The Fundamental Difference
Acoustic modelling predicts acoustic performance based on geometric and material properties of a space that does not yet exist (or has not yet been modified). Acoustic measurement quantifies the actual performance of an existing space. The two activities have completely different roles in an acoustic design workflow, and confusing them — or trying to use one where the other is needed — leads to either over-budget testing or unverifiable compliance claims.
This article establishes the boundary conditions for each approach: when modelling is sufficient, when measurement is required by standards, what accuracy you can expect from each prediction method, and how the two activities relate in a complete acoustic design process.
What Acoustic Modelling Can and Cannot Do
What Modelling Does
Acoustic modelling transforms physical inputs — room geometry, surface areas, material absorption coefficients — into predicted acoustic performance metrics. At its simplest, this is Sabine's formula:
RT60 = 0.161 × V / A (ISO 3382-2:2008 Annex A, Equation A.1)
Where V is room volume (m³) and A is total absorption (sum of surface area × absorption coefficient across all surfaces).
At its most sophisticated, acoustic modelling involves ray-tracing algorithms (ODEON, EASE, Treble) that simulate thousands of individual sound ray reflections through a 3D model, computing the impulse response at each receiver position. From the impulse response, all acoustic parameters are derived: RT60, EDT, C80, C50, D50, STI.
Modelling Limitations
1. Input data accuracy The single largest source of modelling error is the absorption coefficient data used for surfaces. Published NRC values for acoustic panels are typically measured in a reverberation chamber per ISO 354, under specific boundary conditions that differ from field installation. Lab-measured αw may differ from field performance by 0.05–0.15.
2. Diffuse field assumption Sabine's formula assumes perfectly diffuse sound field — that sound energy is uniformly distributed throughout the room at any moment. This assumption holds reasonably well for rooms with multiple reflective surfaces and relatively uniform absorption. It breaks down in: long rooms (L/D > 3), rooms with strongly non-uniform absorption (e.g., highly absorptive floor with reflective ceiling), and rooms below approximately 100 m³ where modal behaviour dominates at low frequencies.
Eyring's correction partially addresses non-uniform absorption:
RT60 = −0.161 × V / (S × ln(1 − ᾱ)) (ISO 3382-2:2008 Annex A, Equation A.3)
Where ᾱ is the mean absorption coefficient and S is total surface area. Eyring is more accurate than Sabine when mean absorption coefficient exceeds 0.2.
3. Low-frequency behaviour All ray-based models (Sabine, Eyring, geometric ray-tracing) become unreliable below approximately 200–300 Hz. Below this range, room modes (standing waves) dominate acoustic behaviour, and prediction requires finite element analysis (FEA) or measurement. For spaces where bass frequency performance is critical (music rooms, recording studios, home cinemas), modelling cannot replace measurement.
4. Construction defects and flanking Models assume ideally constructed partitions with no gaps, penetrations, or flanking paths. In real buildings, workmanship variation and flanking transmission typically reduce measured sound insulation by 3–8 dB compared to the calculated Rw. This gap is irreducible in modelling — it can only be discovered through measurement.
Accuracy of Prediction Methods by Parameter
RT60 (Reverberation Time)
| Method | Typical Accuracy (mid-frequency) | Conditions |
|---|---|---|
| Sabine formula | ±0.10–0.20 s | Well-diffused room, uniform absorption, V > 200 m³ |
| Eyring formula | ±0.08–0.15 s | Rooms with higher absorption (ᾱ > 0.20) |
| Geometric ray-tracing | ±0.05–0.10 s | Mid frequencies (250–2000 Hz), sufficient model detail |
| All methods | ±0.20–0.40 s | At 125 Hz and below, rooms < 100 m³ |
For typical commercial spaces (offices, classrooms, meeting rooms), Sabine/Eyring modelling predicts RT60 to within ±0.15 seconds at mid frequencies — sufficient for design-stage specification. The uncertainty increases for low frequencies and small rooms.
