Production data from a small bakery manufacturing line

Finding an optimal production schedule is a crucial part in improving manufacturing efficiency. Data collected from a real production line serves as the process's starting point. It provides details on the products, the processing procedures, how long they take, and the machines needed to carry out these steps. The production schedule is simulated using a flow shop model. The manufacturing environment determines the sort of flow shop model. The simplest model is a permutation flow shop, where all products are processed by machines in the same order. The majority of production systems in practice contain numerous distinctive constraints that make models quite complex. The "no-wait" flow shop model refers to a particular type of flow shop model where no waiting period is permitted between two successive stages of a product.
A hybrid no-wait flow shop scheduling model can be used to explain the production environment in bakeries. Multiple machines are accessible here for a single processing stage, such as oven A or oven B for baking a product. Products' processing stages may not be organized in the same order depending on the recipe. Occasionally, the same dough is used to make several different bakery products. In these situations, bakers divide the dough after a few preliminary processing steps and carry out the remaining steps independently in accordance with the product recipes. Baking ovens need a lot of energy, and using them even when only partially occupied wastes energy. Therefore, in order to conserve energy, baking of a number of products may occasionally be conducted together that followed separate processing paths. The products are arranged into groups based on their inherent interdependence, such as baking or beginning steps being performed concurrently. At any stage of the production process, one product from this group may be processed. However, it influences the start time for the remaining products in the same group. Small and medium-sized bakeries rely on manual operation for many processing tasks. Therefore, manual tasks must be distributed among staff in order to continue the production without interruption. Some processing tasks need the use of two machines, or one machine and one employee at the same time. To prevent ingredient mixing, a machine may need to be cleaned after processing a product. For instance, after kneading dough containing oil seeds, a kneader is cleaned.

寻找最优生产调度方案是提升制造效率的关键环节。从真实生产线采集的实测数据作为本研究流程的起始基础，其详细涵盖了产品信息、加工工序、各工序耗时以及完成各步骤所需的生产设备。本研究采用流水车间模型（flow shop model）对生产调度方案进行仿真模拟，制造生产环境决定了所采用的流水车间模型类型。最为基础的模型为排列流水车间（permutation flow shop），即所有产品均按照完全一致的顺序经由各台设备完成全部加工工序。实际生产中的多数系统存在诸多独特约束条件，导致模型复杂度显著提升。无等待流水车间模型（no-wait flow shop model）特指一类不允许产品在相邻两个加工阶段之间存在任何等待时长的流水车间模型。
混合无等待流水车间调度模型（hybrid no-wait flow shop scheduling model）可用于阐释烘焙工厂的生产环境：单个加工阶段可配备多台生产设备，例如用于产品烘烤的烤箱A与烤箱B。根据产品配方的差异，各产品的加工工序顺序可能并不统一。部分烘焙产品可采用同种面团进行制作，在此类场景中，操作人员会在完成若干预处理步骤后对面团进行分流处理，随后依据各产品的配方独立完成后续的加工工序。烘烤烤箱的能耗极高，即便仅处于半负载状态运行也会造成严重的能源浪费，因此为节约能源，有时会将加工路径各异的多款产品集中安排烘烤工序。操作人员会基于产品间的内在关联性（例如可同步开展烘烤或起始工序的产品）将其划分为若干批次，同一批次内的产品可在生产流程的任一加工阶段依次完成加工，但任一产品的开工时间都会对批次内其余产品的开工时刻产生影响。中小型烘焙工厂的多数加工任务依赖人工操作完成，因此需将人工任务合理分配至各操作人员，以保障生产持续不间断地进行。部分加工任务需同时使用两台设备，或是一台设备搭配一名操作人员协同完成。为避免不同产品的原料混串，设备在完成某款产品的加工后需进行清洁作业，例如在揉制含油料种子的面团后，需对揉面机进行彻底清洁。