A natural graphite anode sample looks great in the lab. Capacity is on target, first-cycle efficiency is solid, and the particle size sits right where you want it. So the team decides to scale up. A few months later, the first production batches come off the line with a wider particle size distribution, tap density that swings from batch to batch, magnetic contamination above spec, and electrochemical results that differ from one lot to the next, even though the recipe hasn’t changed.
This is a familiar scale-up trap in natural graphite anode production. The lab data usually isn’t wrong. It just answers a different question than the one production asks. The lab shows that a process can work. Production has to show that it works every day, at a cost you can afford. Pilot production is what bridges the two.
For producers new to this industry, the hard calls come early. How many lab rounds are enough? How big should the pilot be? Can you skip it and build the plant? This article doesn’t cover the full plant-building process or an equipment list. It focuses on one question: at each stage from lab to pilot to full scale, what does the work need to prove, what changes, and where do things tend to go wrong?
Scaling up a natural graphite anode process happens in three stages. Lab work proves the process is feasible. Pilot production proves it runs reliably under near-commercial conditions and generates the engineering data plant design needs. Commercial production proves cost, yield, and batch consistency. Strong lab results alone don’t guarantee a successful scale-up.
| Stage | What it needs to prove | Typical failure point |
| Lab | The process route works in principle and the product can meet spec | Small sample sizes and hand-picked feedstock don’t represent real production |
| Pilot | The process runs steadily on near-commercial equipment, and you capture engineering data such as mass balance, energy use, and yield | Pilot feedstock differs from production feedstock, or the run produces samples but no engineering data |
| Commercial | Cost, yield, batch-to-batch consistency, and stable continuous operation | Skipping the pilot leaves the design without a solid data basis, so the plant needs repeated fixes after startup |
What Each Stage Must Prove
First, a point that often gets overlooked: “pilot” has no universal size.
Publicly documented pilots span several orders of magnitude. ANZAPLAN ran a pilot program for Westwater Resources on flake graphite from Alabama, processing about 30 tons of graphite concentrate in a custom-built pilot plant in under three months. The goal was to collect data for engineering a production plant and to make multi-ton quantities of battery-grade material for application testing. The project dates to around 2021, so it’s a useful early example of what a ton-scale pilot looks like, not a picture of where the industry stands today.
At the other end of the range, a 2026 study in Carbon Trends converted natural flake graphite into coated spherical purified graphite in pilot batches of 5 to 10 kg. The team used a chlorination route that avoids hydrofluoric acid, and the final carbon content exceeded 99.95 wt.%.
Both are called pilots, but they answer different questions. One feeds a plant design. The other validates a process route. So before you start, decide what this round has to prove:
- If you’re still confirming that a process route can produce an acceptable product, a kilogram-scale pilot may be enough.
- If you’re already designing a production line and need mass balance, energy consumption, and equipment parameters, you need a pilot that looks more like production equipment and can run continuously.
A U.S. Department of Energy project record offers another reference point. The project aims to scale up processes for converting natural and recycled synthetic graphite into anode material, plans a ton-scale pilot plant, and includes feasibility and front-end engineering design (FEED) studies. In other words, the pilot is tied to engineering design from day one. It isn’t a standalone experiment.
Why Lab Results Don’t Transfer Directly
You can’t multiply lab results by a scale factor and call it a plan. Four things get in the way.
Heat and mass transfer change. In the lab, a few dozen grams to a few kilograms of material heat evenly, and residence time is easy to control. At scale, the bed gets deeper, temperature gradients appear, and the material moves through the equipment in a completely different way. The same temperature profile can produce different products on a lab unit and a production unit.
Feedstock stops being a hand-picked sample. Labs often work with selected, consistent material. Production draws on different lots and sometimes different mines, so impurities, contents, and particle morphology all vary. The process has to tolerate that variation, and you only learn how much it can tolerate by running real feedstock at pilot scale.
Equipment changes shape. Labs typically use batch equipment. Production runs continuously. Batch-to-batch handoffs, feed and discharge methods, dust control, and sealing are problems that simply don’t exist on the lab bench.
Scale itself creates new problems. This is a well-established idea in process scale-up: a pilot surfaces problems that are invisible at small scale and stress-tests both the science and the systems before an expensive commitment to full-scale production. That source discusses scale-up in general terms rather than graphite specifically, but the logic applies to anode materials as well.
