On June 21, 1961, in a dusty small town in Texas, about 60 miles south of Houston, President John F. Kennedy stepped up to the podium to make a speech before the assembled throngs of people below. Or rather, his aides did. The ongoing Cuban missile crisis demanded his presence in Washington, so he instead gave the speech through a telephone call. The matter, however, was deemed of sufficient importance that Vice-President Lyndon Johnson was in attendance.

Behind the stage was a dense, tangled maze of stainless steel piping, tanks, and stairways—the fruits of labor of countless scientists, engineers, and government officials who had been working relentlessly for years. And thanks to their efforts, the project had been completed at breakneck speed. Kennedy’s speech was to commemorate the opening of the facility, and the geopolitical situation did not temper his enthusiasm or his grandiloquence.

“Today,” he said, “was an important step towards securing one of man’s oldest dreams”. With effusive praise for the scientists and engineers involved, he went on to declare that he could not think of a cause that was more important “to people around the globe”, and that this was “a work that in many ways is more important than any other scientific enterprise in which this country is now engaged”.

Kennedy was not talking about space exploration, the Apollo program or the goal of landing a man on the moon, which he had announced a month earlier in an address to Congress. He was talking about the opening of a desalination plant in Freeport, Texas, the first of its kind ever built. Fed by belching coal boilers, it would be capable of producing one million gallons of freshwater per day, providing relief to the drought-ravaged town of Freeport, as well as a nearby chemical plant owned by the Dow Chemical Company.

The practical implications were overshadowed by the symbolic ones. The 1960s were an era of unbridled optimism and the belief in the ability of science and technology to deal the finishing blow in mankind’s eternal war against poverty, disease, drought, and famine. Medical advances and improved sanitation increased the average American lifespan from 53 years in 1920 to 70 years in 1960. Nuclear power promised cheap, abundant electricity. The Green Revolution made crop yields explode, cementing America’s food security. It seemed that the final enemy left to overcome was drought.

In a sense, this plant represented the ultimate triumph of human ingenuity over the vicissitudes of the weather and climate. Finally, mankind would meet its basic needs in “areas in which Nature was not generous”. The deserts would bloom, and men and women in dry climates everywhere would be lifted out of poverty.

The starry-eyed enthusiasm didn’t last. The high spirits of the age were met with crushing economic reality. The mercurial nature of weather goes both ways, and a few years later, the rains came back to South Texas. It cost the government $1.25 to produce one thousand gallons of water, which it sold to the town of Freeport for ¢30, the price of naturally sourced water. Once the funds dedicated to the project ran out, Congress had no inclination to keep producing a now-abundant commodity at a loss. The plant shut down in 1969. The 1970s energy crisis, which more than tripled the cost of oil, also severely curtailed any incentive to build these terribly energy intensive facilities.

---

It’s somewhat of a stroke of irony that 70% of the world’s surface is covered in water, yet the IPCC estimates that 80% of the world’s population faces serious threats to water security. Like the Greek figure Tantalus, there’s a promise of plentiful water that frustratingly forever lies just-so-slightly out of reach.

Most water used by humans comes from the hydrological cycle. In the warm ocean regions, water evaporates and thanks to atmospheric circulation, moves around the globe. Eventually, it condenses into clouds and falls out of the sky as precipitation. Depending on where it lands, the water can accumulate in ice caps or glaciers, flow back into the oceans through rivers, or be stored in lakes or aquifiers. Although the water eventually makes its way back to the oceans, all these serve as sources of fresh water for humans, plants, and animals.

But the availability of these resources is highly location-dependent. There’s no guarantee that your land will be pocked by lakes, that rivers will run free, or that the aquifier will provide useful quantities of water. There are only two ways to increase the available water supply beyond the cycle can provide: water reuse and desalination. And if you didn’t have water to begin with, the first option is not particularly helpful.

Climate change will make all these issues worse. Droughts will increase in severity and the distribution of rainfall, snowmelt and river flows will shift. Potentially billions of people will experience water shortages, and an overall more precarious water supply situation.

