再生冷却作为一种主动热防护形式，被广泛应用于高超声速飞行器的热防护系统。对于超燃冲压发动机，随着飞行马赫数的提升，传统的再生冷却结构面临着冷却裕度不足、结构超温等问题，亟需开展新型的热防护方案设计及优化方法研究。在众多优化方法中，以拓扑优化为代表的新兴方法，因其独特的设计思路、创新的设计构型而逐渐受到研究者的青睐。同时得益于增材制造技术的逐渐成熟，使拓扑优化成为最有发展前途的优化工具之一。

通过对国内外流体力学拓扑优化领域的研究进行总结，发现目前该方法主要应用于常物性、层流等具有简单流态的散热器优化设计，而以发动机再生冷却为背景的拓扑优化相关研究屈指可数。主要难点在于发动机再生冷却结构中的流动换热是具备高雷诺数湍流、燃料物性剧烈变化以及流动可压缩性等特征的强耦合问题，因此在优化难度和计算量方面都极具挑战。为此，本文针对超燃冲压发动机再生冷却结构，在经典的伪密度法拓扑优化模型基础上，提出了考虑变物性以及高雷诺数修正的拓扑优化模型，自主开发了与超临界流体流动换热计算耦合的拓扑优化求解器，实现了针对发动机再生冷却结构的拓扑优化设计。

首先建立了考虑超临界流体变物性的伪密度法流热耦合拓扑优化模型，基于连续伴随法推导得到了灵敏度及伴随方程，随后利用开源CFD平台OpenFOAM开发了多物理拓扑优化求解器MPFTOFoam，并对灵敏度计算的准确性进行了验证。利用该求解器，分别针对导热热沉结构、流动歧管结构以及微通道散热器结构等问题开展了拓扑优化设计，验证了所开发的求解器具备在不同边界条件、不同目标函数以及不同约束条件而自适应生成拓扑优化结构的能力。

在面向发动机再生冷却结构进行拓扑优化设计时，首先提出了考虑高雷诺数修正的拓扑优化模型，即在动量方程中添加了惯性力修正项，在湍流模型方程中添加了惩罚源项，从而在拟多孔介质湍流流动中形成了有效的人工阻力，弥补了经典的拓扑优化模型在高雷诺数湍流流动优化中的不足。随后利用该修正模型分别针对单通道和多通道冷却面板设计域开展了拓扑优化设计。研究发现，相比于传统的等截面直通道，拓扑优化通道中由于固体域的自动分裂，使得冷却剂在流动中发生了多重流动分离和再混合行为，在固体胞元前后缘以及弯曲截面等多个位置产生了二次涡结构，激发了湍动能，进而增强了局部的换热能力。在冷却面板的拓扑优化设计中，发现优化结构不仅降低了结构质量和流动压降损失，还改善了传统直通道中由热加速引起的传热恶化现象，使得局部换热增强，并在非均匀热流分布及不同流量工况下均展现出较好的综合性能。

为了降低拓扑优化的计算量和优化周期，本文进一步发展了再生冷却通道的复合优化方法。首先提出了由典型的形状胞元组合而成的新型再生冷却通道，对其流动换热性能和强化换热机理等进行了分析，随后利用形状优化和拓扑优化方法对胞元冷却通道进行了二次优化。在形状优化中，首先对胞元通道的形状进行参数化，利用回归拟合得到了各目标函数关于形状参数变量的响应面函数，建立了多目标参数优化模型，基于遗传算法计算得到了四种权重分配下的优化方案，经验证该模型具有较好的预测准确性。在拓扑优化中，基于共轭换热拓扑优化模型，分别以等截面直通道以及胞元冷却通道作为初始构型进行了拓扑优化设计，经过优化，在直通道内自动生成了哑铃形的微肋结构，而在胞元冷却通道中则生成了形状各异的新型胞元结构，两类新型通道均展现出较好的流动换热性能。

As an active thermal protection technology, regenerative cooling has been widely applied in the hypersonic vehicle’s thermal protection system (TPS). As for scramjet, with the increase of flight Mach number, traditional regenerative cooling structures are faced with issues, such as insufficient cooling capacity and over-temperature, therefore, it is a crucial task to develop novel thermal protection design and optimization methods. Among the existing optimization methods, topology optimization (TO) is gradually favored by researchers since its unique optimization process and innovative configurations. Additionally, topology optimization has been regarded as one of the most promising optimization tools with the maturity of additive manufacturing (AM) technology.

