Abstract
A number of methodologies have been presented in the literature to meet the demand for environmental comparison of process design alternatives. However, the application of these methodologies in practice has been limited by subjectivity, uncertainty and resource requirements. In this paper, a multi-compartment model, using readily available data and associated assumptions, is presented as a potential tool for environmental comparison of process design alternatives. The suitability of the model to provide a discriminatory basis for comparison is discussed in terms of trends, quantitative uncertainty and limitations.

Introduction
A significant need has been identified1-3 for a method, based on scientific principles, for environmental comparison of process emissions during different stages of design and for compliance with management standards or legislative requirements. In response to this identified need, various regional and global comparison methods have been developed which may be used as add-ons to process design simulation packages.

Regional and global environmental comparison methods are complementary. Global methods provide an indication of environmental impact in terms of ozone depletion, greenhouse effects, etc. Regional methods provide a comparison based on chemical exposure and effect. A methodology, based on scientific principles, is presented as a potential tool for the regional comparison of process emissions during different stages of design.

The suitability of a methodology for comparison can be measured in terms of uncertainty and its ability to provide a discriminatory basis. Many chemical assessment methods are limited for process comparison by their inability to differentiate between the relative environmental behaviour of effluents, intrinsic uncertainty and variation in data4. These limitations are not typically quantified in existing methods. The suitability of the presented methodology to provide a limited resource comparison basis is considered in this paper. Uncertainty is evaluated in terms of intrinsic assumptions, variation in partitioning coefficients and input data.

Relative Impact Potentials (RIPs) are used to provide a regional basis for comparison of process emissions. An RIP is defined as the relative impact, through bioaccumulation, which long term exposure to a process effluent could have on a group of species in a specified compartment, or medium. Long-term exposure concentration may be predicted using dispersion or multi-compartment modelling. 

Typically, dispersion models do not account for compartment transfer or transformation products. The use of a multi-compartment model to provide a limited resource basis for RIP is considered in this paper. To render it readily useable, the model is based on generally available data and associated assumptions. The multi-compartment model is limited to liquid and gaseous releases which may partition between bed sediment, the water column and the troposphere. Application of the model is illustrated with a number of chemicals which represent potential emissions from process plants, e.g. the production of allyl chloride and the HDA process.

Conclusion
The concept of using a multi-compartment model for the long term, regional comparison of process design effluents seems to be very promising. The case studies presented demonstrated that a release in one compartment may have a potential for regional impacts in others which are not typically accounted for using dispersion models or chemical assessment methods. Furthermore, application of the model indicated that Relative Impact Potential (RIP) may be comparable for two effluent components in a given compartment with significantly different concentrations due to differences in residence time.

Assuming equilibrium between compartments, the multi-compartment model results and uncertainty may be predicted using limiting case equations. For specific compartment volume ratios, uncertainty associated with the model will be in a range which corresponds to variance in half-life to that of the product of the half-life with a relevant compartment transformation rate ratio. Uncertainty is significantly magnified when compartment volume ratios are also considered.

It was demonstrated that the multi-compartment model presented may provide a discriminatory basis for the comparison of process effluents for a given 'idealised' situation, i.e. for specific compartment volume ratios. However, estimates of uncertainty indicate that while such models provide an indication of compartments which may be affected, they are not applicable for fine distinction amongst effluents which exhibit similar behaviour unless accurate input data are available.

Further research is required to specify appropriate values for relative compartment volumes based on field data for specific chemicals, to extend the model for inorganic chemicals and to evaluate the inherent uncertainty of the assumption of equilibrium. However it should be noted that data is not readily available for non-equilibrium models and the assumption of equilibrium may be justifiable due to the absence of other methods with greater certainty.