"Material Requirements Planning (MRP) is a plan for the enterprise to formulate material procurement, production, and allocation based on the needs of production, sales, procurement, and forecasting. Whether the material requirements plan is accurate or not can be said to affect the whole body. Initial demand forecasting Minor deviations in links will extend and amplify step by step along the supply chain, causing risks such as material shortages and dead materials for enterprises, and then bringing an unpredictable "butterfly effect" to the entire supply chain.
At the moment when the mass customization production mode is becoming more and more common, factors such as the huge amount of calculation brought about by the intricate material substitution relationship, the optimization problems brought about by multi-objective and multi-constraints, and the high requirements of enterprises for refined management often make the traditional Material Requirements Planning tools can't do enough.
Lenovo, as a high-tech manufacturing company selected in the "Gartner Global Supply Chain TOP25" in 2022, has 35 manufacturing bases around the world, provides products and services to more than 180 markets, and has established cooperative relationships with more than 5,000 suppliers around the world. The business characteristics of large-scale production, large customization requirements, and multi-factory collaboration drive Lenovo to seek more efficient and flexible MRP solutions in material demand planning, so as to reduce the potential risks brought about by the "butterfly effect" in the supply chain from the initial stage .
Recently, Dr. Ouyang Wenli, manager and senior researcher of the Intelligent Supply Chain Team of the Artificial Intelligence Laboratory of Lenovo Research Institute, was invited to conduct a live broadcast on the theme of "Practice of Operations Research in Lenovo's Supply Chain Material Demand Planning", sharing Lenovo's MRP in material demand planning. Technical accumulation and practical experience in the field.
The practice of operations research in Lenovo's supply chain material demand planning
01 Traditional MRP faces unsolved problems
The traditional MRP solution method is usually based on the rules of business experts, such as deduction of materials and calculation of shortage of materials according to the priority of orders and materials. However, more material BOM levels, more complex material substitution relationships, and stricter constraints in actual production scenarios seriously affect the efficiency and accuracy of the solution of material requirements planning. Traditional MRP faces great challenges:
● Agile response is difficult. In the VUCA era, frequent and rapid evaluation of material demand planning is required in the planning business to enhance the resilience of the supply chain. However, traditional MRP is difficult to quickly solve and respond to the dismantling calculation of BOM and the intricate substitution relationship between multiple materials.
● It is difficult to calculate multi-objective and multi-constraint. For multi-objective and multi-constrained material requirements planning such as inventory level, procurement control, and lowest cost, the heuristic solution commonly used in traditional methods cannot obtain global optimal results, and there is still room for optimization in cost control.
● Refined inventory management. Inventory management is the top priority in the supply chain, and how to arrange procurement and consumption of inventory is the core of inventory management, how to buy when needed, how much to buy, to ensure that demand is met while reducing the risk of inventory backlog is the key to inventory management Higher requirements for refinement.
02 The solution to intelligent decision-making
The bill of materials (BOM) is the most important basic data in the MRP system, and the complexity of its structure directly affects the difficulty of solving. Taking electronic products as an example, there are usually dozens of levels in the BOM of finished products, and there are dozens to hundreds of different material types in each level. Especially when there are alternative materials in the BOM, the assembly method of the final product can reach an astronomical figure, which is also a key factor restricting the solution ability and efficiency of the MRP system. Lenovo MRP realizes overall decision-making and comprehensively improves supply chain efficiency by optimizing the solution model.
● Efficient solution: Lenovo MRP greatly simplifies the complexity of the model by optimizing the model. Taking a simple BOM as an example, the finished product is assembled from 15 materials, and each material has 3 alternative materials. If the traditional MRP mixed integer programming (MIP) model is used to solve the problem, the number of assembly methods can reach one billion above. By establishing virtual material nodes between materials, the increase in the number of assembly methods is transformed from a geometric progression to a linear cumulative relationship. After optimizing the MIP model, the number of assembly methods for the finished product is reduced from more than one billion to 6750. Reduced the difficulty of solving and achieved minute-level response;
● Overall decision-making: Lenovo MRP matches according to the order priority, and when consuming materials, it prioritizes the consumption of alternative materials with poor versatility. Through global optimization solutions, it solves the problem of traditional MRP processing supply and demand one by one, each link is isolated, and segmented output Insufficient results and other aspects, through overall decision-making, to achieve refined management of the supply chain;
● Scenario optimization: The incomplete substitution relationship of materials is a pain point in the industry. Traditional methods are difficult to develop and cannot achieve global optimization. Lenovo MRP realizes the global optimal solution of the incomplete replacement group by optimizing the material demand scenario, and at the same time takes planning and inventory allocation into consideration, improving inventory utilization and reducing the overall inventory level.
In addition, in response to the low reuse rate and long function iteration cycle of the traditional MRP model, Lenovo MRP optimized the modeling method based on the full OR algorithm, built a generalized product architecture, and improved the versatility and scalability of the product.
03 Practice shows results
At present, Lenovo MRP has been applied in multiple scenarios in Lenovo's supply chain. While ensuring the accuracy of the results, the advantages of global optimization solutions have shown remarkable results in complex demand scenarios.
● Without considering alternative materials, multi-factory joint planning, and optimization in different time periods, the matching results of supply and demand, material inventory consumption, and new purchase quantities output by Lenovo MRP are consistent with traditional MRP systems;
● Taking the tablet business as an example, in the case of incomplete substitute materials, based on the incomplete substitution relationship of materials and different consumption logics, Lenovo MRP solves through global optimization and consumes more inventory. The number of issued purchases was significantly reduced from 4,643 to 203;
● In the face of multi-factory joint planning that cannot be directly supported by traditional MRP, Lenovo MRP can directly generate decision-making suggestions through global optimization algorithms, realize global material allocation, and reduce new procurement materials.
The successful practice of Lenovo's MRP means that there is a more effective solution to the manufacturing supply chain problem of making production and purchasing plans. So far, the blueprint of Lenovo's supply chain intelligent decision-making solution has also been rolled out: take supply chain decision-making as the entry point, and use intelligence to connect manufacturing and the future.
(Follow the official account ML OR intelligent decision-making, share more dry goods, welcome to communicate~)