I'm trying to design a mobile robot which has to climb over flat 60° inclined wet glass surfaces.

I was thinking to use a large vacuum system composed by a large fan connected to a brushless motor. By looking on google, I found this photo:

enter image description here

As it is possible to see, they use a brushless motor with a fan to generete some kind of vacuum to keep the robot on the wall. This system is very similar to the RC wall car toys which are able to vertically move on flat walls.

My idea is to create a vacuum chamber and to make it connect to the surface by using a soft rubber gasket (to avoid leakage), but I do not know how to design the vacuum generator.

  • How many RPMs do I need?
  • What shape should the fan have?

The robot has a weight of about 30-35Kg and its size is 1000x800 mm. The surface which can be used for the vacuum generator is the space between the tracks and it is about 800x600 mm.

I hope you can help me!

Thank you a lot!

EDIT 2 Information after @ericnutsch answer.

This is my ideal design.

I would like to use a cylider as case for the fan/impeller, use a rubber gasket for the cylider borders in order to avoid leakages (when possible). The cylinder/case will be suspended by springs in order to make it touch the surface by using the weight of the chassis. I do not know if the drawing is clear, but this is what I would like to acquire. The dimensions showed in the figures are realistic and the weight of the robot is about 30-35kg.

enter image description here enter image description here


on #Mark answer:

I've found this impeller which can be customized and can have a diameter of 550 mm. It comes with some specifications:

enter image description here

EDIT1: This is what I tried so far:

During last months, I did several tests with different solutions:

  1. Using rubber tracks with Linatex which offers a great grip over flat surfaces
  2. Multiple suction cups with a vacuum generator

enter image description here

The Linatex works very great up to 60° if the surface is not wet, if it is wet, they works fine up to 18°.

enter image description here

The 4 suction cups works fine only if the robot moves very very slowly (0.1 m/s) otherwise they lose the vacuum.

  • 2
    $\begingroup$ I understand that you have a certain use in mind, but your actual question is very far down the page. $\endgroup$ – hazzey Apr 4 '17 at 13:01
  • $\begingroup$ Sorry, I use a wrong format for my question, I try to edit it. $\endgroup$ – Marcus Barnet Apr 4 '17 at 13:51
  • 1
    $\begingroup$ It should be better, now. $\endgroup$ – Marcus Barnet Apr 4 '17 at 13:54

You do not need full vacuum to accomplish your goal. A small amount of vacuum pressure adds up really quick over a large surface area. Try to make the entire footprint of your robot under vacuum; your sealing perimeter increases linearly with diameter, but area increases with diameter squared. Basically this means the problem is easier to solve the larger you go. And a circle is the best geometry because it has the highest area to perimeter ratio. Its really the design problem of a hovercraft, just in reverse. You will have to plan on some air leakage for it to operate on non-perfect surfaces in the real world.

Some existing equipment that may meet your needs:

  • A hobby rc brushless ducted fan. They do not publish head pressure, but might supply -3.5kPa gauge (-0.5psig) or so at a very high flowrate. You may be able to find one rated for high pressure lower flow. The nice part about these is they are easily battery operated.

  • A vacuum blower. High flow rate and reasonably high pressure. This one has 60 inches of water max, which is approximately 14kPa. They are likely rating in positive pressure. Its vacuum abilities would be less, so lets say -7kPa gauge (-1psig).

  • A "motorized impeller" (the internals of a shop vac). The website does not rate the vacuum pressure but i would estimate it would be -40kPa (-6psig).

  • An electric diaphragm pump. These are typically low flowrate so you would be back to clean smooth surfaces, but the vacuum pressure on this one is -50kPa gauge. Might be more flow than you have on your current vacuum pump if you stack a few in parallel.

I think you mean 80mm in your image not 80cm. 8cm results in an area of 0.005m^3 (7.8in^2). So at perfect vacuum 101kPa (-14.7psig), you are getting 500N (115 pounds of force), which is great until you lose that perfect vacuum. The force drops quickly because the vacuum pump you are using can not maintain the flow rate required by the imperfect surface.