To address these issues, we will need to turn to seawater desalination more and more. Historically, it’s been an incredibly energy-intensive process, resulting in high costs as well as increased environmental impact, which limits its potential usefulness. But could that be improved? How much better could it get, and would more R&D get us there?

A brief history of desalination

The largest, most challenging and complex part of a chemical plant is not, as is popularly thought, the reactors that combine chemicals to produce new ones, but rather it is usually the process of splitting a mixture of chemicals into distinct streams of a specified purity. This discipline is known as separation processes, and it is the bread and butter of chemical engineering.

Fortunately, desalination is a relatively easy process. Humans have done it for millenia—Aristotle noted that “salt water, when it turns into vapour, becomes sweet and the vapour does not form salt water again when it condenses”. But it was never feasible on a large scale prior to the Industrial Revolution, due to the high energy requirements, and was subsequently of little interest.

The idea was revisited during the 1800s with the advent of the steam engine, mostly in military contexts. Militaries were not as concerned about economic viability as they were with the logistical convenience of being able to produce fresh water for their troops in-situ. Research into economic desalination accelerated after WWII, particularly in the United States. In the 1950s, the southern US underwent a severe drought—the worst in the nation’s history—and there was significant political pressure to find a solution. This is when some began to consider the possibility of large-scale, economic distillation of seawater as a source of water security and a bulwark against the fitful caprices of rain. Congress passed the Saline Water Conversion Act in 1952, which led to the creation of the Office of Saline Water in 1955 and its eventual merging with the Office of Water Resources Research in 1974.

The first desalination plants, like the one in Freeport, used multi-stage flash distillation (MSF) to distill the water. The idea is quite simple: boil the water and collect it, leaving the salt behind. A series of heat exchangers captures the heat from the condensing vapor to boil a subsequent stage of water, minimizing energy waste.

They are effective at producing drinkable water, but there’s a downside. They are terribly energy intensive. No matter how you slice it, boiling water takes a lot of energy, and not all of it can be recuperated. A typical MSF plant achieves a Specific Energy Consumption (SEC) of 23-27 kWh per m3 of water. At an electricity price of 20 cents per kWh, that means it would cost $5.00 to produce 1000L of water. But in comparison, the average price of water in the USA is only $0.40 per 1000L!

Paying over 12 times the average price for your water is a tough sell. In many cases, it would be cheaper to simply ship the water in from elsewhere. So it’s unsurprising that MSF desalination never saw the widespread adoption that Kennedy and Johnson would have liked, and indeed, in the 1970s, funding for desalination dried out. Desalination was a niche technology of last resort, only used in locations that had access to abundant energy but for whatever reason could not source water any other way.

The ultimate success of the industry can be credited to Sidney Loeb, an American-Israeli chemical engineer, who invented the first practical reverse-osmosis membrane, a much more energy efficient technology, thereby setting the stage for widespread economic viability. By the 1980s, membrane technology had become the standard for new desalination plants around the world.

How reverse osmosis works

Reverse osmosis is achieved by pushing the water across a semipermeable membrane via a pressure gradient. In a typical setup, the water, after having been filtered for large debris like branches, rocks, algae, etc., is fed through a pump and through a series of concentric tubes.

On the face of it, this just seems to be like any other kind of filtering process. What’s the big deal? I use a “semipermeable membrane” to separate the coffee grounds from my coffee every morning. Why does this have a fancy name attached to it?

The main difference between reverse osmosis and a filtering system is the spontaneity of the process. A fluid filled with undesirable particles will naturally flow through a sieve in the absence of a minimal driving force (often, the only force is gravity). Increasing the pressure will increase the flow rate through the filter, but it’s not a requirement. If I was fine with waiting forever, eventually all of the coffee would trickle down through the filter, absent a small amount retained due to surface tension, resulting in a separation efficiency of nearly 100%.

