Abstract
We study the management of product transitions in a semiconductor manufacturing firm that requires the coordination of resource allocation decisions by multiple, autonomous Product Divisions using a multi-follower bilevel model to capture the hierarchical and decentralized nature of this decision process. Corporate management, acting as the leader, seeks to maximize the firm's total profit over a finite horizon. The followers consist of multiple Product Divisions that must share manufacturing and engineering resources to develop, produce and sell products in the market. Each Product Division needs engineering capacity to develop new products, and factory capacity to produce products for sale while also producing the prototypes and samples needed for the product development process. We model this interdependency between Product Divisions as a generalized Nash equilibrium problem at the lower level and propose a reformulation where Corporate Management acts as the leader to coordinate the resource allocation decisions. We then derive an equivalent single-level reformulation and develop a cut-and-column generation algorithm. Extensive computational experiments evaluate the performance of the algorithm and provide managerial insights on how key parameters and the distribution of decision authority affect system performance.
Introduction: The Challenge of Product Transitions
In high-tech industries like semiconductor manufacturing, "product transitions"—the process of phasing out old products while ramping up new ones—are high-stakes periods. If a firm transitions too slowly, it loses market share; too fast, and it wastes capacity or fails to meet existing demand.
The challenge is that large firms are rarely monolithic. They are composed of autonomous Product Divisions (PDs) that act as independent profit centers. These divisions must compete for two critical, shared resources:
- Factory Capacity: Used for both high-volume production of current goods and the fabrication of prototypes for new ones.
- Engineering Resources: Needed to design and develop the next generation of products.
Existing models often treat the firm as a single entity with perfect information. In reality, Corporate Management (the "Leader") only influences divisions through budget and capacity limits, while the divisions (the "Followers") make operational choices based on their own local constraints and objectives. This creates a hierarchical, decentralized optimization problem that is notoriously difficult to solve.
Methodology: Multi-follower Bilevel Programming
The authors model this environment using a Multi-follower Bilevel Programming framework to capture the interaction between the corporate leader and divisional followers.
- The Upper Level (Leader): Corporate Management (CORP) seeks to maximize total firm profit by allocating budgets and setting overall capacity limits for production and engineering.
- The Lower Level (Followers): Multiple PDs compete for these resources. Because the divisions' decisions are interdependent (one division’s resource use limits another’s), the lower level is modeled as a Generalized Nash Equilibrium Problem (GNEP).
To solve this, the authors reformulated the GNEP into a single-level Mixed-Integer Program (MIP) and developed a Cut-and-Column Generation (CCG) Algorithm. This algorithm iteratively adds only the most relevant constraints ("cuts") and variables ("columns") to find an exact solution efficiently despite the astronomical number of possible budget allocations.
Decision Framework
The complexity of the PDs' decisions is captured through several integer and binary variables representing production, inventory, and development status across different time periods.
Experimental Setup and Results
The study uses randomly generated test instances designed to mimic real-world semiconductor industry conditions, modeling "product rollovers" where demand for old products diminishes while demand for new products ramps up. The experiments tested 36 instance classes with varying numbers of time periods (8–16), divisions (2–8), and products.
Key Insights
- The "Competition Benefit": Counter-intuitively, increasing the number of Product Divisions can actually improve corporate profit. Competition forces the system to streamline budget allocation and incentivizes divisions to find more cost-saving development plans.
- Budget Saturation: There is a clear "diminishing returns" point for budgets. Beyond a certain threshold (a budget fraction of ~0.3 in tests), providing more money to PDs does not improve the firm's objective because physical constraints of factory and engineering capacity become the true bottlenecks.
- Complexity Drivers: Computational difficulty is driven more by the number of divisions and new products than by the length of the planning horizon. The CCG algorithm successfully solved 132 out of 180 test instances within a 3-hour time limit.
Conclusion and Future Work
The model provides a framework for "what-if" analysis, allowing corporate leaders to see how changing a budget or adding a new product line will ripple through decentralized divisions. Future work aims to address information asymmetry (where divisions might hide data), endogenous deadlines (making development timelines a decision variable), and uncertainty (incorporating stochastic demand or yield).
Summary updated