The increase of reliability of high-speed electrical machines is a crucial goal in the industrial point of view. The use of electrical equipment undergoing hard operating conditions (rotational speed, heat dissipations) leads to the development of efficient, passive and reliable device, capable of extracting heat from those systems. As well as being an excellent passive system transferring large quantities of heat, the axial rotating heat pipe satisfies all requirements because of its reduced size and small working fluid loads.
First, we proposed the improvement of existing experimental set up to characterise these two phase-flow devices operating under very high radial acceleration levels. Many parameters are involved in the behaviour of such a complex system: inside geometry, nature and charge in two-phase fluid and external surroundings (transient dissipations, range of temperature of cold source...). Evaporator dissipations were provided by induction while cooling of condenser achieved by air flow in existing high security area. Temperature evolutions of wall heat pipe were measured by Infra-red cameras and heat balances were made at several levels (inductor, heat pipe, cooling air flow…). Different filling ratios and geometry heat pipes were investigated to get deep understanding of heat transfer performances from low to high rotational speed.
In the same time and after a strong state of the art of the modelling of transfer in heat pipe at high rotational speed, numerical modelling was performed at different levels: microscopic liquid/vapour level (finite volume model) then at system level by building and validate nodal networks to reach the objective of certifying the performances of each model approach versus experimental results.
Thanks to these developments, we were able to propose optimisation of high rotational heat pipes for heat transfer in motorised turbomachine context.
An experimental bench was developed for testing the two-phase system under the conditions which are the most similar to the real operating conditions. This bench constituted the demonstrator of the project.
The investigational testing of the smooth heat pipe showed that this system is effective in difficult operating conditions. In addition, a parametric study was carried out to determine the influence of several parameters: the nature of the fluid, water is slightly more effective than ethanol; the filling ratio; over the range studied, the filling rate has little impact on heat pipe performances; the rotational speed, it does not disturb the general operation of the heat pipe; cooling system of condenser, air blast over the condenser improves thermal performances especially at low speed of rotation.
The overall thermal performances of the first smooth rotating heat pipe are very interesting. The nodal model has enabled to point out that the condenser area chiefly is limiting the heat transfer. Indeed, thermal conduction in the liquid film thickness in the condenser area decreased efficiency of two-phase systems. To reduce this phenomenon, the internal geometry of the rotating heat pipe is grooved, grooves acting as fins, and the thermal conductivity of the stainless steel wall is 25 times greater than that of water.
The influence of several parameters has been studied on grooved rotating heat pipe: nature of fluid, rotation speed, power applied and filling mass. The nature of fluid (water, ethanol, FC72) influences weakly the performances of grooved rotating heat pipe. Rotation speed does not influence the system except for rotation speeds lower than 10,000 rpm; below this rotation speed, a part of evaporator area was dried out. The filling mass is the parameter most influencing. Indeed, when it is higher than 100%, temperature oscillations appear. These oscillations seem to be outcome of the non axiality of rotating heat pipe; the period of these oscillations correspond to the rotation speed. The best performances were observed for a filling mass near 50%.
The thermal performances of both rotating heat pipes are very interesting. Furthermore, the second rotating heat pipe which grooved is even more efficiency.
From a numerical point of view, a solver has been developed with the following new features: a new phase change model and the multi-domain. The new phase-change model is a kinetic theory-based model.
The fully three-dimensional smooth rotating heat pipe was simulated using all the necessary recipe ingredients but the simulation time was too prohibitive (in a 40-processor parallel run, simulating 1 physical second took about 7 computation days) to tackle correctly both the complete unsteady regime, after the rotating heat pipe load is applied, and the pseudo-steady state.
An axi-symmetric configuration of the same heat pipe was then considered in order to reduce the computation time. Here, the simulation time was reduced to about 24 computation hours in order to simulate typically 5 physical seconds. This allowed us to unfold the entire unsteady regime, but there the heat pipe was not able to reach a steady state at all, which prevents valuable results to be obtained and useful conclusions to be drawn as the effect of the speed of rotation, wall thickness and shape, wettability, filling ratio and other fluid and solid properties on the overall performance of the rotating heat pipe.
