Decision Frameworks — Energy Procurement

Decision Frameworks — Energy Procurement

This reference provides detailed decision trees, evaluation matrices, financial models, and strategic frameworks for electricity and gas procurement, tariff optimization, demand charge management, PPA evaluation, hedging strategy design, and multi-facility portfolio optimization. It is loaded on demand when the agent needs to make or recommend nuanced energy procurement decisions.

All thresholds, price assumptions, and market benchmarks reflect US commercial and industrial electricity and natural gas markets. Adjust for regional markets, current forward curves, and facility-specific tariff structures.

1. Procurement Strategy Selection

1.1 Pre-Procurement Intelligence Gathering

Before entering any procurement decision — contract renewal, new facility onboarding, or mid-term restructuring — assemble a comprehensive data package.

Data Assembly Checklist

1.2 Fixed vs. Index vs. Block-and-Index Decision Tree

Use this decision tree for each facility or portfolio segment independently — one strategy does not fit all sites.

START: What is the organization's tolerance for energy cost variance?

├── Budget variance >10% triggers executive escalation
│   ├── Contract tenor ≤ 24 months?
│   │   └── YES → Fixed-price full requirements
│   │       - Accept the risk premium (5-12% above forward curve)
│   │       - Negotiate volume tolerance band (±10-15%)
│   │       - Ensure contract includes change-of-use provisions
│   │   └── NO (>24 months) → Fixed-price with annual price resets
│   │       - Lock year 1 at fixed, years 2-3 at a formula (forward + adder)
│   │       - This limits the supplier's long-term risk premium
│
├── Budget variance of 5-10% is manageable
│   ├── Facility load factor > 0.70?
│   │   └── YES → Block-and-index
│   │       - Buy fixed blocks = 70-80% of baseload
│   │       - Float remaining 20-30% at index (day-ahead or real-time)
│   │       - Shape blocks to match base load pattern (ATC vs. on-peak only)
│   │   └── NO (load factor < 0.70) → Shaped block-and-index
│   │       - Buy on-peak blocks only (match production schedule)
│   │       - Float off-peak and shoulder at index
│   │       - Supplement with TOU-indexed product for off-peak
│
├── Organization can tolerate >15% variance (energy is <5% of COGS)
│   ├── Internal capability to monitor wholesale markets?
│   │   └── YES → Index pricing with financial hedges
│   │       - Base product: real-time or day-ahead index + supplier adder
│   │       - Layer financial hedges: buy call options for peak months
│   │       - Set a price ceiling through options ($X/MWh cap)
│   │   └── NO → Index with a price cap product
│   │       - Supplier provides index pricing with a contractual ceiling
│   │       - Cap premium is typically $3-7/MWh above forward curve
│   │       - Simpler than managing separate financial hedges

1.3 Layered Procurement Methodology

Layering eliminates single-point market timing risk. The methodology:

Step 1: Determine the hedging horizon. Most C&I buyers layer 18–36 months ahead of the delivery period. For a January 2028 start date, begin buying tranches in July 2026.

Step 2: Set the number of tranches. Standard approaches:

Step 3: Execution rules.

• Execute each tranche at the prevailing market price on the scheduled date — do not try to time within the tranche window.
• Exception: if the forward curve drops into the bottom 20th percentile of the 5-year range, accelerate by buying 2 tranches immediately ("buy the dip" rule).
• Exception: if the forward curve spikes into the top 20th percentile, defer the current tranche by 30 days (skip and catch up later).
• Never defer more than 2 consecutive tranches — rolling deferrals leave you unhedged.

Step 4: Document and report. Maintain a procurement log showing: tranche date, volume procured, price locked, forward curve price at execution, cumulative weighted average price, and remaining open position. Report to finance quarterly.

