【作者】 宋天一；

【导师】 马学虎； John W Rose；

【作者基本信息】 大连理工大学 ， 化学工程， 2010， 博士

【摘要】 蒸气滴状冷凝传热因其较高的传热系数而受到了研究者们的广泛关注,对其过程特性的研究有助于更加深入地理解其本质机理,这对于冷凝传热传质的强化和过程系统节能具有十分积极的意义。本论文主要研究了蒸气冷凝过程的微观尺度特性,包括空间微尺度和时间微尺度,以及微通道(空间尺寸微细化)内蒸气冷凝压降特性。根据核化过程的基本特征,本文引入分子团聚理论,建立了蒸气冷凝的分子团聚物理模型,对于冷凝表面上初始液滴的形成过程进行了描述：蒸气冷凝核化发生之前,蒸气分子在靠近冷凝壁面的蒸气主体中首先发生团聚,形成团聚体的特征尺寸分布,然后这些夹杂有大量团聚体的蒸气与冷凝壁面接触并在随机位置上发生沉积；沉积后的团聚体能量并未立即湮灭,它们会在冷凝壁面上的一个范围内发生徙动,徙动过程中会与其它团聚体发生融合生长,形成新的体积较大的团聚体(微液滴),并最终停止在其徒动范围内的某一个高能点或表面缺陷处(即核化点)。在上述过程中,这些团聚体表面上始终伴随着蒸气分子或更小团聚体的冷凝与蒸发过程；如果停留在核化点上的微液滴尺寸大于或等于传统滴状冷凝理论定义的最小液滴半径,就形成了冷凝核化中心并逐渐长大,之后的过程就可以用传统滴状冷凝理论进行描述。本文利用分子动力学方法对近冷凝壁面的一个薄层内的水蒸气分子的降温过程中团聚体的形成过程进行了模拟分析。为了证明上述物理模型的合理性,本文首先利用高速摄像和显微技术研究了湿空气露点冷凝的初始过程,分析了冷凝表面上发生明显合并现象之前,半径约为数个微米的液滴尺寸分布特征,发现其符合气相主体中分子团聚体的特征尺寸分布即对数正态分布；然后对文献中利用电子探针技术得到的水蒸气在金属镁膜表面上的初始冷凝图像进行分析,得到了半径为数个纳米的滴状冷凝初始液滴的尺寸分布,发现该分布同样满足对数正态函数,从而说明滴状冷凝中从刚刚形成的初始液滴至合并发生之前的小于临界半径的液滴都符合对数正态尺寸分布,揭示了蒸气初始冷凝过程的空间微尺度特征,并根据分子团聚理论印证了蒸气主体中的分子团聚体的存在,证明了本文提出的蒸气冷凝分子团聚物理模型的合理性。另外,对比不同润湿性表面上的初始露点冷凝过程图像发现,滴状冷凝和膜状冷凝的初始时刻都是“滴状”的,这一现象进一步说明了本文提出的物理模型的合理性；当表面上发生液滴合并之后,其间的差异才表现出来：在疏水表面上,合并后的液滴接触线可以很快地恢复到圆形,而在亲水表面上,合并后的液滴接触线则被滞留在发生合并的位置,最终也呈现出多种不规则形状,表明宏观冷凝形态是由液—固界面效应所决定的,当液—固相间作用力大于液体表面张力时,最终会形成膜状冷凝；反之则形成滴状冷凝,液—固界面效应是通过影响冷凝液接触线的运动行为来决定最终冷凝形态的。本文研究了初始滴状冷凝过程中第一个子循环内的表面液滴尺寸分布的依时性变化,揭示了蒸气初始冷凝过程的时间微尺度特征：冷凝表面上的初始液滴从形成伊始至合并发生之前的第一代液滴都具有对数正态分布特征,尺寸分布图上存在一个数量峰值；随着液滴逐渐长大并发生合并,第一代液滴数量峰的位置向液滴尺寸更大的方向移动并伴随着峰值的减小,与此同时液滴合并产生的空白表面上重新发生核化,大量第二代液滴形成,在最小液滴的位置处组成了另一个数量峰,此时表面上液滴尺寸呈双峰分布；最后冷凝过程逐渐趋于稳定状态,第一个数量峰最终由于液滴合并而逐渐消失,而第二个数量峰则由于冷凝表面上不断有空白表面出现而得以保留并相对稳定,此时冷凝表面上的液滴尺寸分布呈现经典滴状冷凝理论描述的指数分布。另外通过对不同压力蒸气在由空气自然对流冷却的聚碳酸酯表面上冷凝过程图像的表观特征进行分析,发现其瞬态冷凝传热速率不同于稳定滴状冷凝传热随压力升高而增大的规律,存在有一个先增大、后减小、然后再增大的波动特征。对于微小冷凝空间对冷凝过程及其特性的影响,本文主要研究了平行多微通道入口突缩压降和微通道内的冷凝摩擦压降。对比实验测定的压缩空气突缩压降数据和常用的突缩压力损失系数关联式计算结果发现,现有的针对常规尺寸和微小尺寸的单通道突缩压力损失系数计算关联式都不能对其进行很好地预测。本文基于压缩空气的压降数据得到了突缩压力损失系数与单个通道内空气流动Re数之间的关系,并拟合回归了无因次关联式(Re数范围：3100～19000),该关联式同样可以较好地预测水蒸气单相流动突缩压力损失系数。对于微通道内的蒸气冷凝摩擦压降,本文对较为常用的三个模型(Koyama等,Garimella等和Cavallini等)对于相同尺寸单微通道内不同工质(R134a, NH3, FC-72和水蒸气)在不同质量通量(100,300,500和700 kg·m-2·s-1)时的冷凝摩擦压降梯度随蒸气干度变化的预测进行了对比,发现之间差异较大。本文在新搭建的微通道冷凝实验台上,测定了FC-72 (140～750kg·m-2·s-1)和水蒸气(80～200kg·m-2·s-1)的冷凝摩擦压降,并与上述模型的预测结果进行了对比,发现Cavallini模型可以较好地预测FC-72的微通道冷凝摩擦压降；Koyama和Garimella模型则可以较好地描述水蒸气的微通道冷凝摩擦压降。最后本文对Koyama模型进行了修正使其同时适用于FC-72和水蒸气的微通道内冷凝摩擦压降预测。更多还原

【Abstract】 For the excellent heat transfer efficiency, vapor dropwise condensation has been given extensive attention by many investigators, thoroughly studies on the characteristics of this process will help further understand the fundamental mechanism, facilitate the vapor condensation heat and mass transfer enhancement and process energy saving. Microscale characteristics of vapor condensation process have been investigated in the present thesis, including the microscale-space and microscale-time characteristics and the effect of microscale space on condensation pressure drop.In the present work, a physical model in terms of the molecular cluster theory is presented to describe the state of steam molecules in bulk steam phase before condensing on the cooled surface and the resultant forming process of primary droplets. With the existence of surface subcooling, the steam molecules become clusters before condensing on the cooled surface, and a certain cluster size distribution forms in the bulk steam phase close to the surface. After then, the clusters contact the cooled surface and deposit on it randomly, acting