When you get a solar proposal, there is a number near the top of the page that has a hypnotic effect on homeowners. It might say "estimated annual production: 14,800 kWh" or "projected savings: S$4,200 per year." That number is a simulation — the solar equivalent of a car's fuel economy test on a dynamometer in a climate-controlled laboratory. And it will almost certainly not match what your system actually produces in year one.
This is not fraud. It is not deceptive, as long as the caveats are explained clearly. But in the enthusiasm of a sales conversation, those caveats have a habit of shrinking. So this article is the caveats. All of them.
What Is Standard Test Condition Energy — and Why Doesn't It Exist in Singapore?
Every solar panel is rated at Standard Test Conditions (STC): 1,000 W/m2 irradiance, a cell temperature of 25C, and an air mass of 1.5. A 400W panel produces 400W under STC. Here is the problem: STC conditions almost never occur on a real rooftop. In Singapore, where rooftop cell temperatures regularly reach 45–60C, the 25C cell temperature assumption is invalid every single day. Real-world production is routinely 15–25% below STC-rated output before any other factor is considered.
The Simulation Tools: What They Are and Where They Fall Short
Helioscope (Folsom Labs, US) — Industry-standard residential and commercial design tool. Excellent 3D shading modelling and fast UI. Relies on NREL TMY weather files with limited tropical granularity.
PVsyst (PVsyst SA, Switzerland) — The engineering gold standard. Advanced loss modelling, highly configurable, requires calibration with local meteo data. Used for bankable energy reports.
SAM — System Advisor Model (NREL / US Dept. of Energy) — Free, rigorous, with extensive databases. Designed for analysts and researchers. Excellent for sensitivity analysis.
Aurora Solar (Aurora, US) — AI-powered shading from satellite imagery; optimised for fast proposal generation. Strong on sales speed, less deep on engineering detail than PVsyst.
Archelios Pro (Trace Software, France) — Strong European and Asian coverage; less common in Southeast Asia but growing.
PV*SOL (Valentin Software, Germany) — Good battery integration modelling; European-skewed weather data; less Southeast Asia coverage.
Every one of these tools is only as good as the weather data and assumptions fed into it. The backbone of any simulation is a Typical Meteorological Year (TMY) dataset — a synthetic year built from decades of historical climate records. For Singapore, TMY data comes from NASA POWER, Meteonorm, SolarGIS, or NSRDB. These are reputable sources, but they represent historical averages. They do not predict that 2026 will be a La Nina year with 30% more cloud cover, or that Indonesian forest fire smoke will reduce irradiance for three weeks in October.
The Seven Factors That Reduce Your Real-World Output
1. Shading — The Silent Performance Killer
In a traditional string inverter, panels in a string behave like batteries in series — when one panel is shaded and producing 40% of rated output, the entire string is limited to that panel's output. A shadow covering 10% of your array can reduce total string output by 40–60%. Shading in Singapore is also seasonal and time-of-day dependent. As the sun's arc shifts through the year — Singapore sits at 1.3N — shadows from neighbouring buildings, water tanks, and antenna masts fall at different angles depending on the month and time of day. SunMax module-level optimisers (deployed on every Sunollo system) address this directly. Each panel operates independently rather than being limited by the weakest panel in the string. In shading scenarios, SunMax systems routinely recover 8–15% of output that a conventional string would lose.
2. Temperature — The Counterintuitive One
Solar panels produce less electricity in hot weather, not more. The photoelectric conversion process is less efficient at higher cell temperatures. This is quantified in a panel's temperature coefficient, typically -0.30% to -0.45% per degree C above 25C. In Singapore, where rooftop cell temperatures regularly reach 50–65C: at 25C (STC) output is 400W; at 45C it drops to 368W (-8%); at 55C to 352W (-12%); at 65C to 336W (-16%). A system sized for 12,000 kWh/year, without proper temperature derating for Singapore's tropical conditions, may consistently produce 10–13% less from temperature loss alone. Eclipse monocrystalline panels used in Sunollo systems have lower temperature coefficients, meaning they lose less output as temperature rises.
