Sparse-dynamics-based combustion chamber reaction simulation acceleration method for heavy-duty gas turbine

Abstract

The application discloses a heavy gas turbine combustor reaction simulation acceleration method based on sparse dynamics, wherein the method comprises the following steps: dynamically adjusting the intensity of the heat release source term based on the difference between the CFD unit temperature and the corresponding reactor temperature, and eliminating the error caused by the uniform distribution assumption of the heat release through feedback adjustment; constructing a sparse reaction network containing multiple reactor nodes based on the Cantera toolkit; based on the sparse reaction network, real-time evaluation of the reaction intensity in the CFD area where the reactor unit is located is carried out, only for the high active area, the complete PSR reactor unit is activated to perform complete chemical calculation, for the low active and non-active area, the reduced mechanism or the frozen treatment is adopted; dynamically distributing the activated reactor unit to independent computing threads and GPU task blocks, automatically performing cross-scale load balancing adjustment according to the time step, and synchronously maintaining the two-way boundary coupling between the reactor and the CFD, so as to ensure the dynamic balance of communication efficiency and calculation stability.

Description

Technical Field

[0001] This invention relates to the fields of combustion numerical simulation and CFD coupling technology, and in particular to an accelerated simulation method for the combustion chamber reaction of a heavy-duty gas turbine based on sparse dynamics. Background Technology

[0002] Traditional gas turbine combustor reaction simulation relies on detailed reaction mechanisms coupled across the entire region, often requiring the integral of the heat release rate and the solution of the reaction source terms across the entire field. This results in huge computational overhead and low time efficiency. In particular, when dealing with multi-region non-uniform distribution and transient combustion conditions, the solution efficiency becomes a key bottleneck limiting engineering simulation.

[0003] In existing heavy-duty gas turbine combustion simulations, numerical simulation methods based on full-field detailed reaction mechanism coupling are widely used. While these methods offer high physical fidelity, they generally face the following technical bottlenecks when dealing with complex multi-zone combustion structures, unsteady operating conditions, and three-dimensional disturbance evolution processes:

[0004] (1) High computational cost and serious resource consumption: Most current simulations use full-field fine mesh combined with global integration, which requires global calculation of reaction rate, source term and heat release rate of the entire flow field at each time step, resulting in extremely long simulation time, which is not suitable for parameter scanning, structural optimization or real-time prediction in engineering scenarios.

[0005] (2) Unable to adapt to non-uniform distribution and local reactivity characteristics: The fuel distribution, temperature field and chemical reaction in the actual combustion chamber exhibit strong spatial heterogeneity. Existing methods fail to adapt the activation region solution strategy according to the reaction intensity, resulting in high-overhead reaction calculations in low-reaction regions, wasting a lot of resources.

[0006] (3) Sparse dynamics models lack dynamic self-adjustment capability: Most existing sparse reaction networks (such as CKN and DRG-based methods) are based on static reduction mechanisms. The network structure remains unchanged during the simulation process, making it difficult to respond to the evolution of the flow state and the migration of hotspot regions, resulting in the rigidity and non-adaptability of the reaction network. Summary of the Invention

[0007] The main objective of this invention is to provide a method for simulating and accelerating the reaction of a heavy-duty gas turbine combustor based on sparse dynamics.

[0008] Another objective of this invention is to propose a heavy-duty gas turbine combustion chamber reaction simulation acceleration device based on sparse dynamics.

[0009] The third objective of this invention is to provide a computer device.

[0010] A fourth objective of this invention is to provide a non-transitory computer-readable storage medium.

[0011] To achieve the above objectives, a first aspect of the present invention proposes a method for simulating and accelerating the reaction of a heavy-duty gas turbine combustor based on sparse dynamics, comprising: The intensity of the heat release source term is dynamically adjusted based on the difference between the temperature of the CFD unit and the temperature of the corresponding reactor, and the error caused by the assumption of uniform heat release distribution is eliminated through feedback adjustment. Based on the Cantera toolkit, a sparse reaction network containing multiple reactor nodes is constructed. Based on the sparsified reaction network, the reaction intensity within the CFD region where the reactor unit is located is evaluated in real time. Only the fully active region is activated to perform full chemical calculations. For the low-activity and inactive regions, a reduction mechanism or freezing treatment is adopted. Activated reactor units are dynamically allocated to independent computing threads and GPU task blocks, and cross-scale load balancing is automatically adjusted according to time steps. The bidirectional boundary coupling between the reactor and the CFD is maintained synchronously to ensure a dynamic balance between communication efficiency and computational stability.

