What could be reasons to use direct current in the distribution of power?
I know that it is more efficient to distribute current at high voltages to reduce heat loss and that it is easier to step up voltage in AC. Regardless, what would be reasons to not use it (or at least not so wide-spread as it is now)?
Would I only need an alternate history or would I have to change the complete environment?
In the late 19th century, Thomas Edison supported the use of direct current, and his company, Edison General Electric, attempted to use it on a large scale. A major competitor, George Westinghouse, supported the use of alternating current, with his competing company. Both tried to corner the market, but doing so would require the large-scale adoption of one of the two types of current. The resulting battle - which was hard-fought, very public, and world-changing - had a number of significant events:
If you could change the history books such that Edison remained involved with Edison General Electric (or controlled GE after the merger with Thomson-Houston), he could have remained a major player in the market, and would have held the power to continue to advocate for DC. Edison might have won the propaganda war, but if there had been many more accidents, and Westinghouse and other AC companies had lost revenue on a large scale, it's possible that Edison General Electric would have held onto the market, and DC would have been the only real option, financially. The only way to overcome the technical benefits of alternating current is to make it seem dangerous, and unfeasible or unpalatable to adopt.
Actually, DC is superior to AC in almost every way, there's no loss due to induced currents, no frequency synchronization issues, less wiring, etc. And this is totally not counting the fact that all solid state electronics have to be DC. If we had to re-build the electric grid from scratch(or build one on e.g. Mars), it would definitely be DC.
The ONE issue (but it's a doozy) with DC, is that in order to change voltage, you need semiconductors, whereas with AC all you need is a bent metal bar and 2 coils of wire. With no way to step up/down voltage a DC grid is totally impossible until the 50's, by which time the network effects of AC are too big to overcome.
There is one scenario in which it could have happened: in S.M. Stirling's Draka, alt history series, the Draka have a widely distributed compressed air distribution system in their cities before electricity became big (it was a thing IRL too, but was not dominant, since it was almost concurrent with electrical grids). So when electricity becomes big, it's first in the form in the form of small generators being driven from mains compressed air. I should point out that Stirling doesn't extend the concept into a DC grid, but that could be done.
If say Tesla cooked himself while wiring Niagara Falls and burned down Buffalo, killed tourists or disrupted the waterfall in the process that would have been the end of it.
If solar or antenna static worked well enough (they are inherently DC) to power everything the advantages of AC grids would be moot.
The reason AC won out (besides the politics), it because it was more efficient economical* to transmit long distances (at the current time). If humanity decided to make power plant production small scale instead of large scale, it might have gone to DC very easily. However, it would have to be something like solar cells or wind farms, because small steam plants are very inefficient^[citation needed].
Besides the politics mentioned in HDE 226868's answer, power was produced in huge steam plants which were best located outside of town, making transmission efficiency a top priority. If transmission efficiency was not a high priority, then it just as well could have gone to DC.
Fun fact, a lot of our appliances that have a clock use the frequency of the power grid to keep time. This would not be possible on DC. Here's a cool video from Tom Scott about it.
*By this I mean that the feasibility, cost, and technology back when we were creating the national grid meant that AC was better than DC to transmit long distances. Today, they are equitable, and some even argue that they could have very quickly solved the problems that DC had if they wanted to.
DC is efficient at very high voltages (over 500kV ) where the capacitive losses to ground start mattering. It also helps that you don't have to worry about frequency synchronized systems, so failures don't necessarily break the world.
Assuming the OP is talking about a hypothetical future reason to build out the Grid using DC transmission, then the invention of aneutronic fusion reactors would be the reason.
Current research on fusion reactors is focused on D-D or D-T fusion since it is theoretically easier to achieve (although "easier" is a relative term), but the issue there is the energy release is mainly in the form of neutron radiation. In most hypothetical designs, the neutron radiation is captured in something like a "lithium blanket" and the heat energy extracted to create steam to run conventional turbines. This is essentially a nuclear powered tea kettle.
Hypothetical aneutronic machines using reactors like p-B or 3He-3He will produce high energy alpha particles as their outputs, and cleverly designed machines will create high energy beams of alpha particles, essentially a giant direct current beam which can be harvested for energy by decelerating the beam in a reverse particle beam accelerator. The energy release is inherently DC, and converting it to AC would create a great deal of loss. If the promise of these machines can be realized, they will be small and cheap enough to power individual neighbourhoods, or be shipped to places needing energy in ISO containers (realistically, they would probably need to be shipped in several containers, and there are issues with side reactions which don't fundamentally change the premise of the answer).
So the conversion is relatively simple and modular; EnergX builds these things like toasters and signs up subdivisions, shopping malls, industrial parks and individual factories for their product. Each customer that goes on line is wired for DC power, and disconnected from the old grid. Gradually the grid is dismantled and replaced by millions of small DC grids, optimized for their particular situations. There may be some transition issues, and likely the speed of the conversion will be set by the lifespans of legacy generation systems (utilities will be very upset if they lose their customer base while still having 30 year bonds to pay off, so there will be political opposition as well).
