A 7 kW array on a roof in Denver and the same 7 kW array on a roof in Seattle are two completely different power plants. One will push past 10,500 kWh a year. The other will struggle to clear 7,800. Sizing a solar system correctly — whether it is a 6 kW residential job or a 600 kW commercial rooftop — comes down to doing the math in the right order: load first, sun second, losses third, then equipment. We have spec'd and supplied hundreds of systems through PES Supply, and the failures we get called about almost always trace back to somebody skipping one of those four steps. This guide walks the full methodology we use on real bids, with the same worksheets, loss tables, and worked examples our inside sales team runs daily. Everything here scales: the arithmetic that sizes a cabin kit is the same arithmetic that sizes a warehouse, just with more zeros.
1. Load Analysis Methodology
Every sizing conversation starts with energy, not panels. Kilowatt-hours consumed per day is the number that drives everything downstream — array size, inverter rating, battery capacity, and wire gauge. Pull twelve months of utility bills before you draw a single string. One summer bill and a guess will get you a system that looks great in March and leaves the customer short in August. If the utility offers Green Button or interval data, take it even for residential jobs; the shape of consumption tells you things the totals never will, like whether the evening peak argues for a battery from day one.
1.1 Residential Load Calculation
For homes, take the annual kWh from the utility and divide by 365. That is your design target. If the customer wants 100% offset, you size to that number. If they want 80% offset — common when roof space or budget is tight — multiply first, then size. Watch for electrification creep: a customer planning a heat pump, an EV, or an induction range in the next two years needs those loads added now. We learned this the hard way on a 9 kW install outside Salem where the homeowner bought a Model Y three months after commissioning. Adding 3,000 kWh of annual charging load after the fact meant a second array face and a service-panel upgrade nobody had budgeted.
These are the electrification additions we now ask about on every single residential survey, because each one moves the array size materially:
| Planned Addition | Typical Added Annual Load | Extra Array Needed (at 5 PSH, 15% losses) |
|---|---|---|
| EV charging (~12,000 mi/yr) | 3,000–4,000 kWh | +1.9–2.6 kW (5–7 × 400 W panels) |
| Heat pump (replaces gas furnace) | 3,000–6,000 kWh | +1.9–3.9 kW (5–10 panels) |
| Heat pump water heater | 1,200–1,800 kWh | +0.8–1.2 kW (2–3 panels) |
| Induction range | 250–400 kWh | +0.2–0.3 kW (1 panel) |
| Hot tub / spa | 2,000–3,500 kWh | +1.3–2.3 kW (4–6 panels) |
When bills are unavailable — new construction, off-grid cabins, recent purchases — build the load from the appliances. The table below is the same duty-cycle-adjusted worksheet we hand to our design team. Note the refrigerator: 150 W nameplate, but it cycles, so you count roughly 40% duty across 24 hours.
| Appliance | Wattage (W) | Hours/Day | Duty Adjustment | Daily Energy (Wh) |
|---|---|---|---|---|
| Refrigerator | 150 | 24 | ~40% duty cycle | 1,440 |
| LED lighting (10 fixtures) | 100 | 5 | — | 500 |
| Central air conditioner | 3,500 | 6 | — | 21,000 |
| Clothes washer | 500 | 1 | — | 500 |
| Television | 150 | 4 | — | 600 |
| Miscellaneous / receptacle loads | 1,000 | 8 | — | 8,000 |
| Total daily load | — | — | — | 32,040 Wh (32.0 kWh) |
That 32 kWh/day profile is a heavy-use home — the AC dominates. Strip the air conditioning and the same house drops near 11 kWh/day, which is why two homes with identical square footage can need arrays that differ by a factor of three. Anyone who quotes you a system size from your floor area is selling, not engineering.
1.2 Commercial Load Calculation
Commercial work is interval data or nothing. Ask the utility for 12 months of 15-minute interval data (most will export it as CSV under a customer authorization). You are looking for three things: total annual kWh, the monthly peak kW demand, and the shape of the load curve. A warehouse running 7 a.m. to 6 p.m. on weekdays is a nearly perfect solar match — production peaks land inside the consumption window. A cold-storage facility with compressors cycling all night is a battery candidate, not a straight grid-tie candidate.
