Smith College Capstone Project — Vertical Axis Wind Turbine Investigation, Design, and CFD Analysis (Pt. 2)

Introduction

Our capstone team, VAWT Ventures, collaborated with Honda Research Institute (HRI)/99P Labs on the exploration, preliminary design, testing, and performance evaluation of a vertical axis wind turbine (VAWT) for urban rooftop applications.

VAWTs are an emerging renewable energy technology that could be used to take advantage of wind resources in urban areas. Unlike traditional horizontal axis wind turbines (HAWTs), VAWTs have an axis of rotation perpendicular to the wind and can operate in multidirectional and turbulent wind conditions. These characteristics make them promising for urban environments, but they remain underdeveloped in efficiency and power generation compared to HAWTs.

This project represents Phase 1 of a multiphase effort to design and evaluate a rooftop-compatible VAWT. During the first semester, we focused on background research, concept generation, concept elimination, and concept selection as explained in our first blog post. We explored over 200 concepts from academic literature, patents, commercial technologies, and team ideation before narrowing the project down to four selected concepts: Classical Savonius, EN0005, Helical Savonius within a Helical Darrieus, and H-rotor Darrieus with Savonius Compartments. Since our first blog post, we realized that this project is much larger and more complex than we initially thought and have re-framed the scope to be more focused on exploring this emerging technology, conducting initial testing and analysis, and using what we learn to make recommendations for future work. We initially thought of this project as a two phase effort, but we now believe it would need additional phases depending on the direction a future team decides to go. During this semester, we focused on optimization, physical prototyping, wind tunnel testing, and performance evaluation using computational fluid dynamics (CFD), wake comparison, and scaling studies. Our final deliverables include a selected VAWT design, results from initial testing and analysis, and recommendations for future work.

Fig. 1. Classical Savonius (top left) [1], EN0005 (top right) [2], Helical Savonius within a Helical Darrieus (bottom left) [3], and H-rotor Darrieus with Savonius Compartments (bottom right) [4].

From Concepts to Prototypes

Following concept selection, we translated each of the four chosen VAWT concepts into CAD models using Fusion and fabricated small-scale 3D printed prototypes. We selected prototype dimensions of 5 inches tall by 4.5 inches wide so that each design would maintain a consistent swept area and could be fairly compared during testing.

Fig. 2. CAD and 3D printed models of four selected concepts.

We conducted wind tunnel testing to evaluate each prototype’s ability to initiate rotation under low wind speeds. To minimize frictional losses, we designed a common central attachment system that connected to two ball bearings positioned at the top and bottom of the turbine. This allowed the prototypes to rotate freely on a metal rod within the wind tunnel.

Fig. 3. Wind tunnel testing setup.

We increased wind speed incrementally, and at each increment, we categorized turbine behavior into three conditions: no spin, some spin, or constant spin. The wind speed at which the turbine first achieved continuous rotation was recorded as the cut-in speed, and each prototype was tested multiple times to ensure consistency.

During initial testing, the Darrieus with Savonius Compartments prototype was unable to cut-in and lacked sufficient support in existing literature to justify continued optimization within the project timeline. As a result, we removed it from further development. Testing then focused on the three remaining concepts: the Classical Savonius, EN0005, and the Helical Hybrid.

Across multiple rounds of testing, we evaluated several prototype variations that incorporated aerodynamic augmentations identified through literature review. These modifications included changes to blade geometry, overlap ratios, and structural features intended to improve startup behavior and rotational stability.

Fig. 4. Initial 3D models and cut-in speed comparison results for three remaining concepts.

Fig. 5. Helical Hybrid optimization parameters and cut-in speed comparison.

After each testing round, we selected the best-performing prototypes for additional refinement, updated the CAD models, and reprinted them for further testing. Because of project time constraints, the selected prototype represents the most promising configuration tested rather than a fully optimized design.

To select a final design, we compared the remaining concepts using an evaluation matrix that incorporated both experimental results and practical design considerations, including cut-in speed, angular velocity and moment of inertia, which both relate to expected power generation, manufacturability, and potential for future improvement. Based on this evaluation, the Helical Hybrid achieved the highest overall score.

Fig. 6. Final CAD model of the Helical Hybrid VAWT.

The selected geometry combines asymmetrical helical Darrieus blades surrounding a widened internal helical Savonius rotor with an optimized overlap ratio.

We also identified McConnell Hall at Smith College as a potential future testing site for a full-scale VAWT. After evaluating accessibility, structural integrity, rooftop obstructions, and available space, we selected McConnell Hall because of its height, structural accessibility, and existing rooftop solar infrastructure. Based on the available rooftop dimensions, we proposed a full-scale turbine approximately 7 feet in diameter and 7.8 feet tall.

