Why MLD and ZLD projects fail

Failure often begins in framing, not equipment.

Controlled principle Most MLD/ZLD failures are not isolated equipment defects. They are propagated decision errors: an incomplete feed basis, an undefined endpoint, a chemistry or solids constraint that was deferred, a commercial assumption presented as certainty, or an operating burden that the owner was never equipped to carry. The first failed component is often only where the earlier error becomes visible.

Failure begins upstream of the failed component

A useful failure review starts by separating the visible event from the earliest unsupported assumption that made the event consequential. A membrane may foul, a crystallizer may lose capacity, a centrifuge may blind or a storage tank may fill; none of those observations alone establishes the project root cause. Chapter 35 therefore hands this chapter operating evidence—trends, interventions, downtime, corrosion findings, residual bottlenecks and commercial fallback use—rather than a retrospective label of “operator error.” [BC-35]

Current whole-train research reaches the same systems conclusion. ZLD/MLD performance depends on interactions among pretreatment, concentration, crystallization and final residual management; optimizing one block does not optimize the treatment train. The failure question is therefore: which decision or interface allowed a local deviation to become a service, cost, compliance or commercial failure?

Failure layer	Typical initiating error	Visible symptom	What the review must test	Corrective level
Framing	Problem, endpoint or boundary not defined	A technically impressive block solves the wrong duty	Was water, liquid residual, solids and product destination defined before selection?	Reframe objective and boundary
Characterization	Single or non-representative analysis	Unexpected scaling, fouling, yield or chemistry shift	Did the design basis represent normal, design and upset distributions?	Resample and rebuild basis
Chemistry	TDS used as the main descriptor	Scale, corrosion, precipitation or incompatibility	Were ion loads, speciation, additives, temperature and concentration effects modelled/tested?	Redesign chemistry and pretreatment
Integration	Recycle, purge or owner interfaces omitted	Accumulation, bottlenecks, hidden owner scope	Did every stream and interface close under steady and upset states?	Close balances and interfaces
Solids/residuals	Separation proven but solids handling not proven	Storage fills, cake is wet, disposal route fails	Was throughput, dewatering, off-spec and fallback capacity demonstrated?	Redesign solids/residual system
Operating model	Capability assumed rather than designed	Manual workarounds, late detection, long recovery	Can owner staff detect, intervene, maintain and restart within the availability basis?	Change design, staffing, automation or support
Commercial/contract	Revenue, guarantees or exclusions are optimistic	Treatment depends on buyer/vendor/owner not under control	Does the water/residual obligation survive no-product and guarantee-exclusion cases?	Reallocate risk or simplify project

Table 36.1. Failure should be traced from the visible event back to the earliest decision, evidence or interface that allowed it to propagate.

Figure 36.1. Failure tree. Framing, design-basis and integration errors propagate into operating symptoms and project outcomes; the visual is a causal decision map, not a frequency model.

A single analysis converts variability into a design surprise

A single laboratory sheet cannot establish a design feed. ISO 5667-1 treats sampling-programme design as part of obtaining fit-for-purpose water data, while ISO 5667-3 addresses preservation, handling, transport and storage because composition can change between collection and analysis. For industrial wastewater, the additional problem is process variability: production campaigns, cleaning events, reagent changes, recycle conditions and upstream operating modes can create chemistry that never appears in a convenient grab sample.

The failure mechanism is predictable: a nominal analysis is converted into one simulation, one equipment duty and one guarantee envelope; the plant then sees a distribution. The correct basis is not “more samples” in the abstract. It is a sampling plan tied to the operating states that can control scale, corrosion, recovery, product quality or residual generation.

