A novel aerodynamic controllable roof for improving performance of INVELOX wind delivery system

Abstract

INVELOX is a patented structure designed to capture the incident wind from every direction, guide the collected wind to the ground level, and increase its velocity. A novel mechanism called aerodynamic controllable roof structure is introduced to improve the wind stream performance of INVELOX. The advantages of aerodynamic controllable roof structure are twofold: improving efficiency and preventing wind escaping the system. Using aerodynamic controllable roof structure, a performance curve is obtained that can improve the system efficiency. Wind speed and orientation and structural deflection are the parameters that direct the roof orientation. Fuzzy control is utilized in a model structure to control the roof movements using two servo motors. Computational fluid dynamics simulations are done to validate the performance and capability of the aerodynamic controllable roof structure mechanism. Results show that the fuzzy control is successful in controlling the roof orientation and the results from the simulation indicate that the efficiency of the system can be increased up to 12%.

Keywords

Aerodynamic controllable roof structure wind delivery system controllable roof computational fluid dynamics

Introduction

Wind energy has been increasingly utilized to produce electricity in the last decades (Hassanli et al., 2018). Energy crisis and change in the climate led the attention to the renewable energies, especially wind energy. There are parameters, such as blade shape, wind speed, and orientation of blades, which control the conversion of wind power to electrical energy. Most of the time, these parameters require increase in size of wind turbine which is a challenging limit (Hur, 2018; Kadum et al., 2018). Other disadvantages of the conventional wind turbines are their high maintenance cost and excessive downtime.

Many researchers and innovators have developed new technologies related to the wind energy (Al-Bahadly and Peterse, 2011; Chenniappan and Balakrishnan, 2017; Lin and Hong, 2010; Shen et al., 2011). Different types of single and multiple arrays of ducted turbines have been developed so far (Franković and Vrsalović, 2001; Georgalas et al., 1991; McLaren-Gow et al., 2013). Single-ducted turbines have been proved to be effective and promising for small wind applications. On the other hand, multiple-ducted turbines are complex systems that are challenging to deal with. A recently developed technology known as INVELOX (INcreased VELocity) (Allaei and Andreopoulos, 2014) is a wind power harassing and delivery system inspired by the ancient Persian wind-catcher system. In contrast to conventional wind turbines with generators mounted on the top of the structure, INVELOX idea directs the wind to the ground-based generators. The INVELOX wind power delivery captures the wind and funnels it to a passageway that leads to the Venturi and naturally increases the wind speed. This stream drives the generators installed at the ground level (Allaei et al., 2015).

INVELOX suffers from a major structural problem. When the wind speed or density is high, the structure shows instability. If the wind speed reaches a critical level, its force and momentum around the base may destroy the whole structure (Kimura, 2016). In this article, a novel modification has been done to the INVELOX structure. An aerodynamic controllable roof for INVELOX called ACRIS is introduced for the structure, which can improve not only the structural durability but also the efficiency of the electrical energy generation. ACRIS is inspired by airplane spoilers. According to Bernoulli’s equation, as spoilers on the airplane wings decrease the lift, ACRIS also decreases the lift force on the structure which improves the structure stability (studied in section “Performance of the ACRIS mechanism”). The ACRIS mechanism can also be used to change the air flow input area and prevent the wind escape from opposite side of the incident wind. This capability of ACRIS improves the efficiency of the INVELOX system. The idea of ball and plate mechanism (Knuplez et al., 2003) is used to control the system and instead of x and y locations of the ball, wind speed, wind orientation, and structure deflection are the inputs of the control system. The fuzzy concept is utilized to control the ACRIS orientation.

The rest of the article is organized as follows. First, a review on the INVELOX technology and the wind escape performance problem is presented. To improve performance and eliminate escaping air, the ACRIS mechanism is introduced. Then, the implemented control system is discussed. In order to validate the capability of the ACRIS mechanism, a CFD (computational fluid dynamics) simulation is performed. With the help of simulation, a comprehensive study around the performance of the INVELOX with ACRIS is done and performance of the INVELOX with and without ACRIS is compared to each other.

Review on INVELOX delivery system

INVELOX consists of five major parts as shown in Figure 1: (1) intake, (2) pipe carrying and accelerating wind, (3) boosting wind speed by a Venturi, (4) wind energy conversion system, and (5) a diffuser.

