Grid Interconnection Study for a Utility-Scale Generation Facility

1. Executive Summary

The Client, an Interconnection Customer developing a large hybrid generating facility, engaged Keentel Engineering Solutions to carry out the independent engineering that surrounds a large generator interconnection request. The facility comprises approximately 400 MW of solar photovoltaic generation and a co-located 150 MW / 600 MWh battery energy storage system, interconnecting at a 345 kV point of interconnection (POI) on the interconnecting Transmission Provider's system under the pro forma Large Generator Interconnection Procedures established by FERC Order No. 2003 and restructured by FERC Order No. 2023 and Order No. 2023-A.

The question put to Keentel was not "will it connect." It was: what is the network upgrade cost exposure, when is that exposure knowable, and how much of it can be removed by engineering decisions taken before capital is irreversibly committed. Under the first-ready, first-served cluster process, an Interconnection Customer must post a study deposit, post a commercial readiness deposit and demonstrate site control before it receives any study result, and it carries withdrawal penalty exposure once inside the cluster. The commercially decisive information therefore arrives after the money is at risk. Keentel's mandate was to move that information forward in time.

Keentel performed a pre-application screening of five candidate points of interconnection, then a full set of independent shadow studies running in parallel with the Transmission Provider's cluster study: shadow power flow on the regional planning model, contingency screening across summer peak, winter peak, shoulder and light load conditions, ANSI-method short circuit and breaker duty analysis, positive-sequence transient stability, weak-grid screening, and electromagnetic transient (EMT) analysis in the identified low composite short-circuit ratio pocket. A network upgrade cost-exposure model translated each result into a dollar range with a confidence band.

Figures presented are representative of the delivered study and have been generalized to protect client confidentiality.

The screening identified an initial exposure of approximately $245 million at the Client's originally preferred POI with Network Resource Interconnection Service requested at the full 550 MVA plant rating. Moving to a second POI 7.4 miles further along the corridor and requesting Energy Resource Interconnection Service at 400 MW reduced modeled exposure to approximately $54 million, at the cost of $18 million of additional Interconnection Customer-funded gen-tie. That decision was taken in the month before the cluster application window closed, while the deposit was still refundable. The dynamic models were accepted by the Transmission Provider on first submittal, and the Large Generator Interconnection Agreement (LGIA) was executed with a final assigned network upgrade responsibility of $66.0 million and no restudy attributable to Customer-supplied data.

1.1 Study at a Glance

2. Background and Study Drivers

The Client had secured land control, an interconnection-grade resource assessment and a credible path to an offtake agreement. What it lacked was a defensible view of network upgrade cost — a number produced by the Transmission Provider, on its schedule, using its base cases and the queue positions ahead of the project. It is the largest single uncertainty in the capital plan and routinely varies by an order of magnitude between adjacent points of interconnection on the same corridor.

FERC Order No. 2023 changed the risk profile of guessing wrong. The serial process it replaced allowed a developer to enter and leave the queue cheaply; the cluster process does not. An Interconnection Customer must demonstrate site control at application, post a study deposit set by size band, and post a commercial readiness deposit or satisfy an alternative commercial readiness demonstration. Withdrawal after the cluster study begins triggers a penalty calculated on study costs incurred and on the disruption to remaining cluster members, with caps that escalate through the cluster study, the cluster restudy and the facilities study. The decision to enter a particular cluster at a particular POI is therefore close to irreversible, and it is made roughly a year before the cluster study report explains whether it was a good one.

Three drivers shaped the engagement. The Client's board required a capital exposure range before authorizing the deposits, and published queue data alone could not produce one. The corridor serving the preferred POI already hosted a large volume of prior-queued inverter-based resources, implying a crowded thermal constraint set and a real possibility of weak-grid dynamic issues. And the project was to be phased, so the Client needed to know whether Energy Resource Interconnection Service now, with Network Resource Interconnection Service later, beat requesting the higher service level at the outset.

A fourth driver emerged during the study: a neighboring Transmission Owner's system was electrically close enough that an affected system study was probable. Order No. 2023 standardized that process with defined deadlines and a pro forma study agreement, but the engineering exposure remains real, and it is not included in the host Transmission Provider's cost estimate.

3. Study Objectives and Scope of Work

The scope was written to sit alongside the Transmission Provider's studies, not to duplicate them. The cluster study is the only study that produces binding cost assignment; Keentel's studies exist to inform decisions before that assignment exists, to check it when it arrives, and to ensure Customer-supplied data never becomes the reason for a restudy.

