From protocol to production

What the transition from lab to industrial scale actually demands, and why the answer is one roof plus external capacity

Heleen, Head of R&D, NoPalm Ingredients

Most of what is written about industrial biotech treats scale up as a single event. A company develops a process, then scales it, and the difficulty lies in the jump.

That is not how it works. Scale up is a sequence of distinct problems, and each one has to be solved on infrastructure suited to it. Get the sequence wrong, or attempt a stage on the wrong equipment, and you either burn capital proving something you could have proven cheaper, or you commit to a design before you know enough to design it.

We have now been through this sequence several times. What follows is what it actually demands, and why we have concluded the answer is a facility with every step under one roof, with external manufacturing alongside it rather than instead of it.

Four stages, four different questions

The transition from protocol to production is not one transition. It is four, and each asks something different.

Bench. Does the biology do what we think it does. Does the organism accumulate lipid on this feed, at what rate, to what profile. Small volumes, many parallel experiments, fast iteration. Nothing about this stage tells you anything about industrial behaviour, and it is not supposed to.

Pilot. Does the process survive contact with real equipment. Vessels with real geometry, real mixing, real heat transfer, real timing. This is where a protocol becomes a process: where you learn which parameters actually control the outcome and which ones you only thought did.

Industrial validation. Does the process hold at commercial volume. Mixing, oxygen transfer, heat removal and timing behave differently in a large vessel, and you find that out by running it rather than modelling it.

Production. Can you do it again, on specification, on schedule, with material a customer can build a product on.

The mistake the field makes is treating the last two as the same thing. They are not. Proving a process can run at 120,000L and being able to supply 120,000L reliably are separated by everything that makes a supply chain.

Where we are today

In house in Wageningen we take a process from bench up to 400L, with our own downstream processing. That is where a process is built, characterised and proven stable before it goes anywhere near industrial volume.

Industrial validation and customer sample production run externally, at contract manufacturers. We have completed industrial fermentation campaigns at 120,000L several times over the past two years.

Those campaigns matter. They are the reason we can say the biology travels, that the process holds at commercial volume, and that what we make there behaves in a customer's application the way our pilot material does. That is not a projection. It has been run.

What external manufacturing does well

There is a version of this argument that treats contract manufacturing as a compromise to escape from. That is not our position, and it would be a poor reading of the economics.

External capacity gives you volume without capital. It lets you validate at industrial scale years before you could justify building. It gives you access to equipment you would otherwise wait for, and to operators who run that equipment every day. For a company proving a process, it is the correct answer, and we would make the same choice again.

It has also taught us things we could not have learned in our own building: how our process behaves on equipment we did not design for it, which is exactly the robustness test a technology has to pass before anyone builds a plant around it.

What external capacity cannot do is give you control of the sequence.

The constraint that a demo facility removes

The technology required to complete a single industrial run does not exist in one place. Fermentation in one facility. Downstream processing in another. Refining somewhere else again, frequently in a different country.

Each site is good at what it does. The system as a whole is expensive, slow and coordination heavy. Material sits in transit. Schedules at three plants have to align, and none of those plants is scheduling around us. Every handover is a place where a specification can be misread or a parameter can drift.

The visible cost is money and time. The cost that matters is iteration.

A campaign organised across multiple sites takes months to arrange and is expensive enough that it cannot be treated as an experiment. You get very few attempts per year, and every one has to succeed rather than teach you something. That changes what you are willing to try: when an attempt is scarce, you run the version most likely to work rather than the version that would tell you the most.

Process development is fundamentally iterative. Change one variable, observe, change the next. When that loop takes a quarter, progress is measured in years.

A facility with every step under one roof does not primarily save us money, though it saves a great deal. It returns the loop. A campaign stops being a logistics operation and becomes a controlled experiment again, with fermentation, downstream and refining in the same building, under one team, on a schedule we set.

That is what converts a process that works into a specification we can guarantee.

Why both, and not either

The model we are building is not a demo facility replacing external manufacturing. It is the two doing different jobs.

Our own facility carries development, integration and the process definition: the work where iteration speed and control of every step determine how fast we improve, and where the specification is set.

External manufacturing carries volume. Once a process is defined and locked, running it at scale is a capability the industry already has, and we would rather buy that capability than build it twice.

This is what gives the model its shape. We do not have to fund capacity ahead of demand, and we are not dependent on any single external site for supply.

The same logic, one stage further

The thinking behind the first commercial plant follows directly.

The instinct at this point is to design a large standalone facility. We think that is the wrong shape. A greenfield plant means building utilities, logistics and a workforce from nothing, on a site chosen for land availability, then shipping feedstock to it.

The alternative is to site production next to existing food manufacturing: close to the side streams it runs on, the utilities it needs and the infrastructure already there. Lower capital. Shorter logistics. A feedstock supply that arrives rather than travels. We have a co-location agreement at Langemark with FrieslandCampina, formerly Milcobel.

For that to replicate at a new site, three conditions have to hold. Enough feedstock locally. Access to utilities. A host willing to have us there. Everything else we bring.

Replicability, not capacity, is what makes an industrial biotech model investable. A single large plant is one asset. A model that can be reproduced next to existing infrastructure is a system.

Readiness is a team question before it is a capital question

None of this works with scientists alone. It takes technicians, analytical experts and pilot line operators, and that group holds most of the knowledge about what actually happens when equipment runs.

Two examples of what we mean by hiring for it. Jurgen Snelders, our pilot line supervisor, ran the pilot plant at NIZO and was operations manager at Mérieux. Anton Happel, our manager process technology, managed process technology at the Bioprocess Pilot Facility and ran development programmes at DSM.

What that background contributes is not enthusiasm for the technology. It is an instinct for what breaks. People who have already been responsible for equipment that has to work on Monday morning change what a company is willing to promise a customer, and when.

The teams that stall are usually the ones that hire operations after they have already committed to a process.

What readiness actually means

We can develop a process and prove it stable at 400L with our own downstream. We have run it at 120,000L, more than once, and validated the material in customer applications. We have the operational experience to run a facility, and a co-location agreement for the first commercial site.

What we do not yet have is every step under one roof, and that is the last structural gap between a process that works and a supply the industry can rely on.

It is the only thing still standing between the two, and it is the thing we are building next.

NoPalm Ingredients produces REVÓLEO™, a specialty fat made through biomass fermentation of oleaginous yeast on upcycled agrifood side streams.