Besides the development of numerical models and tools, this important work allowed us identifying the main modelling issues linked with the complex phenomena involved in rotating heat pipes. Some of these were addressed (interface conductivity, spurious current, curvature interface accuracy) but others, more complicated (mass conservativeness, flow stability issues), have not been definitely solved during the HIROPEAM project time period. These issues have to be dealt with necessarily but out of the present framework.
In addition, an elaborated liquid film model was developed and simulated. In contrast with the classical and usual lubrication theory, the full Navier-Stokes equations were used to model the liquid flow in order to take into consideration the buoyancy effect under centrifugal effects. The Arbitrary Lagrangian-Eulerian (ALE) method was used in order to describe the moving liquid-vapour interface where capillary and thermocapillary forces are considered. In this case, two main phenomena may compete: the stabilizing centrifugal effect, which tends to maintain the liquid phase farther away from the axis of rotation than the vapour phase and therefore flattens the liquid-vapour interface, on the one hand, and the destabilizing buoyancy and thermocapillary effects, which tend to induce Rayleigh-Bénard/Marangoni instabilities, on the other hand. It was found that mainly Rayleigh-Bénard destabilizing effect significantly affects the system for moderate rotational speeds (~4000 rpm), where convective rolls appear at the evaporator and both: i) disturb the drainage of the liquid from the condenser zone to the evaporator zone and, ii) enhance liquid mixing. An analysis of dependence of the RHP conductance on the rotational speed was also performed. In was shown that the larger the rotational speed, the larger the RHP conductance, other things being equal, with two distinguishable operating regimes; a moderate rotational speed regime ( 4000 rpm), for which significant change in RHP conductance, as the rotational speed increases, is noticed. Note that no of these flow patterns can be captured by the classical lubrication models.
Connection between the liquid film flow pattern and the RHP performance was strongly suggested.
In fact, for the low rotational regime, two convective rolls (one at the evaporator/adiabatic transition region and the other at adiabatic/condenser transition region) get more intense for increasing rotational speed. Nevertheless, the localness of these rolls makes them not an effective contributing factor on the RHP performance. For higher rotational regime, substantial enhancement in rotating heat pipe performance was shown to be influenced also by the occurrence of instabilities at the evaporator area, where several convective rolls develop all along and so mixing is significantly increased. The Critical Rayleigh number associated with this instability was found to be close to 400, which corresponds (for an applied heat power set to 25 W and a film mean thickness about 1 mm) to a rotational speed about 4,000 rpm. So reducing the film thickness by a factor 2 (and so reducing the mass of the working fluid by approximately the same factor) makes instability happen for 16,000 rpm. One must then keep in mind that, as far as RHP is concerned and in addition to drying out increasing risk, it is not straightforward that reducing the film thickness will systematically reduce the film resistance, since convective effects may strongly interfere with pure conductive ones.
Intensification of efforts on the experimental aspect (expertise, more complementary tests) has led today to geometry of rotating heat pipe very effective and producing noticeable performance enhancing. In the near future, the numerical tools to be developed will guide a further optimisation of the rotating heat pipe characteristics and so will help an even more enhancing of its performances.
The numerical developments done so far in this project were disseminated though one national
Conference, one international conference, and one article submitted to peer-reviewed journal, up to now under review.
As a perspective, the numerical tool that was developed in the HIROPEAM framework has to be enhanced in the near future in order to:
- Consider in a more reliable and physical way the critical issue of mass conservation;
- Optimise robustly the parallel processing of the solver in order to handle truncated (and may be full) three-dimensional grooved heat pipe;
- Reduce even more the spurious current and increase accuracy in evaluating the liquid-vapour interface curvature and normal direction;
- Include a more realistic wettability, i.e. contact angle, physics (may be in a multiscale manner).