Weak evidence	What it hides	Operational consequence	Required correction
One grab sample	Temporal and campaign variability	Off-design chemistry becomes routine	Map production states and sample each decision-relevant state
Average only	Short-duration peaks and low-flow concentration	Constraint appears only during peak or upset	Carry normal, design and upset cases
Old sample / poor preservation	Precipitation, volatilization, oxidation or biological change	Laboratory matrix differs from plant matrix	Control preservation, holding time and sample condition
Incomplete ion suite	Charge imbalance and missing scaling species	Speciation and salt balance cannot close	Complete analytical basis and charge/mass checks
No additives history	Antiscalant, coagulant, biocide or CIP residuals	Unexpected fouling, product contamination or reaction demand	Map all routine and non-routine chemical inputs
No flow/time basis	Concentration is mistaken for annual load	Equipment, storage and product scale are mis-sized	Pair concentration with flow distributions and operating hours

Table 36.2. Data failure occurs when sampling does not represent the operating states that control the design.

Equipment-before-endpoint creates a local optimum

A project can buy a technically capable concentrator and still fail because the last liquid, solid or product stream has no acceptable destination. Equipment-first development reverses the required decision sequence. The endpoint should determine how much concentration is valuable, which species must be removed, what residual form is acceptable, and whether ZLD is necessary at all. [BC,1]

Equipment-first statement	Missing endpoint question	How failure appears	Better framing
“Maximize RO recovery”	What is the allowable concentrate endpoint?	Scaling control and pretreatment expand without clear value	Optimize recovery against residual route and whole-train burden
“Install a crystallizer”	What solid quality and disposal/offtake route is required?	Mixed salt or wet solids have no routine destination	Define solid endpoint, moisture, impurity and fallback first
“Recover every mineral”	Which material has a qualified user and positive delivered netback?	Process complexity grows faster than bankable value	Screen product, market, scale, legal route and logistics first
“Zero liquid means zero waste”	Where do sludge, purge, mother liquor and cleaning waste go?	Liquid liability is shifted into unmanaged solids or intermittent waste	Track every routine and non-routine stream to an endpoint

Table 36.3. Equipment selection cannot compensate for an undefined water, liquid-residual, solid or product endpoint.

TDS-only design hides the chemistry that actually fails

TDS describes the aggregate dissolved load but not the reactions that control treatability. Two streams with similar TDS can have very different calcium, sulfate, silica, alkalinity, magnesium, organics, oil, metals, additives and temperature—and therefore very different saturation, fouling, corrosion, precipitation and product-quality behaviour. Brine-composition reviews also emphasize that operating and seasonal variability changes these constraints.

TDS-blind omission	Why it matters at concentration	Possible symptom	Evidence needed
Calcium + sulfate	Ion activity and supersaturation rise as water is removed	Gypsum/anhydrite deposition, plugging, seed burden	Ion-specific balance, speciation/saturation and real-matrix test
Silica / silicate chemistry	Solubility and polymerization depend on pH, temperature and matrix	Hard-to-clean scale, declining heat/mass transfer	Dissolved/colloidal silica basis, pH-temperature path and testing
Alkalinity / carbon chemistry	pH shift and CO2 transfer change carbonate saturation	Unexpected carbonate precipitation or reagent demand	Alkalinity, inorganic carbon, gas transfer and pH control basis
Organics / oil / polymers	Concentration and thermal history can change fouling behaviour	Membrane fouling, foaming, crystal contamination	Fractionated organics/oil, additives and pilot exposure
Chloride + temperature + pH	Materials exposure changes along the train and during cleaning	Localized corrosion, leaks, contamination	Materials/corrosion review under process and cleaning envelopes
Trace metals / impurities	Species can concentrate into product or residual phases	Off-spec product or hazardous residual	Detection limits, partitioning and product/residual analyses

Table 36.4. TDS is a mass descriptor; failure mechanisms are controlled by ion-specific chemistry, contaminants, temperature and operating history.

Scaling, silica, sulfate and corrosion are coupled risks

Scaling risk intensifies as concentration changes activity, supersaturation, temperature and local surface conditions. Modern ZLD/MLD literature treats mineral scaling as a central limit for membrane-based brine concentration, while conventional thermal systems also rely on pretreatment, materials selection and operating controls to keep heat-transfer surfaces and crystallization loops functional.

Corrosion cannot be separated from scale control. Acidification, caustic operation, oxidants, chloride concentration, temperature and cleaning chemicals alter the materials exposure at the same time that they change scaling and fouling. A flowsheet that solves saturation on paper but creates an unqualified metallurgy or cleaning envelope has merely moved the failure mechanism.

