmental Aquatic Chemistry, Research Center for Eco-Environmental Sciences,
Chinese Academy of Sciences, Beijing 100085, P.O. Box 2871, China
c
Department of Environmental Resources Engineering, Waseda University, 3-4-1 Okubo, Shinjuku-ku, Tokyo 1698555, Japan
b
a
Received 7 May 2003; received in revised form 29 March 2004; accepted 11 May 2004




Abstract

A new integrated real-time control system was designed and operated with fluctuating influent loads for swine
wastewater treatment. The system was operated with automatic addition control of an external carbon source, using
real-time control technology, which utilized the oxidation-reduction potential (ORP) and the pH as parameters to control
the anoxic phase and oxic phase, respectively. The fluctuations in swine wastewater concentration are extreme; an influent
with a low C=N ratio is deficient in organic carbon, and a low carbon source level can limit the overall biological
denitrification process. Consequently, a sufficient organic source must be provided for proper denitrification. The feasibility
of using swine waste as an external carbon source for enhanced biological nitrogen removal was investigated. The
real- time control made it possible to optimize the quantity of swine waste added as the load fluctuated from cycle to cycle.
The average removal efficiencies achieved for TOC and nitrogen were over 94% and 96%, respectively, using the
integrated real-time control strategy.
r 2004 Elsevier Ltd. All rights reserved.

Keywords: Denitrification; External carbon source; ORP Real-time control; SBR; Swine wastewater



1. Introduction

Swine wastewater has previously been considered as
one of the major sources of nitrogen pollution dis-
charged into the environment. Traditional biological
removal of nitrogen was achieved by a sequence of
nitrification and denitrification processes. Since the
fluctuations in swine wastewater concentration are
extreme due to the varying practices of manure manage-
ment, in recent years, the real-time control process using
oxidation-reduction potential (ORP) andor pH as
parameters (Lo et al., 1994; Plisson- Saune et al., 1996;

*Corresponding author. Tel.: +81-480-73-8369; fax: +81-
480-70-2031.
E-mail address: a1098356@ (J.-H. Kim).
Chapentier et al., 1998; Fuerhacker et al., 2000) to
control the oxic and anoxic cycles of a system has
received much attention for swine wastewater treatment
(Ra et al., 1998, 1999; Tilche et al., 2001) in sequencing
batch reactors (SBRs). Compared to the traditional
process, real-time control strategy for a batch treatment
process using ORP andor pH was self-adjusted to
various treatment conditions such as influent strength
and treatment status. This resulted in flexible hydraulic
retention time (HRT) from cycle to cycle (Ra et al.,
2000). The high and stable removal rate of nitrogen was
also achieved (Ra et al., 1998; Cheng et al., 2000).
Although real-time control strategy based on ORP
andor pH has been applied to many swine wastewater
treatment systems, until now, the success of the systems
has not been convincing because much effort in the

0043-1354$$ - see front matter r 2004 Elsevier Ltd. All rights reserved.
doi:.2004.05.006

ARTICLE IN PRESS
J.-H. Kim et al. Water Research 38 (2004) 3340-3348

3341
studies has dealt primary with the typical ORP andor
pH profiles of a complete nitrification and denitrification
cycle and focused on aeration control (Ra et al., 1998;
Cheng et al., 2000). In fact, the control set- points on the
ORP or pH profiles would not have appeared in the
acclimated nitrate sludge (Kim and Hao, 2001; Kishida
et al., 2003).
Biological denitrification is known to occur by the
action of heterotrophic bacteria using available carbon
sources (John and Robert, 1985; Lee et al., 1995, 1997).
Because the influent with a low C=N ratio is deficient in
organic carbon and the low carbon source level can limit
the overall biological denitrification process, sufficient
organic source must be provided for proper denitrifica-
tion. Using the fermented swine waste (Lee et al., 1997)
or activated sludge (Ra et al., 2000) as an electron donor
for denitrification in SBRs has been suggested by several
authors, and such external carbon sources are viable
choices for enhancing SBR performance. However, any
excess external carbon added over the amount required
for the process appears in the effluent, and results in
increased cost of operation. Therefore, the addition of
the external carbon source should be optimized with the
fluctuation of wastewater.
The specific objective of this study was to establish an
integrated swine wastewater treatment system and
operating strategies suitable for the fluctuations of
influent loads. Particularly, under low C=N load cycles,
the system can optimize the addition of the external
carbon source to enhance nitrogen removal, as well as to


