Synergistic Infrastructure

[WhatThePhysics]

Solar thermal hydroelectric desalination systems in desert based vertical farms made of line-x coated concrete, which contain rebar and biodegradable capsules full of calcium lactate and bacillus bacteria, with rooms that have reflective and hydrophobic surfaces, shelves of potted plants that are fed by subsurface drip irrigation and nutrient supply tubes, high pressure and heat resistant chambers growing algae to produce synthetic petroleum, air ducts that lead to small wind turbines, and light emitting diodes powered by motors turned by strings and weights cranked up by flowing water.

Relative to existing infrastructure, does this sound feasible, cost effective, and more efficient?

Sources
3D Printed Homes: http://inhabitat.com/3d-printed-house-in-china-can-withstand-an-8-0-earthquake/

Algae Based Synthetic Petroleum: http://www.forbes.com/sites/christopherhelman/2013/12/23/green-oil-scientists-turn-algae-into-petroleum-in-30-minutes/

Cultured Meat: https://culturedbeef.org/

FarmBots: https://farmbot.io/

Gravity Powered Light: http://gravitylight.org/#/solution/

Hydroelectricity: https://water.usgs.gov/edu/hyhowworks.html

Hydrophobics: http://www.neverwet.com/applications/anti-wetting.php

Line-X: http://linex.com/protective-coatings

Regenerating Concrete: http://www.cnn.com/2015/05/14/tech/bioconcrete-delft-jonkers/

Small Wind Turbines: http://www.americanwindinc.com/

Solar Thermal Desalination: https://users.cs.duke.edu/~reif/paper/solar/SolarDesal/SolarDesal.pdf

Solar Thermal Power: http://www.eia.gov/energyexplained/?page=solar_thermal_power_plants

Subsurface Drip Irrigation: http://www.netafimusa.com/agriculture/applications/subsurface

[Trekkin]

Frankly, it sounds like a classic eierlegende Wollmilchsau; it's not at all clear what function this (presumed) structure is intended to perform. You've listed articles describing speculative technologies for power generation, water purification, agriculture and synthetic biology; are you trying to simultaneously generate electric power, water, petroleum, vegetables, meat, and houses?

If you can clarify what you want this thing to do, it might be possible to evaluate it.

As it is, relative to existing infrastructure generally, it's certainly a more expensive way to build a bridge.

[WhatThePhysics]

Frankly, it sounds like a classic eierlegende Wollmilchsau; it's not at all clear what function this (presumed) structure is intended to perform. You've listed articles describing speculative technologies for power generation, water purification, agriculture and synthetic biology; are you trying to simultaneously generate electric power, water, petroleum, vegetables, meat, and houses?

If you can clarify what you want this thing to do, it might be possible to evaluate it.

As it is, relative to existing infrastructure generally, it's certainly a more expensive way to build a bridge.

Yes, the facility is intended to simultaneously purify water, collect hydroelectric, solar thermal, and wind energy, while also producing plant matter and synthetic petroleum.

[Spojaz]

Relative to existing infrastructure, does this sound feasible, cost effective, and more efficient?

No.

Without taking billions of man hours to design these systems to synergize, and absolutely constantly maintain them, all of the bits will interfere with one another, rather than helping.

The algae, desalinization and solar panels fight over every photon of sunlight.
The plants, algae, self repairing concrete, cultured meat, hydroelectric, and especially desalination all need a whole bunch of water, which will be very hard to come by in the desert.The system will certainly not generate any hydroelectric power after you have finished pumping that corrosive saltwater all the way from the coast, and up the tower.
Any fault or failure in the more mechanical sections will leak lubricant and poison the biological, and the living things will be striving (as life does) to grow, fouling the supply lines and mechanical components.
Regulating the temperature, chemistry and pressure to keep all the different kinds of life mentioned alive together would be an enormous challenge.

The only thing that such a design could maximize is space efficiency, and we have a planet far larger than twelve billion people it will eventually support needs. It seems a lot better to use these technologies separately, and placing them where they can do the most good. I.E, producing food near the centers of population, solar power in the deserts, desalination in the ocean, wind power where it is windy.

Perhaps one or two of those functions could be combined into one plant in cities or space colonies, where space is at a premium, but mostly it works better to produce things over there, then bring them over here.