Mechanism	Precursor that should be visible	Common design error	Failure consequence	Control evidence
Sulfate scale	Rising Ca/SO4 load, saturation margin, deposition trend	Assuming average chemistry or antiscalant closes all cases	Pressure drop, heat-transfer loss, plugging, cleaning	Speciation + real-matrix concentration/pilot evidence
Silica/silicate	Silica trend, pH-temperature path, recovery dependence	Treating silica as one fixed solubility number	Irreversible/slow-cleaning deposition and flux loss	Matrix-specific testing and cleaning/recovery evidence
Carbonate/hydroxide	pH, alkalinity, Ca/Mg and reagent control	Ignoring local pH or overdosing effects	Scale, sludge burden, poor separation	Reaction model plus kinetics/solid-liquid testing
Corrosion	Potential, temperature, chloride, pH, oxidants, inspection data	Selecting material from nominal feed only	Leaks, contamination, unplanned replacement	Materials envelope including CIP/upset conditions
Mixed deposit	Simultaneous inorganic/organic/solid exposure	Testing pure salts or synthetic water only	Deposit behaves unlike modelled single-phase scale	Real-feed long-duration test and deposit analysis

Table 36.5. Scaling and corrosion should be controlled as interacting chemistry, materials and cleaning problems—not isolated saturation-index checks.

Antiscalant overconfidence is not a pretreatment strategy

Antiscalants can be useful tools, but their effect is chemistry-, dose-, mineral- and process-specific. Research on high-recovery membrane systems shows that scaling remains a design constraint and that pretreatment choices such as antiscalant addition or ion-selective removal change both recovery potential and downstream disposal duty. DuPont likewise treats feed composition, pH, temperature, recovery and concentration polarization as factors that influence RO performance and recommends normalization to distinguish expected shifts from deterioration.

The failure is not “antiscalant does not work.” It is using antiscalant as permission to omit the precipitation pathway, competing scales, solids route, cleaning burden or upset case. A chemical programme should sit inside a control strategy with validated operating limits and an alternative when those limits are exceeded.

Overconfidence pattern	Why it fails	Required proof	Fallback if margin disappears
One vendor dose at one analysis	Feed and concentration path vary	Dose-response evidence over representative chemistry and recovery	Lower recovery, condition feed or remove limiting ion
Saturation index treated as guarantee	Kinetics, surfaces and mixed deposits are not captured by one index	Pilot/field evidence plus trend monitoring	Trigger condition-based intervention before irreversible loss
No antiscalant residual/product check	Polymer/reagent can partition to solids, product or mother liquor	Residual/product analysis and qualification	Destroy/remove additive or accept non-product residual route
Cleaning assumed to reset performance	Deposit may not be fully reversible and cleaning creates waste	Demonstrated restoration, waste volume and material compatibility	Redesign pretreatment or replacement philosophy

Table 36.6. Antiscalant is one control layer; it should not substitute for chemistry, solids, monitoring and fallback design.

Mother liquor turns a “closed loop” into an accumulation loop

A crystallizer does not automatically create a chemically closed system. Species that do not enter the desired crystal phase remain in solution and can accumulate through recycle. Reagent ions, trace contaminants, organics, antiscalants and off-spec fines may progressively change viscosity, boiling behaviour, crystal purity, corrosion, foaming and downstream product acceptance. The master therefore requires explicit mother-liquor recycle and controlled purge logic. [BC]

Accumulating species / property	Where it enters	What it can change	Required control	Failure signal
Non-crystallizing ions	Feed + reagent additions	Ionic strength, phase behaviour, purge volume	Steady-state salt balance and purge setpoint	Model closes only with zero purge
Organics / antiscalant	Feed and chemical programme	Foaming, nucleation, crystal surface contamination	Long-duration recycle test and product analysis	Pilot purity declines with cycles
Trace contaminants	Feed, corrosion, reagents	Product specification or residual classification	Partitioning analysis and controlled bleed	On-spec early batches become off-spec later
Fines / suspended solids	Nucleation and attrition	Viscosity, filtration, heat transfer and carryover	Particle-size, classification and solids-removal strategy	Slurry density or filterability drifts
Water imbalance	Wash water, condensate losses, cleaning	Concentration and equipment duty	Full water balance including intermittent streams	“Closed loop” requires unexplained water removal/addition

Table 36.7. A recycle loop requires an impurity-accumulation model and a controlled purge; otherwise “closure” is only hydraulic.