Figure 1.

Schematic of INVELOX delivery system.

INVELOX has a fundamental characteristic. It captures a large portion of free stream air flow at the intake section. A pipe carrying path directs the flow to the Venturi section. At the Venturi section, the flow speed is increased. This increases the mass flow rate and then drives a generator installed at the sub-ground level. At the end, flow exits from the diffuser section of the system.

It has been revealed by Anbarsooz et al. (2017) that a considerable portion of the air which has entered INVELOX escapes from the other side and the amount of this escaping air depends on the INVELOX geometry and the upcoming wind velocity. Figure 1 shows the air entering INVELOX and the escaping air. Decreasing the amount of escaping air would clearly increase the speed ratio (SR) (the ratio of the average air velocity in the Venturi to the wind velocity). Omnidirectional design of inlet section of INVELOX leads to this drawback. Although this helps INVELOX accept wind from all directions, it can be inferred that the inlet design must be modified.

Figure 2 shows all dimensional aspects of the design used in this study. The Venturi section is designed, longer in length for the use of multiple turbine generators. The longer the length of Venturi, the more decline in the wind speed would happen in the design (Abe et al.).

Figure 2.

The design used in this study.

Introduction of ACRIS mechanism

Governing equations

As it was mentioned earlier, ACRIS is a mechanism that improves both efficiency and structural behavior of the INVELOX system. As shown in Figure 3, ACRIS is a modification to the roof of the INVELOX.

Figure 3.

Schematic of ACRIS mechanism.

ACRIS is designed for two situations. When the system works at high-speed wind (the speed above the rated level), INVELOX system will not be in the danger of destruction due to the wind speed. In this situation, ACRIS mechanism increases the INVELOX efficiency. When the wind speed is low, ACRIS changes its orientation in a way that widens the area of intake section. This helps the system capture more air flow, which leads to higher wind speeds at the Venturi section. Therefore, the system can have a higher efficiency in comparison with the system without ACRIS.

When the system is working under high-speed winds, it becomes unstable and even could collapse due to the wind force. In this situation, ACRIS can be orientated in a way that it reduces the lift force on the structure. When the wind speed or density (column force) is high, the efficiency of the system is in the rated level and wind speed in the Venturi section is fixed.

The governing equations of the control volume enclosing the fluid inside the funnel can be written as follows (Allaei et al., 2015)

\oint_{A} ρ V \cdot d A = 0

(1)

\oint_{A} u_{z} ρ V \cdot d A = T - \oint_{A} Pd A \cdot e_{z}

(2)

\oint_{A} r u_{θ} ρ V \cdot d A = Q_{T}

(3)

\oint_{A} [\frac{p}{ρ} + \frac{1}{2} | | V^{2} | |] ρ V \cdot d A = P

(4)

where $V = (u_{r} \cdot u_{θ} \cdot u_{z})$ is composed of the air velocity vector components in the axial, radial, and azimuthal directions, respectively. It must be noted that $r$ is the radius, $ρ$ is the density of air, $A$ is the outward-pointing area vector of the control surface, $e_{z}$ is the unit vector in z-direction, $p$ is the pressure, $T$ is the axial force acting on the power generating rotor, $Q_{T}$ is the torque, and $P$ is the power extracted from the rotor.

In this study, a novel roof mechanism is introduced which increases the air flow received by the INVELOX system. From equation (1), it is clear that by increasing the inflow area, the mass flow rate increases. As a result, the air flow speed at the Venturi section will be ameliorated.

Solution methodology

In order to solve the equations of motion and model the turbulent flow, the renormalization group (RNG) $k - ε$ model is utilized. The governing equations of incompressible flow are presented by Reynolds-averaged Navier–Stokes (RANS) equations (Anbarsooz et al., 2019) and RNG $k - ε$ equations as follows (Jeong et al., 2002)

ρ \overset{\cdot}{V} + ρ V \cdot \nabla V = - \nabla p + 2 (μ + μ_{T}) \nabla \cdot s

(5)

\nabla \cdot V = 0

(6)

μ_{T} = ρ C_{μ} \frac{k^{2}}{ε}

(7)

ρ \overset{\cdot}{k} + ρ u \cdot \nabla k = (μ + \frac{μ_{T}}{σ_{k}}) (\nabla \cdot \nabla) k + ρ G - ρ ε

(8)

ρ \overset{\cdot}{ε} + ρ u \cdot \nabla ε = (μ + \frac{μ_{T}}{σ_{k}}) (\nabla \cdot \nabla) ε + C_{1} \frac{ε}{k} ρ G - C_{2} ρ \frac{ε^{2}}{k}

(9)

where $μ$ and $μ_{T}$ are molecular viscosity and turbulence viscosity, respectively. $s$ is the strain rate, and $ρ G$ is kinetic energy production (Yakhot et al., 1992).

