- Email: [email protected]
Numerical analysis for impacts of nozzle end-clearances on aerodynamic performance and forced response in a VNT turbine
CJA 1009 24 February 2018 Chinese Journal of Aeronautics, (2018), xxx(xx): xxx–xxx
No. of Pages 11
1
Chinese Society of Aeronautics and Astronautics & Beihang University
Chinese Journal of Aeronautics [email protected] www.sciencedirect.com
5
Numerical analysis for impacts of nozzle endclearances on aerodynamic performance and forced response in a VNT turbine
6
Ben ZHAO a, Leilei WANG b,*, Hongjun HOU b
3
4
7 8
9
a b
Michigan State University, East Lansing 48824, USA Hebei University of Engineering, Handan 056038, China
Received 26 April 2017; revised 20 June 2017; accepted 17 July 2017
10
12 13
KEYWORDS
14
Clearance; Forced response; Mode analysis; Performance; Reliability; Variable Nozzle Turbine (VNT)
15 16 17 18 19 20
21
Abstract It has been well known that nozzle end-clearances in a Variable Nozzle Turbine (VNT) are unfavorable for aerodynamic performance, especially at small openings, and efforts to further decrease size of the clearances are very hard due to thermal expansion. In this paper, both the different sizes of nozzle end-clearances and the various ratios of their distribution at the hub and shroud sides were modelled and investigated by performing 3D Computational Fluid Dynamics (CFD) and Finite Element Analysis (FEA) simulations with a code of transferring the aerodynamic pressure from the CFD results to the FEA calculations. It was found that increasing the size of the nozzle end-clearances divided equally at the hub and shroud sides deteriorates turbine efficiency and turbine wheel reliability, yet increases turbine flow capacity. And, when the total nozzle endclearances remain the same, varying nozzle end-clearances’ distribution at the hub and shroud sides not only shifts operation point of a VNT turbine, but also affects the turbine wheel vibration stress. Compared with nozzle hub clearance, the shroud clearance is more sensitive to both aerodynamic performance and reliability of a VNT turbine. Consequently, a possibility is put forward to improve VNT turbine efficiency meanwhile decrease vibration stress by optimizing nozzle end-clearances’ distribution. Ó 2018 Production and hosting by Elsevier Ltd. on behalf of Chinese Society of Aeronautics and Astronautics. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/ licenses/by-nc-nd/4.0/).
* Corresponding author. E-mail address: [email protected] (L. WANG). Peer review under responsibility of Editorial Committee of CJA.
Production and hosting by Elsevier
1. Introduction
22
The past decade has seen that the emission standards for various types of vehicles have been becoming more stringent. To achieve less emission, car manufacturers are attempting to downsize an internal combustion engine that can provide the same power of a large engine by adding a boosting device, e.g. turbochargers.1 To better acceleration and deliver more power across a wider range, a Variable Nozzle Turbine
23
https://doi.org/10.1016/j.cja.2018.02.015 1000-9361 Ó 2018 Production and hosting by Elsevier Ltd. on behalf of Chinese Society of Aeronautics and Astronautics. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/). Please cite this article in press as: ZHAO B et al. Numerical analysis for impacts of nozzle end-clearances on aerodynamic performance and forced response in a VNT turbine, Chin J Aeronaut (2018), https://doi.org/10.1016/j.cja.2018.02.015
24 25 26 27 28 29
CJA 1009 24 February 2018
No. of Pages 11
2 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91
(VNT) has been developing and applied.2 A common method of variable nozzle in a radial turbine is the use of the pivoting vanes that are mounted on a flat plate. These vanes are connected with spindles, sometimes together with handles, to achieve the function of rotating axially. Between the flat plate and nozzle vanes, nozzle end-clearances have to be designed. Efforts to reduce the nozzle end-clearances’ size have been doing, and so far the size has been successfully decreased in advanced VNT turbos. But the unfavorable effects of the clearances are still significant and further decreasing the size will be very hard, since the nozzle vanes would get stuck due to thermal expansion. The temperature of exhaust gas is often more than 900 °C for a diesel engine and beyond 1000 °C for a gasoline engine. In addition, complicated sulfide from exhaust gas moves outward from wheel inducer to nozzles under the action of centrifugal force, and then is possible to stick the nozzle vanes as the clearances are too small. It has been reported that the presence of the nozzle endclearances usually has unfavorable effects on aerodynamic performance of a VNT turbine. Hu et al.3 performed numerical simulations on a VNT turbine to study the nozzle endclearances’ effect on turbine performances and revealed that the presence of nozzle end-clearances deteriorated turbine performances, especially at small open condition. Walkingshaw et al.4 carried out numerical simulations of the flow fields in a highly off-design VNT turbine and indicated that the leakage vortices changed the circumferential distribution of the approaching flow to the rotor and then deteriorated aerodynamic performance. The relationship between the nozzle endclearances and the turbine performance was also focused by Tamaki et al.5,6 Besides the aerodynamic performance, the high cycle fatigue is one of the critical issues that the development of an advanced VNT turbine has to address. Shock waves, one of the contributors to the high cycle fatigue failure, weigh heavily on the turbine wheel blade vibration and, sometimes, could seriously damage a turbine wheel, as shown by Chen.7 The intensity of the shock wave was proved to be closely related with nozzle end-clearances in a VNT turbine.8 Therefore, the nozzle end-clearances must have an effect on turbine wheel reliability. In fact, the nozzle clearance leakage flow is also able to directly interact with a turbine wheel and then causes it to vibrate.9,10 Though it has been reported that the nozzle end-clearances are closely related to the turbine performance, the relationship between the forced response of turbine wheel and the changed clearances, i.e. increased clearance size and different distribution, is not fully clear. If the nozzle shroud clearance has a significantly different effect on a VNT turbine from the hub clearance, it will provide a possibility to further improve a VNT turbine in both the aerodynamic performance and the reliability by optimizing the more sensitive clearance, instead of the hard effort to further decrease the size of total nozzle end-clearances. In this paper, a VNT turbine was first modeled and simulated by a 3D steady Computational Fluid Dynamics (CFD) method to validate computational mesh, and then the VNT turbine model with various nozzle end-clearances were predicted by both steady and unsteady CFD methods. After CFD simulations, the information about aerodynamic excitation pressure on a wheel blade surface was transferred from the CFD results to the Finite Element Analysis (FEA) calculation by interpolation so that the harmonic response could be
B. ZHAO et al. predicted and analyzed. Finally, an optimized nozzle endclearances’ distribution with high turbine efficiency and low forced response was suggested.
92
2. Research model
95
The research model is a VNT turbine. In order to reduce the requirement of CFD simulations, the volute was neglected. This simplification should have little influence on the conclusions stated in this paper, since the flow is guided and redistributed in crossing the nozzles. Therefore, the model used just consists of variable nozzles vanes and a mixed-flow rotor. Fig. 1 shows the meridional view of a VNT turbine with detailed illustrations of nozzle end-clearances, and two types of nozzle vanes. Inside the clearances, the spindle and the handle of a nozzle vane are not illustrated. The pivoting nozzle vanes are often fixed and controlled by a spindle or a combination of a spindle and a handle. If the spindle is located round the center of a vane, the handle will be not needed, whereas if it is located near trailing edge of a vane, the handle is often designed and installed near the vane’s leading edge to well control the vanes’ setting angle, as shown in Fig. 1. Regardless of the control styles, VNT turbine always requires the presence of nozzle end-clearances to achieve the relative movement between nozzle vanes and the flat plates (endwalls). Herein, the nozzle shroud clearance is marked with Cshroud and the clearance on another side is represented by Chub, as shown in Fig. 1. In order to clearly identify the research models used, the investigated VNT turbine with various nozzle end-clearances is numbered and listed in Table 1. The letter, h, stands for the normal height of the nozzle end-clearance at each side of a VNT turbine product, and is about 2 percent of the span of flow passage from the hub wall to the shroud one.
96
3. Numerical schemes
125
3.1. CFD model
126
A fully automatic hexahedral grid generator was used to generate the computational mesh for the whole flow field. Multiblock Structured Meshing with an O4H topology including 5 blocks was used to discrete the channels, as shown in Fig. 2. The skin block was an O-mesh surrounding the blade and the H-mesh was applied in inlet, outlet, up and down blocks. The mesh inside the hub and shroud clearances was created using a butterfly topology. The number of grid points had to be adjusted for various clearance sizes to keep a good mesh quality under the condition of the same first cell length. The subsonic inlet condition with total pressure, total temperature, velocity angles and turbulent viscosity were specified at the inlet boundary of the computational domain. At outlet boundary, a radial equilibrium technology was used. With this, the outlet static pressure was imposed on the given radius and integration of the radial equilibrium law along the spanwise direction permitted to calculate the hub-to-shroud profile of the static pressure. Therefore, a constant static pressure would be imposed along the circumferential direction. On the solid surface, the adiabatic walls were used and the velocity vector vanished. The rotational speed of the unsteady simulations
127
Please cite this article in press as: ZHAO B et al. Numerical analysis for impacts of nozzle end-clearances on aerodynamic performance and forced response in a VNT turbine, Chin J Aeronaut (2018), https://doi.org/10.1016/j.cja.2018.02.015
93 94
97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124
128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147
CJA 1009 24 February 2018
No. of Pages 11
Numerical analysis for impacts of nozzle end-clearances on aerodynamic performance
Fig. 1
Table 1
Illustration of nozzle end-clearances (at meridional view) and two types of nozzle vane.
Illustration of research models.
Model
Cshroud
Chub
Spindles
Case Case Case Case Case Case Case Case Case Case Case Case
100%h 100%h 0%h 50%h 100%h 150%h 0%h 20%h 40%h 60%h 160%h 200%h
100%h 100%h 0%h 50%h 100%h 150%h 200%h 180%h 160%h 140%h 40%h 0%h
Spindle Spindle & Handle – – – – – – – – – –
1 2 3 4 5 6 7 8 9 10 11 12
Fig. 2
148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168
3
was selected according to the results of the Campbell Diagram, and other boundary conditions of the unsteady simulations were obtained by running GT-Power software. There is one rotor/stator interface in the whole flow field. For steady simulations, the conservative coupling by pitchwise rows approach was used due to its capacity to provide an exact conservation of mass flow, momentum and energy through the interface. And the full non-matching boundary connecting algorithm was used in order to exchange the flow variables and fluxes between the left and right image. For unsteady simulations, the sliding mesh technology was applied on the rotor/ stator interface. The perfect gas relation and the Spalart-Allmaras turbulence model were employed to close the RANS/URANS equations. The discretization in space was based on a cell centered control volume approach. An explicit 4-stage Runge-Kutta scheme was applied to solve the equations mentioned above. For all the CFD simulations, the Courant-Friedrichs-Lewy (CFL) number was 3. And after calculating a set of computational meshes with the same topology and different number of grid points, the relationship of the mesh density and the chan-
Computational grids.
ged performance suggested the current computational mesh which was within affordable computation source limit. The y plus value was less than 3 in most region and, in nozzle endclearances, the average value was less than 2.68. For unsteady simulation, the total number of the grid nodes was about 4.53 million and the physical time step was about 1.15 108 s. Fig. 3 shows the validation of CFD methods which were performed on a similar VNT turbine. Although the deviation exists between the numerical and experimental data, they are still within an acceptable range.