Weak solid-liquid testing turns chemistry success into a handling failure

Precipitating the right mineral is not enough. The plant must continuously separate, wash, dewater, store and move the resulting solids at the required throughput. Laboratory tests that report removal efficiency but omit settling, filtration, cake moisture, particle-size distribution, rheology and wash demand cannot support a full-scale solids-handling design. Chapter 33 therefore required integrated precipitation, crystal-growth and solid-liquid-separation evidence before scale-up. [BC-33]

Evidence gap	Why bench chemistry can look successful	Scale-up failure	Minimum evidence before design freeze
No kinetics / ageing	Equilibrium removal appears high	Residence time or downstream precipitation is wrong	Time-resolved precipitation and maturation basis
No particle-size distribution	Total mass is correct	Settling, filtration and centrifuge performance are unpredictable	PSD across representative runs and recycle
No dewatering test	Dry-solids calculation looks simple	Wet cake mass, hauling and storage multiply	Cake moisture, filtration/centrifuge cycle and wash demand
No recycle solids exposure	Fresh batch is clean	Seeds/fines/impurities alter nucleation and separability	Continuous/cyclic test with representative recycle
No off-spec route	Every batch is assumed useful	Rejected solids stop the water plant	Quarantine, rework, storage and compliant disposal capacity

Table 36.8. Solids-management evidence must prove throughput and separability, not only precipitation yield.

Assumed salt sales can make an uneconomic residual look profitable

A recovered salt is not automatically a product. Previous chapters require specification, repeatable quality, lawful status, a qualified internal user or buyer, viable logistics and risk-adjusted netback before product revenue enters the base case. This matters because current brine-valorization reviews still identify scale-up, material performance, energy and economic viability as major constraints, with large-scale recovery concentrated mainly in water and bulk salts.

Failure occurs when a treatment obligation is sized around sales that have not been qualified. Product rejection then becomes a water-plant problem: storage fills, crystallizer inventory grows, the buyer outage becomes a process constraint, and the project discovers that the “product” has an ordinary disposal cost.

Commercial assumption	What must be demonstrated	Failure if assumed	Bankable treatment in model
All recovered salt is saleable	Specification yield and on-spec fraction	Off-spec inventory accumulates	Credit only qualified on-spec volume
Commodity price = project revenue	Delivered grade, contract, freight and rejection terms	Netback collapses after logistics/conditioning	Use delivered netback or avoided purchase
One buyer will absorb output	Demand, delivery profile, credit and outage plan	Receiver outage stops production	Multiple outlets, storage or full residual fallback
Waste can become product by naming it	Jurisdiction-specific legal pathway and buyer acceptance	Material remains regulated waste	Retain compliant waste route until status is established
Product train is essential to economics	No-sales/no-revenue case survives	Water project becomes hostage to market	Separate treatment obligation from upside product case

Table 36.9. Product revenue is an evidence-gated upside; it should not conceal the residual fallback obligation.

Missing residual costs do not disappear—they return as operating constraints

Whole-train analysis consistently shows that disposal choices interact with pretreatment, concentration and crystallization. A business case that prices the concentrator but omits sludge, mixed salt, purge, off-spec product, cleaning waste, transport, storage and contingency capacity is not incomplete by a small margin; it has omitted the physical endpoint of the process.

