California Freight Cleanup → Investigation 1-3
Do transport and building electrification compound or conflict?
20 joint transport-and-building combinations evaluated together at 2035, with a direct test of whether the two sectors interfere through shared atmospheric pathways. They don’t, at least not at this model’s resolution — benefits add to within 0.3 deaths. The zero-cost T1+B1 combination remains the rational baseline.
The decision
The transport and building investigations each sized one lever in isolation — necessary for clean accounting but not how programs actually work. CEC dollars are allocated across sectors simultaneously. The real question: given that we are investing in both, which combination produces the most health benefit per dollar?
A secondary question is whether the sectors interfere through shared atmospheric pathways. Both transport and building NOx reductions alter ozone chemistry in VOC-limited airsheds, and the combined effect need not equal the sum of the individual effects. Without a combined run, any advice to “do T2 plus B2” rests on an untested additivity assumption. Investigation 1-3 tests that assumption directly.
Methodology
We cross four transport vectors (T1–T4 from Investigation 1-1) with four building vectors (B1–B4 from Investigation 1-2) to produce 20 combined scenarios. T5 is excluded—its extreme ozone offset profile (93%) warrants separate treatment. Each combined vector is 16-dimensional; costs are additive (T2+B4 = $2.5B).
The combined MC is re-run from scratch under seed 44 rather than composited from stored Investigation 1-1/6 arrays. This ensures internally consistent CRF and VSL draws at the cost of a small baseline discrepancy (14,538 vs. Investigation 1-1’s 14,504 and Investigation 1-2’s 14,582)—MC sampling noise, not a real population difference.
Three upstream values from Investigation 1-1 and Investigation 1-2 are pulled via
upstream_value for cross-validation and drift detection.
Sector decomposition computes transport and building marginals independently
from the joint MC; the additivity check reports the residual in deaths per
scenario. Ozone disbenefit runs deterministically for the top-5 combined
scenarios; VOI covers all 20 alternatives at 2035.
Findings
| Rank | Portfolio | Deaths avoided | Cost ($B) | DAC share |
|---|---|---|---|---|
| 1 | T1 + B1 | 50.2 | $0.0 | 20.8% |
| 2 | T1 + B3 (new construction only) | 35.1 | $0.0 | 20.8% |
| 3 | T1 + B4 (cooking-first) | 52.6 | $0.5 | 20.8% |
| 4 | T4 + B1 (equity-focused transport) | 85.3 | $1.0 | 20.7% |
| 5 | T4 + B3 | 70.2 | $1.0 | 20.7% |
The net-benefit ranking is budget-sensitive by construction: zero-cost portfolios always win when health benefits are compared against program cost. T1+B1 ranks first because it achieves positive benefit at no spending. At a fixed $2B envelope, T2+B1 dominates. T4+B1 at $1.0B is the most defensible equity-priority option—directing transport effort toward disadvantaged communities (LA Basin/SJV at 1.5× the statewide rate) while relying on current building-code enforcement.
Sector additivity confirmed. Summing Investigation 1-1 and Investigation 1-2 sector marginals matches the combined MC output to within 0.3 deaths (<0.06%) for T1+B1, T2+B1, and T4+B1. No detectable cross-sector interference at ISRM linear model fidelity. Additivity is a consequence of the model’s algebra, not an empirical test of atmospheric chemistry—Investigation 4-2 probes the nonlinear ozone interactions separately.
Upstream drift flag. Investigation 1-1 and Investigation 1-2 sha256 values changed between Investigation 1-3’s last run and the current versions. Changes are small (<0.5% in all key fields) and do not alter the qualitative findings, but re-run Investigation 1-3 before citing these combined figures in any CEC presentation.
Caveats
- Additivity is a model property, not an empirical chemistry finding. ISRM sector marginals are applied as independent scalars on the same baseline field. Real VOC-limited cells where transport NOx reductions shift the regime boundary for building NOx would not be captured. Investigation 4-2 investigates this specifically.
- The Investigation 1-3 baseline (14,538) differs from Investigation 1-1 (14,504) and Investigation 1-2 (14,582). This is a seed-44 vs. seed-42/43 artifact. Cross-investigation comparisons will show ~0.2% inconsistencies that are sampling noise, not real phenomena.
- T5 (heavy-duty NOx) is excluded. The 20 combined scenarios cover T1–T4 only. T5’s 93% ozone offset profile makes it a special case requiring separate treatment.
- Net-benefit ranking is VSL-sensitive and budget-conditional. T1+B1 at $0 cost dominates only in the unconstrained-budget framing. CEC program managers evaluating a fixed $2B appropriation should use the fixed-budget frontier.
- No building-load feedback to grid-marginal emissions. Building electrification raises electricity demand; no grid-marginal correction analogous to Inv 20 (EV charging) is applied here. Building benefits may be slightly overstated.
-
Ozone disbenefit inherits the same Sillman-gap code path as Investigation 1-1
and Investigation 1-2.
scenario_o3applies regime-aware sensitivity factors without Sillman (1995) indicator calibration. PM2.5 figures are unaffected.
Provenance
| File | Link | Purpose |
|---|---|---|
results.json | Top-5 rankings, baseline, sector decomposition, additivity residuals | |
analysis.md | Mechanical readout (includes upstream drift warning), inputs audit block | |
scenario.md | Sticky methodology, key anchors, downstream dependency map |
Run provenance: generated 2026-05-04T07:42:25; results.json
sha256 189537563bdc. Note: a stale-upstream flag is active
(Investigation 1-1 sha256 3326583e4445, Investigation 1-2 sha256 99dcff1b4dfb
differ from those recorded at Investigation 1-3’s last run). Re-run Investigation 1-3 before
citing these combined figures as final. Downstream: Inv 11 (CRF-conditional
policy) has a hard dependency on Investigation 1-3 MC arrays; it too requires re-run.