Performance Evaluation

While wind tunnel testing allowed us to compare startup behavior and rotational performance between concepts, it did not provide enough information to fully evaluate power generation or determine how the turbine would behave at full scale on a rooftop. To address these gaps, we integrated several additional methods into our performance evaluation process, including wind data analysis, CFD, wake comparison, scaling studies, and an economic analysis.

Wind Data

To better understand rooftop operating conditions, we analyzed year-round wind data collected from the roof of McConnell Hall. The data came from an omnidirectional anemometer maintained by Professor Nathanael Fortune in the Smith College Physics Department.

Fig. 7. Wind data from McConnell rooftop.

The data showed that McConnell’s rooftop airflow is characterized by low wind speeds with significant variability. We found that the average daily wind speed was approximately 0.67 m/s, while the average daily high wind speed was approximately 4.02 m/s.

The wind data also informed our computational fluid dynamics (CFD) studies. Although the average wind speed was below 1 m/s, we selected an inlet velocity of 2 m/s for CFD because it represented a wind speed that could reasonably occur with some frequency while still being high enough for practical turbine operation.

Computational Fluid Dynamics (CFD)

To evaluate aerodynamic performance and estimate power generation, we used CFD simulations in SimScale. CFD allowed us to visualize airflow around the turbine and estimate the coefficient of power (Cₚ), which measures how effectively a turbine extracts energy from the wind.

A large portion of our CFD work focused on establishing and validating an appropriate simulation setup. We validated our workflow first through airfoil studies and then through a classical Savonius turbine study, comparing our results to published literature. Once we established confidence in the workflow, we applied the same approach to our selected Helical Hybrid design.

Fig. 8. Helical Hybrid velocity field.

For the Helical Hybrid, we estimated a Cₚ of approximately 0.19 at a wind speed of 2 m/s and a tip speed ratio (TSR) of 1. This value aligned reasonably well with values reported in similar studies on hybrid VAWTs, but it also demonstrated that the turbine was still extracting a relatively small amount of power from the wind.

One of the major focuses of our CFD work was TSR analysis. We tested a range of TSR values to determine where turbine performance peaked. However, our simulations showed the Cₚ continuously increasing at unrealistically high TSR values, suggesting that the CFD model was not accurately capturing parasitic losses such as skin friction and pressure drag. Because of this, we were not able to definitively determine the optimal operating TSR for our design.

Despite these limitations, CFD gave us valuable insight into flow behavior around the turbine and helped establish a technical foundation for future computational work.

Wake Comparison

To better understand how accurately our CFD represented physical behavior, we conducted a wake comparison between CFD and wind tunnel testing. We measured airflow velocities behind the turbine using a pitot probe and compared those values to velocities predicted by CFD at the same locations.

Fig. 9. Wake Profile Analysis of small scale helical hybrid.

The overall shape of the wake was reasonably consistent between CFD and the wind tunnel results, particularly in the low velocity region directly behind the turbine. However, the CFD consistently predicted higher velocities than those measured experimentally. We believe these discrepancies may have been caused by differences between the CFD computational domain and the physical wind tunnel geometry, as well as limitations in the amount of experimental data we were able to collect within the project timeline.

Scaling

Another major challenge of this project was understanding how small-scale wind tunnel results relate to full-scale turbine performance. To investigate scaling effects, we printed multiple sizes of the selected turbine and conducted cut-in speed and angular velocity testing.

Our cut-in speed testing suggested that cut-in behavior may scale reasonably well using Reynolds number matching. However, TSR values did not remain consistent across scales, even when Reynolds number was matched. This indicated that small-scale testing alone cannot reliably predict full-scale power generation or aerodynamic performance.

Economic Feasibility

To evaluate whether the selected design could be commercially viable, we conducted an initial economic analysis. We estimated the total cost of a full-scale prototype to be approximately $12,000, with much of the cost associated with large-scale 3D printing.

Using our CFD results, rooftop wind data, and conservative assumptions for TSR, we estimated that the turbine would generate approximately 10 kWh of electricity per year. Based on this predicted power output, the breakeven period would exceed 5,000 years, far beyond the expected lifespan of the turbine.

While there may still be non-monetary motivations for implementing VAWTs, such as energy independence or sustainability initiatives, our analysis suggests that our design in its current phase of development is not yet economically or environmentally viable for low wind speed rooftop applications.

Recommendations and Next Steps

Based on the conclusions from our technical work, we identified three possible directions for future phases of the project.

1. Continue to Target Low, Multidirectional Wind Conditions

One possible direction is to continue focusing on rooftop applications in urban environments. We believe there is still value in exploring technologies that can take advantage of untapped urban wind resources, despite the significant challenges associated with low wind speeds and turbulence.