Residual	Routine quantity basis	Cost / capacity drivers	Downside case	Failure signal
Pretreatment sludge	Stoichiometry + impurities + moisture	Dewatering, reagents, bins, transport, disposal	Higher dose / poorer dewatering	Sludge storage constrains feed
Mother-liquor purge	Steady-state accumulation balance	Hazard, transport, treatment/disposal route	Higher impurity load / no product route	Purge reduced to protect economics
Mixed crystals / salt	Salt mass + retained liquor + on-spec fraction	Drying, packaging, landfill/offtake, dust/corrosion	Buyer rejection / wet product	Crystallizer cannot dispatch solids
Cleaning waste	CIP/cleaning volume, chemicals and frequency	Neutralization, storage, treatment and scheduling	Cleaning frequency increases	CIP delayed because waste route is full
Emergency liquid	Trip/restart volume and duration	Tankage, fallback transfer and permit route	Long outage / simultaneous upstream flow	Plant can fail only by overflowing

Table 36.10. Residual costs and capacities must be tied to physical quantities, moisture, frequency and compliant destinations.

Operators inherit every assumption that engineering did not close

Operations failure is often blamed on people after the project has already imposed an unrealistic operating model. IEC 60300-3-10 treats maintainability and maintenance as lifecycle activities linked to reliability and availability, while IEC 60300-3-14 treats supportability as a planned balance of performance, cost and risk. [7,8] For MLD/ZLD, that means access, isolation, diagnostics, laboratory response, specialist skills, spare lead time, cleaning duration and restart procedures belong in design and procurement—not in post-startup improvisation.

Design assumption	What operators actually need	Failure mode when absent	Design/procurement control
“Automation will handle it”	Reliable instruments, validated logic, alarm priorities and manual fallback	Bad data or alarm flood delays intervention	Critical-tag register, proof tests and degraded-mode logic
“Routine maintenance”	Access, isolation, lifting, wash-down, waste route and staffing duration	Task is deferred or takes far longer than availability model	Maintainability review and executable job plan
“Vendor support available”	Named response time, data access, tools, software and escalation	Minor fault becomes extended outage	Support SLA, data rights, owner diagnostic capability
“Spares are standard”	Criticality, installed base, lead time, storage and obsolescence plan	Single component creates long shutdown	Strategic/operating spare register and replenishment basis
“Operators can restart”	Preservation, sequence, chemistry stabilization, residual capacity and acceptance checks	Repeated trips or long unstable recovery	Demonstrated shutdown/restart procedure

Table 36.11. “Operator error” should not be used where the design assumed unavailable information, access, competence, spares or support.

Guarantees fail when the feed, boundary and exclusions do not match operation

A guarantee protects the owner only when the performance metric, feed envelope, battery limits, test method, duration, exclusions and remedies are measurable on the same basis as the service obligation. Chapter 34 established this procurement rule; Chapter 35 then converted it into operating tasks. A project failure occurs when the plant is routinely outside the guaranteed basis or when owner-supplied pretreatment, chemicals, residual disposal or product offtake are the unpriced conditions that make the guarantee possible. [BC-35]

Guarantee weakness	What it looks like in operation	Why it fails	Required correction
Narrow feed window	Normal plant chemistry is “out of spec”	Guarantee excludes routine variability	Align design/guarantee envelope with representative feed distributions
Peak metric only	Acceptance test passes; sustained duty drifts	Duration, cleaning and availability are not guaranteed	Include normalized sustained performance and operating-state basis
Owner scope hidden	Vendor performance depends on owner systems not in price	Interface failure has no accountable party	Battery-limit and interface responsibility matrix
Residual/product exclusion	Water production assumes uninterrupted disposal/offtake	Third-party outage disables treatment	Fallback route and risk allocation independent of sales case
No measurable remedy	Failure is debated rather than corrected	Metric or test method is ambiguous	Pre-agreed test protocol, tolerances, data rights and remedy path

Table 36.12. Guarantees become operational only when the routine feed, complete boundary, test method and remedy are aligned.

Downtime is often an integration failure rather than a component MTBF problem

Water-service availability can be eroded by many small losses that sit outside the “main” technology package: chemistry excursions, additional cleaning, solids-handling delays, unavailable residual capacity, slow restart, common utilities, laboratory release, a blocked product outlet or a specialist-support dependency. IEC dependability guidance is useful precisely because reliability, maintainability, supportability and availability interact rather than exist as separate vendor numbers.

