Quick answer: Stacking modules accurately is engineering, and not by itself an experiment. Under s 355-25(1) of the ITAA 1997 a core R&D activity needs an outcome that could not be known or determined in advance and could only be determined by a systematic progression of work, conducted to generate new knowledge. At the factory-to-site interface that usually means one specific unknown: whether any joint detail holds its required performance across the gap range that tolerance stack-up and differential movement actually produce. You self-assess.
18 August 2026 — this article describes the current rules. The 2026–27 Federal Budget announced R&DTI changes to apply from 1 July 2028. Until then, the program continues under the current rules.
A module leaves a shed at Wingfield built to plus or minus 3 mm. The transfer slab it lands on was poured to plus or minus 10 mm across its footprint. Between those two numbers sits most of the technical difficulty in modular construction. The factory is a manufacturing environment with jigs and repeatability; the site is a civil environment with survey error, weather and a sequence that cannot be reversed. The interface is where the two error regimes meet and add up.
This article is about that interface specifically: cumulative tolerance across a stack of modules, the behaviour of the connections and joints that absorb it, differential movement over time, and the installation sequence that decides whether error is corrected or accumulated. It does not cover modular, prefabricated or DfMA delivery as a delivery model, novel structural systems on difficult sites, or construction robotics and automated fabrication — each has its own Insight and complements this one rather than repeats it.
The Tolerance on the Drawing Is Not the Tolerance the Building Gets
Tolerance stack-up is the accumulation of individually acceptable deviations into a total that may not be. Each contributor is small and in spec: module height, floor cassette thickness, bearing shim, levelling packer, transfer slab level, setting-out error, crane placement. Over eight levels they compound.
Two ways of adding them give very different answers. Worst-case arithmetic addition assumes every deviation lands at its limit in the same direction. Statistical addition — root-sum-square — assumes the deviations are independent and random, and generally returns a smaller combined value than worst-case arithmetic addition. Both are established methods. The engineering question, and often the genuinely undetermined one, is which model the particular factory and site actually obey, because a drifting jig, an operator habit or a batch of packers cut from one setting produce correlated error, and correlated error does not behave statistically.
The joint absorbs whatever comes out of that, usually while doing several things at once: transferring vertical load, maintaining acoustic discontinuity, holding the weather line, keeping a fire separation. Differential movement widens the range again over time — shrinkage and creep in a concrete core against a lighter module stack, moisture movement in timber, thermal movement on a façade line, differential settlement across a footprint. Sequencing is the fourth variable and the one most often left out: whether a stack is surveyed and corrected at every level, at mid-height or only at the top decides whether error is bounded or compounded, and so does the interval between a factory-applied seal being compressed and the module being landed.
What the Legislation Actually Asks
Section 355-25(1) of the Income Tax Assessment Act 1997 defines core R&D activities as 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 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). AusIndustry's guidance frames what is already known in terms of what a competent professional in the field could determine; that is guidance wording, not the words of the section.
The established-science limb is satisfiable here, which is why it is worth naming rather than skipping: statistical tolerance analysis, structural mechanics, hygrothermal and water-penetration physics, building acoustics and the viscoelastic behaviour of elastomeric seals are established sciences. What it rules out is the shim-and-see loop — land, measure, pack the difference, move on — which can be careful and well documented and still not be an experiment.
Closer to an outcome that could not be determined in advance
Ordinarily design, production or configuration
Whether any joint detail holds water resistance and acoustic separation at once across the gap range the measured, partly correlated stack-up and long-term movement actually produce
Detailing a proprietary joint system inside the gap range and movement capability the supplier has tested
Whether a connection transfers the design action as the bearing area varies across the as-built tolerance range, where published design methods and test data do not cover the behaviour
Designing a connection to a code method for a nominal fit, then specifying an installation tolerance
Whether the error distribution for a factory and erection sequence is independent or correlated, and what that implies for the gap range a detail must survive
Applying a standard tolerance table, or a survey accuracy from an instrument specification
The right-hand column is not a comment on difficulty. Setting out an eight-storey stack to a millimetre budget in a wind is demanding, and is generally unlikely to be a core R&D activity on those facts, subject to the activity's own facts and the statutory tests.
Why the Dominant Purpose Test Bites So Hard on a Construction Site
Supporting R&D activities are activities directly related to core R&D activities. But s 355-30(2) adds a hurdle: if an activity (a) is an activity referred to in s 355-25(2); or (b) produces goods or services; or (c) is directly related to producing goods or services — then it is a supporting R&D activity only if it is undertaken for the dominant purpose of supporting core R&D activities.