Numbered objectives:

1. Rank candidate points of interconnection on network upgrade exposure, deliverability, curtailment risk and schedule, before the cluster application window closed.
2. Produce a shadow power flow assessment anticipating the thermal and voltage constraints the cluster study would identify, including prior-queued generation.
3. Quantify the project's distribution factor on each constraining flowgate and classify each violation as caused or contributed to, with the associated cost allocation consequence.
4. Confirm breaker interrupting duty at and adjacent to the POI using the ANSI method of IEEE C37.010, with correct inverter-based resource fault current representation.
5. Verify transient stability performance and ride-through compliance against NERC PRC-029-1 and IEEE Std 2800-2022, in positive-sequence and, where screening required it, in EMT.
6. Deliver a network upgrade cost-exposure model with explicit confidence bands, updated at each queue event.
7. Support the LGIA technical appendices and the compliance obligations attaching at commercial operation.

3.1 Scope and Deliverables

Deliverables comprised a POI screening report, a shadow system impact report, an ANSI short circuit report, a stability and EMT report, a network upgrade cost-exposure model maintained as a live document, a comment matrix against each Transmission Provider study report, and a compliance obligations register keyed to the LGIA appendices.

4. Regulatory and Standards Basis

The interconnection is governed by the Transmission Provider's Open Access Transmission Tariff, which implements the FERC pro forma Large Generator Interconnection Procedures and Large Generator Interconnection Agreement. The technical performance obligations come from NERC Reliability Standards and, increasingly, from IEEE Std 2800-2022 as incorporated by reference into tariffs and interconnection agreements.

4.1 Regulatory Register

Several cluster process elements mattered commercially. The Transmission Provider must complete the cluster study within 150 calendar days of the close of the customer engagement window, the cluster restudy within 150 calendar days, and the facilities study within 180 calendar days. The "reasonable efforts" standard that previously excused delay is gone, replaced by penalties on the order of $1,000 per business day for a late cluster study, $2,000 for a late cluster restudy or affected system study and $2,500 for a late facilities study, subject to a cap tied to study costs. Study deposits are set by size band, the largest applying above 80 MW. Commercial readiness may be demonstrated by an executed offtake agreement, by evidence of generating facility equipment procurement, or by an additional commercial readiness deposit benchmarked to assigned network upgrade costs. Order No. 2023 also requires evaluation of alternative transmission technologies — dynamic line ratings, advanced power flow control and advanced conductors — as substitutes for conventional network upgrades.

4.2 Standards Register

The inverter-based resource landscape deserves specific treatment because it changed materially during the study. FERC Order No. 901 directed NERC to develop reliability standards addressing IBR performance, data sharing, model validation, and planning and operating studies. The first product relevant here is NERC PRC-029-1, Frequency and Voltage Ride-Through Requirements for Inverter-Based Resources, adopted by the NERC Board in October 2024 and filed with FERC in November 2024, becoming effective in 2026 — twelve months after jurisdictional approval, effective the following calendar quarter. It applies to BES inverter-based resources and to non-BES IBRs of 20 MVA or greater aggregate capacity connected at 60 kV or above, with design compliance for non-BES resources required by January 1, 2027 or the effective date, whichever is later.

The substantive requirements are R1 for voltage ride-through, R2 for reactive current and control behavior during and after disturbances, R3 for frequency ride-through, and R4 for documentation of hardware limitations applicable to legacy resources. The expectations that drove design decisions here were ride-through of phase-angle jumps of at least 25 electrical degrees, the prohibition on ceasing current injection within the ride-through zone, return to pre-disturbance current exchange within approximately five cycles of voltage recovery, and ride-through of frequency rates of change up to 5 Hz/s. Momentary cessation is effectively disallowed inside the ride-through envelope, which reverses the default behavior of a good deal of installed inverter firmware.

Two adjacent points were confirmed early. NERC PRC-024-4 now applies to synchronous generators, synchronous condensers, and Type 1 and Type 2 wind resources only; PRC-029-1 supersedes it for inverter-based resources, so this facility is a PRC-029-1 resource. And PRC-028-1 requires disturbance monitoring equipment on IBRs, supplying the measured event data used to demonstrate operational, as opposed to design, compliance with PRC-029-1, with PRC-030-1 governing reporting and analysis of unexpected performance. IEEE Std 2800-2022 defines the expectations; the PRC standards make an enforceable, evidence-based subset apply. The practical instruction to the Client was that ride-through commitments in the LGIA technical appendices would later be audited against recorded fault data, so they had to be commitments the inverters could meet.