Demand charges change the design target. If the rate tariff includes $15–$25 per kW of monthly peak demand, shaving 100 kW of coincident peak is worth $18,000–$30,000 a year before you sell a single kilowatt-hour. That value depends on whether your peak is coincident with solar noon — afternoon peakers win, dawn peakers do not. We cover the demand-charge math and the full financial modeling in our commercial array ROI deep dive, and the NEC load-calculation side pairs with standard panel-busbar and 120%-rule methods. Check the building's load factor too: annual kWh ÷ (peak kW × 8,760). Below 0.35, storage often beats additional PV on pure economics.
2. Solar Resource Assessment and Peak Sun Hours
Peak sun hours (PSH) is the single most misused number in residential solar. One PSH is one hour of 1,000 W/m² irradiance on the array plane. A day with 5 PSH delivers the same energy as five hours of perfect noon sun — regardless of whether the actual daylight lasted 9 hours or 14. Size everything off PSH, never off daylight hours. The confusion is understandable; long summer days feel productive. But a hazy 14-hour July day in the Pacific Northwest delivers fewer peak sun hours than a crisp 10-hour October day in Albuquerque.
2.1 Peak Sun Hours by U.S. Region
Use annual average PSH for energy-offset sizing, but check the winter column before you promise anything to an off-grid customer. Off-grid systems must be sized to the worst month, not the average — a Phoenix cabin needs roughly 30% more array for December than the annual average suggests, and a Seattle cabin needs close to triple.
| Region | Representative City | Daily PSH (Annual Avg.) | Summer PSH | Winter PSH |
|---|---|---|---|---|
| Pacific Southwest (Desert) | Phoenix, AZ | 6.5–7.0 | 7.5–8.0 | 5.5–6.0 |
| Southwest | Albuquerque, NM | 6.0–6.5 | 7.0–7.5 | 5.0–5.5 |
| Pacific Coast | Los Angeles, CA | 5.0–5.5 | 6.5–7.0 | 4.0–4.5 |
| Mountain West | Denver, CO | 5.0–5.5 | 6.0–6.5 | 4.0–4.5 |
| Texas / Southern Plains | Austin, TX | 4.5–5.0 | 6.0–6.5 | 3.5–4.0 |
| Southeast | Atlanta, GA | 4.0–4.5 | 5.0–5.5 | 3.0–3.5 |
| Mid-Atlantic | Philadelphia, PA | 3.5–4.0 | 4.5–5.0 | 2.5–3.0 |
| Midwest | Chicago, IL | 3.5–4.0 | 4.5–5.0 | 2.0–2.5 |
| Northeast | Boston, MA | 3.5–4.0 | 4.5–5.0 | 2.5–3.0 |
| Pacific Northwest | Seattle, WA | 3.0–3.5 | 5.0–5.5 | 1.0–1.5 |
2.2 Using PVWatts for Production Estimates
NREL's PVWatts calculator is the industry baseline for a reason — it is free, it uses real TMY weather files, and lenders accept its output. Enter the exact tilt and azimuth, not the defaults. A roof at 18° tilt and 225° azimuth (southwest) in Denver produces about 4% less than the same roof at 30° and 180°. On a 10 kW system that is 400+ kWh a year, every year, for 25 years. Run PVWatts with a 14% system loss default first, then refine with Section 3. Set the array type correctly — fixed open rack versus fixed roof mount changes cell temperature assumptions by several degrees, and premium module temperature coefficients earn their money exactly there. For a quick cross-check on battery-backed systems, our solar system calculator and battery sizing calculator use the same PSH logic.
One more PVWatts habit worth building: export the hourly results file and look at the monthly breakdown, not just the annual total. If December production is 45% of June production and the customer's loads are flat, your "100% offset" system is really a summer-overproducer and a winter-shortfall machine. Under net metering with annual true-up, that is fine. Under monthly netting or net billing, it changes the economics enough to matter.
