<!-- ARTICLE_BODY_START -->
Follow the second pipe
A seawater desalination plant is usually introduced through its useful output: a pipe delivering fresh water to a community. The other pipe deserves equal attention. It carries water that has retained much of the incoming salt while losing part of its original water. A proposal can improve the efficiency of the first pipe and still leave difficult questions about the second. Those questions concern coastal circulation, chemistry, organisms, maintenance and the conditions under which a plant must reduce production.
There is a current reason to look beyond a single energy number. The California Energy Commission's 2026 publication, Energy-Water Desalination Hub, describes the National Alliance for Water Innovation's research across materials, process development, and data and modeling. Its stated ambitions encompass distributed desalination and reuse, rather than one universally superior treatment train. The official project summary describes a research program and its aims; it does not establish that every proposed technology has achieved an economical commercial deployment.
The distinction matters to a public discussion often crowded with superlatives. A membrane can perform impressively in a controlled test. A recovery device can reduce a particular pump's electricity demand. A proposed mineral product can have a high market price. None of those observations, alone, gives a community a complete water supply. The community needs dependable delivery, an acceptable receiving environment, a workable operating budget and evidence that the entire arrangement remains credible when conditions change.
This article examines that arrangement through a deliberately hypothetical coastal utility. No fictional figures below describe an actual plant or permit. The utility wants to compare operating choices before selecting a facility, and its decision-makers agree on a simple discipline: every attractive number must retain its denominator, boundary and conditions. That discipline lets a reader understand the main trade-offs without pretending to produce a professional design or environmental assessment from a few calculations.
The sources were checked on October 31, 2026. They include an August 2026 manufacturer manual, a January 2026 research summary, a 2025 regulatory-agency memorandum and older scientific work whose dates remain explicit. Older evidence can still explain a mechanism or reveal a limitation. It cannot honestly be repackaged as a newly announced breakthrough. Nor does the existence of a state agency's guidance establish that the same permitting requirements apply on another coastline.
A recovery percentage needs a boundary
Reverse osmosis separates a pressurized feed into a lower-salinity permeate and a concentrate stream. Recovery is the fraction of feed obtained as permeate; salt rejection describes a different property. DuPont's FilmTec Reverse Osmosis Membranes Technical Manual, version 20 dated August 2026, distinguishes those terms and explains the role of pressure and osmotic pressure. It also describes performance dependence on operating conditions and the need to avoid salt precipitation. These are useful process definitions, not a promise of identical performance across waters or equipment. DuPont technical manual
Start with a simplified accounting exercise. Suppose the hypothetical plant receives 100 units of solution mass in an interval, of which 3.5 units are dissolved salt and 96.5 are water. Suppose it removes 40 units as salt-free product, with no other inputs, outputs, evaporation or accumulated material. Sixty units of concentrate remain, carrying the original 3.5 units of salt. Its salt mass fraction is therefore 3.5 divided by 60, or approximately 5.83 percent. This example uses mass deliberately. Treating concentration in grams per liter as interchangeable with a mass fraction would introduce density assumptions.
If the same idealized feed instead yields 50 mass units of salt-free product, 50 units of concentrate remain and the salt fraction becomes 7 percent. The plant has produced more useful water and less liquid concentrate. It has not made the incoming salt disappear. Nor has the calculation shown that the equipment can achieve either operating point, that the product is genuinely salt-free, or that the resulting discharge is environmentally acceptable. It is a balance, not a design guarantee.
There is another comparison hidden inside this one. If the utility requires a fixed product quantity rather than accepting whatever a fixed intake produces, higher recovery can reduce the necessary intake. At 40 percent recovery, producing 40 mass units requires 100 units of feed. At 50 percent recovery, producing those same 40 units requires 80 units of feed, leaving 40 units of concentrate. At the assumed feed composition, the latter system brings in 2.8 units of salt rather than 3.5. A comparison must say which quantity was held constant.
This is why a statement such as “higher recovery reduces brine” needs explanation. It can mean less concentrate per unit of feed, less concentrate per unit of product, or a smaller absolute discharge from a plant whose output is unchanged. Each can be a sensible measure. They answer different questions about intake demand, plant capacity and receiving-water exposure. A proposal that switches silently between them can make an ordinary accounting change appear to solve a harder environmental problem.