as the nucleate embryos. Since the clusters’energies will not dissipate instantly, some of the deposited clusters are able to migrate on the surface in a limited area and incorporate or be incorporated by the other clusters on their ways, accompanied by the condensation of other clusters from vapor phase. At the same time, the evaporation on the surface is on-going as well. After energies have been dissipated completely, the clusters stay on the locations with high free energies, such as the pits and grooves on the surface. Finally, if the seated, post incorporated clusters were not smaller than the minimal size defined in the classic dropwise condensation theory, they become the primary droplets and grow up through the condensation of the other clusters on their surfaces. Molecular dynamics (MD) method was used to simulate the clustering process of steam molecules within a vapor layer near to the condensing surface.To prove the clustering model for steam condensation, the presupposition is whether the vapor molecules become clusters before condensing on the cooled surface. Inferring from the results of the molecular cluster theory, if the size distributions of the primary droplets and the droplets growing up through direct condensation without coalescence on the condensing surface satisfy the Lognormal distribution function, the same size distribution should exist in the bulk gas phase and steam molecules should become clusters before condensing. To verify the present physical model, the initial dew point condensation process of moist air has been investigated experimentally using high speed camera and microscope, the size distribution corresponding to the droplets with size of several microns before coalescence was obtained. The results show that the droplet size distribution obviously satisfies the typical cluster size distribution, Lognormal distribution. Furthermore, the size distribution corresponding to the primary droplets with size of several nanometers was analyzed based on the experimental image reported in literature, the primary droplet size distribution satisfies Lognormal function as well. These results indicate that the size distribution of the droplets from primary to pre-coalescing size satisfy the Lognormal function. In addition, after comparing the initial condensation processes on the surfaces with different wettabilities, it was found that the initial stages of dropwise and filmwise condensation are identical, from nucleating to growing up through direct condensation and then to coalescing, their discrepancy just occurs after coalescing among droplets. The contact lines of droplets on hydrophobic surface can get back to be circular immediately, while those on hydrophilic surface are pinned and remain the shapes right after coalescing. It can be concluded that the liquid-solid interfacial effect influences the condensation mode through affecting the behavior of contact line of condensate.The transient droplet size distributions in the first sub-cycle of initial dropwise condensation were investigated experimentally, and the characteristics on micro-time-scale of vapor condensation have been revealed. The size distribution from primary droplets to those before coalescence (first generation droplets) satisfy Lognormal function, there is only one droplet number peak on the size distribution plots, resulted from the growing