Control strategy

As mentioned earlier, there are major parameters that affect the orientation of ACRIS. These parameters must be chosen as the input of the control system. Here, a fuzzy controller (Lee, 1990) is used to regulate the orientation of the roof. With the help of fuzzy concept, dealing with multiple control inputs can be more facile.

One of the inputs is the deflection of the mass center of the structure columns from their equilibrium positions. Six states (from very low force to very high force) with the accuracy of 0–10 were considered as the deflection input.

Another input is the wind orientation. Most of the time high-speed winds are in the same direction. However, when dealing with low-speed winds, an omnidirectional analysis must be done. Therefore, combination of five states was considered in the eight main directions.

Wind speed is the last and most important input of the designed fuzzy controller. The wind speed range is from 0 to 150 km/h (0–40 m/s). Four states were considered as the wind speed input. These three inputs are shown in Figure 4.

Figure 4.

Fuzzy inputs: (a) structure deflection (column force), (b) wind orientation, and (c) wind speed.

In Figure 5, the fuzzy output is depicted. Two servo motors are used to change the orientation of the roof. Servo motors outputs vary from –35° to 35° and are divided into six states.

Figure 5.

Fuzzy output.

In the normal situation, when the outdoor wind speed is increasing, air flow speed in the Venturi section must be increased. However, results from simulation and real data show that at some points, when the wind speed increases, the air flow in the Venturi section decreases. This is one of the disadvantages of INVELOX design. This phenomenon is called singularity. Singularity can occur in the INVELOX system due to fluid mechanics, such as turbulence flow and eddy current flow at the downstream of the flow.

The optimum air flow speed at the Venturi section of the generator is about 25 m/s. One of the advantages of the ACRIS mechanism is to reach and maintain the optimum air flow speed in the Venturi section when the wind speed is in the low level range. When the outdoor wind speed is high, the main responsibility of the ACRIS mechanism is to maintain the structure stability. To do so, the orientation of the roof changes in a way that decreases the wind force on the structure. Moreover, when the outdoor wind speed is high, the air flow speed at the Venturi section will cross the optimal speed and at some point it can be even harmful for the generators. The ACRIS orientation for high-speed winds not only decreases the wind force on the structure but also decreases the air flow speed in the Venturi section, so that it can be near the optimal speed. Figure 6 shows the singularities in the INVELOX system for different roof orientations. Also in Figure 6, the optimal curve for the roof orientation is depicted. In the optimal curve, the air flow speed is in the propinquity of 25 m/s. The optimal curve is considered to be the favorable orientation of the roof and the output of the control system must match this curve.

Figure 6.

Singularity and optimal curve.

ACRIS control mechanism prototype

In order to investigate the performance of the ACRIS, a prototype was built and the fuzzy controller was implemented on the setup to orient the roof according to the optimal curve. The model consists of two servo motors to orientate the roof (Figure 7). In Figure 8, an overview of control strategy for the ACRIS mechanism is demonstrated. An Arduino was used as a data logger to acquire the data (wind direction, wind velocity, and column force) from the sensors. Arduino and LabVIEW functioned as conditioning and interfacing units, respectively. The sensor data was sent to the fuzzy controller to adjust the actuators. Output signals were sent to LabVIEW for some graphical representations as well. As mentioned, fuzzy logic was used to control the rotations of the two servo motors.

Figure 7.

Prototype of the ACRIS mechanism.

Figure 8.

Control strategy block diagram.

Results and discussion

In this section, two aspects of the study are investigated, namely: the performance of the fuzzy control system and capability of the ACRIS mounted on the INVELOX to improve its efficiency. The importance and application of the fuzzy controller for the ACRIS is also validated.