Real balances must also include salt passing into the product, side streams, cleaning wastes and changing inventories. A measured discrepancy may be a sensor issue, an omitted outlet or a temporary storage change. The useful response is to investigate the discrepancy rather than adjust a spreadsheet until the totals agree. A small simplified balance is valuable because it teaches the reader what must be accounted for when the actual balance becomes complicated.
The electricity meter does not know the sales pitch
An energy figure becomes interpretable only when the reader knows what water crossed the boundary used in its calculation. Energy per cubic meter of feed differs from energy per cubic meter of product. Energy at the membrane train differs from electricity at the facility meter. Water delivered to a storage tank near the coast may require additional pumping before reaching a customer at elevation. Comparisons can be internally accurate yet misleading if their boundaries differ.
For illustration, suppose a fictional plant uses 1,000 kilowatt-hours over an interval and produces 250 cubic meters of product during that same interval. Its measured consumption across the chosen boundary is 4 kilowatt-hours per cubic meter. If a report divides those 1,000 kilowatt-hours by an assumed 500 cubic meters of intake instead, the result is 2 kilowatt-hours per cubic meter of feed. Neither arithmetic operation is wrong. Presenting the second as product-water efficiency would be wrong. The physical installation has not improved between the two calculations.
Researchers Karan Mistry and John Lienhard developed a framework for comparing separation technologies using thermodynamic efficiency in a 2013 paper. Its abstract explains why electricity, heat and fuel inputs require a consistent basis, and relates useful output to the least work required to separate a product from feed of specified composition. That is the relevant conceptual safeguard here: different forms and conditions of energy cannot simply be treated as equivalent because each is called energy. This article uses the authors' MIT repository abstract, not an independently reproduced evaluation of the paper's full model.
The hypothetical utility therefore asks bidders to present several numbers rather than one celebrated minimum. It wants the membrane system's electricity demand, total facility electricity, and additional delivery requirements reported separately, with an explicitly defined combined total. It asks whether production during commissioning, cleaning or off-specification operation is excluded from the denominator. It also asks for a sustained operating period rather than the best hour selected from a much longer test.
A pressure-energy recovery device adds another accounting question. The DuPont manual describes recovering energy from pressurized concentrate to assist pumping. That opportunity should be credited where it occurs, without assuming the recovered energy exceeds the energy originally supplied or covers every other load. In the utility's spreadsheet, recovery is represented as a reduction in net input within a defined system. The same recovered energy must not then be deducted a second time from the facility total.
Cost and emissions introduce different boundaries again. A lower electricity price does not necessarily mean fewer kilowatt-hours, and a lower-carbon electricity supply does not eliminate pumping work. A credible proposal can improve all three measures, but it should demonstrate each improvement separately. A community deserves to know whether a claimed advantage comes from physical efficiency, a favorable tariff, a particular accounting convention or an assumed future electricity mix.
Higher recovery moves the difficulty
Imagine two candidate operating plans for the hypothetical plant. One aims to minimize concentrate volume; the other accepts more concentrate while operating the membrane system less aggressively. The water balance helps identify their outputs. It cannot rank them by itself. The comparison must include pressure requirements, usable membrane life, treatment needs, downtime and the behavior of the receiving environment. A percentage printed in large type is only one coordinate in that larger decision.
Concentration also concerns individual constituents rather than a generic substance called salt. If two waters have the same total dissolved-solids measure but different chemical compositions, the treatment problem need not be identical. A design calculation that tracks total salt while overlooking the constituents governing precipitation may describe the water balance adequately and the operating risk badly. The utility should ask which constituents constrain the proposed recovery and what evidence supports those constraints.
The Bureau of Reclamation's project 20058, funded in fiscal years 2020 through 2022, illustrates one research direction. Its official description proposes an in-line mixer to promote precipitation and allow additional recovery from concentrate. The project page describes intended development and economic evaluation, with an inland-water emphasis. It does not provide a completed performance dataset on that page. Its ambitions therefore belong in a discussion of research possibilities, not in an assertion that a specific commercial seawater problem has been solved. Reclamation project description
For the hypothetical utility, that distinction changes procurement. A proposed treatment step can be offered a pilot with a clear hypothesis, feed specification and success criteria. It should not receive the same certainty in the base budget as an operation demonstrated on comparable water under comparable conditions. Conversely, a technology should not be dismissed simply because it is unfamiliar. The question is what evidence would move it from a promising option into a dependable part of the service commitment.
