Sure! Here’s a sanitized version of the YouTube transcript:

—

Hello everyone, Paul here from The Engineering Mindset. In this video, I decided to build a basic centrifugal pump and test various blade designs using computational fluid dynamics (CFD). You can download my PDF templates for this pump; I’ll leave links in the video description for how to get your copy.

This is the pump I built. It’s a centrifugal type pump made mostly from wood and plastic, driven by a DC motor with a pulse width modulation speed controller. I ran the pump at various speeds to see how it performed. The pump is used in an open-loop system, circulating water between a reservoir and the pump, then back into the same reservoir. This setup is similar to the condenser side of a water-cooled chiller, which is used for large-scale air conditioning in high-rise commercial buildings. The water is pushed through the heat exchanger of the chiller to pick up unwanted heat from the building, which is then sent to the cooling tower on the roof. The water is sprayed to release heat into the atmosphere and then collects in the cooling tower’s pan, recycling back to the chiller.

Using this type of system helps maintain consistent suction pressure in the pump. If the pump discharged into a separate tank, the water level at the inlet would change, affecting the pump’s performance. To combat this, we would need to top up the same quantity of water being removed, making it easier to loop the water back into the same tank.

A centrifugal pump is fairly simple and consists of a few main parts: the pump casing, impeller, shaft, inlet, outlet, bearings, and an electric motor. My design includes these components, and I chose materials that were readily available and easy to work with. For the pump casing, I used marine wood, which is strong and waterproof, often used in boat construction. I wanted to see inside the pump while it rotates, so I used a thick sheet of acrylic for the front cover with a rubber seal in between.

The impeller was also made from acrylic, as it is easy to work with and can be fused together with a solvent for a strong joint. The shaft was made from stainless steel threaded rod, as stainless steel is more resistant to rust compared to mild steel. The pipes were made from PVC for its low cost and low friction factor. To drive the pump, I used a 775 DC motor with a variable speed ramp. I’ll leave a link in the video description for these materials and parts if you want to check them out.

After selecting the materials, I designed the pump. Centrifugal pumps use a volute, which is an expanding channel around the impeller that converts water velocity into pressure while allowing a flow rate to develop. I based my volute design around some 70-millimeter disks I had for the impeller and sketched a rough volute shape in CAD. If you’d like a PDF copy of the pump plans, you can find a link in the video description.

For the impeller blades, we have three main options: backwards curved, straight, or forwards curved. To keep the design simple, I decided to use segments of 50-millimeter acrylic pipe to form the curves of the blades. I could fit around five blades onto the impeller. I used the inverse of the backward curved impeller design for the forwards curved design. The straight blade impeller design also has five blades made from a thin sheet of acrylic.

To assess the performance of each pump impeller design, I utilized the SimScale CAE platform, which kindly sponsored this video. SimScale provides instant access to online CFD as well as finite element analysis through a user-friendly cloud-based platform. You can try the software for free and edit public projects at SimScale.com or create private projects with enhanced features.

After designing the pump casing and different impeller designs in CAD, I imported them into SimScale for analysis. We can make assumptions and run different operating points simultaneously to see how the pump would perform across a wide range of conditions, such as changing rotational speed, outlet pressure, and flow rate. Once we set up and run our simulations in SimScale for the different impeller types, we can compare the results.

When comparing the results regarding pressure, the backward curved design shows a nice transition from the center out to the edges, where the pressure is greatest. This indicates efficient conversion of velocity into static pressure. The straight blade design doesn’t have such a smooth transition, showing pockets of low pressure at the center, which impacts performance. The forwards curved impeller has large areas of low pressure at the center and sudden changes towards the tips. Thus, the backward curved impeller should be the most efficient at transforming velocity into pressure.

However, the backward curved impeller design is not perfect and needs fine-tuning. There are regions around the tips of each blade that can be improved to reduce losses. Additionally, there are regions of concentrated pressure between the blades leading to recirculation within the pump. We could use a shroud to reduce this and improve performance.

To engineer this design further, we would want to run multiple simulations with different blade thicknesses, angles, diameters, and numbers of blades to find the optimal design. But for now, this is a simple project.

To build the pump, I took sheets of wood measuring 145 millimeters wide, 170 millimeters high, and 12 millimeters thick. I printed out my volute drawing, cut it to size, and glued it to the wood as a template. I screwed two sheets of wood together and used a scroll saw to cut the center out from the template. After removing the center, I glued the template for the backplate to another sheet of wood and used a hole saw to remove some internal segments, allowing me to insert the blade of the saw and cut out the center.