3. System Loss Factors
Between the photon hitting the glass and the electron crossing the revenue meter, you will lose 10–20% of the nameplate energy. Pretending otherwise is how systems underperform their proposals. Here is the loss stack we apply, itemized:
| Loss Category | Typical Range | Field Notes |
|---|---|---|
| Soiling | 2–5% | Dust, pollen, and debris; worst in arid and agricultural zones. Budget a cleaning visit in dusty markets. |
| Shading | 0–20%+ | Trees, parapets, chimneys. Measure on site with a shade tool — never estimate from satellite imagery alone. |
| Snow | 0–5% | Tilt angle and climate dependent; steeper tilts shed faster. |
| Module mismatch | 1–3% | String-to-string variation; module-level electronics recover most of it. |
| DC/AC wiring | 1–3% | Ohmic loss; proper conductor sizing keeps it at the low end. See our NEC compliance guide for ampacity basics. |
| Connections | 0.5–1% | MC4s, lugs, busbars. Torque to spec — a loose lug is a hot lug. |
| Inverter efficiency | 2–4% | Modern string inverters run 96–98% peak; check CEC weighted efficiency, not peak. |
| Temperature derating | 2–5% | ~0.3–0.5%/°C above 25°C cell temperature. Dark roofs in Texas run hot. |
| Availability | 0.5–2% | Maintenance downtime, faults, grid outages. |
| Total system loss | 10–20% | 14% is the PVWatts default and a defensible bid number for clean, unshaded sites. |
The performance ratio (PR) is simply 1 minus total losses. A 15% loss stack means PR = 0.85, and that is the divisor in Step 4 of the worksheet below. A concrete temperature example makes the derating real: on a 95°F afternoon, module cell temperature on a dark shingle roof runs 60–65°C. At a −0.35%/°C power coefficient, that is 12–14% below nameplate before any other loss — which is why summer output per sun hour always disappoints newcomers who size from STC ratings.
4. DC/AC Ratio: Optimizing Inverter Sizing
4.1 Why Oversize the Array?
Modules are cheap; inverters are not. That cost asymmetry is why virtually every well-designed system carries more DC nameplate than AC rating. The array only hits its rated output under lab conditions (STC) — in the field, heat, soiling, and off-noon sun keep it below nameplate most of the year. Oversizing the DC side 15–35% keeps the inverter running closer to its efficient operating window and harvests more energy in shoulder hours. The tradeoff is clipping: on the best cool, clear spring days, the inverter caps output at its AC rating and the excess DC is discarded. Done right, annual clipping losses stay under 2% while annual production rises far more than that.
The clipping math is worth running once by hand to build intuition. Take a 10 kW DC array on a 7.6 kW inverter (ratio 1.32). Across a full year, the array exceeds 7.6 kW of actual output — after temperature and soiling — perhaps 60–90 hours in a temperate climate. The energy clipped in those hours is roughly 150–250 kWh out of ~13,000 kWh of gross potential: under 2%. Meanwhile the fatter array picks up 6–8% more energy in mornings, evenings, and cloudy hours versus a 1.0 ratio design. That asymmetry is the entire argument.
4.2 Recommended DC/AC Ratios
| Application | Recommended DC/AC Ratio | Rationale |
|---|---|---|
| Residential (temperate climate) | 1.15–1.25 | Balanced harvest and clipping; higher inverter $/W argues against extreme ratios. |
| Residential (hot climate) | 1.20–1.30 | Temperature derating justifies more DC headroom. |
| Commercial (fixed tilt) | 1.25–1.35 | LCOE optimization; clipping stays under 2% annually. |
| Commercial (single-axis tracking) | 1.15–1.25 | Trackers already flatten the production curve. |
| Utility-scale | 1.25–1.40 | Aggressive oversizing wins in utility finance models; NREL ATB reference sits near 1.34. |
One caution from the field: check the inverter datasheet's maximum DC input and MPPT voltage windows before you commit to a ratio. We have seen designs at 1.35 that looked great in the model and then needed a last-minute module swap because the string Voc at record-low temperature exceeded the MPPT ceiling. The solar inverter collection lists the spec sheets worth reading before you finalize strings.