5. System Modeling and Data Development

5.1 Model Data Sources

The steady-state model was built on the regional planning base cases, roughly 68,000 buses at full size, with no reduction or equivalencing of the transmission network. Reduction is attractive for run time and dangerous for cost allocation, because distribution factors are sensitive to the very paths a reduction removes, and any constraint identified on a reduced model would be unusable as a comment against the Transmission Provider's results. Run time was managed instead with Python-driven batch execution of PSS®E and parallel case dispatch.

The generating facility was represented at the level of detail the cluster study would use. The collector system was equivalenced into a single collector branch impedance, an equivalent pad-mounted transformer and an equivalent collector line, following the standard collector equivalencing method, with the solar and storage blocks represented as separate equivalent machines behind a common 34.5 kV collector bus. This matters: representing a hybrid facility as one aggregate machine hides the fact that the storage can absorb as well as inject, and the light load cases turn on exactly that.

Facility parameters used were: aggregate inverter rating 550 MVA; two 34.5/345 kV generator step-up transformers, 300 MVA ONAN/ONAF/ODAF each, impedance 10.5 percent on the ONAN base; a radial 345 kV generator lead of 11.6 miles, twin-bundled 1272 kcmil ACSR per phase; and reactive plant comprising the inverters' own ±0.33 pu dynamic capability plus a 15 MVAr shunt reactor at the collector bus to offset gen-tie charging at light load. Plant rated current at 345 kV is 920 A.

Three data gaps had to be closed. Dynamic model behavior under PRC-029-1 depends on inverter firmware, and procurement had not closed, so both candidate revisions were modeled separately until Month 8. GSU impedance was a design value until factory test reports existed; a ±7.5 percent sensitivity was carried and the tested value at Month 24 fell within it. Prior-queued generation dispatch is set by tariff rules rather than engineering judgment; the published rules were adopted exactly, with a sensitivity on the two ambiguous cases.

Model validation was performed three ways. The shadow base case was benchmarked against the Transmission Provider's posted cluster study base case at Month 13, with branch flows on the monitored element set agreeing within 1.8 percent at summer peak and 3.1 percent at light load, the difference attributable to load allocation within the load zone. The positive-sequence facility model was benchmarked against the OEM EMT model across six disturbances, with active and reactive current trajectories agreeing within 5 percent during the fault and recovery windows. The ASPEN OneLiner model was reconciled against published fault duties at four buses, agreeing within 2.4 percent before the project was added.

6. Study Methodology and Assumptions

6.1 Tools and Rationale

Using the same platform as the Transmission Provider is deliberate. A shadow study whose results cannot be reproduced on the reviewer's platform is a debating position, not a comment. Every case submitted in a comment matrix was delivered as a runnable saved case with a documented dispatch.

6.2 Case and Scenario Matrix

Load levels were 100 percent summer peak, 88 percent winter peak, 65 percent shoulder and 42 percent light load of the regional coincident peak. The critical case is not always the peak. Case L2 — light load, full solar output, battery neither charging nor discharging — produced the highest net injection relative to local load and drove both the reverse-flow thermal result and the high-voltage result. Cluster studies that examine only summer peak miss this.

Prior-queued generation was included per tariff: all higher-queued requests in the same and earlier clusters modeled in service at their requested service level, lower-queued requests excluded. Eleven higher-queued projects totalling 4,180 MW were included, of which 3,050 MW were inverter-based. Queue position is not a technical parameter but behaves like one: the same project twenty positions earlier would have seen a materially smaller constraint set, because it would have been the marginal contributor to fewer overloads.

6.3 Acceptance Criteria

Fault clearing assumptions were four cycles for normal clearing at 345 kV and ten cycles for delayed clearing with a stuck breaker. Contingency definitions followed TPL-001 categories P1 through P7, with P6 executed with a system adjustment between the two events.

7. Analysis and Results

7.1 Pre-Application Screening and POI Selection

Five candidate points of interconnection were screened between Month −6 and Month −2, using a reduced monitored element set, one summer peak and one light load case, and a first-order cost model — on the principle that a screen should be cheap enough to run five times and directionally right rather than precise and late.

Option 1 was the Client's original preference because it required only 4.2 miles of generator lead. It sat, however, on the loaded side of a 345 kV corridor where the project's distribution factor on the binding flowgate was 11.4 percent — high enough to be assigned the constraint and the resulting line rebuild. Option 2 lay 7.4 miles further along the corridor, beyond a substation that split the flow, reducing the distribution factor on the same flowgate to 2.1 percent, below the tariff's 5 percent cutoff and therefore outside the constraint set entirely.