Production and site activities
Limbs (b) and (c) catch this field. Installation, craneage, setting out, surveying, shimming, sealing and propping may produce, or be directly related to producing, goods or services depending on the particular activity. Where that is the case, the additional dominant-purpose test applies if the activity is being assessed as supporting R&D.
Each limb is tested against the particular activity — never against the project, the site or the business as a whole. A programme is not "an R&D project" for these purposes; it is a set of activities each separately core, supporting or neither. Where a single lift both resolves a technical unknown and delivers a saleable storey, its dominant purpose is a question of fact.
Routine Testing, Calibration and the Standards Exclusion
Section 355-25(2)(f) excludes activities associated with complying with statutory requirements or standards, including one or more of the following: (i) maintaining national standards; (ii) calibrating secondary standards; and (iii) routine testing and analysis of materials, components, products, processes, soils, atmospheres and other things.
A modular programme may include activities of this kind, but paragraph (f) applies where the testing, calibration or verification activity is associated with complying with a statutory requirement or standard.
Paragraph (f) boundary
"Associated with" is broad, and this article does not assert a carve-out from paragraph (f) for development work. Whether work aimed at meeting a standard falls inside the exclusion is a question of fact, assessed activity by activity, and the company self-assesses. What can be said is narrower: the same rig on the same day can serve a conformance check and a measurement taken against a hypothesis fixed in advance, and those are different activities. The test method does not decide it.
Paragraph (g), which excludes activity related to reproducing a commercial product or process by physical examination of an existing system or from plans, detailed specifications or publicly available information, is nearby: both elements are needed, and a published method is not by itself a commercial product or process.
A Worked Example: An Inter-Module Joint at the Top of an Eight-Storey Stack
Hypothetical and illustrative. The numbers show what a worked investigation looks like; they are not a statement that the activity would be eligible, and the company self-assesses.
Baseline: An Adelaide modular builder has completed four-storey light-gauge steel stacks. Across 20 modules, height deviation had a standard deviation of 1.6 mm within a plus or minus 3 mm spec; bearing shims varied plus or minus 2 mm per interface; the transfer slab came in at plus or minus 10 mm. The horizontal inter-module joint measured 12 to 29 mm against a 20 mm nominal, and the factory-applied single-stage foam seal held.
What changed: The next project is eight storeys on a podium. Worst-case arithmetic addition gives 10 + 8 × (3 + 2) = 50 mm of accumulated deviation; RSS gives roughly 14 mm. The models disagree by a factor of three and a half, and the existing seal's tested envelope is 12 to 30 mm.
The unknown & pre-fixed target: Before any specimen was built, the acceptance criteria were recorded as: no water penetration at a 600 Pa static pressure differential held for 15 minutes; airborne sound insulation across the joint of not less than 55 (weighted level difference); and retained seal contact pressure of at least 20 kPa after six weeks under compression at 40 °C. Those had to hold across a specified test range of 12 to 38 mm, allowing for measured stack-up together with estimated long-term differential and thermal movement. The question was whether any detail satisfies all three at once across that range.
Held constant & varied: Held constant: Module geometry, floor cassette build-up, bearing detail, sealant and gasket product families, specimen size, instrument calibration, and the 600 Pa / 15 minute protocol. Varied: Joint architecture (single-stage face seal / open drained two-stage / compartmented two-stage), seal application point (factory or site), time under compression before landing (0, 14, 42 days), joint gap (12, 26, 34, 38, 42 mm).
Sequence:
Trial A — Existing single-stage factory-applied foam seal extended to wider range: Water held to 26 mm and failed at 31 mm; it also failed at 26 mm where the specimen had sat compressed for 42 days before assembly (contact pressure 11 kPa). Ruled out: single-stage face-sealed approach at this gap range, showing the binding constraint was time under compression before landing (a sequencing variable).
Trial B — Open drained two-stage joint (outer baffle, drained cavity, site-applied inner air seal): Water held at 600 Pa across every gap to 38 mm, but weighted level difference fell from 58 at 12 mm to 49 at 34 mm. Ruled out: continuously drained cavity as detailed, locating loss in a flanking path through the cavity.
Trial C — Compartmented two-stage joint (closed-cell baffles at module boundaries, same inner seal): At 12, 26, 34 and 38 mm: no water penetration at 600 Pa; level difference 58, 57, 56, 55; contact pressure 34 to 22 kPa after six weeks at 40 °C.