An informative pilot would report what happened to the removed material. If an operation produces a solid, the solid needs characterization, handling and an authorized destination. If it creates another liquid stream, that stream belongs in the balance. A reduction in one measured discharge volume can be useful while increasing another waste-management obligation. Describing the new obligation is part of demonstrating success, rather than an admission that the idea failed.
The operating comparison should also include departures from ideal conditions. A recovery setting that works with a favorable feed sample may be less attractive during a difficult season or after equipment has aged. The utility can ask for conditional operating envelopes: what changes when the feed changes, which measurements trigger a response, and how much production is then available? Such questions turn a single-point performance claim into an account of a system that people can operate.
Before water reaches the membrane
The intake creates a separate relationship with the coast. A plant's beneficial output does not erase the consequences of collecting its feed. Where water is drawn, how it enters, and what organisms encounter the intake are design questions in their own right. They also interact with treatment requirements. A project cannot responsibly decide that an intake is suitable solely because it can supply the desired hydraulic flow.
California's State Water Resources Control Board describes its seawater-desalination provisions as addressing site, design, technology and mitigation, following the May 2015 Ocean Plan update. The official program page also records a 2024 scoping process for potential amendments. A scoping process is not itself an adopted amendment. This article uses the agency's program description as evidence of the subjects under review and regulation, without presenting it as legal advice or exporting California's requirements to other jurisdictions.
One especially instructive primary document is the agency's May 12, 2025 memorandum responding to a subsurface-intake panel report submitted in April 2024. The memorandum criticizes aspects of the report's specificity and treatment of feasibility. It calls for stronger practical guidance and objects to reasoning that conflates challenges or costs with demonstrated infeasibility. The memorandum and attached report document a technical disagreement. They should not be summarized as an uncomplicated endorsement of every conclusion in the attachment.
For a reader, that disagreement is useful. It shows why “experts studied the issue” is an incomplete statement. Which experts? What question were they asked? What did the agency accept or criticize? A lengthy report can contain useful observations and still fail to answer the precise decision a public body needs to make. The existence of a PDF, even on an official website, does not eliminate the need to read its status and accompanying response.
In the hypothetical utility's comparison, each intake alternative would have a separate evidence table. The table would identify available site observations, unresolved geological or biological questions, assumptions about achievable flow, and the consequences of uncertainty. A favorable judgment at one site would remain evidence about that site. It would not become a rule that a similar-looking coast must behave the same way.
The team would also resist a false choice between environmental protection and engineering detail. A preference for a lower-impact arrangement needs practical investigation to become a working design. An assertion that such an arrangement is impossible needs evidence rather than repetition. Both the environmental objective and the feasibility question become clearer when the required observations are specified in advance and the decision-maker can see which gaps remain.
Concentrate is a composition, not a color on a diagram
A simplified process diagram often draws concentrate as a single blue arrow. That arrow conceals the question the receiving environment actually encounters: what is in the stream, in what amounts, and over what time? Average salinity is important, but it does not describe every constituent, every operating event or every possible interaction with another discharge. A monitoring plan needs a chemical inventory linked to the actual treatment train.
The hypothetical utility can begin with a deliberately mundane exercise. For each input material and each treatment step, it asks where the material can go. Does it remain dissolved, leave with a removed solid, pass into another stream, accumulate temporarily, or change chemically? The answer does not need to assume that every material is harmful. It needs to make the fate explicit enough that a laboratory program and an operating procedure can address it.
Cleaning events deserve a separate line in that inventory. A measurement taken during stable production cannot automatically characterize a stream generated during maintenance. If those streams are collected and managed separately, the project should show that arrangement and its capacity. If they enter a common treatment or disposal system, the assessment should describe the resulting conditions. A missing event can matter even when the plant's annual average looks uneventful.
A 2012 University of California, Davis study prepared for the State Water Board examined hypersalinity using nine toxicity-test protocols and seven organisms. It included different biological endpoints and tests using brine from the Monterey Bay Aquarium facility. The researchers designed the work to help distinguish salinity effects from additional constituents. This was a particular experimental study, not a universal safe-salinity certificate. Its methods and results demonstrate why the choice of organism, endpoint and test material matters when interpreting a toxicity finding.
That evidence suggests a useful question for the utility's proposed testing: what exactly would a negative result rule out? A result showing no measured effect in one test does not logically rule out effects on every species, life stage or exposure duration. Equally, an effect at an artificially severe exposure does not establish that the same effect occurs at a particular outfall. The bridge between laboratory conditions and environmental exposure must be explained rather than left to the reader's imagination.
