For the early cost estimation of a new process (which targets a 20% accuracy), it is important to know the costs for the waste treatment. This is mainly critical in fine chemistry, where waste treatment costs can reach 10% of the total costs, or even more.
But these ‘ecological’ costs are hard to estimate, due to several factors:
The amount is chiefly significant as the capacities of the various waste treatment facilities are limited, and a big stream containing waste water, for instance, could have to be burnt because your Sewage treatment Plant (STP) is full. This factor is not so important in fine chemistry, as the amounts are normally limited.
As a matter of fact, the pre-treatments can broadly vary according to the composition of the waste. As the required treatments are of course different if your limiting factor is chloride or heavy metals, even the decision between burning or treating in a STP a stream of waste is a multi-criteria's decision. Some polluting-agent could just kill the bacterias and disable the STP, and even a 99% water stream could have to be burnt.
As we target on early cost estimation, most physical or chemical properties are not well known, which complicates the making of a simple model depending of such properties.
Anyway, it is very difficult to find simple laws (even statistical) to explain a price in function of the observables (like TOC, %-water, …). Indeed there are lot’s of interactions between variables, and as it will be shown further in this work, such simple ‘regression’-like laws are impossible to draw for an hypothetical ‘general case’.
More to the point, we could say that there are no general cases.
Nowadays, the early cost estimation of the waste treatment is very straightforward. It principally consists of taking an average price per unit (e.g. m3, kg, …) of waste, and multiply it by the amount of waste. Usually, the recycling is independently handled. This method can of course not handle special cases, surprises can arise when the final process is launched.
Therefore, a software able to handle such a case-by-case study with ‘scenarios’ is fully needed, to better the cost estimations. It would be a useful add-on to P. Dimmer's software of cost estimation.
The aim of this work is to write a software with the following features:
The main challenge of this work is to integrate in one software uncertainty and scenario-handling. The best way to implement these two features will be to use Monte-Carlo’s method, as this method does calculate uncertainty ‘automatically’ if you change (more or less) every number in a matrix, and compute in parallel many simulations.
1. Using the scenario-handling method developped in the chapter ‘Waste Treatment’ and Monte-Carlo’s method to implement uncertainties, the waste treatment simulation was programmed.
The objectives were to get a software with the following features:
The software is flexible enough to let the user modify anything easily, including the whole structure of the program. It is fully integrated in P. Dimmer’s main simulation program, although it can be used by itself.
A wide range of options allows to get the exact behaviour desired, with or without a human interaction.
Therefore the objectives are fully fullfilled.
2. The sensitivity analsis conducted on many samples have shown that no general conclusions are possible concerning the relative significance of the different parameters.
Matter-of-factly, too many parameters and a complex behaviour prevent any general conclusion to be drawn. A case by case study must be conducted for every different waste stream.
Therefore, a software able to handle any stream like ours is fully needed.