[WhatThePhysics]

No.

Without taking billions of man hours to design these systems to synergize, and absolutely constantly maintain them, all of the bits will interfere with one another, rather than helping.

The algae, desalinization and solar panels fight over every photon of sunlight.
The plants, algae, self repairing concrete, cultured meat, hydroelectric, and especially desalination all need a whole bunch of water, which will be very hard to come by in the desert.The system will certainly not generate any hydroelectric power after you have finished pumping that corrosive saltwater all the way from the coast, and up the tower.
Any fault or failure in the more mechanical sections will leak lubricant and poison the biological, and the living things will be striving (as life does) to grow, fouling the supply lines and mechanical components.
Regulating the temperature, chemistry and pressure to keep all the different kinds of life mentioned alive together would be an enormous challenge.

The only thing that such a design could maximize is space efficiency, and we have a planet far larger than twelve billion people it will eventually support needs. It seems a lot better to use these technologies separately, and placing them where they can do the most good. I.E, producing food near the centers of population, solar power in the deserts, desalination in the ocean, wind power where it is windy.

Perhaps one or two of those functions could be combined into one plant in cities or space colonies, where space is at a premium, but mostly it works better to produce things over there, then bring them over here.

I might be wrong, but here are my thoughts regarding the problems you presented:

I think a bunch of neural networks and engineers could shorten the time required to design the systems. Certain architectures could be refined to serve specialized purposes, as I don't expect the initial setup to be universally applicable.

The synthetic petroleum and desalination units are placed in separate rooms, while solar panels aren't meant to be a vital part of the system.

If a pipeline is too impractical to build and maintain, due to resource requirements and thermal expansion, perhaps a fleet of electric powered water trucks would be more practical. The seawater isn't meant to be pumped upwards, but turned into steam that rises to the top of the facility. The steam condenses, flows down to irrigate while providing hydroelectric energy, then restarts the cycle at the bottom of the facility. The salt is meant to remain in the lower levels, and become molten to produce steam explosions when more seawater is introduced.

The rooms are coated with hydrophobics, and have floor drains that recycle any spilled fresh water.

Temperature and pressure can be regulated with water sheets and vacuums. Rooms specialized for certain plants can have the solar mirrors direct specific amounts of light, or be located in the central portions that are illuminated by light emitting diodes.

While I agree that integrating all of these techniques might be impractical, I think there are many benefits to be had by merging them together for the sake of general optimization.

[Knaight]

I think a bunch of neural networks and engineers could shorten the time required to design the systems. Certain architectures could be refined to serve specialized purposes, as I don't expect the initial setup to be universally applicable.
That one could have a bunch of engineers involved doesn't magically make severe fundamental engineering problems not severe. There's no reason to do this.

[QUOTE=WhatThePhysics;21717784If a pipeline is too impractical to build and maintain, due to resource requirements and thermal expansion, perhaps a fleet of electric powered water trucks would be more practical. The seawater isn't meant to be pumped upwards, but turned into steam that rises to the top of the facility. The steam condenses, flows down to irrigate while providing hydroelectric energy, then restarts the cycle at the bottom of the facility. The salt is meant to remain in the lower levels, and become molten to produce steam explosions when more seawater is introduced.[/QUOTE]
Replacing a pipeline with a truck fleet is a terrible method to make resource use more efficient. The fundamental problem is that if you have to transport the water long distance you use more power transporting it than you'll gain using it in a hydroelectric system, which are generally built along rivers for a reason. Swapping pipes for trucks just makes this worse.

The fundamental question is basically asking whether it would be more efficient to lose every localization advantage involved in using natural resources except for solar intensity for solar panels. The answer to that is always going to be no.

[Rockphed]

There are already greenhouses in the desert that use salt water to keep cool and desalinate it to water the plants. It is probably more effective to cover all of the Sahara and Arabian deserts with those and ship the food produced to cities than it would be to build arcologies in the arabian desert.

[WhatThePhysics]

That one could have a bunch of engineers involved doesn't magically make severe fundamental engineering problems not severe. There's no reason to do this.

Replacing a pipeline with a truck fleet is a terrible method to make resource use more efficient. The fundamental problem is that if you have to transport the water long distance you use more power transporting it than you'll gain using it in a hydroelectric system, which are generally built along rivers for a reason. Swapping pipes for trucks just makes this worse.

