DESIGN OF PULSATING HEAT PIPES - A novel non-equilibrium lumped parameter model for transient gravity levels

Marco Marengo - Miriam Manzoni
University of Brighton, UK - Universita' di Bergamo, Italia
Monday 30th November 2015
Aula Semnari Saleri VI Piano MOX Dipartimento di Matematica, Politecnico di Milano
In the last two decades a new concept of capillary heat pipe without wick structure, commonly known as Pulsating Heat Pipe (PHP), entered the domain of the two-phase passive heat transfer devices. Characterized by high effective thermal conductivity and construction simplicity, PHPs may answer to the present demand of high heat transfer capability, efficient thermal control and low cost. Essentially, a PHP consist of a tube, evacuated, partially filled with a working fluid and finally bended to arrange a closed serpentine. Heat is provided in the so called evaporator region and removed in the condenser. The thermal-hydraulic behaviour of this device is chaotic and non linear. The fluid can both circulate and oscillate within the channel. The internal dynamic mainly depends on the interplay between phase change phenomena, capillary and gravity, if present, which may assist or damp the fluid motion. From the very beginning, attempts to numerically predict PHPs performances arose, but only few models are capable of complete thermal-hydraulic simulations. In addition, none are validated for transient operations neither under various gravity levels. Literature, indeed, reports very poor data regarding the effect of gravity loads on the device performances, even if modified-gravity may appears in several applications, from automotive to aerospace, from material synthesis to chemical reactors. The proposed seminar will presents the results of experimental campaigns performed in modified-gravity conditions (ESA SpinYourThesis! 2013, ESA 58th Parabolic Flight Campaign). A bottom heated mode PHP has been investigated under different gravity levels (from 0.01g to 20g) and different heat loads (form 50W to 100W). For the first time, a capillary PHP with circular cross section channels, equipped with 14 thermocouples and a pressure transducer has been fully thermally characterized in such conditions. In addition, comparisons with a novel lumped parameters model are presented. It allows the simulation of two-phase passive thermal systems using an advanced numerical technique to allow fast simulations extending sensitivity analysis and device designs. Even if lumped parameter models are not a novelty for such systems, for the first time this kind of numerical tools has been applied to simulate transient operative conditions removing physical simplified assumptions and embedding directly phase changes processes. Besides, the code has two main blocks: an Eulerian model for the external tube and a Lagrangian model for the internal two-phase flow. A dedicate matrix allows communication between fix and moving domains. Mass, momentum and energy balances are solved for liquid and vapor through a novel hierarchical method: for each time step, first the heterogeneous phase changes are solved, then the homogenous evaporation/condensation phenomena through the interface are accounted for, finally all the other phenomena (e.g. sensible heat exchange with the wall, axial conduction, etc.) are computed. The final mathematical models results in an ODE system which is solved numerically by means of a blocked algorithm consisting of a combination of Adams Bashforth methods of order one and two with the Störmer-Verlet method. The use of the latter method for the discretization of the system block descending from the momentum equation was driven by its well known properties in capturing the long term dynamics of second order ODE systems in an accurate and stable way. Such choice, which represents a novelty with respect to the previous simulation tools, has been shown, by means of numerical experiments to be extremely effective in preventing unphysical phenomena such as the overlapping of adjacent fluidic domains. For the implementation of the numerical method an interpreted code written in GNU Octave was chosen in order to allow for agile development of model modifications and extension; yet, by suitably optimizations, its efficiency was enhanced to reduce global simulation time to a level comparable with more complex tools. This new code showed a very high potentiality, being able to reproduce with good accuracy the experimental recorded data both in steady and transient conditions.