Feature - Winning the ‘scrap lottery’
Plastics manufacturers had a problem.
The typical production line produces hundreds of distinct plastic products of varying weight, shape and size. A single desktop computer can manage the process of deciding which combinations of products should be married to which production lines, but the task increases exponentially when trying to predict how much waste “scrap” is generated.
This is because predicting the amount of scrap is a highly stochastic process, meaning it could have many probable outcomes. Industry insiders describe it as a ‘scrap lottery’, in which the only way to know an outcome is to ‘spin the balls,’ or in this case create the plastic.
The problem was particularly acute at Vitopel, an Argentine company that is one of the world’s five largest plastic manufacturers. It produces biaxially-oriented polypropylene, or BOPP — an important material for making the clear plastic found in some plastic bags, food packaging and laminated goods. (Polypropylene is a thermoplastic polymer of large, covalent and chemically-bonded molecules, which means it is heat resistant, durable and flexible. Making it biaxially oriented makes the polypropylene crystal-clear — essential for some goods.)
Usually there is a lot of waste in manufacturing BOPP, where large volumes of raw material are melted, cooled, molded and cut into specific shapes. Due to the high volume — 300 metric tons per day, or 109,500 metric tons a year — much energy and material is needlessly wasted. Not only is this bad for the environment; it also makes holes in the financial pockets of manufacturers.
To make large-scale plastic manufacturing more efficient, less wasteful — and more profitable — Diego Carvalho and Rafael Barbastefano of the Brazilian Federal Center of Engineering Studies turned to an Industry@Grid application on EELA-2 (‘E-science grid facility for Europe and Latin America,’ soon to become GISELA, or ‘Grid Initiatives for e-Science virtual communities in Europe and Latin America,’ as of 1 September). They developed computer models that use computer algorithms based upon sampling random numbers to calculate results; such ‘Monte Carlo’ simulations produced numerous virtual scenarios of potential scrap.
The tools enabled them to understand how all seven of Vitopel’s production plants worked as a single system; by modeling approximately 50,000 scenarios on the grid, they could predict a cost-effective solution that reduced plastic waste by 2,070 metric tons per year, or 2% of the total manufacturing output of the Vitopel plants. That’s the equivalent of 147 million ‘Ruffles’ potato chip bags per year, minus the contents inside.
Reducing the amount of plastic scrap right at the source — on the factory floor — is key, Carvalho said. Theoretically, every bit of scrap could be sold to another company to be recycled, but the recycling process itself consumes energy: the plastic scrap has to be transported, re-melted and re-manufactured. “By reducing scrap in the industry, we are reducing our CO2 footprint in this process,” he commented.
In the Northern Hemisphere, the British Plastics Federation said it was in favor of such uses of computer-aided design. “We are very interested indeed in minimizing the amount of used plastics going to landfills, and as a UK industry we have a target of zero waste in landfill by 2020.”
—Adrian Giordani, for iSGTW. More about GISELA will be found at the September 16 session on regional initiatives at the EGI Technical Forum.