Example — 10 MW peak load, 60M kWh annual consumption:

Delivery year: 2028
Hedging start: July 2026
Tranches: 6 (bimonthly, ~10M kWh each)

Tranche 1 (Jul 2026): 10M kWh @ $44.50/MWh  — Forward was $45.20
Tranche 2 (Sep 2026): 10M kWh @ $42.80/MWh  — Forward was $43.10
Tranche 3 (Nov 2026): 10M kWh @ $46.30/MWh  — Forward was $46.30
Tranche 4 (Jan 2027): 10M kWh @ $41.20/MWh  — Forward was $41.50 (buy-the-dip rule: 
                                                 also executed Tranche 5 early)
Tranche 5 (Jan 2027): 10M kWh @ $41.40/MWh  — Accelerated from March
Tranche 6 (May 2027): 10M kWh @ $43.80/MWh  — Forward was $44.00

Weighted average: $43.33/MWh
Range of execution prices: $41.20 - $46.30 ($5.10 spread)
If locked all-at-once in Jul 2026: $44.50/MWh → layering saved $1.17/MWh = $70,200

1.4 RFP Process for Deregulated Markets

Timeline and Phases

Supplier Evaluation Scoring Matrix

Bid Comparison Template

For each site, model the annual cost under each supplier's proposal:

Annual Cost = Σ(hourly volume × hourly price) + fixed charges + REC costs + adder fees

Where hourly price depends on product structure:
  Fixed: contract rate for all hours
  Block-and-index: block rate for block volume + index price for excess
  Index: (day-ahead or real-time LMP at load zone) + supplier adder

Always model at three forward price scenarios: base case (current forward curve), low case (forward - 20%), and high case (forward + 30%). A supplier whose index product looks cheapest at base case may be the most expensive at high case.

2. PPA Evaluation Framework

2.1 Physical PPA Evaluation

Physical PPAs involve direct energy delivery and are appropriate when:

• Your load is in the same ISO as the project
• You want both energy and RECs from a specific named facility
• You can manage the operational complexity of scheduling and balancing

Financial Modeling Framework

Step 1: Establish the baseline (no-PPA scenario). Project your energy costs over the PPA term using forward curves for years 1-5 and a long-term price escalation assumption (typically 2-3%/year) for years 6+.

Step 2: Model PPA cash flows.

Year N PPA Net Value = (Market Price at Hub - PPA Strike Price) × Expected Generation
                     - Basis Cost (Hub to Load Zone)
                     - Curtailment Cost (expected curtailed MWh × strike price)
                     - Balancing Costs (firming residual load not covered by PPA)
                     + REC Value (if RECs would otherwise be purchased separately)

Step 3: Sensitivity analysis — run these scenarios at minimum:

Step 4: Calculate NPV, IRR, and levelized cost of energy (LCOE) under each scenario. A PPA is economically justified if NPV is positive under base and bull cases and the loss under bear case is tolerable (typically <$2M cumulative over the PPA term for a mid-size C&I buyer).

2.2 Virtual PPA (VPPA) Evaluation

VPPAs are financial instruments — no physical energy delivery. The key risks differ:

Basis Risk Analysis

Basis risk is the primary financial risk in a VPPA. It arises because the generator settles at its node price and your load settles at your load zone price.

Quantification method:

1. Obtain 3-5 years of hourly LMP data for the generator's node and your load zone from the ISO.
2. Calculate the hourly basis: Load Zone LMP - Generator Node LMP.
3. Filter to hours when the generator would be producing (solar: daylight hours; wind: use historical generation profile).
4. Calculate the generation-weighted average basis.
5. Model the basis impact on PPA settlement:

Annual Basis Cost = Σ(hourly basis × hourly expected generation)

If generation-weighted average basis = $5/MWh and annual generation = 200,000 MWh:
Annual Basis Cost = $1,000,000/year
Over a 15-year PPA: $15M in basis costs (undiscounted)

Red flags for basis risk:

• Basis spread > $8/MWh generation-weighted average → high risk, negotiate basis hedge or reject
• Basis volatility (standard deviation) > $15/MWh → unpredictable, hard to budget
• Basis trend is widening over the historical period → structural congestion, likely to worsen
• Generator is located behind a known transmission constraint → congestion will increase as more generation is added in that zone

Curtailment Risk Analysis

Curtailment occurs when the ISO orders the generator to reduce output due to transmission constraints or oversupply.