169
3.2. FEA model
179
Commercial FEA software with linear analysis method was employed to predict the natural frequencies and harmonic response of a turbine wheel influenced by the unsteady aerodynamic excitation source. In order to reduce the requirement of FEA calculations, only one wheel blade was modelled. The tetrahedral elements were used to discrete the solid field, in which Inconel was assumed as an isotropic elastic material with a Yonge modulus
180
Please cite this article in press as: ZHAO B et al. Numerical analysis for impacts of nozzle end-clearances on aerodynamic performance and forced response in a VNT turbine, Chin J Aeronaut (2018), https://doi.org/10.1016/j.cja.2018.02.015
170 171 172 173 174 175 176 177 178
181 182 183 184 185 186 187
CJA 1009 24 February 2018
No. of Pages 11
4
B. ZHAO et al.
Fig. 3
Validation of the CFD method performed on a similar VNT turbine.
212
and a Poisson ratio. The detailed material properties were selected from a composite material database. The centrifugal force was applied to the FEA analysis, which was calculated according to angular velocity obtained from the Campbell diagram. An assumption of nodisplacement is applied at the center line of the wheel shaft. And a harmonic loading was applied on the wheel blade, which was calculated based on the unsteady CFD method. Once the natural frequencies of the turbine wheel were predicted by the modal analysis and the unsteady CFD simulations were carried out, the information about the aerodynamic excitation acting on the wheel blade was added from the CFD results to FEA analysis by interpolation. The detailed data transmission will be introduced in section of the forced response. The damping parameters significantly affect the accuracy of the vibration solution and the only way to obtain the precise damping parameters is through experimentation. Lu et al.11 reported that the maximum relative error for a computational method with a constant damping ratio was lower than 10 percent. A constant damping ratio, therefore, is used in this research to compute the structure dynamics. The numerical prediction of the investigated model will be compared against experimental results of a similar VNT turbine to verify current numerical methods.
213
4. Results and analysis
214
4.1. Turbine performance
188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211
215
4.1.1. Effects of nozzle vane’s spindles
216
It is well known that nozzle vanes can rotate round their spindles to achieve the function of changing the nozzle opening. This function requires the nozzle end-clearances to be designed and to exist between nozzle vane ends and hub/shroud walls. Meanwhile, the clearance provides an unexpected passage for fluid to cross over nozzles. After exiting the nozzle clearances, the fluid called clearance leakage flow interacts with main flow in nozzle channels, thus generating extra flow loss. Inside the clearances, the presence of spindles reduces the clearance area and therefore, has a contribution to the reduction of the leakage flow. But in lots of research work, the spindles were often neglected during modeling a research model in order to simplify geometry. In this section, the spindles’ effect will be quantified and the simplification will be evaluated.
217 218 219 220 221 222 223 224 225 226 227 228 229
Fig. 4 compares turbine performance of a VNT turbine with and without spindles at both full and small openings. The top two plots compares the flow rate-pressure ratio performance and the bottom two plots represent the com-parisons of the efficiency-speed ratio (U/C) characteristics. The very small deviations among curves at the full opening reveal that the nozzle’s spindles have no significant influence on the turbine performance. At the small opening, however, significant deviations indicate that the presence of spindles can improve slightly the turbine efficiency, but significantly reduce turbine flow capacity. Hu et al.12 studied the effect of the nozzle end-clearances on a VNT turbine and pointed out that, at fully open position, the loading acting on vanes was so small that it was not enough to drive much fluid over the narrow clearance to form strong leakage flow and to significantly deteriorate turbine performance. The reported conclusions well support the current numerical findings. Besides the loading, the ratio of the clearance area and the nozzle throat area becomes gradually larger with the decreased nozzle opening. This trend also takes responsibility for the larger deviations in turbine performance at small opening. With the controlling structure consisting of a spindle and a handle, a VNT turbine always operates with the best aerodynamic efficiency and the lowest flow capacity. The first reason is that for nozzle vanes with a spindle and a handle, the clearance passage area is further decreased so that the nozzle clearance leakage flow is reduced. And the second one is that the spindle is usually located near the vane’s trailing edge where the leakage flow is very strong. Zhao et al.8 investigated the nozzle clearance flow behavior and stated that the flow in the second part of the clearance was stronger because of relatively small vane thickness, high momentum of fluid and relatively large static pressure differential between two sides of a nozzle vane.
230
4.1.2. Effect of nozzle clearance height
264
The above analysis on the effect of nozzle vanes’ spindles proves that the nozzle end-clearance can cause significant deviations in turbine performance at small opening, but the spindle and handle are still able to be neglected in this study, because those curves in Fig. 4 have very similar trends. Consequently, the nozzle vanes’ spindle and handle are neglected in Cases 3– 12. Besides, it is only at the small opening that the following analyses are conducted, since the nozzle clearances’ impact becomes significant at such an opening.
265
Please cite this article in press as: ZHAO B et al. Numerical analysis for impacts of nozzle end-clearances on aerodynamic performance and forced response in a VNT turbine, Chin J Aeronaut (2018), https://doi.org/10.1016/j.cja.2018.02.015
231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263
266 267 268 269 270 271 272 273
CJA 1009 24 February 2018
No. of Pages 11
Numerical analysis for impacts of nozzle end-clearances on aerodynamic performance
Fig. 4
274 275 276 277 278 279 280 281 282 283 284 285 286 287 288
5
Comparison of performance of numerical model with and without spindles.
Fig. 5 shows turbine efficiency variations with the increased nozzle end-clearance size. It can be seen that the maximum flow rate through the turbine gradually increases with the increased size of nozzle end-clearance, but the turbine efficiency deteriorates significantly. The maximum deviation in the flow rate is up to 21 percent, and the efficiency drop is about 16 percentage points as the nozzle end-clearance increases to the largest size. It should be noted that the reason for such significant deviations is the use of the large nozzle clearance size in Case 6. This clearance size is slightly larger than that in current advanced VNT turbines. Meitner and Glassman13 proposed a clearance-flow model and stated that the changes in both turbine efficiency and flow rate became significantly larger with decreasing stator area and varied nearly linearly with clearance.
Fig. 5 Turbine efficiency variations with increased nozzle endclearance, at small opening.
The trend of the efficiency degradation increasing with the clearance size suggests that the nozzle end-clearance is unfavorable to the aerodynamic performance of a VNT turbine. This is why the mitigation of the nozzle end-clearance is still one of the main objects in both academic and industrial fields. But the thermal expansion caused by the high temperature environment makes it hard to further decrease the clearance size. Instead of directly decreasing the nozzle end-clearances, Sun et al.14 developed a VNT turbocharger with sliding vanes, and the totally new nozzle vanes successfully reduced the nozzle end-clearance leakage flow and improved the VNT turbine performance, especially at small opening. Yang et al.15 proposed a VNT turbine with a forepart rotation vane and numerically predicted results showed that the turbine efficiency was increased by 6 percent at small opening by removing partial nozzle end-clearances. These results prove that it is worth studying the nozzle end-clearances to find out more effective ways to improve the aerodynamic performance of a VNT turbine.
289
4.1.3. Effect of nozzle end-clearances’ distribution
309
Fig. 6 shows the turbine efficiency variations with the nozzle end-clearances’ distribution at several operation points within turbine map at the small opening. It is fairly clear that varying the nozzle end-clearances’ distribution not only affects the turbine flow capacity, but also changes the turbine efficiency. For instance, for the turbine with either nozzle shroud clearance (Case 12) or hub clearance (Case 7), the turbine flow capacity is obviously increased. Between those two cases, the maximum
310
Please cite this article in press as: ZHAO B et al. Numerical analysis for impacts of nozzle end-clearances on aerodynamic performance and forced response in a VNT turbine, Chin J Aeronaut (2018), https://doi.org/10.1016/j.cja.2018.02.015
290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308
311 312 313 314 315 316 317
CJA 1009 24 February 2018
No. of Pages 11
6
B. ZHAO et al.
Fig. 6
318 319 320 321 322 323 324 325 326 327 328 329 330 331 332 333 334 335 336 337 338 339 340 341 342 343 344 345 346 347 348 349 350 351 352 353 354 355 356 357 358 359
Turbine performance responses to various nozzle end-clearances’ distributions.
flow capacity occurs when all clearances are located at the hub side (Case 7), because of the relatively large distance between two rows at the hub side. Zhao et al.16 studied the rotor–stator interaction and proved that the blocking effect was closely related with the gap between two adjacent rows. However, when about 80 percent of the total nozzle end-clearance is located at shroud side (Case 11), the turbine flow capacity is the lowest. The maximum deviation in flow rate induced by nozzle end-clearances’ distribution is less than 5 percent. Beside the flow capacity, the turbine efficiency is always higher as small part of the total nozzle end-clearance is located at shroud side. Among those cases studied, no matter which point the turbine operates at, the peak efficiency always occurs when 20 percent of the total nozzle end-clearance size is located at shroud side (Case 8). In contract, the turbine efficiency is always deteriorated seriously as most clearances exist only at shroud side (Case 12). And the maximum deviation in turbine efficiency caused by varying nozzle end-clearances’ distribution is about 5 percentage points. It should be noticed that the use of the mixing plane approach at rotor/stator interface may show higher influence of nozzle end-clearance leakage flow on the turbine efficiency. Zhao et al.17 studied a VNT turbine using commercial code EURANUS and reported that compared with the unsteady simulation with direct interpolation at rotor/stator interface, steady simulations with the mixing plane method showed larger turbine efficiency variations in response to the different nozzle end-clearances. Roumeas and Cros18 tested two configurations: the nozzle gap divided equally at two ends of the nozzle vane and the whole gap located on the hub side. A significant deviation in turbine performance was observed. Walkingshaw and Spence19 tested a radial turbine with either hub side or shroud side stator vane clearance at full, middle and small openings, and revealed that the turbine performance differed for different nozzle clearances’ distribution. Those literatures provide evidences for that the nozzle end-clearances’ distribution significantly change the aerodynamic performance of a VNT turbine. In spite of that the current steady simulations may show higher influence of nozzle end-clearances, the effect on the turbine efficiency of the various nozzle end-clearances is still worth focusing.
