I am searching for the next thrusters for my ROV, about 25kg when neutral, but all the ones I see have propellers optimized for one direction only.
Taking the T200 for example, there is a about 30% diference in thrust between FW or REV thrust availability.
In an ROV with 6DOF that needs to move fast in all directions this means that I can have a very large diference in the max speed I can do moving one direction or the other. Apart from that the stability algorithm will also suffer with that diference in available thrust.
Am I missing something here or a symmetrical propeller design would be way better for ROV usage?
I was trying to find any thruster in the market with those and fund nothing…
Hi @papaxiclas -
A symmetric thrust propeller would eliminate the airfoil in the blade, and be a flat blade - a shape that’s okay both ways isn’t great either way…
A propeller is a spinning wing. Each blade is shaped like a small airplane wing. It has a rounded front edge, a thin sharp back edge, and a slight curve. That shape is built to slice through water in one direction.
In short, the physics just don’t work out to make a symmetric direction thruster without losing a ton of efficiency! Most ROV designs compensate for this by mounting two horizontal thrusters optimized for forward travel, and two for reverse - look closely at the BlueROV2 for an example!
Out of interest, are you trying to make a completely uniformly omnidirectional vehicle? It’s usually the case that at least some directions are prioritised over others (in both the motor placements/orientations and the vehicle design), and ArduSub is not currently optimised for maintaining completely arbitrary attitudes.
This is technically correct, but feedback control algorithms are quite versatile in what they can handle, even if they need to work a bit harder to do so.
Hello!
To be honest I didn’t had the time to test it by my self neither for looking hard enough for any study/test or any comparison between the 2 forms but thinking aerodynamically It’s hard to be convinced that even a simple flat paddle will be worst than an aerodynamic profile been run backwards. I now for experience that there are airfoils that improve the aerodynamic over a flat wing pushing up or upside down like on aerobatic airplanes but never encountered an example of the airfoil running backwards as it happens on these thrusters. If for instance you have remember anything for someone studying this difference please tell me so that I can check it.
Yes, it should be uniform. It’s an ROV for use in cinema with the cinema cameras and etc that should be able to move in any direction needed to do the shoot. This way doesn’t mater if I can go 10knots in forward and 6 backwards because most of the times it will not be shooting in pursuit of the actor.
I know that the algorithms used are very versatile, to be honest I was very surprised with how much abuse ardusub is able to cope with, but thinking of the main 6 axles of movement and thruster performance.
If I have 10knots forward speed and 6knots backward speed there is only one movement in all 6 axels and 12 movements were I can reach the full 10knots. That will be the one were all motors are going in the forward direction. All the other movements will be caped by the 6knots motor that is going to be backwards and in order to keep everything stable and in the correct direction the algorithm will throttle down the ones going forward. Am I thinking correctly?!
My thinking is that if a thruster going reverse doing 6knots with a backward spinning airfoil is worse than a flat paddle thruster going weither way, even if it only does 6.2knots, it might be an improvement for my case.
Correction first: I was wrong earlier about symmetric thrusters, sorry about that. They do exist commercially. They use a symmetric airfoil section, like aerobatic planes, not a flat paddle. Copenhagen Subsea sells both versions of the same thruster: asymmetric for efficiency in one primary direction, symmetric for equal thrust both ways. The trade is even thrust in exchange for less thrust per watt.
Speed in each direction: Your reasoning holds if all the thrusters face the same way, but a vectored layout avoids most of it. On the BlueROV2, two horizontal thrusters face forward and two face backward. Every surge, sway and yaw command uses two thrusters pushing in their strong direction and two in their weak one, so both directions on each axis match. You can see it in ArduSub’s motor matrix and in our matching 9 kgf forward and lateral bollard thrust.
The speed difference is also smaller than the thrust difference. Speed scales with the square root of thrust, so the T200’s 0.78 reverse/forward ratio works out to about 0.88× the speed in the weak direction, not 0.6×.
Choosing which axis gets more thrust: The BlueROV2 frame has mounting holes for angling the horizontal thrusters toward more forward power or more lateral power. Each axis stays balanced either way. A 30° angle instead of 45° gives about 22% more thrust on the favored axis and 29% less on the other. For cinema, favoring lateral might suit tracking shots.
Vertical: This axis is asymmetric, because our vertical thrusters all face the same way. You have two ways to even it out:
With four verticals, face two each way.
In depth hold, set PILOT_SPEED_UP and PILOT_SPEED_DN to the same speed, one the weaker direction can reach. The stick will then respond the same up and down.
You could design a frame that way, but as you’ve been thinking that would bias the control towards one direction, at the expense of either the (uncapped) stability / more undesired axis coupling, or (capped) performance of the other directions. If you want balanced performance between the axes you can use symmetric pairings of the thrusters to counteract the asymmetry of the individual thrusters (as @tony-white mentioned).
With an 8-thruster spherical omnidirectional frame (where every thruster is engaged in equal magnitudes for each motion axes) that difference looks like this:
In that case there would be no single movement where all thrusters are running forwards (since doing so would not move the vehicle, because the thrusters would cancel each other out), and you instead have full speed along any axis, but the full speed is averaged between the asymmetric thrusts.
It’s worth noting that active compensation provides stability by using extra power, and frame designs that make heavy use of it effectively waste a significant portion of their energy usage just cancelling out unwanted thrust from the other side. Frames like the BlueROV2 are a less severe compromise, with the vertical motion kept independent from the horizontal motion, but still some trading off between forward and lateral thrust for the convenience of either prioritised forward/backward motion, or prioritised yaw (depending on whether the thrusters are mounted at 30° or 45°).
Note that if you really care about the most balanced possible control authority between different motion axes you’ll need a vehicle that’s spherically symmetric in both external shape (to avoid uneven drag) and mass distribution (to avoid uneven inertia), and will need carefully tuned neutral buoyancy for the water you’re operating in (to avoid the vehicle wanting to sink), and some form of wireless control like autonomy or an acoustic modem (to avoid tether drag).
That said, most applications do not have highly equal movement capabilities as the top priority, so most vehicle designs you’ll find make different tradeoffs to improve factors like efficiency, primary motion axes, payload capacity, ease of construction, etc.