Option 5 is the option developers most often overlook. Surplus interconnection service, established by FERC Order No. 845 and expanded by Order No. 2023, allows unused service at an existing facility's POI to be taken up by another resource through an expedited process without a full cluster study. Where it fits it is the cheapest interconnection available; here the surplus was 120 MW against a 400 MW requirement, and the screen documented the rejection so the option would not be reopened later.

7.2 Steady-State Power Flow and Thermal Results

The shadow power flow monitored 1,840 elements against 6,750 contingencies — roughly 12.4 million element-contingency evaluations per case and 74 million across the matrix — executed as a Python-driven batch in PSS®E with AC contingency solution and a DC pre-screen to prioritize the AC set.

Three observations shaped the commercial outcome. The 345 kV corridor segment remains the binding regional constraint, but at Option 2 the project's contribution falls below the assignment cutoff, so the 118.7 percent figure is driven by the cluster in aggregate rather than by this project. The 345/138 kV transformer overload is unambiguously caused by this project: base loading of 88.6 percent is well inside criteria and the project alone takes it to 104.1 percent, so it was assigned in full and produced the transformer uprate. The 138 kV loop element is already over criteria at 101.8 percent without the project, which contributes 4.4 percentage points — a pre-existing condition with a wholly different cost allocation treatment.

7.3 Distribution Factors, Cause versus Contribution

The distinction between causing an overload and contributing to one is where most of the money is, and it is decided by distribution factors rather than by narrative.

The tariff applies a 5 percent distribution factor cutoff for inclusion in the constraint set: a project below the cutoff on an element is not assigned responsibility for it, however overloaded it may be. Above the cutoff, treatment depends on the base case. Where the element is within criteria without the cluster and violates with it, the responsible customers are assigned the upgrade. Where it already violates in the base case, cluster members are assigned a pro rata share of the incremental worsening and the underlying condition remains a transmission planning obligation.

Two engineering points follow. Distribution factors are a function of topology, not project size, so a smaller project at a bad POI can be assigned a very large upgrade while a larger project at a good POI is assigned nothing; 7.4 miles of gen-tie produced a 9.3 percentage point swing here. And the cutoff creates a cliff — a project at 5.2 percent is assigned and one at 4.8 percent is not, for a difference well inside base case modeling uncertainty. Keentel therefore reported distribution factors with their sensitivity band attached and treated any value within 1 percentage point of the cutoff as assigned for planning purposes.

7.4 Reactive Power and Voltage Performance

FERC Order No. 827 requires non-synchronous facilities to provide reactive power across a 0.95 leading to 0.95 lagging power factor range at the POI, which at 400 MW of service is ±131.5 MVAr. The inverters' aggregate ±0.33 pu dynamic capability on 550 MVA is 181 MVAr at the collector bus, covering the requirement plus GSU and gen-tie losses across the operating range, with margin at reduced solar output because the storage inverters retain full reactive capability when not exporting real power.

The light load post-contingency value of 1.061 pu exceeded criteria, caused by full solar export into a lightly loaded network combined with gen-tie charging following loss of a parallel circuit. The resolution was operational rather than capital: the plant controller holds a POI voltage schedule with droop rather than a fixed power factor, absorbing 62 MVAr post-contingency, and the 15 MVAr collector-bus reactor reduces steady-state charging at minimum export. Voltage schedule control at the POI is also what the Transmission Provider requires under its VAR-002 obligations, so the change aligned both requirements.

7.5 Short Circuit Analysis and Breaker Duty

Short circuit analysis was performed in ASPEN OneLiner using the ANSI method of IEEE C37.010, with ratings from IEEE C37.06 and the Transmission Provider's equipment database. Forty-six breakers within two buses of the POI at 345 kV and 138 kV were screened.

The representation of the inverter-based resource is the technical crux. A synchronous machine subjected to a close-in fault delivers current governed by its subtransient, transient and synchronous reactances — typically five to seven times rated current initially, decaying over hundreds of milliseconds. An inverter does not. Its output is a controlled current source bounded by the semiconductor's thermal and safe operating area limits, with the control acting within a fraction of a cycle, so the contribution is roughly 1.1 to 1.3 times rated current, sustained rather than decaying, at a phase angle set by the control strategy rather than machine physics. At a rated current of 920 A, 1.2 pu was used, or 1.10 kA.

