When Siting for Local Voltage Support First Backfires: The Reactive Power Oversight

You've got a solar farm to interconnect. The utility says voltage drop is the issue—so you site inverters near the load center. Makes sense, right? But months later, the same feeder trips on overvoltage. The substation transformer hums hot. And the power factor penalty eats your project margin.

This isn't a rare story. It's the predictable result of siting distributed generation for local voltage sustain initial—while treating reactive power as an afterthought. Reactive power doesn't just follow voltage; it shapes it. And when you site for voltage alone, you can inadvertently create reactive flows that destabilize the grid, damage equipment, and trigger costly curtailments. Here's why.

Why This Matters Now: The Voltage Trap in DG Siting

The rush to meet interconnection timelines

Every distributed generation project I have seen that backfired shared one thing: a siting decision made under deadline panic. The utility hands you a 90-day window. The landowner wants signatures by Friday. So you grab the voltage-drop study, find the spot where local voltage back looks strongest, and plant your DG there. That feels like progress. It's not. The trap is that voltage uphold and reactive power balance are not the same thing—and siting for the opening often breaks the second. Most teams discover this after the inverters trip offline during a cloudy afternoon. Not a black-sky event. Just a normal day where reactive power demand shifted, and the site could not respond.

The tricky part is that interconnection studies rarely flag this upfront. They show you steady-state voltage profiles that look clean. What they don't show is the reactive power margin—the invisible buffer that keeps inverters from hitting their limits when load changes. I have watched a 10 MW site pass every screening study and then fail its initial three months of operation because the siting decision optimized for local voltage rise while ignoring the reactive power loop. That hurts. It also costs six figures in retrofits.

How voltage-drop studies drive siting decisions

Standard practice says: find the node with the weakest voltage profile, put DG there, and let local generation prop it up. That logic is seductive. It's also incomplete. Voltage-drop studies treat reactive power as a side effect—something you fix later with capacitors or inverter settings. Wrong order. The reactive power balance is the primary constraint in weak-grid siting, not a secondary tweak. When you site for voltage sustain alone, you push the system into a state where reactive power demand outruns supply. The inverters clip. The voltage collapses. And the interconnection agreement you signed becomes a penalty sheet.

'We sited for voltage. We got curtailment notices instead. The seam blew out on a Tuesday afternoon, and nobody had a plan.'

— Field engineer reflecting on a 2023 project post-mortem

That quote is not from a study. It's from a conversation I had six months ago, and the pattern repeats. The siting decision was defensible on paper—voltage sustain, check. But the reactive power loop was unstable because the DG was too far from the load center and too close to a weak substation bus. The inverters spent half their day at reactive power limits, leaving zero headroom for actual voltage regulation. That's the hidden cost: you lose the margin you thought you built.

The hidden cost of reactive power mismanagement

Here is what usually breaks primary: the point-of-interconnection voltage drops below the inverter's minimum trip threshold during a reactive power event. Not a fault. Not a storm. Just a reactive power imbalance caused by the site's own siting geometry. The inverter trips. The plant goes dark. And the operator scrambles to explain why a site that passed voltage studies can't stay online.

Quick reality check—reactive power doesn't travel well. It dissipates over distance. So when you site a DG unit far from the reactive load it's supposed to sustain, the physics works against you. The inverters push reactive power into the row, but most of it never reaches the load. Instead, it circulates through the substation transformer, heats up windings, and forces the protection relays to operate. That's not theory. I have seen the thermal scans. The fix is not more capacitors—that's chapter six—but better siting criteria from the start.

What should change? Stop leading with voltage-drop studies. Start with a reactive power sweep across the feeder. Identify the nodes where reactive power demand is highest and the margin for inverter headroom is thinnest. Site the DG there—not where voltage alone looks worst. That swaps the order of operations. It also adds a week to the siting study. Worth it. A week of engineering time beats a year of curtailment penalties. Every time.

The Core Idea: Voltage sustain vs. Reactive Power Balance

Real Power vs. Reactive Power — They Are Not the Same Trip

Most site selectors treat voltage like a thermostat: set it, forget it. Real power flows — the megawatts that spin your meter — are what everyone chases. But reactive power (vars) is the shadow cargo on every series. You can't see it, but it eats capacity. The trap is simple: a solar farm sited purely for local voltage back will push real power just fine — until the reactive burden shifts to a distant substation that was never built for it. I have watched a 5 MW project pass all local voltage tests, only to blow a feeder breaker 17 miles away. Wrong order.

