Brewery industry processing plants are made-up of tanks, heat exchangers and columns, which are factory-built, interconnected by thin-walled pipes and bends, assembled together on-site. However, pipes are inaccessible from inside so onsite welding is more difficult to control, leading to poor or inadequate welded joints that can compromise product quality for correctly designed plants. One factor leading to poor welds is pipe geometry as pipes are manufactured to outer diameter (OD), wall thickness and ovality tolerances that affect the alignment of pipe ends. A mathematical algorithm was developed to assess interconnecting pipes and bends from a brewery fabrication site to achieve a minimum of 80% area overlap around the pipe circumference. The 80% was simply a reference point based on a 90% overlap ±10%. 316 L stainless steel pipes of different OD were tested together with 53 mm OD 90°elbows. It was discovered that well-performing welds were difficult to achieve with random orientation of pipes. Better results were achieved by aligning major axes of pipes as there was an increase from 51 to 91% after major axis alignment on welds for straight pipes. There was an increase from 22 to 59% and from 58 to 85% for manufacturers 1 and 2, respectively, for straight pipes-to-bends welds. The higher chances of success for manufacturer 2 were attributed to better OD, wall thickness and ovality control. Whilst the welder is unable to control manufacturer tolerances, they are able to manage orientation of pipes to achieve better welds.

No abstract

This paper discusses the stock size selection problem (Chambers and Dyson, 1976), which is of relevance in the float glass industry. Given a fixed integerN, generally between 2 and 6 (but potentially larger), we find theNbest sizes for intermediate stock from which to cut a roster of orders. An objective function is formulated with the purpose of minimizing wastage, and the problem is phrased as a combinatorial optimization problem involving the selection of columns of a cost matrix. Some bounds and heuristics are developed, and two exact algorithms (depth-first search and branch-and-bound) are applied to the problem, as well as one approximate algorithm (NOMAD). It is found that wastage reduces dramatically asNincreases, but this trend becomes less pronounced for larger values ofN(beyond 6 or 7). For typical values ofN, branch-and-bound is able to find the exact solution within a reasonable amount of time.