Overall, it’s an interesting project and I hope they can solve all issues for a real deployment.
They don’t mention the length penalty that comes from wrapping around the cable or any impact of loss to fiber itself. Given the helical pattern, I’d imagine that they are adding substantial length to any distance they need to achieve. Generally you have about 20dB of loss before you need an amplifier or a terminal node. A typical number for field installed fiber is 0.25dB/km. So there’s some distance penalty needed here. I’m not clear on how you’d add amplifiers to these routes or if they are simple short hops.
Generally regular SMF-28 has a bend radius of 25mm. This is partially for loss reasons as bends below this radius induce loss and partially because bending fiber below that radius creates micro cracks in the fiber. The micro cracks can spread with time and cause the fiber to break. Given the rather large CTE mismatch between the power cable and the fiber cable, the 2 will expand at different rates with temperature, putting additional stress on the fiber, potentially leading to failure.
It’s possible to have a less bend sensitive fiber but it’s generally used more in specialty applications and not as transmission fiber. So it’ll be more expensive per km.
Along the lines of mechanical failure, it would seem that there would be the potential for abrasion between the power cable and the fiber cable jacket.
Finally, it’s a little unclear how repairs would be made. Doesn’t seem very safe to have to splice the fiber on a live cable like this.
While the length is an issue isn’t the expansion problem resolved by it being wrapped in a helical pattern? You basically have a spring at that point which can contract and expand with less stress than a pretensioned straight wire fixed at two points since it can slide across and rotate around the power line.
For repairs I’m guessing there will be no or very limited splicing you can have connectors at each pole and simply run a new fiber if you have a fiber break mid line the main reason for splicing fibers today in below ground installations is that pulling them out requires you to dig up a whole bunch of dirt if it’s just plug and let the robot rewrap the fiber along the power line I think it’s cheaper to do just that.
It's medium voltage lines which are up to tens of kilometers long, not high voltage hundreds of km. The spans between pylons are also much shorter than HV, so adding slack to allow for CTE mismatch seems doable without excessive abrasion.
On the illustration the winding is very sparse, it definitely does not look like running into minimal bend radius problem or adding much length.
I’m not sure what the diameter is of these cables but I found a data sheet that shows MV (33kV) cables are <50mm in diameter. If that’s true then, the bend radius is <25mm.
If the winding is sparse, let’s say 1 turn every 1m on a 38mm diameter cable, then the amount of length added will be ~1% of the total linear length (neglecting the cable sag). So the amount of added length due to winding it will negligible.
At 1000 bends per km, if we assume loss is 0.001dB per bend, then we’d add 1dB per km.
How do you know they really know their stuff? Is it because they used a lot of words you're not typically familiar with? It's straight out of the "blind them with bullshit" playbook and the modus operandi for a lot of HN comments these days.
OP is concerned about extra cable length that is needed with this approach (completely unfounded), drops a few attentuation loss over distance calculations, but really, the practical engineer says: "we'll have only as many winds of fiber per meter as is neccessary"
Not trying to dazzle anyone with anything. Just trying to understand the problem And have discussion. I’m not a cable engineer but these are very practical considerations.
I did a quick calculation with a simple assumption and admitted that my initial estimation of added length was incorrect and it’s not an issue.
The bends around a narrow diameter cable are a potential issue for both added loss and cable reliability as are the different rates of thermal expansion.
As far as the reliability goes, I wasn’t sure so I looked it up, see Table 1 on page 6 below. There’s an example of cable failure for a typical FTTH network. If you see this, then you’ll note that the number of turns and the radius of those bends is important in predicting the reliability of the cable plant. It’s worth noting that the cable plant failure mechanisms are dominated by breakage due to other things and not mechanical fatigue.
As long as the bend radius is >10mm the failure rate should be <0.5ppm per turn. For a 38mm diameter cable the failure rate <0.1ppm/ turn.