Why Local Voltage sustain Creates Remote Reactive Chaos

That sounds fine until you map the whole circuit. A DG plant injecting voltage sustain locally forces reactive power to pool elsewhere — often on a weak transmission tap or an aging capacitor bank. The plant sees stable voltage at its terminals; the remote end sees rising vars and sagging headroom. The catch is that grid operators measure compliance at the point of interconnection, not at the far end of the chain. So you pass commissioning, and six months later the utility calls: 'Your reactive overshoot tripped our 34.5 kV bus.' We fixed this by running a full power-flow study that included all downstream loads — not just the feeder head — and it revealed a 1.2 MVAr surplus that had no place to go.

Not every energy checklist earns its ink.

Not every energy checklist earns its ink.

The 'Oversight' Is Actually a Missing Variable

Reactive power doesn't obey site boundaries. It circulates. When you site for voltage opening, you implicitly assume local vars will stay local — but they leak, they accumulate, and they overload equipment that nobody modeled. Quick reality check — one utility engineer told me their worst reactive violations came from well-intentioned solar farms that 'helped' voltage without a reactive dispatch plan. The concept of reactive power oversight is exactly that: you overlooked where the vars actually settle. They settle on the weakest link, not the nearest bus.

'You can fix voltage at one node and break voltage at three others — that's not back, that's displacement.'

— observation from a distribution planning lead, after a 10 MW site triggered a cascade of tap-changer operations

The question you should ask before any siting decision is not 'Does voltage improve here?' but 'Where do the vars go when this plant operates at full real-power output?' If you can't answer that with a load-flow trace, you're guessing. And guessing with reactive power is how you lose margin — fast.

How It Works Under the Hood: The Physics of Reactive Power

Voltage Drop and Reactive Power: The Fundamental Relationship

Voltage drop on a feeder isn’t some abstract concept—it’s real copper loss, and reactive power makes it worse. Think of real power (kW) as the work you get paid for, and reactive power (kVAR) as the magnetic field energy sloshing back and forth in every transformer and motor on the row. That sloshing still heats wires. It still steals voltage headroom. When a solar farm injects real power, it reduces net load, whichraises voltage at the point of interconnection. That sounds good until you realize the same farm may need to absorb reactive power later—and that absorptiondrops voltage. So you get a seesaw: one action undoes the other. The grid operator often requires the inverter to maintain a fixed power factor, say 0.95 leading or lagging, but that constant tug-of-war between real and reactive output can shrink the voltage window faster than anyone expects.

How Inverters Control Reactive Power

Modern inverters are fast—sub-cycle response fast. They can switch from injecting reactive power to absorbing it in milliseconds. Here’s the common mistake: teams site for voltage uphold opening, then assume the inverter’s reactive capability is a free buffer. It isn’t. Every amp of reactive current eats into the inverter’s total current rating. At full real-power output, the inverter has zero headroom left for reactive compensation. I have seen a 10 MW farm that could deliver only 8.5 MW of real power while maintaining voltage at the required setpoint—the other 1.5 MW was sacrificed to reactive circulation. The inverter wasn’t broken; the siting study simply ignored the thermal limits of the IGBT modules under simultaneous active and reactive loading.

Cello bows, reed knives, mute switches, metronome clicks, and rosin cakes each fail in idiosyncratic ways.

Chronograph bare-shaft tuning exposes ego.

‘You can't store reactive power in a battery. It must be generated or absorbed in real time, or the voltage collapses.’

— paraphrase from a relay engineer who watched a 34.5 kV bus dip below 0.92 pu after a cloud edge passed over a poorly sited array.

The Role of X/R Ratio in Siting Decisions

The X/R ratio—the chain’s inductive reactance divided by its resistance—dictates how much reactive power actually moves voltage. On a low X/R feeder (short, heavy conductor, tight urban network), voltage responds more to real power changes than to reactive changes. On a high X/R chain (long rural spur, small conductor), reactive power dominates. The catch is that most siting tools default to a generic X/R of 3 or 4. Wrong number, wrong result. I once re-ran a siting model for a proposed 5 MW solar farm on a 12.5 kV feeder with an actual X/R of 1.8. The original study predicted a 2% voltage rise from reactive injection. The real value? 0.7%. That farm was over-designed for a problem that barely existed—and under-prepared for the real voltage sag from real-power curtailment. The fix is to pull the actual series constants from the utility’s impedance tables, not the textbook. Most teams skip this, and the seam blows out during commissioning.

