Miiller-Steinhagen, HansSchou, Grant Brian2020-07-082020-07-081994https://hdl.handle.net/2292/52247Full text is available to authenticated members of The University of Auckland only.A test rig has been assembled to investigate the thermodynamic and hydraulic performance of a cylindrical graphite block heat exchanger consisting of three vibration moulded graphite blocks inside a steel shell. The flow pattern is triple cross-flow on the shell-side (service side) with one pass per block and single pass on the tube-side (process side). Overall heat transfer coefficients and pressure drops have been measured for a range of operating conditions. Initial experimental parameters were the Reynolds and Prandtl numbers. Further tests were performed on a modified set up of the apparatus to investigate the effect of the resin layer on the surface of the channels on the overall heat transfer coefficient. Additionally, the effects of tube-side natural convection, the outside surface of the graphite blocks, an unheated entrance length and bypass flow on the heat transfer rate were investigated. There was no observable transition from laminar to turbulent flow on the tube-side down to a Reynolds number of 1000. The resin layer on the surface of the channels was determined to have a combined thermal resistance of 26×10⁻⁵ m²K/W. Depending on the film coefficients in the heat exchanger channels, the effect of the resin layer on the overall heat transfer coefficient can be considerable. The effects of natural convection, the outside surface of the blocks, and the unheated entrance length on the heat transfer rate were negligible. Bypass flows were significant, with values up to 19% with water and 38% with oil. Corresponding increases in the overall heat transfer coefficient without bypass flow were 8% and 21%, respectively. However, without bypass flow the shell-side pressure drop increases by up to 15% for water and up to 38% for oil. Flow visualisation experiments performed on a plexiglas model of the shell-side of the heat exchanger showed that there was a certain amount of water bypassing the channels and flowing over the vertical baffles. A mathematical analysis of an assumed shell-side velocity distribution showed that the effect of a flow maldistribution on the heat transfer rate would be small. Accurate data have also been obtained on the variation of the thermal conductivity of various grades of graphite with orientation, manufacturing process and impregnant content. The thermal conductivity of non-impregnated, extruded graphite showed significant and uniform anisotropy, ranging from 140 W/mK to 100 W/mK parallel and perpendicular to the direction of extrusion, respectively. Impregnation of extruded graphite with pitch increases the thermal conductivity at all angles by approximately 20 W/mK, while still maintaining its uniform anisotropy. Both vibration moulded and isostatically pressed graphite exhibit little anisotropy. The thermal conductivity of vibration moulded graphite with and without a synthetic resin impregnant was determined to be approximately 97 W/mK and 103 W/mK, respectively. For isostatically pressed graphite, the thermal conductivity was significantly less, at approximately 56 W/mK and 63 W/mK for the nonimpregnated and resin impregnated grades, respectively. A three-dimensional finite element model has been set-up to determine the effective thermal resistance of the graphite matrix. A resistance of 9.59×10⁻⁵ m²K∕W was calculated for the investigated block type. This value was then used together with the theoretical shell and tube-side film coefficients to predict the overall heat transfer coefficients. The mean difference between the theoretical and experimental overall heat transfer coefficients for the runs with bypass flow does not exceed 7% with water as the shell-side liquid. With oil the error does not exceed 6.5%. The theory Underpredicts the overall heat transfer coefficient and so gives a safety margin for the required heat transfer area. Pressure drop models were developed for both the shell-side and tube-side of the heat exchanger. The tube-side model predicted the pressure drop with a mean error of 5%. Errors within 20% can be expected for the shell-side pressure drop model. The pressure drop on the shell-side depends on a number of different losses and it is therefore difficult to be modelled accurately. Le Carbone-Lorraine's thermal design procedure neglects shell-side bypass flow and uses the same wall resistance for all their different blocks. In the assumed transition and laminar flow regimes on the tube-side, the design equations from Le Carbone-Lorraine overestimate the overall heat transfer coefficient by up to 40% and so underpredict the heat transfer area. A user-friendly computer program has been written for SGL Carbon GmbH, the sponsors of this project, for the design of their DK series graphite block heat exchangers. Using the mathematical models developed in this investigation, the program calculates the number of blocks, of a specified type, required to perform a given heat duty and then calculates the pressure drop across both sides of the heat exchanger.Items in ResearchSpace are protected by copyright, with all rights reserved, unless otherwise indicated.Restricted Item. Full text is available to authenticated members of The University of Auckland only.https://researchspace.auckland.ac.nz/docs/uoa-docs/rights.htmThesisCopyright: The authorQ112854002