Repairing the fiber is one thing, but what about when repairs on the power-carrying cable is needed? Wrapping them up in each other certainly seems like it would complicate things...
Yeah, they'd need to bring in a specialist to repair the fiber optics as well, that's a specialist job.
Of course, that aside, cables generally don't need repair that much / often if they aren't in areas with extreme weather. And if they are, medium-power lines can go underground (as can the fiber optics).
I know it's not really relevant to the meat of the article, but I can't trust any of the numbers listed in the article. They claim that "Just 28 per cent of people living in Africa are internet users and two-thirds of those are located in South Africa."
If we take the population estimates of just 4 countries: Nigeria (206 million), Ethiopia (109 million), South Africa (60 million), and Kenya (47 million) we get 422 million people. 28% of that is 118 million, and 2/3 of that is 79 million. Clearly South Africa can't have 79 million internet users if they only have a population of about 60 million.
Found a source Google cited for the 28% figure in this [1] article. Not claiming it's accurate since they only source their own "ITU estimates". I also don't see a mention of "South Africa" in the stats:
Perhaps it is an estimate of Internet users using a conventional computer. I am pretty certain that the figure must be higher than 28% these days as pretty much everyone has a smart phone, even poor people in the slums of Kenya. There are however still quite a few feature phones out there though, often with keypads nearly worn down to the circuit board!
One thing I started to worry about and which wasn't addressed is that the technique of wrapping the cable around the power line hinders power line repair or could preclude future applications like this. Can the same robot traverse the same line with the cable already applied? Can a second cable be applied over the first?
Hah, I've never seen a utility schedule end of life for cables. They just replace them as they break, they'll replace chunks if the cable itself is looking bad, but otherwise they'll just splice it together with a coupler.
If fiber optic is in the way, they'll just cut it.
My first through is why is Facebook doing this? Why do they have a division that's developing Robotics?
Promising though. Much of the cost of running cable is having someone mount the cable, drive 30 feet down the road, and mount the cable. Over and Over. If they can slash the cost of last mile, that's big.
Google and Facebook have both at times invested a lot in expanding internet access or improving the quality of public internet. It makes sense given that they can't monetize their customers unless their customers have internet access.
More data as well, considering they can use browsing data from the internet connection itself to influence ad targeting even when browsing non-Facebook properties.
My laypersons assumption is: once a company gets big enough it makes sense to burn some cash on research projects as most(?) countries tax incentivise it.
Without getting into a lot of detail I don't see this being very useful. First off it only works on wires suspended via pin or post insulators on cross arms. There are plenty of other suspension methods used for MV distribution so good luck accommodating them. Most runs are normally interrupted by strain insulator sets on poles containing switch gear or poles that make sharp turns. It also needs to accommodate a wide variety of insulator designs which have varied over the years. Some are tall and narrow, some are short and mushroom shaped. Some have clamps at the top, some are tied with wire.
Any intervention with the robot also requires a highly trained and certified linesman crew to manipulate the thing on a live conductor with a potential upward of 34.5kV line to line. So if the robot gets stuck or breaks down you need a trained crew to handle it. And how easy is it to repair a broken fiber on a downed power line?
Interesting idea, but I don’t see any mention on how they deal with the power line stretching. As power draw goes up, the electric company pumps more juice, which heats up the cable and in summer months it gets so hot (everyone using A/C), that the wires start to stretch and sag. The fiber would need to deal with that.
I don't think overhead power lines heat up much in normal operation, given the economics involved. They will definitely expand/contact in line with ambient conditions however, and also must be flexible enough to sway in the wind without causing excessive abrasion between the cable and fibre.
Edit: the helical wrap would probably expand/contact more than the electrical cable, as it'd be "wider" overall.
If it is wrapped, the spiral form will provide the necessary slack. You just need to wrap tighter in summer and looser in winter when installing the fiber.