Relay box (On Off)
efficiently remove the pollutants from wastewater. For
this purpose, swine waste as external carbon source for
denitrification of nitrate was examined, and a pulsed
pattern of addition was determined. In addition, ORP
and pH as practical real-time control parameters were
evaluated. The SBR with an integrated strategy of real-
time control and a pulsed input control of swine waste
were designed and continuously operated for swine
wastewater treatment.
2. Methods
2.1. Sequencing batch reactor and operating strategies
The SBR was operated as shown in Fig. 1. The water
temperature was maintained at 2372 C. The reactor
was constructed using Plexiglas and had a working
volume of 9 L. A mechanical agitator was installed in it
for complete mixing. Air (2.4 Lmin) for the reactor was
provided by an aerator through an air stone placed at
the bottom of the reactor. The reactor had five
sequences: influent feeding, anoxic phase, oxic phase,
sludge settling, and effluent transfer. The anoxic and
oxic times were automatically controlled by the compu-
ter depending on the variable process, while the times of
influent feeding, sludge settling and effluent decanting
were fixed at 5, 55 and 5 min, respectively. For every
cycle, 0.3 L of the influent wastewater was fed into the
reactor.
PC
IO PC Card
Monitoring the time
variations of ORP
and pH

Effluent

Influent




Swine waste



Air
Aerator






SBR
Air stone
ORP probe
pH probe
DO probe
Effluent bucket
Fig. 1. Schematic diagram of a single batch sequencing reactor with real-time control strategy.

ARTICLE IN PRESS
3342 J.-H. Kim et al. Water Research 38 (2004) 3340-3348

The ORP, pH and DO probes were inserted into the
SBR. The output signal was directed to a computer. The
influent pump, effluent pump, aerator, agitator and
swine waste pump were controlled by a relay box
connected by electrical cable.
Our experiment was started with high C=N ratio load
(TOCTN>1.5) influent, which was collected after
screening treatment. After one week of operation, the
fluctuating influent was continuously used for 8 months
in our experiment. Swine waste as an external carbon
source was added into the SBR for complete denitrifica-
tion after a time determined by our experiment if there
was insufficient carbon source in the influent.
For the purpose of easy automatic control, a pulsed
input pattern of swine waste was used to compensate for
the external carbon source. During low C=N load
influent cycles, the diluted swine waste was pumped into
the SBR by a pulse- metering pump with a pulse of 1 g
cycle, and the time interval between additions was
designed to be 10 min. If the quantity of swine waste
added was deficient for complete denitrification, the next
addition cycle was started, and when the nitrate knee
point, which indicates the end of denitrification,
appeared on the ORP profile, the addition of swine
waste was stopped. Therefore, the quantity of swine
waste added adapted to fluctuations in the wastewater,
and the optimization was easily achieved by the pulsed
input pattern.


2.2. Influent swine wastewater, waste and seed sludge

The swine wastewater used in this study was obtained
from a local farm in Saitama, Japan. The practical swine
wastewater with high C=N ratio (TOCTN ratio: more
than 1.2) and low C=N ratio (TOCTN ratio: less than
0.8), which was obtained before and after coagulation
treatment, respectively, was used alternately in the
experiment. The C=N ratio of the raw wastewater was
markedly changed by the separation of feces and urine.
The TOCTN ratio in the wastewater was varied in the
range of 0.45-1.53. The swine wastewater was stored at
4 Cuntil required.
The swine waste was obtained from the same farm.
Prior to use, the waste was screened using a sieve with
0.5 mm mesh openings to remove large solids, diluted
with tap water, and then used as an external carbon
source in low C=N ratio load periods for complete
denitrification. The characteristics of the diluted swine
waste are listed in Table 1.
The average concentration of mixed liquor suspended
solids (MLSS) in the system was maintained at
approximately 7000 mg L
1
. When the concentration
MLSS in the reactor was more than 8000 mg L