The most valuable single result concerned two 345 kV breakers at the adjacent station, rated 40 kA symmetrical interrupting. The Transmission Provider's initial screen represented all cluster inverter-based resources at 2.0 pu of rated current — a conservative default used where OEM data is unavailable — with a simple E/X calculation and no X/R-dependent adjustment. That produced 41.2 kA, above rating, and assigned replacement of both breakers at an estimated $6.8 million.

Keentel's ANSI-method study used the OEM-confirmed 1.2 pu limit, a two-cycle contact parting time for the installed three-cycle breakers, and the C37.010 multiplying factor appropriate to the local-to-remote source ratio at that bus, predominantly remote with an X/R of 13.6. Total interrupting duty was 38.1 kA, 95.3 percent of rating, with the project contributing 0.9 kA. Submitted with the runnable OneLiner case and the OEM fault current declaration, the comment was accepted during the facilities study and the $6.8 million assignment removed. Both breakers remain flagged above the 90 percent threshold, which is a genuine finding: the next interconnection on that bus will very likely be assigned their replacement.

7.6 Stability Analysis and Ride-Through Verification

Positive-sequence transient stability was performed in PSS®E across 42 disturbances in four dispatch scenarios, for 168 simulations: three-phase faults with four-cycle normal clearing at seven 345 kV buses, single line-to-ground faults with ten-cycle delayed clearing and a stuck breaker, loss of the largest local generating unit, and loss of the largest local load block. Dynamic models were the industry-standard second-generation renewable energy models — renewable generator, electrical control and plant control modules — supplied by the OEM with parameters matched to the procured firmware, plus the storage-specific electrical control module representing state-of-charge limits and bidirectional operation. Data submission followed NERC MOD-032 and the Transmission Provider's data request; MOD-033 validation obligations fall on the Planning Coordinator but rely entirely on the quality of what the Interconnection Customer submits.

The first submitted model failed. On the three-phase fault at the adjacent 345 kV bus with 0.42 pu retained voltage it exhibited momentary cessation: the inverters blocked current injection for 0.10 seconds, then ramped back, taking 0.38 seconds to return to pre-disturbance current. That is inside the PRC-029-1 ride-through envelope and therefore not compliant — R1 requires ride-through, R2 requires continued reactive current exchange and return to pre-disturbance current within approximately five cycles of voltage recovery, and momentary cessation is effectively disallowed within the envelope.

The fix was a parameter and firmware configuration change, not hardware: momentary cessation disabled, the low-voltage current-injection threshold lowered below the deepest retained voltage in the study set, reactive current priority enabled during the fault, and the post-fault active current recovery ramp shortened. The revised model returned to pre-disturbance current exchange within 3.4 cycles. Phase-angle jump ride-through was verified at 32 electrical degrees against the 25 degree expectation, and frequency ride-through at a rate of change of frequency of 5 Hz/s. Had the deficient model gone to the Transmission Provider it would have been rejected and the cluster study paused for the Customer; instead the defect was found at Month 7 in Keentel's own runs and the corrected model accepted on first submittal.

7.7 Weak-Grid Screening and EMT Analysis

Short-circuit ratio at the POI, pre-project short circuit MVA divided by facility rating, is 10,993 / 550 = 20.0. On its own that is a strong grid. It is also the wrong number.

Composite short-circuit ratio accounts for the fact that inverter-based resources electrically close to one another share the same short circuit strength and do not contribute meaningfully to it. Over the pocket containing this project and the existing and prior-queued inverter-based resources within it, CSCR is 3.09 intact and 2.40 under the governing N-1 — at or below the Transmission Provider's threshold of 3.0, which triggers a mandatory EMT study.

EMT analysis was performed in PSCAD/EMTDC at a 10 microsecond time step using the OEM's encrypted, firmware-matched model, with the surrounding network represented by a frequency-dependent network equivalent extended far enough to retain the neighboring IBR plants explicitly rather than as sources. Twenty-eight cases covered fault ride-through, weak-grid steady-state stability, control interaction screening and energization.

The significant finding was an 18 Hz weakly damped oscillation in the facility's reactive output under the N-1-1 condition where CSCR falls to 2.40, with the plant-level voltage controller at default settings: damping 1.2 percent against a 3.0 percent criterion. The mechanism is the classic weak-grid control interaction — the voltage regulation loop sees a high grid impedance, so reactive output produces a large voltage response and loop gain approaches instability. Retuning the plant-level voltage controller, proportional gain from 4.0 to 1.8 pu and reactive loop time constant from 0.02 to 0.08 seconds, raised damping to 6.4 percent at the cost of a slower steady-state response that remained within the tariff requirement. No interaction with neighboring plants remained after the retune. The positive-sequence model was then re-benchmarked against the retuned EMT model and its plant controller parameters updated to match, because MOD-026 verification will later be performed against the positive-sequence model and a divergence between the model on file and the equipment as configured is a finding waiting to happen.