Contract protection: Negotiate a curtailment threshold (e.g., first 5% is developer risk) and a compensation mechanism for excess curtailment (developer provides replacement RECs or a price adjustment). Never accept "buyer bears all curtailment risk" on a VPPA — this transfers a risk the buyer cannot manage or influence.

Credit and Accounting Requirements

2.3 Physical vs. Virtual PPA Decision Matrix

3. Demand Charge Optimization

3.1 Load Analysis Methodology

Step 1: Download 15-minute interval data. Request a minimum of 12 months of 15-minute kW demand data from the utility or your meter data management system. For facilities with sub-metering, obtain interval data at the system level (HVAC, production, compressed air) in addition to the main meter.

Step 2: Identify peak demand intervals. Sort all 15-minute intervals by kW descending. Focus on the top 50 intervals (the top 0.15% of all intervals in a year). These intervals drive your demand charges.

Step 3: Characterize peak drivers. For each of the top 50 intervals, identify:

• Date and time of day
• Day of week
• Outdoor temperature (proxy for HVAC load)
• Production schedule (was the line running?)
• Any anomalous events (equipment startup, testing, maintenance)

Typical findings for manufacturing facilities:

Step 4: Calculate the demand charge cost of peak intervals.

Monthly Demand Charge = Peak kW × Demand Rate ($/kW)

If normal operating peak is 4,000 kW and the actual peak is 4,800 kW:
Excess peak cost = (4,800 - 4,000) × $15/kW = $12,000/month

With an 80% ratchet:
Minimum billing demand for next 11 months = 4,800 × 0.80 = 3,840 kW
If normal peak drops to 3,500 kW next month, you're still billed at 3,840 kW
Annual ratchet cost = (3,840 - 3,500) × $15/kW × 11 months = $56,100

3.2 Peak Shaving ROI Framework

Battery Energy Storage System (BESS)

Sizing methodology:

1. Determine the target peak reduction (kW to shave).
2. Calculate the required energy capacity: target kW × duration of peak events. For demand charge management, 1-2 hours of duration is typically sufficient.
3. Apply round-trip efficiency (88-92% for lithium-ion): size the battery 10% larger than the calculated energy requirement.

Example — 500 kW peak shaving at a manufacturing plant:

Target reduction: 500 kW
Peak event duration: 2 hours (based on interval data analysis)
Battery size: 500 kW / 1,000 kWh (with 10% efficiency buffer: 500 kW / 1,100 kWh)

Installed cost (2025): $800-$1,200/kWh for C&I BESS
Total capital: $880,000-$1,320,000 (using 1,100 kWh at midpoint $1,000/kWh = $1,100,000)

Annual savings stack:
  Demand charge savings: 500 kW × $15/kW × 12 months = $90,000
  Capacity tag reduction: 500 kW × $60/kW-yr (PJM example) = $30,000
  TOU energy arbitrage: charge off-peak ($0.04/kWh), discharge on-peak ($0.08/kWh)
    1,100 kWh × $0.04/kWh spread × 250 days × 90% efficiency = $9,900
  Demand response revenue: 500 kW × $40/kW-yr (PJM Economic DR) = $20,000

Total annual value: $149,900
Simple payback: $1,100,000 / $149,900 = 7.3 years
With ITC (30% for standalone storage as of IRA): payback = $770,000 / $149,900 = 5.1 years

Decision thresholds:

• Payback < 5 years (with stacked value + incentives): strong economic case, proceed
• Payback 5-7 years: viable if aligned with sustainability goals or if demand charges are rising
• Payback 7-10 years: marginal, requires additional strategic justification
• Payback > 10 years: economics don't support investment without regulatory mandate

Demand Response Program Evaluation

Not all DR programs are equal. Evaluate on these dimensions:

3.3 Staggered Startup Protocol

The single lowest-cost demand charge reduction strategy — no capital required:

Problem: Morning startup creates a demand spike when HVAC, compressors, lighting, and production equipment all energize simultaneously between 5:30-6:30 AM.