of
1
,
sludge was drawn out. During the experiment period,
the average SRT was 32 days.
Table 1
Characteristics of the diluted swine waste
Parameters Mean Min-max Std. dev.
mg L
1
(n ¼ 15)
TOC 26,167 11,410-55,640 17,010
BOD
5
90,280 46,370-172,200 31,850
TN 4529 2418-6882 1741
TP 2600 1500-3810 821
TSS 917 240-3950 43,720
2.3. Sampling and analytical methods
Parameters routinely assayed included TOC, BOD
5
,
total nitrogen (TN), NH
4
-N, NO
3
-N, NO
2
-N, total
phosphorous (TP), PO
4
-P, MLSS, mixed liquor volatile
suspended solids (MLVSS), and total suspended solids
(TSS). Track analysis that covered the entire cycle was
carried out at high and low C=N ratio load. Mixed-
liquor samples were taken during track analysis.
Analysis for NH
4
-N, NO
3
-N and NO
2
-N was carried
out for each track study. Analysis for BOD
5
, TSS,
MLSS and MLVSS was performed in accordance with
the standard method (APHA, 1995). The NH
4
-N, NO
3
-
N, NO
2
-N and PO
4
-P were analyzed with an ion
chromatograph (Yokogawa IC 7000). The TOC was
analyzed with a Shimadzu total organic carbon analyzer
(TOC 5000). TN and TP were analyzed with a total
nitrogenphosphorous analyzer (TN-30, TP-30, Mitsu-
bishi Chemical Corp.)
3. Results and discussion
3.1. Real-time control point in high CN ratio load cycles
In high C=N ratio load cycles, with real-time control
technology using ORP and pH as anoxic and oxic
control parameters, a treatment process can be operated
effectively without the addition of an external carbon
source to enhance the denitrification. During the initial
period with the high C=N ratio load influent relative
constant final effluents were obtained along with high
nutrient removal. The typical control set-points are
shown in Fig. 2. Point A is the feeding point, and after
5 min the anoxic phase was started. From the nutrient
profile, it can be seen that NO
3
-N is completely
denitrified to nitrogen gas through NO
2
-N within
75 min, using the influent organic materials as a carbon
source. Point B is known as the nitrate knee in the ORP
curve, which represents the complete removal of nitrate.
Reportedly, sulfate reduction that produces sulfides
starts just after denitrification is complete, and causes
this sudden decrease in the ORP (Plisson-Saune et al.,
1996). Point C signifies the beginning of the oxic phase.

25


20
l
-
1
)


g
·
C
o
n
c
e
n
t
r
a
t
i
o
n

(
m
m



15


10


5



8.1

8.0
7.9
ARTICLE IN PRESS
J.-H. Kim et al. Water Research 38 (2004) 3340-3348

3343
Influent Anoxic phase Oxic phase
NH
4
-N
NO
2
-N
NO
3
-N



































150
100
50
0
O
R
P

(
m
V
)


p
H

A
B
d
c
f
7.8
7.7

7.6

7.5
7.4

7.3
0 25 50
pH
ORP
-50
-100
e
-150
-200
-250









C
-300
-350
75 100 125 150
175 200 225
250 275 300 325 350
375 400
Time (min)
Fig. 2. Real-time control points in high C=N ration load cycles (TOCTN ratio of the influent: 1.4) A: Feeding, B: Nitrate knee point,
C and c: Beginning of the oxic phase, e: Ammonia valley point, f: End of the oxic phase.

