Quick answer: A process that has worked once is not yet a production process. Commissioning is generally unlikely to be a core R&D activity where established supplier guidance and existing knowledge allow the required operating outcome to be determined in advance. A core R&D activity may exist where no established basis predicts whether an operable window — temperature, residence time, shear, moisture — exists at production scale that holds output within specification across varying feedstock, because scale changes the transport phenomena and only a systematic progression of work conducted to generate new knowledge can determine it. Which of the two a given campaign is turns on a dated pre-trial knowledge review, and each activity has to be assessed against the statutory tests by the company, which self-assesses its own claim.
24 August 2026 — this article describes the current rules. The 2026-27 Federal Budget announced R&DTI changes; the ATO states the measure is not yet law, and industry.gov.au states the changes will apply to income years starting on or after 1 July 2028.
Two facts sit behind almost every difficult scale-up. A process specification is not a set of numbers but a region — the volume in the space of controllable variables inside which the product still meets specification — and what matters is its width, not its centre. And it moves when the equipment gets bigger, because heat, mass and momentum transport do not scale together. A pilot result shows a point inside the region exists on that machine; it does not show the region exists on the next one, nor that it is wide enough to absorb lot-to-lot variation in the feed.
What this article covers
This is about process-window characterisation and continuous manufacturing scale-up under variable feedstock composition, in general. It excludes food and beverage processing, where reformulation, shelf life and sensory specification raise a different set of questions — see food and beverage process development — and construction 3D printing and robotic build processes, where the workpiece is the structure itself — see construction robotics and additive manufacturing. The examples here are deliberately non-food.
The Statutory Test the Activity Has to Meet, Stated in Full
Eligibility under the R&D Tax Incentive is assessed activity by activity, not project by project. Under s 355-25(1) of the Income Tax Assessment Act 1997, core R&D activities are experimental activities whose outcome cannot be known or determined in advance on the basis of current knowledge, information or experience, but can only be determined by applying a systematic progression of work that is based on principles of established science and that proceeds from hypothesis to experiment, observation and evaluation, and leads to logical conclusions; and that are conducted for the purpose of generating new knowledge, including new knowledge in the form of new or improved materials, products, devices, processes or services (business.gov.au; ITAA 1997).
Process engineering supplies a settled body of principle for the established-science limb: viscous dissipation and the energy balance, residence time distribution theory, film and penetration theories of mass transfer, and degradation kinetics are settled principles a hypothesis can be built on — although whether the limb is satisfied turns on how the particular progression is designed, conducted and evaluated, not on the maturity of the discipline. The first limb is the hard one, because much commissioning work has a determinable outcome: the supplier has a recommended envelope, the material a published processing guide, and the task is to find setpoints inside a region everyone knows is there. Turning knobs until the product passes is not a systematic progression based on established science, however carefully the runs are logged.
Why the Laboratory Window Does Not Survive the Change of Scale
Energy generation is a volume term; energy removal is a surface term: In a screw extruder, stirred reactor, mill or kneader, mechanical work dissipates throughout the process volume, which in geometrically similar equipment grows roughly with the cube of the characteristic diameter, while heat leaves through a wall whose area grows roughly with the square. Specific heat-transfer area therefore falls roughly in inverse proportion to diameter: on a small machine the wall sets the material temperature, on a large one the material increasingly sets its own. A window recorded in barrel setpoints — the way it is almost always recorded — does not transfer at all. Only one expressed in measured material temperature has a chance.
Degradation follows the tail of the residence time distribution, not the mean: Chain scission, hydrolysis and unwanted side reactions are kinetic: rate rises steeply with temperature and accumulates with time at temperature. Larger equipment often carries proportionally more slow-moving volume near walls, in transitions and in die adapters, so the distribution can broaden and its tail lengthen — but that is a risk arising from the particular geometry, flow regime and transitions, not a consequence of size alone, and it has to be measured rather than assumed. Where it does happen, a stabiliser package qualified against a narrow laboratory distribution can be exhausted by the tail while the mean residence time still looks similar.