I would say a motorized impeller would be your best bet. I don't know your scale here, but lets say putting a shroud around your footprint puts you at a 0.4m (16 inch) diameter.

climbing robot footprint

The corresponding area would be 0.126m^2 (200in^2). At -40kPa (-6psig) you now have roughly 5000N (1,200 pounds of force). And since this fan can handle a high flow rate, even large leaks are not a problem. You will want to design the shroud such that it is strong enough to not collapse under this force and design it such that it does not apply too much load to your climbing tracks.

| improve this answer | |
  • $\begingroup$ Thank you for your very detailed answer! I really appreciated it! Please, take a look at these pictures I added in my first message which show how I would like to design the vacuum system. I would like to use a cylider as case for the fan/impeller, use a rubber gasket for the cylider borders in order to avoid leakages (when possible). The cylinder/case will be suspended by springs in order to make it touch the surface by using the weight of the chassis. I do not know if the drawing is clear, but this is what I would like to acquire. Do you think is it possible? $\endgroup$ – Marcus Barnet Apr 5 '17 at 14:28
  • 1
    $\begingroup$ I attempted to add an answer to complement yours, with a few more technical points on fluid dynamics. Ultimately since sealing against surface is an empirical design, this just needs a lot of testing to figure out the leakage and design from there - I just wanted to add something to help when the design reaches that next stage. This was very well done. $\endgroup$ – Mark Apr 5 '17 at 17:47
  • $\begingroup$ Thank you, Mark, I think the sealing system will be the last one to be designed since I'm trying to figure out which blade and case I should use. I will keep you updated when I'll reach this step. $\endgroup$ – Marcus Barnet Apr 5 '17 at 18:18
  • $\begingroup$ @Mark, I received few quotations from some manufacturers of centrifugal fans. They can supply fans with diameter of 550-600 mm, but they asked me that they need to know the air volume and the static pressure in order to give me the best quotation and the proper fan. What do you suggest? I would like to give them general data as starting point. Can you help me, please? $\endgroup$ – Marcus Barnet Apr 6 '17 at 8:43
  • $\begingroup$ I would target -40kPa max static vacuum and perhaps 0.05m^3/s max flow since you are counting on a pretty good seal. Don't go too low on the flow or the device will not be reliable. The physical size of the fan is not really important. Pressure and flow are your key design criteria here. $\endgroup$ – ericnutsch Apr 6 '17 at 16:13

An important point in the fluid dynamics design is about control points. A simple experiment - hold your hand over a hose on the vacuum cleaner, you'll notice a light vacuum pulling on it, but hold your hand right on the top of the hose, and it feels like it is pulling significantly harder. This is because fans operate on a load response curve, and when there is less air flow, it pulls with more pressure.

There are a few different fan designs. Forward, Radial and Backwards curved blades:

enter image description here

These lead to different kinds of performance curves:

enter image description here

As expected, with no flow, you have nearly the highest pressure, and with a very low pressure, you have almost no pressure. The problem is you need to operate in a portion of the curve with a negative slope. The area with the positive slope is called the surge region:

enter image description here

While forward curved fans have a lot more pressure available, they typically have very large surge areas because of the large area with a positive slope, but there is usually the small area with a negative slope right before the surge area. In this application, a forward curved blade design could work so long as it stays above the local maxima in the mid-flow range, However,a radial blade design would be more conservative, and usually more cost effective, but the unit will necessarily have to be bigger to obtain large pressures.

A comprehensive stability analysis also needs to look at the power requirements. For power design, we'd need to look at the power curve for the fan. Notice the power requirements for the motor increase drastically as the flow rate goes up. Since a user can always peel the robot off the wall while in operation, the design will need to account for the fact that a constant RPM fan will demand more torque as more air is moved while the system stabilizes to the lower pressure requirements. Potential solutions could include:

  • Electronics could detect this torque and turn the motor off
  • Size the motor to handle the maximum airflow scenario
  • Something new and creative that's mechanically governed to limit torque
| improve this answer | |
  • $\begingroup$ Thank you for your answer and for your data!! I tried to search on google where I can buy a centrifugal fan with forward curved blade, but I found nothing. I think I should find a fan with a diameter of 40-50 centimeters at least in order to maximize the pressure. I think that I can detect when the torque change (reading the current) and, in any case, size the motor for the maximum flow scenario. Can you suggest me where I can find these fans, please? If we find the datasheet, we can calculate the RPMs, is it true? $\endgroup$ – Marcus Barnet Apr 5 '17 at 17:52
  • 1
    $\begingroup$ Here's one fun trick - run the wheel in reverse rotation and a backwards curved blade becomes forwards curved. Your best bet is probably going to be 3D printing, to be honest. yeggi.com/q/centrifugal+fan As you play with parameters, you could run a few tests on the printed fan, then use fan laws engineeringtoolbox.com/fan-affinity-laws-d_196.html to resize the fan and RPMS. Note - Pressure is related to RPM^2, while Flow is related to RPMs. So more RPMs is more Pressure v. Flow, so aim for high RPMs. This means faster response on your current detection though. $\endgroup$ – Mark Apr 5 '17 at 18:13
  • $\begingroup$ Thank you for your tips! Since I know nothing about fluid dynamics, how is it possible to imagine a relationship between the robot's weight and the vacuum against the wet surface? The vacuum should be enough to keep the robot on the surface, but also to let the robot move in any direction without block it. I think this is not so easy to calculate, isn't it? $\endgroup$ – Marcus Barnet Apr 5 '17 at 18:24
  • $\begingroup$ I added EDIT3 on my first message since I've found a centrifugal fan with a diameter of 550 mm and it is for high pressures. What do you think about? I can find a way to connect it to a large brushless motor. $\endgroup$ – Marcus Barnet Apr 5 '17 at 18:30
  • $\begingroup$ I received a quotation for [this fan] (robo-dyne.com/pics/fan.pdf) RM630/2 which costs 250USD and has a pressure of 6500Pa and a flow of 5500m3/h. Do you think it can work well as starting point for my application? $\endgroup$ – Marcus Barnet Apr 6 '17 at 15:27

Well if you would like to create your own vacuum, I would not recommend designing the actual pump yourself. You would need to design the turbomachinery components like the impeller (fyi an impeller is not a simple fan, it is much more complicated in terms of it's geomtery and the material used to make it), casing and various seals from scratch and those things are NOT easy to analyse, design and test in terms of the fluid dynamics, stress analysis and design engineering that goes into them. I say go on a distributor like RS Components and find one that suits your needs. Of course the part that you would design is the way you would channel the air, create the suction chamber, so on and so forth. That will depend on your engineering/DIY knowledge and your specifications.

| improve this answer | |
  • $\begingroup$ The impeller is something like this one? uk.rs-online.com/web/p/impeller-blades/7404471 and its case something like this? img0129.psstatic.com/… $\endgroup$ – Marcus Barnet Apr 4 '17 at 16:09
  • $\begingroup$ My point is not many people who are working on a robotics project like yours would design and manufacture their own vacuum pump. I would recommend estimating your operating pressure and suction flow characteristics and then simply choose a pump from say uk.rs-online.com/web/c/pneumatics-hydraulics-power-transmission/… and go with that. You will still need to make fairly thorough analysis of the channel flow inside the piping and chambers of your robot which is hard enough. $\endgroup$ – thephysicsguy Apr 4 '17 at 16:20
  • $\begingroup$ You also will need to learn some fluid dynamics to tackle this problem, at least some primitive estimations of pressures and simple Couette flow, etc. If you do insist on designing your own pump with impeller and other components you simply need significant background in mathematics, science or engineering. Only then you would be at a level to look into literature, perform computational analysis and design then test a working pump. Saying that, if you are really confident, you could buy the components like you mentioned and put one together which like I said will require experience/maths. $\endgroup$ – thephysicsguy Apr 4 '17 at 16:24
  • $\begingroup$ By looking at the picture I attached in my previous message, it doesn't seem that they are using a commercial pump or am I wrong? I think they made a case with a gasket and put inside a fan activated by a brushless motor. I saw the pump you suggested me, but I already have a pump like that and it is compressed air driven. I used it to control the suction cups which I placed at the bottom of my robot. $\endgroup$ – Marcus Barnet Apr 4 '17 at 16:46
  • $\begingroup$ Yes you could use a compressor in junction but I admit that would be quite pricey and not even sure if it would fit in a small-sized robot. Well I would recommend then to first look into some literature if you can eg: Introduction to Turbomachinery (Japikse et al. 1952) or Fundamentals of Turbomachinery (Peng 2008, see chapter 5 for example for axial pumps) and many many others. Then after having a rough idea regarding your geometrical requirements (blade angle, etc.) you could pick an impeller from RS and confirm it will suit your needs from the performance curves in it's datasheet. $\endgroup$ – thephysicsguy Apr 4 '17 at 17:00

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.