The initial rise on the pH curve (from c point to d point)
is caused by carbon dioxide stripping from the system
and the rapid consumption of VFA that is produced
during the anoxic phase (Ra et al., 1998). Under oxic
condition, NH
4
-N decreases with time. Nitrate concen-
tration increases with time as ammonia is converted
through nitrification. The decrease in pH is caused by
the removal of ammonia from the system. Point e
represents the end of nitrification and it is known as the
ammonia valley. During nitrification, NH
4
-N is con-
verted into NO
3
-N, as shown in Eqs. (1) and (2) (EPA,
1975).
55NH
4
þ 76O
2
þ 109HCO
3
-C
5
H
7
NO
2
þ 54NO
2
þ 57H
2
O þ 104H
2
CO
3

400NO
2
þ NH
4
þ 4H
2
CO
3
þ HCO
3
þ 195O
2
-C
5
H
7
NO
2
þ 3H
2
O þ 400NO
3
ð1Þ
Alkalinity is required in the ammonia-nitrate oxida-
tion process (7.14 mg of alkalinity as CaCO
3
to 1 mg of
ammonia-N). The reduction of alkalinity and the acid
production during nitrification decrease the pH. The
complete removal of ammonia indicates the end of
alkalinity consumption in the wastewater, hence the end of
further pH decrease.
3.2. Real-time control point in low CN ratio load cycles
The designation of a control point in low C=N ratio
load cycles was very important for integrated real-time
control strategy. The track analysis with low C=N load
influent is shown in Fig. 3. Point A is the beginning of
the anoxic phase. From the nutrient profile, it can be
seen that NO
3
-N that is produced from nitrification
during the previous oxic phase is slowly denitrified using
the carbon source provided by the feed. After 2 h,
complete denitrification was not reached due to insuffi-
ð2Þ

ARTICLE IN PRESS
3344


J.-H. Kim et al. Water Research 38 (2004) 3340-3348

25


20


15


10


5


8.1

8.0

7.9

7.8
Influent Anoxic phase Oxic phase
NH
4
-N
NO
3
-N
NO
2
-N

l
-
1
)


C
o
n
c
e
n
t
r
a
t
i
o
n

(
m
g
·
S










150
c
A
Swine waste addition

S
B
f
100
50
0
-50
-100
O
R
P

(
m
V
)


p
H

7.7
7.6
7.5
7.4
pH
ORP
e
-150
-200
C
0 25 50 75 97 122 147 172 197 222 247 272 297 322 347 372 397
-250
-300

7.3
Time (min)
Fig. 3. Real-time control points in low C=N ratio load cycles (TOCTN ratio of the influent: 0.7) A: Feeding, B: Nitrate knee point, C
and c: Beginning of the oxic phase, e: Ammonia valley point, f: End of the oxic phase, S: Beginning of the swine waste addition.


cient carbon source provided by the influent, and at
point S, the addition of swine waste was started. The
NO
3
-N is gradually denitrified using swine waste with
every pulsed addition, and the denitrification rate
rapidly increases. Ten minutes after the first addition
of swine waste, the concentration of NO
3
-N in the
1
reactor decreased from 11.7 to 9.6 mg L . The complete
denitrification was not reached, and the second addition
of swine waste was started. The program was cycled to
provide external carbon source for complete denitrifica-
tion. After the third addition of swine waste, the
concentration of nitrate was decreased to zero, and an
abrupt change in the slope of the ORP curve was
appeared at point B, denoting the complete disappear-
ance of the NO
x
-N through denitrification in the anoxic
phase. Until point B, the program of swine waste
addition was stopped automatically. Integrated strategy
of real- time control with pulsed input control of the
swine waste based on the nitrate breakpoint that
occurred in the ORP-time profile enables the optimiza-
tion of swine waste addition. Point C signifies the
beginning of the oxic phase. The d point apparently did
not appear because insufficient carbon dioxide and VFA
were produced in the anoxic phase due to low C=N ratio
influent. Under oxic conditions, NH
4
-N decreases with
time. Nitrate concentration increases with time as
ammonia is converted through nitrification. The de-
crease in pH is caused by the removal of ammonia from
the system, as ammonia is strongly related to alkalinity
of the wastewater. Point e represents the end of
nitrification and it is known as the ammonia valley.
The complete removal of ammonia indicates that the
alkalinity consumption in the wastewater has stopped,
hence the end of further pH decrease. The rise in pH
beyond point e might be caused by air stripping of
carbon dioxide (Chen et al., 2002).