7.8 Affected System Study and Queue Events

The affected system study was anticipated at Month −2 by the screening, which identified a 138 kV terminal on the neighboring Transmission Owner's system where the project's distribution factor was 8.7 percent. Under Order No. 2023 affected system studies follow a standardized process with defined deadlines and a pro forma study agreement, but the resulting cost is assigned separately from the host Transmission Provider's network upgrades and is often discovered late. Here it was budgeted at $7.4 million at Month −2 and confirmed at $7.4 million by the affected system study report at Month 19.

At Month 15 two cluster members withdrew, one a 780 MW project, and the tariff required a cluster restudy. The restudy raised the assigned host-system network upgrade cost from $47.0 million to $65.4 million, of which $6.8 million was the disputed breaker replacement subsequently removed, giving $58.6 million; the withdrawing customer's penalty returned a partial credit to remaining members, and the restudy cost four months of schedule. Keentel's shadow restudy, run within three weeks of the withdrawal notice, projected $61.2 million against the eventual $65.4 million — a 6.4 percent difference, and enough for the Client's board to confirm rather than revisit its investment decision four months before the Transmission Provider produced a number.

8. Sensitivity and Scenario Analysis

Two sensitivities dominated and four did not, which is itself the useful finding.

Interconnection service level was the largest lever after POI selection. Requesting Network Resource Interconnection Service at the full 550 MVA plant rating triggered deliverability testing against a broader set of conditions and assigned $139 million of upgrades even at Option 2. Energy Resource Interconnection Service at 400 MW assigns only the upgrades needed for the facility to inject energy safely and reliably; it does not confer the deliverability status NRIS provides. FERC Order No. 845 explicitly permits an Interconnection Customer to request interconnection service below the generating facility capacity, which is what allowed a 550 MVA plant to request 400 MW of service, with the plant controller enforcing the limit at the POI. The Client's economics did not depend on capacity accreditation in the first phase, and a later NRIS request remains available as a separate interconnection request.

The inverter fault contribution sensitivity produced money, as described in Section 7.5. The spread between 1.1 pu and 2.0 pu is 3.3 kA at the adjacent 345 kV bus — the difference between 94.8 percent and 103 percent of a 40 kA rating, and therefore between no assignment and $6.8 million. There is no engineering basis for the 2.0 pu figure once OEM data exists; it is a placeholder for missing data, and the cure is to supply the data.

The insensitive variables were equally worth documenting. GSU impedance at ±7.5 percent changed no constraint classification. Load allocation at ±5 percent moved the binding distribution factor by 0.7 percentage points, which mattered only because one element sat near the cutoff. Dispatch interpretation moved cost by $4.2 million, inside the exposure model's confidence band. Reporting these explicitly stopped the project team spending effort refining inputs that could not change the answer.

9. Findings and Root Cause Assessment

F1 and F2 share a root cause worth stating plainly: the commercial terms of an interconnection are set by two decisions — where to connect and how much service to request — made before any study result exists. Neither is a design decision in the ordinary sense. Both are taken in the pre-application window and effectively locked at submission, because changing either afterwards is a material modification with queue-position consequences.

F3 arises because inverter firmware defaults were established under an earlier regulatory regime in which momentary cessation was an accepted, and in some tariffs a required, response to depressed voltage. PRC-029-1 reverses that expectation, and any model delivered in a default configuration should be assumed non-compliant until tested.

F4 arises because short-circuit ratio computed for a single plant is a poor indicator of dynamic robustness where inverter-based resources cluster geographically, as they do wherever the resource is good. A single-plant SCR of 20.0 and a composite SCR of 3.09 describe the same bus.

F5 reflects a structural feature of cluster studies: with dozens of projects and limited time, the Transmission Provider applies conservative defaults where data is absent, and those defaults become assigned cost unless a customer supplies better data with a reproducible case. F8 is not a defect but how the cluster process works, and the reason withdrawal penalties exist; it should be carried as a budgeted contingency, not treated as a surprise.

10. Mitigation Options and Recommendations

The recommendation was Option 2 with Energy Resource Interconnection Service at 400 MW, with four supporting measures.