The fundamental question is basically asking whether it would be more efficient to lose every localization advantage involved in using natural resources except for solar intensity for solar panels. The answer to that is always going to be no.

Though, I imagine engineers with some neural nets on hand could chip away at the problems more rapidly. One example that comes to mind would be finding more efficient turbine designs with evolutionary algorithms and physics engines.

Would building a seawater pipeline shaded beneath a half cylinder coated in solar panels be more effective? Or, what if the solar paneled shade is raised, and has small wind turbines beneath it to exploit the artificial draft it'd create?

Besides water and fertilizer, I'm not sure what other resources would need to be imported. Also, do you believe using solar thermal energy to produce synthetic petroleum is not cost effective?


There are already greenhouses in the desert that use salt water to keep cool and desalinate it to water the plants. It is probably more effective to cover all of the Sahara and Arabian deserts with those and ship the food produced to cities than it would be to build arcologies in the arabian desert.

I agree, though I think there's some merit to building multifunctional and long lasting structures that can withstand earthquakes. If many of these are constructed, they might serve as resilient nodes to an anti-desertification grid, and supply local communities with various types of renewable resources.

[Trekkin]

Neither neural networks nor engineers are trivial to parallelize. Neural networks in particular are not omniscient magic oracles; they are best used to explore large state spaces, but they need careful programming and benchmarking not to output garbage. That takes considerable time. Engineers are less likely to do that, but there's still the issue of making sure all their designs fit together, which in a system like this is going to require real-world testing.

A lot of the systems you're calling synergistic are actually working at cross purposes. The hydroelectric desalination plant, for example, is necessarily going to impede the flow of hot water away from the boiler to generate power with it, which raises the pressure in the boiler and requires more solar input for the same water flow out. You aren't getting any free power out, just building a bigger boiler/mirror array instead of solar panels.

The far more pervasive obstacle, though, is that synergized systems are inherently vulnerable to cascade failures, like Spojaz said. Redundancy offers some defense, but it limits your peak efficiency when everything's packed in together like you've got it here -- and this isn't just a problem at startup. Consider, for example, what happens when you shut off the boilers to remove the salt (which you will have to do before it physically overflows the chambers, if nothing else) -- you've lost not just water, but also electricity and heat regulation. Now your culture vats aren't being hydrated, stirred, aerated, or heated, so they're going to start rotting, which is going to foul your harvesting systems...see where I'm going with this? This isn't even taking random malfunctions into account, nor that redundant biological systems need to be kept alive even when nominally idle. Physically separating them doesn't help when you need them all running and supporting each other.

Now consider a distributed system, like we actually have and like Knaight and Spojaz have suggested. If you have a bunch of desalination plants piping water into a common network, you can shut them off at random and possibly borrow water from elsewhere in the grid and generally juggle the demands however you want so long as the flow rate to the farms keeps up. Likewise, a petroleum farm can leak and not get all over your wind turbines, and it's possible to totally isolate and sterilize it if it gets infected.

The question is not whether the facility can work, it's whether it can work at midnight in freezing acid rain during an earthquake while shot full of holes and lit afire by irate townsfolk and infested with bloodthirsty radioactive bees infected with horrible zoonotic viruses while your staff are all actively trying to destroy everything you've ever built and the coffee machine is broken.

Anything less is a catastrophe waiting to happen.

[WhatThePhysics]

Neither neural networks nor engineers are trivial to parallelize. Neural networks in particular are not omniscient magic oracles; they are best used to explore large state spaces, but they need careful programming and benchmarking not to output garbage. That takes considerable time. Engineers are less likely to do that, but there's still the issue of making sure all their designs fit together, which in a system like this is going to require real-world testing.

A lot of the systems you're calling synergistic are actually working at cross purposes. The hydroelectric desalination plant, for example, is necessarily going to impede the flow of hot water away from the boiler to generate power with it, which raises the pressure in the boiler and requires more solar input for the same water flow out. You aren't getting any free power out, just building a bigger boiler/mirror array instead of solar panels.