ARTICLE IN PRESS
J.-H. Kim et al. Water Research 38 (2004) 3340-3348

3345
3.3. Designation of integrated real- time control strategy

The designed program for the integrated strategy of
real- time control and pulsed input control of swine
waste is outlined in Fig. 4. In anoxic phase control,
extended time r was used as a selector to identify the
necessity of an external carbon source for denitrification;
if complete denitrification was not attained using the
carbon source provided by influent beyond this time, the


Influent (Agitator on)
addition of swine waste would be automatically started (S
point). The ORP and pH profile controls were applied
during anoxic and oxic phases, respectively. The values
of dORPdt and dpHdt were used to detect the real-
time control point. The values of ORP (pH) were
collected at intervals of 1s and were averaged over
5 min. The dORPdt (dpHdt) values were calculated
between two averaged ORP (pH) values. In fact, the
interval of 5 min was suitable to calculate the dORPdt
Anoxic Phase Control
Stay 5 min

Turn on swine waste addition
Read
dORPdt



dORPdt>-5
Stay 10 sec


no
no
yes
Read
dORPdt
Sum time < r

Sum time < w
Turn off swine waste addition, and stay 10 min
yes
no




yes
Read
dORPdt
no
no
dORPdt<-5


dORPdt<-5

yes
Stay 30 min
yes
r: 120 min
w: 90 min




Air on Oxic Phase Control
Read dpHdt

dpHdt < 0

no


yes
Read dpHdt
no
dpHdt >0


yes
Stay 20 min
Air off, agitator off

Settling and effluent


Fig. 4. Real-time control strategy.

ARTICLE IN PRESS
3346

Influent feeding
0.035

0.030
S
e
t
t
l
i
n
g

a
n
d

e
f
f
l
u
e
n
t


S
e
t
t
l
i
n
g

a
n
d

e
f
f
l
u
e
n
t


S
e
t
t
l
i
n
g

a
n
d

e
f
f
l
u
e
n
t


0.025
d
p
H

d
t

(
m
i
n
-
1
)


0.020

0.015
0.010

0.005
Anoxic phase
A
n
o
x
i
c

p
h
a
s
e

A
n
o
x
i
c

p
h
a
s
e

Oxic
phase
Oxic
phase
S
e
t
t
l
i
n
g

a
n
d

e
f
f
l
u
e
n
t


Anoxic phase
Oxic
phase
Oxic
phase
J.-H. Kim et al. Water Research 38 (2004) 3340-3348
Influent feeding Influent feeding Influent feeding Influent feeding

0.000
e
e
e
e
-0.005
-0.010
25
20
15
10
5
0
-5
m
i
n
-
1
)


d
O
R
P

d
t

(
m
V
·
-10
-15
-20
B B B B
0 250 500 750 1000 1250 1500 1750
Time (min)
Fig. 5. Designation of real- time control strategy with high C=N ratio load (TOCTN ratio of the influent: 1.22-1.53).


(dpHdt) for real-time control strategy due to instability of
the sensors, and real- time control would fail if the
interval was shorter than 3 min.
The dORPdt (dpHdt) profiles with high and low
C=N ratio influent were shown in Figs. 5 and 6,
respectively. Although the pH sensor was unstable,
and the B point was not very clear, dpHdt could be
used as a nitrification control parameter because of the
sharply changing value from negative to positive at the
ammonia valley (e point). The change in value of dpHdt
was more apparent than that of dORPdt at point e;
therefore, it would be better to use pH as the
nitrification control parameter. During low C=N ratio
influent cycles, swine waste was added to provide a
suitable carbon source for complete denitrification, and
it caused the control point on the ORP profile to appear
clearly. Near the B point, the value of dORPdt
calculated by the program markedly decreased (the
1 1
range was from 1.6 mV min to 16.6 mV min),
and after this point, the value of dORPdt continuously
decreased. The point refers to complete denitrification,
and a value less than 2 mV min could be selected as a
real-time control point to reflect the denitrification
condition of the system. To prevent erroneous process
1
control, 5 mV min was set as the control value at the B
point, and the integrated real-time control strategy was
programmed to recognize each feature step by step, until
the designated control point appeared.
3.4. Integrated real-time control system performance
Using the integrated real- time control strategy with
automatic control of swine waste addition, a number of
cycles occurred each day. This system could provide
near optimal conditions for continuous denitrification
followed by nitrification. The removal efficiencies of
nutrient are summarized in Table 2. By using the
integrated strategy of real-time control and pulsed input
control of swine waste with the extreme fluctuations in
influent, a relatively constant final effluent was obtained.
The average removal efficiencies of TOC and nitrogen
were over 94% and 96%, respectively.
1