Sound Insulation (Rw, DnT,w)
| Method | Typical Accuracy | Notes |
|---|---|---|
| Mass Law (simple) | ±4–6 dB | Homogeneous constructions only; ignores stiffness, coincidence |
| Calculation per EN 12354-1 | ±2–4 dB | For standard constructions with published test data inputs |
| Laboratory test (ISO 10140) | ±1 dB (repeatability) | Single element, ideal flanking isolation; no real-building penalty |
| Field measurement (ISO 16283-1) | ±1–2 dB (repeatability) | Includes flanking; representative of in-situ performance |
The key point: laboratory Rw is not the same as field DnT,w. The field penalty (flanking correction) is typically 3–8 dB. EN 12354-2 provides a calculation method for estimating field performance from laboratory data, but the result is still an estimate with ±3–5 dB uncertainty.
STI (Speech Transmission Index)
| Method | Typical Accuracy | Notes |
|---|---|---|
| Calculation from RT60 and S/N | ±0.05–0.10 (STI units) | Using Houtgast-Steeneken estimation method |
| ODEON/EASE simulation | ±0.04–0.08 | If impulse response model is accurate |
| STIPA measurement (IEC 60268-16) | ±0.02 (repeatability) | Direct field measurement, most accurate |
When Standards Require Measurement (Not Modelling)
England and Wales: Building Regulations Part E
Part E of the Building Regulations (Resistance to the Passage of Sound) requires post-construction field measurements for:
- All new-build dwellings (separation walls, separation floors)
- Conversions and material changes of use creating dwellings
- Hotels and student accommodation
- Registration with Robust Details Ltd before work commences
- Correct installation following the RD specification (no deviations)
- Random testing by RD inspectors (approximately 1 in 100 plots)
Schools: Building Bulletin 93:2015
BB93 states in its Technical Standards section: "Compliance with the acoustic performance standards in BB93 shall be demonstrated by measurement after the building is complete and before occupation." RT60 and background noise measurements are required in all spaces with acoustic performance requirements.
Note that BB93 also requires pre-design acoustic modelling to demonstrate that the specification will meet targets — both stages are required, not one or the other.
NHS Healthcare: HTM 08-01
NHS England's Health Technical Memorandum 08-01 (Acoustics) requires:
- Pre-design acoustic calculations
- Post-construction RT60 and background noise measurements in clinical spaces
- Post-construction sound insulation measurements (DnT,w and L'nT,w) for sensitive interfaces
BREEAM: Acoustic Assessment
BREEAM Man 05 (Construction Site Impacts) and Hea 05 (Acoustic Performance) both require either acoustic measurement or third-party assessment. Hea 05 specifically requires post-construction acoustic testing to achieve credits above threshold level.
The Standard Workflow: Modelling and Measurement Together
The correct workflow is not a choice between modelling and measurement. Both are required, at different stages:
RIBA Stage 1–2 (Concept/Feasibility) Acoustic modelling to check that the brief is achievable: can the room volumes accommodate the required RT60 targets? Is the site noise environment compatible with the indoor noise targets? Use Sabine/Eyring calculation or AcousPlan for quick iteration.
RIBA Stage 3 (Spatial Coordination) Acoustic modelling to verify the spatial design against all performance targets. Are partition lengths sufficient for the required DnT,w? Is the HVAC specification compatible with background noise targets? Issues identified at this stage cost almost nothing to resolve.
RIBA Stage 4 (Technical Design) Acoustic specifications prepared with product performance requirements. Modelling outputs justify the specification to the contractor. Any departures from Stage 3 design that affect acoustic performance require re-modelling.
Construction Stage Acoustic consultant site visits to verify acoustic-critical details: partition head/base seals, acoustic door hardware, service penetrations through acoustic barriers. These are the flanking paths that cause measured DnT,w to be worse than predicted.
Post-Construction (Pre-Occupation) Field measurements per the relevant standard (Part E, BB93, HTM 08-01). Compare against both the design-stage model predictions and the regulatory targets. Document any deviations and their causes for the O&M manual.
A Note on Using AcousPlan for Design-Stage Modelling
AcousPlan implements Sabine and Eyring calculations (ISO 3382-2 Annex A) with the 5,600+ material database to give you RT60 predictions across all octave bands from 125 Hz to 4000 Hz. This is sufficient for:
- Verifying that proposed specifications will meet BB93, BREEAM Hea 05, or WELL Feature 74 targets
- Identifying which surfaces require treatment and how much area is needed
- Comparing material options and their relative effectiveness
- Producing calculations to include in a pre-design acoustic report
What AcousPlan provides is the rigorous, standard-referenced calculation that every acoustic specification should be built on — replacing spreadsheet approximations with a defensible, documented prediction model.