The far more pervasive obstacle, though, is that synergized systems are inherently vulnerable to cascade failures, like Spojaz said. Redundancy offers some defense, but it limits your peak efficiency when everything's packed in together like you've got it here -- and this isn't just a problem at startup. Consider, for example, what happens when you shut off the boilers to remove the salt (which you will have to do before it physically overflows the chambers, if nothing else) -- you've lost not just water, but also electricity and heat regulation. Now your culture vats aren't being hydrated, stirred, aerated, or heated, so they're going to start rotting, which is going to foul your harvesting systems...see where I'm going with this? This isn't even taking random malfunctions into account, nor that redundant biological systems need to be kept alive even when nominally idle. Physically separating them doesn't help when you need them all running and supporting each other.

Now consider a distributed system, like we actually have and like Knight and Spojaz have suggested. If you have a bunch of desalination plants piping water into a common network, you can shut them off at random and possibly borrow water from elsewhere in the grid and generally juggle the demands however you want so long as the flow rate to the farms keeps up. Likewise, a petroleum farm can leak and not get all over your wind turbines, and it's possible to totally isolate and sterilize it if it gets infected.

The question is not whether the facility can work, it's whether it can work at midnight in freezing acid rain during an earthquake while shot full of holes and lit afire by irate townsfolk and infested with bloodthirsty radioactive bees infected with horrible zoonotic viruses while your staff are all actively trying to destroy everything you've ever built and the coffee machine is broken.

Anything less is a catastrophe waiting to happen.

Please forgive and correct me if I mistype.

I imagine that starting with simple schematics can help train the neural networks to optimize basic designs, or individual components that are inserted into a more complex final product. As they reach specific efficiencies, they can be reapplied to blend components together into a virtual gestalt. A team of engineers could first test the simple designs and individual components, then work their way up to a miniature model of the final product. If the gestalt proves ineffective, at least they've refined a wide variety of systems that could be marketed, open sourced, or repurposed for greater efficiency.

I'm not sure if I've made an error, or that you're misunderstanding the concept. The desalination produces steam that rises and moves turbines, and isn't an energy input for the desalination system. If you're agreeing with this, and stating that the steam inhibits desalination, then perhaps ditching steam power is more efficient. In that case, there are still hydromechanical and gravitic irrigation benefits, and the steam could be condensed by water tanks that are cooled by frigid desert nights.

Regarding salt removal, why not flush it out during the night with pressurized hydromechanics? I see your point about backup power, and suggest that the hydromechanical cranks are used for more than just light emitting diodes. As for random malfunctions and redundant biological systems, could you list some examples?

About distributed systems, the facility is intended to have modular rooms with individual subnetworks. We're not talking about a small building, but one large enough to generate a return in the short and long term. This would likely require the reserve systems you referenced, such as common water networks. Concerning the petroleum farms and wind turbines, the two systems are in completely different compartments, and I'm having trouble envisioning how ideally contained petroleum will spray upwards into grated air ducts. I'll admit that I made the mistake of saying the algae is grown in the pressure chambers, but it's clear that growing them in segmented aquariums that feed into the chambers is more efficient.

As far as I know, line-x resists corrosion, abrasion, kinetic impacts, freezing, and heating. This makes it useful in an apocalyptic scenario involving simultaneous freezing acid rain, a barrage of bullets, and plenty of fires. Also, the concrete is full of capsules with dormant bacteria and the nutrients necessary for self-healing, and the rebar is coated with line-x to improve structural integrity. On the subject of radioactive bees with zoonotic viruses, you could deny them entry with compartmentalized rooms that have grated air ducts, and sterilize interior surfaces by spraying them down with boiling water that's drained through the floor.

[Knaight]

Now consider a distributed system, like we actually have and like Knight and Spojaz have suggested. If you have a bunch of desalination plants piping water into a common network, you can shut them off at random and possibly borrow water from elsewhere in the grid and generally juggle the demands however you want so long as the flow rate to the farms keeps up. Likewise, a petroleum farm can leak and not get all over your wind turbines, and it's possible to totally isolate and sterilize it if it gets infected.

On top of that, the various distributed systems can be put in places that make sense. Those desalination plants can be put on coastlines, where the water is right there for use. The wind turbines can be put in areas that get heavy winds, the solar panels in areas that get lots of sunlight, the biofuel farm* somewhere that fits temperature and water needs and which is easy to do bioremediation on, the hydroelectric dams at rivers. The redundancy alone is worth a lot, but even if that wasn't a problem being able to select sites that fit the various aspects of the hypothetical giant facility is worth a lot.