This is interesting, I've never heard of this. How much stretching/sagging are we talking about? (I.e. a couple of inches difference, or is it a large amount)
The article mentions the main costs with deploying fibre "lie within the construction of new poles and other infrastructure". Is there any reason why fibre can't be strung between existing power poles? They aren't conductive, so you don't need to worry about distances to buildings (i.e. they could be run lower than power lines).
Fiber is routinely hung on utility poles, in the same way hardline coax was deployed in previous decade, and open pair telegraph was hung in previous centuries. Actually comms cables on poles are much older than power cables.
You can actually purchase MV/HV cables with fiber optic built right into it - There's a few power companies in Canada that have built their infrastructure such that they can also provide fiber optic internet service to communities up north who otherwise wouldn't have internet access.
The costs involved have to do with labor needed to string cables. You need a crew of humans in bucket trucks moving from one set of poles to the next. Whereas this robot clamps to the conductor which acts as a rail and the little robot can string the fiber on its own for the length of the cable.
Right that's what I thought. Maybe in North America where labour costs are high this will be popular, but the article talks about developing countries where lower labour costs and less regulations make deployment of fibre a lot cheaper.
A telco was established in the UK by the National Grid during the 1990s using the high voltage distribution network as the backbone, and wrapping fibre around the neutral conductor. They called it Energis, and it had a great slogan at one point: "dial on the pylon"
IIRC after a while they moved to integrating the fibre bundle in the neutral conductor cable.
In the early 2000s I was on a project that was planning on using the fibre that was embedded in the neutral conductor in the local grid here in eastern Australia. It was used for telemetry and power control, but they thought they could sell access to ISPs and the like. I don't think it ended up going beyond a few pilot deployments and then our National Broadband Network got the go-ahead
We’ve wrapped fiber around ground wire in the past. Though here they’re doing it around the conductor. Also their robot is an advancement and pretty cool. But when I’ve worked in the power industry in the past the guys who had experimented with this stuff wound up really hating it for the difficulties it adds to maintenance and the like. So this may be cool but I wouldn’t be surprised to see it rarely used in actual practice. Strikes me a lot like project loon and all these other things where people are trying invent unnecessary shortcuts. If you really want to get internet to people with a reasonable bandwidth/coverage/and cost trade off start building large towers dotted across the continent and use wireless access, or wait for star link.
I don't see this being used other than by electrical grid operators themselves, there's no way they would let third party telecom companies do this, for grid-reliability and safety reasons.
I've seen some gnarly utility line webs in south america. There's something beautiful about the chaos, like a cyberpunk puzzle box for anyone that wants to get signals from it.
How do they pass the more complex transitions between line sections? I vaguely recall that being a significant limitation of existing line-crawling robots.
They don’t mention the length penalty that comes from wrapping around the cable or any impact of loss to fiber itself. Given the helical pattern, I’d imagine that they are adding substantial length to any distance they need to achieve. Generally you have about 20dB of loss before you need an amplifier or a terminal node. A typical number for field installed fiber is 0.25dB/km. So there’s some distance penalty needed here. I’m not clear on how you’d add amplifiers to these routes or if they are simple short hops.
Generally regular SMF-28 has a bend radius of 25mm. This is partially for loss reasons as bends below this radius induce loss and partially because bending fiber below that radius creates micro cracks in the fiber. The micro cracks can spread with time and cause the fiber to break. Given the rather large CTE mismatch between the power cable and the fiber cable, the 2 will expand at different rates with temperature, putting additional stress on the fiber, potentially leading to failure.
It’s possible to have a less bend sensitive fiber but it’s generally used more in specialty applications and not as transmission fiber. So it’ll be more expensive per km.
Along the lines of mechanical failure, it would seem that there would be the potential for abrasion between the power cable and the fiber cable jacket.
Finally, it’s a little unclear how repairs would be made. Doesn’t seem very safe to have to splice the fiber on a live cable like this.