If you’re evaluating a zero liquid discharge process, the core train usually looks like this: reverse osmosis (RO) concentrates the wastewater first, mechanical vapor recompression (MVR) evaporates the RO reject, and a crystallizer turns the final brine into solid salt. The logic is straightforward — each stage removes water more aggressively than the last, so the volume of liquid you must handle drops by an order of magnitude at every step. What isn’t straightforward is making the three stages work together without scaling, fouling, or blowing your operating budget. I’ve commissioned and troubleshot enough of these trains to know that the design decisions at the RO stage determine whether the MVR and crystallizer run smoothly or become a maintenance nightmare.
Why ZLD Trains Are Built in Stages
The reason we don’t just send raw wastewater straight to an evaporator is economics. Evaporation is thermally driven, and the energy cost scales with the volume of water you’re boiling off. If you can reject 70 to 85 percent of the feed as clean permeate using membrane pressure — which costs a fraction of thermal energy — you shrink the evaporator and crystallizer dramatically.
Think of it as a volume reduction cascade:
- RO stage: Feed at, say, 3,000–8,000 mg/L TDS gets concentrated to 30,000–70,000 mg/L in the reject. Permeate goes to reuse or discharge.
- MVR stage: The RO reject is boiled, and the vapor is recompressed to reuse its latent heat. Brine concentrates further to near-saturation.
- Crystallizer stage: The saturated brine is cooled or further evaporated so dissolved salts precipitate as solid crystals you can dewater and landfill or sell.
Each stage has a different failure mode, and they interact. Get the RO recovery too aggressive and you’ll precipitate gypsum or silica in the last elements. Push the MVR too hard and you’ll scale the heat exchanger in weeks. Feed the crystallizer a brine with the wrong composition and you’ll get a sticky, unfilterable sludge instead of clean crystals.
Stage 1: Getting RO Recovery Right
RO is where most of your operating cost savings live, so it’s worth spending design time here. Standard brackish RO elements top out around 1,200 psi (about 83 bar) in practical industrial service, but for ZLD service you’re usually looking at high-pressure RO or disc-tube RO (DTRO) because the feed is already concentrated or has scaling potential.
Key design parameters
| Parameter | Typical range | What drives it |
|---|---|---|
| System recovery | 60–85% | Feed TDS, scaling ions, pretreatment quality |
| Operating pressure | 400–1,200 psi | Osmotic pressure of concentrate |
| Flux | 8–14 gfd | Fouling tendency, element type |
| Concentrate TDS | 30,000–70,000 mg/L | Solubility limits of CaSO₄, SiO₂, BaSO₄ |
I always run a scaling projection before fixing recovery. The three ions that bite you most are calcium sulfate, silica, and barium sulfate. If your feed has 200 mg/L silica, you cannot concentrate past roughly 2,000 mg/L without risking silicate scaling unless you dose a silica inhibitor or run at elevated temperature.
Antiscalant selection matters more than most people think. A generic phosphonate won’t hold up against high calcium sulfate saturation. I’ve seen systems where switching to a purpose-built high-saturation antiscalant let the operator push recovery from 70 to 78 percent — a meaningful reduction in downstream evaporator load.
For high-salinity or high-fouling feeds, DTRO membrane systems handle suspended solids and scaling better than spiral-wound elements because the open-channel design resists plugging. They cost more per unit of permeate, but they keep running where spirals would foul out in weeks.
Stage 2: MVR Evaporator Design and Operation
Mechanical vapor recompression is the workhorse of the ZLD train. The concept: boil the RO reject, compress the resulting vapor to raise its temperature and pressure, then use that compressed vapor as the heating medium for the same evaporator. You’re recycling latent heat, so the energy input is mostly the compressor motor plus whatever you lose to the environment.
A well-designed MVR system runs at a specific energy consumption of roughly 15–40 kWh per cubic meter of distillate produced, depending on boiling point elevation, compressor efficiency, and how much heat you recover from the blowdown. Compare that to a multi-effect evaporator, which might need 0.3–0.5 kg of steam per kg of water evaporated — the MVR wins on operating cost whenever electricity is reasonably priced relative to steam.
What determines MVR performance
- Boiling point elevation (BPE): As brine concentrates, its boiling point rises. High BPE means the compressor has to work harder to create a useful temperature difference across the heat exchanger.