Solution: Stagger equipment startup over a 60-90 minute window:

5:00 AM  — Lighting (50-100 kW)
5:15 AM  — HVAC pre-cooling/heating (500-800 kW, ramps over 30 min)
5:45 AM  — Compressed air system (200-400 kW, staged compressor starts)
6:00 AM  — Production Line 1 (300-500 kW)
6:15 AM  — Production Line 2 (300-500 kW)
6:30 AM  — Auxiliary systems, battery chargers, water heating

Result: Peak during startup drops from 2,200 kW (simultaneous) to 1,600 kW (staggered)
Savings: 600 kW × $15/kW × 12 months = $108,000/year at zero capital cost

Implementation: Program the building automation system (BAS) to enforce startup sequencing. Set hard interlocks that prevent the next system from starting until the prior system has reached steady state.

4. Market Analysis Framework

4.1 Regulated vs. Deregulated Strategy Map

4.2 Forward Curve Analysis

What the forward curve tells you:

• Market consensus on future energy prices (adjusted for risk premium)
• Seasonal price patterns (summer/winter spreads)
• Year-over-year price trajectory (escalation or decline)

What the forward curve does NOT tell you:

• Actual future spot prices (forwards are not forecasts — they include a risk premium)
• Short-term price spikes (forwards are averages, not tails)
• Regulatory changes, plant retirements, or transmission additions not yet priced in

Using forward curves for procurement decisions:

4.3 Capacity Market Exposure

In organized capacity markets (PJM, ISO-NE, NYISO), capacity charges are a significant cost component — $30–$120/kW-yr depending on the zone and auction results.

PJM Reliability Pricing Model (RPM):

• Auction held 3 years ahead of delivery year (Base Residual Auction)
• Incremental auctions adjust quantities closer to delivery
• Your capacity obligation is based on your PLC (Peak Load Contribution)
• PLC is set by your metered load during the 5 highest system coincident peak hours (5CP) in the prior delivery year

Managing capacity exposure:

1. Track PJM system peak alerts. PJM issues "hot weather alerts" and "emergency alerts" when system peaks are expected. Curtail discretionary load during these hours to reduce your PLC for the following year.
2. Install peak notification systems. Subscribe to PJM's demand response alerts. Deploy load curtailment controls that can drop 10-20% of facility load within 30 minutes of a peak alert.
3. Behind-the-meter generation. Running backup generators during coincident peak hours reduces your metered load and thus your PLC. Ensure generators are permitted for non-emergency operation and emissions-compliant.
4. Capacity tag trading. In some markets, capacity obligations can be traded or offset through financial instruments. Your supplier may offer capacity tag management as a service.

Example — capacity charge impact:

Facility peak: 5,000 kW
PLC (measured during prior year 5CP hours): 4,200 kW
PJM BRA clearing price for your zone: $85/MW-day

Annual capacity charge: 4,200 kW × $85/MW-day × 365 / 1,000 = $130,305/year

If you had curtailed 500 kW during the 5CP hours:
Reduced PLC: 3,700 kW
Annual capacity charge: 3,700 kW × $85/MW-day × 365 / 1,000 = $114,793/year
Savings: $15,512/year from 5 hours of load curtailment

5. Hedging Strategy Design

5.1 Hedging Instruments Available to C&I Buyers

5.2 Hedging Strategy by Risk Profile

5.3 Option Pricing and Evaluation

When buying call options to cap index pricing exposure, evaluate:

Option value = Max(0, Spot Price - Strike Price) × Volume

Cost: Premium per MWh × Contracted Volume
Annual premium for a $50/MWh cap on day-ahead pricing: $2-5/MWh (varies by market volatility)

Example — protecting 50,000 MWh annual index volume:
  Call option strike: $50/MWh
  Premium: $3/MWh
  Total premium cost: $150,000/year

  If spot averages $42/MWh: option expires worthless, total cost = $42 + $3 = $45/MWh
  If spot averages $65/MWh: option pays $15/MWh, effective cost = $65 - $15 + $3 = $53/MWh
  If spot spikes to $200/MWh (weather event): option pays $150/MWh, effective cap = $53/MWh

  Maximum effective rate: strike + premium = $53/MWh regardless of market price

When to use options vs. fixed contracts:

• Options when you want to participate in downside moves but protect against spikes
• Fixed contracts when the premium for options exceeds the cost of just locking in a fixed price (this happens when volatility is high and options are expensive)

6. Sustainability Procurement Alignment

6.1 Mapping Procurement to RE100 and SBTi

RE100 progress calculation:

RE% = (Renewable MWh procured) / (Total electricity consumption MWh) × 100

Acceptable renewable MWh sources (in order of additionality):
1. On-site generation (strongest claim)
2. Physical PPA with new project (strong additionality)
3. Virtual PPA with RECs from new project (good additionality)
4. Utility green tariff (varies by program design)
5. Unbundled RECs (weakest claim — RE100 tightening requirements)

SBTi trajectory alignment:

• SBTi requires absolute Scope 2 emissions reductions on a defined trajectory (typically 4.2%/year for 1.5°C alignment).
• Lock in long-term renewable procurement (PPAs) that deliver emission reductions year over year.
• Avoid procurement strategies that increase fossil dependence (long-term fixed contracts with fossil-heavy grid mix and no REC component).

6.2 Cost-Effective Sustainability Procurement Path

7. Multi-Facility Portfolio Optimization

7.1 Portfolio Aggregation Strategy

When to aggregate:

• 3+ sites in the same ISO/utility territory
• Total volume > 20 GWh/year (attracts competitive supplier attention)
• Sites have complementary load profiles (some peak summer, others peak winter)

Aggregation benefits:

• Volume leverage: 5-15% lower supply pricing than individual site procurement
• Load diversity: combined portfolio has higher load factor than individual sites, reducing supplier risk premium
• Administrative efficiency: single contract, single invoice, single relationship

When NOT to aggregate:

• Sites in different ISOs with different market structures (PJM and ERCOT should be procured separately)
• One site has unique requirements (e.g., real-time pricing needed for a demand response strategy) that would constrain the entire portfolio
• Sites have vastly different contract expiration dates (stagger expirations to avoid all-at-once recontracting risk)

7.2 Portfolio-Level Risk Metrics

Track at the portfolio level, not just site-by-site:

7.3 Site Prioritization for Demand-Side Investment

With limited capital for demand charge mitigation, prioritize sites using this scoring model:

Investment priority: highest composite score first. A site scoring >80 is a strong candidate for battery storage or demand response. A site scoring <40 has limited demand charge optimization potential — focus on supply-side procurement instead.

8. Natural Gas Procurement

8.1 Gas Procurement Structures

Natural gas procurement for C&I consumers (boilers, CHP, process heat, backup generation) follows similar principles to electricity but with distinct market mechanics.

8.2 Basis Differentials for Natural Gas

Natural gas prices vary by delivery point. Henry Hub (Louisiana) is the benchmark, but delivered cost depends on the basis differential between Henry Hub and your local city gate or utility delivery point.

Common basis differentials (approximate):

Key insight: A facility in New England on index pricing faces dramatically different winter risk than a facility in Texas. Basis in New England during a cold snap can exceed $15/MMBtu, tripling the delivered gas cost. New England gas procurement requires winter hedging with firm pipeline capacity or LNG backup — index pricing without protection is reckless in that market.