Of course the real key is the development of working aneutronic fusion systems in the first place.
What type of power (AC or DC) is used is determined by three major factors.
How easy it is to produce the power.
Electric power is typically produced from mechanical energy. Usually some arrangement of spinning magnets/electromagnets, and coils. The voltage waveform produced by a spinning magnet/coil is a sine-wave (AC).
To turn it into DC requires a rectifier of some kind. In Edison's time there was not any efficient means of rectifying AC electricity.
How easy it is to move the power from where it is produced to where it is used.
Our current model is that electric power is produced in large power plants, and then distributed to the individual buildings that use the power.
Often times power plants are located away from cities so that they can be near a natural power source. One example is the Niagara Falls power plant. Another example would be a geo-thermal power plant. In other cases power plants are located away from cities because they produce pollution (such as fossil fuel burning plants).
If power is produced far away from the point of use then there must be a way to efficiently transmit it. Typically this involves transmitting the power at very high voltage and then stepping down the voltage at the point of use.
For AC power all that is required to step/down the voltage is a transformer. Transformers are fairly simple. They are typically two insulated wires wrapped around an iron core and stored in an oil filled metal can. A transformer that supplies power to an entire house sits comfortably on a utility pole and has a volume of just couple of cubic feet. In todays prices the power company can buy one for a few hundred dollars and it may last several decades. Its hard to beat the price.
To change the voltage level on DC power you need a DC to DC converter. This typically involves some high speed power switches inductors, and capacitors. In Edison's time diodes and solid state switches had not yet been invented. His only choice for a switching element would have been relays or vacuum tubes. Both of those have limited lifespans, especially at high power. Such systems would need constant maintenance and be quite large.
There are a couple of things that would have made DC power feasible in Edison's time.
How easy it is to use the power at its destination.
For an electric power company to be viable you need someone to want to buy the power. The first and most important applications of electric power were light bulbs, and electric motors.
For a an incandescent light bulb its equally easy to use AC or DC power.
The typical DC electric motor requires brush contacts that tend to wear out. DC motors with brush contacts are less reliable than AC motors.
The typical AC reluctance motor is just a set of stationary wire coils and a rotating iron core attached to a shaft. There is no brush contacts so these types of motors can last decades and are very cheap and low maintenance.
A lot of AC electric machinery can also use the AC waveform as a timing source to regulate speed or output power.
In some situations, using DC currents is more efficient. Practically all electronic circuits use DC, so if you have a lot of electronic together you may wish to convert your mains to DC. Most companies that keep datacenters nowadays do that. Here is an article explaining the technicalities involved - TL;DR, you lose some efficiency when converting from AC to DC, so having it all DC from the start is best in this scenario:
DC distribution is not just for the giants
In a world where Datacenters become increasingly more popular and surpass everything else in terms of demand for energy, DC will be more common than AC.
DC current used in distribution networks may make a comeback in a hypothetical future when perhaps all electricity will be generated by fusion plants and the power plants will be located very closely together so that the failure of any one will only cause local outages. In such a situation the use of AC will not have as many advantages and the use of DC will not cause massive pollution from non existing fossil fuel power plants spaced very closely together..
If you never need to transmit large amounts of power over long distances, DC might win out. No matter what you make your wires out of, there will always be some amount of resistance. The longer the wire, the more the resistance adds up. You can make the wire fatter, but that gets more expensive. So, to carry more power, you do it at higher voltage rather than higher current so you can keep the wire thickness reasonable. That means you need a way to step voltage up from your generating voltage and down to your consuming voltage at each end of the transmission line. Using alternating current, voltages can be easily and reasonably efficiently stepped up and down with transformers, which don't require advanced technology or exotic materials to construct.
If there was never a need to distribute large amounts of power over hundreds of miles - say just a few miles at most - the economics that favor A/C might not exist.
Beyond the politics of the war of the currents, there is the reality of of technological evolution. The need to distribute large amounts of power over long distances emerged when the only loads were incandescent lamps and motors - at least a couple of decades before the emergence of electronics. Such loads were indifferent to whether the supplied power was A/C or D/C; only voltage mattered. Without the benefit of semiconductor devices which only emerged around 1950, and more importantly, solid-state device capable of switching high current at high voltage (which have really only been around a few decades now), efficient high power D/C to D/C conversion just isn't possible.
If you're going to build out a large-scale power infrastructure without the benefit of modern thyristors and triacs, chances are, you'll end up going A/C.
RADICALLY EDITED ANSWER AFTER CONSIDERATIONS:
A DC power plant could work if you had a much smaller population to care for. A plant like that could provide the needs of a small community and be used especially as recharger.
This scenario implies that society here evolved radically toward portability and lower power usage. Optimization of power banks and the awareness regarding their usage would prevent waste.