If a future team continues in this direction, we would recommend redesigning the wind tunnel setup to measure torque in addition to angular velocity so that coefficient of power can be evaluated experimentally. Future work should also continue optimizing the Helical Hybrid design, perform additional scaling studies, and eventually conduct rooftop testing to compare real-world performance against wind tunnel results.

2. Change Application to Areas with Higher Wind Speeds

A second possible direction is changing the intended application environment to areas with higher wind speeds and fewer size constraints. Our work showed that wind speed, swept area, and coefficient of power are the dominant factors influencing energy generation.

Instead of targeting low-rise urban rooftops, future work could investigate installations on taller buildings or other environments where wind conditions are more favorable. In these cases, a different turbine type, such as a Darrieus-focused design, may ultimately be more effective than the Helical Hybrid configuration we selected.

This direction would shift the focus away from startup behavior and toward maximizing aerodynamic efficiency and power generation.

3. Focus on Computational Modeling for a Digital Twin

The third possible direction is to focus primarily on computational modeling and the development of a future digital twin. While we established an initial CFD workflow, there is still significant work needed to improve model fidelity and validation.

A future team pursuing this direction could transition to a more advanced CFD platform such as ANSYS, improve boundary layer meshing, and focus on validating simulations against wind tunnel data under controlled conditions. This work would contribute to the long-term goal of creating a high-fidelity virtual replica capable of predicting turbine performance.

Rather than focusing directly on turbine optimization, this pathway would prioritize computational modeling and simulation methodology.

Conclusion

VAWT Ventures completed Phase 1 of a multiphase project investigating vertical axis wind turbines as an emerging distributed energy technology. Through concept generation, prototyping, wind tunnel testing, CFD, wake comparison, scaling studies, and economic analysis, we developed and evaluated a helical hybrid VAWT designed for low wind speed rooftop applications.

While our selected design is not currently viable for urban rooftop energy generation due to low predicted power output and economic limitations, this project established a strong technical foundation for future work. We identified key challenges related to startup behavior, efficiency, CFD validation, and scaling, while also developing workflows and recommendations that future teams can build upon.

VAWTs remain an emerging technology with significant unanswered questions and opportunities for innovation. Our work contributes to the early-stage exploration of how these turbines could someday support distributed renewable energy generation in urban environments.

Acknowledgements

We as a team would like to acknowledge our 99P Labs liaisons Duane Detwiler, Ryan Lingo, and Rajeev Chhajer for their consistent support and guidance, as well as the opportunity to explore this emerging technology and expand our area of knowledge. Their positive attitudes, encouraging advice, and confidence in our abilities have allowed us to grow as both engineers and as a team. Additionally, this project would not have been possible without our Design Clinic coach Professor Sussanah Howe, who has been fundamental to the success of this project. Her feedback has always been thorough, insightful, and uplifting, and her guidance has been invaluable as we’ve navigated the nonlinear process of this project. We would also like to acknowledge R Koh for sharing their technical expertise regarding wind turbines, and Katy Kaproth-Gerecht for regular feedback during Design Reviews and presentations. Across multiple branches of Honda, we would like to thank Zhenyu Wang, Tom Ramsay, and Phillip Aquino for their technical assistance. From Smith College, we’d like to thank Eric Jensen and Dale Renfrow for helping us set-up our wind tunnel testing, Professor Nat Fortune and Professor James Lowenthal for access to wind data from McConnell roof, and the facilities team including Beth Hooker, Charles Dougherty, Kevin Ulrick, and Corey Lynch. Finally, we’d like to thank our families, friends, and the Picker Engineering Program 2025–26 senior cohort.

References

[1] A. Dewan, A. Gautam, and R. Goyal, “Savonius wind turbines: A review of recent advances in design and performance enhancements,” Materials Today: Proceedings, vol. 47, pp. 2976–2983, 2021. [Online]. Available: https://doi.org/10.1016/j.matpr.2021.05.205

[2] Batista NC, Melício R, Mendes VMF, Calderón M, Ramiro A. On a self-start Darrieus wind turbine: Blade design and field tests. Renew Sustain Energy Rev 2015;52:508–22.

[3] O. Ciobanu, “Analysis and Optimization of a Savonius- Darrieus Hybrid Wind Turbine,” Academia.edu, 2021. https://www.academia.edu/103232008/Analysis_and_Optimization_of_a_Savonius_Darrieus_Hybrid_Wind_Turbine?utm_source=chatgpt.com (accessed Dec. 03, 2025).

[4] A. Ghosh, “Hybrid vertical axis wind turbine,” U.S. Patent 9,890,768 B2, Feb. 13, 2018.