A resistance type vertical axis wind turbine for building

16 The Persian or Sistan wind mill is possibly the oldest wind energy device. It consists of a vertical axis 17 with six blades, and an outer shroud which encases half the rotor against the wind. The wind only 18 acts on one half of the runner, providing the driving force. The efficiency of this machine was 19 assumed to be between 5 and 14%, too low for practical application today. The concept however is 20 interesting since it has the potential for integration into buildings. Recently, the Sistan wind mill was 21 re-visited to assess whether the performance could be improved. Initial exploratory tests reported in 22 the literature indicated potential. At Southampton University, a series of tests was conducted with a 23 model of 0.6 m diameter and 0.5 m high runner employing an improved measurement and data 24 acquisition system. Two geometries were investigated. Qualitative tests indicated that a gap 25 between blade and axis is essential for functionality. Performance tests with an improved geometry 26 resulted in efficiencies of 0.4 to 0.5, similar to e.g. Darrieus-type VAWTs, for blade to wind speed 27 ratios of 0.82 to 1.8. The modified resistance-type vertical-axis wind turbine appears to have 28 potential for further development. 29


33
The integration of wind turbines into buildings constitutes an interesting aspect of wind energy 34 application. The direct proximity of generation and end user would create very favourable conditions 35 for wind energy usage. A brief overview over current concepts of wind energy integration into 36 building is given in [1]. Vertical Axis Wind Turbines (VAWTs) are hereby considered as better suited 37 for this purpose. The most common VAWT today is the Darrieus-turbine, which employs the forward 38 acting forces generated on an airfoil. Such turbines have an efficiency of up to 0.40 [2]. They do not 39 need active components since they rotate in the same direction irrespective of the direction of the 40 wind although speed control is necessary. There are however several factors which negatively affect 41 the integration of wind turbines into buildings and built-up areas: 42 1. Architecture: propeller turbines and vertical axis wind turbines (VAWTs), which employ the 43 Darrieus principle, are difficult to integrate into buildings. 44 2. Turbulence: both propeller and Darrieus type turbines are affected by the turbulent air flow 45 created by buildings and surface obstacles. 46 3. Environment: the fast moving tips of propeller blades are considered as dangerous for many 47 flying animals such as birds and bats. In addition, the blades create moving shadows and low 48 frequency noise, which may affect people working or living near these installations. 49 Today, there is no turbine type which is easy to integrate into buildings. The principle of the oldest 50 wind energy converter, the Persian or Sistan wind mill, may however offer a solution. 51 2 Literature review 52 The Persian or Sistan wind mill is the oldest type of wind mill, it origins go back to the 9 th Century, e.g 53 [3]. This wind mill constitutes resistance type wind energy converter. It consists of a vertical axis 54 runner, often 3-4 m high, with six to eight blades of 2 to 3 m width, Fig. 1a. One side of the runner, 55 the side moving against the direction of the wind, is encased in a semi-circular wall. On the other 56 side, a guide wall focuses the wind onto the blade, Fig. 1b. This arrangement was effective, since in 57 Khorasan the wind blows in one direction only for 120 days per year. The Persian wind mill was until 58 recently considered as of historical interest only. The efficiency is given in the literature as η = 0.14, 59 and sometimes as low as 0.05, e.g. [4]. These values were determined using the blade area only for 60 the calculation of the available wind energy. In order to make them comparable with other wind 61 turbines, these factors need to be reduced by at least 50% to take account of the total area of the 62 mill exposed to the wind. 63 The Sistan turbine