Plant Combined Simulation for Compost Production starting from Sewage Sludge
From Proceeding of Summer Computer Simulation Conference,SCSC95, Ottawa, July 24-26 1994
Pietro Giribone, Agostino G.Bruzzone, P. Cereda, R.F. Dell'Acqua, Domenico Rivarolo

---

THE COMPOST PRODUCTION ENVIRONMENT

The entire world is beleaguered by the problems regarding waste treatment since social needs often conflict with a lack of economic advantages. Therefore, these two requirements must be reconciled to reduce the service cost, making it as remunerative as possible.
In one interesting method, waste is treated to obtain compost (fertilizer) that can be sold at low cost to create new terrain and landscapes (e.g. golf courses) and as fertilizer for agricultural and floricultural applications.
The Italian scenario is particularly complex. In fact, many of the plants which are currently operating utilize processes which are not very economical nor effective in guaranteeing the quality of the final product. This situation creates mistrust and makes it difficult for the market to accept new and modern plants.
The other major problem concerns the aleatory nature of the prices of the raw material (conditioned sludge, etc.) and the finished product; therefore operative and commercial management becomes an extremely complex process.
Moreover, many attempts are being made to solve this problem. In fact, efforts are being made in schools to educate new floriculturists to use compost as fertilizer, new plants are focusing on guaranteeing the quality of the product, consortiums are being created to market fertilizer, etc.
Another possible action, aimed at improving technical-economic management, involves the use of the simulation. In fact, detailed models can be used to meet the many different design and managerial requirements, while also providing important information.
The authors are involved with a Consortium dealing with waste treatment activities. This Consortium operates over a vast area and is currently completing the start-up phase for this type of plant using an innovative approach (maturation in automatic compost lines). The authors also work with an Academic Institute that applies these methods in the plant engineering field.
This paper has set the objective to evaluate the plant during start-up and to build a simulation model that can be re-used in many different applications.
The process in question was reproduced by referring to an Object-Oriented Design simulator. Event logic was used, but since this involves a combined process, it was necessary to implement continuous/discrete hybrid operations in the model.
In fact, some logic phases of the process are evidently controlled by events such as picking up and placing the material on the maturation aisles using bulldozers, turning the material over, entrance the weighing area and unloading trucks at the plant which transport raw materials. Instead, other process phases are continuous, for example maturation using forced ventilation or in aisles, and depend on endogenous (characteristics of the layers in the compost piles) and exogenous variables (humidity and temperature conditions). Other times the continuous component is simply related to the mixing and blending of different materials or transporting material by conveyor belt.
Given the uncertainty about the measurement, but especially about the relations which link the continuous variables, it was decided to implement fuzzy rules to correlate such data. It was possible to immediate action within the model to modify these relations and/or to replace them with other analytical types.
To correctly reproduce this operation it was decided to build the model using a multi-purpose language, C++, which guarantees maximum flexibility and combines continuous parts with others based on events, whenever required.
The personal computer was selected as the hardware platform. An important factor that led to this choice is the availability of this type of computer in the plant (in fact, the plant is equipped with a remote control system run by personal computers).
The study continued by illustrating the construction logic of this combined discrete/continuos simulator, implementation of the relations between the continuous variables with fuzzy logic and the statistical validation and calibration phase of the existing situation which is currently in the advanced start-up phase.
The initial objectives considered were linked to verification of the production capacity and the quality of the product under real operating conditions, thus considering faults, malfunctions, delivery delays, etc.
To do this methods were used which were developed for studies relative to similar plants created by the research team. Therefore, meta-models were built which can define a relation between the plant and production characteristics, correctly identifying the uncertainty band relative to the results obtained.
The second objective was to evaluate factors relative to operating management in relation to the production objectives (quantities, use of resources, costs). Also in this case the model can predict the behavior of the plant, therefore highlighting possible lines of action.
To this regard the simulation detail level must be enhanced in order to consider the problem of plant internal handling operations. In fact, these handling operations are performed by bulldozers and trucks. These devices must perform a large number of tasks (transfering material, turning over piles, weighing operations, etc.) and the number of such devices is rather low (2 bulldozers and 2 trucks) in the current configuration. Therefore, it is was necessary to follow the movement of these vehicles in the plant, step by step, to determine whether an increase in the mechanical bulldozers would really have made it possible to increase productivity or if the interference between the devices would turn out to be counterproductive beyond a certain limit.
This aspect made it possible to create a simulation model with very different components: event machines, logic events, handling vehicles while considering interference, continuous processes. Therefore, the simulator processing speed is not very high, but still satisfactory. In fact, about thirty minutes is needed to perform a test over one year.
Having completed the analysis, this model still remains a powerful tool that can be used along with specific plant management and new projects can be evaluated by the consortium for similar plants. At the same time it is currently being used as a logic base to develop training equipment for plant managers and may be used to create an on-site training center for compost plant handlers.