You're compressing to the wrong pressure and evaporating from the wrong geometry.
- The dairy MVR industry solved this decades ago: you compress vapor only to the condensation pressure needed for heat rejection — not to atmospheric.
- Your system compresses to 1013 mbar because that's what a standard vacuum pump does.
- But you only need to condense the vapor, which at a nighttime-cooled surface (28°C) requires only 40 mbar.
- This cuts compression work by 75%.
- Separately, your 1-2% flash fraction from bulk liquid is a nucleation kinetics problem, not a thermodynamic one — spray atomization to 100-200 μm droplets achieves 5-8% flash per pass, halving the recirculation requirement.
- These are engineering changes, not research.
- The Carnot COP for the required temperature lift (30→48°C) is 16.8, and compress-to-condensation MVR practice from desalination proves that 50-60% of Carnot is achievable at industrial scale — placing COP 8-10 within reach using known engineering.
If you prioritize speed to market and proven components, start with the optimized pressure-swing system (sol-primary). If you're building a multi-generation product roadmap and want to own the IP on a paradigm shift, run the pervaporation membrane compatibility test in parallel — it costs $3-5K and takes 4 weeks, and the result determines whether your long-term architecture is thermal-only or membrane-hybrid.
Optimized Pressure-Swing with Compress-to-Condensation and Nighttime Time-Shifting
Compress vapor to condensation pressure (not atmospheric), use polymer microchannel HX for tighter pinch, shift 75% of regeneration to nighttime — achieves COP 10-15 in arid climates but needs roots blower validation for condensable vapor service.
Pervaporation Membrane Pre-Concentrator with Thermal MVR Finish
Replace bulk evaporation with selective molecular transport through a pervaporation membrane for the easy concentration step (30→35%), achieving COP 15-25 for 55% of the water removal — but membrane survival in LiCl is untested.
If this were my project, I'd split my effort into two parallel tracks starting Monday morning. Track 1 is the money track — get sol-primary built and running. I'd email Busch Vacuum and Pfeiffer Vacuum with the duty spec (14 mbar suction, 80-150 mbar discharge, water vapor, 100-300 m³/hr) and ask for a vapor-service roots blower recommendation. Simultaneously, I'd order a Kelvion PVDF polymer plate HX sample and a 150L PP tank. The compress-to-condensation insight alone is worth the price of admission — it's the single biggest COP lever in the entire analysis, and it requires zero new science. While waiting for the blower, I'd build the air-side pre-concentrator bench test in my parking lot: 10 liters of Brentwood packing, a box fan, 5 kg of LiCl, and a kitchen scale. Three days of testing in Phoenix summer conditions tells me whether I get the free COP boost from ambient air. If the answer is yes (and I think it will be), that's a $200 add-on module that boosts weighted COP by 3-5 points. Track 2 is the IP track — run the membrane experiments. I'd contact Sulzer for a PERVAP 4101 coupon and Sterlitech for a CF042 test cell with 0.2 μm PTFE membrane. Run both experiments in parallel: pervaporation compatibility (200-hour soak in 35% LiCl) and VMD stability (100-hour operation at 42% LiCl). Total cost: $5-8K. Total time: 3-4 weeks. These two experiments collectively determine whether my second-generation product is a membrane hybrid (COP 10-14) or thermal-only (COP 8-10). If either membrane survives, I file a provisional patent on membrane-based desiccant regeneration before publishing anything.
- The thing I would NOT do is pursue the ejector or the staged architecture yet. The ejector is a $10-15K CFD+prototype investment that I can't justify until sol-primary is deployed and generating revenue. The staged architecture is a beautiful long-term vision, but it's a third-generation product — I need to walk before I run.
- I would also NOT pursue the ultrasonic atomization concept. Mechanical spray nozzles get me 80% of the flash improvement at 5% of the complexity. The ultrasonic throughput limitation is fatal.
The strategic insight I keep coming back to: the HVAC desiccant community has been solving this problem in isolation, when the dairy industry, desalination industry, and membrane science community each hold pieces of the answer. The compress-to-condensation insight from dairy MVR is worth 3-4 COP points. The membrane insight from chemical engineering could be worth another 3-5 points. And the air-side insight from lithium mining is essentially free energy. My job is to connect these dots, not to invent new physics.