【摘要】 建立了可进行壳管式换热器动态特性试验研究系统,通过试验研究的方法对水-油为换热工质的连续螺旋折流板管壳式换热器动态特性进行了试验研究,进口流量扰动为等百分比流量特性,研究了4种流量扰动方式下水和油出口温度的动态响应。同时研究了在一定Re数下,不同的流体扰动量对换热器进出口温升的影响,得到了换热器进出口温升与流体扰动量之间的关联式。实验表明,液液换热系统温度的动态响应时间比较长,研究发现在正负的流量扰动下,换热器进出口温度变化呈现线性变化,进出口温升在正负流量扰动下其变化曲线具有对称特征。分别建立了有限差分数值预测模型及人工神经网络模型对换热器油侧的出口温度进行了动态预测,预测结果与试验值符合良好,人工神经网络的预测结果要好于数值模拟预测,其偏差绝对值在1.3%以内,表明人工神经网络在进行复杂的系统辨识时具有一定的参考及应用价值。更多还原

【Abstract】 A system has been established to conduct the experimental study of dynamic characteristics of shell-and-tube heat exchangers.Through experimental methods,an experimental study has been performed of the dynamic characteristics of heat exchangers with continuous spiral baffle plates,which have water and oil to serve as heat exchange working media.With the inlet flow disturbance representing equal percentage flow rate characteristics,studied were the dynamic responses to water and oil temperature at the outlet in four modes of flow rate disturbance.In the meantime,the effect of fluid disturbing quantity on the temperature rise at heat exchanger inlet and outlet at a certain Re number was also studied with a correlation equation between the temperature rise at the inlet and outlet of the heat exchanger and fluid disturbing quantity being obtained.The tests show that the dynamic response to the temperature of a liquid-liquid heat exchange system needs a comparatively long time.It has been found through the study that under a disturbance with a positive or negative flow rate,the change of temperature at the inlet and outlet of spiral-baffle plate heat exchangers assumes a linear relationship and the variation curve featuring temperature rise at the inlet and outlet has a symmetric feature at a positive or negative flow rate disturbance.A finite-difference numerical forecast model and artificial neural network one have been established respectively to dynamically forecast the outlet temperature at the heat exchanger oil side.The forecast results are in good agreement with the test values.The forecast results obtained from the neural network is better than those from the numerical simulation with the absolute value of the deviation being less than 1.3%,indicating that artificial neural network has a definite merit for engineering reference and applications when performing the discrimination of complicated systems.更多还原