5. Battery Storage Sizing and Integration
5.1 Battery Sizing Methodology
Battery sizing is a separate calculation from array sizing, and confusing the two is the most common design error we see in permit packages. The array is sized to annual energy (kWh/year). The battery is sized to the critical loads it must carry (kWh/night or kWh/outage) and to the inverter's power rating (kW). Three questions settle it: What must stay on? For how long? And can the inverter start the largest motor in that list? A well pump that draws 1,000 W running can demand 3,500 W for half a second at start; if the inverter's surge rating cannot cover it, the battery's kWh rating is irrelevant. For chemistry and lifecycle tradeoffs, our off-grid battery sizing guide goes deeper, and the battery and energy storage collection shows the current LFP options we stock.
5.2 Worked Example: Residential Backup Battery
Take the critical-loads panel from a real Portland-area job: refrigerator (1.4 kWh/day), furnace blower (0.9 kWh/day), lighting and receptacles (2.2 kWh/day), well pump (1.1 kWh/day), network and misc (0.4 kWh/day). That is 6.0 kWh per day of critical load. For one night of autonomy with lithium iron phosphate at 90% usable depth of discharge and a 95% inverter efficiency:
Required battery = 6.0 kWh ÷ 0.90 ÷ 0.95 = 7.0 kWh nominal. Two days of autonomy doubles it to 14.0 kWh. That is why a single ~10 kWh wall-mount battery covers one night comfortably, and why customers who want multi-day outage protection end up stacking two or three units. A unit like the Fortress eFlex Max 5.4 kWh scales exactly this way — modular 5.4 kWh blocks on a 51.2 V bus. If the site has a generator already, hybrid designs that use the battery for overnight quiet hours and the generator for extended outages routinely cut the required battery bank in half; our generator collection covers the genset side of that pairing.
5.3 Commercial Battery Sizing
Commercial storage is usually sized against demand charges, not outages. The math: battery kW must meet or exceed the kW you intend to shave off the monthly peak, and battery kWh must sustain that shave for the duration of the peak window (typically 2–4 hours). A 100 kW shave for 2 hours needs a 100 kW / 200 kWh block — the class covered by our C&I battery collection. Round-trip efficiency (~88–92% for LFP) belongs in the ROI model, not the sizing model. And confirm the tariff: some utilities ratchet demand off the highest 15 minutes in the past 12 months, which means one missed discharge event can erase a year of savings.
6. Step-by-Step Sizing Worksheet with Example
This is the exact eight-step sequence our design desk runs. Work through it in order — each step feeds the next. Skipping Step 7 is the one that bites: a string that looks fine at STC can over-volt the inverter on the coldest morning of the year, and NEC 690.7 makes that calculation mandatory, not optional.
- Determine daily energy requirement. Annual kWh ÷ 365. Example: 10,950 kWh/year ÷ 365 = 30 kWh/day.
- Determine peak sun hours. Denver annual average = 5.0 PSH/day.
- Calculate required array size before losses. 30 kWh ÷ 5.0 h = 6.0 kW DC.
- Apply the system loss factor. 6.0 kW ÷ 0.85 (PR) = 7.06 kW DC.
- Apply the DC/AC ratio. Target 1.25 → inverter AC rating = 7.06 ÷ 1.25 = 5.6 kW; a 6.0 kW unit gives a final ratio of 1.20 after module rounding.
- Determine module count. 7.06 kW ÷ 0.400 kW = 17.6 → 18 modules × 400 W = 7.2 kW DC.
- Verify string configuration. 2 strings × 9 modules. Check cold-temperature Voc against the inverter's 600 V ceiling: 9 × ~49 V (Voc at −10°C for a typical 400 W mono) ≈ 441 V — comfortably inside the window.
- Verify with PVWatts. Model the final array at the actual tilt/azimuth; expect ~11,500 kWh/year, slightly over the 10,950 target — exactly the margin you want for degradation and soiling.
| Parameter | Value |
|---|---|
| Annual consumption | 10,950 kWh |
| Daily average load | 30 kWh/day |
| Peak sun hours (Denver) | 5.0 h/day |
| System loss factor | 15% (PR = 0.85) |
| Calculated DC array size | 7.06 kW |
| DC/AC ratio | 1.25 (design) / 1.20 (as built) |
| Inverter AC rating | 6.0 kW |
| Module selection | 400 W mono, 18 modules |
| Final DC array size | 7.2 kW |
| String configuration | 2 strings × 9 modules |
| Estimated annual production | ~11,500 kWh |
7. Commercial vs. Residential Considerations
The physics is identical; everything else changes. Voltage classes, interconnection paths, NEC article scope, and financing all fork at the residential/commercial boundary.