3. Soiling — Dust, Pollen, and Grime Accumulation
In Singapore, soiling includes: red laterite dust, pollen, bird droppings, diesel particulate matter near major roads, salt aerosol near the coast, and during haze events, fine particulate matter from Indonesian agricultural burning. Typical soiling loss without cleaning: 2–5% per year under normal conditions, rising to 8–15% during severe haze periods. Simulation tools apply a flat soiling loss factor (often 2–3%) as a single annual deduction. They do not model haze events, localised dust sources, or bird droppings concentrated under a nearby tree.
4. Seasonal Irradiance Variation — Singapore Is Not Flat Year-Round
Singapore has two monsoon seasons that create meaningful irradiance variation. The Northeast Monsoon (December–March) brings more cloud cover, higher rainfall, and production typically 10–20% below the annual average. The Southwest Monsoon (June–September) is drier and sunnier but also the period of highest haze risk from Sumatran fires. Inter-monsoon periods (April–May, October–November) see convective thunderstorm activity in afternoons that reduces daily peak production hours. Year-to-year variation driven by ENSO (El Nino/La Nina) cycles can be plus or minus 8–12% from the modelled typical year.
5. Inverter Efficiency and Clipping
Modern string inverters are rated at 97–98.5% peak efficiency, but at low irradiance (early morning, late afternoon, overcast days) efficiency drops to 92–95%. Clipping occurs when the array produces more power than the inverter is rated to handle — a deliberate, economically rational design trade-off that simulation tools model, but that varies between installers depending on exact sizing choices.
6. System Losses — Wiring, Connectors, and Diodes
DC wiring losses (1–2%), AC wiring losses (0.5–1%), connector and junction box losses (0.5–1%), and bypass diode losses under partial shading add up to 4–8% even in a well-designed, correctly installed system. The actual values depend on cable run lengths, connector quality, and correct cable sizing.
7. Panel Degradation Over Time
Most tier-1 panels guarantee a minimum of 80% rated power after 25 years — implying roughly 0.5–0.7% degradation per year. In Singapore's tropical climate, LID (Light-Induced Degradation) in year 1 typically causes an additional 1–2% loss before stabilising. Over 25 years, your system in year 25 produces roughly 15–20% less than in year 1.
P50 and P90: The Numbers Your Installer Should Be Quoting
When a solar simulation produces a figure like 14,800 kWh/year, that is almost always a P50 estimate — the level of production that will be met or exceeded in 50% of years. Half the time you will do better; half the time you will do worse. P90 is the risk-adjusted figure: the production level that will be met or exceeded in 90% of years. P90 is typically 10–15% lower than P50. For Singapore, a P90 figure might be 12% below P50 — so if P50 is 14,800 kWh, P90 is approximately 13,000 kWh. P99 (met or exceeded 99% of years) is used for project finance on large commercial systems and might be approximately 11,500 kWh in this example.
The practical takeaway: When your installer gives you an annual energy figure, ask: Is this P50 or P90? If they do not know the answer — or if the concept is unfamiliar to them — that is a meaningful data point about the rigour of their engineering process. At Sunollo, every proposal includes both P50 and P90 figures.
The Car Analogy: Why the Lab and the Road Are Different Places
The fastest way to explain all of the above is this: your solar system's rated output is exactly like a car's official fuel economy or top speed — technically accurate under controlled conditions, and reliably different from what you experience daily.
When Volkswagen says the Golf GTI has a combined fuel consumption of 6.8L/100km, that figure comes from a standardised EU WLTP test cycle conducted indoors, at a specific speed profile, with no air conditioning, no roof rack, and a warm engine. In Singapore — with stop-start traffic on the PIE at 6pm, air conditioning running continuously from April to November — the real-world consumption is closer to 9.0–11.0L/100km. The difference is not that Volkswagen lied. The difference is that the test protocol cannot replicate the real world.
Solar simulation tools are the WLTP of the rooftop energy business. Helioscope, PVsyst, and Aurora are excellent at calculating what happens when you apply accurate solar physics to a well-characterised roof model using historical climate data. What they cannot do is: know that your neighbour will build a new storey that shades your southwest array from October 2027; predict that a 2028 El Nino event will reduce Singapore's irradiance by 11% for eight months; account for dust accumulating from a construction site three doors down; or model the branch that falls on panel row 3 and creates a partial bypass diode fault. The simulation and the reality are always in conversation. The simulation sets the expectation; the real world delivers the result.