constitutes a visually solid block, contrary to most other wind turbines. This 69 makes the architectural integration of this turbine type into buildings much easier. As a resistance or 70 drag type machine, it can be expected that the Sistan turbine is less affected by turbulence than 71 other VAWTs such as the Darrieus turbine. This is also considered a potential advantage although it 72 has not yet been demonstrated experimentally. Today, the adjustment of the VAWT to the wind 73 direction is possible e.g. by providing a movable outer section or shroud. Compared with propeller 74 type Horizontal Axis Wind Turbines (HAWT), this has the advantage that the rotating part of the 75 turbine does not have to be moved. Fig. 1c shows such an arrangement. In addition, the 76 comparatively slow speed of the rotor combined with its high visibility will very probably mean 77 reduced noise emission and a better ecological performance. Propeller type wind turbines are 78 coming increasingly under criticism for damages to birds, and other flying species of animals [7]. 79 Müller et al. were the first to actually test a model of the Sistan-type turbine [5]. Their motivation 80 was to assess its energy generation and also the potential for building integration. The model tests 81 resulted in efficiencies of 15% for maximum power for a blade to wind speed ratio of 0.3, using the 82 total front area of the model as basis for the determination of the input energy (the efficiency was 83 0.42 when calculated using the blade area only). This was substantially more than the values given in 84 the literature, and was considered a promising starting point. The tests were however preliminary, 85 since the measurement equipment was not adapted to deal with strong variations in wind speed. 86 Following this paper, several modifications of the theme were investigated by researchers: 87 A problem of the Sistan type VAWTs for present-day application is, that they need to be adjustable 88 to the wind direction. One possibility to do this, is to rotate the outer shroud [5]. In [8], a static 89 outer flow guidance structure, termed external power-augmentation-guide vane (PAGV) for a Sistan-90 type rotor is described, which aligns the direction of the air flow with the turbine blades irrespective 91 of the wind direction. The freely rotating three blade rotor reached a blade to wind speed ratio of vB 92 / vw = 0.44. A combined experimental and numerical analysis was performed in order to estimate the 93 power production. This resulted in a maximum efficiency of 18.3% [9]. 94 A modification of the principle, with a 1.64 m outer diameter rotor with eight blades of 0.24 m 95 breadth, and a shroud to cover the returning side of the rotor was tested recently. Efficiencies were 96 17% at a blade to wind speed ratio of vB / vw of 0.37 [10]. They conducted wind tunnel tests, and numerical simulations. This work was subsequently extended, 101 looking at more detail into the effect of the stator [12]. Further numerical studies were conducted 102 with a 9-blade, Sistan-type turbine with external guide structure and a closed top [13]. The rotor had 103 a diameter of 6.5 m, with the 1 m wide blades fixed at the end of the radius. The inlet section was 104 11.50 m wide, and funnels the wind into the 3 m wide inflow opening. The authors investigated the 105 flow inside the turbine housing in detail, but it is unclear whether any negative pressures resulting 106 from flow separation at the downstream end were considered. Also, the counter-rotating side of the 107 downstream end of the turbine appears not to have been covered. The numerical model predicted a 108 blade speed to wind speed ratio of vB / vw = 5.4. The authors also comment on the formation of 109 vortices inside the turbine, their dynamics and potential negative effects. 110 From the available literature it can be concluded that resistance type VAWTs have the potential to 111 be effective wind energy converters which can more easily be integrated into buildings than e.g. 112 Darrieus type VAWTs. 113 3 Experimental set-up 114