| Factor | Residential | Commercial |
|---|---|---|
| Typical system size | 5–15 kW | 50–1,000+ kW |
| DC voltage | 300–600 V | 600–1,500 V |
| AC voltage | 120/240 V single-phase | 208/480 V three-phase |
| Inverter type | String or microinverter | Large string or central |
| Interconnection | Net metering (state-dependent) | Net metering, feed-in tariff, or PPA |
| Mounting | Pitched roof, penetrating or rail-less | Ballasted flat roof, ground mount, carport |
| NEC scope | Article 690 (< 5,000 kW) | Article 690, or 691 at ≥ 5,000 kW |
| Financing | Cash, loan, lease | PPA, lease, direct purchase with depreciation |
For racking on either side of that divide, our racking and mounting collection covers residential rails through commercial ground mounts, and the roof vs. ground vs. pole mount comparison walks the structural tradeoffs. Three-phase service also changes inverter selection and revenue metering — if 208 V versus 480 V three-phase is fuzzy territory, our kWh-to-amps conversion guide has the three-phase formulas with worked examples.
8. ROI Calculations and Economic Analysis
8.1 Simple Payback Period
Payback = net installed cost ÷ annual savings. Using the Denver example: 7.2 kW at $2.80/W = $20,160 gross; after the 30% federal ITC, $14,112 net. At 11,500 kWh/year and $0.16/kWh retail, annual savings run about $1,840. Payback ≈ 7.7 years. Every assumption in that line deserves scrutiny — especially the retail rate, which should be the customer's marginal rate, not the utility's advertised average. Tiered and time-of-use rates move the real number by 20% or more.
8.2 Return on Investment (ROI)
Over 25 years with 0.5%/year degradation and a 3%/year utility escalation, that same system saves roughly $58,000 against $14,112 net cost — a lifetime ROI above 300%. We model these in the solar ROI calculator, which lets you stress-test escalation and degradation assumptions. Escalation is the silent giant in every solar model: dropping it from 3% to 1.5% cuts 25-year savings by nearly $12,000 on this example, so agree on the assumption with the customer in writing.
8.3 Commercial ROI Considerations
Commercial models add demand-charge reduction, accelerated depreciation (MACRS), and sometimes PPA structures. A 500 kW rooftop with a $128,000 combined annual energy-plus-demand savings line can clear a 13% unlevered IRR — the full sensitivity tables are in our commercial sizing deep dive. Commercial buyers also think in hedge terms: a fixed LCOE below $0.06/kWh against a utility rate with 25 years of unknown escalations is a risk-management decision as much as a financial one.
8.4 LCOE (Levelized Cost of Energy)
LCOE = lifetime cost ÷ lifetime kWh. For the Denver system: roughly ($14,112 + $3,000 lifetime O&M) ÷ ~265,000 lifetime kWh ≈ $0.065/kWh. When your LCOE lands under half the retail rate, the project sells itself. LCOE is also the honest way to compare bids — a cheap system with a 22% loss stack can carry a worse LCOE than a premium one at 13%. If two proposals differ, normalize both to LCOE before comparing anything else.
9. Common Sizing Mistakes to Avoid
- Sizing from square footage. Roof area tells you what fits, not what is needed. Load first, always.
- Using daylight hours instead of PSH. A 14-hour summer day in Seattle still only delivers ~5 PSH.
- Ignoring the winter month on off-grid jobs. Annual-average sizing fails in December. Size to the worst month or add a generator.
- Forgetting degradation. Budget 0.4–0.55%/year; year-20 production is what the customer remembers.
- Undersizing the inverter AC for future DC expansion. If the customer may add panels, leave MPPT and breaker headroom now.
- Skipping the shade study. One vent pipe shadow across a string can cost 15% of that string's annual yield without module-level electronics.
- Net-zero promises on a gross-zero roof. If the roof only fits 80% offset, say so in the proposal. Under-promise, over-deliver.
