banner image

11, Mar 2019 / rebound

Pilot Results

It has been just over a year since we shipped out our first paid pilot to Oxnard, California and an update is in order. Before I get to that, I want to take a moment to thank the lineage engineering and analytics teams for helping get this Pilot installed along with the entire Oxnard staff for helping us operate it.

Now, let’s get to some results, with more than 5000hrs of operation, the Oxnard pilot was a success. After reviewing its performance, Lineage Logistics agreed to deploy the first, full-scale IcePoint® unit. Additionally, the pilot was used to as a showcase for investors. To summarize pilot performance, lets go over some high-level numbers:

Achieved Value
Total Runtime [hr] >5,000
Strawberries Accelerated [lb] 1,700,000
Largest Burst of Cooling [kWt] (TR) 37.5 (10.5)
Tank Temperature [C] (F) -30 (-22)
COP* [kWt/kWe] 3.3
* The COP calculation was complicated, as I will describe later. This value is an adjusted COP taking into account a condenser issue that, for practical purposes, was left unsolved.


The overwhelming majority of the pilot system performed well over the entire operating period

1. There were no major pump failures, tank ruptures, or spills.

2. The unit never negatively impacted the performance of any on-site equipment or caused any lost time.

3. The controller, cloud-based user interface, and remote-control features never failed to respond.

For all these reasons, the pilot was a success. That being said, the most valuable aspect of the pilot were the many lessons learned. I think it is valuable to discuss a few of the key findings here.

Vapor compression isn’t easy: There is the perception out there that because Freon and Ammonia refrigeration is so ubiquitous it must be simple. That couldn’t be further from the truth. By far the least reliable, and most difficult to control, component was the off-the-shelf condensing unit. Why? Because vapor compression is extremely mechanically complicated. We dramatically underestimated this and thus were never able to operate at the ice maker’s designed suction temperature. As such, we had to adjust our COP values for the extremely low suction temperatures encountered. For the past year, ice maker suction temps have been a primary development focus and the few people we have shown our new top secret ice maker to are… excited.

Waste heat isn’t economically worth it: This one ended up being pretty simple. We can mechanically separate our freeze point suppressant for a fraction of the cost of capturing waste heat. In the end, capturing waste heat is just too expensive. While running a refrigeration cycle at -22°F using 140°F waste heat is a thermodynamic marvel, the economics just don’t line up. Luke and Alex are hard at work building out our mechanical vapor recompression separator at this very moment.

In-house commissioning is key: One of the hard lessons we walked away with is how much easier it is to fix a problem in your own shop before shipping a product. We ran into a lot of small issues with this pilot. All of these issues would have been a 1-2 day fix in our shop but diagnosing them remotely and fixing them on-site took months. For the full-size deployment, we set aside significantly more in-house commissioning time and Chance is hard at work preparing the facility’s infrastructure for the task. Some of the small problems we spent a lot of time fixing in the field were:

1. the water recuperator froze often

2. managing lubricant oil levels in the ice maker was difficult

3. the check valve seals on the ice maker water lines failed

4. the vacuum pump required a soft-start relay to startup under some ambient conditions

5. the column solenoid caused illusive water hammer issues

6. the ice maker lubricant caused materials compatibility issues with seals throughout the system.

It is great to walk away from a pilot with a customer making follow on purchases and investors excited to help us scale. However, from a strictly technical development perspective, the satisfaction, despite those long hours cramped in the hot container, was pushing this technology forward and learning what it takes to create a reliable, large-scale, industrial refrigeration technology.

I want to close with a host of photos covering the pilot. This was a huge undertaking for our small team and I want to share that journey as honestly as possible.

Fig1. We started with a standard ISO shipping container.

Fig2. We modified the walls, painted, and sealed the container.

Fig3. Framing, insulation, and tanks were next to go in

Fig4. Due to a miscommunication with our mechanical contractor, we initially thought we could put the entire unit on the roof and thus built the separation system inside the container. This was typical of the kinds of lessons we were learning both about mechanical contractor communication and about system design.