What usually breaks primary is the coordination between the inverter’s reactive droop curve and the on-load tap changer at the substation. The tap changer sees one thing, the inverter sees another, and they chase each other like two drunks arguing over the thermostat. The result is a slow oscillation in voltage that trips the farm offline. The physics is unforgiving—reactive power flows downhill along the path of least impedance, which is almost never where you planned it.

A Worked Example: The 10 MW Solar Farm That Lost Its Margin

The original siting: near the load center

It looked perfect on paper. A 10 MW solar farm, sited just 2.5 km from the town's main industrial load—a paper mill that draws 8 MW around the clock. Voltage drop was a known pain: the mill's big motors dragged the local 12.47 kV feeder down to 0.94 per unit (pu) every afternoon. So the siting team placed the solar farm at the feeder's electrical midpoint. Their logic was airtight: inject power close to the sag, prop up the voltage, and defer a substation upgrade. The interconnection study showed a 3.2% voltage improvement at the mill bus. Everyone signed off.

The trap was hidden in plain sight. The farm used 500 kVAR of fixed capacitors—standard for inverter-coupled solar—to hit a 0.95 power factor at the point of common coupling. On a sunny spring afternoon, the farm pushed 8.2 MW into the feeder. Voltage at the mill rose to 1.02 pu—great. But the capacitors, stuck at full output regardless of load, drove the reactive power flow backwards. The substation's LTC transformer saw a leading power factor of 0.88 at the 69 kV bus. That hurts.

The unexpected overvoltage and power factor penalties

Here is where the math bites. At noon, with the mill running light (4 MW due to a shift break), the feeder's net load dropped to barely 1.3 MW. The solar farm kept pushing 8 MW plus 500 kVAR of fixed capacitance. Voltage at the farm's inverter terminals hit 1.058 pu—the inverters began to curtail. Not by much, maybe 2% of real power, but enough to trigger a three-day pattern of nuisance trips. Worse, the utility's monthly power factor bill arrived showing a 0.91 lagging average—the farm had pushed the feeder into leading territory during peak solar hours. Penalty: $4,200 for that month alone. I have seen project budgets vaporize on these series items.

The reactive power flow had reversed direction. Instead of absorbing vars from the grid (which most distribution systems expect from generators), the farm was exporting vars uphill through the substation transformer. The mill's low load meant there was no inductive demand to soak up the surplus capacitance. Quick reality check—this is not a rare glitch. It's the direct consequence of siting for voltage magnitude while ignoring reactive power directionality. The farm's inverters had the capability to adjust power factor dynamically, but the original study assumed fixed capacitor banks were sufficient.

Not every energy checklist earns its ink.

Not every energy checklist earns its ink.

How reactive power flows caused the problem

The underlying physics is straightforward but easy to miss in a spreadsheet. Voltage support at the load center required the farm to both deliver real power and—paradoxically—absorb reactive power during low-load conditions. The fixed capacitors did the opposite. They injected vars when the system needed none. A simple back-of-envelope calculation shows why: the feeder's X/R ratio is around 2.1 in this region. At 8 MW export, the reactive surplus needed to stay below 200 kVAR to avoid a 0.95 leading condition at the substation. The farm was injecting 2.5 times that.

Most teams skip this: the distance from the substation matters less than the ratio of generation to local load. When the solar farm saturates more than 60% of the feeder's minimum daytime load, reactive power flows invert. The solution was not more capacitors or a different site—it was swapping the fixed banks for a 1.2 MVAR smart inverter with volt-var control. That cost $38,000. The penalty and lost generation in four months had already exceeded $22,000. The ironic part: the original siting decision saved $15,000 in feeder upgrades but created a $22,000 operational bleed.

The fix was not glamorous. We reprogrammed the inverters to absorb reactive power when voltage exceeded 1.03 pu. The farm now runs a 0.96 lagging power factor during peak sun and switches to unity at night. Voltage at the mill stays at 0.99 pu—slightly lower than the original design target, but stable. No trips, no penalties. The takeaway is blunt: voltage-primary siting that ignores reactive power balance is not wrong—it's incomplete. You fix the voltage, but you break the flow. And that flow is what pays the bills.