Of course, a bg city couldn't be born around one single power plant if we wanted to power up all houses. And a network of smaller generators would increase pollution, so there'd be this 'industrial' village built around the plant, a first ring where production of basic goods and technology is concentrated, while the rest of the city, in a second, larger ring, would be powered with solar cells and technology with batteries.
For the sake of completeness, I add this kind of late, bland answer:
Home appliances are mainly DC, industrial ones are more prone to AC. If a way to build, operate and maintain cheap superconductors was invented, maybe DC transmission would be more widespread.
There is lots of research about room-temperature superconductors, and a graphene-based solution could be near.
DC can do everything AC can do, just that DC needs a rotating machine for voltage conversion, where AC can use a no-moving-parts transformer.
In DC, that is irritating and expensive because rotating DC machines need commutators, which is a cylinder of copper bars brushed by carbon brushes. From time to time, the surface needs be smoothed, brushes changed, and mica insulators between commutator bars notched down a bit so they are lower than the copper. (Otherwise abrasive mica will tear up the brushes pretty fast). Carbon brush dust must also be removed from the mica notches to keep it from shorting.
This is all a big pain.
Now imagine a slightly semiconducting teflon coating which could be painted on top of the copper bars. Where it is thick (between bars) its resistance is very high. Where it is thin (over copper bars) its resistance is nil. It's laid in a continuous surface - no mica gaps. This, coupled with a magic brush material, means the DC commutator just doesn't wear. Bearings are well mastered, so the upshot is that a rotating machine has about the same reliability as a transformer.
They even put the bigger ones in sealed cylinders and pull vacuum on them to eliminate windage loss and increase insulation strength.
Voltage conversion can now be easily handled with an M-G set (if you need isolation) or a dynamotor (if you don't).
Now, why put up with the drawbacks of AC (e.g. Having to synchronize grids)? If someone really really wants AC, an M-A set can do that.
Until rectifiers came along, railroads preferred DC power for railway electrification. Series-wound DC motors pull like a mountain goat, and locomotives are big enough to solve the handling issues.
Suppose the railroads electrified very, very early, so the very first electricity in any town showed up in the 1880's as the railroad's trolley wire. All the depots and freight houses would be electrified first, followed very quickly by the rich people. Since 3000 volts DC is an awkward voltage to use for residential lighting, the railroads would provide M-G sets to knock it down to 100/200 split, 100 for light bulbs, 200 for motor loads. (Later bumped to 105 and 110V to increase system capacity).
This would spread like wildfire and the whole country would soon be electrified. 110 volts would be a household word. (Even though later they bumped to 115 and 120).
Before electricity, every town with a river had a mill pond. Those mill pond operators decide to install generators and backfeed the system, since it's much, much easier to synchronize into a DC grid than AC.
And then, Aermotor comes out with an electric generation kit for their windmills. While they only officially support off-grid local use, they intentionally design it so it's trivial to make it generate into the grid. Sales explode. The year is barely 1900 and home generation is already a thing.
High Voltage Direct Current (HVDC)
This is a method of transmission using DC at voltages of 100 kV to 1500 kV.
See Wikipedia (https://en.wikipedia.org/wiki/High-voltage_direct_current):
A long distance point to point HVDC transmission scheme generally has lower overall investment cost and lower losses than an equivalent AC transmission scheme. HVDC conversion equipment at the terminal stations is costly, but the total DC transmission line costs over long distances are lower than AC line of the same distance. HVDC requires less conductor per unit distance than an AC line, as there is no need to support three phases[clarification needed] and there is no skin effect.
This doesn't require futuristic technology. It's currently in use in some places, and may become more popular.
There are disadvantages (see that Wikipedia page), but these might be reduced by future technological development.
How strong the case for using it is depends on how the power is generated (the cost is reduced if the power is already DC, which might be the case, as other answers suggest) and the length of the transmission line (with greater benefits for longer lines). Using it for shorter lines might be somewhat more justified if a society was willing to invest a higher capital cost to reduce ongoing costs.
High Voltage DC (HVDC) for power transmission is already quite common. If you want more detail, you may be better off here asking precisely why we use HVDC.
The main advantage of it, is that it can join asynchronous grids - that is, ones that run at different frequencies and/or voltages. AC grids must be syncronised. If one side of the transmission line has poor quality of supply, then people may be hesitant to connect via an AC interconnector.
Countries around the world use a wide variety of frequency and voltage. No country wants to change over - it would simply involve throwing out too much existing stuff. Think of the costs and benefits that would result from the USA swapping cars and roads so they could drive on the correct side of the road. It'd be similar with grid frequency.
So if you're joining up two different electricity grids, you use HVDC, so neither side has to change grid frequency.
So if you want HVDC for transmission (the big power lines between cities), then simply make states, counties, or cities more independent of each other, and make it so that all of them established their own local grids before connecting them together.
Note that you will still have AC distribution (the small power lines in each city) in this situation.
If your world built high-quality portable batteries much sooner (think lithium ion) and had people amazed with portable power that might be enough to make DC king.
Also make people horrified by the sight of transmission lines marring the sky/cityscape.