Lab data is necessary, but it isn’t sufficient. It tells you which direction to go. It doesn’t give you engineering parameters.
What Changes at Each Process Step
A typical natural graphite anode process includes crushing and classification, spheroidization, purification, coating and granulation with carbonization, and demagnetization and screening. The principles behind each step are covered in separate articles on our site, so this section only covers what changes at scale. For the full process flow, see our guide to the complete natural graphite anode material production process.
| Process step | Lab focus | New variables at scale | Pilot data to collect |
| Crushing and classification | Reaching the target particle size | Feed rate, equipment load, dust, wear | Throughput, energy use, stability of size distribution |
| Spheroidization | Sphericity, tap density | Continuous feeding, recirculating load, fines generation | Yield, fines fraction, tap density variation |
| Purification | Reaching the target purity | Reagent volume, waste liquid and gas handling, equipment corrosion resistance | Batch-to-batch purity variation, reagent consumption, waste volume |
| Coating, granulation, and carbonization | Coating uniformity, first-cycle efficiency | Bed depth, atmosphere control, residence time distribution | Temperature profile, batch consistency, energy use |
| Demagnetization and screening | Magnetic contamination level | Balance between throughput and removal efficiency | Contaminant level, screening efficiency |
Purification deserves a closer look, because it faces the tightest environmental and cost constraints at scale.
Publicly available sources describe three broad approaches, each with trade-offs:
- Chemical purification. It scales easily. Hydrofluoric acid, for example, dissolves the quartz, feldspar, and other silicate phases common in natural graphite concentrates. But HF is highly toxic and corrosive, which creates serious worker safety risks, and the process generates large volumes of acidic liquid and solid waste.
- Thermal purification. This method runs at extreme temperatures, typically between 2,400°C and 3,000°C depending on the impurity profile and target purity, and it consumes a lot of energy.
- HF-free routes. The chlorination route mentioned above has been demonstrated at kilogram scale, but moving it to ton scale will take more engineering data.
The takeaway: don’t choose a purification route based on lab purity numbers alone. Waste treatment capacity, local environmental requirements, and energy cost all affect whether a route is viable at scale. Those factors barely show up in the lab, yet they often decide whether a project gets built.
For more on this topic, see our articles on purity control for natural graphite anode materials, spheroidization, and coating and granulation.
Designing a Pilot Line: What to Include and What to Skip
A good pilot line isn’t a shrunken production line. It’s the smallest set of equipment that can validate your key variables.
Design principles
- Decide what you need to validate first, then choose the equipment. If the goal is to understand the yield relationship between spheroidization and purification, you don’t need packaging and finished-product handling on day one.
- Build in a wide adjustment range. The value of a pilot is finding the process window, so speed, temperature, and feed rate should be adjustable rather than fixed at a single operating point.
- Plan sampling points and data collection up front. You should be able to sample before and after every key step and log temperature, flow, and energy use. When the pilot ends, you should have a data set, not just a few bags of samples.
- Favor flexibility. Pilots often test different feedstocks and recipes, so equipment should be easy to change over and clean to avoid cross-contamination.
A note on rotary kilns. In coating, granulation, and carbonization, the rotary kiln is the core thermal equipment. From pilot to production, the differences show up in throughput and bed depth, zoned temperature control, atmosphere management and sealing, and how material goes in and out. On a small unit you can often get away with compromises in these areas. At scale, they directly affect batch consistency. So the pilot kiln should reflect the key characteristics of the production kiln, not just be able to produce samples. For the basics, see what a rotary kiln is and our article on the laboratory rotary kiln.
Site and permitting. The site is part of pilot design too. The Department of Energy project mentioned earlier is planned for a brownfield site with existing chemical and waste disposal permits, with parallel lab research on reducing waste, reclaiming water and acid, cutting heat-treatment energy, and developing low-cost carbon coatings. A pilot isn’t only an equipment question. It involves permits, waste handling, and utilities.