8.3 Gas-Electric Interdependency

For facilities with both electricity and natural gas loads, recognize the coupling:

• When gas prices spike, electricity prices spike. Natural gas is the marginal fuel for electricity generation in most US ISOs. A $2/MMBtu increase in Henry Hub translates to approximately $10-$15/MWh increase in wholesale electricity prices (depending on the average heat rate of marginal gas plants, typically 7,000-8,000 BTU/kWh).

• CHP economics are gas-price dependent. A CHP system generating electricity at a heat rate of 6,500 BTU/kWh has a fuel cost of $6.50 × gas price per MWh. At gas $3/MMBtu, generation cost is $19.50/MWh. At gas $8/MMBtu, generation cost is $52/MWh. If your grid electricity cost exceeds your CHP generation cost, run the CHP. If grid electricity drops below CHP cost (e.g., during spring shoulder months with mild weather and low grid demand), consider shutting down CHP and buying from the grid.

• Dual-fuel hedging: When hedging gas and electricity simultaneously, recognize that fixing gas costs and leaving electricity at index (or vice versa) creates a cross-commodity basis risk. If gas prices drop but electricity stays high (due to transmission constraints or non-gas generation tightness), your gas hedge underperforms while your electric bill remains high. Consider hedging both commodities on a correlated basis — many energy suppliers offer combined gas+electric portfolio management.

9. Tariff Optimization in Regulated Markets

9.1 Rate Schedule Selection

In regulated markets, the available tariff options may seem limited, but switching between rate schedules can save 5-15% on the total bill without changing consumption.

Step 1: Identify available rate schedules for your demand level and voltage. Most utilities offer 2-4 rate options for large C&I customers:

• Standard demand rate (flat energy + demand charge)
• Time-of-use rate (lower off-peak energy, higher on-peak energy + demand)
• Real-time pricing pilot (if available)
• Interruptible service rate (lower cost, utility can curtail during emergencies)

Step 2: Model 12 months of actual interval data against each available rate schedule.

For each rate schedule:
  Monthly cost = Σ(energy_charge_component) + demand_charge + customer_charge + riders

Where:
  energy_charge_component = Σ(kWh_per_interval × applicable_rate_per_kWh)
  demand_charge = max(15-min kW interval in month) × demand_rate
  For TOU rates: separate on-peak demand charge may apply

Step 3: Compare annual totals.

Step 4: Evaluate non-financial factors.

• TOU: requires ability to shift load or accept higher on-peak costs
• RTP: requires market monitoring and tolerance for price volatility
• Interruptible: requires ability to curtail load on short notice (typically 30-60 min)

9.2 Rate Case Monitoring and Response

When to intervene in a rate case:

Rate case timeline (typical):

Month 0: Utility files rate case with state PUC
Month 1-2: Intervenors file to participate
Month 3-4: Discovery (interrogatories, data requests to utility)
Month 5-7: Intervenor testimony filed
Month 8-9: Hearings
Month 10-12: PUC issues order
Month 13-15: New rates take effect (may be retroactive to filing date)

What to challenge in a rate case:

1. Rate of return on equity (ROE): Utilities typically request 10-11% ROE. Current authorized ROEs are trending down (9-10%). Challenge excessive ROE requests.
2. Rate base additions: Utilities earn their ROE on their rate base (invested capital). Challenge excessive or imprudent capital investments included in the rate base.
3. Cost allocation between rate classes: Utilities allocate total revenue requirement across residential, commercial, and industrial rate classes. Ensure your rate class is not subsidizing residential or other classes above cost causation.
4. Rate design: Even if the total revenue is approved, fight for demand-based rate design (rewards load factor management) rather than pure volumetric rates (punishes high-consumption customers regardless of load shape).

10. Emergency Procurement Protocols

10.1 Supplier Default / Bankruptcy

If your retail energy provider files for bankruptcy or fails to perform:

**Immediat

…(truncated)