Brentan Alexander, Ph.D.

Scientist / Entrepreneur / Engineer

Resolutions: Don’t jump from lab to commercial scale

I’m ringing in the new year by doing something completely new on the internet: Resolutions. The first resolution for 2019 is also the most common mistake I see innovators make: trying to jump from lab to commercial scale in one hop. Let’s set the scene: Our brilliant inventor has been toiling away in the lab for months, years even, working to prove and fine-tune the latest breakthrough innovation. Finally, with great excitement, our hero stands back and with a confident nod flips the switch.

It works! Gears spin, steam hisses, lightbulbs flash. The inventor speaks: The theory is proven, the science is sound! Now its just a matter of scaling up…that’s just an engineering problem, trivial to solve! Cue the voiceover: it won’t be trivial to solve.

In my experience, breakthrough technology companies are making a critical mistake when they do not build pilot demonstration facilities or hardware with similar scale, features, and equipment as their intended commercial product.

It all boils down to a misunderstanding on the diligence needs of various sources of capital. When a company is making the turn to commercial, the usual early-stage capital sources (venture etc.) that small companies are familiar with are generally poor choices. The transaction size for most industrial businesses at this point is often too large for the early-stage crowd, and if they would do it, the cost of that capital is exorbitant. If the technical risk is gone, why give up the company?

Lower-cost capital providers and mainstream customers, however, are by nature more risk averse (I know, I’m dropping super insider knowledge here). That means they want more assurances that they will get their money back with the expected return on their purchase or capital, and (here is the key) their burden of proof that the technology works is more stringent than what you, the innovator, are used to, or even need for yourself.

We see companies on a near daily basis fail to understand this, and they universally have a hard time convincing skeptical financiers and potential customers that what they’re selling is going to work as described. To access lower-cost capital or mainstream clients, you have to prove performance beyond the basic science and show that the system works as engineered.

At NER, we rarely write policies for companies like these either. We have found that they generally won’t get financed even with the help of insurance, and we’ve seen that the companies that have done demonstration testing learn and improve their processes through the testing program itself. Let’s break this down in nuts-and-bolts fashion:

Scale appropriately

In general, any per-unit scale-up of over 10x for key components of the technology or process will raise eyebrows and dampen enthusiasm.  If you have a chemical process in a beaker, don’t claim it will work in a 10,000 gallon mixing vessel. If you’ve made liquid fuel from 10 pounds of trash in a batch-pyrolysis vessel, don’t tell me you plan to build a plant to consume 100 tons a day in continuous fashion. If you’ve made a new battery cell, don’t claim your next step is mass production of a stack with hundreds of cells many times the size. Exceptions can be made for proven kit from third parties, such as pumps and valving or standard operations like heat exchange. Utilize testing resources as much as possible, such as for-hire processing facilities, to get the scale testing you need while minimizing up-front investment.

Keep on keeping on

You need to demonstrate a reasonable fraction of the expected life of the technology (or expected period of a loan) to access capital or make a sale. The requisite duration of a test campaign is dependent on the technology and expected maintenance cycle: once-through systems or systems with independent unit operations will need less time on test than systems with recirculation loops or highly integrated systems.  In general, anything less than a 100-hour test campaign is immediately disqualifying, and more complex systems will need many times that to satisfactory demonstrate system reliability. 1,000 hours of stable operations is a safe target for industrial technologies with 3-5 year overhaul cycles. Stable doesn’t necessarily mean problem free: downtime in a testing environment that would be handled with redundancies or sparing at a true commercial facility is perfectly acceptable.

Key degradation mechanisms also need to be thoroughly examined through highly-accelerated life tests (HALT). Working on a new battery design? Cycle that thing like crazy to artificially show years of usage in just a few months. Working with a proprietary catalyst formulation? Spike the process stream with known poisons to simulate long term exposures. Make sure to take enough data to show a stable, non-accelerating degradation trend for at least 50% of the expected life using HALT tests, and validate this with real-time tests for the first 5-10% of the product life.

And remember that even this isn’t enough

Some banks we work with at NER want to see a like-size and like-kind facility with at least three years of operational history before lending. They are used to funding Siemens, or BASF, or GE, where the technical risk is virtually non-existent and proving that is straight-forward. That’s stiff competition…but in our experience at NER, if you test to the standards outlined above, you’ll have a fighting chance to convince customers and low-cost financiers that the technical risk has been retired.

And (shameless self-promotion!) if you need a little help getting over that hump, NER will be ready and willing to take a look and help provide comfort to the customer or investor by bridging the gap between the perceived and actual technology risk.


Also published on Medium.