Quality Metrics to Track from Lab to Plant
Track at least the following metrics throughout scale-up, and use the same test methods at every stage. Otherwise you can’t compare the data.
| Metric | Why track it | Common problem at scale |
| Particle size distribution (D10/D50/D90) | Affects electrode processing and cell performance; see particle size and battery performance | Distribution widens, or fines and coarse fractions increase |
| Tap density | Reflects the effect of spheroidization and particle packing | Batch-to-batch swings |
| Specific surface area | Linked to first-cycle efficiency and side reactions | Uneven coating pushes it too high or too low |
| Fixed carbon and impurity content | Determines purity grade | Feedstock variation or unstable purification efficiency |
| Magnetic contaminants | Affects cell safety | Equipment wear, higher demagnetization load |
| First-cycle efficiency and capacity | The end result in the cell; see electrochemical performance | Gaps between lab samples and production material |
For newcomers, the more important question is batch consistency. Don’t look at the best result from a single batch. Look at the range across many consecutive batches. A simple place to start is to record key metrics for every batch, plot them as a trend, and check whether they hold steady within an acceptable band. Stability is worth more than a single peak result.
Common Scale-Up Pitfalls
Skipping the pilot and building the plant. This is the costliest mistake. Without pilot data, equipment selection and line design have no solid basis, and the plant often needs repeated rework after startup.
Running the pilot on different feedstock than production. A pilot on hand-selected material produces data that doesn’t represent real production. If production will use several sources, the pilot should cover them.
Focusing only on electrochemical performance. Meeting product specs is a necessary condition, not a sufficient one. Yield, energy consumption, and waste volume determine unit cost.
Ignoring environmental and waste constraints. This matters most in purification. Hazardous reagents, high energy use, and waste generation are drawing more regulatory attention worldwide, which is one reason cleaner purification routes are getting attention.
Underestimating pilot time and budget. A pilot isn’t just the time it takes to make samples. It includes equipment design, procurement, installation, commissioning, operation, and data analysis. In the ANZAPLAN case above, the “under three months” refers to the processing run itself, not the design and construction that came before it.
From Pilot Data to Engineering Design
The end product of a pilot is data you can use directly to design a production line:
- Mass balance: input, output, and losses at every step.
- Energy data: consumption of electricity, fuel, water, and reagents.
- Equipment parameters: throughput, speed, temperature profile, and residence time.
- Quality window: which process changes push product metrics out of spec.
- Waste data: the types and volumes of liquid, gas, and solid waste.
These numbers drive equipment selection, line layout, and utility design. They also form the basis for judging investment and risk. Two articles already on our site show real-world examples from different angles: the Falcon Energy Materials anode plant study looks at project economics, and the GreenRoc pilot plant launch looks at a pilot startup.
Once you have enough pilot data and a clear process window, you’re ready to move on to full plant development.
As an equipment manufacturer and EPC contractor, ZD’s role at this stage is to turn validated pilot parameters into a production line that can be manufactured, built, and operated. The stronger the early data, the fewer changes you’ll need later.
Frequently Asked Questions
How long does pilot production take for a graphite anode process? There’s no single answer. It depends on the goal and scale of the pilot. In the ANZAPLAN case, the processing run for about 30 tons of graphite concentrate took under three months, but that doesn’t include the time to design and build the pilot plant. Kilogram-scale validation pilots are typically shorter.
How big should a graphite anode pilot line be? It depends on what you need to prove. If you’re validating a process route, kilogram scale may be enough. If you need data for plant engineering, you’ll want a scale and equipment format closer to production, one that can run continuously. Define the pilot’s goal first, then work backward to the size.
Can I skip the pilot stage? In theory, yes, but the risk is high. Without a pilot, your production line design rests on lab data alone, and lab data can’t reflect the effects of continuous operation, feedstock variation, and scale.
What is the difference between lab-scale and pilot-scale? Lab scale shows that a process works in principle, with the focus on whether it can make an acceptable product. Pilot scale shows that the process runs steadily under near-commercial conditions, with the focus on collecting engineering data and exposing scale-up problems.
Next Steps
From lab to full scale, what separates successful projects from struggling ones is rarely a single machine. It’s whether the team asked the right question at the right stage: What does this step need to prove? Is the data strong enough to support the next investment?
If you’re at this point, here are three ways to move forward:
- Self-check first. Request our Pilot Validation Checklist and see whether your lab data is ready for scale-up.
- Talk it through. Book a free technical consultation, and our engineers will help you define pilot goals, scale, and a sampling plan.
- Move to execution. If you’re already planning a pilot line or a full production line, contact us for an equipment and EPC evaluation that connects pilot data to production line design.