Elsevier

Expert Systems with Applications

Volume 38, Issue 9, September 2011, Pages 11538-11546
Expert Systems with Applications

An approach to interval programming problems with left-hand-side stochastic coefficients: An application to environmental decisions analysis

https://doi.org/10.1016/j.eswa.2011.03.031Get rights and content

Abstract

An interval programming with stochastic coefficients (IPSC) model is developed for planning of regional air quality management. The IPSC model incorporates stochastic coefficients with multivariate normal distributions within an interval parameter linear programming (ILP) framework. In IPSC, system uncertainties expressed as stochastic coefficients and intervals are addressed. Since stochastic coefficients are the left-hand-side (LHS) parameters of the constraints in IPSC, a left-hand-side chance-constrained programming (LCCP) method is developed to solve the problem. The developed IPSC model is applied to a regional air quality management system. Uncertainties in both abatement efficiencies expressed as stochastic coefficients and environmental standards expressed as intervals are reflected. Interval solutions associated with different violation probability levels and/or different environmental standards have been obtained. Air quality managers can thus analyze the solutions with appropriate combinations of the uncertainties and gain insight into the tradeoffs between the abatement costs and the risks of violating different environmental standards.

Highlights

► The IPSC model incorporates stochastic coefficients with multivariate normal distributions within an interval parameter linear programming (ILP) framework. ► A left-hand-side chance-constrained programming (LCCP) method is developed to solve the problem. ► The developed IPSC model is applied to a regional air quality management system. ► Interval solutions associated with different violation probability levels and/or different environmental standards have been obtained.

Introduction

Optimization techniques were widely used in the field of environmental management and pollution control (Huang & Chang, 2003). Many innovated optimization techniques were developed to tackle uncertainties existing in environmental systems (Chanas and Zielinski, 2000, Chang and Kashani, 2009, Chang and Wang, 1997, He and Huang, 2008, Huang et al., 1992, Huang et al., 1994, Huang et al., 1995, Huang et al., 1996, Huang et al., 2007, Liu et al., 2009, Luo et al., 2005, Qin and Huang, 2009, Yeomans, 2008, Yeomans and Huang, 2003). Among them, Huang et al. (1992) developed an interval linear programming (ILP) approach for municipal solid waste management system. In ILP, all parameters in objective function and constrains were interval numbers. A two-step method was developed to solve the ILP problem. It has the following advantages: (1) it was applicable to practical problems because it was relatively easy for planners to define fluctuation intervals of uncertain parameters instead of specifying their probability distributions; (2) the two-step method did not lead to high computational requirements; (3) in ILP, interval solutions were obtained which could reflect the inherent system uncertainties and were convenient for the managers to interpret and adjust the decision schemes according to practical situations. However, some parameters do have certain probability distributions which should be incorporated into the optimization process.

It is thus desired that distribution information be integrated into the ILP framework Based on ILP and chance-constrained programming (CCP), Huang introduced an inexact chance-constrained programming method (ICCP) for water quality management within an agricultural system (Huang, 1998). In this method, the left-hand-side (LHS) parameters were interval numbers, and the right-hand-side (RHS) ones were allowed to be stochastic variables. Compared with ILP, ICCP can reflect stochastic coefficients with known probability density distributions in RHS parameters of the constraints. Moreover, solutions of ICCP are intervals under different violation probability levels. However, ICCP cannot deal with problems where the LHS parameters of the constraints are stochastic variables.

Therefore, the objective of this research is to develop a method of interval programming with stochastic coefficients (IPSC) that can address the above problem. In IPSC, uncertainties could be presented as stochastic coefficients in the LHS of the constraints. Meanwhile, intervals will be introduced as the coefficients in both the objective function and constraints. A left-hand side chance-constrained programming (LCCP) method was developed to solve the IPSC problem. The solutions will be presented as intervals which will supply a number of alternatives such that the decision makers could incorporate their preferences and experiences when deciding the final management strategy. The developed model will then be applied to the planning of a regional air quality management system, and the results would be used for helping regional air quality managers to identify desired decision schemes.

Section snippets

Interval programming with stochastic variables (IPSC)

In interval-parameter linear programming (ILP), interval values are allowed to be directly communicated into the optimization process (Huang et al., 1995). An ILP model can be formulated as follows (Huang et al., 1992):Minf±=c1±x1±++cn±xn±Subject toa11±x1±++a1n±xn±b1±,ai1±x1±++ain±xn±bi±,am1±x1±++amn±xn±bm±,x1±,,xn±0,where cj±,aij±,bi±R±, and R± denotes a set of interval numbers. An interactive solution algorithm was proposed to solve the above problem. The solution for model (1a),

Left-hand-side chance-constrained programming (LCCP)

Suppose that the constraints are independent with each other, and all the parameters (cj±) in the objective function are not less than 0. Taking the ith (1  i  l) constraint as an example, a method for handling constraints with stochastic variables in its LHS and intervals in its RHS will be illustrated as follows.