Influent feeding
ARTICLE IN PRESS
J.-H. Kim et al. Water Research 38 (2004) 3340-3348


Influent feeding Influent feeding




S
e
t
t
l
i
n
g

a
n
d

e
f
f
l
u
e
n
t


Anoxic phase
S
e
t
t
l
i
n
g

a
n
d

e
f
f
l
u
e
n
t


Oxic
phase
3347
Influent feeding
0.040
0.035
d
p
H

d
t

(
m
i
n
-
1
)


0.030
0.025
0.020
0.015
0.010
0.005
0.000
-0.005
-0.010
20

15

10

5

0

-5
Anoxic phase Oxic
phase





e










B B





500 750
e e
m
i
n
-
1
)


d
O
R
P

d
t

(
m
V
·

-10

-15

-20
0 250
B





1500 1000 1250
Time (min)
Fig. 6. Designation of real-time control strategy with low C=N ratio load (TOCTN ratio of the influent: 0.45-0.79).


Table 2
Nutrient removal performance with real-time control strategy
Parameters mg L
1
Influent (n ¼ 52)
Mean
TOC
BOD
5
TN
NH
4
-N
NO
3
-N
TP
PO
4
-P
TSS
—: No detection.


864
3206
722
589
—
46
18
917
Min-Max
250-3288
1257-5588
408-1138
385-971
—
13-93
9-13
240-3950
Std. dev.
573
1544
173
137
—
40.4
6.8
1902
Effluent (n ¼ 52)
Mean
40
15
26
o0.1
18
23
19
16
Min-max
32-48
8-23
15-39
o0.1
5-26
11-32
7-24
2-27
Std. dev.
4
6
6
5
4
5
7
Removal rate (%)
94.7
99.6
96.2
50.0
98.9
4. Conclusions
With integrated real-time control of the system, a
number of cycles occurred each day. This control
strategy could provide near optimum conditions for
bacterial growth and performance. Based on the
results obtained in this study, the practical impor-
tance of the real-time control for nitrogen removal of
S
e
t
t
l
i
n
g

a
n
d

e
f
f
l
u
e
n
t


Anoxic phase Oxic
phase

ARTICLE IN PRESS
3348 J.-H. Kim et al. Water Research 38 (2004) 3340-3348

swine wastewater treatment can be summarized as
follows:
1. The ordinary real-time control would be defeated by
the incomplete denitrification and accumulation of
nitrate in the SBR for the swine wastewater treatment
process. Using the swine waste as electron donor for
denitrification in SBRs was emphasized in this
research, and the optimization of swine waste
addition was controlled by an integrated strategy of
real- time control with pulsed input control on the
basis of the nitrate breakpoint occurring in the ORP-
time profile.
2. High removal of nitrogen was obtained in the system.
By achieving the integrated strategy of real-time control
with pulsed input control of swine waste, a constant
effluent quality can be obtained, despite variations in
the characteristics of influent wastewater. The advan-
tages of real-time control can be developed using the
integrated strategy, for example, a relatively complete
removal of nutrients is always ensured, because the
optimum external carbon source could be provided for
complete denitrification, and a flexible HRT can be
achieved, depending on the biological activity inside the
system and the influent characteristics. This system
program could be chosen for practical swine waste-
water treatment with fluctuating influent.
3. Although both ORP and pH could be control
parameters for complete denitrification, the control
point on the pH profile was not clear using the pulsed
pattern of swine waste addition for investigation of
denitrification, and with the more sharply changing
values of dpHdt on the control profiles for detection of
nitrification, it was suggested that ORP and pH
should be chosen as denitrification and nitrification
control parameters, respectively.