Basically, I agree with everything said here except the misspelling of my name.

*Which is still in research stages anyways, and which

[Trekkin]

Regarding neural networks: I'm afraid that isn't how neural networks work in practice. Despite the name, neural networks are just another approach to machine learning; they're loosely "inspired by biology" in a way that lets the popular press write amusing articles for the laity but determines nothing about their capabilities. Machine learning is itself a way around the usual strictures of rule-based programming in cases where we can evaluate the final result but aren't sure how to explicitly encode the underlying logic. This is why they're used for classification. Consider face recognition: you and I know what a face looks like, but it's hard to express something like "the set of all possible eyebrows" mathematically. The neural net lets us evaluate the performance of many candidate algorithms post ex facto; provided there is already an algorithm for finding faces somewhere in the initial state space, the network will eventually favor it over the others.
The above is a gross oversimplification, but the key is that machine learning lets us program from the top down rather than the bottom up. In cases like this design, where we legitimately don't know what the right design looks like, it becomes much more difficult to train the network -- and these networks are task-specific, so we can't train it to design things we already know how to build and then somehow reapply it to something else. If you train it to design a hammer and then feed it a set of candidate greenhouse plans, it will return to you the best hammer from among them; getting it to do anything else starts with resetting the network.

As far as the desalination plant, the plans you've outlined to extract energy from either steam or output water require that you put more energy into the steam. Desalination plants work most efficiently with low-temperature low-pressure steam; they can extract energy from the output water via a countercurrent heat exchanger, but extracting mechanical power would require raising the system pressure. Steam turbines in particular need hot live steam; even something as simple as letting the steam rise a long way to power waterwheels would need a tall boiler with lots of surface area through which to lose heat and on whose walls the steam would inevitably condense. There's just no way to pull energy out of the output water that won't require adding energy in -- and even if you rinse out the boilers at night, that's just that much more energy it's going to need to start back up in the morning.

For a list of random malfunctions, I'd start with the problems faced by aeroponic and aquaponic systems. Anything affecting the flow rate adversely affects your ability to regulate the environment, which is a problem in algae tanks when dead algae get into the water flow. Pipes and nozzles and pumps and aerators can all get clogged, at which point the whole system starts to deoxygenate and/or starve. This is going to change the physical and chemical properties of the algal solution as well as its production efficiency, all of which are going to need ad hoc correction. Then, too, pathogens and fungi can get into the tanks, and even in a sterile system you can get random DNA damage sabotaging your synthetic metabolic pathways. Remember, making petroleum is not good for algae; the useless strains are going to proliferate faster, ceteris paribus.

That same metastability pervades all biological systems. Let's say you want a backup algae tank in case you need to sterilize and repopulate your main system; that backup tank needs light and heat and air just like the main tank or else it's going to die. Thus it makes petroleum, which is either inefficiently unused or used and depended upon, at which point a malfunction in either tank is a shortfall.

Furthermore, metastability underlies the cascade failure problems everyone has been pointing out. It doesn't take a complete failure to chill the algae tanks below optimal temperature, at which point they don't make enough petrol and start fouling the water with byproducts; anything dependent upon that is likewise now dealing with suboptimal conditions. If you're going to go to the trouble of isolating the systems from each other so that a failure in any one doesn't impede the others because they're all ganged together, then you might as well put them all where the local environmental conditions most closely match their optimal conditions and just accept the relative inefficiencies of pumping all the reagents around.

I guess the thesis of my argument is that what you gain in shorter transport infrastructure you lose in more complicated maintenance procedures and more interdependent failure conditions, so by the time you separate them enough to be safe you've already made half of a distributed system and you gain in efficiency by moving them farther away. My exact disaster scenario was hyperbole born of conversations with nuclear engineers, intended more to illustrate how multiple failures can be exponentially harder to fix as they impede your efforts to address any one of them.

One final point: if you're serious about understanding the technologies you've posted about here, I'd stay away from anything written for lay consumption or put on a corporate website and go for primary literature. Science journalism tends to elide all the complexities of implementation in favor of affected breathless awe about how this latest whiz-bang technology is just like the movies; if you want accuracy, read things that have gone through peer review.

Knaight: sorry for the typo. It's fixed now!