With the main parts of the pump casing made, I used strong wood glue to form a seal between each sheet and let it set. Once ready, I screwed all three sheets together and used a file and sandpaper to ensure a smooth internal surface. For the front cover, I cut out the paper template and glued it to a sheet of acrylic, which will be bolted to the pump casing. I drilled holes slightly larger than the diameter of the bolts I would be using.

I then used a 22-millimeter hole saw to create a hole in the material for the PVC inlet pipe. To ensure a tight fit, I filed down some wood until it fit inside the PVC pipe. I heated the pipe with a heat gun until it was malleable and pushed the front cover over it to form a nice seal.

For the rubber seal, I used a two-millimeter thick sheet of rubber, traced the outline of the volute onto it, and cut it out to ensure a seal around the edges. For the impeller, I took a 70-millimeter diameter acrylic disc, found the center, and drilled through it using a drill bit the same diameter as the threaded shaft. I took some 50-millimeter acrylic tube, wrapped and taped some white paper around it, and measured the height for the blades.

After cutting the segments free from the tube to form the blades, I applied solvent to the base of each blade and positioned them. For the bearing housing, I cut out the CAD template and glued it to a piece of wood, then attached it to two more sheets of wood and used a hole saw to cut a hole for the bearings. Once the glue was dry, I filed the excess glue and widened the hole just enough for the bearings to fit tightly.

I placed two bearings and a spacer onto the threaded shaft and secured them in position. I used stainless steel threaded rod and flange locking nuts to hold the impeller in place. With the blade temporarily installed, I ensured it rotated well with a small gap between the blade and the bearing housing wall. The volute casing and bearing housing were glued together to form a seal and held with extra-long screws. The wood was covered with a white primer and waterproof coating.

To assemble the pump, I placed the shaft and impeller into the casing and gave it a quick spin to test it. I then used a flange locking nut and a normal nut on the back to secure them. This prevents the impeller from moving back and forth and allows for later removal to change the impeller. To attach the front cover, I used self-tapping screws along with metal and rubber washers to reduce stress on the acrylic sheet.

For the outlet, I inserted the 22-millimeter pipe and added a piece of rubber for a tight joint, covering these in hot glue to hold them in place. I fitted a pressure gauge to the inlet and outlet of the pump for measurements. The pump is driven by a 775 DC motor controlled by our pulse width modulation speed controller, powered by a DC bench power supply.

To test the pump, I created a simple open-loop setup with a water tank and PVC pipe running through a bend and ball valve into the pump’s inlet. The pump is driven by the DC motor and speed controller, powered by the bench power supply. The outlet of the pump rises through bends and returns to the supply tank. I used a water cup to measure the flow rate.

The pump worked quite well, achieving around 16 liters per minute of maximum flow rate. However, the tools and methods used to test the pump were not accurate enough for comparison with my simulations. The gauges did not read any pressure, making performance assessment difficult, so it will have to be done through manual calculations and assumptions. There were some leaks from the pump, which could have been prevented with waterproof grease, but I didn’t have any at the time. The water cup is not a precision instrument, but it was all I had available.

A significant problem I faced was cavitation, as air was being sucked into the impeller through small gaps around the inlet pipe due to the low-pressure region created at the impeller’s eye. The returned water falling into the tank also caused small bubbles in the supply.

This is a working prototype, so issues like these are expected. Now that we know these problems, we can rectify them in a future model. Comparing the performance of the impellers, we found that no flow developed until the shaft reached around 1,000 RPM. The backward curved impeller was the most efficient, converting electricity into useful mechanical work, resulting in a higher flow rate compared to the other designs.

To assess the efficiency of the power, I accounted for the losses of the electrical motor from the manufacturer’s data, which shows a minimum efficiency of around 40% and a maximum of around 72%. Using these figures for a rough estimate, the backward curved impeller had a peak efficiency range of between 15.4% and 27.8%. The straight blades ranged from 13.3% to 23.9%, and the forwards curved blades ranged from 12.5% to 22.57%.

These are just rough estimates, as there are many inaccuracies in the data. One conclusion from this experiment is that designing the pump and impeller with 3D CAD models and testing them through the SimScale CFD package was far more accurate and time-efficient than manually building and testing multiple impeller designs. Setting up the simulations took only a few minutes, and obtaining accurate results took one to two hours. This is why more engineering companies are moving to this technology; it truly is the future of design engineering.

That’s it for this video! To continue learning, check out one of the videos on screen now, and I’ll catch you in the next lesson. Don’t forget to follow us on social media and visit The Engineering Mindset.

—

This version removes any informal language and maintains a professional tone while preserving the essential information.