Elsevier

Environmental Pollution

Volume 151, Issue 3, February 2008, Pages 460-469
Environmental Pollution

IPCS: An integrated process control system for enhanced in-situ bioremediation

https://doi.org/10.1016/j.envpol.2007.04.010Get rights and content

Abstract

To date, there has been little or no research related to process control of subsurface remediation systems. In this study, a framework to develop an integrated process control system for improving remediation efficiencies and reducing operating costs was proposed based on physical and numerical models, stepwise cluster analysis, non-linear optimization and artificial neural networks. Process control for enhanced in-situ bioremediation was accomplished through incorporating the developed forecasters and optimizers with methods of genetic algorithm and neural networks modeling. Application of the proposed approach to a bioremediation process in a pilot-scale system indicated that it was effective in dynamic optimization and real-time process control of the sophisticated bioremediation systems.

Introduction

Enhanced in-situ bioremediation is an approach for cleaning up petroleum-contaminated sites (Schoefs et al., 2004). Enhanced in-situ biodegradation is used in conjunction with groundwater pumping and soil flushing systems to circulate aerated and nutrient-enriched water through a contaminated aquifer and associated soils through a series of injection wells. The process usually involves stimulating the indigenous bacterial with appropriate nutrients or introducing non-indigenous bacterial to degrade the organic contaminants in groundwater (Huang et al., 2007). Generally, to improve remediation efficiency and cost-effectiveness, many factors, such as pumping rate, oxygen addition, nutrient supply, and groundwater temperature, can be adjusted to reach the desired levels for microorganisms to degrade the hydrocarbon effectively (Lenczewski et al., 2003). The designed system must be able to facilitate on-line adjustment and control of these factors according to varying site conditions (Rutherford and Johnson, 1996). Therefore, an integrated process control approach for enhanced in-situ bioremediation system is desired to make these systems more efficient.

Previous approaches of model predictive control (MPC) and intelligent process control contributed to improving product quality and efficiency and reducing the costs of industrial production and pollution control (Khalid, 1993, Tan, 1996, Ferrer et al., 1998, Rao and Rawlings, 2000, Zhou et al., 2002, Guh, 2003). MPC is capable of dealing with linear and simple non-linear systems. However, for non-linear predictive control of a subsurface remediation system, the optimization problem that needs to be solved on-line may require tremendous computational efforts, and in many cases this cannot be completed even by fast computing systems. Although intelligent process control does not have to deal with complex non-linear issues, it has a high requirement of prior operation knowledge that highly depends on expert experiences and may not be available for practical in-situ bioremediation (Huang et al., 2003a). It is desirable to develop a process control system for site remediation practices using a more effective simulation-optimization hybrid by capitalizing the advantages of both model predictive control and intelligent process control.

As an extension of previous efforts, Huang et al. (2006a) developed an integrated numerical and physical modeling system for simulating an enhanced in-situ biodegradation process coupled with three-dimensional multiphase multicomponent flow and transport within a multi-dimensional pilot-scale physical model. With the developed integrated modeling system. A forecasting system was then developed for optimal remediation design and process control based on stepwise-cluster analysis (Huang et al., 2006b). In this paper, based on Huang et al., 2006a, Huang et al., 2006b, a framework will be proposed to show how an integrated process-control system (IPCS) can be developed for a complex subsurface remediation system, based on techniques of physical and numerical models, stepwise cluster analysis, non-linear optimization and artificial neural networks (ANN). The development of the optimization model and the ANN controller will be emphasized for the bioremediation process within a pilot-scale system, targeting on realizing integrated process control. A complete optimization model will be developed, with a set of optimization results being generated through the genetic algorithm. Then an artificial neural networks (ANN) model will be developed to establish the relationship between various remediation conditions and responding benzene concentrations. Finally, a framework of process control system will be developed.

Section snippets

Research framework

Fig. 1 shows the general procedure for developing a process control system for enhanced in-situ bioremediation. It consists of seven steps. First, an integrated numerical and physical modeling system is developed through designing a three-dimensional (3-D) pilot-scale model to support the operation of enhanced in-situ bioremediation after a hydrocarbon spill, and developing a multi-phase, multi-component subsurface flow and transport model with biodegradation kinetics that reflects the in-situ

Pilot-scale model

A physical pilot-scale system was developed to support calibration and verification of numerical models with interior dimensions of 3.6 meters (m), 1.2 m, and 1.4 m for the length (L), width (W), and height (H), respectively. Stratified soil in the system contained four layers. In total, 25 wells were installed in the system for monitoring purpose. Some experimental design factors are listed in Table 1. Detailed design of the pilot-scale system is presented in Huang et al. (2006a). Tap water in a

Development of a forecasting system

A forecasting system based on stepwise cluster analysis (SCA) was then developed to reflect the relationships between system performance and operating condition. A stepwise cluster analysis method can effectively deal with discrete and non-linear complexities (Huang, 1992). The solutions of the numerical model (benzene concentrations at selected locations, denoted as x=(x1,x2,,xp), where p=6) were considered as dependent variables in the stepwise cluster analysis. The operation conditions were

Non-linear optimization through genetic algorithm

For each contamination-level scenario (with the initial benzene concentrations of (x10,x20,x30,x40,x50,andx60), the resulting forecasting system could be directly incorporated into a non-linear optimization model as follows:Minz=w1(η1)2+w2Sη2+w3i=14Ui+w4i=58Uisubject to:Ui=uiULiUUiULii=1,2,,8ηi=(xixi0)/xi0i=1,2,,6η=i=16ηi/6Sη=i=16(ηiη)2/6xi=fi(u1,u2,u3,u4,u5,u6,u7,u8)i=1,2,,6u3+u4u1u20u3+u4u1u220ULiuiUUii=1,2,,8where ui is the operating-conditions variable, i=1,2,,8; xi is