Performance of fuzzy controller

As mentioned in the previous section, an optimal curve for the orientation of the roof was introduced. The objective of using fuzzy control for the ACRIS mechanism is to help the roof orientation reach the optimal curve. In order to evaluate the performance of the fuzzy controller, the control system was simulated in MATLAB and the model was simulated in ANSYS. These two simulations were done simultaneously in order to evaluate the performance of the control system and investigate the output results. To do so, synthetic inputs for each three fuzzy inputs were considered and the output results of the simulation were also investigated.

Figure 9 shows the output results of the control system. As shown, the orientation of the roof has followed the optimal curve designed for the roof. Therefore, it validates the performance of the implemented fuzzy controller. To make the simulation understandable, the wind direction was considered to be fixed. Therefore, servo 2 did not require rotation; however, due to the fuzzy rules, servo 2 had some rotation near zero. Servo 1 almost followed the optimal curve (red dash line). It should be noted that, when the roof orientates according to the optimal curve, it will preclude the occurrence of singularity in the system.

Figure 9.

Fuzzy outputs from the simulation (red dash line is optimal curve).

Mesh and grid validation

In this section, the mesh independence study of five different grid sizes is studied. Figure 10 shows that a resolution of 2.2 million mesh is the finest grid size for the mesh to be qualified in this simulation. The mesh selection criteria are based on applying free stream wind to the structure and monitoring the flow speed at the Venturi section. First layer size is set at 5 mm with a 1.2 growth rate and the maximum cell size of 2 m. A suitable value for viscous sublayer is a y-plus value less than 20 in all locations of the INVELOX wall. In Figure 11, the mesh for the simulation environment and around the structure is shown. In the vicinity of the structure, a denser mesh was used to correctly capture the effect of the fluid on the structure. A boundary layer condition is set in three-dimensional (3D) domain, where the left surface is the velocity inlet and the opposite side is set for pressure outlet, all other lateral faces are set as slip walls. The inlet velocity ranges from 1 to 40 m/s and an algorithm called SIMPLE is picked with the turbulent model of $k - ε$ , for solving the partial differential equation in the simulation.

Figure 10.

Mesh validation in million elements.

Figure 11.

Mesh specification in the simulation.

Performance of the ACRIS mechanism

Here, the effectiveness of the ACRIS mechanism in improving the performance of INVELOX and its efficiency is investigated. In order to compare the performance of the INVELOX with and without the ACRIS mechanism, CFD simulations were performed.

First, two systems were simulated at a wind speed of 1 m/s. The results of this simulation are shown in Figure 12. The INVELOX can reach the air flow speed at the Venturi section to about 1.8 m/s; however, the air flow speed at the Venturi section of the system with ACRIS was about 2.2 m/s. It shows that ACRIS mechanism has enhanced the performance of the INVELOX system up to 18% in increasing the wind speed inside the Venturi and prevented the wind escaping from the opposite side of the incident wind.

Figure 12.

Comparison of two mechanisms at wind speed of 1 m/s.

Most of the time, the wind speed is between 2 and 4 m/s. Figure 13 shows the performance of INVELOX and ACRIS at the wind speed of 2 m/s. As it can be seen, the air flow speed at the Venturi section for the ACRIS mechanism is about 4.46 m/s, while the air flow speed for the INVELOX system is about 3.8 m/s. Again, the ACRIS mechanism validated its capabilities. It indicated 20% improvement in the performance of the system and prevented wind escaping.

Figure 13.

Comparison of two mechanisms at wind speed of 2 m/s.

It was noted before that the optimal air flow speed for the generators inside the Venturi section is about 25 m/s. Therefore, it is vital to investigate which system can reach the optimal air flow speed at the Venturi section earlier. To do so, another simulation was done at a wind speed of 10 m/s and the results from the simulation were compared to each other (Figure 14). As shown, the air flow speed at the Venturi section of the system with ACRIS is near to 25 m/s; however, the air flow speed at Venturi section for the INVELOX system is about 22 m/s. This result shows that ACRIS can work at an optimal air flow speed in a lower wind speed compared to INVELOX system. It indicates 12% improvement in the performance of the system.

Figure 14.

Comparison of two mechanisms with wind speed of 10 m/s.

ACRIS turbine performance evaluation

After analyzing the Venturi wind speed in the simulation, it is necessary to determine the energy output of the system. Figure 15 shows wind indicator for a period of approximately 10 years of Kish Island in Iran, where the 2018–2019 period is adopted to apply the test for a regional estimation energy output. The reason for selecting this period is the diversity in wind speed compared to more steady wind flow in previous years in this location.