4.2. Aerodynamic excitation
360
4.2.1. Turbine wheel resonant
361
A FEA calculation of a single blade was performed to predict the natural frequencies of turbine wheel. Since the critical modes for blade fatigue are mainly low order modes20, no higher order modes are analyzed in this paper. The numerically predicted frequencies are used as input for the Campbell diagram shown in Fig. 7. The aerodynamic excitation frequencies are plotted as a function of rotational speed of a turbine wheel. The two horizontal lines drawn in Fig. 7 represent the natural frequencies of Mode 1 and Mode 2, respectively. In the plot, the rotational speed and frequency was normalized by the turbine top speed and the natural frequency at Mode 1, respectively. The intersection points of natural frequencies with rotational speed orders are identified as potential operation points at which the blade resonance may occur. For the research turbine, resonances of the 1st speed order may excite blade Mode 1 whereas the 2nd speed order may excite Model 2 within a realistic operating range from the rotational speed of 0 to the top speed limit. According to the Campbell diagram, the critical rotational speed is easily known. And other parameters of the critical operation point can be obtained by running GT-Power software. In consequence, unsteady CFD simulations with those boundary conditions were carried out on a VNT turbine with
362
Fig. 7
Campbell diagram with Modes 1 and 2.
Please cite this article in press as: ZHAO B et al. Numerical analysis for impacts of nozzle end-clearances on aerodynamic performance and forced response in a VNT turbine, Chin J Aeronaut (2018), https://doi.org/10.1016/j.cja.2018.02.015
363 364 365 366 367 368 369 370 371 372 373 374 375 376 377 378 379 380 381 382 383 384
CJA 1009 24 February 2018
No. of Pages 11
Numerical analysis for impacts of nozzle end-clearances on aerodynamic performance 385 386
387 388 389 390 391 392 393 395 396 397 398 399 400 401 402 403 404 405 406 407 408 409 410 411 412 413 414 415 416 418 419 420 421 422
424 425 426 427 428 429 430 431 432 433
various nozzle end-clearances to predict the aerodynamic excitation pressure. 4.2.2. Loading variations with increased nozzle end-clearance size The aerodynamic loading, fðx; y; z; tÞ, is the differential of static pressure on both surfaces of a turbine blade, and represents the aerodynamic force that acts on blade surfaces to transfer energy from the fluid to the turbine wheel. fðx; y; z; tÞ ¼ pðx; y; z; tÞ pðx0 ; y0 ; z0 ; tÞ
ð1Þ
where f and p stand for the loading and pressure at an arbitrary point on blade surface, respectively; x, y, z represent the coordinate point at one surface of turbine blade and the x0 , y0 , z0 represent the coordinate point at another surface of turbine blade; the t is time. If the aerodynamic loading is steady, the turbine wheel will operate smoothly. In fact, the flow inside a VNT turbine is always highly unsteady, and therefore, there is serious aerodynamic interaction between the nozzle vanes and turbine wheel. This rotor–stator interaction must cause the aerodynamic loading to be high unsteadiness. The unsteady loading is a dominant excitation source to the vibration of a VNT turbine wheel. The aerodynamic loading can be divided into two parts, steady and unsteady parts. The steady part represents the time-average loading. And the unsteady part is corresponding to the fluctuating loading caused by the unsteady interaction and other complicated flow, which transfers exciting energy from the fluid to the turbine wheel to have the wheel deformed repeatedly and vibrated. They are expressed as follows: y; z; tÞ þ fðx; ~ y; z; tÞ fðx; y; z; tÞ ¼ fðx;
ð2Þ
where f stands for the time-average value of the loading and f~ represents for the unsteady part of the loading. The time-average loading is calculated as follows: Z T y; zÞ ¼ 1 fðx; fðx; y; z; tÞdt ð3Þ T 0 where T is for the time period that the pressure was monitored and recorded. Fig. 8 shows the distribution of the steady part of aerodynamic loading on wheel blade surface for four nozzle end-clearance sizes. It is fairly clear that high loading is centrally located in inducer. The reason is that at small opening the nozzle vanes are pivoted at the large setting angle so that the positive incidence is formed at inducer inlet. As the nozzle end-clearance size increases, the loading density near inducer blade tip and root obviously increases, but it decreases in mid-
Fig. 8
7
dle spanwise section in inducer. In exduer, the steady loading also increases with the increased clearance size. It is well known that the distribution of the steady loading is closely related with turbine efficiency and torque. When a turbine operates at small opening, the high loading density in inducer is in favor of recovering energy from exhaust gas, because an internal combustion engine is often to operate at part loads. Though the increased nozzle end-clearance size promotes the loading density in exducer, the decreased loading density in whole inducer deteriorates the turbine efficiency. Fig. 9 shows the variations of aerodynamic excitation intensity at both the 1st and the 2nd orders with the increased size of nozzle end-clearances. They are obtained by using Fast Fourier Transform (FFT) algorithm and expressed as follows: ~ y; z; tÞÞ Forceð1st;2ndÞ ¼ FFTðfðx; 0
ð4Þ
434 435 436 437 438 439 440 441 442 443 444 445 446 447 448 450
where Force is for the excitation force at the 1st order and/or the 2nd order. For the model without nozzle end-clearances, the aerodynamic excitation acted on the wheel blade is very weak in most regions. After adding nozzle end-clearances, however, the aerodynamic excitation intensities including the 1st and 2nd orders is reinforced not only in inducer tip and root regions but in exducer as well. Compared with the 2nd order aerodynamic excitation, the excitation intensity of the 1st order is stronger. Therefore, the 1st order excitation should transfer more energy to the wheel blade to have it vibrated and do much for the turbine wheel high cycle fatigue failure. And with the increased nozzle end-clearance size, the intensity of the 1st order excitation increases in the region near the blade tip round 70 % of blade chord. It should be noticed that the span from blade tip to root is so large in exducer that the high excitation intensity at the 1st order mode greatly increases the risk of the turbine wheel suffering from high cycle fatigue failure.
451
4.2.3. Loading variations with nozzle end-clearances’ distribution
469
Fig. 10 shows the variations of the steady loading distribution on a wheel blade surface with the nozzle end-clearances’ distributions. When most nozzle end-clearances, or all, are located at the hub side, the obviously high loading occurs near inducer blade root. And, with the decreased nozzle hub clearance and increased nozzle shroud clearance, the steady loading in the region near the exducer blade tip gradually increases. Meanwhile, the average loading density in the whole inducer decreases. The loading style in Case 12 is unfavorable to the recovery rate of energy from exhaust gas. Therefore, it can be drawn that the nozzle end-clearances’ distribution is able to change the distribution of the loading on a wheel blade.
471
Distribution of time-averaged loading on wheel blade.
Please cite this article in press as: ZHAO B et al. Numerical analysis for impacts of nozzle end-clearances on aerodynamic performance and forced response in a VNT turbine, Chin J Aeronaut (2018), https://doi.org/10.1016/j.cja.2018.02.015
452 453 454 455 456 457 458 459 460 461 462 463 464 465 466 467 468
470
472 473 474 475 476 477 478 479 480 481 482
CJA 1009 24 February 2018
No. of Pages 11
8
Fig. 9
B. ZHAO et al.
Variations of aerodynamic excitation at both the 1st and 2nd orders on wheel blade with increased size of nozzle end-clearances.
Fig. 10
Variations of steady loading acted on wheel blade with nozzle end-clearances’ distribution.
Fig. 11 Variations of aerodynamic excitation at both the 1st and 2nd orders on wheel blade surface with nozzle end-clearances’ distribution.
483 484 485 486 487 488 489
Fig. 11 shows the variations of aerodynamic excitation intensity at both the 1st and 2nd orders on the wheel blade surface with various nozzle end-clearances’ distribution. It is fairly clear that the aerodynamic excitation intensity in the region near blade tip around 70 % of chord is very weak when all nozzle end-clearances are located at the hub side, as shown in Case 7 in Fig. 11. However, as the nozzle shroud clearance
increases, the 1st order aerodynamic excitation intensity in that region gradually increases and is up to the maximum when all nozzle clearances are located at shroud side. Therefore, it can be concluded that the nozzle shroud clearance leakage flow is very sensitive to the 1st aerodynamic excitation intensity. Beside the excitation at the 1st order, the aerodynamic excitation with the 2nd order frequency also varies with the nozzle
Please cite this article in press as: ZHAO B et al. Numerical analysis for impacts of nozzle end-clearances on aerodynamic performance and forced response in a VNT turbine, Chin J Aeronaut (2018), https://doi.org/10.1016/j.cja.2018.02.015
490 491 492 493 494 495 496
CJA 1009 24 February 2018
No. of Pages 11
Numerical analysis for impacts of nozzle end-clearances on aerodynamic performance
502
end-clearances’ distribution. For instance, the excitation intensity at the 2nd order at inducer blade root gradually decreases with the decreased nozzle hub clearance. Meanwhile, it increases in the inducer blade tip region. But, compared with the 1st order excitation, the excitation intensity of the 2nd order is far weaker.
503
4.3. Forced response
504
4.3.1. Turbine wheel dynamic stress distribution
505
Independent calculations of the internal flow field and the harmonic response are presented in this paper, respectively. The calculations were carried out with two different computational meshes. The length scale of grid cells used in the CFD simulations is fully different on the wheel blade surface due to the use of clustering technology. However, the mesh used in the harmonic response calculation is almost uniform on the blade surface except the fillet region. With the Matlab code based on the cubic spline function interpolation, the data transmission was performed to transfer the information of the loading fluctuation on the grid nodes of the flow field to the grid points of the solid field. The evaluation of the accuracy of the interpolation is illustrated in Fig. 12. The comparison of two plots shows very small deviations, so that the interpolation method used in this paper could be accurate enough to meet the requirements of
497 498 499 500 501
506 507 508 509 510 511 512 513 514 515 516 517 518 519 520
Fig. 12
Fig. 13
9
current research work. After the interpolation, the aerodynamic loading was further analyzed by FFT Algorithm to obtain the real and imaginary components which would be used for the FEA calculations. Based on the analysis of the harmonic response, the stress distributions at both the 1st and the 2nd orders are illustrated in Fig. 13, respectively. The magnitude is normalized by dividing the vibration stress value by the peak value in the case with normal nozzle end-clearances (Case 5). From the contour distribution in two plots in Fig. 13, it can be seen that the maximum stress at the 1st order mode is about 2.5 times than the peak stress at the 2nd order and occurs at exducer blade root. This numerical result points out that the vibration stress of the 1st order induced by the 1st harmonic loading can be considered as the major alternating stress on the turbine wheel and the exducer blade root region may suffer from high cycle fatigue failure. The numerically predicted region with high risk is in agreement with the experimental result of a similar turbine reported by Hu et al.12 and illustrated Fig. 14.
521
4.3.2. Forced response of turbine wheel
540
Fig. 15 shows the variations of the peak vibration stress at both the 1st and 2nd orders with the increased height of the nozzle end-clearances which were equally divided at two ends of nozzle vanes. It is fairly clear that the presence of nozzle end-clearances does increase the peak stress of the 1st order
541
Validation of the interpolation approach.
Distribution of stress on wheel blade (Case 5).