Reporting should preserve that bridge. The concentration tested, the exposure period, the endpoint and the measurement uncertainty belong near the conclusion. A laboratory report can legitimately answer a narrow question. Trouble arises when its narrow answer is rewritten as a broad slogan. A useful public summary would make clear which risks have been investigated, which were not addressed by the chosen tests, and what field observations will complement them.
Dilution has a place and a time
Combining concentrate with other water lowers its concentration without destroying its dissolved material. This elementary distinction is easy to lose in discussions of outfalls. The location and speed of mixing matter because organisms encounter local conditions rather than a mathematical average over the entire ocean. A large receiving water body does not guarantee that a particular pocket of water exchanges rapidly with the rest of it.
Consider another idealized mass example, independent of the earlier plant calculation. One unit of concentrate at a 7 percent salt mass fraction mixes completely with nine units of seawater at 3.5 percent. The combined ten units contain 0.07 plus 0.315 units of salt, giving a 3.85 percent salt mass fraction. The result exceeds the original seawater concentration by 0.35 percentage points. This calculation assumes complete mixing and additive masses. It does not establish how far from an outlet that mixture forms or what environmental response follows.
That missing spatial information is the point. A drawing can show ten parts of water combined instantaneously while the actual flow contains gradients, changing currents and boundaries. A numerical dilution ratio only becomes meaningful when linked to a location, time scale and operating condition. A reader should ask whether the reported value is an immediate near-outlet estimate, a time average at a monitoring station, or a modeled value at a specified boundary.
A March 2012 science advisory panel report for the State Water Board distinguished discharge scenarios and recommended site-specific monitoring, including before-and-after observations and multiple reference locations. It emphasized validating modeled behavior after operation begins. The document explicitly states that its recommendations represent the panel's views rather than State Water Board policy. The panel report therefore supports the methodological concern here without establishing a present-day permit condition for any particular facility.
The hypothetical utility would ask its modelers to identify the observations that most strongly constrain their prediction. If a conclusion depends heavily on a poorly characterized current pattern, collecting better current observations may be more useful than running a more visually elaborate simulation. If a conclusion changes under plausible alternative assumptions, the range should appear in the decision document. A polished plume image should not conceal how much of its shape comes from an assumption.
The project also needs to state what happens when another discharge used for blending becomes unavailable or changes its flow. An arrangement dependent on a neighboring facility has an operational dependency, not merely a convenient source of dilution on paper. The response could involve storage, reduced desalination production, a different authorized operating mode or shutdown. Whatever it is, it should be designed before the dependency becomes an urgent problem.
A baseline is part of the plant
The first day of production is too late to begin learning what the surrounding environment looked like before production. A baseline gives future measurements something meaningful to be compared with. It also reveals how much natural variability the monitoring design must handle. Without that context, a later increase or decrease can invite confident stories that the data cannot distinguish.
For the hypothetical utility, consider three possible observations: salinity changes near the outlet, similar changes at a reference location, and a broader seasonal shift in the region. A monitoring design that observes only the outlet may attribute all three to the plant. A design that compares locations and time periods can ask whether the local change differs from the background change. The logic is comparative rather than magical; reference locations themselves must be chosen with care.
An example makes the distinction concrete. Suppose an ecological indicator falls by ten units near a discharge and by nine units at reference sites during the same period. The plant could still have contributed to the local outcome, but the raw ten-unit decrease is not automatically its effect. Conversely, a stable regional average could hide a local change concentrated close to the outlet. These are hypothetical values illustrating inference, not results from a desalination facility.
Measurement quality also shapes the inference. Sensors can drift, samples can be mislabeled, and gaps can cluster during difficult weather or equipment failures. A dataset with many entries is not necessarily representative of the conditions of greatest concern. The monitoring plan should say how instruments are checked, how missing observations are recorded, and how samples remain connected to their time, location and operating state.
Public access should include enough context to make the data interpretable. A column labeled salinity needs units or the defined measurement scale, location, depth, time and quality flags. A plotted average needs its averaging interval and treatment of missing values. Corrections should be documented rather than silently replacing an earlier series. These are ordinary data-governance choices, but they determine whether disagreements can be investigated using a shared record.