Surface-renewal-limited steps lose specific capacity: Stripping moisture or residual volatiles depends on how much fresh interfacial area is created per unit time in the partially filled vent. That rate is set largely by geometry and rotation rather than by mass flow, so stripping capacity per unit of throughput tends to fall as the machine grows. Moisture a small machine removes without effort can survive to the die at production scale, and the voids and surface tearing it causes present as a forming problem rather than a transport problem.
The similarity criteria are over-determined: Specific mechanical energy, mean residence time, peak shear rate, specific heat-transfer area and specific surface-renewal rate cannot all be held constant across a change of scale, because they scale with different powers of the diameter. Any scale-up rule holds one or two and lets the rest move. Which of them dominates for a particular material and specification is a property of that material, and no general theory supplies it.
That is only half the argument, and the weaker half. Imperfect predictability in the abstract does not establish that a particular outcome was not determinable in advance. The first limb asks about current knowledge, information or experience, so the practitioner's first step is a knowledge review, done and dated before the trials: what the equipment supplier's own scale-up data covers and where it stops; what the resin, additive and equipment suppliers publish for the grades and geometries actually in use; what the literature and any comparable installation give; what the company already knows from its own lines; and what a competent professional in the field could reasonably have predicted from all of that. Where the review shows the answer was already available, the work is commissioning however novel it feels. Where it shows the published basis runs out — typically at the point where a specific material, a specific contamination profile and a specific geometry meet — the review is what makes "the outcome could not be determined in advance" a finding rather than an assertion.
Variable feedstock turns a point problem into a width problem. When the incoming material varies in molecular weight distribution, contaminant fraction, moisture, mineral content or prior thermal history, the requirement is not that a setting works but that the window is wider than the disturbance. Two lots can share the same headline index — a single-point melt flow rate, say — and behave differently, because that index is one low-shear point and the difference lives in the distribution behind it. Establishing the window means establishing its edges on the worst lots in the envelope, and top and bottom edges are usually set by different mechanisms.
Much scale-up sits outside all of this: a characterised single-source feed, a well-understood material and a supplier-supported line is a commissioning exercise with a determinable outcome. See what does not qualify.
A Worked Example (Hypothetical and Illustrative Only)
Invented to show where the boundary falls. It is not a real project and the numbers are illustrative. It says nothing about whether any actual claim would be accepted.
An Adelaide compounder makes an injection-moulding grade for materials-handling crates and electrical conduit — non-food-contact applications. The baseline runs at 15% post-consumer recyclate on a 92 mm twin-screw line. The company wants 60%, using kerbside-derived material, a stream whose scale and recovery pathways are set out in national plastic flows reporting (DCCEEW).
The knowledge review & feedstock envelope (before any trial): Line supplier's scale-up data covered virgin homopolymer and a single filled grade at contamination levels an order of magnitude lower. Twelve incoming lots characterised: melt flow rate 0.55–1.35 g/10 min; moisture 0.09–0.62%; PP contamination 0.8–4.6%; mineral ash 0.4–3.1%; carbonyl index 0.008–0.041; OIT 2–11 min at 200 °C.
The target, recorded before the first run: At 60% recyclate and ≥ 800 kg/h: melt flow rate 0.72 ± 0.08 g/10 min; notched Izod impact ≥ 9.0 kJ/m²; gels ≤ 40 per m² in 100 µm film; volatiles ≤ 250 ppm; pellet OIT ≥ 22 min at 200 °C. Pass/fail on all 12 lots, with triplicate runs at both envelope extremes.
Held and varied: Held: screw geometry per run block, die/adapter, 60% loading, melt temp via immersion probe in adapter (not barrel setpoints). Varied: screw speed (220–420 rpm), barrel profile, throughput (600–900 kg/h, SME 0.17–0.31 kWh/kg), vent vacuum, drying, stabiliser, and compatibiliser (0, 0.5, 1.0% MAPE).
Run A — Constant SME scale-up rule (Failed): 0.24 kWh/kg matched production scale at 840 kg/h gave adapter melt temp of 238 °C (vs 209 °C on 26 mm machine). OIT fell to 6–9 min, impact to 5.1–7.8 kJ/m². Morphology showed PP domain size improved (1.4 vs 2.6 µm), proving degradation, not dispersion, was the root cause. Ruled out: constant SME as a transfer rule and morphology as the primary failure mode.