Please cite this article in press as: ZHAO B et al. Numerical analysis for impacts of nozzle end-clearances on aerodynamic performance and forced response in a VNT turbine, Chin J Aeronaut (2018), https://doi.org/10.1016/j.cja.2018.02.015
522 523 524 525 526 527 528 529 530 531 532 533 534 535 536 537 538 539
542 543 544 545
CJA 1009 24 February 2018
No. of Pages 11
10
Fig. 14 Example of high cycle fatigue failure at exducer blade root, observed on a similar turbine wheel.
Fig. 15 Turbine wheel maximum stress variations with increased nozzle end-clearances.
Fig. 16 Turbine wheel maximum stress variation with nozzle end-clearance distribution.
546 547 548 549 550 551 552 553 554 555 556 557 558 559 560 561
vibration. And, with the increased size of nozzle clearances, the peak value of the stress at the 1st order mode gradually increases. Compared with the model without nozzle endclearances (Case 3), the 1st order stress is increased by twice as much. This trend means that the nozzle clearance leakage flow is one of the major resources contributing to the damage of a turbine wheel due to the high cycle fatigue. Besides the 1st order excitation, the peak stress of the 2nd order vibration is slightly changed with the increased nozzle end-clearances. This decreased trend reveals that the nozzle clearance leakage flow acts as the 1st order aerodynamic excitation source for the current turbine wheel. Fig. 16 shows the variation of the peak stress at both the 1st and 2nd orders with the various nozzle end-clearances’ distribution. As was analyzed above, with the decreased nozzle end-clearance on the shroud side the turbine efficiency signifi-
B. ZHAO et al. cantly increases. Herein, the curve of the peak stress at the 1st order mode clearly shows that the peak stress obviously decreases with the decreased nozzle shroud clearance, and the maximum deviation is about 30%. These results reveal that the nozzle shroud end-clearance is more sensitive not only to the turbine efficiency but to the forced response of the current turbine wheel at the 1st order mode as well, compared with the nozzle hub clearance. In contrast, the forced response at the 2nd order mode firstly increases and then slightly decreases with the decreased nozzle shroud clearance. The narrow variation range of the peak stress at the 2nd order means that the effect of the nozzle end-clearances’ distribution on the wheel vibration at the 2nd order mode is insignificant. According to the above analysis on the harmonic response, not only the presence of the nozzle end-clearances but also the increased size of nozzle shroud clearance does increase the risk of the turbine wheel suffering from high cycle fatigue failure and, between the shroud and hub clearances, the nozzle shroud clearance has more contribution for the vibration of the turbine wheel at the 1st order mode. These results, therefore, suggest a possibility to improve both the turbine efficiency and the reliability of a turbine wheel by reducing the unfavorable effect of the nozzle shroud clearance. For instance, Wang et al.21 proposed a controlling structure to decrease the nozzle shroud clearance to improve the aerodynamic performance of a VNT turbine. This method, or other approaches with similar principles, may also benefit the turbine performance, while also solving the high cycle fatigue issue that the design and development of an automotive turbocharger is trying to address. Besides, an introduction of 3D nozzle vane with oblique trailing edge may be possible to mitigate the unfavorable impact of the nozzle shroud clearance. The differential of pressure on both sides of a nozzle vane is the force to drive the air over the narrow nozzle end-clearances. Decreasing the pressure difference at shroud side by increasing the nozzle opening only at shroud side is likely to reduce the nozzle shroud clearance leakage flow; in the meantime, the opening of the nozzle vanes at the hub side needs to be decreased in order to keep the same nozzle throat.
562
5. Conclusions
601
This paper numerically investigated the effects of various nozzle end-clearances on both the aerodynamic performance and the forced response of a VNT turbine in order to find out the possibility to improve both the stage efficiency and the reliability of a turbine wheel via optimizing nozzle end-clearances. Research models are a VNT turbine with different total sizes of nozzle end-clearances which are equally divided at two ends of nozzle vanes or with various distributions of the clearances at the hub and shroud sides which have the same total size.
602
(1) As the total nozzle end-clearances remain the same, the nozzle end-clearances’ distribution is closely related to the turbine efficiency, the flow capacity, and the vibration stress at the 1st order mode at small openings. When most nozzle end-clearances are located at shroud side, a VNT turbine has the minimum flow capacity while the flow capacity is the maximum when no nozzle clearance is located at shroud side. When small part of the total nozzle end-clearances is located at shroud side,
611
Please cite this article in press as: ZHAO B et al. Numerical analysis for impacts of nozzle end-clearances on aerodynamic performance and forced response in a VNT turbine, Chin J Aeronaut (2018), https://doi.org/10.1016/j.cja.2018.02.015
563 564 565 566 567 568 569 570 571 572 573 574 575 576 577 578 579 580 581 582 583 584 585 586 587 588 589 590 591 592 593 594 595 596 597 598 599 600
603 604 605 606 607 608 609 610
612 613 614 615 616 617 618 619
CJA 1009 24 February 2018
No. of Pages 11
Numerical analysis for impacts of nozzle end-clearances on aerodynamic performance 620 621 622 623 624 625 626 627 628 629 630 631 632 633 634 635 636 637 638 639 640 641 642
the peak efficiency always occurs and the peak stress of the 1st order vibration at exducer blade root can be decreased greatly. (2) The nozzle end-clearance leakage flow has stronger effects on vibration stress at the 1st order mode for the current model than that of the 2nd order. Compared with the nozzle hub clearance, the nozzle shroud clearance is more sensitive to the turbine efficiency degradation and the forced response of a turbine wheel at the 1st order, but has less contribution to increase the turbine flow capacity. (3) It has reached broad agreement that decreasing total size of the nozzle end-clearances can benefit the aerodynamic performance of a VNT turbine. But due to thermal expansion of the material, it is very hard to further decrease the size in the development of an advanced VNT turbine. According to the results reported in this paper, controlling the nozzle end-clearances’ distribution, or decreasing the nozzle shroud clearance’s effect, may be possible methods to design an advance VNT turbine with higher efficiency and lower forced response in the future, instead of the hard efforts to further decrease the total size of the nozzle end-clearances.
643
644
Acknowledgements
645
This study was co-supported by the Natural Science Foundation of Hebei Province of China (No. E2017402135) and the Program of Science and Technology Research and Development of Handan of China (No. 1621212047-2).
646 647 648
649 650 651 652 653 654 655 656 657 658 659 660 661 662 663 664
References 1. Tang H, Pennyycott A, Akehurst S, Brace AJ. A review of the application of variable geometry turbines to the downsized gasoline engine. Int J Engine Res 2015;16(6):810–25. 2. Feneley AJ, Pesiridis A, Andwari AM. Variable geometry turbocharger technologies for exhaust energy recovery and boosting – a review. Renew Sustain Energy Rev 2017;71(1):959–75. 3. Hu L, Yang C, Sun H, Zhang J, Lai M. Numerical analysis of nozzle clearance’s effect on turbine performance. Chin J Mech Eng-en 2011;24(4):618–25. 4. Walkingshaw J, Spence S, Ehrhard J, Thornhill D. A numerical study of the flow fields in a highly off-design variable geometry turbine. Glasgow: ASME; 2010, Report No.: GT2010-22669. 5. Tamaki H, Goto S, Unno M, Iwakami A. The effect of clearance flow of variable area nozzles on radial turbine performance. Berlin: ASME; 2008, Report No. GT2008-50461.
11
6. Tamaki H, Unno M. Study on flow fields in variable area nozzles for radial turbines. Int J Fluid Mech Syst 2008;1(1):47–56. 7. Chen H. Turbine wheel design for Garrett advanced variable geometry turbines for commercial vehicle applications. In: Proceedings of the 8th international conference on turbochargers and turbocharging, 2006 May 28–29; London, UK; 2006. 8. Zhao B, Yang C, Hu L, Sun H, Yi J, Eric C, et al. Understanding of the interaction between clearance leakage flow and main passage flow in a VGT turbine. Adv Mech Eng 2015;7(2):652769. 9. Liu Y, Yang C, Qi M, Zhao B. Shock, leakage flow and wake interactions in a radial turbine with variable guide vanes. Dusseldorf: ASME; 2014, Report No.: GT2014-25888. 10. Zhao B, Hu L, Sun H, Yang C, Shi X, Yi J, et al. Numerical investigation of a novel approach for mitigation of forced response of a variable geometry turbine during exhaust braking mode. Seoul: ASME; 2016, Report No.: GT2016-56342. 11. Lu FA, Qi DT, Wang XJ, Zhou Z, Zhou HH. A numerical optimization on the vibroacoustics of a centrifugal fan volute. J Sound Vib 2012;331(10):2365–85. 12. Hu L, Sun H, Yi J, Curtis EC, Morelli A, Zhang J, et al. Investigation of nozzle clearance effects on a radial turbine: aerodynamic performance and forced response. Detroit: SAE; 2013, Report No.: 2013-01-0918. 13. Meitner PL, Glassman AJ. Off-design performance loss model for radial turbine with pivoting, variable-area stators. NASA Technique paper 1708. 14. Sun HH, Zhang J, Hu L, Hanna DR. Sliding vane geometry turbines: United States patent US 20130042608; 2013 Feb 21. 15. Yang D, Lao D, Yang C, Hu L, Sun H. Investigations on the generation and weakening of shock wave in a radial turbine with variable guide vanes. Seoul: ASME; 2016, Report No.: GT201657047. 16. Zhao B, Yang C, Chen S, Qi M, Zhou M. Unsteady flow variability driven by rotor-stator interaction at rotor exit. Chin J Aeronaut 2012;25(6):871–8. 17. Zhao B, Ma C, Hu L, Yang C. Numerical investigation on effects of uncertain nozzle vane clearance on variable geometry radial turbine performance. J Propul Technol 2014;35(4):492–8 [Chinese]. 18. Roumeas M, Cros S. Aerodynamic investigation of a nozzle clearance effect on radial turbine performance. Copenhagen: ASME; 2012, Report No.: GT2012-68835. 19. Walkingshaw JR, Spence SWT, Thornhill D, Ehrhard J. An experimental assessment of the effects of stator vane clearance location on an automotive turbocharger turbine. London, UK: Woodhead Publishing Limited; 2012, Reprot No.: C1340/ 007. 20. Ma C, Huang Z, Qi M. Investigation on the forced response of a radial turbine under aerodynamic excitations. J Therm Sci 2016;125(2):130–7. 21. Wang L, Hou H, Wang K, Yang Y. An adaptive device and method to control nozzle endwall clearances in a variable nozzle turbine. China Patent CN103352729; 2016 Oct 16 [Chinese].