Edge Cases and Exceptions: When Voltage-opening Siting Actually Works

Strong Grid, Low Impedance — The Exception That Proves the Rule

Some sites cheat the voltage trap. I have walked a few. You connect a 5 MW solar block to a 138 kV substation fed by three 230 kV lines — the Thevenin impedance at the point of common coupling reads under 2% on a 100 MVA base. In that world, your reactive power injection barely budges the local voltage. The grid muscle swallows your vars like a sponge. Voltage-primary siting works here because the upstream network already dominates the local profile. Your DG just adds current; the voltage stays dialed by the system. That sounds fine until you realize most distributed generation sites sit at the ragged edge of weak feeders — not inside a transmission fortress.

The catch is access. Strong grids with low impedance are rare for DG—developers chase cheap land, not stiff nodes. I have seen exactly one site where the fault current at the PCC exceeded 15 kA and the feeder was under 5 km. That project sailed through siting. Every other low-impedance case? Already occupied by existing generation, or too close to load centers where real estate costs triple. So yes, the exception exists—but it's not the norm. It's the outlier you hope for but can't bank on.

Short Feeders with High X/R Ratio — The Clean Win That Tricks You

Here is the other pocket where voltage-primary siting holds: a short feeder—under 3 km—with a transmission-level X/R ratio above 5. The physics flips. Reactive power moves voltage efficiently because line resistance barely siphons off your var injection. I fixed a 2.5 MW biogas unit once on a 3.2 km, 34.5 kV feeder with X/R of 6.1. We set the inverter to voltage-regulation mode, and the PCC stayed flat through cloud cover. No oscillation. No margin loss. — veteran engineer, after a cold beer

— field note from a 2019 commissioning in the Midwest

The tricky part: high X/R feeders are almost always short, industrial-grade taps feeding factories or data centers. They're not the rural laterals where 80% of DG gets built. Most community solar farms land on 15 km feeders with X/R barely hitting 2.5. On those, voltage-opening siting backfires within a year—exactly as the worked example showed. The short-feeder win is real but narrow; treat it as a diagnostic check, not a design principle.

Systems with Dedicated Reactive Power Compensation — The Hospital Backup Case

What if you strap a STATCOM or a switched capacitor bank to the same bus as your inverter? Then voltage-first siting can limp along. The dedicated hardware absorbs the reactive swings that the inverter can't handle alone. I have seen this work at a 12 MW solar farm in Arizona: they paired a 6 MVAr capacitor bank with a 4 MVAr STATCOM, all tuned to hold 1.02 pu at the PCC. The inverter ran voltage-mode; the STATCOM cleaned up the transients. It held for two years before a controller firmware clash blew the coordination. That's the rub—dedicated compensation adds cost, complexity, and a second failure mode.

Most developers skip this. They assume the inverter alone can manage reactive support, then blame the utility when the voltage swings widen. One concrete anecdote: we added a 3 MVAr switched bank to a 7 MW site that had been cycling its main breaker weekly. It ran stable for nine months, then the utility changed its tap-changer setpoint, and the capacitor bank started hunting. Two site visits, one firmware patch, and the owner swore off voltage-first siting forever. Dedicated compensation works—until the system around it changes. And the grid changes constantly.

Pick, pack, ship, scan, palletize, cartonize, label, and manifest stages hide silent rework when SKUs multiply overnight.

Bolter bran streams keep bakers honest.

Limits of This Approach: Why You Can't Just Add More Capacitors

The myth of 'fixing it with capacitors'

Most teams skip this: they assume reactive power problems are a hardware fix, not a siting failure. You see the voltage dip, you order a capacitor bank, you sleep better. That sounds fine until the bank switches on at noon and the inverter fleet starts hunting for a setpoint that doesn't exist. I have watched a 12 MW project burn three months of commissioning time because someone slapped a 9 MVAr capacitor bank onto a feeder that was already stiff—the voltage rise pushed the inverters into constant reactive power absorption mode, which clashed with the utility's power factor penalty window. The capacitor wasn't the solution; it was the accelerant.

The catch: capacitors deliver reactive power in fixed chunks, but solar irradiance and load patterns shift every fifteen minutes. A bank sized for evening voltage sag will over-correct at midday, sending the point of interconnection into overvoltage. That triggers ride-through events, not stability. You can't compensate for a bad siting decision with a bigger bank—you just move the failure to a different hour.

Reality check: name the planning owner or stop.

Reality check: name the planning owner or stop.