Reverse osmosis, on the other hand, is not spontaneous. If you place a concentrated salt solution next to a reverse osmosis membrane without applying a driving force, nothing will happen. The water will never migrate across the membrane on its own. To make this happen, you need to fight against the osmotic pressure generated by the difference in salt concentration across the membrane. This is where the majority of the energy requirements come from: pressurizing the water to force it across. The amount of pressure needed depends on the membrane permeability as well as the concentration of salt in the seawater.

The pressure requirements are where the bulk of the energy consumption for desalination come from. But how much of that is actually needed? Would it be possible to create some kind of magical filter that would let water pass through really easily, but keep the salt out, or to invent some new technique that would be able to sidestep the need for high pressure? Wouldn’t that allow us to have cheap and easy freshwater anywhere in the world, and thus eliminate the threat of thirst for humankind once and for all?

As it turns out, the answer is no. There is a fundamental thermodynamic limit to how energy efficient desalination—and other separation processes—can be.

The energy of unmixing

So how much energy is actually needed to separate a mixture’s components from one another? What is, so to speak, the theoretical minimum? We’re going to actually mathematically derive it in this section, but if you are not interested in that, you can skip to the next section and just look at the graphs instead, without too much loss.

First of all, let’s start by defining what we mean by “mixture”. We are referring to a fluid consisting of multiple components (e.g. water, salt, sugar, ethanol) but which has homogenous physical properties throughout the fluid. By contrast, a fluid with heterogeneous physical properties (e.g. water with sand and sticks in it) is called a suspension. There are some additional subtleties¹, but this definition is good enough for us.

Secondly, is it even a given that separating two fluids from each other requires energy? The answer is an emphatic yes. At a very high level, one can make a somewhat handwavy argument using entropy: two separate pure chemicals have lower entropy than a single vat with both chemicals mixed together. Since, from the second law of thermodynamics, entropy must always increase, then we could make the argument that separating the two requires work to be applied so that entropy increases elsewhere (most likely from the process used to generate useful work)².

So we need to expend energy to separate fluids, because we are fighting entropy. It’s a solid argument, but it’s not a particularly useful one—it doesn’t really provide any intuition about the relevant physics nor does it us tell anything about how much energy is needed. Fortunately, it does give us an idea where to look: the entropy of mixing³, which, for a two-component mixture, is given by the following equation: $$ \Delta S_{mix} = -nR[x\ln x + (1-x)\ln (1-x)] $$ $x$ is the mole fraction of the first component, and since there are two components, $x_1 + x_2 = 1 \rightarrow x_2 = x_1-1$. $n$ and $R$ are the number of moles in the system and the ideal gas constant, respectively. As $0 \leq x \leq 1$, then $\ln x$ and $\ln(x-1)$ are both negative and thus $\Delta S_{mix}$ is always positive, as expected.

Next, let’s look at a high-level block diagram of a hypothetic reverse osmosis unit.

While we want to extract fresh water from the process, we can’t extract all of the water, or there would be nothing to transport the salt out of the system. Thus we have one stream—the seawater—going into the unit and two streams exiting it: a stream of desalinated water, and a stream of concentrated saltwater (called the brine). The ratio of the flow rate of the fresh water stream to the inlet stream is called the recovery ratio. $$ R_w = \frac{\text{freshwater flow rate}}{\text{feed flow rate}} $$ The recovery ratio roughly captures a tradeoff between capital expenditures and operating expenditures. Desalination plants commonly use a recovery ratio of 50%.