A monitoring contract should also fund interpretation. Collecting samples without a plan for evaluating them can produce an expensive archive and little operational guidance. The utility needs qualified people to examine trends, revisit assumptions and recommend additional investigation when the original design proves inadequate. Monitoring is therefore an ongoing component of the water supply, with staffing and budget implications, rather than an appendix purchased once to complete an approval process.
An alarm needs an operating consequence
A public commitment to monitor sounds reassuring until someone asks what will happen after a concerning result. The answer should be more specific than “we will review the data.” Some observations may require immediate operational attention; others may justify confirmatory sampling or a longer investigation. The distinction depends on the parameter, confidence in the measurement, applicable requirements and the consequence of delay.
The hypothetical utility can describe a response ladder without inventing a universal numerical threshold. A quality-control problem first calls for checking the measurement while preserving the original record. A confirmed departure from an operating envelope calls for a predefined operating response and notification through the appropriate channels. An unexplained ecological pattern may require expanded investigation even if a routine chemical measurement remains within its expected range. These different cases should not collapse into one generic alarm.
The ladder needs authority as well as instructions. Who can reduce production? Who receives the information outside normal business hours? What happens if the operator and environmental reviewer disagree? A responsibility chart with named roles and alternates is more useful than a promise that all stakeholders will collaborate. Independence matters when the decision to protect a receiving environment also reduces a facility's revenue or water output.
Production planning must accommodate those decisions. If every financial forecast assumes uninterrupted full operation, the utility has implicitly treated its environmental response commitments as events that never occur. A more candid forecast includes operating constraints and contingency supply. The point is not to assume failure. It is to avoid designing a water system that can honor its safeguards only by creating an immediate supply emergency.
Some responses will change as evidence improves. A model may overpredict one effect and underpredict another. A monitoring location may prove poorly chosen. Revising a program can be responsible, provided the revision and its justification are visible. The utility should distinguish evidence-led changes from changes made merely because a measurement became inconvenient. Keeping the original rationale and the new evidence in the public record helps outsiders evaluate that distinction.
This approach gives a community something more durable than a declaration that the plant is safe or unsafe in the abstract. It creates a way to identify problems, assign responsibility and adjust operations. The quality of that arrangement can be assessed before the plant starts. Its actual performance can then be examined over time, using observations rather than confidence alone.
Selling the salt requires a second business case
Resource recovery offers an appealing alternative story for concentrate: instead of paying to manage a residual stream, sell valuable materials from it. The attraction is reasonable. A material leaving one process can sometimes become another process's input. But value depends on what is recoverable at the required purity, in a useful form, with a reliable buyer and an acceptable production cost.
The hypothetical utility should separate a laboratory separation result from a market transaction. A test that identifies or concentrates a constituent does not establish the expense of continuous recovery, the handling of impurities, product qualification or long-term demand. The proposed buyer may require a specification that changes the processing burden. A high quoted market price may apply to a purer product or a much smaller sales volume than the plant could produce.
A deliberately simple calculation shows the fragility of gross-revenue claims. Suppose a fictional recovery process could sell 100 units of product for 2 currency units each, generating 200 units of revenue. If producing, qualifying and delivering those units costs 250, the operation loses 50 before considering any avoided disposal expense. If avoided disposal is genuinely worth 80, the combined change could instead be favorable by 30. Both the expense and the avoided expense must be evidenced; neither can be inferred from the product's selling price alone.
There is a timing issue as well. The water service may need to operate continuously while the materials buyer purchases irregularly. Storage then becomes part of the project. If the buyer declines a batch, the utility still needs a destination for it. A binding sales arrangement can reduce that risk, but it should be examined for specifications, volumes, rejection rights and what happens if the buyer stops operating.
The same discipline applies to claims of zero liquid discharge. Even if a particular arrangement eliminates a liquid outlet within its defined boundary, it still requires a complete account of energy, solids and other outputs. The phrase describes a boundary condition, not the disappearance of matter or responsibility. A community comparing options should ask for the whole material flow and operating cost, rather than treating the phrase as an automatic environmental ranking.
A sensible development path can preserve the opportunity without relying on it prematurely. The water project can evaluate a recovery pilot while keeping its core financial case viable without speculative revenue. If the pilot demonstrates consistent quality, real sales and a defensible net benefit, the business case can be updated. That sequence rewards evidence and avoids making dependable water delivery contingent on an unproven commodity business.
