Santiago

Santiago is the demo of a closed basin with a real monitoring network. The Metropolitan Region has winter temperature inversions, a critical-episode regime that is declared every year, and SINCA stations with decades of data — the place where a forecast can be compared with what was measured, not just shown.

Time zone	America/Santiago
Standard evaluated	Chile, MMA (cl-mma)
Grid	109 × 109 cells of 750 m, about 82 × 82 km
Forecast	4.5 days, every 15 minutes
Species modelled	SO₂, NO₂, PM₁₀, PM₂.₅
Sources	22 in 4 groups: 10 stacks, 9 volume sources, 3 area sources
Receptors	10, all SINCA stations
Monitoring stations	43, with measured air-quality and weather series

The stations, and the receptors

The demo carries 43 monitoring stations of the SINCA network, at their official coordinates, with their measured series — air quality and weather — beside the forecast. Ten of them are also modelled receptors: Pudahuel, Quilicura, Cerro Navia, Cerrillos II, Parque O’Higgins, Las Condes, La Florida, Puente Alto, El Bosque and Talagante. The reason is simple: a receptor that coincides with a station lets the modelled curve sit beside the measured one, and that comparison is what makes a forecast believable. On the map such a place is one marker with two roles, and in the Series panel the measured curve is one Origin choice away. The coordinates come from the monitoring registry itself, not a hand-written list — nine of them sat 1.5 km north of their station until that source was corrected.

The sources, in four groups

The groups are not an administrative classification; each exists so that one difference stays legible in the contribution breakdown.

Renca thermoelectric complex — two adjacent plants on the north bank of the Mapocho: Nueva Renca, a 379 MW combined cycle on natural gas with diesel as backup, and Renca, a diesel backup plant with two 50 MW units. It is the domain’s dominant source of nitrogen oxides and the only one that switches fuel between scenarios. Being dual-fuel is what earns it more than one scenario: sulphur in combustion depends almost entirely on the fuel, so going from gas to diesel multiplies SO₂ by more than two orders of magnitude while NO₂ rises only modestly. The plume changes composition, not just size. Unit 1, disconnected in 2023, is kept at zero rate on purpose: a retired asset is still part of the installation.

Industry of Quilicura, Pudahuel and Cerrillos — eleven sources across the three industrial corridors, the only mix of stacks with volume sources. An industrial park does not emit through an identifiable stack but from a set of buildings; modelled as a volume it stays near the ground and affects the nearby receptors, which is what actually happens.

Santa Marta landfill — in Talagante, far from the thermoelectric group, with three distinct geometries: the working face as an area, the biogas flare as a stack, the leachate plant as a volume. It is the group that separates geometry from magnitude: it emits at ground level and its footprint reads completely differently from a 60 m stack.

Maipo aggregates, fugitive emissions — extraction and crushing on the bank of the Maipo. Particulate matter only, no plume rise, highest near the source and governed by low-level wind. Kept separate, it explains why PM₁₀ behaves differently from SO₂ and NO₂ in the south of the domain.

The scenarios: three operating plans

Normal operation on gas is the case the others are read against: Nueva Renca on gas, negligible SO₂, a NOₓ-dominated profile; Renca’s diesel backup is why the base SO₂ is not zero.

Natural gas restriction — Nueva Renca switches to diesel. Not a failure but the plant’s design behaviour: it keeps diesel precisely so it can keep running when gas is short. Worth modelling because the change is not uniform: SO₂ multiplies by more than two hundred, NO₂ rises far less.

Environmental pre-emergency (GEC) — a declared critical-episode day: industry and fugitive sources cut back under the decontamination plan while power generation and the landfill keep operating. It is the scenario that inverts the argument: the others make things worse, this one makes them better, and it answers a question a Chilean regulator or operator genuinely asks — does the restriction help, and at which receptors?

The scenarios: three emergencies

Three more scenarios model an event rather than a plan: an emission that changes hour by hour, with a start, a peak and a tail. The installations are real; the events are hypothetical and match no incident that occurred. All three start at the same hour — the worst hour of dispersion in the modelled window, a winter morning inversion with a 46 m mixing height and wind under 1 m/s — so they can be compared against the same weather.

Baghouse failure (Quilicura) — the kiln’s fabric filter tears: half an hour with no control at all while the kiln is still at load, an hour and a half at reduced load while the filter is isolated, then the kiln stopped, its residual emission small but its plume no longer rising. That last phase often matters more than the first: a smaller emission with no buoyancy reaches the receptors. Chosen here because the SINCA Quilicura station sits practically on the kiln — the one point where the model and a real measurement can share a chart.

Biogas peak at the flare (Santa Marta) — the opposite of the obvious “the flare goes out”: the flare is what turns the biogas’s sulphur into SO₂, so what makes an SO₂ peak is more biogas reaching a flare that stays lit. Forty-five minutes of peak at 1273 K, an hour and a quarter above design flow, then the flare trips on overpressure and the plume loses both its SO₂ and its rise.

Fire at the working face (Santa Marta) — a surface source, and that is the point: the other two are elevated stacks, this is 87.6 hectares burning at ground level. An hour of development, two of declared fire — the peak, and the moment of most thermal lift — then five hours of smouldering after extinction, still emitting particulate matter but with no buoyancy left. That last phase emits less than the second and reaches far more at receptor level.

The alerts

The demo’s alert levels — warning, alert, critical — are its own instantaneous thresholds, set per species for SO₂, NO₂, PM₁₀ and PM₂.₅ and read in Domain → Species: operational, site-relative levels, not the critical episodes the authority declares. An alert appears on the map and in the panel when a receptor crosses one; already-observed past and forecast are told apart by their position relative to the forecast cycle. The Chilean standard (MMA) is a separate reading, in the Regulatory panel, with the hourly averaging it prescribes for SO₂ and NO₂. Against the stations’ measurements, at 1 and 8 hours, it is read in Monitoring → Standards, where the SINCA SO₂, NO₂ and O₃ rows are flagged unit unverified for now.

A note on the figures

Emission rates are screening values derived from standard emission factors and published plant specifications, calibrated so the model sits below the measurements at the great majority of station/species pairs. The emergency rates are the team’s own, representing a total loss of control followed by an orderly stop. They serve to compare scenarios with each other; they are not permit figures nor regulatory declarations.

→ The demos · Air quality standards