SPP's HILL and HILLGA Framework: The In-Depth Guide for Large-Load Developers

What the tariff says, how the process works, and the questions that decide your project — from Keentel Engineering's working knowledge of the framework

Data centers, hydrogen facilities, and industrial electrification are arriving on the transmission system faster than any load class in modern history and in the Southwest Power Pool (SPP) footprint, they arrive into a framework that did not exist before 2026. SPP's High Impact Large Load (HILL) process, and its generation companion the HILL Generation Assessment (HILLGA), received FERC approval in January 2026. That makes this one of the youngest major interconnection frameworks in the country: the tariff text is new, the study practice is forming in real time, and the market's understanding is uneven.

Keentel Engineering works inside this framework daily advising large-load developers, building the load and generation models the process requires, and engaging the process's open questions as they resolve. This guide consolidates what we know: the tariff structure, the study mechanics, the economics, and most valuably the practical clarifications that are emerging as the framework meets real projects. It is written for developers, and it is deliberately thorough.

Part 1: What HILL Is — and Why It Exists

Transmission planning was built around load that grows slowly and predictably. A single campus that adds hundreds of megawatts in eighteen months breaks that model: it can consume delivery capability that other customers were counting on, stress voltage and stability margins, and arrive faster than network upgrades can be built.

The HILL process is SPP's answer: a structured study-and-integration pathway for loads large enough to matter. In broad terms, a load qualifies for HILL treatment based on size and interconnection voltage with thresholds capturing large loads connecting at transmission voltages (loads of roughly 50 MW and above at higher transmission voltages, with a lower threshold at lower voltages). If your data center or industrial facility is in the size class this article cares about, you are in the process.

The HILL Delivery Point Study (HDPS)

The center of the load-side process is the HILL Delivery Point Study a comprehensive evaluation of whether, and how, the transmission system can serve your load at your chosen delivery point. Key features:

• A 90-day study clock that begins when your application is complete and validated — not when it is first submitted. Deficient applications do not start clocks. This single fact drives the most important discipline in the process: submittal quality is schedule.
• Study content: thermal loading, voltage performance, short-circuit duty, and dynamic performance at and around the delivery point, under system-intact and contingency conditions.
• A system-strength screening using composite short-circuit-ratio criteria and critical-clearing-time thresholds. Falling short of the screening criteria does not end the project — but it routes the request into supplemental electromagnetic-transient (EMT) analysis, which runs outside the 90-day window on its own timeline. Projects with significant power-electronic load (or co-located inverter-based generation) should treat this screening as a first-class risk item and consider proactive EMT model readiness.

The load models the process requires

SPP's HILL practice requires dynamic load models that actually represent the load's behavior — not generic placeholders. For data centers, this means composite load model (CMLD) and performance-based (PERC1-class) representations parameterized to the facility's real composition: the UPS/rectifier share, motor-driven mechanical load, power-factor characteristics, and ride-through behavior. Facilities with distinct load classes on distinct feeders can be represented with the appropriate model on each feeder within a single planning case an architecture that mirrors the physical design and reconciles cleanly to the requested megawatts.

The modeling requirement is a quiet filter: developers who arrive with well-parameterized models move; developers who arrive with defaults get comment cycles.

Part 2: HILLGA — When Generation Comes With the Load

Many large-load strategies pair the load with co-located generation: batteries to shave constrained hours, gas turbines or solar to supply energy behind the fence. The HILLGA HILL Generation Assessment is the generation-side companion process for resources supporting a HILL.

The jurisdictional fork: the first decision that matters

Before anything else, a co-located generation strategy must answer one question: does the generation interconnect to facilities under SPP's functional control, or entirely outside them? The answer selects the lane, and the lanes are profoundly different.

The jurisdictional lane (generation connected to SPP-tariff facilities) runs through the HILLGA's formal machinery, with its published economics: an application fee in the low tens of thousands of dollars, study deposits in the several-hundred-thousand-dollar class scaled by project size, and most notably a security requirement computed per megawatt of Maximum Injection Capability. That last phrase deserves attention, and we return to it below.

