Summary
You may have come across the term ‘fluid flux barrier’, or ‘fluid barrier’, or in any case fluid-something, in the context of synchronous reluctance motors.
I certainly have, on multiple occasions. In fact, I encountered the term multiple times over a period of multiple years, without ever really looking into what fluid flux barriers are.
Read on to find out.
What are synchronous reluctance motors
Synchronous reluctance motors (SynRM) are electric motors that operate based on the saliency of the rotor, aka magnetic asymmetry, aka the flux density finding it easier to travel into one direction versus other. The stator is essentially a standard slotted stator with a polyphase stator, fed with (ideally) sinusoidal waveform. Not like a switched reluctance motor, where the stator is also salient and fed with trapezoidal-type currents.
Most commonly, the saliency is achieved by means of flux barriers, i.e. (magnetic) voids in the rotor steel laminations. These direct the flux to travel along the parts that do have iron, often called flux guides.
These motors, unsurprisingly, have both pros and cons.
The rotor has no current-carrying conductors nor large conductive bodies like magnets susceptible to eddy currents, and hence can have rather low losses (surface iron losses can be significant due to the short airgap, though). Nor does it have anything overtly sensitive to temperatures, again like magnets. These factors make the thermal management of the rotor easy – it’s comparatively low-loss, and could be allowed to heat up quite a bit.
On the flip side, SynRMs do have their drawbacks, too. Their power factor is generally poorer than PM or induction motors, their torque capacity is often nothing stellar, and they do constant-power field-weakening rather abysmally, more often than not.
Furthermore, the electromagnetic and mechanical aspects are very contradictory, perhaps more so than in any other topology. The electromagnetic side would love magic flux guides floating in air, connected by nothing ferromagnetic. This is often not feasible (although perhaps plastic fillers and epoxy glue might work at very low speeds) – instead the guides are typically joined by thin iron bridges. Obviously, magnetics would like those bridges to be as thin as possible, which is not something mechanics likes.
Finally, specifying the shape of the flux barriers quickly gets very difficult, especially if you are thinking about numerical optimization. As in, how to define decent barrier shapes without having to use a million parameters/dimensions is a non-trivial question.
This is where fluid barriers come in – read on to find out more.
What are fluid barriers
Defining ‘optimal’ barrier shapes – and a lot does depend on the barrier shape – impossible without actually simulating the entire motor geometry (and defining what we mean by optimal in the first place). That’s given.
But, fluid barriers take an a-priori blind stab at the problem by making some (or quite many) simplifying assumptions.
Simply put, fluid barriers follow the flux density as it would flow in a solid rotor, with no flux barriers (or saturation for that matter). In other words, determine the flux density in a solid rotor, and then draw a flux line starting from a desired spot, and you have one side of your barrier. Repeat, starting from a different spot to get the other side, and then join the sides with suitable curves and fillets and you have your barrier.
You can modify this approach to include e.g. the shaft, or even pole shaping because why not, but the overall methodology remains the same.
What are fluid barriers good for
The obvious benefit of this approach is the low number of parameters needed.
The the minimum, you could just define the combined barrier width (or flux guide width) at the mid-axis of the pole for instance, and the number of barriers used, and then spread those uniformly (along the pole mid-axis, again). That’s it. Well, you’d probably need to specify the minimum thickness for the outer bridge, and some filleting parameters for the corner, but still – that’s only a few parameters to get a really nice-looking rotor.
The other benefit is that since the barriers do, in a limited way, follow a ‘natural’ flux pattern, they will probably perform okay…ish. At least if you compare the results to any more-naiive approach with equally few parameters (for example the simple V- or W-shaped barriers seen in early SynRMs).
What are fluid barriers NOT good for
Despite their apparent physical background, fluid barriers are generally not optimal in any respect (saliency ratio, torque ripple, mechanics, you name it). How the flux would flow in a solid rotor sadly does not bear that much relevance to what happens once you add the some barriers into the mix, or saturation, or stator slotting, or or or…You get the point.
Another thing is that the ‘fluid’ principle does not define an entire barrier on its own. You can trace some curves based on it, but you’ll still need to connect two curves to form a single barrier, and so on.
How are the barrier shapes actually defined
Now, for actually defining the barrier shapes, we need to be able to solve the flux density distribution in a solid rotor. The traditional, perhaps mathematically prettier approach has been to use a conformal mapping; first solving the flux density in an easier geometry and then using a mapping to transform it into a cylindrical geometry.
This approach has the potential of being very lightweight computationally, and probably not lightweight at all when you actually need to code it for the first time.
Another approach is to throw some CPU time on it (many computational problems have a habit of disappearing once you do this, just like many real-world problems disappear if you can throw money at them).
I have personally read an article by Ali Jamali Fard that used the finite difference method for solving the flux pattern. Myself, I’m way more versed in finite element analysis, so I used that instead when implementing fluid barriers to EMDtool.
The choice of the underlying tool is not critical here – it’s the choice of using a numerical approach in the first place. It can have the benefit of getting started faster, and the ability to handle arbitrarily-shaped geometries. Having a shaped pole-face springs to mind as a possible example, or extending the principle to the small pocket-barriers often found in IPM traction motors.
On the con side, a numerical approach is a numerical approach, and needs its share in CPU time. Granted, creating the rotor geometry will be a small extra cost on top of the actual analysis (provided the actual analysis is anything remotely realistic), but even smallish costs can and do add up once you do a million of them.
Additionally, making the numerical approach work requires the ability to interface with a numerical solver in a rather detailed fashion. Having to fire up another instance of a Commercial Software, creating the geometry via scripting, solving the problem, extracting the flux lines, saving them do a .dxf or .csv file, and then importing said lines to another instance to actually begin with the actual analysis sounds like something of a pain.
By contrast, having access to the fundamental-level FEA utilities like meshing and matrix assembly (ehrm, EMDtool again) lets you (or me, in this case) encapsulate the barrier geometry creation behind a hassle-free interface. Work once, use forever in other words.
Check out EMDtool - Electric Motor Design toolbox for Matlab.
Need help with electric motor design or design software? Let's get in touch - satisfaction guaranteed!
