Available online at www.sciencedirect.com ScienceDirect Energy Procedia 101 (2016) 758 – 765 71st Conference of the Italian Thermal Machines Engineering Association, ATI2016, 14-16 September 2016, Turin, Italy Rotating heat transfer measurements on realistic multi-pass geometry Fabio Pagnaccoa, Luca Furlania, Alessandro Armellinia, Luca Casarsaa*, Anthony Davisb a b Università degli Studi di Udine, via delle Scienze 206, Udine 33100, Italy Siemens Industrial Turbomachinery Ltd, Ruston House, Waterside South, Lincoln LN5 7FD, England Abstract In this contribution, a novel rig was used to assess the heat transfer coefficients on a full internal multi pass cooling scheme Transient liquid crystal technique was used for the measurement of the heat transfer coefficient (HTC) on channel’s internal surfaces A first set of experiments were performed at engine similar conditions of Re=21000 and Ro=0.074 In order to assess the reliability of the measurement methodology and to explore the thermal behavior at higher rotation numbers, tests were also carried out at Re=17000 and Ro=0.074-0.11 From the spatially resolved HTC maps made available, it is possible to characterize the thermal performances of the whole passage and to highlight the effect of rotation © Published by Elsevier Ltd This ©2016 2016The TheAuthors Authors Published by Elsevier Ltd is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/) Peer-review under responsibility of the Scientific Committee of ATI 2016 Peer-review under responsibility of the Scientific Committee of ATI 2016 Keywords: blade cooling; internal cooling; transient liquid crystal; rotation; multi-pass Introduction Safe and efficient operation of gas turbine engines strongly relies on effective blade cooling Consequently, the performance required to the cooling systems is ever increasing, which leads to the definition of passage geometries that are increasingly complex A commonly adopted solution is a multi-pass geometry with different types of turbulent promoters used to enhance heat transfer The choice of which kind of turbulator to install is constrained by the channel * Corresponding author Tel.: +39 0432 558010 E-mail address: luca.casarsa@uniud.it 1876-6102 © 2016 The Authors Published by Elsevier Ltd This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/) Peer-review under responsibility of the Scientific Committee of ATI 2016 doi:10.1016/j.egypro.2016.11.096 Fabio Pagnacco et al / Energy Procedia 101 (2016) 758 – 765 cross section dimensions and aspect ratio, which in turns depends upon the blade region where the passage is located Several studies have been carried out in order to define the behavior and the efficiency of different turbulent promoters [1] Another important aspect that has to be taken into account in the design process of a cooling passage is the combined effect of turbulators, rotation, and channel orientation Fundamental contributions are given by Hart [2], Lezius and Johnston [3], Speziale [4], and Speziale and Thangam [5] These contributions describes the Coriolis effects on the flow field inside basic channel geometries (square or rectangular channels with outward flow in orthogonal rotation, i.e the rotation axis is parallel to the channel height) In addition, in real turbines blades, channels are not always in orthogonal rotation, and since gas turbine blades are manufactured by casting, it is difficult to ensure that the internal passages have sharp edges or perfectly squared ribs Experimental studies have been performed in order to study the heat transfer behaviour in channels rotating with different orientations with respect to the rotation axis [6] Only few contributions can be found in literature regarding the behaviour in terms of enhancement and pressure losses of realistic turbulators geometries and all demonstrated a deterioration in heat transfer due fillets at the base of the rib [7-9] or pin-fin [10] It should be therefore clear that the complexity of the modern internal cooling designs makes difficult to predict the expected thermal performances starting from the available knowledge Indeed, the majority of the data refers to stereotyped geometrical configurations or the results cannot be projected straightforward to the engine because of lack of full similarity between experiments (real or virtual) and real engine conditions The present work investigates the rotational effects inside a three-pass channel for cooling of land based gas turbine engine The geometry still brings some simplifications with respect to the real application, but the main features are considered, such as realistic cross section with rounded walls, angled ribs and filleted pin fins Channel orientation is consistent with the real application as well as the imposed flow split among the different coolant discharges Nomenclature Bo