[0012] In this embodiment of the invention, the dynamic adjustment of the intensity of the heat release source term based on the difference between the CFD unit temperature and the corresponding reactor temperature, and the elimination of errors caused by the assumption of uniform heat release distribution through feedback adjustment, includes: In the initial iteration, it is assumed that the local unit has no reactive heat source; In subsequent iterations, the source term size is dynamically adjusted based on the temperature difference between the CFD unit and the reactor it is connected to. The formula defining the source term is expressed as:

[0013] in For CFD cell temperature, For reactor temperature, This is the convergence control factor.

[0014] The sparsification reaction network described in this embodiment of the invention includes: The PSR module is used to simulate local mixing and reaction processes, serving as the basic unit for mesh and reaction coupling. The MFC module is used to control the mass exchange between reactors and to build the topological connections between reactors. The Values ​​and Reservoirs modules are used to define the reactor's initial conditions, boundary information, and physical properties.

[0015] In this embodiment of the invention, the dynamic allocation of activated reactor units to independent computing threads and GPU task blocks, and the automatic cross-scale load balancing adjustment, includes: Activated reactors are allocated to independent computing threads and GPU task blocks as needed; Each time step automatically recycles inactive reactor units and adds reactor units in highly active areas, achieving dynamic adjustment of load balancing.

[0016] To achieve the above objectives, a second aspect of the present invention provides a heavy-duty gas turbine combustor reaction simulation acceleration device based on sparse dynamics, comprising: The heat release source term dynamic adjustment module is used to dynamically adjust the intensity of the heat release source term based on the difference between the CFD unit temperature and the corresponding reactor temperature, and eliminate the error caused by the assumption of uniform heat release distribution through feedback adjustment. The sparse reaction network building module is used to build sparse reaction networks with multiple reactor nodes based on the Cantera toolkit. The reactor node dynamic state switching module is used to evaluate the reaction intensity in the CFD region where the reactor unit is located in real time based on the sparsed reaction network. It activates the complete PSR reactor unit only in the high-activity region to perform complete chemical calculations, and adopts the order reduction mechanism or freezing treatment for the low-activity and inactive regions. The adaptive parallel deployment and update mechanism module is used to dynamically allocate activated reactor units to independent computing threads and GPU task blocks, automatically perform cross-scale load balancing adjustments, and synchronously maintain the bidirectional boundary coupling between the reactor and the CFD to ensure a dynamic balance between communication efficiency and computational stability.

[0017] To achieve the above objectives, a third aspect of this application provides a computer device comprising a processor and a memory; wherein the processor runs a program corresponding to the executable program code stored in the memory, for implementing a method for accelerating the simulation of heavy-duty gas turbine combustion chamber reactions based on sparse dynamics as described in the first aspect embodiment.

[0018] To achieve the above objectives, the fourth aspect of this application proposes a non-transitory computer-readable storage medium storing a computer program thereon, which, when executed by a processor, implements a method for accelerating the simulation of heavy-duty gas turbine combustion chamber reactions based on sparse dynamics as described in the first aspect embodiment.

[0019] The embodiments of the present invention have the following beneficial effects: The methods, apparatus, electronic devices, and computer-readable storage media of the present invention significantly improve the computational efficiency of heavy-duty gas turbine combustion chamber reaction simulation, reduce redundant computational resource consumption, and maintain the physical consistency and numerical stability of thermal-fluidic-chemical coupling. Attached Figure Description

[0020] The above and / or additional aspects and advantages of the present invention will become apparent and readily understood from the following description of the embodiments taken in conjunction with the accompanying drawings, wherein: Figure 1 A flowchart illustrating a method for simulating and accelerating the reaction of a heavy-duty gas turbine combustor based on sparse dynamics, provided for an embodiment of the present invention; Figure 2 A schematic diagram of an adaptive reaction network for a method to accelerate the simulation of combustion chamber reactions in a heavy-duty gas turbine based on sparse dynamics, provided for an embodiment of the present invention; Figure 3 This is a structural diagram of a heavy-duty gas turbine combustion chamber reaction simulation acceleration device based on sparse dynamics, provided as an embodiment of the present invention. Detailed Implementation

[0021] It should be noted that, unless otherwise specified, the embodiments and features described in the present invention can be combined with each other. The present invention will now be described in detail with reference to the accompanying drawings and embodiments.

[0022] To enable those skilled in the art to better understand the present invention, the technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments of the present invention. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort should fall within the scope of protection of the present invention.

