Layer 1 reads a building's configuration—its geometry, materials and load paths. Layer 2 reads its climate history. Where configuration is a static input, the environment is a sequence: decades of temperature and humidity acting on each component. This brief gives institutional reviewers a traceable way to scrutinise that second layer.

That claim is this: the same building, in two different climates, is not the same building. A condition methodology that returns one answer for a given age and typology has not modelled the environment at all — it has modelled the average. Layer 2 refuses the average. It computes, from raw time series, the specific cycles a structure has actually lived through, and lets those cycles drive material-specific decay through models that have stood for forty years. Five figures follow, each its own visual story.

01 — PipelineFrom raw weather to a condition map

The layer is a pipeline with five stages. Raw environmental time series enter at the top; a per-component condition trajectory — and a current condition map for the whole building — emerge at the bottom. The crucial thing to read in this first figure is the direction of time: nothing here is a snapshot. Every stage is an integration over decades.

FIG. 1
Layer 2 pipeline — environmental time series to per-component condition
Data / signal Rule engine Propagation
Time axis
decades of hourly–daily records · integrated, never averaged
01 Raw inputs Temperature & relative humidity
hourly or daily · over decades
Series A · °C
Temperature time series
Series B · % RH
Relative humidity time series
02 Derived signals aggregate statistics computed
from the raw series
FTCFreeze-thaw cycles0°C sign-crossings / yr
TATTime above thresholdhours > 80% RH / yr
ΔTThermal amplitudedaily & seasonal range
CONCondensation eventssurface < dew point
WDCWetting-drying cycles80% RH crossings / yr
AFIAccumulated frost indexΣ freezing degree-days
03 Per-material attack rules each signal damages each material
at a known rate
Every derived signal is mapped to the materials it attacks, and the rate of attack is drawn from an established degradation model — not invented for this engine.
Tuutti Fagerlund Vesikari ISO 9223 Powers
04 Chain-reaction propagation damage spreads to connected
neighbours · ×multiplier at joints
●
○
○
×1.0 → ×1.8 amplification at cracks & joints
Damage in one component raises the local severity for its neighbours — a spalled joint admits water, which lifts freeze-thaw and corrosion rates for everything it touches.
05 Output per-component trajectory over time
+ current condition map
Condition trajectory · score vs. years
Current condition map · per component
each tile = one componentdark green → deep red
Read top to bottom — and notice the clock. Stages 1 and 2 turn a forty-year weather record into six aggregate signals. Stage 3 routes each signal to the materials it attacks through established models; Stage 4 lets damage propagate along the structure, amplified at every crack and joint. The output is not a grade but a trajectory: the calculated path each component took to its present state, and the map that path produces.

This is the whole posture of Layer 2 in one diagram. A static survey asks what condition is this in? Layer 2 asks what did it live through, and what does forty years of that do to this material? The rest of this brief opens each stage in turn — beginning with the signals themselves.

02 — SignalsPatterns hidden inside the time series

The simulation never reads raw temperature or humidity directly. It reads five derived signals — aggregate statistics that compress decades of records into the handful of patterns that actually drive degradation. The distinction matters to a physicist: these are not new measurements but structures already present in the series, isolated because they carry the damage information. Each chart below shows the signal as it is extracted, with the typical Swedish range it falls in.

FIG. 2
Five derived signals, extracted from temperature & humidity
Measured series Threshold Counted event
Freeze-thaw cycle count FTC · n/yr
Swedish range50–150 / yr
Sign-crossings through 0°C, weighted by the humidity state at the crossing.
Time above threshold TAT > 80% RH · h/yr
Unheated industrial1 500–3 000 h
Total hours where RH exceeds the corrosion-enabling threshold.
Thermal amplitude ΔT · °C
Winter / summer15–25 / 10–18
Daily & seasonal max–min range; drives differential expansion stress.
Condensation events CON · n/yr
Episodicsurface < dew pt
Surface temperature drops below ambient dew point — full electrolyte on steel.
Wetting-drying cycle count WDC · n/yr
Swedish typical40–70 / yr
Humidity crossings through the ~80% RH threshold — accelerates carbonation and salt crystallisation.
These are derived patterns, not new measurements. Each signal is an operator applied to the raw series: a crossing-counter for FTC and WDC, an integral above a line for TAT, an envelope for ΔT, a comparison against a computed dew point for condensation. The simulation works on these because they — not the instantaneous readings — are what forty years of weather actually leaves behind in a material.

