Overview
The problem of finding a fluid for two phase capillary pumped cooling systems is a multiobjective one, considering the numerous properties to match. Besides we can expect to find a fluid within the tens of thousands of existing molecules, which have not been tested for this application, or a fluid that could be easily synthesised from an existing fluid. The critical point is finding the fluid among databases and/ or identifying chemical functions that would enable an existing fluid to match the specifications. Finally, we can expect mixtures to be suitable as much as pure fluids.
Rather than undertaking an inefficient trial and error search, we proposed to implement a computer aided molecule and mixture design approach based on reverse engineering. Such a strategy combines bottom-up and top-down approaches to find fluids that can match a large set of specification together.
The first task (WP1) consisted in building a mathematical performance function encompassing all the property specifications and screen potentially interesting chemical families. Second, a systematic computer-based search was run to issue a list of candidate fluids (WP2). It combined two existing computer tools from the partners: a bottom-up approach to account for feasible chemical synthesis pathways, and top-down search based on group contribution property estimation methods to explore new pure compounds and mixtures. Third, the candidate list was narrowed by refining property calculations by using accurate first principle methods (WP3). They also provided an electronic signature of the ideal fluid. The fourth task concerned the fluid choice and is possible synthesis (WP4). Fifth, experimental measurements were performed to validate a few candidates (WP5).
Funding
Results
Executive Summary:
The DIPHASICFLUID project aimed at finding new fluids for two phase capillary pumped cooling systems (CPL). Those systems being used under severe conditions during operation, from very cold weather situations to very hot conditions, the specifications that the fluid must meet are very challenging with respect to phase transition temperatures. Properties to be considered are also numerous and the problem to be solved ends up being a multi-objective one. Among the phenomena to be taken care of, heat transfer is related to the vaporization enthalpy and the heat capacity; viscosity and density are related to the fluid flowing through the pipes; surface tension is related to the ability to wet the porous wick where heat transfer occurs and to pump the fluid. None of the existing fluids (acetone, methanol, ethanol, ammonia for instance) are satisfactory. Besides those fluids also display inappropriate effects in terms of toxicity, safety, which must be improved and add new products requirements.
As an all-around search in databases and suppliers portfolio could be very time consuming and unable to cover all properties at the same time, the project implemented a strategy for finding solutions based on the use of property prediction tools and first-principle thermodynamic models, to orient the search towards suitable chemical families. Since it was forecasted that pure compound fluids may not be found, mixture candidates were also looked at. This defined an additional challenge because mixtures are reputedly exhibiting non ideal behaviour which can be described only by nonlinear property predictions models and those are difficult to run in a large scope search. The project was split in several tasks. The first task (WP1) consisted in building a mathematical performance function encompassing all the property specifications and in screening out unsuitable chemical families in order to define an electronic signature of the ideal fluid. Second, a systematic computer-based search was run to obtain a list of candidate fluids (WP2). It combined two existing computer tools: a bottom-up approach that constructs molecules with feasible chemical synthesis pathways starting from a core molecule; and a top-down search based on group contribution property estimation methods to explore new pure compounds and mixtures built from a pool of chemical fragments. The third task (WP3) intended to narrow the candidate list by updating the performance function of each candidate through the refining of property values either extracting them from databases or predicting them with more accurate first-principle methods. The fourth task concerned the fluid choice (WP4). The fifth task was devoted to experimental measurements of selected fluids (WP5). A coordination tasks (WP6) oversaw the whole project and took care of the dissemination issues.
The project has been successful. With the help of the efficient strategy combining computer-based calculation and experiments, several tens of thousands of molecules covering all chemical families were screened. Not a single molecule achieved the maximum performance regarding all properties, but suitable chemical families were identified, and drawbacks of the most promising single molecules were pointed out. Hence, mixtures were screened with the idea of benefiting from synergies in the mixture where one compound property could compensate the far-from-target property value of another compound in the mixture. With that in mind, several hundred mixtures were imagined, their property predicted and verified through experiments. Ultimately, the project found several fluid mixture candidates which performance was validated. Additionally, new cooling fluids suitable for other applications have been found. Both fluids are undergoing patenting at the time of the project end and could be used in future exploitation.