The non-jurisdictional lane applies to generation that interconnects entirely behind the customer's meter, outside SPP functional control the classic behind-the-meter (BTM), non-exporting configuration. This lane runs under the non-jurisdictional generator provisions of SPP's business practices, and its economics are materially lighter: study costs run on an actual-cost basis with deposit-and-true-up mechanics at a small fraction of the jurisdictional lane's published deposit schedule, and the heavyweight security instruments of the jurisdictional lane do not attach.

The fork is not a formality. It determines your study path, your cost structure, your security exposure, and your process obligations. Getting the lane determination right and documented before spending anything else is the single highest-leverage act in a co-located strategy.

The zero-injection insight

For a non-exporting configuration, notice what the security formula implies: security scales with Maximum Injection Capability the maximum real power the facility can inject at the point of interconnection. A facility that is protection-limited to zero injection where relay schemes with defined trip logic make export physically impossible, not merely contractually promised has a maximum injection capability of zero. A declared zero is an argument; a protection-backed, demonstrable zero is a fact. Configurations built on the latter, with the protection scheme documented in the application (one-line, relay elements, trip logic, failure modes), put themselves in the strongest possible posture on both the jurisdictional determination and everything downstream.

One threshold worth engraving: under the non-jurisdictional provisions, small incidental injection (at or below roughly one megawatt under normal system-intact configuration) is tolerated within the lane but injection capability above that level changes the treatment of affected-system studies and pulls the configuration toward the standard interconnection machinery. The design implication: engineer the export limit below the threshold, enforce it in protection, and document it relentlessly.

Part 3: The Pathway Menu — Seven Ways to Connect, One Underlying Study

The framework's revision requests (RR696 and its companions) built out a menu of connection pathways, and understanding what each one permits and what it does not is where strategy begins. Every pathway runs the same underlying delivery-point study; they differ in what they allow before firm capacity exists and in how they eventually firm up:

Three strategic observations:

Pathways can run in parallel. A load may pursue more than one route simultaneously a conditional-service track for early energization alongside a firm track for the end-state. The combinations, their study interactions, and their decision forks are where a well-advised strategy separates from a default one.

Attachment AX is the co-location pathway. Its distinguishing power counting planned, not-yet-designated generation in the study is precisely what a load-plus-generation strategy needs. Most of Part 5's modeling questions, and all of Part 6's endgame, live inside AX.

The conditional pathways trade certainty for speed. BC and CHILL/CHILLS energize sooner and firm up later which makes the firm-versus-conditional megawatt split a genuine commercial decision, and one that pre-application analysis (testing alternate sizes at the same point of interconnection before filing) can inform rather than leave to the study results.

Part 4: How the Studies Actually Sequence — What Practice Is Revealing

The tariff describes the studies; practice is revealing how they run together. Several clarifications have emerged that materially improve what developers can plan for:

Concurrent filing is available

A HILLGA request must have a corresponding HILL request the generation assessment points at a load but the two may be filed at the same time. Early market readings assumed strict sequence; the framework as practiced allows the requests to move together, and the assessments share study assumptions in ways that can shorten combined timelines.

The battery charging study can run alongside the rest

Energy storage in these configurations is evaluated for its charging behavior (as negative generation, on customer-specified charging windows, under the system conditions that matter for charging), and parallel execution across the framework's evaluations is broadly available. The practical coupling worth knowing: these studies share modeling foundations, so as a matter of workflow, the model build is the shared critical path. The lesson generalizes: in this framework, data readiness and model quality not tariff clocks are the true schedule drivers. Every formal clock starts on complete, validated inputs; the developer who shows up with an airtight data package controls the schedule.

End-to-end duration is what you make it

SPP has, sensibly, declined to promise end-to-end timelines data-gathering and verification phases vary too much by project. The tariff anchors are the defined study windows (the 90-day delivery-point study; a 90-day stability-analysis window on the non-jurisdictional side). Around those anchors, a well-prepared project pursuing everything promptly on complete data can credibly plan in months, not years which, compared to the standard generation interconnection queue experience, is the entire point of the framework.