[0023] To overcome the problems of existing technologies, this invention proposes an accelerated simulation method for heavy-duty gas turbine combustor reactions based on sparse dynamics. This method aims to address the issues of heavy computational burden and redundant computation in invalid response regions inherent in traditional global reaction integration methods for complex multi-scale, multi-regional combustion simulations. This invention does not rely on predefined geometric region partitioning but instead achieves efficient solution of complex reaction networks through a heat release integration mechanism, a dynamic reactor activation strategy, and an adaptive parallel scheduling mechanism, thus constructing an Adaptive Sparse Reaction Computational Framework (Adaptive CKN).

[0024] The following description, with reference to the accompanying drawings, describes a method and apparatus for simulating and accelerating the reaction of a heavy-duty gas turbine combustor based on sparse dynamics, according to an embodiment of the present invention.

[0025] Example 1 This embodiment provides an accelerated simulation method for the combustion chamber reaction of a heavy-duty gas turbine based on sparse dynamics. For example... Figure 1 As shown, the method includes the following steps: S1 dynamically adjusts the intensity of the heat release source term based on the difference between the CFD unit temperature and the corresponding reactor temperature, and eliminates the error caused by the assumption of uniform heat release distribution through feedback adjustment.

[0026] Traditional heat release coupling methods, involving global integration and fixed source term inputs, suffer from problems such as high computational cost, low efficiency, significant resource waste, and inaccurate coupling, especially in dense reaction networks and explicit time-progression scenarios. Their main drawbacks are: frequent and complex updates of heat source terms leading to high computational overhead; static inputs failing to reflect the dynamic evolution of reaction-flow-heat transfer; the involvement of numerous inactive regions in the calculation, resulting in wasted resources; and overly strong assumptions regarding heat release distribution, affecting accuracy.

[0027] Therefore, this invention implements a source term construction strategy driven by heat release integral. This strategy does not require a pre-defined heat release distribution form, but rather adaptively adjusts it based on the temperature difference, which improves local physical consistency and reduces overall computational cost, adapting to the rapid advancement requirements of explicit solvers.

[0028] Furthermore, by employing a source term construction strategy driven by heat release integral, this invention avoids the unreasonable assumptions introduced by traditional methods regarding the uniform distribution of heat release in CFD units, achieves feedback regulation of heat source intensity and local temperature deviation, enhances the physical consistency of thermal-fluidic-chemical coupling, and exhibits better numerical stability, especially under explicit propagation methods.

[0029] Specifically, step S1 is performed through the following steps: S11, in the initial iteration, the local unit is assumed to have no reactive heat source.

[0030] S12, in subsequent iterations, the source term size is dynamically adjusted based on the temperature difference between the CFD unit and its connected reactor (such as PSR). The formula defining the source term is expressed as:

[0031] in For CFD cell temperature, For reactor temperature, This is the convergence control factor.

[0032] S2, based on the Cantera toolkit, constructs a sparse reaction network containing multiple reactor nodes.

[0033] In this embodiment of the invention, in order to improve the simulation efficiency of complex reaction networks in the combustion chamber of heavy-duty gas turbines, a dynamic activation strategy for reactors is introduced to achieve on-demand scheduling and local precision control of reaction solution resources.

[0034] Specifically, the sparsification reaction network includes the following modules: The PSR module is used to simulate local mixing and reaction processes, serving as the basic unit for mesh and reaction coupling. The MFC module is used to control the mass exchange between reactors and to build the topological connections between reactors. The Values ​​and Reservoirs modules are used to define the reactor's initial conditions, boundary information, and physical properties.

[0035] It should be noted that even when a large number of reactors are inactive, this invention ensures the connectivity of the CKN topology and the consistency of decoupling iteration through a structure preservation mechanism, avoiding the risk of information loss or convergence failure caused by traditional order reduction methods, and ensuring the connectivity and stability of the sparse reaction network.

[0036] S3. Based on the sparsified reaction network, the reaction intensity within the CFD region where the reactor unit is located is evaluated in real time. Only the fully active region is activated to perform full chemical calculations. For the low-activity and inactive regions, a reduction mechanism and freezing treatment are used.

[0037] The reactor dynamic activation strategy proposed in this invention, combined with the modular sparse network (CKN) structure constructed by Cantera, evaluates the reaction intensity (such as reaction rate, heat release rate, temperature gradient, etc.) within the CFD region of each reactor unit in real time. It activates complete PSR reaction units only in highly active regions (regions with significant reaction intensity) and performs complete chemical calculations. For low-activity or inactive regions, it automatically switches to a frozen state (without updating the reaction) or uses a reduction mechanism for rapid estimation. This effectively compresses more than 90% of redundant reaction calculation units, improves overall computational efficiency by several times, and significantly reduces unnecessary computational burden.