From the entropy of mixing, we want to next get an expression for the thermodynamic favorability of the process. The desired outcome doesn’t happen because it would decrease the entropy of the system. This means that we need to add energy to make the process thermodynamically favorable. This quantity is given by the Gibbs free energy of mixing. $$ \Delta G_{mix} = \Delta H - T\Delta S_{mix} = nRT[x\ln x + (1-x)\ln (1-x)] $$ Where $\Delta H$ is enthalpy and $T$ is temperature (in absolute units). Without adding energy to the system $(\Delta H = 0)$, the Gibbs free energy is always negative so mixing will spontaneously occur. But we’re interested in the Gibbs free energy of separation, not mixing. This is calculated by taking the difference between the free energy of mixing of the two exit streams and the free energy of mixing of the initial feed: $$ \Delta G_{sep} = \Delta G_{mix, brine} + \Delta G_{mix, fresh} - \Delta G_{mix, feed} $$ We’re starting to get somewhere useful, but this is still too abstract. We want to substitute in values that we can measure in order to get an physical intuition about the system. From here, we can do two things: normalize $\Delta G$ to a per-mole basis and then multiply by the relative flowrates of each stream, and then convert the per-mole basis to a per-volume basis, giving us energy per volume of water, i.e. the Specific Energy Consumption. With some manipulations we can get the following equation:

$$ SEC = 2RT \left[\frac{c_{feed}}{R_w}\ln\left(\frac{c_{brine}}{c_{feed}}\right) - c_{fresh}\ln\left(\frac{c_{brine}}{c_{fresh}}\right)\right] $$ Where $c$ is the concentration of ions (in mol/m3) in the stream. Note that salt dissociates into two ions, which is why the equation is multiplied by two. If we further assume that the freshwater stream has perfect rejection (completely removes salt), then $c_{fresh} = 0$ and so we get the simplified version: $$ SEC = \frac{2RTc_{feed}}{R_w}\ln\left(\frac{c_{brine}}{c_{feed}}\right) $$ At this point, this is looking like something we can work with, but we can make some further substitutions. The first is from the definition of the recovery ratio. Since $$ \frac{c_{brine}}{c_{feed}} = \frac{1}{1- R_w} $$ Secondly, while these values are all things we can directly measure, it would be useful to relate them into a single value that is relevant to the process at hand: osmotic pressure. Osmotic pressure is a colligative property, meaning that only solute concentration will affect it. The relationship between solute concentration and osmotic pressure for dilute solutions can be estimated using the van’t Hoff equation: $$ \Pi = icRT $$ Where $c$ is the solute concentration and $i$ is an index term that corrects for the actual concentration of particles produced when a solute is dissolved. Hmm, looks familiar! Thus the SEC needed to separate a water-salt mixture is: $$ SEC = -\frac{\Pi}{R_w}\ln(1-R_w) $$ For a typical seawater salt concentration of 35 g/L, (with a resulting osmotic pressure of 29.7 bar), and assuming a recovery ratio of 50%, we get a value of 1.1 kWh/m3 of water. This means that no matter the separation technique, it is thermodynamically impossible to desalinate water using less energy than this⁴.

Note that this doesn’t work the other way around. Certain separation techniques will be more energy efficient than others, and there’s no reason that any specific separation technique would be able to reach the theoretical minimum. It turns out that reverse osmosis in particular is capable of doing so.

The other, perhaps unintuitive conclusion is that osmosis is a natural consequence of the second law of thermodynamics. Without it, you would theoretically be able to create a net reduction in entropy simply by using a reverse osmosis membrane as a sieve.

Additional limitations

This analysis applies to a thermodynamically reversible process, but this can’t be done in practice. The problem is that as the water gets transferred across the membrane, the concentration of the brine increases, which leads to a corresponding increase in the osmotic pressure.

In theory, to achieve perfect efficiency we would need to continuously increase the pressure along the membrane proportionately to the increase in salt concentration. This is, to say the least, an impractical arrangement. The best we could do, assuming no pressure losses, would be to maintain the pressure constant over the length of the membrane. This means that in order to achieve the recovery ratio we want, the pressure needs to be at least equal to the final osmotic pressure of the brine.

If we plot the equilibrium pressure vs. water recovery on a graph, and the area below the graph is “thermodynamic minimum” energy and the area above is the “wasted” energy.