Run B — Lower barrel setpoints: Dropped melt temp to 221 °C. Gels jumped to 260/m² on high-PP lots (unmelt). Established window bounds: lower bound set by PP melting floor, upper bound set by thermal oxidation. Window on worst lot was only ~3 °C wide (222–225 °C), impossible to control in production without widening.
Runs C to F — Window widening: Distributive mixing elements reduced SME (224 °C at 780 kg/h). Two-stage venting in series at −70 kPa dropped volatiles to 210 ppm. 0.35% phosphite/hindered-phenol package raised oxidation ceiling by ~6 °C (to 231 °C). 0.5% compatibiliser lowered unmelt bound by ~3 °C (to 219 °C) and boosted impact to 9.6 kJ/m². Combined additive package widened worst-lot window from ~3 °C to ~12 °C.
Result: Operable window defined in measured adapter melt temp 219–231 °C at 0.19–0.23 kWh/kg with two-stage venting, 0.35% stabiliser, 0.5% compatibiliser, and a feed-forward rule (700 kg/h for PP > 3% or carbonyl index > 0.030). Pre-drying adopted for 2 extreme lots. All 12 lots passed physical targets.
Throughput result & boundary: Sustained throughput was 700–780 kg/h (missing 800 kg/h target due to thermal ceiling). Candidate core activity spans the dated unknown and Run A through to window characterisation. Routine lot testing, equipment modifications, and production pellet sales require separate supporting-activity analysis under s 355-30.
Where the Exclusions and the Supporting-Activity Test Bite
Two exclusions in s 355-25(2) sit close to this work:
Paragraph (f) — Routine testing and standards
Paragraph (f) covers activities associated with complying with statutory requirements or standards, including routine testing and analysis of materials, components, products and processes. Incoming-lot acceptance testing and batch release testing may fall within that exclusion where they are undertaken in connection with an applicable statutory requirement or standard; otherwise, their treatment must be assessed on their own facts.
Paragraph (g) — Reproduction of a commercial product or process
Paragraph (g) covers activities related to reproducing a commercial product or process by physical examination of an existing system, or from plans, blueprints, detailed specifications or publicly available information — where a scale-up that is really the implementation of a licensor's package lands. Activities of that kind are generally unlikely to be core R&D activities on those facts, subject to each activity's own facts and the statutory tests.
Exclusion from core does not by itself establish whether an activity satisfies the separate supporting-activity tests. A supporting R&D activity must be directly related to core R&D activities (s 355-30(1)), and s 355-30(2) adds a dominant purpose bar for activities of a kind referred to in s 355-25(2), activities that produce goods or services, and activities directly related to producing goods or services (business.gov.au). In the hypothetical above, a production-line campaign that turns out hundreds of tonnes of saleable pellet, and the changeovers and packing around it, would each have to be considered against that bar — a question for the company and its tax adviser, activity by activity.
Two practical notes: Measurement traceability to national standards can strengthen comparability across runs and scales and the reliability of the resulting technical record (National Measurement Institute). And the R&DTI uses "feedstock" in its own sense — expenditure on goods transformed or processed in R&D activities to produce marketable products — which is not the sense used above; the ATO explains how it is treated (expenditure you can claim).
Where an RSP Fits
An RSP is a scientific or technical service provider registered in specific research fields that a company can engage to conduct R&D activities on its behalf (business.gov.au). Ignition Research is registered in Manufacturing Engineering (ANZSRC 4014) among other fields, and our work on a campaign like the one above sits upstream of the line: running the knowledge review that decides whether there is a question to answer at all, defining the feedstock envelope so the worst lot is designed for rather than discovered, choosing which similarity criterion the campaign holds and which it lets move, and specifying the measurements — material temperature rather than wall setpoint — at the resolution the criteria require. See manufacturing engineering research.