Figure 15.

Decennial wind report of Kish Island, Iran and the adopted 2018–2019 period.

By looking at the annual Venturi wind speeds of the ACRIS and INVELOX turbines, a noticeable increase can be observed in the ACRIS output (Figure 16). Therefore, it is plausible that for years with lower wind speeds, like those prior to 2017, the power output would even show a higher difference.

Figure 16.

Annual Venturi wind speed comparison in Kish Island and the adopted 2018–2019 period.

A wind power chart from the Venturi can be obtained using equation (10) and the wind speed profile of the region is shown in Figure 17. In equation (10), $C_{p}$ is the efficiency of the windmill, and in this case study, the ideal efficiency of Betz limit which is 59.3% is used. This shows a great improvement for ACRIS versus INVELOX

P = \frac{1}{2} C_{p} ρ A V^{3}

(10)

Figure 17.

Annual ideal power output (W) comparison in Kish Island, according to Betz limit.

In sum, the new ACRIS design can increase the output power of the INVELOX up to three times, and this shows that the roof adjustability for capturing wind stream is working effectively and it can even be improved with an optimized design of the roof mechanism itself (Dehghani-Sanij et al., 2015).

Structure resistance performance

As the law of aerodynamics implies, when the angle of attack increases in ACRIS roof mechanism through the wind stream, a potential amount of drag can be forced to the inlet section, which results in a large moment to the foundation and then a destructive uplift of the structure is possible (Chen and Letchfortd, 2004).

For this test, a 40 m/s wind stream was blown and then the pressure applied to the structure by the stream was captured. Figure 18 shows a larger area of pressure is applied to the INVELOX structure, while the ACRIS design with its omnidirectional angle adjustability can reduce this effect on the structure.

Figure 18.

Load analysis of ACRIS mechanism under 40 m/s wind stream.

For a better feasible study, the pressure distribution results in Figure 18 was implemented on an axial force test simulation in Figure 19. In this comparison, the result shows that in INVELOX structure an ample tensile force of 4.17e5 N, depicted in blue (darker color) column rods, has an effect through the area confronting the wind stream; this proves the uplift pressure on the INVELOX system. However, ACRIS does not permit the uplift force to influence the structure, but it transmits a portion of it to the opposite side of the wind flow and creates a down lift on the structure (Luo et al., 2009).

Figure 19.

Axial force of ACRIS mechanism under 40 m/s wind stream load.

Conclusion

In this article, a novel mechanism called ACRIS was introduced to improve the performance of INVELOX delivery system. ACRIS is a roof for the INVELOX that orientates omni-directionally. This roof can improve the efficiency and structural strength of the system. One of the main challenges of the ACRIS mechanism is the control of the roof orientation. For this purpose, a fuzzy controller was utilized. One disadvantage of INVELOX delivery system is the occurrence of singularity between 15 and 20 m/s. To avoid the singularity, with the help of ACRIS, an optimization curve was introduced. This curve showed the orientation of the roof in a way that not only stopped the occurrence of singularity but also let the system work at its optimal efficiency. A total of 240 rules in the fuzzy controller enabled the roof to successfully orientate according to the optimal curve. After the implementation of the control system, the capability of ACRIS was evaluated in ANSYS and the results were compared for the ACRIS and fixed roof conditions (INVELOX). Simulation results validated the advantages of ACRIS mechanism over the INVELOX wind delivery system over 20% flow efficiency in the Venturi. ACRIS let the system work at higher air flow speeds in the Venturi section, made it work at the propinquity of optimal curve, and prevented wind escaping problem. On the other hand, while the ACRIS showed an uplift prevention, the design needs to be more efficient in its aerodynamic effect through the structure. The suggestion would be a better bio-inspired mechanism solution for aerodynamic 360° angular control mechanism system.

Footnotes

Acknowledgements

Special appreciation is credited to Mr Mohammad Hassan Ranjbar for assistance with the CFD simulations of the project and his comments.

Declaration of conflicting interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author(s) received no financial support for the research, authorship, and/or publication of this article.