115
At Southampton University, a series of model tests on a Sistan-type, resistance VAWT was 116 conducted. The aim of the project was to determine conversion efficiencies experimentally, and to 117 improve the geometry. 118 The experiments were conducted with a model VAWT as shown in Fig. 2 The bottom (downstream) half of the shroud can be removed, so that both constellations -with and 127 without downstream shroud -can be tested. This prevents turbulence at the inflow, and reduces 128 the counteracting effect of negative pressures at the outflow. Test were conducted outside, in a 129 narrow road between two buildings, and in a large wind tunnel facility. 130

131
A dedicated LabView based stand-alone measurement system was developed to allow for 132 simultaneous acquisition of wind speed, runner speed and friction force data. The wind speed was 133 monitored using a UNI-T UT361 anemometer placed at roughly 0.5m in front of the model at 1m 134 height from ground level. The friction force Ffr on the Prony brake was measured with a TEDEA 135 Model 0355 load cell. The rotational speed n of the friction wheel was recorded using a compact 136 digital optical tachometer A2103/LSR/001 from Compact Instruments pointed at the friction wheel. 137 It has speed range of 3 to 99,999 rpm. The data from all three sensors was collected digitally with a 138 laptop using LabVIEW (LV2014) based measurement and data analysis software. The anemometer 139 was checked against the wind tunnel Pitot tubes, the tachometer against a stopwatch and the force 140 transducer was calibrated at the beginning of each test series and checked against drift at the end. 141 All three devices were found to be accurate within the manufacturer's information. 142 The torque MT generated by the model could be determined by multiplying the friction force Initially, the blades of the runner had a width of 370 mm and extended directly from the axis to the 163 shroud, with a 5 mm gap between the blade and the outer wall. This geometry was deployed outside 164 for first exploratory tests. It appeared however that the model was not self starting. For lower wind 165 speeds below 3 m/s the runner oscillated back-and forward, for higher wind speeds it moved slowly. 166 An assessment of the situation led to the assumption that, with a static rotor, a hydrostatic pressure 167 situation was generated at the inflow whereby the stagnation pressure of the air flow acted on one 168 blade in the direction of rotation, but on the following blade against that direction. An analysis of 169 pictures of Persian Wind Mills showed that there always was a gap of around 1/6 of the blade width 170 or more between blade and axis. This, it was realised, would create a very different flow condition 171 where the air always flows through the box, see also [13]. The blade width was subsequently 172 reduced to 300 mm with a 50 mm gap between blade and axis, and a 25 mm gap between blade and 173 outer wall. The first tests were conducted with a torque of 0.  In this section, the experimental results will be shown as original and non-dimensional data. The 231 load had to be pre-set, and the wind speed could not be controlled. This means that every test run 232 contains a multitude of conditions. In order to identify the main system characteristics, it was 233 therefore decided to show not just power out against wind speed and efficiency against non-234 dimensional blade speed, but also the relationships of actual and non-dimensional parameters. Fig.  235 5a shows the power generated as a function of wind speed for both geometries. of vB / vw = 0.8 to 1 (Fig. 6a, Geometry 1) to vB / vw = 1.0 to 2 (Fig. 6b, Geometry 2). The efficiency as function of the blade to wind speed ratio is finally shown in Fig. 8. Again, geometry 262 2 shows better results for higher blade to wind speed ratios vB / vw. Maximum efficiencies reach 0.42 263 to 0.5 for vB / vw = 0.75 to 1.25 (Fig. 8a, Geometry 1), and 0.42 to 0.5 for vB / vw = 0.85 to 1.8 (Fig. 8b,  264 Geometry 2). Geometry 2 also allows for higher operational blade to wind speed ratios of up to 2.7. 265 The downstream shroud therefore allows for higher runner speeds, better power conversion and 266 increased power output. 267 6 Discussion 268 6.1 Qualitative tests 269 The initial tests, which were conducted with only friction torque applied, resulted in a maximum 270 blade to wind speed ratio of 1.5. The gap between blade and axis was found to be an essential detail 271 of the system. The results from a three blade rotor without gap between blades and axis reported in 272 the literature may also have been influenced by this effect [8], It had been decided to conduct experiments in a narrow channel between two buildings since the 297 costs for tests in a wind tunnel were too high. The disadvantages of outside tests are mainly the 298 variable and unpredictable wind speed, and the uncertainty about wind direction. To some extent 299 the latter factor could be balanced by choosing an appropriate site with temporary unidirectional 300 flow conditions. The measurement system was designed accordingly so that wind speed, runner 301 speed and friction force could be measured simultaneously. 302 The tests resulted in efficiencies 0.45 to 0.5 for wind speed ratios of around 1 to 1.8, and a 303 maximum blade to wind speed ratio of 2.5. This indicates the effect of the deflector, which funnels 304 the air flow into the operating section. The efficiency is however not just a function of the applied 305 torque blade speed ratio, but also of the actual wind speed, reaching a maximum for lower wind 306 speeds and reducing to 0.4 for a wind speed of 5 m/s. This indicates that wind speed related 307 turbulent losses are present. 308 Geometry 2 with a downstream shroud led to a significant improvement of the performance. With 309 efficiencies of up to 0.5, the values from the tests reported here are significantly larger than those 310 reported in the literature (0.15). The experiments reported in the literature were however all 311 conducted in wind tunnels, so that vortex shedding and the associated detrimental effects cannot be 312 excluded. In addition, the effect of a downstream shroud to avoid counteracting forces on the rotor 313 was not considered. In the numerical models it appears that the effect of downstream separation 314 was not included either, which must lead to lower efficiencies. 315  in this study of 1.31 to 3.15 (rows 9 and 10), but close to those from the wind tunnel tests (row 8). 329

