5 ms·
> With a ~20% learning curve, that is ~5 doublings from today, so more than net electricity demand Just to pick up on that point, electricity demand will have
by Brakenshire 8y ago
> With a ~20% learning curve, that is ~5 doublings from today, so more than net electricity demand
Just to pick up on that point, electricity demand will have to be a lot higher in future, because we're going to need to electrify transport and heating. That would be worth perhaps another doubling or more in electricity demand. And road transport at least looks like it's locked in to happen over the course of 2-3 decades through pure market forces.
- brianbreslin 8y agoGlobal electricity demand will skyrocket as places that now have less than ideal grids come online and local proliferation of cheaper electronics continues.
- benj111 8y ago"proliferation of cheaper electronics" Electronics use very little electricity. The biggest users of electricity are things with motors or heaters. Washing machines, dryers, ovens, kettles, etc. Electronics by comparison are a rounding error.
- dreamcompiler 8y agoExcept for places where there are a LOT of electronics, like datacenters.
- spydum 8y agoEven there - the lions share of electricity is consumed COOLING those electronics, but it is a fair point.
- D_Alex 8y agoThat is not correct, cooling uses maybe 20% of the electricity used by the electronics themselves.
- raducu 8y agoBut aren't all electronics just converting 99% of electricity to heat? Electronics don't just consume electricity by themselves -- they convert it to heat/light or noise.
- D_Alex 8y agoTrue, but an air conditioning system removes multiple kWhrs of heat for each kWhr of electricity. See for example here: http://www.currentforce.com.au/coefficient-of-performance/ http://www.currentforce.com.au/coefficient-of-performance/
- benj111 8y agoAs I understand it, cooling the data centres uses as much electricity as the servers themselves use[1]. Of course you could make a very good case for including that in the electronics usage totals anyway. [1] http://large.stanford.edu/courses/2018/ph240/mangu2/ http://large.stanford.edu/courses/2018/ph240/mangu2/
- JumpCrisscross 8y ago> The biggest users of electricity are things with motors or heaters I'd guess it's something industrial. Chemicals, in particular, are energy-intensive processes.
- benj111 8y agoMost of the energy in an industrial process would be heat? With second being mixing and other movement? So that's back to motors and heaters. I cant think of any exceptions apart from electrolysis.
- JumpCrisscross 8y ago> Most of the energy in an industrial process would be heat? I don't believe so. When industrial processes require heat, I've usually seen the fuel burned at the plant. Electricity is used for compressors, pumps and such.
- benj111 8y agoYou said energy, I assumed electrical energy. So yes most heat probably comes from burning fuel direct. Would be interesting to know what proportion of electrical energy is used in heating. I would guess its non negligible. Edit: Making make sense, and addition.
- petre 8y agoAluminium production requires lots of electricity. Other metal recycling and heat treating processes too, not to mention extrudation processes.
- jabl 8y agoAFAIK a lot of steel recycling around the world uses electric arc furnaces.
- salty_biscuits 8y agoYeah, electrolysis of aluminum is a huge user of power. You've got the heating to several hundred degrees Celsius and electrolysis combined.
- jayalpha 8y ago"Electronics by comparison are a rounding error" Big rounding error! By 2040, computers will need more electricity than the world can generate https://www.theregister.co.uk/2016/07/25/semiconductor_industry_association_international_technology_roadmap_for_semiconductors/ https://www.theregister.co.uk/2016/07/25/semiconductor_indus...
- benj111 8y agoDouble check the table on that article, particularly the scale. Then as a quick sanity check, consider whether the thing you are actually predicting will in fact use more electricity than the world uses, and still carry on growing exponentially.
- jayalpha 8y agoDouble check before you post if you are really smarter than the guys at ITRS (https://en.wikipedia.org/wiki/International_Technology_Roadmap_for_Semiconductors https://en.wikipedia.org/wiki/International_Technology_Roadm... ) or if you are just maybe incapable to understand the graph.
- benj111 8y agoThe scale on the left is orders of magnitude 1.00E+3 is 1000, 1.00E+4 is 10000, do you have a different interpretation? The benchmark line on the graph shows that by 2040 we'll be using 1.00E22 electricity versus 1.00E21 world usage. So that means we'll be using 10 times more electricity on computing than everything else. And the lines seem to cross at about 2037. Do you think we can increase electricity production 10 fold in 3 years?
- jayalpha 8y agoWow, what a smart fuck you are. Make this "we'll be using" a "would be using" and try to thing about the other graphs.
- 8y ago
- semi-extrinsic 8y agoWe're talking as you say a doubling or tripling of demand for electricity in that case. The costs to replace the current transmission and distribution grids in the US are estimated at $500 billion and $1800 billion, respectively. So we're talking ballpark $5 trillion to enable a tripling in electricity demand. Over a ten year period, this corresponds approximately to the entire US defense budget. We're literally talking about digging up all of the streets, everywhere. I don't want to be negative, but that's what we are talking about.
- aidenn0 8y ago> We're literally talking about digging up all of the streets, everywhere. I don't want to be negative, but that's what we are talking about. Aren't the vast majority of power lines above ground?
- torpfactory 8y agoI was thinking about this very problem the other day and I think it actually may be the best reason why residential storage batteries (as opposed to grid-scale) make sense. It might be easier to keep your transmission system and then trickle charge a whole bunch of small batteries to power end uses.
- wongarsu 8y agoThe increase wouldn't be evenly distributed though. Industrial areas where huge buildings need to be heated and a lot of freight traffic happens might see a much larger increase in power demand than a remote farm. If we manage to coordinate the charging patterns of cars (for example with electricity prices that change by the minute based on local demand) we might not even need to replace most of the residential grid: Personal vehicles are mostly charged over night when load is currently at 1/4 of the peak handeled in the evening. A lot of other uses like heating can also be shifted with the right incentives to not occur at times of peak usage, further reducing the load increase.
- Brakenshire 8y agoI don’t think that’s actually true, transmission grids are built to handle peak demand, and a doubling or a tripling of total electricity use does not mean a doubling or tripling of peak usage. In temperate climates for instance charging of cars will almost entirely happen overnight, with a significant rate discount. The basic existence of such a large battery will smooth out the peaks as well as add to them. In fact you’re likely to get far better utilization out of the existing infrastructure, and any new infrastructure which is built, which will bring the per unit costs down. And in any case, the electricity grid pays for itself, if people are willing to pay the going rate then the infrastructure can be built to satisfy that demand. What is the amount of money spent each year on maintaining and upgrading oil and gas infrastructure? For that matter what is spent maintaining the existing grid? These are huge figures, electricity companies will not be doing this for free, they are happy to maintain the grid in exchange for the revenues that come from it, and they will be only too happy to grow into competitor markets and use what was their competitors’ revenues to fund their expansion.