- Compressor type: Centrifugal for large flows, positive displacement (roots or screw) for smaller or higher-pressure-ratio duty.
- Heat exchanger surface and material: Titanium or duplex stainless for chloride service; the surface area must account for fouling margin.
- Concentration limit: You stop evaporating when the brine reaches saturation for the least soluble salt, or when viscosity makes circulation impractical.
The single biggest operational headache with MVR is scaling on the heat exchanger tubes. If you’re evaporating a brine that’s saturated in calcium sulfate, you will form scale unless you seed the system, run a high circulation rate, or operate at a temperature where the salt stays soluble. I’ve seen plants lose 30 percent of their capacity in three months because nobody accounted for the retrograde solubility of calcium sulfate above 40°C.
In my experience, the difference between an MVR that runs 8,000 hours a year and one that shuts down monthly is almost never the compressor. It’s the pretreatment and the brine chemistry management upstream.
For difficult brines with high organic content or oil, evaporation with oily wastewater requires special attention to defoaming and surface skimming, otherwise the organics concentrate in the brine and cause foaming that carries over into the compressor.
Stage 3: Crystallizer Selection and Salt Handling
The crystallizer takes the near-saturated brine from the MVR and pushes it past saturation so salt drops out as solid crystals. There are two main approaches:
Forced circulation crystallizer
This is the most common choice for ZLD because it handles high solids loading and viscous slurries. The brine is circulated through a heat exchanger at high velocity to prevent scaling, then flashed into a vapor-liquid separator where supersaturation is generated and crystals grow. Typical crystal size is 0.2–0.5 mm, which dewaters reasonably well on a centrifuge or filter press.
Fluidized bed / Oslo crystallizer
Better for producing large, uniform crystals of a single salt. The tradeoff is that it’s less tolerant of multiple salts and organics. If your brine has a mixed salt composition — which is common in ZLD from complex industrial wastewater — the forced circulation design is usually more forgiving.
The crystallizer is where the whole train’s chemistry comes home to roost. If the RO and MVR stages didn’t remove enough hardness or silica, you’ll get co-precipitation that produces a sludge instead of crystals. That sludge is hard to dewater, high in volume, and expensive to dispose of. I’ve seen projects where the crystallizer output went from clean salt to a sticky paste simply because the upstream softening was underperforming.
Solid-liquid separation after the crystallizer typically uses a pusher centrifuge for coarse salts or a filter press for finer material. The mother liquor is recycled back to the MVR or crystallizer feed, which means impurities that don’t precipitate will build up over time. You need a small purge stream to control that, and that purge has to go somewhere — often back to the front of the plant or to a dedicated disposal route.
Energy and Cost Reality Check
ZLD is expensive to build and expensive to run. Anyone who tells you otherwise is selling something. But the cost is manageable if you design the train correctly. Here’s a rough breakdown of where the money goes:
| Cost element | Share of total (indicative) | Notes |
|---|---|---|
| RO / pretreatment | 20–30% of CAPEX | Membranes, pumps, chemical dosing |
| MVR evaporator | 35–45% of CAPEX | Compressor is the single biggest item |
| Crystallizer | 15–25% of CAPEX | Includes solid handling |
| Electrical power (OPEX) | 50–65% of OPEX | Mostly MVR compressor |
| Chemicals (OPEX) | 15–25% of OPEX | Antiscalant, cleaning, pH adjustment |
| Maintenance (OPEX) | 10–20% of OPEX | Compressor overhaul, heat exchanger cleaning |
These ranges are project-dependent. A plant with cheap electricity and a high-silica feed will have a different cost profile than one with expensive power and a clean brine. The point is to model it honestly before you commit.
One thing I always tell clients: the cheapest ZLD train is the one that doesn’t have to evaporate water you could have removed with a membrane. Every percentage point of RO recovery you gain reduces the MVR and crystallizer size, and that saving compounds over the life of the plant.
If you’re at the early design stage and want to sanity-check your process configuration, our engineers can review your water analysis and propose a train that matches your actual brine chemistry rather than a generic template.
Maintenance and Common Failure Points
I’ve been called out to more ZLD plants than I can count, and the failures cluster around a few predictable issues:
RO membrane fouling and scaling
Usually caused by inadequate pretreatment or antiscalant overdose/underdose. Monitor normalized pressure drop and salt passage weekly. A 15 percent rise in differential pressure across a stage means it’s time to clean or investigate.