d dh HTC k Nu Nu0 Pr R Re buoyancy parameter pin fin diameter hydraulic diameter heat transfer coefficient air thermal conductivity Nusselt number reference Nusselt number Prandtl number rotation radius Reynolds parameter Ro Tb The Ti Tw Ub µ ρ Ω Rotation parameter bulk flow temperature temperature at the outlet of heat exchangers initial flow temperature wall temperature bulk velocity air dynamic viscosity air density rotational speed Experimental setup 2.1 Channel geometry The test section is a model of a full internal cooling scheme of an idealized blade Figure reports a drawing of the test section and a view of its installation on the rig The cooling scheme has a multi-pass design First two passages are two-sided ribbed channel with ribs arranged in a line-on-line configuration At the blade tip, part of the coolant is discharged through a dust hole (10% of inlet mass flow rate), while the greater part of it goes through a 180° bend and flows radially inwards inside the 2nd passage characterized by a higher aspect ratio cross section with respect to the first one At the end of the second passage, another 180° bend diverts the flow inside the rd leg, where low aspectratio cylindrical pin fins (d/dh = 0.34) are arranged in a staggered configuration to promote flow turbulence Inside the 3rd passage, the coolant is progressively discharged through holes at the blade trailing edge (55% of inlet mass flow rate) and finally at the blade tip (35% of inlet mass flow rate) The geometrical features of the ribs used in the st and 2nd legs are: square section with rounded edges and base fillet, rib pitch/height of and inclined at 60° with respect to 759 760 Fabio Pagnacco et al / Energy Procedia 101 (2016) 758 – 765 Fig Test section sketch (dimensions in mm) the radial direction Their blockage effect inside both st and 2nd leg is about 10% The test model (made of 20mm thick plexyglass) differs with respect to the real design only because there is no curvature due to the blade camber Conversely, the correct stagger angle of the real blade (60° with respect to the peripheral velocity) is preserved 2.2 General overview of the test rig In the following, only a general description will be provided, the details about all rig components, operation and procedure are provided in [11] A view of the rig and a simplified sketch of the air circuit are shown in Figs 2a and 2b, respectively The test section is installed on one side of the facility rotating arm while on the opposite side counterweights are used in order to balance the whole structure during rotation A kW electric motor is fastened to a load bearing structure and drives the rotating arm up to 400 rpm Flow is drawn through the rotating test section by a six stage centrifugal fan that sucks air from the test section through the rotating joint The air flow rate is continuously monitored by an orifice flow meter installed in between the rotating joint and the fan Process flow temperature is lowered by means of fin and tube heat exchangers supplied with liquid nitrogen This allow to generate a sudden drop of flow temperature which is required by the application of the measurement technique and to maintain the correct behaviour of the buoyancy forces [11] A correct flow split between the model multiple discharges must be ensured during both stationary and rotating test This constrain is satisfied by a careful design of the air circuit (made of multiple volumes, control valves and (a) (b) Fig (a) Test rig; (b) Air circuit sketch 761 Fabio Pagnacco et al / Energy Procedia 101 (2016) 758 – 765 flow measurements stations) that allows to set the correct mass flow split in static condition and to keep it constant in rotation independently from the rotational velocity [11] Due to the very large variation of flow temperature that occurs during the test phase, physical properties of the cooled fluid change, which causes a variation of both Re and Ro parameters In order to maintain constant Re and Ro, a feedback control system has been developed This system has the task to control simultaneously the air-cooling flow rate and the rotational speed of the test section [11] 2.3 Measurement technique, test condition and data reduction Heat transfer measurements on the test section are carried out by means of thermochromic liquid crystals with the well-known transient technique [13-15] This technique is based on the computation of the heat equation solution on the measurement surface Duhamel superposition approach is used due to the non ideal temperature step of the fluid, hence: ܶ௪ ൌ ܶ σୀଵ൫ܶǡ െ ܶǡିଵ ൯ ൈ ቈͳ െ ݁ ݔቀ డమ ் డ௫ మ మ ሺ௧ିఛ ሻ ఘ ቁ ݁ ݂ܿݎቆට మ ሺ௧ିఛ ሻ ఘ ቇ (1) Bulk temperature is computed in according to [14] using the Laplacian diffusion equation డమ ் డ௬మ ൌ Ͳ (2) More details about the specific methodology used in this work is reported in [12] Detailed heat transfer maps have been obtained on both pressure and suction sides of the channel In a first experimental campaign tests have been carried out at Re = Ubdhρ/μ = 21000 and Ro= Ωdh/Ub = and 0.074 In order to explore the thermal behavior at higher rotation numbers, a second measurement campaign was performed at Re = 17000 and Ro = - 0.074 - 0.11 All test parameters are defined at test section inlet Both Reynolds and rotation numbers values are consistent with those that can be found inside land based gas turbines Concerning the buoyancy parameter the definition of a reference Bo = ((Tw-Tb)/Tb)Ro2(R/dh) of the tests is not straight forward in view of the continuous change in fluid and wall temperature, both during time and along the passages, as imposed by the transient measurement approach However, for the present experiments, buoyancy effects are expected to be negligible The computed heat transfer coefficient distributions are non-dimensionalize as Nusselt number (Nu=HTCdh/k) where the specific hydraulic diameter of each legs was used as reference length for the calculations and the thermal conductivity of the fluid was modelled at every positions considering the time-averaged fluid temperature evolution until the liquid crystal indication took place [11] Data normalization is performed using as reference Nusselt number the smooth channel value from classical Dittus-Boelter correlation The Nu0=0.023Re0.8Pr0.4 value is computed at the inlet of the test section and kept constant throughout the channel Dedicated tests have been also performed in order to asses the measurments accuracy [12], which allow to estimate the current data uncertainty at 10% Results 3.1 Flow condition Re = 21000, Ro = 0-0.074) Figure reports the Nu distribution maps on both PS and SS obtained for the static channel condition at Re=21000 At first, an overall agreement between data on PS and SS can be appreciated, as expected in view of the perfect symmetry of the passage Concerning the 1st leg, the typical path of the heat transfer field promoted by the inclined ribs is observed: regions of strong enhancement (up to 4.5) are found downstream of each ribs, an overall enhancement with respect to the smooth channel case is observed on the whole channel surface thanks to the turbulent mixing produced by the ribs [16,17] The same enhancement (about 1.5) is also found in the first 180° turn, at least in the measured area (the outer portion of the curve was not optically accessible) Inside the nd leg, the behaviour is similar but affected by the flow evolution inside the bend 180° turns without vanes (as in the present case) determine an acceleration of the fluid layer on the external side of the path and flow separation close to inner corner [18,19] In the present data, this is shown by: the low Nu values found close to the lateral wall that separates from the 1st leg, the minimum being located close to the U-bend corner; by the higher extension of the separated flow region downstream 762 Fabio Pagnacco et al / Energy Procedia 101 (2016) 758 – 765 (a) (b) Fig Nu/Nu0 maps on pressure (a) and suction (b) sides of the channel for the static case of the first ribs of the leg (wider high Nu areas, which are associated to the higher flow velocity) Downstream the turn, the flow will progressively recover from separation and it will develop towards a more uniform flow distribution over the channel span Inside the 3rd leg, again flow separation is found close to the corner of the bend (very low Nu values that are found around the first pin fin (Fig 3)) The Nu values are higher at the beginning of the leg and they lower moving towards the blade tip due to the progressive flow discharge at the trailing edge that reduces the local Reynolds number, in agreement with previous observations [10] The literature about pin fin roughened channels with lateral discharge [10,20,21] reports a local augmentation of heat transfer coefficient in proximity of each exhaust hole/slot, due to the local flow acceleration In the present channel configuration, the number of discharge holes is very high (40) and the local amount of exhausted flow rate is consequently low Local flow acceleration in the hole proximity is therefore not that strong to leave a footprint on the heat transfer field The same main features that were commented about the static case characterize the rotating heat transfer Nusselt distributions (Fig 4) However, the expectation is to find differences in some local features that can be directly related to rotational effects Briefly, by comparing static and rotating data with reference to the available literature about rotating channels [4-5] one should expect: - higher Nu in rotation on PS of 1st and 3rd leg, and on SS of 2nd leg; - lower Nu in rotation on SS of 1st and 3rd leg, and on PS of 2nd leg These