Part 5: The Modeling Question — How a Behind-the-Meter Battery Is Actually Studied

For co-located battery configurations, the deepest technical question in the framework is deceptively simple: when your load application includes a BTM battery as its supporting planned resource, how does the battery actually appear in the study cases? The answer determines whether your 100 or 200 MW of behind-the-fence capacity actually counts.

The modeling categories

SPP's model development practice distinguishes several treatments for distributed and storage resources: bulk-system-scale storage modeled as generating units (with discharge capability as maximum output and charging as negative minimum); registered network-resource-class units that are always modeled as generating units; unregistered resources submitted as generation, which may be modeled in-service or out-of-service with model-governance concurrence; storage serving as transmission-only assets, modeled offline in all cases (a category inapplicable to behind-the-meter facilities); and, where explicit generator modeling criteria are not met, representation within the load record through distributed-generation fields never as a silent reduction of forecast load. Netting generation against load is not how this framework works: resources are represented explicitly.

The questions that decide it and how to resolve them



Which category applies to a behind-the-meter battery submitted as the planned resource supporting a large load — and therefore which study states apply (producing in the cases where it supports the load; out-of-service in stress-test conditions; charging treated in the dedicated charging evaluation) is exactly the class of question a developer should resolve with SPP at the scoping stage rather than assume. The framework rewards those who ask it concretely: which case types apply which dispatch state, and what the load is assumed to be served by in each. In Keentel's working experience, the treatments the framework applies to well-documented, protection-limited configurations align with what a well-designed project wants but the discipline is to establish that treatment on the record for your configuration before study assumptions lock.

The practical corollary for application design: carry the battery's charging demand inside the requested load megawatts and the ten-year forecast from the first draft, so the charging evaluation and the load study rest on one consistent data set and pre-agree the dispatch and study-state assumptions at the scoping call rather than discovering them in study results.

Part 6: The Endgame From Provisional Service to Firm

Here is the part of the framework most often misunderstood, and where recent clarification has been most decisive.

The load-process pathway grants what is, in essence, provisional service a structured way to energize and operate while the supply picture completes. The framework's intent is a path toward firm service. And the transition to firm has a specific requirement: the planned generation supporting the load must itself receive firm service through the aggregate transmission service study process, or through the expedited designation process the tariff provides.

Two implications deserve emphasis:

First, market-registration alone does not carry the transition

The tariff provides registration constructs for resources that do not inject and do not seek network-resource status useful for what they are but the provisional-to-firm transition runs through the designation machinery, not around it. Strategies built on the assumption that registration alone completes the picture should be revisited.

Second, the tariff provides more than one designation route and route selection is a genuine strategy decision

Alongside the aggregate study process, an expedited designation provision exists, and for purpose-built co-located resources the choice between routes their timing, their study constructs for a resource with no network point of receipt, and their fit against the load-process milestones deserves early, deliberate evaluation rather than default assumption. The load-bearing questions a developer should resolve early, in writing: when the designation process must commence relative to the load-process milestones, how a zero-interface resource is studied within the designation machinery, and how long provisional service may run before the transition must complete. Those answers convert the endgame from a surprise into a schedule line.

Part 7: The Technical Bar — Ride-Through, System Strength, and the EMT Question

Two published requirement sets deserve their own section, because they are where large power-electronic loads most often meet the framework's hard engineering edges.

The fault ride-through requirements

SPP's HILL fault ride-through requirements (now in their second published version) set performance obligations that a data-center-class load must demonstrate, not merely assert:

• A voltage ride-through envelope at the point of interconnection — the load must remain connected through defined voltage-sag regions rather than tripping on the first disturbance.
• Transient overvoltage withstand referenced to the IEEE 2800-2022 cumulative-duration logic the load must tolerate defined overvoltage magnitudes for defined cumulative durations.
• Frequency ride-through on profiles aligned with the NERC PRC-029 lineage.
• Constant-current behavior during disturbances — the load's power electronics must behave predictably through the event, not chaotically.
• Recovery discipline — return to at least ninety percent of pre-disturbance consumption within one second of voltage recovery.
• Reclosing endurance — ride through multiple fault-clearing (reclosing) attempts in rapid succession up to six events inside ninety seconds.