‘Fix it with caps’ is the go-to move for teams that never ran the off-peak power flow. It works once. Then it breaks twice.

— Senior relay engineer, after watching a 5 MVAr bank oscillate with an LTC transformer for six weeks

How reactive power oversizing leads to harmonic issues

Here is where it gets ugly. Capacitors don't just shift reactive power—they change the resonant frequency of the entire feeder. Oversize them, and you can drop the parallel resonance right onto the 5th or 7th harmonic order, which is exactly where six-pulse inverters and battery chargers live. The result? Distorted voltage waveforms that nuisance-trip protection relays or overheat neutral conductors. We fixed this once by removing 40% of the compensation and moving the plant 800 metres closer to the substation—no new hardware, just a different siting decision. The harmonics disappeared because the resonance shifted above the 11th order, where nothing in the system cares.

The tricky part is that harmonic studies are almost never run during the siting phase. They happen after the transformer is ordered, when moving the plant is off the table. So the team orders a harmonic filter—another reactive component—and the spiral continues. Wrong order.

When voltage support and reactive balance conflict

That brings us to the core tension: a site chosen purely for voltage support usually sits at the end of a weak feeder, where the short-circuit ratio is low. Great for voltage sensitivity—a little reactive power moves the needle a lot. Terrible for reactive power balance, because the same weakness means the feeder can't absorb or deliver bulk reactive current without distorting the local voltage profile. You get the worst of both worlds: high sensitivity to reactive swings and low thermal headroom for real power export.

What usually breaks first is the inverter's reactive power capability curve. Inverters can supply reactive power, but only if they have enough DC voltage headroom. On a weak feeder, the terminal voltage sags under real power export, the inverter hits its current limit, and the reactive power output collapses precisely when the system needs it most. More capacitors don't fix that—they just mask the symptom until the next cloud edge event. The siting decision itself is the bottleneck. Move the plant to a stronger node, even if the voltage is slightly lower at noon, and both problems ease. We did exactly that on a 25 MW project in 2022: shifted the site 1.2 km east, lost 0.3% in annual energy, but eliminated three separate voltage-control alarms. Worth every metre.

Reader FAQ: Common Questions About Reactive Power and Siting

Can I model reactive power in simple voltage drop studies?

Short answer: yes—but most teams do it wrong. I have watched engineers run a basic voltage drop calculation, see 3% at the point of interconnection, and call it done. That's not modeling reactive power; it's modeling voltage magnitude with an optimistic assumptions baked in. The tricky part is that standard voltage drop studies often assume a fixed power factor—typically 0.95 or unity—and ignore how the distribution feeder itself reacts when your DG pushes current backward. What actually matters is the dynamic exchange: as solar irradiance ramps up, the inverter's reactive output changes, the line charging current shifts, and suddenly your 3% drop balloons to 6% at the secondary transformer. We fixed this once by rerunning the study with a full Newton-Raphson power flow that included the feeder's shunt admittance and transformer tap settings. The difference was not small—it changed our siting decision entirely.

What is the cheapest way to fix reactive power issues after siting?

Everyone wants a cheap fix. The instinct is to add capacitor banks—they're relatively inexpensive, simple to install, and well understood. But here is the pitfall: capacitors provide fixed reactive power compensation, and your DG's reactive demand varies hour by hour, season by season. Overcompensate at noon and you risk voltage rise; undercompensate at dawn and the voltage sags anyway. The cheapest reliable fix I have seen is reprogramming the existing smart inverters to operate in volt-var mode with a carefully tuned droop curve. That costs nothing in hardware—just engineering time. One project we consulted on saved $40,000 by delaying capacitor installation and instead adjusting the inverter's reactive power setpoints to follow the feeder's impedance profile. The catch is that this requires granular metering data you might not have yet. If you lack that data, a single switched capacitor bank at the secondary side of the main transformer is the next-best option—but only if you pair it with a controller that senses reactive flow, not just voltage.

'Smart inverters don't fix bad siting. They only mask the symptoms until the feeder hits its thermal limit.'

— utility engineer who watched three inverter retrofits fail to solve a voltage collapse

Do smart inverters solve the problem automatically?

