QTZ40塔式起重機(jī)——臂架優(yōu)化設(shè)計,qtz40,塔式起重機(jī),優(yōu)化,設(shè)計 Ocean Engineering 29 (2002) 1463–1477 Analysis of Wells turbine design parameters by numerical simulation of the OWC performance A. Brito-Melo, L.M.C. Gato * , A.J.N.A. Sarmento Mechanical Engineering Department, Instituto Superior Te′cnico, Technical University of Lisbon, Av. Rovisco Pais, 1049-001 Lisbon, Portugal Received 22 May 2001; accepted 30 August 2001 Abstract This paper investigates by numerical simulation the influence of the Wells turbine aerody- namic design on the overall plant performance, as affected by the turbine peak efficiency and the range of flow rates within which the turbine can operate efficiently. The problem of match- ing the turbine to an oscillating water column (OWC) is illustrated by taking the wave climate and the OWC of the Azores power converter. The study was performed using a time-domain mathematical model based on linear water wave theory and on model experiments in a wave tank. Results are presented of numerical simulations considering several aerodynamic designs of the Wells turbine, with and without guide vanes, and with the use of a bypass pressure- relief valve. ? 2002 Elsevier Science Ltd. All rights reserved. Keywords: Wave energy; Oscillating water column; Equipment; Wells turbine 1. Introduction The Wells turbine has been the most commonly adopted solution to the air-to- electricity energy conversion problem in oscillating water column (OWC) wave energy converters. These essentially consist of a capture pneumatic chamber, open at the bottom front to the incident wave, a turbine and an electrical generator. The incident wave motion excites the oscillation of the internal free surface of the entrained water mass in the pneumatic chamber, which produces a low-pressure reci- * Corresponding author. Tel.: +351-21-841-7411; fax: +351-21-841-7398. E-mail address: lgato@hidro1.ist.utl.pt (L.M.C. Gato). 0029-8018/02/$ - see front matter ? 2002 Elsevier Science Ltd. All rights reserved. PII: S 00 29 -8018(01)00099-3 1464 A. Brito-Melo et al. / Ocean Engineering 29 (2002) 1463–1477 procating flow that drives the turbine. A few full-scale turbine prototypes have been built and installed in grid-connected power plants in European countries, e.g. the 500 kW Wells monoplane turbine with guide vanes installed in the Island of Pico, Azores (Falca?o, 2000), and 2×250 kW biplane contrarotating turbine of the LIMPET plant, at Islay, Scotland (Heath et al., 2000). The greatest challenges to designers of equipment for wave energy converters are the intrinsically oscillating nature and the random distribution of the wave energy resource. These features are absent or much less severe in other competing energy technologies. The air turbine in an OWC converter is subject to flow conditions (randomly reciprocating flow), which, with respect to efficiency, are much more demanding than in turbines in almost any other application. The Wells turbine, while reaching only a moderately high peak efficiency as compared with conventional tur- bines, can operate in reciprocating flow without the need of a rectifying valve system. The turbine, on the one hand, is required to extract energy from air whose flow rate, in each of the two directions, oscillates between zero and a maximum value, which in turn has an extremely large variation from wave to wave and with sea conditions. On the other hand, at fixed rotational speed, turbines in general, and Wells turbines in particular, are capable of operating with good efficiency only within a limited range of flow conditions around the peak efficiency point. The power output of Wells turbines is known to be low (or even negative) at small flow rates (the flow rate passes through zero twice in a wave cycle) and it drops sharply for flow rates above a critical value due to aerodynamic losses produced by rotor blade stalling. Therefore, the turbine is expected to perform poorly in very energetic sea-states or whenever violent wave peaks occur. Mounting a bypass pressure-relief valve on the top of the air chamber as proposed in the Azores plant may prevent this problem. The valve is controlled to limit the maximum pressure and suction in the chamber (depending on the turbine rotational speed) to prevent the instantaneous air flow rate through the turbine from exceeding the values above which aerodynamic stalling at the rotor blades would produce a severe fall in power output. Numerical simulations (Brito- Melo et al., 1996; Falca?o and Justino, 1999) indicate that a reduction in turbine size and a substantial increase in the annual production of electrical energy might be achieved by the use of a bypass pressure-relief valve. Moreover, recent studies (theoretical and model testing) indicate that blade sections especially designed for Wells turbine rotors can significantly enlarge the range of flow rates within which the turbine operates efficiently and reduce aerodynamic losses under partially stalled flow