Two practical notes. First, these requirements flow down into the load models: the CMLD/PERC1 parameterizations must encode the trip, transfer, and recovery behavior the requirements demand, so that simulated performance matches the obligation which is why a fault ride-through gap review (stated equipment capability versus required behavior) belongs at the front of any modeling engagement. Second, the requirements contemplate targeted exemptions variable-speed-drive equipment, for example, may qualify for relief from selected requirements where the delivery-point study demonstrates no adverse reliability impact. An exemption strategy, properly supported, is real compliance-cost money.

The system-strength screen and the EMT decision

The delivery-point study screens every request for system strength: composite short-circuit-ratio measures and a critical-clearing-time test. Published screening thresholds are demanding composite ratios in the mid-single digits and clearing times measured in fractions of a second and a large power-electronic load at a weak point of interconnection can fail them.

Failing the screen does not kill the project. It routes the request into a supplemental EMT study stage electromagnetic-transient dynamic performance, sub-synchronous oscillation screening (per the applicable CIGRE guidance), converter-driven stability, emergency power control, power quality, fault ride-through demonstration, and model verification performed outside the 90-day window, on a timeline SPP does not fix. Industry experience with supplemental EMT stages elsewhere suggests months, not weeks.

This is why the framework's own guidance points toward providing a PSCAD EMT model at the time of application: a request that arrives EMT-ready cannot be schedule-hostaged by the screen. The decision build the EMT model proactively or risk the supplemental stage is a genuine cost-versus-schedule tradeoff, and it should be made deliberately, on a screening-informed basis, before filing. An EMT model built “supplemental-stage-ready” from day one stable initialization, operation against network equivalents, suitability for oscillation screening, documented quality checks converts the framework's biggest schedule risk into a contingency you have already covered.

The verification expectation

Where both RMS (positive-sequence) and EMT models are submitted, expect them to be compared: the framework's model-verification expectations run the model sets through common disturbances and look for consistency. The models are a matched pair, not independent artifacts building them as one internally consistent package, benchmarked against each other before submission, pre-empts the least productive comment cycle in the process.

Part 8: The Disciplines That Separate Winning Applications

Across everything above, five disciplines recur:

1. Determine the lane first, in writing. The jurisdictional fork drives everything; document the determination before committing spend.

2. Make zero injection a fact, not a claim. Protection-enforced, documented, demonstrable with the small-injection threshold engineered against.

3. Treat data readiness as the schedule. Every clock starts on validated inputs. Model quality is submittal quality is schedule.

4. State your premises explicitly in every filing. New frameworks default to their oldest interpretations; applications that name their lane and their configuration foreclose the generic treatment.

5. Convert interpretations into written confirmations before filings harden. On a framework this young, the difference between an interpretation and a confirmed position is the difference between a plan and a hope. Build the confirmation step into your schedule.

Frequently Asked Questions: The In-Depth Set

Closing Thought

SPP's HILL/HILLGA framework is the most consequential development in large-load interconnection in years — a genuine fast lane, for developers disciplined enough to use it well. The winners in this framework will not be the ones with the oldest brochures. They will be the ones who arrive with protection-backed configurations, real models, airtight data packages, and every load-bearing interpretation confirmed in writing before it matters.

That is the practice we have built.

Keentel Engineering is a power systems engineering firm headquartered in Tampa, Florida, with offices in Austin, Sacramento, and Baltimore, providing Professional Engineering services in all 50 states. Our practice spans SPP and ERCOT large-load interconnection, HILL/HILLGA advisory, dynamic load and facility modeling (CMLD/PERC1, PSS\u00aeE, PSCAD), delivery-point and injection studies, and interconnection application support.

Contact: 813-389-7871 \u00b7 contact@keentelengineering.com \u00b7 keentelengineering.com

This article is provided for general informational purposes only. It reflects the SPP tariff, business practices, and Keentel's professional experience as of publication; the framework is new and evolving, and specific provisions, thresholds, and figures should be verified against current tariff documents. Nothing herein constitutes engineering, legal, or investment advice for any specific project, and no engineering studies underlie this publication. Project decisions should rest on project-specific engineering and written confirmations from SPP and the applicable Transmission Owner.