Suppose that ξ = (ξi1, ξi2, …, ξin)T obeys an n-dimensional normal distribution:ξN(μ,V),where μ = (ai1, ai2, …, ain)T is the expectation of ξ and V = (vij)n×n is the variance-covariance matrix of ξ

Overview of the study system

The phenomenon of regional air pollution involves many processes, such as pollutant generation, emission, transport, transformation and removal. A number of impact factors like properties of pollutants, locations of emission sources and receptors, meteorological conditions, and control measures make the management of air quality complicated (Ellis, 1991, Ellis and Bowman, 1994, Fortin and McBean, 1983, Guldmann, 1986, Haith, 1982, Liu et al., 2003). Since it is generally either economically

Result analysis

Table 4 presents the solutions obtained form the IPSC model under different environmental restrictions and violation probabilities. In general, the LWS method should be adopted during the first three years (period 1). In period 2, the methods of LWS and OPA should be used. The amounts of emission as treated by those technologies would vary with the environmental standards and violation probabilities. In period 3, the LWS method would be out of use due to its low efficiency. Instead, both OPA

Conclusions

An interval programming with stochastic variables (IPSC) approach was developed for air quality management under uncertainty. Stochastic variables with multivariate normal distributions were introduced as left-hand-side-parameters. Other parameters and decision variables in the IPSC were expressed as intervals to reflect the uncertainties. A left-hand-side chance-constrained method (LCCM) was developed to solve the IPSC model. Compare with the existing methods, IPSC was advantageous in

Acknowledgments

This research was supported by the Major State Basic Research Development Program (2005CB724200), and the Natural Science and Engineering Research Council of Canada. The authors are grateful to the editor and reviewers for their insightful comments.

References (35)

  • D.R. Cox et al.

    Testing multivariate normality

    Biometrika

    (1978)
  • J.H. Ellis

    Stochastic programs for identifying critical structural collapse mechanisms

    Applied Mathematical Modeling

    (1991)
  • H. Ellis et al.

    Critical loads and development of acid rain control options

    ASCE Journal of Environmental Engineering

    (1994)
  • R.C. Flagan et al.

    Fundamentals of air pollution engineering

    (1988)
  • J.M. Guldmann

    Interactions between weather stochasticity and the locations of pollution sources and receptors in air quality planning: A chance-constrained approach

    Journal of Geometric Analysis

    (1986)
  • A.D. Haith

    Environmental systems optimization

    (1982)
  • L. He et al.

    Optimization of regional waste management systems based on inexact semi-infinite programming

    Canadian Journal of Civil Engineering

    (2008)
  • Cited by (12)

    • Explicit cost-risk tradeoff for optimal energy management in CCHP microgrid system under fuzzy-risk preferences

      2018, Energy Economics
      Citation Excerpt :

      In contrast, for interval parameter programming where uncertainties are expressed as interval numbers with lower and upper bound, accurate probability distribution information is not necessary. Consequently, it requires less information and easier algorithm, and has become a popular tool for reflecting uncertainties in complex energy system (Cao et al., 2011; Ji et al., 2015). For example, Boloukat and Foroud (2016) presented an interval linear programming model for long-term optimal generation expansion planning of grid-connected microgrid containing various renewable energy generation.

    • Multi-criteria selection of an Air Quality Model configuration based on quantitative and linguistic evaluations

      2014, Expert Systems with Applications
      Citation Excerpt :

      In this respect, expert-based approaches are important in air quality studies. For example: in forecasting air pollution with fuzzy time series methods (Domanska & Wojtylak, 2012); classifying concentrations of criteria pollutants in Mexico (Barrón-Adame, Cortina-Januchs, Vega-Corona, & Andina, 2012); for controlling emissions (Zhou, Huang, & Chan, 2004), identification of geographical distribution of particle pollution (Li & Shue, 2004); or in suggesting a system for air quality management at regional scale based on interval programming with stochastic variables (Cao, Huang, & He, 2011). Another approach consists in the application of deterministic models which solves the conservation equations of different physical and chemical processes (Zhang, Bocquet, Mallet, Seigneur, & Baklanov, 2012).

    View all citing articles on Scopus
    View full text