1. What are the applications of the rectangular cutting problem (RCP)?
The rectangular cutting problem (RCP) has various applications in real-life industrial scenarios. It is essential in industries such as glass, plastic, and metal, where large rectangular stock objects need to be cut into smaller rectangular items to produce parts of end products. This process helps minimize waste and optimize resource utilization. Additionally, the RCP finds relevance in the shipping and transportation industries, where rectangular packages need to be arranged on a two-dimensional surface to minimize empty space. Variants of the RCP arise when additional constraints are imposed, such as technological constraints, guillotine cuts, limited number of cuts, and fixing items' orientation. The RCP is a crucial component of more complex packing and cutting problems, making it a vital area of research and optimization.
read more
2. What is the focus of the research on oriented guillotine stage-unrestricted 2DSKP and 2DSLOPP?
The focus of the research on oriented guillotine stage-unrestricted 2DSKP and 2DSLOPP is on the cutting pattern of items with a fixed orientation and obtainable only via guillotine edge-to-edge cuts. The research explores the state of the art solution techniques, including the first tree search exact method for 2DSLOPP proposed by Christofides and Whitlock (1977) and revised by Velasco and Uchoa (2019). The research also investigates the 'cut-and-plate' principle, which is commonly used in guillotine cutting approaches. Additionally, the research introduces a novel CP model to the 2DSLOPP by modeling the tree of guillotine cuts, with cuts applied first to the original plate in the root node, and then recursively in each internal node to the resulting sub-plates. This model takes advantage of cumulative scheduling relaxations intrinsic to 2D packing/cutting problems, making it efficient and tight.
read more
3. What is the cutting process in Constraint Programming Model?
The cutting process in Constraint Programming Model involves deciding on the necessary cuts, their types, and how they should split the plate to allocate space for the items during the search process. In guillotine cutting, a single item or a rectangular region with a set of packed items can be obtained from the original stock plate (W, H) or from its sub-plate (a parent region) via a sequence of exactly two cuts: One cut must be vertical and the other horizontal. There are two options to cut a parent region r*: first, it can be cut horizontally, resulting in two horizontal strips (r' and r''), and then a vertical cut is necessary to separate r from r''. Alternatively, it can be cut vertically, resulting in two vertical strips (r'' and r'), and then a horizontal cut is needed to separate r and r''. Each pattern contains exactly one waste region (r'' or r'), and the model assumes that the waste region must remain empty. The cutting process replicates the binary tree structure of guillotine cutting, where obtaining n items requires building exactly m = n - 1 patterns, with each internal node of the tree being either an a or b pattern and leafs representing the items. The modeling approach is generic and can be translated to other solution techniques like MILP.
read more
4. What are the decision variables used in the CP model for guillotine cutting and how are they implemented?
The decision variables in the CP model for guillotine cutting are categorized into four groups. The first group consists of binary variables e {0, 1}^n x m, where e ir = 1 if item i I is assigned to region r R, and 0 otherwise. The second group includes integer variables u N n 0, representing the number of copies of each cut item, with domain D(u i ) = {0, . . ., d i }. The third group comprises binary variables a {0, 1}^m x 3m, where a pr = 1 if pattern p P is assigned to region r R p R, and 0 otherwise. The fourth group consists of auxiliary binary variables z {0, 1}^m, indicating when a pattern is assigned to a region, and sets g ' {0, 1}^m-1 and g '' {0, 1}^m-1, signaling when a pattern is assigned to the top or right region of its predecessor. Additionally, binary variables v {0, 1}^m and integer variables W N 3m 0 and H N 3m 0 represent the layout and sizes of the regions, respectively. These decision variables are implemented to model the recursive nature of guillotine cutting, assign items to regions, and determine the layout and sizes of the regions in the cutting pattern.
read more