We chose a Horizontal Axis Wind Turbine (HAWT) for its higher efficiency in harnessing energy from higher wind speeds. Specifically, we designed around Needham's average wind speed, 11 MPH. While small VAWTS have a higher RPM output, HAWTS have a higher power output due to the mechanical advantage of the blades being much larger. Additionally, HAWTs are a proven, widely studied technology, and the modeling software we used, QBlade, is generally more accurate for HAWTS than VAWTS. They handle turbulent flow more effectively, which is important since Needham has variable wind conditions, and also, their horizontal alignment minimizes turbulence around the blades, which ensures greater lift and energy capture. Lastly, HAWTs are what most of the team just wanted to design. They are what everyone thinks of when they think of turbines!
QBlade is a software tool that is specifically designed for the design and optimization of wind turbine blades. There are two specific modes, HAWT and VAWT, which each allow different tweaking and parameter optimization at the modeling stage.
The process of modeling in Qblade follows a couple steps:
Step 1: Select the airfoil, define the Reynolds number and lift-drag ratio.
We found airfoils that optimized for a lot of different parameters, but since we wanted a high output rpm from the blades spinning, we searched for airfoils that were designed to maintain consistent lift across varying wind speeds. We assumed a high Reynolds number of about 500,000, meaning that our airfoil would be in a turbulent flow state. This was a valid assumption because of the inconsistent wind speeds that Needham experiences. At any given moment, the wind speed can range from 5 to 12 m/s. Given this range of expected wind speeds, QBlade generates different angles of attack that the blades experience and then simulations of the lift and drag forces acting at each angle of attack for a given airfoil profile. This gave us a lift to drag ratio, and we cycled through a couple different airfoils that optimized for small angles of attack before landing on using NACA Airfoil 2412 because it had a very high lift to drag ratio for angles of attack between 0 and 10 degrees.
Step 2: Make the model!
After selecting the NACA Airfoil 2412, we used QBlade to create the 3D blade geometry by defining the chord length, blade twist, and number of blade sections. We divided the blade into multiple sections along its span to account for variations in aerodynamic performance from the root to the tip. From the two blade iterations we designed in Qblade, we had to specify a length and then define these sections accordingly. We divided them somewhat evenly, but QBlade suggests having a lot of fidelity near the larger end of the taper, because this is where STL geometry is most likely to fail. Using QBlade’s parameterization tools, we specified a linear taper for the chord length, starting with a larger chord near the root for structural support and tapering to a smaller chord at the tip to reduce drag and rotational inertia.To further optimize the blade, we applied a twist distribution along the span to ensure that each section of the blade maintained an optimal angle of attack relative to the incoming wind. You can see optimal angles of attack from blade geometry here, and Qblade suggests matching them with the angles of attack that the airfoil was designed for, so in our case, as long as the outputted angles of attack from a given blade geometry were between 0 and 10 degrees, the twist is valid. This step was essential for maximizing the lift-to-drag ratio across the entire blade.
Step 3: Simulations and Outputs
After creating the blade geometry, we ran several simulations in QBlade to evaluate the performance of the design. Using the blade element momentum (BEM) theory built into the software, we simulated the turbine's performance over Needhams wind speed range (5 m/s to 12 m/s) QBlade outputted key metrics such as the power coefficient (Cp), torque, and rotational speed (rpm) for each wind speed. These were the main outputs we were looking for and also became the subject of a lot of different iterations. To validate the design further, we ran a tip speed ratio analysis, which compares the rotational speed of the blade tips to the wind speed. It was important to have a TSR of about 6, because that is the defined optimal TSR for a 3-bladed NACA 2412 airfoil. We also fine-tuned the blade design by adjusting parameters like twist and taper. We only physically made 2 of the qblade designs but went through the entire process a couple of times to find a high amount of output power. This part was especially tricky because it was based on unknowns from other parts of our project, like the power requirement of the water pump.
Description:
Our first Qblade iteration was a desktop-scale blade that mounted to our test stand setup. This was our bench test of QBlade, and we wanted to see if the program's predictions were accurate. We tested the blade with a tachometer and some reflective tape and verified that the RPM was somewhat close to the expected one from the simulation. The mounting fixture was made with a circular airfoil shape instead of the NACA 2412.
Important Conclusions:
Since our actual RPM value was close to the expected value, we gained a little more faith in QBlade's predicted outputs. These blades performed much better than our own custom-designed blades, so it was an important validation step for making our final blades with QBlade. Another important conclusion was based on these blades being extremely thin. Intuitively, it made sense that thinner blades would be easier to spin, so for the next iteration, we worked on optimizing the values while also keeping the chord lengths such that blades would be thin.
Limitations:
It was very evident after testing the blades that there wasn't a lot of torque coming off of the shaft. We had no real way of validating this, other than the blade being incredibly easy to stop. This to us was an intuitive enough reason to iterate on the blade and try to make it better. We used a 4-blade setup, which, after looking into HAWT test setups, was a poor choice because it increases drag with only a small improvement in lift. The tradeoff between these two important parameters was significant enough to switch to a 3-blade setup.
Setup:
Length: 11 Inches
Max Chord: 2 Inches
Expected Output RPM: 140
Actual RPM: 120
Expected Output Power: 2.3E-04 W
Expected Output Torque: 2.7E-05 Nm
Angle of twist: 15 Deg
Description:
The final physical blade we made was after experimenting many different blade designs. Firstly, the backwards taper from the original designs was removed to create more surface area. This was optimized based on the lift/drag ratio and TSR, and that is why the longest chord is much straighter than the original. This blade is significantly larger than the original. Increasing the blade length allowed us to capture more wind energy due to the larger swept area, which is proportional to the square of the blade radius. This increase in blade length also provided higher torque, enabling more consistent RPMs to power the gear pump. The blade mount to the hub has been altered, inspired by a mortise and tenon joint. The joint has been toleranced for a pressfit, and we also included a through hole for a bolt that will secure the blade from slipping out of the hub. The blade is also made up of two parts, and the two blade haves slot together with circular pegs located at the middle of the cross section of the full blade.
Setup:
Length: 19 inches
Max Chors: 5.5 inches
Expected Output RPM: 50
Actual RPM: PROBABLY OVER 100
Expected Output Power. 1.7E-01 W
Expected Output Torque: 1.8 Nm
The hub supports a three-bladed turbine design and has a 1/4" shaft passthrough. We used a keyed connection to lock the hub's rotation about the shaft, and bolt holes were made to secure the blades to the hub. The hub design itself is quite simple, but we needed to verify if it could support the loads from the turbine, so we performed fea on one blade and the hub as a cantilevered beam. The blade geometry was too complex to even run a simulation, so we structured the simulation around a straight bar attached to the hub. This was a valid assumption because we only needed to find out the bending moment of the hub part, and this would have worked with either geometry. This point force loading assumes a worst-case scenario, and as the hub did not yield with ABS, our assumption that the part will not fail is rooted in a simple analysis. In addition, though the actual stresses from the FEA may not be reflective of the real stresses, the simulation also helped identify where stress areas were and what features on the blades to support. This meant increasing the fillet radii of the slot feature closest to the hub and making a small part of the middle blade geometry more smooth than the qblade design.
Before resorting to QBlade, we attempted to make our own blades using 2D CFD studies. Unfortunately, despite many hours of simulation, the first iteration spun the wrong way, and the second iteration barely spun at all.