Figure 36.2. Scenario-only service-erosion graph. Starting from 100 normalized service units, illustrative losses of 8, 5, 4 and 3 units show how interface burdens can reduce delivered service to 80 units. The values are explicit assumptions, not empirical failure rates or availability benchmarks.

Downtime event	Primary cause class	Secondary dependency	What should be recorded	Design decision informed
Scale/fouling clean	Chemistry / pretreatment	CIP waste, chemicals, labour, restart	Trigger, duration, restoration and residual waste	Pretreatment, recovery, cleaning philosophy
Solids system stop	Solids / mechanical	Storage, transport, wash water, upstream hold	Throughput, inventory, moisture and recovery time	Redundancy, buffer and dewatering capacity
Corrosion repair	Materials / chemistry	Isolation, spare material, contamination	Location, mechanism, exposure and repair	Material upgrade and inspection interval
Instrument/control trip	Automation / measurement	Bad data, interlocks, vendor support	Tag quality, alarm sequence, false/true trip	Sensor redundancy and degraded-mode control
Receiver/disposal outage	Commercial / residual	Storage and alternate route	Inventory growth, time-to-limit, fallback use	Independent residual capacity and contract structure
Restart delay	Procedure / integration	Chemistry stabilization, heat-up, product release	Safe-state duration and step that controlled restart	Procedure, automation, storage and staffing

Table 36.13. Downtime records should identify the interface that extended loss of service, not only the component that tripped first.

A failure review should change the project—not merely assign blame

The output of a failure review should be a controlled change to the evidence, design, operating model or commercial basis. A root-cause statement that cannot change a decision is not enough. The review should distinguish immediate cause, enabling conditions and systemic cause; quantify service and residual impact; test whether the same mechanism exists elsewhere; and then update the basis of design, procedures, spares, monitoring, training, guarantees or economics through management of change.

Review field	Minimum record	Weak answer	Decision output
Event definition	Service lost, duration, stream state, environmental/product impact	“Unit failed”	Common event boundary and consequence
Immediate cause	Observed physical/control mechanism with evidence	Component name	Repair action and verification
Enabling condition	Why protection, detection or fallback did not contain the event	“Operator did not respond”	Control, alarm, access, spare or procedure change
Systemic cause	Earlier assumption, interface or governance gap	“Unexpected chemistry”	Design-basis, procurement or commercial correction
Recurrence check	Other equipment/streams exposed to same mechanism	Only failed asset inspected	Fleet/train-wide inspection or model update
Change control	Owner, due date, test, documentation and acceptance	Informal workaround	MOC, updated SOP/model/drawing and closure evidence

Table 36.14. A failure review is complete only when it changes the controlled project basis and verifies the corrective action.

Failure-prevention acceptance gate

Before a project leaves Part VII, the owner should be able to show that the recurring failure mechanisms in this chapter are either designed out, bounded by monitored operating limits, or provided with a finite fallback. The gate is not a promise that the plant will never fail; it is evidence that foreseeable failure does not automatically become loss of the water/residual obligation.

Gate field	Proceed	Conditional / recycle	Stop	Required record
Feed / variability	Representative normal/design/upset basis is controlled	One non-fatal state needs closure testing	Guarantee/design still rests on one nominal analysis	Sampling and design-basis register
Chemistry / materials	Scale, fouling, corrosion and cleaning envelopes are demonstrated	Bounded uncertainty has monitor/trigger plan	Critical mechanism has no validated control or compatible material	Chemistry/materials risk register
Recycle / mother liquor	Accumulation and purge balances close	Pilot duration leaves one bounded scale-up uncertainty	Closed-loop claim depends on zero/unpriced purge	Recycle and purge mass balance
Solids / residuals	Routine, off-spec and upset routes have finite capacity	One contingency route is time-limited but controlled	Water service depends on indefinite storage or assumed salt sale	Residual-capacity and fallback plan
O&M / support	People, monitoring, access, cleaning, spares and restart are executable	Early-life support closes a defined competence gap	Availability depends on unavailable skills/spares/vendor access	O&M readiness and support plan
Guarantees / commercial	Measured guarantee and no-revenue fallback preserve service	One commercial allocation awaits bounded closure	Routine operation falls outside guarantee or buyer outage stops treatment	Guarantee, interface and risk-allocation matrix
Learning / failure control	Events, deviations and MOC update the controlled basis	Data integration improvement remains open	Recurring workaround is accepted as normal operation	Failure/MOC register

Table 36.15. Failure risk is controlled only when data, chemistry, residuals, O&M and commercial dependencies have observable limits and fallback routes.