conditions, in comparison with other blade designs which give a higher peak efficiency within a narrower range of flow rates through the turbine. This raises the question of whether, in view of the total annual produced electrical energy and taking into account the hydrodynamic performance of the OWC device, it is more appropri- ate to select a turbine aerodynamic design which allows an enlarged range of flow rates at which the turbine operates efficiently or whether it is better to adopt a turbine design which gives a higher peak efficiency value with a reduced range of flow rates at which the turbine operates efficiently. Furthermore, it is of interest to know to what extent this issue might be dependent on the use of a pressure-relief valve. The main objective of the present work is to investigate the influence of the Wells 1465A. Brito-Melo et al. / Ocean Engineering 29 (2002) 1463–1477 turbine aerodynamic design on the overall plant performance, as affected by the turbine peak efficiency and the range of flow rates within which the turbine can operate efficiently. Realistic characteristics are assumed for the turbine and the use of a bypass pressure-relief valve is also considered. Since the resulting pressure changes in the chamber are dependent on the turbine characteristics and the pressure- relief valve influences the turbine operation, the hydrodynamic process of energy extraction is also modified. The hydrodynamics of the conversion of wave energy into pneumatic energy is predicted by using a time-domain mathematical model based on linear water wave theory and on model experiments in a wave tank as described in Sarmento and Brito-Melo (1996). The conversion of pneumatic energy into electrical energy is predicted by a suitable computational model of the power take-off equipment based on the results extrapolated from aerodynamic tests on a scale-model and on empirical approximations for the generator losses (Brito-Melo et al., 1996). This paper presents the results of numerical simulations considering several aerodynamic designs of the Wells turbine, with and without guide vanes, and the use of the pressure-relief valve. The problem of matching the turbine to an OWC is illustrated by taking the wave climate and the OWC of the Azores wave power converter. 2. Wave-to-wire model 2.1. Plant operation The wave-to-wire model concerns the operation of an OWC equipped with a Wells turbine, a bypass valve of unlimited capacity and a variable speed turbo-generator, under a set of representative sea-state conditions. The Wells turbine is known to exhibit an approximately linear relationship between the turbine pressure drop p(t) and the flow rate q t (t). Then we may write the turbine characteristic as K H11005 p(t)/q t (t) H11005 p s (H9024)/q s (H9024), where p s (H9024), and q s (H9024) are maximum values of pressure and flow rate (prior to the onset of aerodynamic stall at the turbine rotor blades), which (for a given turbine) depend on the turbine rotational speed H9024. The use of a properly controlled bypass pressure-relief valve prevents the occurrence of stall at the turbine rotor blades. The valve is controlled to ensure that |p(t)|H11349p s (H9024). Then |q s (t)|H11349q s (H9024). The excess flow rate q v (t) passes through the valve to (or from) the atmosphere. The inertia of the rotating parts is assumed large enough so that rotational speed H9024 may be considered approximately constant over the time-intervals simulating a given sea-state (about 15 minutes). This allows H9024 to be optimized for each represen- tative record of the sea-state, in order to maximize the electrical energy production. The turbine rotational speed is allowed to vary between the synchronous speed of the generator and twice its value. Summing the product of the time-averaged electri- cal power output with the occurrence frequency for all data records gives the overall annual average electrical power output. 1466 A. Brito-Melo et al. / Ocean Engineering 29 (2002) 1463–1477 2.2. Hydrodynamic model The hydrodynamic model is based on the pressure model presented in Sarmento and Falca?o (1985). According to the OWC performance description presented in Section 2.1, the mass balance across a control surface enclosing the pneumatic chamber is given by p(t) K H11001 q v (t) H11005 q(t)H11002 V 0 gP a dp(t) dt (1) where q(t) is the volume flow rate displaced by the free-surface inside the chamber, V 0 denotes the volume of the air in the chamber under undisturbed conditions, P a is the atmospheric pressure and g is the ratio of specific heats. As stated in Section 2.1, q v (t) H11005 0if|p(t)| H11021 p s (H9024) (i.e. when the valve is not operating). According to the linear water wave theory, the volume flow rate displaced by the free-surface inside the chamber may be decomposed as q(t) H11005 q d (t) H11001 q r (t), where q d (t) is the diffraction flow rate, due to incident wave action assuming the internal and the exter- nal free-surfaces at