Optimization of Heat Exchangers

Abstract

Executive SummaryThe objective of this research is to develop tools to design and optimize heat exchangers (HE) and compact heat exchangers (CHE) for intermediate loop heat transport systems found in the very high temperature reactor (VHTR) and other Generation IV designs by addressing heat transfer surface augmentation and conjugate modeling. To optimize heat exchanger, a fast running model must be created that will allow for multiple designs to be compared quickly. To model a heat exchanger, volume averaging theory, VAT, is used. VAT allows for the conservation of mass, momentum and energy to be solved for point by point in a 3 dimensional computer model of a heat exchanger. The end product of this project is a computer code that can predict an optimal configuration for a heat exchanger given only a few constraints (input fluids, size, cost, etc.). As VAT computer code can be used to model characteristics (pumping power, temperatures, and cost) of heat exchangers more quickly than traditional CFD or experiment, optimization of every geometric parameter simultaneously can be made. Using design of experiment, DOE and genetic algorithms, GE, to optimize the results of the computer code will improve heat exchanger design.For optimization to be accomplished, one needs to separate conduction and convection at its natural interface. This means friction and heat transfer at the interface needs to be treated in a way that allows each side to be independently varied. If you increase the flow rate, the heat transfer coefficient will increase and you need to take advantage of this by changing the geometry to modify the conduction side of the problem. Of course you try to do this while keeping the friction factor as small as possible. All of this is done within a complex heterogeneous hierarchical structure.iii Several issues arise when we set out to solve such a problem. One of the more interesting is how you compare CFD calculations to measured values of the two parameters that are needed.Another is how to directly use experimental data. The simple equations contain coefficients that are essentially averages of some kind and we are usually unsure of what they are. Measured friction factors are fairly straightforward to interpret although calculated values are not.Measured heat transfer coefficients are usually based on some kind of upper level assumptionsand often not what we need and calculated values can be difficult to interpret correctly. Using VAT to derive the simple equations leads us to a rigorous definition of the needed coefficients whether obtained from experiment or through the use of CFD.Determination of flow-variables and scalar transport in a heterogeneous (or porous) media is difficult even when subject to simplifications allowing the specification of medium periodicity or regularity. Linear or linearized models fail to intrinsically account for transport phenomena, requiring dynamic coefficient models to correct for short-comings in the governing models. Allowing inhomogeneities to adopt random or stocha...