effects are only partially confirmed by a qualitative comparison of Fig with Fig A better insight can be obtained by analysing the data from an area-average point of view Figure confirm an overall reduction of heat transfer on the suction side of the first leg, conversely an augmentation of heat transfer is highlighted in the correspondent pressure side, the opposite behavior is confirmed for the second passage However, the observed differences between static and rotating data are lower than expected if compared with the available literature [22] A possible explanation is provided in [12] and is ascribed to the channel stagger angle (not in orthogonal rotation) and to the particular channel cross section The wider aspect ratio cross section of the nd leg with respect the 1st one, weakens even more the rotational effects (a) (b) Fig Nu/Nu0 maps on pressure (a) and suction (b) sides of the channel for the rotating case Fabio Pagnacco et al / Energy Procedia 101 (2016) 758 – 765 Fig Detailed distribution of Nu/Nu0 area averages for each inter-rib region: Re = 21000 The thermal behavior of the 3rd leg is not affected by rotation (compare Fig with Fig 4), which does not agree with available literature on a similar geometry [10] This different behavior may be due to the different inlet conditions to the pin fin channels (smooth and radial for [10], 3rd passage fed by a 180°bend in the present case) and the higher Ro that characterizes the analysis performed in [10] 3.2 Flow condition Re = 17000, Ro = 0- 0.074-0.11 In order to investigate higher rotation conditions, due to actual rig limitations, it has been necessary to reduce the coolant mass flow rate, hence the Reynolds number at the section inlet at Re = 17000 Unfortunately, by doing this only inside the 1st leg a complete activation of liquid crystals is achieved For this reason all further analysis and comparisons are limited to the 1st leg of the test section Figure shows the Nusselt contour plots inside the 1st leg on both pressure and suction side for Ro = 0-0.074-0.11 A first look confirm the different behaviour on pressure and suction side due to rotation As it can be seen on the suction side, high Nusselt areas decreases increasing the rotation number, on the contrary on pressure side at first sight there are no large variations in the contour plots In order to better highlight the Nusselt trends along the channels area averaged values are represented in Fig 6-g Fig Nu/Nu0 maps for Re = 17000: PS-Ro=0 (a), SS-Ro=0 (b), PS-Ro=0.074 (c), SS-Ro=0.074 (d), PSRo=0.11 (e), SS-Ro=0.11 (f), area averaged values (g) 763 764 Fabio Pagnacco et al / Energy Procedia 101 (2016) 758 – 765 This analysis, as well as confirming the above considerations, reveals a seemingly counterintuitive effect On suction side Nu/Nu0 values decreases monotonically increasing the rotation number, as normally expected On the contrary, on pressure side a non-monotonic trend is observed: Nu values shows a general increase when moving from static to the lower rotation condition (Ro=0.074) as expected, while a further increase of rotational speed (from Ro=0.074 to 0.11) does not determine higher Nu but values that are comparable to the static case The same phenomenon was observed on the pressure data acquired along the channel on the pressure side In order to carry out these measurements, pressure taps were installed at the beginning and end of each leg By the use of a 16 channels pressure scanner (NetScanner System mod.9116) (see [11]), it was possible to acquire the pressure values at different test conditions Test conditions were chosen in order to replicate test conditions of thermal tests; given that these tests were performed at a later time, it was not possible to replicate the test condition of Re=17000 and Ro=0.11 because structural reasons This condition has been replaced with a pair Reynolds number and Ro=0.091 which still allows to perform a qualitative analysis of the pressure data Pressure data are summarized in figure As it can be seen pressure data at Re=17000 highlight the same behavior of Nu/Nu0 data presented in figure 6-g Since the pressure field is generated by the velocity field, which is the same that influences the thermal field it is evident that the two behaviors should be similar It is clear that the described behavior is dominated by rotational effects that are not easy to be argued in view of the complexity of the geometry and working condition (non-orthogonal rotation) A close look to the flow behavior (either by experiment or CFD) it is mandatory to gain more insight about the aero-thermal behavior of the device Fig.7 Pressure distribution along the channels Conclusions The contribution report the results obtained on a new rig that has been specifically designed to perform heat transfer measurements inside complex