Hindsight is 20/20: Some preliminary conclusions (Reflections from El Paso, Post 2)

Things got a bit hectic during the last few weeks in Honduras and it has taken me a while to get situated at home again. I do want to take the time to finish up on our activities, so this is the second of at least three that will summarize what we did in September and what we are planning to do in the future.

===============================================

After having a few weeks to think about what we have learned, I keep thinking "If I only did this..." or "If I just would have started with..." or many other things that are only obvious after time has passed and one has learned from experience, mistakes, and observations. This post will be my way of organizing what we learned from the pilot from a technical point of view. This will be a long post, but I will put some pictures here and there to balance the text.

It is probably helpful to review what our initial objectives were for our pilot demonstration in Tegucigalpa. A few months ago, I summarized how we were using ZDD to desalinate brackish water using energy from the sun. We received funding from USAID to:

• Demonstrate that high efficiency and zero liquid discharge are possible using ZDD powered by renewable energy
• Transfer knowledge to our partners at UPi
• Provide training for Honduras farmers about water quality and desalination technology (including ZDD, of course!)
• Assess the commercial potential for ZDD in Honduras

Just in case you forgot - Here's our pilot site, located at the East end of Morazán Blvd
The PV panels were on the ground in the left. The desalination equipment is in the container in the middle of the picture. The salt recovery and enhanced evaporation system was behind the wall with the interesting art on the right of the picture.

In addition to these general objectives, we were also produce 8,000 liters per day of irrigation water and 250 liters per day of drinking water with certain water quality parameters that were set by the Desal Prize judges. The drinking water requirements were relatively simple, in that the salinity had to contain no more than 600 parts per million of dissolved salts (and meet other World Health Organization water quality requirements). Most reverse osmosis and some nanofiltration membranes can meet these requirements. The irrigation water quality requirements were more challenging. The salinity requirement was slightly lower (no more than 550 parts per million of dissolved salts), but other requirements included a sodium absorption ratio less than 3 (this requirement generally ensures better water infiltration into the soil), a calcium to magnesium ratio greater than 1, pH less than 8, and boron no higher than 0.5 parts per million. [If you would like to learn more about how water quality can affect soil quality and crop health, here are some resources from Texas A&M, Colorado State University, and North Carolina State University.] 

Relationship between SAR and Conductivity.
High SAR (high sodium relative to calcium and magnesium) generally reduces infiltration, however it depends also on the salinity (measured using conductivity) also. Some groundwater may have the same SAR, but the one with higher conductivity is more acceptable for irrigation (For more info: Montana State University). 

Most reverse osmosis membranes produce very high quality water from a salinity perspective, however, because the membranes are very good at rejecting large ions and large dissolved species, the concentration of calcium and magnesium in the product water is very low. This means that the water is generally unacceptable for irrigation without adding some minerals back to the water. Our team usually uses a blend of "loose" reverse osmosis and nanofiltration membranes to achieve suitable water quality. The number and orientation of membranes is chosen based on the amount of silica in the water. All of the waste stream from the RO/NF is fed to another system, called EDM, that removes some of the species that limit desalination efficiency using RO or NF alone (read more here). However, EDM does not remove uncharged species like silica, so the concentration will build up between the RO/NF and EDM systems unless the silica has another place to go. Using loose membranes allows us to have the silica pass into the product water. However, there is a balance: as more silica is allowed to pass into the product water, more of other dissolved species also pass into the product water. This means that the product (irrigation water) will have higher salinity. This means we balance the product water quality and silica concentration. If the silica (and/or other sparingly soluble species) reach levels above saturation, they will precipitate on the RO/NF membranes. Sometimes the precipitated salts (sometimes called "scale") can be removed with a chemical cleaning. Sometimes the precipitation is permanent and reduces the membrane performance.

Hindsight lesson #1: Bring more field testing kits and monitor silica more often. We knew silica in the brackish water had the potential to be problematic. However, previous pilot experience had led me to think I could monitor things well enough without having a full analytical laboratory. I. Was. Wrong. We scaled up the first set of membranes permanently, mostly because we weren't adding antiscalant, which mitigates scale formation in the membranes (we added checking on the pump to our regular monitoring to make sure this didn't happen again). We loaded new membranes and we thought they were doing well. But, as it turns out we scaled them up again. I need some more time to do some calculations, but the RO/NF membrane combination seemed to remove more silica than expected, which led to more silica in the RO/NF loop. We ended up installing a bleed line to ensure the silica stayed below a safe concentration. However, the scale was permanent and affected our production. We did a cleaning, but it seemed to affect the salt rejection, so our product water quality wasn't as great. Had we started with the bleed, we probably would never have had the silica problem in the system (good old hindsight...).