03 — MechanismEach material is attacked differently

A signal is only half of a rule. The other half is the material it acts on — and the same humidity that quietly carbonates concrete will rot timber and corrode steel on entirely different clocks. The engine encodes a distinct attack mechanism for every material, and each is grounded in an established model rather than fitted to our own data. The point a technical partner should take from this figure is the citation in the corner of every card: these rules are not invented.

FIG. 3
Material attack mechanisms — dominant signal, timescale, model
Dominant signal Secondary
Reinforced concreteStructural
FTCWDCTAT
Freeze-thaw microcracking once degree of saturation exceeds the critical threshold; carbonation accelerated by wetting-drying cycles; rebar corrosion initiates once the carbonation front reaches the bar.
Timescaledecades
Tuutti two-phase
+ Fagerlund ScrModel
Structural steelExposed
TATCON
Atmospheric corrosion proportional to time above 80% RH; condensation events provide full electrolyte contact, the worst case for internal surfaces.
Timescaleyears–decades
ISO 9223
corrosivity C1–C5Model
MasonryBrick + mortar
FTCWDC
Freeze-thaw spalling driven by cycle count; mortar-joint deterioration concentrates at saturated joints; orientation matters — north walls thaw slower, taking more damage per cycle.
Timescaledecades
Vesikari
Nordic service lifeModel
Sandwich panelsEnvelope
ΔTCONTAT
Facing-to-core bond fatigue from thermal cycling; fastener corrosion at penetration points; mineral-wool core wetting reduces both thermal and structural performance.
Timescaleyears–decades
Thermal-fatigue
+ bond degradationModel
TimberThreshold material
TATMC
Stable below 20% moisture content, decay-active above — a hard threshold, not a gradient. Brown rot dominates in Swedish conditions; sustained exposure is required to initiate decay.
Timescalethreshold-gated
Moisture-threshold
brown-rot kineticsModel
Roofing membranesEnvelope top
ΔTTAT
Thermal-cycling fatigue plus UV ageing; lap-seal failure is the typical first failure mode; trapped moisture then accelerates a substrate-damage cascade below.
Timescaleyears
Thermal-fatigue
+ UV ageingModel
Read the corner of each card. Concrete answers to Tuutti and Fagerlund, masonry to Vesikari, steel to ISO 9223 — decades of building-physics research, each routed to the signal it responds to. Timber is the outlier: it is a threshold material, stable until moisture content crosses 20%, then decay-active. The engine models that discontinuity explicitly rather than smoothing it into a rate.
× The same humidity that quietly carbonates concrete will rot timber and corrode steel — on three entirely different clocks. EGG Layer 2 brief · §3

04 — SensitivitySame building, different environment

A model is only as useful as it is sensitive. If the output barely moves when the input changes, the input was never really driving anything. So here are three controlled comparisons — same building, same age, one environmental variable changed — and in each, the output diverges sharply. The divergence is the result.