This relationship is, in a sense, the “rocket equation” of desalination, because there’s very few ways to actually improve it. The equilibrium curve is a fundamental property of the fluid, and so there’s no technological solution that can change it. There’s only two things we can do: change the target recovery ratio, or, just like with rockets, add staging by changing the number of “steps” we take to reach the final recovery fraction.

This means instead of having a single large reverse osmosis unit, we place multiple smaller units in series, with pumps in between each unit to boost the pressure to the new equilibrium pressure. However, this tends to significantly increase capital and maintenance costs, so this isn’t a lightly made design decision.

How well can we do in practice?

All of this theoretical discussion is interesting, but meaningless unless we’re at energy consumption levels where the theory is actually relevant. And if we had remained stuck at 1970s-era membrane technology, this would be the case. But fortunately, modern membranes have come a very long way, with an almost 90% reduction in energy consumption since then.

They can now sustainably achieve a thermodynamic efficiency of around 80%, which means they only require about 20% more energy than the thermodynamic limit. A single-stage RO process with 50% recovery requires ~1.6 kWh/m3 of energy in theory, and so the best single-stage RO systems available today consume about 2.0 kWh/m3 in practice.

What does that number actually mean? Canada consumed about 4.86 billion cubic metres of potable water in 2019. If all of that water came from desalination, it would represent an additional 9.7 TWh of electricity per year. Canada’s annual electricity production is about 632 TWh, so desalination would roughly represent 1.5% of total consumption. By comparison, if MSF desalination was used instead, it would require over 120 TWh to supply all of Canada’s water!

And of course, the reverse osmosis step is but one of several needed to obtain drinkable water, but state-of-the-art facilities like in Israel are extremely efficient. The Sorek plant, which opened in 2013, consumes about 740 GWh annually and has a capacity of 624,000 m3/day, enough water for 1.5 million people. This corresponds to a total energy consumption of 3.3 kWh/m3. So on top of the ~2.0 kWh/m3 needed for reverse osmosis, they only need 1.3 kWh/m3 for pre-filtration, post-treatment, intake, discharge, pressure losses, and the works. It is an incredible achievement that we are able to get this close.

These systems are much more complex than the simple process flow diagram previously shown, with parallelized pipelines, circulation loops, heat exchangers, and energy recovery devices. A typical piping and instrumentation diagram for a reverse osmosis system will look like this:

A very common way to improve the energy efficiency is to attempt to reuse as much of the pressure energy from the brine as possible, since it exits as high pressure but no longer has any need to be pressurized. Pressure exchangers are nifty devices that can do this with very high efficiency—in a sense they fulfill a similar function as a gas generator turbopump in a rocket engine⁵.

And this is only a fraction of the plant's functionality! Depending on the quality of the membrane and permeate, sometimes an additional pass is needed, called Brackish Water Reverse Osmosis (BWRO). In addition, the plant requires systems for prefiltering the water and handling all the sludge that gets separated out, a post-treatment system that adds minerals, disinfects and adjusts the pH, and antiscaling systems that flush out minerals from the RO membranes to prevent fouling, and more!

Can we improve on this?

The energy intensity of desalination is one of the primary reasons why it hasn’t seen more widespread adoption. Electricity costs account for almost half of the operating costs of a typical desalination facility.

And unfortunately, that means there’s no silver bullet to make desalination cheaper. There are only three ways to improve the energy efficiency of desalination:

1. Use a membrane that is closer to thermodynamic limits
2. Reducing the recovery ratio
3. Introducing a pressure ramp during the reverse osmosis process.

For #1, the thermodynamic limit places a fundamental restriction on its energy efficiency. We’re not at a local maximum, we’re approaching the global maximum. Even if you invented a groundbreaking, “perfect” membrane, it would be at best 20% more energy efficient than the state of the art today. But in practice diminishing returns are likely to apply here as well: individual percentage point improvements to efficiency are going to become harder and harder to chase, while R&D costs will soar. As this sobering review by Patel et al. notes, there is a “universal insignificance of novel materials in further enhancing the energy efficiency of desalination”.