Fig5. Simultaneous to system assembly, we were developing system components. This parallel development was extremely time sensitive and required a phenomenal amount of hours from our team. Here is a photo of our 8th R&D ice maker we developed.

Fig6. Ice maker version 8 makes ice for the first time!!!

Fig7. Back on the assembly side, the separation system was taken out of the container and assembled into its own roof-mounted system.

Fig8. I am still amazed by how we were able to build this system in such a small space. We had to plan the assembly down to the inch so that we were sure we wouldn’t block the path of equipment.

Fig9. The system arrives in Oxnard!!! Kevin did a great job managing the on-site relations during the install while the tech team prepared to commission the system.


Fig10. The separation system is installed. It will start to become obvious to anybody who has ever worked with hot pipe why waste heat capture isn’t economically viable.


Fig11. The condensers to the left in this image were the source of heat and cooling for the separation system.

Fig 12. The pipe welding is complete!

Fig13. The system went live Feb 13th!! It actually wasn’t that exciting. The system just starts getting cold. Here is some video of the ice makers dropping ice into the tank. Sorry for the terrible video quality, this shot is very difficult to take… After this point we started commissioning the system and troubleshooting problems.

Fig14. The work for the next several months was adjusting our controls to deal with the unpredictable nature of the waste heat from ammonia compressors. Like all refrigeration plants, Oxnard’s compressors produce varying amounts of waste heat sometimes with very quick changes. The image here shows Oxnard’s compressors shutting off because the freezer reached temp at 12:30am. We had to build in a lot of controls robustness to handle these issues.

Fig15. The tank temperatures started to come down over a few days, but took a while because the separator was only operating a few hours each day. These screen shots were taken from our cloud based dashboard that was critical in seeing how the system was running remotely and while we were onsite.

Fig 16. The day we delivered our first burst of cooling! More that 30kWt delivered to the ammonia.

Fig17. We eventually got the system to perform at design. If figures are in orange, they are operating off-design. Note that none are orange. Good day!

Fig18. Once we got the system performance stabilized, our on-site work turned to improving performance and O&M. The issues we ran into were diverse. I am including this image because it shows how difficult this task was onsite. Here I was trying to diagnose a mixing issue in the ice tank. The ice tank was insulated with 6” of polystyrene insulation and completely sealed. I had to stand on a latter, insert a snake camera through an inspection port, and see what it looked like from that angle. Then, remove the camera, re-orient the lens, and repeat. What took 10’s of hours to solve on-site, might have taken 15 minutes in our facility, hence the importance sufficient internal commissioning.

Fig19. This Buna gasket and tube have an “A” grade compatibility with the ice maker lubricant oil. Both of them were embrittled and failed due to contact with the oil. We dealt with these kind of issues throughout the pilot operation. Luckily, Josh and Luke found these two issues during regular inspections and we were able to replace them both without loosing vacuum.

Fig20. Here is another great example. We used these small gear motors to manage lubricant flow. This pump was supposed to have an AFLAS seal. It didn’t and immediately failed. Josh took this image one week after installation.

Fig21. Another issue we ran into with this design was frost buildup. A few components needed to be accessible and couldn’t be insulated. As such we built up a lot of frost and the unit was constantly full of water. This isn’t actually that unusual in the Refrigeration industry, but our future systems won’t have this issue because there will not be components which need to be bare.

Fig22. One last point about the materials compatibility issues. Initially we used control valves with Buna seals. Again, this material had an A compatibility rating from the manufacturer. As the seals swelled and shrank the flow would change. Here are the flows one day after all being set to 4.5. As you can see some are as low as 1 and some are as high as 6.

Fig23. Regardless of the ups and downs of working out the system bugs, we had a lot of fun showing people the operating equipment. Here is Kevin pointing out the liquid subcooler HX which can be seen in the upper portion of the shot.

Fig24. Its easy to get lost in the pump failures and late-night calls to the Oxnard maintenance staff (Alex Mendoza, you are a champion), but one of my favorite parts of running the pilot was seeing the trucks roll up and get filled with strawberries that we helped freeze.


Leave a Reply

Your email address will not be published. Required fields are marked *