models of cooling channels for gas turbine blades The facility has been designed in order to accommodate large test articles, to perform measurements at both static and rotating condition by exploiting the capabilities of the liquid crystal thermography in transient approach The investigated geometry is a three-pass channel, rib roughened in the first two legs and provided with short pin fins and distributed coolant discharge in the 3rd leg Geometrical features such as channel cross section, rounded edges, contoured rib, and filleted pin fins have been used in order to resemble as close as possible a realistic geometry A realistic channel orientation was also considered by spinning the test section at 60° with respect to the peripheral velocity The results here reported are the first obtained with this new rig, and therefore further efforts have to be put in the investigation of this complex geometry The preliminary results however, shows already a number of important features of the thermal field and highlight non trivial rotational effects Fabio Pagnacco et al / Energy Procedia 101 (2016) 758 – 765 References [1] Han JC, Zhang YM, Lee CP Augmented heat transfer in square channels with parallel, crossed, and V-shaped angled ribs Journal of Heat Transfer 1991; Vol.113, p.590-596 [2] Hart JE Instability and secondary motion in a rotating channel flow J Fluid Mech 1970; Vol 45, pp 341–351 [3] Lezius DK, Johnston JP Numerical study of viscous flow in rotating rectangular ducts J Fluid Mech 1976; Vol 77, pp 153–176 [4] Speziale CG Numerical study of viscous flow in rotating rectangular ducts J Fluid Mech 1982; Vol 122, pp 251–271 [5] Speziale CG, Thangam S Numerical study of secondary flows and roll-cell instabilities in rotating channel flow J Fluid Mech 1983; Vol 130, pp 377–395 [6] Dutta S, Han JC Local Heat Transfer in Rotating Smooth and Ribbed Two-Pass Square Channels with Three Channel Orientations ASME J Heat Transfer 1996; Vol 118, pp 578-584 [7] Bailey JC, Bunker RS Heat Transfer And Friction in Channels With Very High Blockage 45° Staggered Turbulators Proc ASME TurboExpo 2003; GT2003-38611 [8] Taslim ME, Spring SD Effects of Turbulator Profile and Spacing on Heat Transfer and Friction in a Channel AIAA J Thermophysics and Heat Transfer 1994; Vol 8(3), pp 555-562 [9] Chandra PR, Fontenot MC, Han JC Effect of Rib Profiles on Turbulent Channel Heat Transfer AIAA J Thermophysics and Heat Transfer 1998; Vol 12, pp 116-11 [10] Rallabandi AP, Liu YH, Han JC Heat transfer in trailing edge wedge-shaped pin fin channel with slot ejection under high rotation numbers Proc ASME Turbo-Expo 2010; GT2010-22832 [11] Pagnacco F, Furlani L, Armellini A, Casarsa L, Davis A Rotating heat transfer measurements on a multi-pass internal cooling channel – I Rig Development Submitted to ASME Turbo Expo 2016; GT2016-56308 [12] Pagnacco F, Furlani L, Armellini A, Casarsa L, Davis A Rotating heat transfer measurement on a multi-pass internal cooling channel - II Experimental tests Submitted to ASME Turbo Expo 2016; GT2016-56307 [13] Ekkad S, Han JC A transient liquid crystal thermography technique for gas turbine heat transfer measurements Meas Sci Technol 2000; Vol 11, pp 957-968 [14] Poser R, von Wolfersdorf J, Lutum E Advanced evaluation of transient heat transfer experiments using thermochromic liquid crystals Journal of Power and Energy 2007; Vol 221 DOI: 10.1243/09576509JPE464 [15] Ireland PT, Jones TV Liquid crystal measurements of heat transfer and surface shear stress”, Meas Sci Technol 2000; Vol 11, pp 969986 [16] Zhou F, Acharya S Heat Transfer at High Rotation Numbers in a Two-Pass 4:1 Aspect Ratio Rectangular Channel with 45 deg Skewed Ribs ASME J Turbomachinary 2008; Vol 130, pp 021019-1-021019-2 [17] Huh M, Lei J, Han JC Influence of Channel Orientation on Heat Transfer in a Two-Pass Smooth and Ribbed Rectangular Channel (AR=2:1) under Large Rotation Numbers ASME Paper 2010; No GT2010-22190 [18] Lei J, Su P, Xie G, Lorenzini G The effect of a hub turning vane on turbulent flow and heat transfer in a four-pass channel at high rotation numbers International Journal of Heat and Mass Transfer 2016; Vol 92, pp 578-588 [19] Lei J, Li SJ, Han JC, Zhang L, Moon HK Heat transfer in rotating multi-pass rectangular smooth channel with and without a turning vane in hub region Proc ASME Turbo-Expo 2013; GT2013-94269 [20] Willett F, Bergles A Heat transfer in rotating narrow rectangular pin-fin ducts Experimental Thermal and Fluid Science 2002, Vol 25(7), pp 573–582 [21] Chang S, Yang T, Hong GF, Liou T Heat Transfer in Radially Rotating Pin-Fin Channel at High Rotation Numbers Proc ASME TurboExpo 2008; GT2008-50514 [22] Wagner JH, Johnson BV, Graziana RA, Yeh FC Heat Transfer in Rotating Serpentine Passages with Trips Skewed to the Flow ASME J Turbomachinery 1994; Vol 116, pp 113-123 765