Freiburg is a town rife with history and culture, according to Wikipedia. Everything I know about the German town I learned by walking to and from the Fraunhofer Institute each day over the course of a week. Needless to say, the 15 minute walk didn’t provide the best german immersion experience. That said, here is what I did learn: Boulder is not the only town where most people commute by bike, most Germans are simultaneously completely fluent in and apologetic about their English, and it is possible to pack a months worth of membrane testing into a week long trip.
Rebound started working with the Fraunhofer Institute informally in March 2013 after I reached out to Joachim Koschikowski, the head of Thermal Desalination and Separation team, to get some information about their membrane development activity. After several phone calls, a few Skypes, and a damaged membrane care of the US Postal Service, we decided that a visit was probably in order. Thanks in part to our SunChill grant, we were able to pay for an entire week of dedicated testing and experimentation using our salts.
The week followed a perfect, 3-part story:
Fast Start – The first day included a whirlwind of activity that set the tone for the week. That morning I met Dani Winter, one of the scientists who would help me throughout the week, and we dove right into the technology. By the day’s end we had created our test samples, initiated liquid entry pressure testing, and started flow testing with our salts. At some point that day we also began discussing our independently created numerical models.
Crisis –Everything was on track following Day 1, but then the schedule took a bit of a turn. First there were the usual testing issues: a some broken lab equipment, some bad math on my part, etc. But Dani, Rebecca Schwantes (another scientist joining us for Days 2-5), and I put in some long lab hours and worked through the mechanical issues that arose. The real killer came via the small changes to my membrane performance model suggested by Fraunhofer. Although my model was “pretty” accurate, it turns out a higher level of accuracy was required. In just a couple days Fraunhofer helped me replace my original assumptions with more accurate ones, but unfortunately the results were not encouraging. To resolve these membrane issues, the final four days folllowed an intense regime: wake up, work an hour on the code, walk, 11 hours in the lab, walk, 3+ hours on the code, minimal sleep, repeat.
Climax/Resolution – Wednesday ended up being a pretty perfect climax of mechanical and system design issues. In the end though, it worked out like it always does and most of our membrane issues were resolved. Once we identified why the system struggled adapting to the new membrane model, we found a solution. Arriving at this level of understanding took several long nights, many lines of code, and not just a few cups of espresso.
By the week’s end we had multiple salt trials completed, test results confirming compatibility of our salts up to high pressures, several other material trials for Rebound’s other technology, IcePoint*, and, most importantly, our discovery helped construct a strategy to adapt both technologies.
A month later, SunChill is now cheaper and has a higher efficiency than before. Overall the journey was a storybook learning experience about putting yourself out there, being open to help from others, adapting to change, and staying focused in the face of uncertainty. It was also great to work with Joachim’s flexible and dedicated team.
Highlights for the tech-savvy:
The Big Issue – Originally I was using an empirical permeability coefficient independent of temperature and vapor pressure changes along the membrane length. This is not that great an assumption for SunChill, which operates at a unique set of vapor pressures.
The Solution – To fix this I had to switch to a Knudsen diffusion based model and re-work the code to calculate the flux from first principles.
Once I had this in the model it became clear (with help from Dani and Rebecca) that deaerated forced circulation modules would be necessary.
This solves most of our problems, but does mean our system now needs about 3ft2 of solar panels, roughly the area of a pizza box.
*Disclaimer: USAID funding was only used for SunChill testing. Work related to IcePoint was funded using Independent R&D funds.