MVR heat exchanger scaling
Calcium sulfate, calcium carbonate, and silica are the usual suspects. Clean-in-place with the right chemistry — not just acid — is essential. Silica needs an alkaline clean; calcium sulfate needs a chelant or specific scale dissolver. Using the wrong cleaner can make it worse.
Compressor vibration and bearing wear
Usually a sign of carryover from the evaporator. If liquid droplets reach the compressor, you’ll get erosion and imbalance. Proper demister sizing and regular inspection of the mist eliminator prevents this.
Crystallizer plugging
Happens when supersaturation is too high or crystal residence time is too short. The fix is usually to increase circulation rate, adjust the flash temperature, or seed the crystallizer with fine crystals to provide growth surface.
Purge stream buildup
Impurities that don’t crystallize — organics, ammonia, certain ions — concentrate in the recycle loop. If you don’t purge enough, they’ll eventually kill the process. If you purge too much, you waste water and salt. Finding that balance is a commissioning and operating skill.
When ZLD Makes Sense — and When It Doesn’t
I’ve turned down projects where ZLD was the wrong answer. If your plant has access to a municipal sewer with capacity and your discharge limits are achievable with conventional treatment, ZLD is a very expensive way to solve a problem you don’t have. The economics only work when:
- Discharge permits are unavailable or prohibitively expensive
- Water reuse value is high enough to offset the energy cost
- You’re in a region with strict zero-discharge requirements for your industry
- The salt byproduct has a market or a low-cost disposal route
For a deeper look at how evaporation fits into broader industrial wastewater treatment, the process overview here covers the full range of separation technologies and where each one fits.
If you’re dealing with landfill leachate, which is one of the hardest feeds for ZLD because of its high organic load and variable composition, the design has to account for biological pretreatment or advanced oxidation before the RO stage. Otherwise the organics will foul your membranes and carry over into the evaporator. The chemistry of landfill leachate is worth understanding before you size anything.
Frequently Asked Questions
What recovery can I realistically expect from the RO stage in a ZLD train?
Most industrial ZLD systems run RO recovery between 60 and 85 percent. The upper end requires good pretreatment, effective antiscalant, and feed chemistry that doesn’t hit solubility limits too early. If you have high silica or high calcium sulfate, expect to be closer to 60–70 percent unless you add softening or use DTRO.
How much does an MVR evaporator cost to operate?
Electricity is the dominant cost. A typical MVR consumes 15–40 kWh per cubic meter of distillate, depending on boiling point elevation and compressor efficiency. At industrial power rates, that translates to a few dollars per cubic meter. Maintenance and chemical costs add another 20–40 percent on top of the power bill.
Why does my MVR heat exchanger scale so quickly?
Almost always because the brine is saturated in a salt with retrograde solubility — calcium sulfate is the classic example — or because the concentration limit was set too high. Check your brine chemistry against solubility curves at your actual operating temperature, and consider seeding or reducing concentration.
Can I skip the crystallizer and just dispose of the MVR brine?
Only if you have a permitted disposal route for liquid brine, such as a deep well or an approved hauling service. If your goal is true zero liquid discharge, the crystallizer is what gets you there. Without it, you’re running a volume reduction system, not ZLD.
What’s the most common cause of ZLD plant downtime?
In my experience, it’s scaling and fouling in the evaporator and crystallizer, usually traceable to inadequate pretreatment or poor brine chemistry management. Compressor failures are less common but more expensive when they happen. A well-designed plant with proper instrumentation and a disciplined cleaning schedule can achieve 90 percent or better availability.
How do I choose between MVR and a multi-effect evaporator?
MVR wins when electricity is cheap relative to steam and when you have a steady, high-volume flow. Multi-effect wins when you have waste heat available or when power is expensive. For very high boiling point elevation brines, a hybrid — MVR plus a finishing effect — sometimes makes sense. The decision should be based on a lifecycle cost model, not a rule of thumb.
What pretreatment do I need before the RO stage?
At minimum: suspended solids removal (often ultrafiltration), hardness removal if needed, pH adjustment, and antiscalant dosing. For feeds with high organics, add biological treatment or advanced oxidation. For feeds with oil, add coalescing or DAF. The cleaner the RO feed, the longer your membranes last and the less often you have to clean.