ORCID iD

Farzad A Shirazi

References

Abe

Nishida

Sakurai

, et al. (2005) Experimental and numerical investigations of flow fields behind a small wind turbine with a flanged diffuser. Journal of Wind Engineering and Industrial Aerodynamics 93: 951–970.

Al-Bahadly

Peterse

AFT

(2011) A ducted horizontal wind turbine for efficient generation. In: Al-Bahadly

(ed.) Wind Turbines. Rijeka: InTech Open, pp. 87–106.

Allaei

Andreopoulos

(2014) INVELOX: Description of a new concept in wind power and its performance evaluation. Energy 69: 336–344.

Allaei

Tarnowski

Andreopoulos

(2015) INVELOX with multiple wind turbine generator systems. Energy 93: 1030–1040.

Anbarsooz

Amiri

Rashidi

(2019) A novel curtain design to enhance the aerodynamic performance of INVELOX: A steady-RANS numerical simulation. Energy 168: 207–221.

Anbarsooz

Hesam

Moetakef-Imani

(2017) Numerical study on the geometrical parameters affecting the aerodynamic performance of INVELOX. IET Renewable Power Generation 11: 791–798.

Chen

Letchfortd

(2004) A deterministic–stochastic hybrid model of downbursts and its impact on a cantilevered structure. Engineering Structures 26: 619–629.

Chenniappan

Balakrishnan

(2017) Container wind energy for wind turbines. In: 2017 international conference on computer communication and informatics (ICCCI), Coimbatore, India, 5–7 January, pp. 1–5. New York: IEEE.

Dehghani-Sanij

Soltani

Raahemifar

(2015) A new design of wind tower for passive ventilation in buildings to reduce energy consumption in windy regions. Renewable and Sustainable Energy Reviews 42: 182–195.

10.

Franković

Vrsalović

(2001) New high profitable wind turbines. Renewable Energy 24: 491–499.

11.

Georgalas

Koras

Raptis

(1991) Parametrization of the power enhancement calculated for ducted rotors with large tip clearance. Wind Energy 15: 128–136.

12.

Hassanli

Fletcher

, et al. (2018) Potential application of double skin façade incorporating aerodynamic modifications for wind energy harvesting. Journal of Wind Energy and Industrial Aerodynamics 174: 269–280.

13.

Hur

(2018) Modelling and control of a wind turbine and farm. Energy 156: 360–370.

14.

Jeong

Koh

Lee

(2002) Finite element formulation for the analysis of turbulent wind flow passing bluff structures using the RNG k–ε model. Journal of Wind Energy and Industrial Aerodynamics 90: 151–169.

15.

Kadum

Friedman

Camp

, et al. (2018) Development and scaling of a vertical axis wind turbine wake. Journal of Wind Energy and Industrial Aerodynamics 174: 303–311.

16.

Kimura

(2016) Wind loads. In: Pipinato

(ed.) Innovative Bridge Design Handbook. Oxford: Butterworth Heinemann, pp. 37–48.

17.

Knuplez

Chowdhury

Svecko

(2003) Modeling and control design for the ball and plate system. In: IEEE international conference on industrial technology, Maribor, 10–12 December, pp. 1064–1067. New York: IEEE.

18.

Lee

(1990) Fuzzy logic in control systems: Fuzzy logic controller. IEEE Transactions on Systems, Man, and Cybernetics: Systems 20: 404–418.

19.

Lin

W-M

Hong

C-M

(2010) Intelligent approach to maximum power point tracking control strategy for variable-speed wind turbine generation system. Energy 35: 2440–2447.

20.

Luo

Wang

Jiang

(2009) Wind-induced responses of a high-rise transmission tower to thunderstorm downbursts. In: The seventh Asia-Pacific conference on wind engineering, 8–12 November, Taipei.

21.

McLaren-Gow

Jamieson

Graham

JMR

(2013) A comparison between ducted turbine theory and inviscid simulation. In: 2nd IET renewable power generation conference (RPG 2013), Beijing, China, 9–11 September, p. 3.48. London: Institution of Engineering and Technology.

22.

Shen

Zhu

(2011) Wind turbine aerodynamics and loads control in wind shear flow. Energy 36: 1424–1434.

23.

Yakhot

Orszag

Thangam

, et al. (1992) Development of turbulence models for shear flows by a double expansion technique. Physics Fluids A Fluid Dynamics 4: 1510–1520.