316
In the tests described in [9], the blade extends from the axis to the outer rim of the augmenting vane 330 section. The gap between blade and axis, which the experiments described in this article showed to 331 be very important, is missing. This may to some extent balanced by the fact that the model 332 described in [9] only had three blades. 333 6.5 Operational conditions 334 Operational conditions in a natural environment may affect the performance of a wind turbine. Two 335 aspects, the effect of cross winds or wind components normal to the prevailing wind direction and 336 the effect of wind speeds exceeding the design wind speed will be discussed here briefly: 337 Cross winds or wind vector components normal to the turbine axis can have negative effects on 338 HAWTs, leading e.g. to additional dynamic loadings on the turbine blades. The VAWT described in 339 this article is expected to tolerate large wind vector angles of up to 45 degrees to the rotor channel 340 without additional effects, apart from a reduction in power output. The wind can enter the rotor 341 channel from the deflector side for an angle of 45 degrees, from the rotor channel side at even 342 larger angles. Here, the reduction in power would be more significant since a vector component of 343 the air flow points against the direction of rotation. This may need to be addressed by a slight 344 change in geometry. The turbulence associated with cross winds should not have any significant 345 effect on power generation since the VAWT relies on resistance drag conversion rather than 346 aerodynamic effects. 347 348 The rotor speed is limited by the design speed of the generator. An further increase of the wind 349 speed above design speed could theoretically be compensated by the rotor speed remaining 350 constant, so that the blade speed ratio and the efficiency reduces. However, say for a 20% wind 351 speed increase the available power would increase by 73% whilst the efficiency only drops by 4%. 352 Reducing the rotor speed further to match the power rating at a lower speed would overload the 353 generator. A solution for this problem could be to turn the shroud, the external hull, slightly out of 354 the wind direction. This will reduce the wind energy input, so that the effective power generated by 355 the rotor and its speed can therefore remain constant. Contrary to HAWT's, stall does not occur for 356 resistance type wind turbines. It can be expected that the turbine will remain operational for wind 357 speeds significantly higher than the design wind speed. The effect of an increased wind speed and an 358 oblique wind angle on flow induced vibrations will need to be assessed in wind tunnel experiments. 359 If the flow induced vibrations are linked to the flow through the rotor, then it may be necessary to 360 shut down operation by turning the outer shroud into the wind direction. 361 In general, the authors expect that the resistance-type VAWT has a wider operational range of wind 362 speeds than airfoil-type VAWTs and HAWTs. 363 364

365
With mechanical efficiencies of 0.5, the resistance VAWT is in interesting alternative to propeller 366 type turbines. A 6 m diameter, 10 m high rotor could produce approximately 30 kW electrical energy 367 for the wind speed of 12.5 m/s. Since the resistance type machines do not rely on aerodynamic uplift, 368 stall -i.e. the sudden loss of uplift under high air flow velocities / high angles of incidence -cannot 369 occur. Operation in higher wind speeds may be possible by turning the rotor channel slightly out of 370 the wind as described in the previous section. Also, it can be expected that the effect of turbulence 371 will be small, making the application in built-up areas easier. All these aspects require further 372 investigation. 373 The original Sistan wind mills were built for wind coming from just one direction. The modern 374 VAWTs are envisaged to have a movable outer shroud, which can turn the opening into the 375 prevailing wind direction. Contrary to current propeller type HAWTs, the runner will not have to be 376 moved, reducing structural complexity. 377

378
A series of quantitative and qualitative model tests with a vertical axis, resistance wind turbine were 379 conducted in order to determine geometric parameters and to assess the performance. Qualitative 380 tests served to identify important design details. Wind tunnel tests led to the discovery of 381 unexpected aerodynamic effects. Outside tests showed the actual performance under variable 382 conditions. The following conclusions were drawn: 383  A gap between blades and axis of approximately 1/6 of the blade width is essential for the 384 functionality of the resistance type VAWT. 385  In wind tunnel test, periodic vortex shedding occurred, which created very high forces on 386 the turbine model and strongly reduced the performance 387  Two geometries -Geometry 1 with an open downstream side, and Geometry 2 with a closed 388 downstream side -were investigated. 389  With minimum torque applied, blade velocities can reach up to 2.5 times the wind speed. 390  Geometry 2 with showed the better performance. Efficiencies ranged from 0.42 to 0.5 and 391 0.25 for blade to wind speed ratios of vB / vw = 0.85 to 1.8 and 2.5. 392 It appears that the Sistan type VAWT could be an interesting wind energy converter for effective 393 building integration and possibly even for application in tower-like structures. 394 395