Lesson Learned: Measure silica regularly (and check antiscalant addition)

Hindsight Lesson #2: Use acid for pre-treatment of the brackish groundwater (i.e. trust  intuition and past experience). The Desal Prize judging rules penalized teams for using chemicals, like acid, as part of the desalination process. So, we decided to try to operate without acid, even though the alkalinity present in the Los Almendros well was higher than any other groundwater the team had ever worked with in past pilot tests (I wrote about this here). Alkalinity present in the RO/NF concentrate ends up in the EDM's Mixed Na stream at a much higher concentration. Because the compounds formed (mostly sodium bicarbonate, but also some sodium carbonate) could eventually be supersaturated and precipitate, the stream is eventually diluted with a very low salinity water. Finding the setpoint for dilutions without an alkalinity test kit involves trial and error, and some indirect measurements. We operated for the first few weeks this way, but eventually decided to install an acid feed line on the brackish feed. Had we done this from the beginning, this could have been incorporated into the automatic controls. I chose not to do this partially for budgetary reasons, but also because I really wanted to see how the system would perform without acid. In previous pilots I have been able to find the appropriate setpoint without needing to do a lot of field tests, however, field tests ended up being necessary. We thought we would have access to a laboratory with alkalinity test capabilities for the pilot, so we didn't bring reagents with us. After receiving a test kit that was sent from UTEP, we were able to monitor alkalinity in the RO/NF and EDM systems and were able to get mostly stable operation. While the process was not smooth, we did find a pH setpoint that seemed to work for both the RO/NF and EDM systems. Thankfully, we had UPi students available to help with the field tests.

Monitoring Alkalinity.
Top Left: Using the trial & error method involves testing various concentrations in the Mixed Na stream. If the concentration is too high, scale forms in the Mixed Na compartments. We could confirm that the scale was carbonate-based by adding acid and seeing if it bubbled.
Bottom Left: Indirect measurements involved taking a sample from the various streams with suspected supersaturated alkalinity. If precipitation was visible, we had gone too far.
Right: Measure alkalinity analytically. Here, UPi student Victoria Amador is performing the alkalinity test

Hindsight Lesson 3: Prepare for the worst in terms of particle removal when you only have one water sample to base a design from. Another thing that most water engineers know is that you can't base a design off of a single water sample. But, that was all we had, so we did. Also, we didn't know a lot about the well design before arriving in Honduras (another bad idea, in terms of design). One thing we found out when we got here is that the well produces much more water than we need, and the well isn't currently used, so the well was cycled on & off several times during the day to fill a storage tank that we pumped from for the pilot. Each time a well is turned on, particles are brought to the surface. This is usually not a problem, since wells are not normally operated in this manner. Also, many desalination plants will install a strainer to filter out sand and other particles from the well. We did have cartridge filters ahead of our pilot equipment, but we initially had filters with too large of pores installed. Even the smallest pore size filters we had (1 micron) still allowed some particles to leak by. These particles caused fouling of the membranes and are suspected to have also contributed to the precipitation issues found in the RO/NF. We didn't have enough budget to install additional particle removal equipment on this project, but it will be included with the system for further work so that better results can be expected.

Particle Removal.
Left: RO/NF feed tank after 20-micron cartridge filters
Middle: Cartridge filters after one week, one day, and no use (clean)
Right: RO/NF feed tank after 1-micron cartridge filters (there was a vast improvement, but particles were still leaking past the filters)

Bottom Line: We worked hard to be successful

So, what did we learn? And did we meet our stated goals and objectives? I think so. 

Between competing in the Desal Prize and operating in Honduras, we reduced our energy needs by 40%! This means that a smaller PV system could be purchased or that the extra energy could be used for farmers' homes (or possibly sold to recover additional costs). Further energy savings are possible if more of the AC load can be moved to the DC side.

We were able to achieve very high desalination efficiency (94%) with a very challenging brackish water source and only used power produced from our PV system. We produced solid gypsum (the first sample sent to our UTEP lab indicated about 98% purity) and liquid sodium chloride using a salt recovery and enhanced evaporation system. We weren't able to concentrate the sodium chloride enough prior to evaporating it, but previous experiments suggest that we can achieve good enough purity for use in the EDM. Before the RO/NF membranes scaled, we were able to meet the water quality objectives, as estimated from the field analyses. We have identified conservative operational setpoints for how a ZDD system could operate at the Los Almendros location.

And, while we weren't able to demonstrate everything running at the same time, our enhanced evaporation system seems to be capable of evaporating enough of the waste stream to maintain zero liquid waste discharge. With a fan installed and oeprating, we were able to evaporate almost a gallon per hour, which is about double the combined waste flow. 

This project was so much more than the technical objectives. We started the process of transferring knowledge to our UPi partners and to various groups in Honduras. Over the course of two months, we held three training sessions and a public tour at the end of the piloting, we met various UPi professors and students, and we had a chance to meet with Mr. James Watson, the USAID Mission Director Honduras. I am grateful for the opportunity to visit the good people of Honduras and hope that we will be able to continue this work.

Developing Desalination in Honduras. From Left to Right: Clara Borrego, Chancellor Jance Carolina Funes (UPi), Mr. James Watson (USAID), Malynda Cappelle, Ing. Luis Eveline, and Isaac Ferrera (USAID)