Please cite this article in press as: ZHAO B et al. Numerical analysis for impacts of nozzle end-clearances on aerodynamic performance and forced response in a VNT turbine, Chin J Aeronaut (2018), https://doi.org/10.1016/j.cja.2018.02.015
665 666 667 668 669 670 671 672 673 674 675 676 677 678 679 680 681 682 683 684 685 686 687 688 689 690 691 692 693 694 695 696 697 698 699 700 701 702 703 704 705 706 707 708 709 710 711 712 713 714 715 716 717
No. of Pages 11
1
Chinese Society of Aeronautics and Astronautics & Beihang University
Chinese Journal of Aeronautics [email protected] www.sciencedirect.com
5
Numerical analysis for impacts of nozzle endclearances on aerodynamic performance and forced response in a VNT turbine
6
Ben ZHAO a, Leilei WANG b,*, Hongjun HOU b
3
4
7 8
9
a b
Michigan State University, East Lansing 48824, USA Hebei University of Engineering, Handan 056038, China
Received 26 April 2017; revised 20 June 2017; accepted 17 July 2017
10
12 13
KEYWORDS
14
Clearance; Forced response; Mode analysis; Performance; Reliability; Variable Nozzle Turbine (VNT)
15 16 17 18 19 20
21
Abstract It has been well known that nozzle end-clearances in a Variable Nozzle Turbine (VNT) are unfavorable for aerodynamic performance, especially at small openings, and efforts to further decrease size of the clearances are very hard due to thermal expansion. In this paper, both the different sizes of nozzle end-clearances and the various ratios of their distribution at the hub and shroud sides were modelled and investigated by performing 3D Computational Fluid Dynamics (CFD) and Finite Element Analysis (FEA) simulations with a code of transferring the aerodynamic pressure from the CFD results to the FEA calculations. It was found that increasing the size of the nozzle end-clearances divided equally at the hub and shroud sides deteriorates turbine efficiency and turbine wheel reliability, yet increases turbine flow capacity. And, when the total nozzle endclearances remain the same, varying nozzle end-clearances’ distribution at the hub and shroud sides not only shifts operation point of a VNT turbine, but also affects the turbine wheel vibration stress. Compared with nozzle hub clearance, the shroud clearance is more sensitive to both aerodynamic performance and reliability of a VNT turbine. Consequently, a possibility is put forward to improve VNT turbine efficiency meanwhile decrease vibration stress by optimizing nozzle end-clearances’ distribution. Ó 2018 Production and hosting by Elsevier Ltd. on behalf of Chinese Society of Aeronautics and Astronautics. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/ licenses/by-nc-nd/4.0/).
* Corresponding author. E-mail address: [email protected] (L. WANG). Peer review under responsibility of Editorial Committee of CJA.
Production and hosting by Elsevier
1. Introduction
22
The past decade has seen that the emission standards for various types of vehicles have been becoming more stringent. To achieve less emission, car manufacturers are attempting to downsize an internal combustion engine that can provide the same power of a large engine by adding a boosting device, e.g. turbochargers.1 To better acceleration and deliver more power across a wider range, a Variable Nozzle Turbine
23
https://doi.org/10.1016/j.cja.2018.02.015 1000-9361 Ó 2018 Production and hosting by Elsevier Ltd. on behalf of Chinese Society of Aeronautics and Astronautics. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/). Please cite this article in press as: ZHAO B et al. Numerical analysis for impacts of nozzle end-clearances on aerodynamic performance and forced response in a VNT turbine, Chin J Aeronaut (2018), https://doi.org/10.1016/j.cja.2018.02.015
24 25 26 27 28 29
CJA 1009 24 February 2018
No. of Pages 11
2 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91
(VNT) has been developing and applied.2 A common method of variable nozzle in a radial turbine is the use of the pivoting vanes that are mounted on a flat plate. These vanes are connected with spindles, sometimes together with handles, to achieve the function of rotating axially. Between the flat plate and nozzle vanes, nozzle end-clearances have to be designed. Efforts to reduce the nozzle end-clearances’ size have been doing, and so far the size has been successfully decreased in advanced VNT turbos. But the unfavorable effects of the clearances are still significant and further decreasing the size will be very hard, since the nozzle vanes would get stuck due to thermal expansion. The temperature of exhaust gas is often more than 900 °C for a diesel engine and beyond 1000 °C for a gasoline engine. In addition, complicated sulfide from exhaust gas moves outward from wheel inducer to nozzles under the action of centrifugal force, and then is possible to stick the nozzle vanes as the clearances are too small. It has been reported that the presence of the nozzle endclearances usually has unfavorable effects on aerodynamic performance of a VNT turbine. Hu et al.3 performed numerical simulations on a VNT turbine to study the nozzle endclearances’ effect on turbine performances and revealed that the presence of nozzle end-clearances deteriorated turbine performances, especially at small open condition. Walkingshaw et al.4 carried out numerical simulations of the flow fields in a highly off-design VNT turbine and indicated that the leakage vortices changed the circumferential distribution of the approaching flow to the rotor and then deteriorated aerodynamic performance. The relationship between the nozzle endclearances and the turbine performance was also focused by Tamaki et al.5,6 Besides the aerodynamic performance, the high cycle fatigue is one of the critical issues that the development of an advanced VNT turbine has to address. Shock waves, one of the contributors to the high cycle fatigue failure, weigh heavily on the turbine wheel blade vibration and, sometimes, could seriously damage a turbine wheel, as shown by Chen.7 The intensity of the shock wave was proved to be closely related with nozzle end-clearances in a VNT turbine.8 Therefore, the nozzle end-clearances must have an effect on turbine wheel reliability. In fact, the nozzle clearance leakage flow is also able to directly interact with a turbine wheel and then causes it to vibrate.9,10 Though it has been reported that the nozzle end-clearances are closely related to the turbine performance, the relationship between the forced response of turbine wheel and the changed clearances, i.e. increased clearance size and different distribution, is not fully clear. If the nozzle shroud clearance has a significantly different effect on a VNT turbine from the hub clearance, it will provide a possibility to further improve a VNT turbine in both the aerodynamic performance and the reliability by optimizing the more sensitive clearance, instead of the hard effort to further decrease the size of total nozzle end-clearances. In this paper, a VNT turbine was first modeled and simulated by a 3D steady Computational Fluid Dynamics (CFD) method to validate computational mesh, and then the VNT turbine model with various nozzle end-clearances were predicted by both steady and unsteady CFD methods. After CFD simulations, the information about aerodynamic excitation pressure on a wheel blade surface was transferred from the CFD results to the Finite Element Analysis (FEA) calculation by interpolation so that the harmonic response could be
B. ZHAO et al. predicted and analyzed. Finally, an optimized nozzle endclearances’ distribution with high turbine efficiency and low forced response was suggested.
92
2. Research model
95
The research model is a VNT turbine. In order to reduce the requirement of CFD simulations, the volute was neglected. This simplification should have little influence on the conclusions stated in this paper, since the flow is guided and redistributed in crossing the nozzles. Therefore, the model used just consists of variable nozzles vanes and a mixed-flow rotor. Fig. 1 shows the meridional view of a VNT turbine with detailed illustrations of nozzle end-clearances, and two types of nozzle vanes. Inside the clearances, the spindle and the handle of a nozzle vane are not illustrated. The pivoting nozzle vanes are often fixed and controlled by a spindle or a combination of a spindle and a handle. If the spindle is located round the center of a vane, the handle will be not needed, whereas if it is located near trailing edge of a vane, the handle is often designed and installed near the vane’s leading edge to well control the vanes’ setting angle, as shown in Fig. 1. Regardless of the control styles, VNT turbine always requires the presence of nozzle end-clearances to achieve the relative movement between nozzle vanes and the flat plates (endwalls). Herein, the nozzle shroud clearance is marked with Cshroud and the clearance on another side is represented by Chub, as shown in Fig. 1. In order to clearly identify the research models used, the investigated VNT turbine with various nozzle end-clearances is numbered and listed in Table 1. The letter, h, stands for the normal height of the nozzle end-clearance at each side of a VNT turbine product, and is about 2 percent of the span of flow passage from the hub wall to the shroud one.
96
3. Numerical schemes
125
3.1. CFD model
126
A fully automatic hexahedral grid generator was used to generate the computational mesh for the whole flow field. Multiblock Structured Meshing with an O4H topology including 5 blocks was used to discrete the channels, as shown in Fig. 2. The skin block was an O-mesh surrounding the blade and the H-mesh was applied in inlet, outlet, up and down blocks. The mesh inside the hub and shroud clearances was created using a butterfly topology. The number of grid points had to be adjusted for various clearance sizes to keep a good mesh quality under the condition of the same first cell length. The subsonic inlet condition with total pressure, total temperature, velocity angles and turbulent viscosity were specified at the inlet boundary of the computational domain. At outlet boundary, a radial equilibrium technology was used. With this, the outlet static pressure was imposed on the given radius and integration of the radial equilibrium law along the spanwise direction permitted to calculate the hub-to-shroud profile of the static pressure. Therefore, a constant static pressure would be imposed along the circumferential direction. On the solid surface, the adiabatic walls were used and the velocity vector vanished. The rotational speed of the unsteady simulations
127
Please cite this article in press as: ZHAO B et al. Numerical analysis for impacts of nozzle end-clearances on aerodynamic performance and forced response in a VNT turbine, Chin J Aeronaut (2018), https://doi.org/10.1016/j.cja.2018.02.015
93 94
97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124
128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147
CJA 1009 24 February 2018
No. of Pages 11
Numerical analysis for impacts of nozzle end-clearances on aerodynamic performance
Fig. 1
Table 1
Illustration of nozzle end-clearances (at meridional view) and two types of nozzle vane.
Illustration of research models.
Model
Cshroud
Chub
Spindles
Case Case Case Case Case Case Case Case Case Case Case Case
100%h 100%h 0%h 50%h 100%h 150%h 0%h 20%h 40%h 60%h 160%h 200%h
100%h 100%h 0%h 50%h 100%h 150%h 200%h 180%h 160%h 140%h 40%h 0%h
Spindle Spindle & Handle – – – – – – – – – –
1 2 3 4 5 6 7 8 9 10 11 12
Fig. 2
148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168
3
was selected according to the results of the Campbell Diagram, and other boundary conditions of the unsteady simulations were obtained by running GT-Power software. There is one rotor/stator interface in the whole flow field. For steady simulations, the conservative coupling by pitchwise rows approach was used due to its capacity to provide an exact conservation of mass flow, momentum and energy through the interface. And the full non-matching boundary connecting algorithm was used in order to exchange the flow variables and fluxes between the left and right image. For unsteady simulations, the sliding mesh technology was applied on the rotor/ stator interface. The perfect gas relation and the Spalart-Allmaras turbulence model were employed to close the RANS/URANS equations. The discretization in space was based on a cell centered control volume approach. An explicit 4-stage Runge-Kutta scheme was applied to solve the equations mentioned above. For all the CFD simulations, the Courant-Friedrichs-Lewy (CFL) number was 3. And after calculating a set of computational meshes with the same topology and different number of grid points, the relationship of the mesh density and the chan-
Computational grids.
ged performance suggested the current computational mesh which was within affordable computation source limit. The y plus value was less than 3 in most region and, in nozzle endclearances, the average value was less than 2.68. For unsteady simulation, the total number of the grid nodes was about 4.53 million and the physical time step was about 1.15 108 s. Fig. 3 shows the validation of CFD methods which were performed on a similar VNT turbine. Although the deviation exists between the numerical and experimental data, they are still within an acceptable range.