There is also a threshold point for smaller programmes: Eligible R&D expenditure incurred to a registered research service provider may be included even where the entity's total notional R&D deductions are below the usual $20,000 threshold, subject to the requirements in s 355-100(2), including that the expenditure is deductible under s 355-205, the provider is not an associate, and the services fall within a research field for which the provider is registered. In mechanism, where a company's total notional R&D deductions for an income year come to less than $20,000, s 355-100(2) works the offset out on a substituted base instead: certain kinds of expenditure, including expenditure to a registered research service provider that is not an associate, for services within a research field the provider is registered in. See claiming R&D under $20,000. The offset tiers are covered in our article on the refundable and non-refundable offset.
Talk to Ignition Research before the first production-scale campaign is run, while the hypothesis, the acceptance criteria and the instrumentation can still be settled against the question they are meant to answer. As a Registered Research Service Provider at Lot Fourteen in Adelaide, we design and conduct the experimental programme and produce the technical record while the work is happening. We are not a registered tax agent: your company self-assesses and remains responsible for its own claim. Get in touch.
Frequently Asked Questions
Q: Is scaling a process from pilot to production an eligible R&D activity?
A: Often not, as a core R&D activity. Where the equipment supplier's envelope, the material's processing guide and the licensor's package between them predict the operating conditions, the outcome is determinable in advance and finding the setpoints is commissioning. A core activity may exist in the narrower case where no established basis predicts whether an operable window exists at production scale that holds output within specification across the feedstock variation, and that can only be determined by a systematic progression of work conducted to generate new knowledge. A dated pre-trial knowledge review — supplier data, published processing guidance, literature, comparable installations, internal experience — is what distinguishes the two. You self-assess.
Q: Why does a laboratory result not transfer to a production line?
A: Because the transport terms scale differently. Viscous dissipation and reaction heat are volume terms that grow roughly with the cube of the characteristic diameter; heat removal through the wall is a surface term that grows roughly with the square, so specific heat-transfer area falls as equipment gets larger. Depending on the geometry, the flow regime and the transitions involved, the residence time distribution can also broaden and its tail lengthen — degradation follows the tail rather than the mean — though larger size alone does not establish that, and it has to be measured. Surface-renewal-limited steps such as devolatilisation lose capacity per unit of throughput. And because specific energy, residence time, peak shear and specific area cannot all be held constant at once, any scale-up rule fixes some and lets the others move.
Q: What is a process window in manufacturing?
A: The region in the space of controllable variables — temperature, residence time, shear rate, moisture, dose — inside which the output still meets its specification. Its practically important property is its width, not its centre, because width is what absorbs variation in the incoming material and in the equipment. The edges are usually set by different mechanisms: an upper bound from degradation or side reaction, a lower bound from incomplete melting, dispersion or conversion.
Q: Is commissioning a production line a core R&D activity?
A: Generally unlikely, subject to the activity's own facts and the statutory tests. Commissioning demonstrates that installed equipment performs as specified, which is ordinarily a determinable outcome, and where it amounts to implementing a process from plans, detailed specifications or publicly available information the exclusion in s 355-25(2)(g) is also relevant. Commissioning work may still be considered as a supporting R&D activity. Where the particular activity falls within s 355-30(2) — including because it produces goods or services or is directly related to producing them — it must also satisfy the dominant-purpose test.
Sources & Further Reading
legislation.gov.au — Income Tax Assessment Act 1997 — Div 355, incl. ss 355-25, 355-30 and 355-100
industry.gov.au — National Measurement Institute: Australia's measurement system
industry.gov.au — Advanced manufacturing and materials technologies
ABS — Australian and New Zealand Standard Research Classification (ANZSRC)
Related: manufacturing engineering research · what does not qualify · what an RSP is · claiming R&D under $20,000 · refundable vs non-refundable offset · R&D Tax Incentive in Adelaide · more Insights
This article is general information from a Registered Research Service Provider about the R&D Tax Incentive. It is not tax, legal or financial advice; eligibility depends on your circumstances and you should self-assess and seek your own advice.
Thinking about a project like this?
If you're weighing up an AI, software or technical improvement project and can't tell yet whether it's implementation or research, start with a quick read on where it sits.