All professional DC, BLDC or PMSM motor controllers that I have seen (Sevcon, etc.) have large numbers of DC bus capacitors connected in parallel. Their capacitances range around 100 µF - 220 µF. Wouldn't a single capacitor of a large value, like 4700 µF or 10000 µF, be more convenient?
Is it because of the large surge current whenever these controllers are connected to batteries or other high current power sources?
Sure having enough capacitance is one parameter. But capacitors have series resistance which limits how much peak current can be drawn from a capacitor. Capacitors also have series inductance which limits how fast you can get the peak current out. Having multiple smaller capacitors in parallel reduces both series resistance and inductance.
Higher ripple current capability, lower ESR and sometimes better form factor (eg. shorter) to fit in a convenient spot in the enclosure are likely reasons.
More surface area of the capacitor means more power dissipation capability, all other things being equal.
Other answers have already mentioned the main factors which determine that choice: lower total ESR, lower total inductance, better heat handling capability, etc.
I'll add one more aspect that has been neglected: reliability.
If you have just one big capacitor, once it fails, you are left with a nonworking system. Moreover, a bigger cap can do more damage to nearby components if it fails spectacularly.
Having multiple caps in parallel helps mitigate the effects you have when a cap fails open, because the others will still be there. You could even design the system with redundancy in mind, i.e. adding more caps than the minimum you would need given the other constraints.
There are also issues with endurance against vibrations (this is particularly relevant when dealing with big motors). A single, big capacitor can be stressed mechanically more heavily when subjected to vibrations. The big mass of the cap can resonate mechanically and exert a bigger stress on its terminals or its mounting points, leading to mechanical failure of the cap itself or the PCB it is attached to.
Smaller capacitors, since they have less mass, have less inertia, so they experience and cause less mechanical stress due to vibrations or shocks. Therefore it's also easier (and cheaper) to design appropriate strain reliefs to avoid mechanical stress and shocks causing problems.
The capacitors help in filtering and decoupling noise. But each single value of capacitor is only good at one particular frequency. It has least ESR (higher ability to mitigate noise) Using a range of values provides that good filtering ability over wide frequency range.
Reduced Heating due to ESR. As the ripple currents flow through the capacitors to and fro, the ESR opposes the current flow (similar to resistor). The higher ESR means higher power dissipation (as heat). This effectively raises the temperature of the capacitors. Higher the temperature lower the capacitance they can provide. Hence, low ESR over multiple frequency band is one desired parameter which can be effectively received by combining multiple capacitors than one single bug capacitor.
This could also be a production optimization thing. If a product already uses 220uF capacitors, using them instead of an additional 4700uF may make sense (though replacing one cap with 20 seems a bit extreme). A 4700uF cap is likely to be through-hole, and if it's the only through-hole component in a product, you save a whole manufacturing step if you can avoid it. Even if it's not, your stock becomes easier to manage because there are less part types to order, and you reduce the risk of having to redesign a product because that capacitor model goes out of production.
A single, custom capacitor optimized for the needs of that drive would probably have some advantages, if that was the only product you were building. But if you build dozens of different drives, as all drive manufacturers do, you want to optimize the supply chain across the entire product line. That means standardizing on as few building blocks as possible, and using them in various combinations to get the voltage and capacitance ratings you need.
This model needs two caps in parallel, another needs two in series, another needs four, another needs twenty, but you still only have to stock one part. You get economies of scale in purchasing, lower likelihood of running out of a part you need, and lower stocking costs overall. Bonus points if it's the same part a dozen other drive manufacturers are using, since they're probably building exactly the same drive frame sizes you are.
Now, if we could just get the power magnetics industry to work this way...
I think it is the best option for the manufacturer. After all, whatever cost less I think will be the preferred option for it.