169
3.2. FEA model
179
Commercial FEA software with linear analysis method was employed to predict the natural frequencies and harmonic response of a turbine wheel influenced by the unsteady aerodynamic excitation source. In order to reduce the requirement of FEA calculations, only one wheel blade was modelled. The tetrahedral elements were used to discrete the solid field, in which Inconel was assumed as an isotropic elastic material with a Yonge modulus
180
Please cite this article in press as: ZHAO B et al. Numerical analysis for impacts of nozzle end-clearances on aerodynamic performance and forced response in a VNT turbine, Chin J Aeronaut (2018), https://doi.org/10.1016/j.cja.2018.02.015
170 171 172 173 174 175 176 177 178
181 182 183 184 185 186 187
CJA 1009 24 February 2018
No. of Pages 11
4
B. ZHAO et al.
Fig. 3
Validation of the CFD method performed on a similar VNT turbine.
212
and a Poisson ratio. The detailed material properties were selected from a composite material database. The centrifugal force was applied to the FEA analysis, which was calculated according to angular velocity obtained from the Campbell diagram. An assumption of nodisplacement is applied at the center line of the wheel shaft. And a harmonic loading was applied on the wheel blade, which was calculated based on the unsteady CFD method. Once the natural frequencies of the turbine wheel were predicted by the modal analysis and the unsteady CFD simulations were carried out, the information about the aerodynamic excitation acting on the wheel blade was added from the CFD results to FEA analysis by interpolation. The detailed data transmission will be introduced in section of the forced response. The damping parameters significantly affect the accuracy of the vibration solution and the only way to obtain the precise damping parameters is through experimentation. Lu et al.11 reported that the maximum relative error for a computational method with a constant damping ratio was lower than 10 percent. A constant damping ratio, therefore, is used in this research to compute the structure dynamics. The numerical prediction of the investigated model will be compared against experimental results of a similar VNT turbine to verify current numerical methods.
213
4. Results and analysis
214
4.1. Turbine performance
188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211
215
4.1.1. Effects of nozzle vane’s spindles
216
It is well known that nozzle vanes can rotate round their spindles to achieve the function of changing the nozzle opening. This function requires the nozzle end-clearances to be designed and to exist between nozzle vane ends and hub/shroud walls. Meanwhile, the clearance provides an unexpected passage for fluid to cross over nozzles. After exiting the nozzle clearances, the fluid called clearance leakage flow interacts with main flow in nozzle channels, thus generating extra flow loss. Inside the clearances, the presence of spindles reduces the clearance area and therefore, has a contribution to the reduction of the leakage flow. But in lots of research work, the spindles were often neglected during modeling a research model in order to simplify geometry. In this section, the spindles’ effect will be quantified and the simplification will be evaluated.
217 218 219 220 221 222 223 224 225 226 227 228 229
Fig. 4 compares turbine performance of a VNT turbine with and without spindles at both full and small openings. The top two plots compares the flow rate-pressure ratio performance and the bottom two plots represent the com-parisons of the efficiency-speed ratio (U/C) characteristics. The very small deviations among curves at the full opening reveal that the nozzle’s spindles have no significant influence on the turbine performance. At the small opening, however, significant deviations indicate that the presence of spindles can improve slightly the turbine efficiency, but significantly reduce turbine flow capacity. Hu et al.12 studied the effect of the nozzle end-clearances on a VNT turbine and pointed out that, at fully open position, the loading acting on vanes was so small that it was not enough to drive much fluid over the narrow clearance to form strong leakage flow and to significantly deteriorate turbine performance. The reported conclusions well support the current numerical findings. Besides the loading, the ratio of the clearance area and the nozzle throat area becomes gradually larger with the decreased nozzle opening. This trend also takes responsibility for the larger deviations in turbine performance at small opening. With the controlling structure consisting of a spindle and a handle, a VNT turbine always operates with the best aerodynamic efficiency and the lowest flow capacity. The first reason is that for nozzle vanes with a spindle and a handle, the clearance passage area is further decreased so that the nozzle clearance leakage flow is reduced. And the second one is that the spindle is usually located near the vane’s trailing edge where the leakage flow is very strong. Zhao et al.8 investigated the nozzle clearance flow behavior and stated that the flow in the second part of the clearance was stronger because of relatively small vane thickness, high momentum of fluid and relatively large static pressure differential between two sides of a nozzle vane.
230
4.1.2. Effect of nozzle clearance height
264
The above analysis on the effect of nozzle vanes’ spindles proves that the nozzle end-clearance can cause significant deviations in turbine performance at small opening, but the spindle and handle are still able to be neglected in this study, because those curves in Fig. 4 have very similar trends. Consequently, the nozzle vanes’ spindle and handle are neglected in Cases 3– 12. Besides, it is only at the small opening that the following analyses are conducted, since the nozzle clearances’ impact becomes significant at such an opening.
265
Please cite this article in press as: ZHAO B et al. Numerical analysis for impacts of nozzle end-clearances on aerodynamic performance and forced response in a VNT turbine, Chin J Aeronaut (2018), https://doi.org/10.1016/j.cja.2018.02.015
231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263
266 267 268 269 270 271 272 273
CJA 1009 24 February 2018
No. of Pages 11
Numerical analysis for impacts of nozzle end-clearances on aerodynamic performance
Fig. 4
274 275 276 277 278 279 280 281 282 283 284 285 286 287 288
5
Comparison of performance of numerical model with and without spindles.
Fig. 5 shows turbine efficiency variations with the increased nozzle end-clearance size. It can be seen that the maximum flow rate through the turbine gradually increases with the increased size of nozzle end-clearance, but the turbine efficiency deteriorates significantly. The maximum deviation in the flow rate is up to 21 percent, and the efficiency drop is about 16 percentage points as the nozzle end-clearance increases to the largest size. It should be noted that the reason for such significant deviations is the use of the large nozzle clearance size in Case 6. This clearance size is slightly larger than that in current advanced VNT turbines. Meitner and Glassman13 proposed a clearance-flow model and stated that the changes in both turbine efficiency and flow rate became significantly larger with decreasing stator area and varied nearly linearly with clearance.
Fig. 5 Turbine efficiency variations with increased nozzle endclearance, at small opening.
The trend of the efficiency degradation increasing with the clearance size suggests that the nozzle end-clearance is unfavorable to the aerodynamic performance of a VNT turbine. This is why the mitigation of the nozzle end-clearance is still one of the main objects in both academic and industrial fields. But the thermal expansion caused by the high temperature environment makes it hard to further decrease the clearance size. Instead of directly decreasing the nozzle end-clearances, Sun et al.14 developed a VNT turbocharger with sliding vanes, and the totally new nozzle vanes successfully reduced the nozzle end-clearance leakage flow and improved the VNT turbine performance, especially at small opening. Yang et al.15 proposed a VNT turbine with a forepart rotation vane and numerically predicted results showed that the turbine efficiency was increased by 6 percent at small opening by removing partial nozzle end-clearances. These results prove that it is worth studying the nozzle end-clearances to find out more effective ways to improve the aerodynamic performance of a VNT turbine.
289
4.1.3. Effect of nozzle end-clearances’ distribution
309
Fig. 6 shows the turbine efficiency variations with the nozzle end-clearances’ distribution at several operation points within turbine map at the small opening. It is fairly clear that varying the nozzle end-clearances’ distribution not only affects the turbine flow capacity, but also changes the turbine efficiency. For instance, for the turbine with either nozzle shroud clearance (Case 12) or hub clearance (Case 7), the turbine flow capacity is obviously increased. Between those two cases, the maximum
310
Please cite this article in press as: ZHAO B et al. Numerical analysis for impacts of nozzle end-clearances on aerodynamic performance and forced response in a VNT turbine, Chin J Aeronaut (2018), https://doi.org/10.1016/j.cja.2018.02.015
290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308
311 312 313 314 315 316 317
CJA 1009 24 February 2018
No. of Pages 11
6
B. ZHAO et al.
Fig. 6
318 319 320 321 322 323 324 325 326 327 328 329 330 331 332 333 334 335 336 337 338 339 340 341 342 343 344 345 346 347 348 349 350 351 352 353 354 355 356 357 358 359
Turbine performance responses to various nozzle end-clearances’ distributions.
flow capacity occurs when all clearances are located at the hub side (Case 7), because of the relatively large distance between two rows at the hub side. Zhao et al.16 studied the rotor–stator interaction and proved that the blocking effect was closely related with the gap between two adjacent rows. However, when about 80 percent of the total nozzle end-clearance is located at shroud side (Case 11), the turbine flow capacity is the lowest. The maximum deviation in flow rate induced by nozzle end-clearances’ distribution is less than 5 percent. Beside the flow capacity, the turbine efficiency is always higher as small part of the total nozzle end-clearance is located at shroud side. Among those cases studied, no matter which point the turbine operates at, the peak efficiency always occurs when 20 percent of the total nozzle end-clearance size is located at shroud side (Case 8). In contract, the turbine efficiency is always deteriorated seriously as most clearances exist only at shroud side (Case 12). And the maximum deviation in turbine efficiency caused by varying nozzle end-clearances’ distribution is about 5 percentage points. It should be noticed that the use of the mixing plane approach at rotor/stator interface may show higher influence of nozzle end-clearance leakage flow on the turbine efficiency. Zhao et al.17 studied a VNT turbine using commercial code EURANUS and reported that compared with the unsteady simulation with direct interpolation at rotor/stator interface, steady simulations with the mixing plane method showed larger turbine efficiency variations in response to the different nozzle end-clearances. Roumeas and Cros18 tested two configurations: the nozzle gap divided equally at two ends of the nozzle vane and the whole gap located on the hub side. A significant deviation in turbine performance was observed. Walkingshaw and Spence19 tested a radial turbine with either hub side or shroud side stator vane clearance at full, middle and small openings, and revealed that the turbine performance differed for different nozzle clearances’ distribution. Those literatures provide evidences for that the nozzle end-clearances’ distribution significantly change the aerodynamic performance of a VNT turbine. In spite of that the current steady simulations may show higher influence of nozzle end-clearances, the effect on the turbine efficiency of the various nozzle end-clearances is still worth focusing.