Buying a measurement rather than a promise
Before committing to construction, the hypothetical utility can turn unresolved claims into a limited set of informative tests. The order matters. A test is most valuable when its result could change the choice between alternatives. Measuring an easily observed quantity repeatedly may create confidence without resolving the uncertainty that determines feasibility. A difficult but decisive observation can deserve more attention than a large volume of routine data.
Suppose two candidate designs differ mainly in how much stable production they can sustain during a particular feed condition. A pilot conducted only during favorable conditions cannot resolve that difference, however precise its instruments. The test plan should identify the condition of interest, explain whether it can be observed or represented credibly, and distinguish direct measurements from extrapolation. If the relevant condition never occurs during the study, that absence belongs in the conclusion.
There is also a distinction between repeatability and representativeness. Repeated trials using the same controlled feed can show that an apparatus behaves consistently under those conditions. They do not automatically show how it responds to the variability of an actual intake. Conversely, an uncontrolled field trial may encounter realistic variability while making it difficult to identify why performance changed. A useful program can combine controlled and field observations without pretending they answer exactly the same question.
The utility should specify the analysis before seeing the most flattering result. That includes the operating periods to be counted, the handling of failures, and the conditions under which a test is considered inconclusive. A failed pump, interrupted sample series or unusable product batch may be relevant operational evidence rather than an inconvenience to remove from the dataset. Exclusions can be legitimate, but their reasons should remain visible and consistent.
Costs deserve the same treatment as physical measurements. An estimate should identify the price year, assumed utilization, major replacements, energy assumptions and financing boundary. Different bidders may use different assumptions even when they quote the same unit of water cost. The utility can request a common comparison case alongside each bidder's preferred case. That does not eliminate uncertainty; it prevents hidden assumptions from masquerading as superior equipment.
A sensitivity exercise then asks which assumptions have the largest effect on the decision. If a modest change in electricity price reverses the ranking, the choice is exposed to that price. If the ranking is stable across the tested range but reverses when expected utilization changes, demand and operating strategy deserve closer scrutiny. These are hypothetical analytical questions, not forecasts of prices or demand. Their purpose is to reveal dependence, not to predict one exact future.
Finally, the utility should retain the option to decline an attractive proposal whose decisive evidence remains unavailable. The absence of evidence need not mean the underlying technology cannot work. It can mean the project has not yet earned the level of confidence required for that particular public commitment. A further pilot, a smaller deployment or a revised design may be the appropriate next step. Treating uncertainty as a decision variable makes that intermediate outcome legitimate, rather than forcing every promising idea into an immediate approval or permanent rejection.
The public decision is about a service under conditions
The hypothetical utility now has a different decision document from the one it began with. Instead of a single production figure and a celebrated energy minimum, it has linked accounts: water and salt, energy and delivery, intake evidence, concentrate composition, discharge behavior, monitoring, operational responses and financial dependencies. The accounts do not make the decision effortless. They make disagreement more precise.
A supporter can identify a demonstrated advantage without claiming that every issue has been settled. A critic can point to an unresolved measurement or fragile assumption without having to argue that desalination is impossible everywhere. An engineer can explain the conditions under which an option works. A public official can see which conditions would require a revised budget, a different design or a different supply strategy.
The comparison should include the actual service the utility needs. A facility designed for steady output may answer a different need from one expected to provide occasional drought backup. Financing, staffing and equipment readiness still matter during periods of low production. A plan to operate intermittently needs evidence about that operating pattern, rather than borrowing a continuous-operation performance figure and assuming it transfers unchanged.
Likewise, a larger nominal capacity is not automatically a larger dependable contribution. If an intake constraint, discharge condition or maintenance requirement limits operation during a critical period, that limitation belongs in the supply assessment. The useful quantity is water available under the conditions the community must plan for. A capacity label can be accurate while giving an incomplete picture of that contribution.
The practical standard is therefore demanding but intelligible. Show the balances. Keep assumptions attached to the calculations. Identify the status and date of the supporting evidence. Explain what remains uncertain and what observation would change the decision. Fund the people and instruments needed to learn after commissioning. Give protective commitments a real operating consequence and a financially credible contingency.
Desalination can be evaluated without either treating the ocean as a limitless disposal space or treating every concentrate stream as an identical ecological disaster. The second pipe forces attention to place, time and composition. Following it carefully reveals why water production, energy efficiency and environmental performance must be judged together. A useful fresh-water supply is the entire system that makes delivery possible while managing what it leaves behind.
<!-- ARTICLE_BODY_END -->