#2 is a classical engineering trade study. But the tradeoff becomes increasingly steep at the extremes. At 100% recovery, you would need an infinitely high pressure to achieve the desired flow rate, and at 0% recovery, you would need an infinitely large feed stream to get any kind of freshwater whatsoever. Could R&D conceivably reduce the capital expenditure of building high-flow systems? Sure, but at this point the technological surface area that needs to be covered is massive. Could you improve pipes by adding a coating to reduce friction? Find ways to make the coating less expensive? Find ways to manufacture pumps for cheaper? And these will only chip away at the overall costs, since their impact is directly proportional to their percentage of the total cost of the system. These incremental improvements could happen incidentally, but are not worth seeking out specifically.

#3 actually has some potential for improvement. There’s a fair bit of unexplored territory—adding multiple stages is not the only way to accomplish #3. Another approach is called batch reverse osmosis. Instead of continually flowing the water through in a steady-state configuration, you perform the reverse osmosis in big, static vats, where you gradually increase the pressure over time.

This lets you, in principle, approach the thermodynamic efficiency, although there are other inefficiencies with batch processes that need to be overcome. And as you can imagine, throughput becomes a problem since you have to perform all the operations sequentially: filling the vat, pressurizing the vat, and emptying the vat. This means you spend only a fraction of your time actually producing freshwater.

For example, this press release by Purdue University describing a new reverse osmosis ‘breakthrough’ is simply a system that increases the throughput of a batch process by using alternating chambers and a reciprocating piston to ensure water is always being pressed through a membrane. They were able to achieve a SEC of 1.88 kWh/m3, which is a modest improvement over traditional continuous processes.

In practice, this would probably end up shifting the recovery ratio “sweet spot”, so that the plant could use smaller feed streams to produce the same quantity of potable water, resulting in small reductions in capital and operating costs. There’s still a long way to go, though. These systems haven’t proven that they can scale up well, or that they can operate with the same longevity as continuous systems. And at the end of the day, there’s still a limit that can’t be overcome, regardless of the approach used.

What does this mean?

Desalination has reached its endgame. As energy efficient as desalination has become, it’s not going to improve much beyond this point. And while the costs have dramatically improved, achieving low costs requires a serious amount of domain expertise and capital, and these requirements are unlikely to ever go away. The dreams of making deserts bloom remain that—literal pipe dreams.

Today, about 1% of the world’s drinking water is provided through desalination. This is likely going to rise as water security issues become more pressing. And the amount of desalination plants being built is rising, thanks to the dropping costs. But capacity is growing steadily, with no evidence of an “exponential” adoption curve—if anything, growth is slowing down slightly, as the lowest-hanging fruits get picked.

There’s a common idea that droughts are acute, but short periods of no rainfall. While this can be the case, in reality they are often prolonged periods of reduced rainfall that can last years. And while a plant can be built, commissioned, and run profitably for a few years over the duration of a long drought, these plants are typically planned and designed to operate over timescales of 30+ years. And just like what happened with Freeport, this makes the economic projections of these ventures fraught with uncertainty. Desalination is a long-term solution, and thus the political and economic incentives also need to be long-term ones.

Previously, I used Canada as an example of the “reasonable” energy demands of modern desalination, but it should be pointed out that Canada is one of the most energy intensive countries per capita as well as one of the world’s most developed economies. Even at these very high efficiencies, poorer countries, particularly those in regions most susceptible to water insecurity, would likely still struggle to provide a sufficient amount of power.

Does this mean desalination R&D is a dead-end as a career field? That depends on your idea of a dead-end. We focused exclusively on energy efficiency, but there are a whole host of other considerations with RO systems that we barely even touched upon, with membrane longevity and fouling resistance being key among them. There are still a lot of avenues to research. And there’s always the holistic approach. As Patel et al. argues, we need to begin to focus more on “the importance of process engineering—in place of innovation of advanced materials—for the sustained optimization of desalination energy efficiency.” If the world is to truly lean on desalination—and it is good enough to do so—then we will need lots of experienced people who know how to design and build efficient desalination plants in a cost-effective manner. Someone capable of doing so will likely have many times more impact than someone who comes up with a 2% more energy efficient membrane.