What This Means for Your Monthly Monitoring
The metric that matters is Performance Ratio (PR) — the ratio of your system's actual output to its theoretical maximum given the irradiance it received. A well-performing Singapore rooftop system should have a PR of 0.74–0.82 (74–82%). Below 70% sustained over several weeks warrants investigation. Sunollo's LiveTrack app shows this in real time and flags anomalies — individual panel underperformance, inverter alerts, and production trends against the modelled baseline. A PR drop of -2 to -5% typically means soiling: schedule a clean. A drop of -5 to -10% means soiling plus seasonal effects: clean panels and check inverter logs. A drop of -10 to -15% may indicate a new shading obstruction. A drop of greater than -15% suggests a fault: contact Sunollo support, covered under SunolloCare.
Panel Cleaning: Not Optional, More Often Than You Think
In Singapore, the recommended minimum is once per year, timed after the Northeast Monsoon ends in March/April. However, more frequent cleaning is warranted in certain conditions: near major roads or expressways (every 6 months); post-haze event where PSI exceeds 100 (within 2–4 weeks of haze clearing); under trees with significant pollen or leaves (every 6 months); near Changi Airport or industrial areas (every 4–6 months); standard residential with minimal local pollution (annually in March/April); near bird nesting (after each nesting season).
How to know when panels need cleaning without going on the roof: LiveTrack trend analysis — a gradual, consistent decline in PR over 4–6 weeks without obvious weather change is the clearest signal. Compare rainy-day production to sunny-day production — rain partially cleans panels, so if PR recovers after heavy rain and then gradually declines again, soiling is the culprit. After any PSI over 100 event, expect measurable soiling losses within 48–72 hours.
A note on DIY cleaning: tap water with a soft brush or sponge and a long-handled squeegee works well. Avoid abrasive materials or high-pressure jets — they can scratch the anti-reflective coating or damage frame seals. For roofs requiring ladder access or steep pitches, use a professional.
In Singapore field studies, panels cleaned after a moderate soiling period (6–8 months without cleaning) show a 5–9% production recovery immediately after cleaning. Post-haze cleaning can recover 10–15%. Over a year, consistent cleaning versus no cleaning is the difference between a system performing at 95% of its potential versus 87–90% — a gap worth S$280–S$420/year on a typical 10 kWp system, for a cleaning service that costs S$150–S$300. The economics are clear.
Other Maintenance Factors Worth Knowing
Inverter degradation: Modern inverters have operating lifespans of 10–15 years. Most systems will need at least one inverter replacement over a 25-year system life — a planned cost, not a surprise.
Connector and wiring inspection: MC4 connectors can develop micro-corrosion in humid tropical environments. Annual thermal imaging inspection (offered under Sunollo's SunolloCare premium tier) identifies hot spots from failing connections before they become faults or fire risks.
Panel microcracks: Physical damage from falling debris can create microcracks invisible to the naked eye but reducing individual panel output by 5–15%. Module-level monitoring via SunMax optimisers is the earliest detection method.
Bird nesting and pest management: Birds nesting under panels — a common issue in Singapore with myna birds and pigeons — can cause soiling, wiring damage, and overheating. Anti-bird mesh around the panel perimeter is a low-cost preventive measure.
So What Is Your System Actually Going to Produce?
A well-designed, well-installed, regularly maintained solar system in Singapore will typically produce 85–92% of its simulated P50 annual output in a typical year, and may fall to 78–85% in a year with adverse weather (La Nina, heavy haze) or if cleaning is deferred. This is not a reason to be pessimistic about solar — the economics are still very strong even at 85% of simulation. It is a reason to: ask for P90 figures when evaluating a proposal; plan on 85–90% of the simulated number as your baseline; clean your panels at least annually; monitor in real time via LiveTrack; and choose SunMax module-level optimisation for any system with any shading.
The car goes faster in the brochure than in traffic. The solar system produces more on the datasheet than on a hazy Tuesday in December. Both are still excellent products — the key is knowing the gap between the brochure and the reality, and taking the steps — cleaning, monitoring, optimisers — that close it.
Want to know your system's P50 and P90 figures, and how it compares to actual monthly production? Sunollo's LiveTrack shows real-time performance ratio against your modelled baseline — and SunolloCare includes annual panel cleaning, thermal inspection, and performance assurance. Get a free solar assessment.
