Scaling a manufacturing process from lab to production is rarely linear. Variables like material flow, equipment behavior, and throughput constraints compound at volume in ways that small-scale testing can't predict. The companies that scale successfully follow a structured crawl, walk, run approach (feasibility, pilot, production), and never skip the pilot. One successful run is luck. Three repeatable runs is a process.
What works in the lab doesn't always work in production, and that gap can be costly.
Scaling a manufacturing process from R&D to full production is one of the most complex transitions a company can face. In a recent episode of Toll Manufacturing Minute, CPS Vice President of Business Development Albert Medoro and host Jen Lepore explore why scale-up is inherently non-linear, where companies most often go wrong, and what a structured approach to process development actually looks like.
From material flow and equipment differences to throughput, yield, and cost constraints, this conversation highlights what separates successful scale-ups from expensive failures.
Most processes perform well at lab scale because conditions are controlled and forgiving. But as volume increases, variables compound quickly.
At larger scale:
“Your margin for error really gets a lot tighter at scale,” Albert explains. “What worked at 1–5 kg may behave completely differently at thousands of kilograms. Assuming otherwise is likely the most common and costly mistake in manufacturing.”
One of the most persistent assumptions in manufacturing is that a process will scale proportionally. Theoretically it should work. In reality, it rarely does.
A process that performs perfectly in a small mill may lose consistency at higher throughput, produce wider particle size distributions, or require entirely different equipment to meet the same specs. This is especially true in milling, where equipment type in addition to size can determine success or failure.
“‘The mill is the mill’ is not a safe assumption,” Albert points out. “For example, a D50 under ten microns with a tight distribution isn’t a hammer mill job. It's a pretty complex milling job. Misidentification means building a process around the wrong equipment, and nothing will fix a foundational mismatch.”
Equipment selection. Not all mills or processing systems are interchangeable. Choosing the wrong technology early can lock in inefficiencies or make specs unattainable at production volumes. Most companies default to running mills at OEM standard which works well for high-volume, single-product runs. But as Albert notes, “There's a whole range of performance that you really can unlock if you run outside of parameters and beyond standard.”
Material behavior. A material that flows freely in a small batch may bridge or clog in a larger system. For instance, instead of top feeding, material may be run through a screw conveyor from the floor, and be moved 20 feet on the line. That presents an entirely different scenario than when that same material was run on a benchtop scale. What seemed like a minor handling detail at feasibility can become the one of the hardest parts of the entire production process.
Packaging. In the feasibility phase, material is generally packed out into a one kilogram liner strictly for testing purposes. At commercial scale, pack outs may be into a 55-pound valve bag. The packaging change itself is a constraint.
Yield loss. Lab yields can drop significantly at scale. “You could go from something in a lab scale that’s at 98-99%, but in production that drops to 84%. Some companies can manage that, but most companies can’t since that margin will catch up to you pretty fast,” Medoro said.
Throughput constraints. Throughput is often the defining factor in commercial viability. “Throughput drives yield,” Albert explains. “There are instances where you’re trying to hit a throughput rate that’s viable in the market. In order to hit that, you’ve sacrificed 15–20% yield.” Finding the right balance between speed and output quality is exactly what feasibility and pilot trials are designed to solve.
A process can be technically successful and still fail commercially.
CPS worked with one customer whose validated, technically sound process was commercially impossible at market cost. The only path to a viable price was collapsing a three-step process down to one or two steps, which the customer initially resisted. CPS ran the simplified process on their own and delivered two side-by-side samples with identical performance at half the cost. “Once we provided proof of results at 50% of the cost, the decision to move forward was easy,” says Medoro.
The lesson: Companies need to stay open to process simplification, not solely optimization.
Pilot trials are where scale-up success is actually determined. They’re also the step most commonly skipped, typically because they cost money and time.
Companies can usually justify feasibility spending since the process must be proven to work before sending samples to a customer. But the pilot stage often gets cut. Albert pushes back on that reasoning, stating:
“Pilot testing is where you catch all of the things that are going to happen in production, like material buildup, wear patterns on the equipment, inconsistencies in the feed rates.”
Albert cautions that pilot trial isn’t a one-and-done. “One successful pilot run could very much just be luck. Three successful runs is repeatable. And that’s what you want to strive for.”
On the cost question, Albert offers a reframe: “Don't look at a dollar per pound price from a pilot perspective. Look at it from the totality of the cost of the product. I’d much rather spend $10 per pound and only produce 100 pounds versus having to be liable for 100,000 pounds at $5 a pound."
The most successful projects follow a structured progression:
Crawl (Feasibility): Can the process work at all?
This stage is about confirming viability, not perfecting parameters. CPS sometimes runs what Albert calls a “dirty trial,” meaning knowingly allowing minor cross-contamination that won’t affect performance to confirm whether the target particle size is achievable.
Walk (Pilot): Can the process work consistently and at commercially viable cost?
This stage is where multiple batches are run, edge cases are tested, and the process is pressure-tested against real production conditions. “You can't just run one pilot test and assume you have enough data to provide a commercially viable cost,” explains Medoro. “Humidity levels may have changed, or a feeder drifted. If data from three different runs and three different batches isn’t captured, it leaves room for some potential downfalls.”
Run (Production): Can the process scale reliably and profitably?
By this stage, the unknowns have been identified and addressed, and the project is cleared for production.
The projects that go well are the ones that go through each of these steps.
Skipping steps may appear to save time upfront, but it usually creates far more expensive delays later. CPS will sometimes decline to take on a project if a customer insists on skipping straight to production: “We're simply not willing to take that risk. You can't predict what you can't predict.”
HOW TOLL PROCESSORS ADD VALUE IN SCALE-UP
Experienced toll processors bring something internal teams often lack: pattern recognition across hundreds of materials and processes.
“Tollers know, over time, which raw materials are going to behave differently than others. And customers don’t always get that with their internal teams, especially a procurement team that’s cutting costs by swapping out materials,” Albert continues. “You may save a couple pennies on the price per pound, but the effects it has on the process could be substantial.”
The ideal time to bring in a tolling partner is during feasibility, not during a production crisis. An experienced toller can surface downstream issues early, or provide alternatives quickly to save both time and money before the stakes get high.
Ready to SCALE without the costly surprises?
Scaling a process from lab to production doesn’t have to mean expensive setbacks. The manufacturers that get it right aren’t lucky, they’re prepared. They know their critical constraints going in, they test before they commit, and they work with partners who’ve seen the patterns before.
If you’re planning a scale-up or evaluating your current process, don’t start from scratch, and don’t do it alone. Find the right partner to help navigate the jump from R&D to full production using our Toll Processor Evaluation Checklist.