[0038] Furthermore, to ensure the overall network topology and computational sustainability, all inactive reactors retain their node roles within the network, guaranteeing the connectivity integrity and iteration continuity of CKN, and preventing global computational interruptions due to local deactivation. For example... Figure 2 As shown, Figure 2 The left side shows the application of dynamic activation strategies to structured reactive networks. Figure 2 The right side shows the application of dynamic activation strategies to unstructured reactive networks, illustrating the effects of applying dynamic activation strategies to structured reactive networks (i.e., Adaptive Structured CKN) or unstructured reactive networks (i.e., Adaptive Untructured CKN).

[0039] This strategy effectively solves the problems of redundancy and low efficiency in reaction solving under traditional global activation, ensuring the controllability and efficiency of sparse reaction networks under complex regional conditions.

[0040] S4 dynamically allocates activated reactor units to independent computing threads and GPU task blocks, automatically performs cross-scale load balancing adjustments according to time steps, and synchronously maintains bidirectional boundary coupling between the reactor and the CFD to ensure a dynamic balance between communication efficiency and computational stability.

[0041] It should be noted that, considering the strong spatial non-uniformity of reactor activation, this invention constructs a parallel deployment framework suitable for Adaptive CKN: dynamically allocating activated reactor units to independent computing threads and GPU task blocks, automatically performing cross-scale load balancing adjustments according to time steps, and synchronously maintaining the bidirectional boundary coupling between the reactor and the CFD to achieve a dynamic balance between communication efficiency and computational stability. During the allocation process, this embodiment also supports cross-scale scheduling to adapt to different grid densities and load characteristics.

[0042] The source term construction and reactor activation processes in this embodiment of the invention exhibit a high degree of decoupling, making them naturally compatible with parallel deployment on GPU or multi-core platforms. Simultaneously, the source term update mechanism supports real-time dynamic adjustment based on local states, demonstrating excellent adaptability and scalability. This mechanism, combined with a local load balancing strategy, effectively supports large-scale parallel reaction computation under high-resolution CFD, making it suitable for deployment on modern high-performance computing (HPC) platforms.

[0043] Specifically, step S4 is performed through the following steps: S41 allocates activated reactors to independent compute threads and GPU task blocks as needed.

[0044] S42 automatically recycles inactive reactor units and adds reactor units in highly active areas at each time step, achieving dynamic adjustment of load balancing.

[0045] In summary, the embodiments of the present invention effectively solve the problems of coupling instability, redundant calculation and efficiency bottlenecks in the traditional CFD-CKN thermo-chemical information fusion process. It is applicable to multi-physics fine simulation scenarios of complex combustion fields such as heavy-duty gas turbines and aero engines, and has broad engineering application value.

[0046] Example 2 This invention also provides a heavy-duty gas turbine combustor reaction simulation acceleration device based on sparse dynamics, such as... Figure 3 As shown, the device 10 includes: The heat release source term dynamic adjustment module 100 is used to dynamically adjust the intensity of the heat release source term based on the difference between the CFD unit temperature and the corresponding reactor temperature, and eliminate the error caused by the assumption of uniform heat release distribution through feedback adjustment. Sparse reaction network building module 200 is used to build sparse reaction networks containing multiple reactor nodes based on the Cantera toolkit; The reactor node dynamic state switching module 300 is used to evaluate the reaction intensity in the CFD region where the reactor unit is located in real time based on the sparsed reaction network. It activates the complete PSR reactor unit only in the high-activity region to perform complete chemical calculations, and adopts a reduction mechanism or freezing treatment for the low-activity and inactive regions. The adaptive parallel deployment and update mechanism module 400 is used to dynamically allocate activated reactor units to independent computing threads and GPU task blocks, automatically perform cross-scale load balancing adjustments, and synchronously maintain the bidirectional boundary coupling between the reactor and the CFD to ensure a dynamic balance between communication efficiency and computational stability.

[0047] Example 3 To implement the methods of the above embodiments, the present invention also provides a computer device, which includes a memory and a processor; wherein the processor runs a program corresponding to the executable program code by reading executable program code stored in the memory, so as to implement the various steps of the methods described above.

[0048] Example 4 To implement the above embodiments, this application also proposes a non-transitory computer-readable storage medium storing a computer program thereon, which, when executed by a processor, implements the method described in the foregoing embodiments.

[0049] The above description is merely a preferred embodiment of the present invention and is not intended to limit the invention. Various modifications and variations can be made to the present invention by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the scope of protection of the present invention.