constant atmospheric pressure, and q r (t) is the radiation flow rate due only to the pressure oscillation p(t) in otherwise calm waters. Under the assump- tions of the linearized wave theory, we may apply the convolution theorem to obtain the solution of a linear problem in terms of an impulse response (Pipes and Harvill, 1970), as follows: q r (t) H11005 H20885 H11002H11009 t h r (tH11002t)pH11032(t)dt (2) where pH11032(t) is the time-derivative of the pressure inside the chamber and t represents a time-lag. The upper limit of the integral in Eq. (2) represents the present instant t because the process is causal (Cummins, 1962). The impulse response function h r (t) can be obtained from the hydrodynamic coefficients of the OWC computed with a numerical model, such as the WAMIT (Lee et al., 1996) or the AQUADYN- OWC (Brito-Melo et al., 1999), or by tank testing. Here we use an estimate of the impulse response function obtained in free-oscillation transient experiments from 1:35 scale testing of the Azores OWC wave power plant (see Sarmento and Brito- Melo, 1996, for details). Time series for the diffraction flow, q d (t), have also been obtained in energy extrac- tion experiments with the scaled model subject to a set of 44 sea-states representative of the Azores power plant site. In these experiments a device producing an equivalent air pressure drop simulated the turbine. The flow rate q t (t) could be obtained as a function of p(t) from the device calibration curve. The diffraction flow time-series for each of the 44 sea-states was estimated by solving Eq. (1) (with q v (t) H11005 0) using the pressure records from the energy extraction experiments, and the experimental estimate of h r (t) previously obtained in the transient experiments. 1467A. Brito-Melo et al. / Ocean Engineering 29 (2002) 1463–1477 2.3. Power take-off equipment The power take-off sub-model is based on results extrapolated from small-scale turbine tests (Gato et al., 1996; Webster and Gato, 1999a,b) and on empirical data for the turbine and generator losses (Brito-Melo et al., 1996). The average power at the turbine shaft for a period T is given by W s H11005 H9024 T H20885 0 T [L(H9024,q t (t))H11002L m (H9024)] dt (3) where L is the aerodynamically produced turbine-torque and L m the torque due to mechanical losses (especially bearing losses). For stall-free conditions, L is approxi- mated by a second-order polynomial. In order to provide the necessary performance data to study the matching of the power take-off equipment and the pneumatic chamber, the data from small-scale turbine tests are modified using a simple mean- line turbine flow analysis method to take into account the rotor solidity S and the hub-to-tip ratio. Ignoring the postponement of stall when the Reynolds number is increased, scale effects are taken into account by correcting the torque curve of the turbine model. This is done multiplying (dividing) the positive (negative) values of L by f H11005 0.8/0.706. This corrects the torque curve of the unswept NACA 0015 bladed rotor with guide-vanes to match a peak efficiency of h max H11005 0.80. For the preliminary design of the turbine a maximum blade tip speed of 160 ms H110021 is assumed. The average electrical power output is obtained by subtracting the generator losses from the average power at the turbine shaft. The model for the generator losses includes the Joule losses, the iron losses, the ventilation losses and the mechanical losses (Brito-Melo et al., 1996). 3. Results and discussion Experimental research on different types of rotor blades has been conducted recently to improve the aerodynamic performance of the Wells turbine (Raghunathan, 1995; Gato et al., 1996; Curran and Gato, 1997; Webster and Gato, 1999a,b). Among these types, we consider two turbine blade configurations, which may give a wider range of flow rates within which the turbine can operate with fairly good efficiency, in comparison with that of the more standard NACA 0015 unswept bladed turbine rotor: they are the backward-swept NACA 0015 blades (Webster and Gato, 1999a), Fig. 1, and the optimized HSIM-15-262123-1576 unswept blades (Gato and Hen- riques, 1996), Fig. 2. For comparison we take results for the NACA 0015 unswept blades (Gato et al., 1996). Figs. 3 and 4 show experimental results from unidirectional-flow small-scale test- ing at the IST rig (Webster and Gato, 1999a,b). Results presented in Figs. 3 and 4 refer to high-solidity Wells turbine rotors (rotor outer radius R H11005 0.295 m, constant chord c H11005 125 mm, rotor solidity S H11005 0.64, equipped with the blades referred to 1468 A. Brito-Melo et al. / Ocean Engineering 29 (2002) 1463–1477 Fig. 1. Rotor blade sweep angle. Fig. 2. The NACA 0015 and HSIM 15-262123-1576 sections. above, with and without guide vanes. The figures show, in dimensionless form, experimental results for the efficiency h H11005 LH9024/(q t p), pressure drop p ? H11005 p/(rH9024 2 R 2 ), and torque L ? H11005 L/(rH9024 