Engagement state at handoff
§ 01 · STATEMENT OF WORKWhat's in, what's out.
Tier 3 has the largest scope of the three engagements, and an explicit SOW prevents drift mid-engagement. The Phoenix engagement was scoped at 8 weeks fixed-fee, with a written status cadence and event-driven decision calls. The full SOW is in the bundle; the operationally-critical part is what's not in scope.
Out of scope by design: direct utility or counterparty negotiation (Saguaro counsel runs negotiations, we supply the package); EPC bid solicitation, scoring, or contract negotiation (Saguaro procurement runs the bid, we supply the assumption pack); construction-phase support beyond the engagement window; operating-phase model maintenance (the analyst owns it post-handoff); permit applications.
This division matters because it sets up the cadence rules. Sgridworks is the model's author, not the project's owner; the model's outputs feed the owner's decisions, and once handed off, the model leaves Sgridworks's hands.
§ 02 · UTILITY + EPC PACKSTwo artifacts that leave the model.
Two of the engagement deliverables are not internal documents — they go to counterparties. The Utility Negotiation Package is what counsel takes into APS. The EPC Assumption Pack is what procurement puts in front of three named bidders. Both are produced by deterministic generators in the BTM-Optimize repo, so a refresh after a counterparty position changes is a single CLI invocation, not a rewrite.
Utility Negotiation Package
Translates the model into utility-language: peak demand, deliverable MW, ETA months, BTM offsets, flexibility commitments. Backs the legal posture; does not replace it. Counsel reads it before each negotiation milestone. Refresh trigger: any counterparty position change.
EPC Assumption Pack
Pre-empts the "three comparable bids each modeling a different stack" problem. PDF checklist organized around bid sections (gas, solar, BESS, BoP, controls, sequencing), Excel BOM with asset-level capex and source. Pack ships before the RFQ goes to bidders.
§ 03 · CADENCE TAXONOMYEvent-driven, not scheduled.
Tier 3's risk-register update cadence is event-driven, not on a calendar. A scheduled-check-in cadence has a sponsor-team-flavored failure mode where snapshots arrive on time and contain nothing of substance — the project hasn't moved, so the snapshot duplicates the previous one. We trigger updates only when something material changes. The taxonomy below is the pre-commitment to what counts.
Counterparty actions
- Equipment LOI signed / revoked / expired
- Utility milestone (study completion, IA execution, energization)
- Permit decision / extended hearing
- Offtake LOI / lease signed or amended
- EPC contract executed
- Construction milestone delay > 30 days
Exogenous shocks
- Fuel-price excursion > ±25% from base
- Forward strip shift > +15% on the 5-yr
- Tariff filing on the relevant utility
- BESS chemistry / cost shift > 15%
- Lender repricing > 100 bps
Model-driven findings
- Re-run flips a top-3 risk
- Sensitivity sweep returns P90 outside Tier 2 documented range
- Methodology error discovered post-handoff (rare)
Default: none
- Default cadence has no calendar trigger
- IC may request quarterly governance check-in
- Documented in cadence file if requested
§ 04 · SNAPSHOTS IN FLIGHTThree triggers, three updates.
During the 8-week Phoenix engagement, three triggers fired. None warranted a Tier 2 reversion. The first two were favorable (project-execution risks resolving on schedule); the third was an adverse market event that compressed the IRR margin from 0.7 pp to 0.5 pp without flipping the recommendation.
2026-07-08-
bess-loi
Second BESS-vendor LOI signed
Saguaro procurement closed a Fluence LOI complementing the existing Tesla Megapack 2 LOI. Both LOIs price-locked through 2026-12-31 with cell allocation guarantees for 2027-Q3 delivery. The BESS supply risk that dominated the Tier 2 IC memo as the outstanding gating item is now mitigated through dual sourcing.
BESS supply chain: rank 1 → 4 · contribution −17.1 pp
Air permit (NOx): rank 2 → 1 · contribution +1.2 pp
Construction execution: rank 4 → 3 · contribution +1.7 pp
2026-07-22-
permit-clear
Maricopa air-permit pre-application clean
Maricopa County Air Quality Department confirmed the 4×12.5 MW Wartsila configuration fits within the existing major-source NOx ceiling for the site's airshed. SCR retrofit recommended but not required. Formal application filed 2026-07-20; hearing on docket 2026-09-15.
Air permit: rank 1 → 3 · contribution −9.0 pp
Fuel price excursion: rank 2 → 1 · contribution +1.6 pp
Construction execution: rank 3 → 2 · contribution +2.4 pp
2026-08-04-
fuel-strip
Henry Hub forward strip widened +18%
Sustained Gulf Coast LNG export uplift + Permian production deferral pushed the 5-year forward strip from $4.20/MMBtu to $4.96/MMBtu base case. Project IRR moves from 8.7% to 8.5%; NPV from +$11.8 M to +$9.7 M. Both still clear hurdle and IC's NPV floor — but the load-bearing-assumption test from the Tier 2 memo (fuel as the largest tornado mover) just had its first real-world stress.
Fuel price excursion: rank 1 → 1 · contribution +5.5 pp
Construction execution: rank 2 → 2 · contribution +4.3 pp
Air permit: rank 3 → 3 · contribution −1.4 pp
The 08-04 trigger was the only one to fire a synchronous decision call. Top-mover Δ contribution (+5.5 pp) was below the +10 pp threshold, but Saguaro project finance requested a call independent of the threshold rule because the IRR margin compressed from 0.7 pp to 0.5 pp. Cadence escalation rules are floors, not ceilings — the client can always pull a call without triggering the formula.
§ 05 · HANDOFF BUNDLEWhat ships on day 56.
The bundle is one tarball. It contains every Tier 1 + Tier 2 + Tier 3 artifact, the BTM-Optimize repo at the engagement commit, all three update logs, all four risk-register snapshots, and a SHA-pinned reproducibility manifest. The acceptance test for the engagement: the client analyst regenerates the Tier 2 IC memo PDF byte-identically against the handed-off repo.
saguaro-phoenix-dc-tier3-bundle.zip
├── README.md # Bundle table of contents
├── 01-statement-of-work.md
├── 02-model-handoff-spec.md
├── 03-utility-negotiation-package.md
├── 04-epc-assumption-pack.md
├── 05-risk-register-cadence.md
├── 06-update-log-template.md
├── 07-engagement-close.md
├── METHODOLOGY.md
├── REPRODUCE.md
├── repo/ # BTM-Optimize at engagement commit
├── outputs/
│ ├── tier1/ (Site Screen artifacts)
│ ├── tier2/ (Power Stack Model artifacts)
│ └── tier3/
│ ├── utility-negotiation-pack.{pdf,xlsx}
│ └── epc-assumption-pack.{pdf,xlsx}
├── engagement/
│ ├── scenarios/
│ ├── runs/
│ ├── risk-register-snapshots/
│ │ ├── manifest.json
│ │ ├── ...kickoff.json
│ │ ├── ...bess-loi.json
│ │ ├── ...permit-clear.json
│ │ └── ...fuel-strip.json
│ └── update-logs/
│ ├── 2026-07-08-bess-loi.md
│ ├── 2026-07-22-permit-clear.md
│ └── 2026-08-04-fuel-strip.md
└── handoff-bundle.tarball.sha256 # Reproducibility manifest
After handoff, the model is the client's. We make no commitment to operate, update, or maintain the bundle post-engagement. The cadence transfers with the bundle; the analyst owns trigger detection from day one. Sgridworks remains available on a follow-on retainer if the client requests it, scoped and priced separately.