First, the additional 7.4 miles of 345 kV generator lead, at approximately $18 million, is an Interconnection Customer-funded Interconnection Facility rather than a Network Upgrade. It is not subject to the cost allocation and reimbursement mechanics of network upgrades, it is within the Customer's construction control, and it is a known cost rather than an estimate that can escalate through restudy. Trading $18 million of controllable cost for $167 million of uncontrollable cost was the central recommendation of the engagement.

Second, request ERIS now and preserve the NRIS path. The first phase economics rest on energy revenue and storage arbitrage, neither requiring deliverability status. A subsequent NRIS request can be made as a separate interconnection request once the offtake structure justifies it, and will be studied against a network that by then includes the upgrades this project funded.

Third, pursue provisional interconnection service. The 345/138 kV transformer uprate carried a 34-month equipment lead time, beyond the Client's target commercial operation date. FERC Order No. 845 established provisional interconnection service, allowing interconnection and operation on a limited basis prior to completion of network upgrades, subject to a provisional interconnection study defining the operating limits. Keentel's shadow analysis identified a stable operating limit of 235 MW at the POI pending transformer completion, enforced through the plant controller and backed by a remedial action scheme that reduces facility output on loss of the constrained element. The Transmission Provider's provisional study subsequently confirmed 235 MW.

Fourth, adopt the retuned plant controller settings and the POI voltage schedule as design commitments written into the LGIA technical appendices rather than left as commissioning-time settings. A control parameter that exists only in a commissioning record will be changed by a firmware update without anyone recognizing the compliance consequence.

11. Implementation Support and Field Validation

Keentel supported the LGIA technical appendices, the compliance obligations register, and the commissioning-stage verification that closes the loop between study and reality. Appendix A recorded the Interconnection Facilities, Network Upgrades and Distribution Upgrades, distinguishing the Customer-funded gen-tie and POI equipment from the assigned network upgrades. Appendix B recorded milestones, including the provisional service milestone and the transformer completion date that lifts the 235 MW limit. Appendix C recorded the interconnection details including metering, telemetry and communications. The technical appendix addressing non-synchronous generating facility requirements recorded the reactive capability, the POI voltage schedule and droop, the primary frequency response settings, and the ride-through commitments.

MOD-025 real and reactive capability verification was performed at commissioning through staged testing at three output levels, confirming the 400 MW interconnection service limit at the POI and reactive capability slightly exceeding the ±131.5 MVAr requirement, the leading value being binding because gen-tie charging works against absorption. The tested GSU impedance of 10.72 percent fell inside the ±7.5 percent sensitivity band and required only a parameter correction.

The most informative validation came from an external transmission fault approximately five months after commercial operation. The facility's disturbance monitoring equipment, installed to meet PRC-028-1, captured the event: retained voltage at the POI of 0.47 pu for four cycles, followed by recovery. The recorded response showed no cessation of current injection, reactive current priority during the depression, and return to pre-disturbance current exchange within 3.6 cycles of voltage recovery, against 3.4 cycles simulated and the approximately five-cycle expectation of PRC-029-1 R2. That recording is the evidence of operational compliance; the study was the design case, and PRC-028-1 data is what demonstrates the design was real. No PRC-030-1 report was required because performance was as expected.

Remaining compliance obligations were placed on a register with owners and periodicities: MOD-026 voltage control model verification and MOD-027 active power and frequency control model verification on the periodicity the standards set and following applicable equipment or control changes; PRC-019 coordination of the plant and inverter voltage-regulating controls with equipment capabilities and protection settings, which for an IBR means demonstrating that the ride-through envelope is not cut short by protection; PRC-025 generator protection loadability applied to the GSU and gen-tie protection; and VAR-002 operation to the voltage schedule.

12. Results and Value Delivered

The headline number is the $179 million reduction between the $245 million screened exposure at the Client's original intent and the $66.0 million finally assigned in the executed LGIA, of which $18 million was reinvested in Customer-controlled gen-tie. The more durable value is in the last row: the decision was made at Month −2, before the study deposit, the commercial readiness deposit and the withdrawal penalty exposure attached. Every element of the analysis could have been performed at Month 14 instead, and would have been worth nothing, because by then the only available responses were to accept the cost or to withdraw and pay the penalty.

The curtailment figures come from a congestion screening informed by historical flow duration on the constraining elements, not from a full production cost simulation, and were reported with that limitation stated. At Option 1 under ERIS the project would have faced dispatch-down of up to 165 MW during approximately 610 hours per year, roughly 74 GWh. At Option 2 with the assigned upgrades in service, modeled exposure falls to 38 MW peak and roughly 9 GWh per year.