Trial D — Trial C at 42 mm (run to find the edge): The inner seal lost continuous contact at two of six points and water penetrated at 600 Pa.
The result actually reached: A detail satisfies all three criteria at gaps up to 38 mm under the conditions tested, but failed at 42 mm. The knowledge generated was the relationship between accumulated gap, cavity compartmentation and time under compression — plus finding that deviation tracked closer to arithmetic than RSS on upper levels due to correlated jig error, requiring a mid-height survey-and-correct step.
Where the boundary falls: The candidate experimental activity begins at the dated statement of the unknown and the search establishing that supplier envelopes, published tolerance tables and the company's own four-storey data did not answer it, and ends at the evaluation of Trial D. Outside it: routine dimensional QA of production modules, calibration of survey instruments, podium survey and setting out, production installation and sealing, and handover testing. Whether any of those qualifies as a supporting R&D activity depends first on whether it is directly related to a core R&D activity and, where s 355-30(2) applies, whether it is undertaken for the dominant purpose of supporting that core activity.
Where an RSP Fits
Research Service Providers are organisations registered to provide scientific or technical services to companies conducting R&D, in specific fields of research (business.gov.au). For a modular builder, an RSP supplies measurement and experimental structure: characterising the factory's actual error distribution rather than assuming one, and designing a trial sequence capable of resolving the unknown.
One point on the money side is specific to RSPs: qualifying expenditure incurred to a non-associate RSP may still be taken into account in determining R&D tax offset entitlement where total notional deductions are below the usual $20,000 threshold, provided the services are within a research field for which the RSP is registered — see claiming R&D under $20,000 — and using an RSP does not guarantee eligibility — you still self-assess. Offset rates, tiers and how they are worked out are covered in refundable vs non-refundable.
Talk to Ignition Research before the first module is landed — as a Registered Research Service Provider at Lot Fourteen in Adelaide, we help modular builders measure what their tolerance stack-up really does and design a trial sequence capable of resolving an interface question. We are not a registered tax agent: your company self-assesses and remains responsible for its own claim, with advice and lodgement handled by your tax adviser. Get in touch.
Frequently Asked Questions
Q: Is solving tolerance stack-up in modular construction eligible for the R&D Tax Incentive?
A: Not as a category. Managing tolerance is normal construction engineering. A core R&D activity requires an outcome that could not be known or determined in advance on the basis of current knowledge, information or experience, determinable only by a systematic progression of work based on principles of established science and conducted to generate new knowledge. At a modular interface that usually means a specific question — whether any detail holds its required performance across the gap range that measured stack-up and differential movement produce — rather than the accumulation problem in general. You self-assess.
Q: Is surveying and setting out modules an R&D activity?
A: Setting out, levelling and shimming to the project's stated tolerances using established methods is generally unlikely to be a core R&D activity on those facts, subject to the activity's own facts and the statutory tests. It may still be relevant as a supporting activity, but because it is directly related to producing a building it engages s 355-30(2)(c) and is a supporting R&D activity only if undertaken for the dominant purpose of supporting core R&D activities.
Q: Is dimensional QA of modules routine testing under s 355-25(2)(f)?
A: Section 355-25(2)(f) excludes activities associated with complying with statutory requirements or standards, including (i) maintaining national standards, (ii) calibrating secondary standards and (iii) routine testing and analysis of materials, components, products and processes. Checking each production module against the factory's tolerance specification sits close to subparagraph (iii). "Associated with" is broad, and whether a particular measurement activity falls inside the exclusion is a question of fact assessed activity by activity — the company self-assesses.
Q: Does installing modules on site pass the dominant purpose test?
A: It has to be answered activity by activity. Installation produces goods, and craneage, setting out and sealing are directly related to producing them, so limbs (b) and (c) of s 355-30(2) are engaged and the work is a supporting R&D activity only where it is undertaken for the dominant purpose of supporting core R&D activities. A lift that delivers a saleable storey while also generating experimental data has a dominant purpose that is a question of fact, tested against that activity and not against the project as a whole.
Sources & Further Reading
legislation.gov.au — Income Tax Assessment Act 1997 — Div 355, incl. ss 355-25 and 355-30
Related: R&D for property and construction · what does not qualify · refundable vs non-refundable · what an RSP is · claiming R&D under $20,000
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.