With the continuous development of lean and digital manufacturing, the Manufacturing Execution System (MES) has emerged, comprising 11 modules: 1. Production Planning and Scheduling; 2. Operator Management; 3. Production Unit Allocation; 4. Resource Status Management; 5. Product Tracking Management; 6. Quality Management; 7. Document and Drawing Management; 8. Equipment Maintenance Management; 9. Equipment Performance Analysis; 10. Shop Floor Data Acquisition; and 11. Manufacturing Process Management.
MES is the core of a smart factory, integrating product data management in the front-end product design and process definition stages with production data management in the back-end manufacturing stage, to achieve closed-loop collaborative full lifecycle management of product design, production process, and maintenance services.
APS stands for Advanced Planning and Scheduling. Originally a module of MES (Manufacturing Execution System), APS was perhaps developed into a standalone software component due to the critical importance and high technical barrier to production optimization. APS must meet resource constraints, balance various production resources during the production process, provide optimal production scheduling plans at different bottleneck stages, and achieve rapid scheduling and quick response to changes in demand.
The planning and scheduling problem of a single workshop or factory should not be underestimated. From an academic perspective, it is an optimization problem of a large and complex system.
Scheduling is about prioritizing tasks, determining the order of operations. But imagine this: hundreds of machines of varying sizes and hundreds of people are simultaneously performing various tasks. How can they optimally achieve their goals (delivery time, equipment utilization, lowest cost, etc.) under various constraints (equipment capacity, personnel, time, space, materials, etc.) that are dynamically changing?
To illustrate with a simple sorting example: Suppose a computer can process 1,000,000 sequences per second, and we want to build an optimal scheduling system where 9 jobs can be completed in less than a second, 11 jobs in a minute, and given 20 jobs, finding the optimal schedule would take 77,147 years! Real-world scheduling problems involve hundreds of devices and thousands of jobs, demonstrating the extreme complexity of optimizing scheduling for large systems. Of course, people wouldn't resort to brute-force calculations.
For years, system administrators and computer scientists have been searching for a fast method to solve optimization problems in large systems. Systematic methods, heuristics, rule-based methods, simulation methods, genetic algorithms, and so on—each has its own characteristics for specific needs. Some are fast, but the results are not optimal; others converge very slowly and are impractical. Even the academic community once doubted the existence of an optimal solution. Until a few years ago, an American applied mathematician (EYUANSHI) invented the partitioning nested (NP) algorithm, proving that it generates Markov chains, achieves global convergence, and can provide confidence intervals from the optimal solution. This has become a shortcut to solving optimization problems in large and complex systems.
Current Status of the APS Industry
APS has many successful applications in enterprises, especially when integrated with MES modules. Process industries such as steel and chemical engineering have relatively simple planning and scheduling problems, making schedule optimization easier to implement.
In discrete manufacturing, due to the complexity of scheduling problems, almost all current APS systems employ rule-based or heuristic algorithms. The biggest advantage of rule-based or heuristic algorithms is their ability to quickly obtain a feasible scheduling result, but they cannot guarantee an optimal solution or quantify the scheduling result. For simple processes with a small number of orders, the results obtained by any algorithm are roughly similar. However, for complex scheduling problems, the presence or absence of optimization capabilities will lead to significantly different results.
The core of Advanced Planning and Scheduling (APS) lies in the word "advanced." Otherwise, it's just planning and scheduling. Extensive research data shows that scheduling results obtained using rule-based or heuristic methods can differ from optimal scheduling by 30% to 150%. Even with the goal of minimizing delayed orders, an optimized APS, when handling 100 orders, may still have 30 delayed deliveries. Over time, this can result in significant losses for the company. Due to the limitations of optimization algorithm technology, most APS products on the Chinese market currently cannot "calculate" the optimal solution and therefore have to incorporate significant human intervention (e.g., manually creating many rules that may themselves be suboptimal) or ignore certain issues.
In terms of price, APS (Advanced Planning and Scheduling) products range from tens of thousands to hundreds of thousands of yuan. For some small businesses with simple processes, transitioning from manual scheduling to APS scheduling is a step forward and provides auxiliary decision-making support. However, many companies' actual production processes are extremely complex. APS is the most technologically advanced enterprise management software, and its application can improve enterprise production efficiency by several to tens of percent. A truly optimized scheduling APS should be priced at least several hundred thousand yuan. This indicates that the Chinese APS market and technology are still immature.
What kind of APS is needed?
APS is enterprise management software with highly intelligent production planning and scheduling capabilities. It can maximize the utilization of enterprise resources and find the optimal scheduling results in complex production processes with multiple tasks and numerous constraints. The core of APS is its optimization engine, which finds the best possible outcome.
In actual production, discrete manufacturing enterprises (small batches, multiple varieties, and highly variable orders) face complex tasks, resources, and processes with numerous constraints, and the process is entirely dynamic. What these enterprises need is to generate an optimal scheduling plan within a tolerable timeframe (e.g., 10 minutes). Furthermore, the degree of optimization in this scheduling plan should be measurable and quantifiable, and its future impact should be predictable (e.g., the situation three months from now).
APS must be highly adaptable. A company's actual operations may encounter different requirements at different times. For example, sometimes the shortest delivery time is required, sometimes the best equipment utilization rate is required, sometimes the minimum inventory is required, and sometimes there are urgent orders. APS must be able to easily meet the needs of the company at different times.
The user interface of APS must conform to the mindset and scheduling habits of enterprise dispatchers. It is unacceptable for enterprise users to undergo extremely complex training to adapt to computer requirements.
MES and APS are integrated
APS and MES overlap in their production scheduling functions. However, the current trend is for APS and MES to be integrated to achieve four closed loops:
1. Closed loop of demand forecasting and order commitment.
2. Closed-loop planning and production scheduling.
3. Production scheduling and execution closed loop.
4. A closed loop of order commitment, order fulfillment, and delivery. This forms a self-governing system with self-feedback and self-decision-making capabilities.
Disclaimer: This article is a reprint. If there are any copyright issues, please contact us promptly for deletion (QQ: 2737591964). We apologize for any inconvenience.