13. Lessons Learned and Engineering Insights

The commercially decisive engineering happens before the queue application. By the time a cluster study report exists, the two variables that set the cost — point of interconnection and requested service level — are fixed, and changing either is a material modification with queue-position consequences. A pre-application screen that is directionally correct and delivered in eight weeks is worth more than a precise study delivered after the application window closes.

Single-plant short-circuit ratio is not a measure of grid strength for a modern IBR pocket. This project's SCR of 20.0 would, on a naive reading, have ruled out any weak-grid concern. Its composite SCR of 3.09 falling to 2.40 under N-1 put it squarely into mandatory EMT territory and produced a genuine control interaction finding. Wherever the renewable resource is good, inverter-based resources cluster, and the appropriate screening metric is the composite or weighted ratio computed over the pocket, evaluated under contingency as well as intact conditions.

Conservative defaults in a Transmission Provider's screen are a data problem, and the customer owns the data. The 2.0 pu inverter fault current assumption that produced a $6.8 million breaker assignment was a reasonable placeholder in the absence of OEM information. The cure was not argument but evidence: an OEM fault current declaration, an ANSI-method calculation with defensible contact parting time and multiplying factors, and a runnable case the reviewer could execute. Comments supported that way were accepted; comments elsewhere in the cluster that consisted of assertions were not.

Inverter firmware defaults lag the regulatory requirement, and models inherit the defaults. The first dynamic model submitted for this facility exhibited momentary cessation inside the ride-through envelope, which PRC-029-1 effectively disallows. This was not an OEM failing so much as a configuration correct under a previous regime. Any positive-sequence or EMT model received from an OEM should be tested against the PRC-029-1 envelope before it is submitted anywhere, and the firmware revision recorded alongside the parameters, because a firmware update can silently invalidate both the model and the compliance position.

Distinguish Customer-controlled cost from Provider-assigned cost, and prefer the former. An extra 7.4 miles of gen-tie is a construction problem with a bid price. A network upgrade assignment is an estimate that moves at every restudy, is exposed to the withdrawal behavior of unrelated projects, and is outside the Customer's control. The recommendation to spend $18 million to avoid $167 million was primarily a cost-certainty argument, and the subsequent $18.4 million restudy increase demonstrated the point.

Report insensitivity as carefully as sensitivity. Documenting that GSU impedance, load allocation and dispatch interpretation could not change the constraint classification stopped a project team spending three weeks refining inputs. Honest treatment of what does not matter is what makes statements about what does matter credible.

14. Keentel Capability Summary

• Pre-application interconnection screening: candidate POI ranking, first-order upgrade exposure, curtailment and deliverability screening
• Shadow cluster studies: steady-state power flow, AC contingency analysis across TPL-001 P0 through P7, distribution factor and cost allocation analysis
• Short circuit and equipment duty: ANSI-method fault studies per IEEE C37.010, breaker interrupting duty screening, IBR fault contribution characterization
• Transient stability: positive-sequence dynamic simulation, ride-through verification against PRC-029-1 and IEEE Std 2800-2022, voltage recovery and damping assessment
• Weak-grid and EMT analysis: SCR, composite SCR and weighted SCR screening, PSCAD/EMTDC studies, control interaction identification and controller retuning
• Model development and quality: MOD-032 data packages, library and OEM model configuration, positive-sequence to EMT benchmarking, first-submittal model acceptance
• Interconnection process support: cluster application readiness, comment matrices against Provider study reports, affected system study support, provisional and surplus service evaluation
• Commercial analysis: network upgrade cost-exposure modeling with confidence bands, restudy impact projection, service level right-sizing
• Compliance support: LGIA technical appendices, MOD-025 test planning and witness, MOD-026 and MOD-027 verification planning, PRC-019, PRC-025 and PRC-028-1 obligation registers

15. Frequently Asked Questions

16. Glossary of Terms and Abbreviations

17. Confidentiality and Use Statement

This case study has been prepared for informational purposes. All client identities, project locations, contract details, and proprietary data have been withheld or generalized. Technical parameters, study results, and figures presented are representative of work performed by Keentel Engineering Solutions and have been adapted so that no individual project, owner, or facility can be identified. Nothing in this document constitutes a design recommendation for any specific installation. Any reuse of the methodologies described requires project-specific engineering analysis by a qualified professional engineer.