- Ignoring the 120% rule. NEC 705.12 limits backfed breakers against the panel busbar rating; a 200A panel with a 200A main tops out at 40A of solar without a main breaker derate or line-side tap. Check the busbar before you promise the array size.
10. Tools and Resources
The working stack: PVWatts for feasibility, a shade tool on site, interval data for anything commercial, and the system sizing calculator for quick client-facing numbers. When you are ready to price hardware, start at the solar panel collection and the residential starter kits, or get a firm number through our quote desk. For charge-controller-based off-grid designs, the charge controller sizing guide picks up where this article stops.
Field Notes from Real Installs
We have pulled thousands of feet of 4/0 through EMT on commercial services, and the lesson that repeats is simple: the load calc is the job. Everything else is hardware. On a 12 kW residential job in Beaverton last spring, the homeowner's bills showed 11,800 kWh/year — but two months of that was a hot tub they had already sold. We sized to 9,600 kWh instead and saved them six panels of roof space for a future addition. I have also watched a competitor's crew discover at commissioning that nobody checked the winter PSH on an off-grid cabin; the system ran a generator 40% of December. And on a 300 kW warehouse, our interval-data catch — a 4 a.m. refrigeration ramp the owner had forgotten — moved the design from straight grid-tie to grid-tie plus 200 kWh of storage before a single module was ordered. Measure twice, order once. That is the whole trade. One more that cost somebody real money: a 20 kW array promised against a 200A panel whose busbar turned out to be 200A-rated with a 200A main. The 120% rule capped backfeed at 40A, the design needed 60A, and the fix was a main-breaker derate plus a new line-side tap detail that delayed inspection by six weeks.
Frequently Asked Questions
How many solar panels do I need for a 2,000 sq ft house?
Square footage does not size a system — energy use does. A 2,000 sq ft home using 900 kWh/month (10,800 kWh/year) in a 5-PSH climate needs about 7.1 kW after losses: 10,800 ÷ 365 ÷ 5.0 ÷ 0.85 ≈ 6.96 kW, or 18 × 400 W panels. The same house in a 3.5-PSH climate needs about 26 panels. Start with the utility bill, not the floor plan.
What is a good DC/AC ratio for a home system?
1.15–1.25 in temperate climates, up to 1.30 in hot ones. That keeps annual clipping losses under about 1.5% while harvesting meaningfully more energy in morning and evening hours. Confirm the inverter's maximum DC input before finalizing the ratio.
How do I size a battery for backup power?
Total your critical loads in kWh per day, multiply by the days of autonomy you want, then divide by usable depth of discharge (0.90 for LFP) and inverter efficiency (~0.95). Six kWh/day of critical load with one day of autonomy needs a ~7 kWh nominal battery; two days needs ~14 kWh.
Does system size affect my payback period?
Yes, but not linearly. Larger systems enjoy lower $/W installed cost, which improves payback — until production exceeds your usage and export compensation drops below retail. Size to your load under your utility's net metering rules; oversizing past ~110% of consumption rarely pays.
What system losses should I use in a proposal?
Use 14% for a clean, unshaded, well-ventilated site — the PVWatts default. Add points for shading, heavy soiling environments, or steep temperature derating. A defensible loss number beats an optimistic one: the customer judges you against the proposal at year one.
Can I expand my system later?
Usually, yes — if the inverter has spare MPPT capacity, the service panel has breaker and busbar headroom (check the 120% rule under NEC 705.12), and the interconnection agreement allows it. Planning expansion at design time costs almost nothing; retrofitting for it costs real money.
How does shading affect the array size I need?
Shade does not change your load — it raises the array required to meet it. A site with a 10% shade loss needs roughly 11% more DC capacity than an open site (divide by 0.90 instead of leaving it out of the loss stack). Module-level electronics recover mismatch but cannot manufacture photons; serious shade sometimes means the better engineering answer is fewer panels on a better face.
Related Products & Collections
- Grid-tied solar systems — complete packaged kits sized to your load
- DIY solar kits — starter bundles with design support
- Fortress eFlex Max 5.4 kWh battery — modular LFP backup block
- Request a quote — firm pricing for contractors and homeowners