FIG. 4
Three sensitivity comparisons
Healthy Degraded Critical
A Freeze-thaw cycle count on masonry condition Fixed: same building · 61 years
Three FTC scenarios, from a sheltered south wall to an exposed north wall with condensation. Mortar — the saturated, jointed component — diverges far harder than the brick around it.
Mortar condition · score / 100
Low FTCsheltered S · 30 / yr
55
Medium FTCexposed E · 70 / yr
35
High FTCexposed N + cond. · 110 / yr
20
Brick condition · score / 100
Low FTC30 / yr
75
Medium FTC70 / yr
65
High FTC110 / yr
55
B Time above threshold on steel section loss Fixed: same steel beam · 40 years
Doubling the annual hours above 80% RH does not double the damage — it does worse, because more of those hours arrive with condensation and full electrolyte contact.
Low TAT
1 500 h / yr above 80% RH
1.0× baseline loss
SECTION LOSS1.7×MULTIPLIER
High TAT
3 000 h / yr above 80% RH
1.7× baseline loss
C Wetting-drying cycles on carbonation rate Fixed: same RC column · 61 years
Frequent wetting-drying doubles the carbonation coefficient. At 61 years that is the difference between a column still in its initiation phase and one that has been actively corroding for years.
Constant RH regime
k = 6 mm/yr0.5 · standard
47mm depth · initiation
CARBONATION DEPTH2×AT 61 YEARS
Frequent wet-dry cycles
k = 12 mm/yr0.5 · accelerated
94mm depth · active corrosion
Same building, different environment, dramatically different outcome. Mortar falls from 55 to 20 across the FTC range while the brick beside it barely moves; steel section loss runs 1.7× higher under high humidity exposure; a carbonation front reaches 47 mm or 94 mm depending only on how often the concrete wets and dries. None of these are tuned — they fall out of the established rate models when the derived signals change.

05 — UniquenessSensitive to building and climate

The final figure is an experiment, not a diagram. Four scenarios are constructed so that each comparison isolates exactly one axis — orientation, construction quality, or climate — while everything else is held constant. If the methodology produced generic outputs, these four would converge. They do not: each axis moves the result on its own, which is what it means for an output to be building-specific and climate-specific.

FIG. 5
Four-scenario uniqueness demonstration
Held constant Variable changed Critical outcome
Variable Scenario ABaseline Scenario BOrientation axis Scenario CConstruction axis Scenario DClimate axis
Frame Heavy RC frame Heavy RC frame Lightweight steel Heavy RC frame
Cover 35 mm 35 mm 20 mm · poor 35 mm
Orientation South wall North wall South wall South wall
Climate Stockholm
moderate FTC · mid-RH		 Stockholm Stockholm Sundsvall
high FTC · higher TAT
Mortar condition 38 / 100 18 / 100 36 / 100 22 / 100
RC capacity loss 8% 9% 18% 14%
A vs B
Orientation isolated
Building variable changes, environment constant. North-wall mortar lands 15–25 pts below south — here 38 → 18.
A vs C
Construction isolated
Frame & cover change, climate constant. RC column capacity reduction runs 8% (A) against 18% (C).
A vs D
Climate isolated
Only the climate file changes. Masonry mortar falls from 38 to 22 — a 42% reduction from location alone.
The experimental design is the argument. Each column changes exactly one thing from the baseline — the yellow cells mark the single variable, the red cells the outcome it moves. Because orientation, construction and climate each shift the result independently, the methodology cannot be producing a generic answer keyed to age and typology. It is producing an answer keyed to this building, in this place.

That is the case for Layer 2 in full. The configuration a scan recovers tells us what a building is made of and how it is connected; the environmental signals tell us what forty winters did to those materials. Neither alone is enough. Held together — building variable and climate variable, each demonstrably driving the output — they produce a condition estimate specific enough to defend.

The scientific foundation
Layer 2 does not invent rules. It integrates established models from building physics into a chain-reaction propagation model applied to scanned building configurations.
Tuutti1982
Two-phase model for carbonation and rebar corrosion.
Fagerlund1977
Critical degree of saturation, Scr, in freeze-thaw.
Vesikari1988
Nordic masonry service-life prediction.
Powers1945
Hydraulic pressure theory of frost damage.
ISO 9223std
Atmospheric corrosivity categories C1–C5 for exposed steel.
The simulation's contribution is not a new degradation law — it is the integration: routing each derived signal to the right model, propagating the resulting damage along the structure scanned in Layer 1, and amplifying it at the cracks and joints where real buildings fail first. Every rate in the engine traces back to a citation above.
Read the Condition Brief
— END —