References

Chapentier, J., Martin, G., Wacheux, H., Gilles, P., 1998. ORP
regulation and activated sludge: 15 years of experience.
Water Sci. Tech. 38 (3), 197-208.
Cheng, N., Lo, K.V., Yip, K.H., 2000. Swine wastewater
treatment in a two-stage sequencing batch reactor using
real-time control. J. Environ. Sci. Health B35 (3), 379-398.
Chen, K.C., Chen, C.Y., Peng, J.W., Houng J, Y., 2002. Real-
time control of an immobilized-cell reactor for wastewater
treatment using ORP. Water Res. 36, 230-238.
Fuerhacker, M., Bauer, H., Ellinger, R., Sree, U., Schmid, H.,
Zibuschka, F., Puxbaum, H., 2000. Water Res. 34 (9),
2499-2506.
John, C., Robert, L., 1985. Nitrogen removal in a low-loaded
single tank sequencing batch reactor. J. Water Pollut.
Control. Fed. 57 (1), 82-86.
Kim, H., Hao, O.J., 2001. pH and Oxidation-reduction
potential control strategy for optimization of nitrogen
removal in an alternating aerobic- anoxic system. Water
Environment Research 73 (1), 95-102.
Kishida, N., Kim, J.H., Chen, M., Sasaki, H., Sudo, R., 2003.
Effectiveness of oxidation-reduction potential and pH as
monitoring and control parameters for nitrogen removal in
swine wastewater treatment by sequencing batch reactors.
J. Biosci. Bioeng. 96, 285-290.
Lee, S.I., Koopman, B., Park, S.K., Cadee, K., 1995. Effect of
fermented wasters on denitrification in activated sludge.
Water Environ. Res. 67 (7), 1119-1122.
Lee, S.I., Parkm J, H., Ko, K.B., Koopman, B., 1997. Effect of
fermented swine wasters on biological nutrient removal
in sequencing batch reactors. Water Sci. Tech. 31 (7),
1807-1812.
Lo, C.K., Yu, C.W., Tam, N.F.Y., Trayor, S., 1994. Enhanced
nutrient removal by oxidation-reduction potential (ORP)
controlled aeration in a laboratory scale extended aeration
treatment system. Water Res. 28 (10), 2087-2094.
Plisson-Saune, S., Capdeville, B., Mauret, M., Deguin, A.,
Baptiste, P., 1996. Real-time control of nitrogen removal
using three ORP bending points: signification, control
strategy and results. Water Sci. Tech. 33 (1), 275-280.
Ra, C.S., Lo, K.V., Mavinic, D.S., 1998. Real-time control of
two-stage sequencing, batch reactor system for the treat-
ment of animal wastewater. Environ. Tech. 19, 343-356.
Ra, C.S., Lo, K.V., Mavinic, D.S., 1999. Control of a swine
manure treatment process using a specific feature of
oxidation reduction potential. Bioresource Technology 70,
117-127.
Ra, C.S., Lo, K.V., Shin, J.S., Oh, J.S., Hong, B.J., 2000.
Biological nutrient removal with an internal organic carbon
source in piggery wastewater treatment. Water Res. 34 (3),
965-973.
Standard Methods for the Examination of Water and Waste-
water., 1995. 19th Ed. American Public Health Association
American Water Works AssociationWater Environment
Federation. Washington DC, USA.
Tilche, A., Bortone, G., Malaspina, F., Piccinini, S., Stante, L.,
2001. Biological nutrient removal in a full-scale SBR
treating piggery wastewater: results and modeling. Water
Sci. Tech. 43 (3), 363-371.
US Environmental Protection Agency 1975. Process design
manual for nitrogen control. EPA Technology Transfer,
Washington DC, USA.