[0050] In the description of this specification, the references to terms such as "one embodiment," "some embodiments," "example," "specific example," or "some examples," etc., refer to specific features, structures, materials, or characteristics described in connection with that embodiment or example, which are included in at least one embodiment or example of the present invention. In this specification, the illustrative expressions of the above terms do not necessarily refer to the same embodiment or example. Furthermore, the specific features, structures, materials, or characteristics described may be combined in any suitable manner in one or more embodiments or examples. Moreover, without contradiction, those skilled in the art can combine and integrate the different embodiments or examples described in this specification, as well as the features of different embodiments or examples.

[0051] Furthermore, the terms "first" and "second" are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of technical features indicated. Thus, a feature defined as "first" or "second" may explicitly or implicitly include at least one of that feature. In the description of this invention, "a plurality of" means at least two, such as two, three, etc., unless otherwise explicitly specified.

Claims

1. A method for simulating and accelerating the reaction of a heavy-duty gas turbine combustor based on sparse dynamics, characterized in that, Includes the following steps: The intensity of the heat release source term is dynamically adjusted based on the difference between the temperature of the CFD unit and the temperature of the corresponding reactor, and the error caused by the assumption of uniform heat release distribution is eliminated through feedback adjustment. Based on the Cantera toolkit, a sparse reaction network containing multiple reactor nodes is constructed. Based on the sparsified reaction network, the reaction intensity within the CFD region where the reactor unit is located is evaluated in real time. Only the fully active region is activated to perform full chemical calculations. For the low-activity and inactive regions, a reduction mechanism or freezing treatment is adopted. Activated reactor units are dynamically allocated to independent computing threads and GPU task blocks, and cross-scale load balancing is automatically adjusted according to time steps. The bidirectional boundary coupling between the reactor and the CFD is maintained synchronously to ensure a dynamic balance between communication efficiency and computational stability.

2. The method according to claim 1, characterized in that, The method of dynamically adjusting the intensity of the heat release source term based on the difference between the CFD unit temperature and the corresponding reactor temperature, and eliminating the error caused by the assumption of uniform heat release distribution through feedback adjustment, includes: In the initial iteration, it is assumed that the local unit has no reactive heat source; In subsequent iterations, the source term size is dynamically adjusted based on the temperature difference between the CFD unit and the reactor it is connected to. The formula defining the source term is expressed as: in For CFD cell temperature, For reactor temperature, This is the convergence control factor.

3. The method according to claim 1, characterized in that, The sparsification reaction network includes: The PSR module is used to simulate local mixing and reaction processes, serving as the basic unit for mesh and reaction coupling. The MFC module is used to control the mass exchange between reactors and to build the topological connections between reactors. The Values ​​and Reservoirs modules are used to define the reactor's initial conditions, boundary information, and physical properties.

4. The method according to claim 1, characterized in that, The dynamic allocation of activated reactor units to independent computing threads and GPU task blocks, and the automatic cross-scale load balancing adjustments, include: Activated reactors are allocated to independent computing threads and GPU task blocks as needed; Each time step automatically recycles inactive reactor units and adds reactor units in highly active areas, achieving dynamic adjustment of load balancing.

5. A heavy-duty gas turbine combustion chamber reaction simulation acceleration device based on sparse dynamics, characterized in that, include: The heat release source term dynamic adjustment module is used to dynamically adjust the intensity of the heat release source term based on the difference between the CFD unit temperature and the corresponding reactor temperature, and eliminate the error caused by the assumption of uniform heat release distribution through feedback adjustment. The sparse reaction network building module is used to build sparse reaction networks with multiple reactor nodes based on the Cantera toolkit. The reactor node dynamic state switching module is used to evaluate the reaction intensity in the CFD region where the reactor unit is located in real time based on the sparsed reaction network. It activates the complete PSR reactor unit only in the high-activity region to perform complete chemical calculations, and adopts the order reduction mechanism or freezing treatment for the low-activity and inactive regions. The adaptive parallel deployment and update mechanism module is used to dynamically allocate activated reactor units to independent computing threads and GPU task blocks, automatically perform cross-scale load balancing adjustments, and synchronously maintain the bidirectional boundary coupling between the reactor and the CFD to ensure a dynamic balance between communication efficiency and computational stability.

6. A computer device, characterized in that, Including processor and memory; The processor reads executable program code stored in the memory to run a program corresponding to the executable program code, so as to implement the heavy-duty gas turbine combustor reaction simulation acceleration method based on sparse dynamics as described in any one of claims 1-4.

7. A non-transitory computer-readable storage medium having a computer program stored thereon, characterized in that, When the program is executed by the processor, it implements the method for accelerating the simulation of heavy-duty gas turbine combustion chamber reaction based on sparse dynamics as described in any one of claims 1-4.

Images