Neural networks for supporting process control

During past years, significant progress in the fields of non-linear dynamic pattern recognition and system control theory was made through advances of artificial neural network modeling (ANN). ANN is a non-linear and dynamic mathematical structure with a unique ability of recognizing underlying relationships between input and output events (Hsu et al., 1995). ANN is composed of many simple interconnected neurons which have two components (Dayhoff, 1990): (1) a weighted sum which performs a

Process control system for in-situ bioremediation

The developed NN were trained to establish relationships between optimal operation decisions and subsurface-contamination situations. Therefore, it could be used as a process controller for operating the remediation system. As the last two steps in Fig. 1, a simple control system for enhanced in-situ bioremediation was designed (Fig. 4). First, benzene concentrations at the selected wells at the beginning of time period t were monitored. The concentration levels were then used as inputs for the

Conclusions

A framework was proposed to develop an integrated process-control system (IPCS) for a complex subsurface remediation system, based on techniques of physical and numerical models, stepwise cluster analysis, non-linear optimization and artificial neural networks (ANN). Process control for enhanced in-situ biodegradation was accomplished through incorporating the developed forecasters and optimizers with methods of genetic algorithm and neural networks modeling. The development of the optimization

Acknowledgements

The authors would like to thank the anonymous reviewers for their insightful comments and suggestions that were helpful for improving the manuscript. This paper was supported by the “National Basic Research (973) Program” Project (No. 2005CB724202) of the Ministry of Science and Technology of China, National Natural Science Foundation of China (No. 50509010 and No. 50221903), and Tsinghua University Basic Research Foundation (JCQN2005008).

References (27)

  • G.L. Chierici

    Principles of Petroleum Reservoir Engineering

    (1994)
  • A.J. Chipperfield et al.

    Genetic Algorithm Toolbox User's Guide

    (1994)
  • L. Davis

    Handbook of Genetic Algorithms

    (1991)
  • Cited by (10)

    • Identifying remedial solutions through optimal bioremediation design under real-world field conditions

      2021, Journal of Contaminant Hydrology
      Citation Excerpt :

      the extraction (Fan et al., 2014; He et al., 2010; Li et al., 2015; Mategaonkar and Eldho, 2012) and the injection rates (Zou et al., 2009) applied at each well, the concentrations of electron acceptors (Huang et al., 2008) and nutrient concentration (Hu et al., 2007; Hu and Chan, 2015).

    • In-situ bioremediation for petroleum contamination: A fuzzy rule-based model predictive control system

      2015, Engineering Applications of Artificial Intelligence
      Citation Excerpt :

      But due to the complex computation involved in generating the solutions, these systems cannot operate in real-time. In Huang et al. (2008), the theoretical framework of an integrated control system was proposed to improve performance of the remediation control system, which involves a combination of numerical model, stepwise cluster analysis, optimization and artificial neural networks. Model predictive control (MPC) is one of the most compelling advanced process control algorithms in recently years.

    • Quasi-Monte Carlo based global uncertainty and sensitivity analysis in modeling free product migration and recovery from petroleum-contaminated aquifers

      2012, Journal of Hazardous Materials
      Citation Excerpt :

      This implies that a short-term pumping scheme would not be suggested due to its sharp decline in recovery rates. A feasible way to enhance recover efficiency is to introduce multistage pumping [3] or process control strategies [34]. It should be mentioned that this comparison was conducted under the operating conditions of 1 m3/h for water pumping and −4 m H2O column for vacuum pumping.

    • A coupled simulation-optimization approach for groundwater remediation design under uncertainty: An application to a petroleum-contaminated site

      2009, Environmental Pollution
      Citation Excerpt :

      Groundwater remediation systems can be used to control the flow and transport of contaminants in groundwater and prevent uninterrupted expansion of contamination zones (USEPA, 1994; Huang et al., 2006). Historically, a number of coupled simulation–optimization approaches have been developed for optimal design of such systems (Abriola and Pinder, 1989; Tam and Byer, 2002; McKinney and Lin, 1996; Minsker and Shoemaker, 1996, 1998; Mulligan and Ahlfeld, 2002; Mayer et al., 2002; Zheng and Wang, 2002; Baú and Mayer, 2006, 2007; Huang et al., 2007). In addition, many efforts have been taken to address various parameter uncertainties in remediation design (Wagner and Gorelick, 1987; Andricevic and Kitanidis, 1990; Tiedeman and Goreilick, 1993; Aly and Peralta, 1999; Freeze and Gorelick, 1999; Guan and Aral, 2004; Yan and Minsker, 2006; Lu et al., in press-a, in press-b).

    • Simulation and optimization technologies for petroleum waste management and remediation process control

      2009, Journal of Environmental Management
      Citation Excerpt :

      The control system included an optimization tool that consisted of a simulation model and an optimization function. Huang et al. (2008) proposed a framework of an integrated process control system for improving remediation efficiencies and reducing operating costs based on physical and numerical models, stepwise cluster analysis, non-linear optimization and artificial neural network. In brief, a great number of process control studies exist but very few of them addressed soil and groundwater remediation processes.

    View all citing articles on Scopus
    View full text