4.2. Aerodynamic excitation
360
4.2.1. Turbine wheel resonant
361
A FEA calculation of a single blade was performed to predict the natural frequencies of turbine wheel. Since the critical modes for blade fatigue are mainly low order modes20, no higher order modes are analyzed in this paper. The numerically predicted frequencies are used as input for the Campbell diagram shown in Fig. 7. The aerodynamic excitation frequencies are plotted as a function of rotational speed of a turbine wheel. The two horizontal lines drawn in Fig. 7 represent the natural frequencies of Mode 1 and Mode 2, respectively. In the plot, the rotational speed and frequency was normalized by the turbine top speed and the natural frequency at Mode 1, respectively. The intersection points of natural frequencies with rotational speed orders are identified as potential operation points at which the blade resonance may occur. For the research turbine, resonances of the 1st speed order may excite blade Mode 1 whereas the 2nd speed order may excite Model 2 within a realistic operating range from the rotational speed of 0 to the top speed limit. According to the Campbell diagram, the critical rotational speed is easily known. And other parameters of the critical operation point can be obtained by running GT-Power software. In consequence, unsteady CFD simulations with those boundary conditions were carried out on a VNT turbine with
362
Fig. 7
Campbell diagram with Modes 1 and 2.
Please cite this article in press as: ZHAO B et al. Numerical analysis for impacts of nozzle end-clearances on aerodynamic performance and forced response in a VNT turbine, Chin J Aeronaut (2018), https://doi.org/10.1016/j.cja.2018.02.015
363 364 365 366 367 368 369 370 371 372 373 374 375 376 377 378 379 380 381 382 383 384
CJA 1009 24 February 2018
No. of Pages 11
Numerical analysis for impacts of nozzle end-clearances on aerodynamic performance 385 386
387 388 389 390 391 392 393 395 396 397 398 399 400 401 402 403 404 405 406 407 408 409 410 411 412 413 414 415 416 418 419 420 421 422
424 425 426 427 428 429 430 431 432 433
various nozzle end-clearances to predict the aerodynamic excitation pressure. 4.2.2. Loading variations with increased nozzle end-clearance size The aerodynamic loading, fðx; y; z; tÞ, is the differential of static pressure on both surfaces of a turbine blade, and represents the aerodynamic force that acts on blade surfaces to transfer energy from the fluid to the turbine wheel. fðx; y; z; tÞ ¼ pðx; y; z; tÞ pðx0 ; y0 ; z0 ; tÞ
ð1Þ
where f and p stand for the loading and pressure at an arbitrary point on blade surface, respectively; x, y, z represent the coordinate point at one surface of turbine blade and the x0 , y0 , z0 represent the coordinate point at another surface of turbine blade; the t is time. If the aerodynamic loading is steady, the turbine wheel will operate smoothly. In fact, the flow inside a VNT turbine is always highly unsteady, and therefore, there is serious aerodynamic interaction between the nozzle vanes and turbine wheel. This rotor–stator interaction must cause the aerodynamic loading to be high unsteadiness. The unsteady loading is a dominant excitation source to the vibration of a VNT turbine wheel. The aerodynamic loading can be divided into two parts, steady and unsteady parts. The steady part represents the time-average loading. And the unsteady part is corresponding to the fluctuating loading caused by the unsteady interaction and other complicated flow, which transfers exciting energy from the fluid to the turbine wheel to have the wheel deformed repeatedly and vibrated. They are expressed as follows: y; z; tÞ þ fðx; ~ y; z; tÞ fðx; y; z; tÞ ¼ fðx;
ð2Þ
where f stands for the time-average value of the loading and f~ represents for the unsteady part of the loading. The time-average loading is calculated as follows: Z T y; zÞ ¼ 1 fðx; fðx; y; z; tÞdt ð3Þ T 0 where T is for the time period that the pressure was monitored and recorded. Fig. 8 shows the distribution of the steady part of aerodynamic loading on wheel blade surface for four nozzle end-clearance sizes. It is fairly clear that high loading is centrally located in inducer. The reason is that at small opening the nozzle vanes are pivoted at the large setting angle so that the positive incidence is formed at inducer inlet. As the nozzle end-clearance size increases, the loading density near inducer blade tip and root obviously increases, but it decreases in mid-
Fig. 8
7
dle spanwise section in inducer. In exduer, the steady loading also increases with the increased clearance size. It is well known that the distribution of the steady loading is closely related with turbine efficiency and torque. When a turbine operates at small opening, the high loading density in inducer is in favor of recovering energy from exhaust gas, because an internal combustion engine is often to operate at part loads. Though the increased nozzle end-clearance size promotes the loading density in exducer, the decreased loading density in whole inducer deteriorates the turbine efficiency. Fig. 9 shows the variations of aerodynamic excitation intensity at both the 1st and the 2nd orders with the increased size of nozzle end-clearances. They are obtained by using Fast Fourier Transform (FFT) algorithm and expressed as follows: ~ y; z; tÞÞ Forceð1st;2ndÞ ¼ FFTðfðx; 0
ð4Þ
434 435 436 437 438 439 440 441 442 443 444 445 446 447 448 450
where Force is for the excitation force at the 1st order and/or the 2nd order. For the model without nozzle end-clearances, the aerodynamic excitation acted on the wheel blade is very weak in most regions. After adding nozzle end-clearances, however, the aerodynamic excitation intensities including the 1st and 2nd orders is reinforced not only in inducer tip and root regions but in exducer as well. Compared with the 2nd order aerodynamic excitation, the excitation intensity of the 1st order is stronger. Therefore, the 1st order excitation should transfer more energy to the wheel blade to have it vibrated and do much for the turbine wheel high cycle fatigue failure. And with the increased nozzle end-clearance size, the intensity of the 1st order excitation increases in the region near the blade tip round 70 % of blade chord. It should be noticed that the span from blade tip to root is so large in exducer that the high excitation intensity at the 1st order mode greatly increases the risk of the turbine wheel suffering from high cycle fatigue failure.
451
4.2.3. Loading variations with nozzle end-clearances’ distribution
469
Fig. 10 shows the variations of the steady loading distribution on a wheel blade surface with the nozzle end-clearances’ distributions. When most nozzle end-clearances, or all, are located at the hub side, the obviously high loading occurs near inducer blade root. And, with the decreased nozzle hub clearance and increased nozzle shroud clearance, the steady loading in the region near the exducer blade tip gradually increases. Meanwhile, the average loading density in the whole inducer decreases. The loading style in Case 12 is unfavorable to the recovery rate of energy from exhaust gas. Therefore, it can be drawn that the nozzle end-clearances’ distribution is able to change the distribution of the loading on a wheel blade.
471
Distribution of time-averaged loading on wheel blade.
Please cite this article in press as: ZHAO B et al. Numerical analysis for impacts of nozzle end-clearances on aerodynamic performance and forced response in a VNT turbine, Chin J Aeronaut (2018), https://doi.org/10.1016/j.cja.2018.02.015
452 453 454 455 456 457 458 459 460 461 462 463 464 465 466 467 468
470
472 473 474 475 476 477 478 479 480 481 482
CJA 1009 24 February 2018
No. of Pages 11
8
Fig. 9
B. ZHAO et al.
Variations of aerodynamic excitation at both the 1st and 2nd orders on wheel blade with increased size of nozzle end-clearances.
Fig. 10
Variations of steady loading acted on wheel blade with nozzle end-clearances’ distribution.
Fig. 11 Variations of aerodynamic excitation at both the 1st and 2nd orders on wheel blade surface with nozzle end-clearances’ distribution.
483 484 485 486 487 488 489
Fig. 11 shows the variations of aerodynamic excitation intensity at both the 1st and 2nd orders on the wheel blade surface with various nozzle end-clearances’ distribution. It is fairly clear that the aerodynamic excitation intensity in the region near blade tip around 70 % of chord is very weak when all nozzle end-clearances are located at the hub side, as shown in Case 7 in Fig. 11. However, as the nozzle shroud clearance
increases, the 1st order aerodynamic excitation intensity in that region gradually increases and is up to the maximum when all nozzle clearances are located at shroud side. Therefore, it can be concluded that the nozzle shroud clearance leakage flow is very sensitive to the 1st aerodynamic excitation intensity. Beside the excitation at the 1st order, the aerodynamic excitation with the 2nd order frequency also varies with the nozzle
Please cite this article in press as: ZHAO B et al. Numerical analysis for impacts of nozzle end-clearances on aerodynamic performance and forced response in a VNT turbine, Chin J Aeronaut (2018), https://doi.org/10.1016/j.cja.2018.02.015
490 491 492 493 494 495 496
CJA 1009 24 February 2018
No. of Pages 11
Numerical analysis for impacts of nozzle end-clearances on aerodynamic performance
502
end-clearances’ distribution. For instance, the excitation intensity at the 2nd order at inducer blade root gradually decreases with the decreased nozzle hub clearance. Meanwhile, it increases in the inducer blade tip region. But, compared with the 1st order excitation, the excitation intensity of the 2nd order is far weaker.
503
4.3. Forced response
504
4.3.1. Turbine wheel dynamic stress distribution
505
Independent calculations of the internal flow field and the harmonic response are presented in this paper, respectively. The calculations were carried out with two different computational meshes. The length scale of grid cells used in the CFD simulations is fully different on the wheel blade surface due to the use of clustering technology. However, the mesh used in the harmonic response calculation is almost uniform on the blade surface except the fillet region. With the Matlab code based on the cubic spline function interpolation, the data transmission was performed to transfer the information of the loading fluctuation on the grid nodes of the flow field to the grid points of the solid field. The evaluation of the accuracy of the interpolation is illustrated in Fig. 12. The comparison of two plots shows very small deviations, so that the interpolation method used in this paper could be accurate enough to meet the requirements of
497 498 499 500 501
506 507 508 509 510 511 512 513 514 515 516 517 518 519 520
Fig. 12
Fig. 13
9
current research work. After the interpolation, the aerodynamic loading was further analyzed by FFT Algorithm to obtain the real and imaginary components which would be used for the FEA calculations. Based on the analysis of the harmonic response, the stress distributions at both the 1st and the 2nd orders are illustrated in Fig. 13, respectively. The magnitude is normalized by dividing the vibration stress value by the peak value in the case with normal nozzle end-clearances (Case 5). From the contour distribution in two plots in Fig. 13, it can be seen that the maximum stress at the 1st order mode is about 2.5 times than the peak stress at the 2nd order and occurs at exducer blade root. This numerical result points out that the vibration stress of the 1st order induced by the 1st harmonic loading can be considered as the major alternating stress on the turbine wheel and the exducer blade root region may suffer from high cycle fatigue failure. The numerically predicted region with high risk is in agreement with the experimental result of a similar turbine reported by Hu et al.12 and illustrated Fig. 14.
521
4.3.2. Forced response of turbine wheel
540
Fig. 15 shows the variations of the peak vibration stress at both the 1st and 2nd orders with the increased height of the nozzle end-clearances which were equally divided at two ends of nozzle vanes. It is fairly clear that the presence of nozzle end-clearances does increase the peak stress of the 1st order
541
Validation of the interpolation approach.
Distribution of stress on wheel blade (Case 5).
