Objective: This paper aims to advance numerical simulation methodologies for combustion by integrating computational fluid dynamics (CFD) with chemical reactor networks (CRNs). The primary goal is to significantly enhance the predictive accuracy of NO_x emissions in gas turbine combustors. Building on existing numerical simulations of full-parameter and full-scale combustion processes within the combustor, this study provides a method for improved NO_x emission predictions. Methods: The study investigates an automated partitioning/solving method and its programmability within the CRN framework. It evaluates strategies for partitioning the combustion space and establishes general criteria. The combustion reaction zone is a key area for the extensive generation of NO_x. Among the calculations, the division of the recirculation zone is relatively straightforward, with "axial velocity v_a=0" serving as the partition criterion. The envelope area of its isosurface defines the recirculation zone. In addition to the recirculation zone downstream of the central nozzle, a smaller recirculation zone exists at the rear of the Venturi structure at the outlet of the annular zone. However, the partition criteria for the flame front and main flame zones are more complex. This study finds that using the "burnout rate η" and the "average equivalence ratio φ_av before combustion" as criteria is the most reasonable approach. The combustion zone is subdivided in the axial direction using the η criterion. Specifically, the zone is divided into two sections along the axial direction, with partition boundaries corresponding to η=90% and η=99%, respectively. The φ criterion is then applied to partition the cross-section, and each zone is further divided into two sub-zones. The partition boundary φ is the average equivalence ratio (φ_av) of each combustion zone. Results: The findings indicate that under varying loads, combustion modes, and structural conditions, the prediction error for NO_x emissions does not exceed 6.4%. This error is considerably lower than those associated with NO_x predictions made using post-processing models in current commercial softwares. Compared with the commonly used T partition criterion, the η-φ criterion requires fewer reactors and offers higher accuracy. Conclusions: This paper develops an automatic CRN partitioning/solving method and utilizes the XML data format to standardize the storage of input and output data. The method demonstrates good versatility across different combustion chamber types and operating conditions. In addition, the paper proposes a CRN partition strategy and general criterion for gas turbine combustors, specifically the η-φ partition criterion. This criterion reflects the structural characteristics of the combustor, aligns with the principles of basic combustion theory, and has a clear physical meaning. The CRN method based on the η-φ partition criterion is applicable to multiple load conditions, different combustion operating modes, and combustion chamber variations in the local structure, significantly improving the accuracy of NO_x emission predictions. The method can replace the post-processing calculation model for NO_x emissions used in current commercial software, greatly enhancing both calculation efficiency and accuracy. The developed numerical simulation approach provides a more robust tool for research and development related to combustors.