Furthermore, just because desalination isn’t becoming more energy efficient doesn’t mean that it can’t become more economic, which is ultimately what we care about. One way of viewing it is that desalination technology transforms a water scarcity problem into an energy scarcity problem. And while in some respects the same limitations remain—access to capital, skilled labor and political will—in others it makes the solution less dependent on the local geographical situation, since access to the sea for non-landlocked nations is usually not a big deal.

Perhaps the most effective way to make desalination more accessible is to go work in photovoltaics R&D and deployment. One way of making the process less expensive is to reduce the cost of the inputs—solar is cheap and only getting cheaper. And perhaps desalination could even have a role to play with offload excess renewable production from the grid. Desalination facilities coupled with solar power could be a cost-effective way of using excess power output, and the product, water, doesn’t spoil and is easy to store and transport. In fact, there have been proposals to do just that at the US-Mexico border.

Some closing thoughts

If I had to choose one discipline to represent the art of engineering, desalination would be near the top of my list. In many ways, it’s emblematic of the engineering method’s greatest strengths, but also its biggest limitations.

First, it’s a strong testament to the power of first principles reasoning and a scientifically-minded analysis. Starting from the statistical mechanics definition of entropy, something so abstract and high-level that we would assume that it would only be good for ivory-tower theoreticians, we were able to infer a whole bunch of the design considerations and limitations of desalination plants, and also understand the motivations behind current desalination research.

Second, it demonstrates how engineering can be used to improve the human condition. Stable access to water is a fundamental building block of civilization. It is necessary for life itself, yes, but it also feeds crops and carries away diseases like cholera and typhoid fever. It is the difference between an uninhabitable land and a hotbed of human activity. Millions of people, especially in the Middle East, rely on desalination to ensure that they can drink, cook, clean, and bathe on a regular basis.

Third, it shows how continuous, compounding iterative improvement can lead to astonishing advances. In the history of desalination research, there are very few specific points that could be considered “breakthroughs”. Instead, it’s been a dogged and persistent effort by many scientists and engineers to bring the cost to the point where desalination is now a viable, cost-effective solution to many water supply issues rather than a niche, uneconomic venture wholly reliant on government subsidies.

But on the other side of the coin, the very scientific principles that have led to these tremendous improvements also show that we’re near the end of the road. Better designs and better technology will help—to a point. But desalination will never become the technological silver bullet some people hoped for. Finding natural sources of water will almost always be cheaper. Instead, it will remain a tool in the arsenal of policymakers and governments around the world—a very useful tool, to be sure, but one with specific strengths and weaknesses.

The history of desalination shows how strongly its success has been intertwined with politics and economics. In the 1950s and 60s, research was reliably championed by Senator-and-eventually-President Lyndon Johnson, who had grown up in the Texas Hill Country and had felt the punishing consequences of drought firsthand. But many plants in the 1970s and 1980s were shut down, unable to justify their costs. And in 1982, Ronald Reagan shuttered the Office of Water Resources Research, as part of his agenda to defund most (non-military) government science programs.

When confronted with a large, complex, and multifaceted problem without straightforward solutions, politicians, at least in North America, have this tendency to punt the issue under the belief that “innovation” will find a solution. Eventually, someone will invent a cost-effective technology that will solve the problem so well that applying it will be a no-brainer. There will be no tough calls about resource allocation, no exposure to potential gaffes or weaknesses exploitable by the opposition, no need to spend valuable political capital. This is the standard response to the pressing issues of our day—calls for “increased investment” and “enabling innovators” and other techno-optimist speak.

But that playbook won’t work here. There is no technological solution coming. The technology we have now is what we have to work with. It is up to us to decide whether we care about the problem or not and whether we are willing to expend a portion of our very limited resources on it.

The engineering method won’t solve this. And it never will.