Please cite this article in press as: ZHAO B et al. Numerical analysis for impacts of nozzle end-clearances on aerodynamic performance and forced response in a VNT turbine, Chin J Aeronaut (2018), https://doi.org/10.1016/j.cja.2018.02.015
522 523 524 525 526 527 528 529 530 531 532 533 534 535 536 537 538 539
542 543 544 545
CJA 1009 24 February 2018
No. of Pages 11
10
Fig. 14 Example of high cycle fatigue failure at exducer blade root, observed on a similar turbine wheel.
Fig. 15 Turbine wheel maximum stress variations with increased nozzle end-clearances.
Fig. 16 Turbine wheel maximum stress variation with nozzle end-clearance distribution.
546 547 548 549 550 551 552 553 554 555 556 557 558 559 560 561
vibration. And, with the increased size of nozzle clearances, the peak value of the stress at the 1st order mode gradually increases. Compared with the model without nozzle endclearances (Case 3), the 1st order stress is increased by twice as much. This trend means that the nozzle clearance leakage flow is one of the major resources contributing to the damage of a turbine wheel due to the high cycle fatigue. Besides the 1st order excitation, the peak stress of the 2nd order vibration is slightly changed with the increased nozzle end-clearances. This decreased trend reveals that the nozzle clearance leakage flow acts as the 1st order aerodynamic excitation source for the current turbine wheel. Fig. 16 shows the variation of the peak stress at both the 1st and 2nd orders with the various nozzle end-clearances’ distribution. As was analyzed above, with the decreased nozzle end-clearance on the shroud side the turbine efficiency signifi-
B. ZHAO et al. cantly increases. Herein, the curve of the peak stress at the 1st order mode clearly shows that the peak stress obviously decreases with the decreased nozzle shroud clearance, and the maximum deviation is about 30%. These results reveal that the nozzle shroud end-clearance is more sensitive not only to the turbine efficiency but to the forced response of the current turbine wheel at the 1st order mode as well, compared with the nozzle hub clearance. In contrast, the forced response at the 2nd order mode firstly increases and then slightly decreases with the decreased nozzle shroud clearance. The narrow variation range of the peak stress at the 2nd order means that the effect of the nozzle end-clearances’ distribution on the wheel vibration at the 2nd order mode is insignificant. According to the above analysis on the harmonic response, not only the presence of the nozzle end-clearances but also the increased size of nozzle shroud clearance does increase the risk of the turbine wheel suffering from high cycle fatigue failure and, between the shroud and hub clearances, the nozzle shroud clearance has more contribution for the vibration of the turbine wheel at the 1st order mode. These results, therefore, suggest a possibility to improve both the turbine efficiency and the reliability of a turbine wheel by reducing the unfavorable effect of the nozzle shroud clearance. For instance, Wang et al.21 proposed a controlling structure to decrease the nozzle shroud clearance to improve the aerodynamic performance of a VNT turbine. This method, or other approaches with similar principles, may also benefit the turbine performance, while also solving the high cycle fatigue issue that the design and development of an automotive turbocharger is trying to address. Besides, an introduction of 3D nozzle vane with oblique trailing edge may be possible to mitigate the unfavorable impact of the nozzle shroud clearance. The differential of pressure on both sides of a nozzle vane is the force to drive the air over the narrow nozzle end-clearances. Decreasing the pressure difference at shroud side by increasing the nozzle opening only at shroud side is likely to reduce the nozzle shroud clearance leakage flow; in the meantime, the opening of the nozzle vanes at the hub side needs to be decreased in order to keep the same nozzle throat.
562
5. Conclusions
601
This paper numerically investigated the effects of various nozzle end-clearances on both the aerodynamic performance and the forced response of a VNT turbine in order to find out the possibility to improve both the stage efficiency and the reliability of a turbine wheel via optimizing nozzle end-clearances. Research models are a VNT turbine with different total sizes of nozzle end-clearances which are equally divided at two ends of nozzle vanes or with various distributions of the clearances at the hub and shroud sides which have the same total size.
602
(1) As the total nozzle end-clearances remain the same, the nozzle end-clearances’ distribution is closely related to the turbine efficiency, the flow capacity, and the vibration stress at the 1st order mode at small openings. When most nozzle end-clearances are located at shroud side, a VNT turbine has the minimum flow capacity while the flow capacity is the maximum when no nozzle clearance is located at shroud side. When small part of the total nozzle end-clearances is located at shroud side,
611
Please cite this article in press as: ZHAO B et al. Numerical analysis for impacts of nozzle end-clearances on aerodynamic performance and forced response in a VNT turbine, Chin J Aeronaut (2018), https://doi.org/10.1016/j.cja.2018.02.015
563 564 565 566 567 568 569 570 571 572 573 574 575 576 577 578 579 580 581 582 583 584 585 586 587 588 589 590 591 592 593 594 595 596 597 598 599 600
603 604 605 606 607 608 609 610
612 613 614 615 616 617 618 619
CJA 1009 24 February 2018
No. of Pages 11
Numerical analysis for impacts of nozzle end-clearances on aerodynamic performance 620 621 622 623 624 625 626 627 628 629 630 631 632 633 634 635 636 637 638 639 640 641 642
the peak efficiency always occurs and the peak stress of the 1st order vibration at exducer blade root can be decreased greatly. (2) The nozzle end-clearance leakage flow has stronger effects on vibration stress at the 1st order mode for the current model than that of the 2nd order. Compared with the nozzle hub clearance, the nozzle shroud clearance is more sensitive to the turbine efficiency degradation and the forced response of a turbine wheel at the 1st order, but has less contribution to increase the turbine flow capacity. (3) It has reached broad agreement that decreasing total size of the nozzle end-clearances can benefit the aerodynamic performance of a VNT turbine. But due to thermal expansion of the material, it is very hard to further decrease the size in the development of an advanced VNT turbine. According to the results reported in this paper, controlling the nozzle end-clearances’ distribution, or decreasing the nozzle shroud clearance’s effect, may be possible methods to design an advance VNT turbine with higher efficiency and lower forced response in the future, instead of the hard efforts to further decrease the total size of the nozzle end-clearances.
643
644
Acknowledgements
645
This study was co-supported by the Natural Science Foundation of Hebei Province of China (No. E2017402135) and the Program of Science and Technology Research and Development of Handan of China (No. 1621212047-2).
646 647 648
649 650 651 652 653 654 655 656 657 658 659 660 661 662 663 664
References 1. Tang H, Pennyycott A, Akehurst S, Brace AJ. A review of the application of variable geometry turbines to the downsized gasoline engine. Int J Engine Res 2015;16(6):810–25. 2. Feneley AJ, Pesiridis A, Andwari AM. Variable geometry turbocharger technologies for exhaust energy recovery and boosting – a review. Renew Sustain Energy Rev 2017;71(1):959–75. 3. Hu L, Yang C, Sun H, Zhang J, Lai M. Numerical analysis of nozzle clearance’s effect on turbine performance. Chin J Mech Eng-en 2011;24(4):618–25. 4. Walkingshaw J, Spence S, Ehrhard J, Thornhill D. A numerical study of the flow fields in a highly off-design variable geometry turbine. Glasgow: ASME; 2010, Report No.: GT2010-22669. 5. Tamaki H, Goto S, Unno M, Iwakami A. The effect of clearance flow of variable area nozzles on radial turbine performance. Berlin: ASME; 2008, Report No. GT2008-50461.
11
6. Tamaki H, Unno M. Study on flow fields in variable area nozzles for radial turbines. Int J Fluid Mech Syst 2008;1(1):47–56. 7. Chen H. Turbine wheel design for Garrett advanced variable geometry turbines for commercial vehicle applications. In: Proceedings of the 8th international conference on turbochargers and turbocharging, 2006 May 28–29; London, UK; 2006. 8. Zhao B, Yang C, Hu L, Sun H, Yi J, Eric C, et al. Understanding of the interaction between clearance leakage flow and main passage flow in a VGT turbine. Adv Mech Eng 2015;7(2):652769. 9. Liu Y, Yang C, Qi M, Zhao B. Shock, leakage flow and wake interactions in a radial turbine with variable guide vanes. Dusseldorf: ASME; 2014, Report No.: GT2014-25888. 10. Zhao B, Hu L, Sun H, Yang C, Shi X, Yi J, et al. Numerical investigation of a novel approach for mitigation of forced response of a variable geometry turbine during exhaust braking mode. Seoul: ASME; 2016, Report No.: GT2016-56342. 11. Lu FA, Qi DT, Wang XJ, Zhou Z, Zhou HH. A numerical optimization on the vibroacoustics of a centrifugal fan volute. J Sound Vib 2012;331(10):2365–85. 12. Hu L, Sun H, Yi J, Curtis EC, Morelli A, Zhang J, et al. Investigation of nozzle clearance effects on a radial turbine: aerodynamic performance and forced response. Detroit: SAE; 2013, Report No.: 2013-01-0918. 13. Meitner PL, Glassman AJ. Off-design performance loss model for radial turbine with pivoting, variable-area stators. NASA Technique paper 1708. 14. Sun HH, Zhang J, Hu L, Hanna DR. Sliding vane geometry turbines: United States patent US 20130042608; 2013 Feb 21. 15. Yang D, Lao D, Yang C, Hu L, Sun H. Investigations on the generation and weakening of shock wave in a radial turbine with variable guide vanes. Seoul: ASME; 2016, Report No.: GT201657047. 16. Zhao B, Yang C, Chen S, Qi M, Zhou M. Unsteady flow variability driven by rotor-stator interaction at rotor exit. Chin J Aeronaut 2012;25(6):871–8. 17. Zhao B, Ma C, Hu L, Yang C. Numerical investigation on effects of uncertain nozzle vane clearance on variable geometry radial turbine performance. J Propul Technol 2014;35(4):492–8 [Chinese]. 18. Roumeas M, Cros S. Aerodynamic investigation of a nozzle clearance effect on radial turbine performance. Copenhagen: ASME; 2012, Report No.: GT2012-68835. 19. Walkingshaw JR, Spence SWT, Thornhill D, Ehrhard J. An experimental assessment of the effects of stator vane clearance location on an automotive turbocharger turbine. London, UK: Woodhead Publishing Limited; 2012, Reprot No.: C1340/ 007. 20. Ma C, Huang Z, Qi M. Investigation on the forced response of a radial turbine under aerodynamic excitations. J Therm Sci 2016;125(2):130–7. 21. Wang L, Hou H, Wang K, Yang Y. An adaptive device and method to control nozzle endwall clearances in a variable nozzle turbine. China Patent CN103352729; 2016 Oct 16 [Chinese].
Please cite this article in press as: ZHAO B et al. Numerical analysis for impacts of nozzle end-clearances on aerodynamic performance and forced response in a VNT turbine, Chin J Aeronaut (2018), https://doi.org/10.1016/j.cja.2018.02.015
665 666 667 668 669 670 671 672 673 674 675 676 677 678 679 680 681 682 683 684 685 686 687 688 689 690 691 692 693 694 695 696 697 698 699 700 701 702 703 704 705 706 707 708 709 710 711 712 713 714 715 716 717