Debut of an MVP

Author: Luke Erickson

Phase I of the National Science Foundation project wrapped up in January. Phase II is still an idea in a document, patiently awaiting the NSF’s green light. What’s a startup to do for six months without funding to build newer and better freeze point suppression cycles?

Connect with customers, of course! The age of the mad scientist toiling away in the lab for years to reveal a great product are over. In his place, we have the lean startup and customer development. If you’re not familiar with these ideas or minimum viable products (MVP), here’s a short primer or the definitive source that will help put the rest of this discussion in context. We’re using our six-month gap to develop an MVP to test our hypotheses about our customers and technology.

We hypothesize that grocery stores will upgrade their low temperature refrigeration systems to less conventional FPS cycles if they saves them 60% on their electricity bills. To test that hypothesis, we’ve spent the past few months building a stripped down system. The system’s core is still a freeze point suppression (FPS) cycle. We dropped some of the efficiency-boosting components in order to move from design to rollout in less than six months. We swapped integrated waste heat capture for a simple system where we provide the heat for the customer to simulate cost savings. In the end, the system will allow the grocery store customer to get comfortable with an FPS system running in their store while understanding the cost savings. In short, we’ll get to measure response to our hypothesis without added complication.

                                   Russell Muren with the MVP.

                                   Russell Muren with the MVP.

By stripping the IcePoint™ system down to an MVP, we’ve put together a commercial-ready system in just a few months. The biggest challenge so far has been developing the controls system. We coddled the Phase I system; running it was a full time job whenever it was on. The MVP, on the other hand, needs to run 24/7 without an on-site operator. Check out this time series of the system temperatures from an early 3-hour-long run.

                                                                           Time series from early MVP operation.

                                                                           Time series from early MVP operation.

You can extrapolate from the instability of those temperatures that the initial runs were a bit bumpy. The flow rates were up and down, with the boiler drying out twice. The separation wasn’t effective so the icemaker started making FPS-filled slush instead of solid nuggets. Ultimately, the system heated up the freezer instead of cooling it.

Since then, the MVP has run for a month in the lab. We’ve tweaked a few things mechanically and implemented a PID controller. The performance is a lot better. Notice how smoothly everything runs now in this recent 18-hour run.

                                                                            That is some smooth looking data!

                                                                            That is some smooth looking data!