Stop rule Stop, simplify or re-stage when the project still depends on one analysis, an undefined residual endpoint, TDS-only chemistry, an untested scale/corrosion mechanism, zero mother-liquor purge, unproven solids handling, assumed salt sales, omitted residual costs, unavailable operator/support capability, or a guarantee that excludes routine operating conditions. Do not use a more complex flowsheet to hide an earlier unresolved failure mechanism.

Handover to Chapter 37 — power and cooling systems

Chapter 37 begins the sector playbooks with power and cooling systems. It should not restart from a generic ZLD flowsheet. It should receive the failure controls developed in Part VII and apply them to cooling-tower blowdown, boiler blowdown and flue-gas-desulfurization wastewater, where utility integration, water-reuse value, silica/hardness/sulfate chemistry and the residual route can determine the practical MLD/ZLD endpoint. [BC]

The handover should therefore identify the sector-specific reason to act, operating and chemistry distributions, pretreatment constraints, concentration limits, solids/residual route, owner utility interfaces, operability basis and the failure mechanisms that require explicit validation. The sector chapter can then show how the common decision framework changes when the industrial process—not the brine technology—sets the boundary.

Handover field	Required basis for Chapter 37	Failure lesson carried forward
Source stream / duty	Cooling-tower, boiler, FGD or other defined stream; normal/design/upset flow	Do not merge chemically different streams without testing the benefit
Utility context	Make-up water value, steam/electricity, heat integration, site outages	Energy and water value are plant-system variables, not package constants
Chemistry distribution	Silica, hardness, sulfate, chloride, alkalinity, organics/metals/additives as relevant	Do not design from TDS or one analysis
Pretreatment / concentration envelope	Constraint-removal logic, recovery limits, cleaning and materials basis	Antiscalant or vendor recovery claim does not replace evidence
Solids / residual endpoint	Sludge, concentrate, crystals, purge and fallback capacity	ZLD is not achieved by creating an unmanaged solid stream
O&M / availability	Operator tasks, monitoring, cleaning, spares, emergency and restart basis	Sector solution must fit owner capability and outage philosophy
Economics / commercial	Avoided water/disposal, full residual costs, no-product-revenue case	Salt/product upside cannot carry the essential water obligation

Table 36.16. Chapter 37 should apply the common failure controls to the specific utility, chemistry and residual realities of power and cooling systems.

Chapter conclusion

MLD and ZLD projects rarely fail for one reason. A visible membrane, evaporator, crystallizer, centrifuge or corrosion event often sits at the end of a longer chain: the wastewater was characterized too narrowly; the endpoint was not defined; TDS substituted for chemistry; scale, silica, sulfate or materials exposure was underestimated; recycle accumulation and solids handling were not proven; residual costs were omitted; product revenue was assumed; or the operating and contractual model could not sustain the designed duty.

The most useful failure lesson is therefore upstream. Define the problem and endpoints before equipment. Design for chemistry distributions, not one analysis. Treat antiscalant as one control layer, not permission to ignore precipitation. Close mother-liquor and residual balances. Pilot the solids and integrated recycle. Qualify products before revenue. Design maintenance, spares, monitoring and restart with the owner. Make guarantees measurable on the routine feed and complete boundary. When these controls are present, equipment failures can remain bounded events instead of becoming project failures.

Chapter 36 in one sentence Failure often begins in framing, not equipment: the project becomes resilient when every assumption that can stop water service has representative evidence, an observable limit, an owner and a fallback.