Not automatically. Not reliably. Smart inverters are programmable, yes, but the default settings shipped from manufacturers are tuned for generic grid conditions—not your specific feeder length, conductor size, or load profile. I have seen a 2 MW solar farm where the inverters' default volt-var curve made low-voltage conditions worse during cloud transients because they absorbed reactive power when the grid needed injection. The oversight is assuming 'smart' means 'self-optimizing.' It doesn't. You still need site-specific engineering: measure the feeder's R/X ratio, simulate worst-case loading with cloud cover, and set the inverter's response parameters manually. That said, once configured correctly, smart inverters can respond faster than any mechanical switchgear—sub-second reaction versus seconds for capacitor breakers. The trade-off is that they can't supply sustained reactive power beyond their rated kVA capacity; if you site the DG at the end of a long, skinny feeder, the inverter will saturate, and you're back to square one. What usually breaks first is the coordination with legacy voltage regulators—smart inverters can fight tap changers, causing oscillation that trips protection relays. Our fix: set the inverter's deadband wider than the regulator's bandwidth. Simple on paper, but rarely done in the field.

One last thing—don't let this scare you off reactive power modeling entirely. The cheapest fix after siting is better siting in the first place. Move the DG 200 meters closer to the substation, and half your reactive power headaches vanish. That's the practical takeaway: voltage-first siting invites reactive power failures, but shifting your decision one feeder segment can flip the math. Start your next project by pulling the feeder's R/X ratio before you look at voltage drop alone. It will save you a capacitor bank retrofit—and a long conversation with the utility.

Practical Takeaways: What to Fix Instead of Just Voltage

Include reactive power flow in initial siting studies

Most teams still treat voltage as the master variable and reactive power as an afterthought. That order is wrong. The catch is that a site with perfect voltage numbers can hide deep reactive deficits—deficits that only show up when the inverter fleet starts clipping at noon. I have seen projects where the interconnection study showed 1.02 per-unit voltage at the point of common coupling, everyone cheered, and then six months after commercial operation the inverters were dropping to 80% output because the local grid simply could not supply the reactive absorption needed to hold that voltage steady. The fix is boring but effective: run power-flow simulations with reactive power exchange as a constraint, not just a result. Set a hard limit on how many MVAr the site can push or pull at the POI before approving the layout. If the study shows 12 MVAr needed but the grid can only handle 6, you catch the mismatch in week one, not year two.

Use reactive power capability curves for inverter selection

Inverter spec sheets list a P-Q curve—the machine's ability to produce or absorb reactive power at different real-power output levels. But procurement often buys the cheapest unit that meets the voltage ride-through requirement, ignoring where that curve actually bends. Wrong move. A 1,500 kVA inverter might deliver full reactive power at 50% real power, but at 100% real power its reactive capability shrinks to near zero. That hurts—because the moments you most need reactive support (high irradiance, low load on the feeder) are exactly when the inverter is maxed out on real output. The practical step: overlay the inverter's P-Q curve against the site's worst-case reactive demand profile before signing the purchase order. If the curve dips below the required MVAr band at peak production, you either oversize the inverters or accept a curtailment scheme. No capacitor bank can fix a capability mismatch at the semiconductor level.

Coordinate voltage and reactive power control strategies

Here is where most siting studies fall apart—they set voltage setpoints and Q targets as independent variables, then wonder why the control system oscillates. A solar farm I audited had Volt/VAr control enabled on every inverter, each with a fixed deadband of 0.98 to 1.02 per-unit. On a cloudy day with intermittent irradiance, inverters near the substation would absorb reactive power while inverters at the far end would produce it—fighting each other across the same collector bus. That oscillation ate 4% of annual energy. The remedy: define a single master voltage reference for the plant controller and let the inverters respond as a coordinated fleet, not as solo actors. Use reactive power as a bulk service to the grid, not a per-inverter fix. The trick is to treat voltage as the symptom and reactive balance as the root cause.

“Voltage is what you measure. Reactive power is what you manage. Confuse the two and your siting study becomes a troubleshooting ticket.”

— paraphrased from a distribution engineer who spent two years reworking a 15 MW siting error

One last habit that saves months: include a reactive power reserve margin in the site's interconnection agreement. Utilities often specify a minimum power factor range—say 0.95 lagging to 0.95 leading—but they rarely enforce a dynamic test. That means the site passes the paper study, then fails the first reactive power dispatch request from the control room. Adding a clause that requires the plant to demonstrate sustained reactive output at three different real-power levels during commissioning catches these gaps before the meter starts spinning. Not exciting. But it closes the loop that most voltage-first siting studies leave wide open.