2 R 5 ) as functions of the flow rate coefficient U* (r is the air density). Results in Fig. 3 for the turbines without guide vanes show that the NACA 0015 unswept rotor has h max H11005 0.583 at U ? H11005 0.114, and stalls at U ? H11005 0.21. The NACA 0015 30° backward-swept rotor has a lower h max H11005 0.583, with a lower flow rate for the onset of stall, U ? H11005 0.17, but without exhibiting the sharp decrease in the torque that occurs in the unswept rotor. Furthermore, under stall conditions, the torque of the swept rotor becomes negative at a much higher flow rate, U ? H11022 0.45, whereas for the unswept blades the efficiency becomes nega- tive for U ? H11022 0.3. The unswept HSIM bladed rotor shows a h max similar to that of the backward-swept rotor, but produces a soft progressive stall of the flow through the rotor blades, with notably higher efficiency for a wide range of flow rates after the onset of stall. Fig. 4 shows a corresponding plot for the same turbine rotors when equipped with a double row of guide vanes. The experimental results plotted in Fig. 4 show that the use of guide vanes increases h max for any of the above geometries, i.e. from 0.583 to 0.706, 0.551 to 0.613 and 0.553 to 0.669, for the NACA 0015 unswept and 1469A. Brito-Melo et al. / Ocean Engineering 29 (2002) 1463–1477 Fig. 3. Unswept and 30° backward-swept NACA 0015 and unswept HSIM bladed rotor turbines, without guide vanes: measured values of efficiency (a), pressure drop (b) and torque (c) against flow rate coef- ficient. 1470 A. Brito-Melo et al. / Ocean Engineering 29 (2002) 1463–1477 Fig. 4. Unswept and 30° backward-swept NACA 0015 and unswept HSIM bladed rotor turbines, with guide vanes: measured values of efficiency (a), pressure drop (b) and torque (c) against flow rate coef- ficient. 1471A. Brito-Melo et al. / Ocean Engineering 29 (2002) 1463–1477 backward-swept rotors and the HSIM unswept rotor, respectively. Furthermore, we find that the use of guide vanes narrows the range of flow rates within which the turbine works with positive torque. Table 1 summarizes the performance data for the six turbines, where U ? a and U ? b are the minimum and maximum flow rate coefficients respectively, at which the efficiency is nominally h H11005 0.5h max . Therefore, H9021H11005U ? a /U ? b and H9004H11005U ? a H11002U ? b give an indication of the operational range while (H9004p ? 0 /U ? ) h H11005 h max is the pressure–flow ratio in the approximately rectilinear region. In the above performance comparison, constant overall solidity was assumed for the different turbine configurations. Results in Table 1 show that the rotor blade geometry has a remarkable influence on the turbine performance. In particular, some rotor geometries give a considerable wider range of flow rates within which the turbine operates efficiently, in comparison with others that have higher peak efficiency within a narrower range of flow rates. Figs. 5–7 plot the average electrical power output as given by the numerical simul- ation for the set of the 44 representative records of the wave climate for the Azores Plant site, taking into account the frequency of occurrence of each sea-state. The results give the turbine characteristic K for several values of the rated power W 0 H11005 p s q s . Table 2 indicates the values of the flow coefficient U ? s at which the different types of turbine rotor were designed and the bypass pressure-relief valve is actuated. 3.1. NACA 0015 unswept bladed rotor with and without guide vanes Fig. 5 presents the results of the numerical simulation to study the effect of the use of guide vanes with the NACA 0015 unswept bladed rotor. Fig. 5 shows that the use of guide vanes provides a significant increase in the average electrical power output, both with and without the presence of the bypass pressure-relief valve. The curves plotted in Figs. 3 and 4 for the unswept NACA 0015 rotor, with and without guide vanes, respectively, show that the turbine with guide vanes has h max H110150.72 Table 1 Peak efficiency, useful flow rate range and damping ratio for several turbine models (overall solidity S=0.64) Turbine rotor With guide vanes Without guide vanes NACA 0015 NACA 0015 HSIM NACA 0015 NACA 0015 HSIM unswept swept-back unswept unswept swept-back unswept h max 0.706 0.613 0.669 0.583 0.551 0.553 (U ? ) h H11005h max 0.124 0.137 0.154 0.114 0.129 0.131 U ? a 0.050 0.062 0.057 0.051 0.058 0.059 U ? b 0.197 0.209 0.275 0.251 0.232 0.360 H9021 0.254 0.297 0.207 0.203 0.250 0.164 H9004 0.147 0.147 0.218 0.200 0.174 0.301 (H9004p ? 0 /U ? ) h H11005h max 2.19 1.87 2.38 2.54 2.04 2.79 1472 A. Brito-Melo et al. / Ocean Engineering 29 (2002) 1463–1477 Fig. 5. Unswept NACA 0015 bladed rotor turbine with and without guide vanes working (a) with and (b) without the bypass valve: average electrical power conversion as a function of the turbine characteristic K, for several values of the turbine-rated power. whereas for the turbine without guide vanes