By summarizing the research in the field of fluid mechanics topology optimization, it is found that topology optimization is mainly applied to the optimization design of heat sinks with constant physical properties and laminar flow, while the topology optimization of scramjet’s regenerative cooling structure is few and far between. The main difficulty is that flow and heat transfer in regenerative cooling channels is a complicated fluid-thermal coupled problem, which combines high Reynolds number turbulent flow and dramatic variation and compressibility of the fuel’s properties, causing its calculation and optimization to be extremely challenging. Therefore, based on the classical pseudo-density based topology optimization model, a modified topology optimization model considering variable physical properties and high Reynolds number’s correction was proposed in this paper, and a complete topology optimization numerical solver coupled with supercritical fluid flow and heat transfer was developed, which fulfills the topology optimization design for engine’s regenerative cooling structure.

Firstly, a pseudo-density based fluid-thermal coupled topology optimization model, which considers the thermophysical properties of supercritical fluid, was established, and the corresponding sensitivity and adjoint equations were derived based on the continuous adjoint method. An in-house topology optimization solver MPFTOFoam was then built on OpenFOAM, which is an open-source CFD platform. By utilizing this solver, the topology optimization of three physical problems were conducted, i.e. the heat conduction structure, the flow manifold structure, and the heat sink structure, respectively. The function and feasibility of the developed solver were verified by adaptively generating optimized structures under different boundary conditions, different objective functions, or different unconstraint conditions.

In the topology optimization design for engine’s regenerative cooling structure, a topology optimization model with high Reynolds number correction was first proposed. Specifically, an inertia correction term was added to the momentum equation and a penalty term was added to the turbulent model, to form an effective artificial resistance in the turbulent flow of porous media, it also makes up for the deficiency of the classical topology optimization model in the optimization of high Reynolds number turbulent flow. The model was then utilized to conduct topology optimization designs for a single channel and a multi-channel panel. It was found that, in the topology-optimized channels, the automatically split solid cells resulted in multiple flow separations and re-mixing behaviors of the coolant. It further resulted in the generation of secondary vortex structures at various positions, including the front and rear edges of solid cell elements as well as in curved sections, thereby stimulating turbulence energy and enhancing local heat transfer performance. Furthermore, in the topology optimization of multi-channel panel, it was observed that the optimized structure not only reduces the structural mass and flow pressure drop but also ameliorates the heat transfer deterioration (HTD) caused by thermal acceleration in traditional straight channels. Additionally, the optimized structure demonstrated favorable comprehensive performance under non-uniform heat flux distribution and various flow rate conditions.

To reduce the computation and optimization period of topology optimization, the composite optimization method of the regenerative cooling channel was further researched. Firstly, a series of cooling channels composed of cellular structures were proposed, the flow and heat transfer performance and the enhancement mechanism were numerically investigated, then the shape optimization and topology optimization methodologies were employed. In the shape optimization, the structure of the cells-shaped channel was first parameterized. By utilizing regression fitting, the response surface functions for each objective function regarding design variables were obtained, and a multi-objective optimization model was established. Four optimization designs with different weights were obtained based on genetic algorithms (GA), and the model shows favorable prediction accuracy. In the topology optimization, based on the conjugate heat transfer (CHT) topology optimization model, the topology optimization designs were conducted on the straight and the cell-shaped cooling channels, respectively. After the optimization, the dumbbell-shaped micro-rib structures were automatically generated within the straight channel, and the cellular structures with various shapes were automatically generated within the cell-shaped cooling channel, and the novel channels showed favorable flow and heat transfer performance.