and coalescing, the position of the droplets number peak corresponding to the first generation droplets shifts towards the direction with larger droplet size and however with its value decreasing. At the same time, on the bare areas appearing on the condensing surface resulting from coalescences among droplets, nucleation occurs again, then another droplets number peak appears at the droplet radius of about 0.5 to 1μm constantly, the bimodal size distribution forms. Also, the droplet number decreases rapidly due to the coalescence of the first generation droplets, resulting in lower and lower value of first droplet number peak, and it disappears finally. On the other hand, the initially nucleated droplets periodically form on the bare areas always have the same size, so the location of second droplet number peak keeps constant relatively. After long enough time evolution, the droplet size distribution approaches to the steady state as described in classic dropwise condensation. In addition, the investigations on initial dropwise condensation of different pressure steam on polycarbonate (PC) surface indicate that the evolutions of transient condensation stages on low thermal conductivity surface are affected by steam pressure obviously, demonstrating different features from the steady state dropwise condensation process. The abrupt contraction pressure drop at entrance of and the condensation frictional pressure drop in the parallel multi-microchannels were investigated experimentally in the present work to elucidate the effect of micorscale space on condensation pressure drop. Pressure and temperature before and after the entrance of parallel multi-microchannels have been measured using compressed air and steam as working fluids. The results indicate that the existing correlations which were commonly used to predict the abrupt contraction pressure drop coefficient in single channel with commercial and small size cannot reasonably predict the trend for the case concerned in this work. A new empirical formula for the abrupt contraction pressure drop coefficient against Reynolds number (range from 3100 to 19000) in one single micro-channel for air has been correlated, and its availability has been tested by the pressure drop data of saturated steam single phase flow. Microchannel condensation frictional pressure drop gradients predicted by the most accepted models, such as those by Koyama et al, Garimella et al, and Cavallini et al for R134a, NH3, FC-72 and steam have been compared at different mass fluxes of 100,300,500 and 700 kg·m-2·s-1, the obvious discrepancies among the predictions were found. Frictional pressure drop of steam and FC-72 condensation in parallel multi-microchannels (consist of 6 parallel 1 mm×1.5 mm channels with 1.5 mm in space between each other machined in aluminium substance) were measured very accurately and compared with the three models calculations, mass flux ranges of FC-72 and steam were 140～750 kg·m-2·s-1 and 80～200 kg·m-2·s-1, respectively. Koyama and Garimella models can estimate steam condensation frictional pressure drop within the deviation of±20%, Cavallini model overestimates it for about 3 folds; while for FC-72, Cavallini model has a better agreement with our measurements. Koyama model was modified to be able to predict frictional pressure drop of FC-72 and steam condensation in microchannels finally.更多还原