HPS Street Lighting to LED: How the US Got It Wrong and What Australia Should Specify

The energy case: HPS versus LED in numbers

High pressure sodium lamps dominated street and area lighting globally for five decades because they were efficient and long-lived by the standards of their time. A 250W HPS lamp produces roughly 27,500 lumens, giving an efficacy of around 110 lm/W. That was impressive in 1975. A modern LED street light at the same lumen output runs at 160 to 180 lm/W, consuming under 160W for equivalent light delivery.

On top of the energy saving, HPS lamps require replacement every 12,000 to 24,000 hours. LED street lights are rated for 50,000 to 100,000 hours. For a council managing thousands of poles, that maintenance cost difference alone often justifies the capital expenditure before the energy saving is even counted. Ausgrid has reported that its NSW LED rollout, covering over 82% of its network, is cutting energy consumption by around 60% against the legacy mixed portfolio it replaced.

The financial case is clear. The specification case is where it gets complicated.

What the US got wrong: the blue light problem

Between 2010 and 2020, American cities replaced tens of millions of HPS streetlights with LEDs at pace. The technology was new, costs were falling fast, and councils and utilities prioritised the highest-efficacy product available. The highest-efficacy LEDs at the time ran at 5000K to 6000K colour temperature. The blue-white light they produced looked bright and modern. It also produced a public health backlash that is still playing out.

In 2016, the American Medical Association adopted a formal policy statement on LED street lighting. The AMA's core findings were that blue-rich LED light, specifically LEDs operating at 4000K and above, was causing three measurable problems: disruption of human circadian rhythms through melatonin suppression, increased glare that reduces driver and pedestrian safety at night, and harm to wildlife through disruption of migratory and behavioural patterns.

The complaints from residents in Seattle, New York and Houston were loud enough that cities began reversing decisions. Montreal had planned to retrofit 132,000 streetlights with 4000K LEDs. After sustained community pushback, the city switched its specification to 3000K before the bulk of the rollout began. Davis, California replaced its newly-installed 4000K LED streetlights with warmer alternatives just months after installation, absorbing the cost twice. Chicago installed dimmable 4000K lights and faced years of complaints before introducing adaptive controls to reduce output after midnight.

These were not fringe concerns from lighting enthusiasts. The issue was widespread enough to prompt US Department of Energy guidance, International Dark-Sky Association campaigns, and eventually a shift in how the lighting industry itself frames colour temperature selection for outdoor use.

What colour temperature actually means

Correlated colour temperature (CCT) is measured in Kelvin and describes the apparent warmth or coolness of a light source. It is not a measure of brightness, efficacy or quality. Two LEDs at identical CCT can have very different blue light content depending on their phosphor formulation. CCT is a useful shorthand but it is not a complete specification.

The orange glow of a 2000K HPS lamp looks nothing like the sun, but its spectral output is strongly red-shifted. There is very little blue. The human eye and brain evolved to associate blue-spectrum light with daytime and daylight. When a 5000K LED comes on at 10pm on a residential street, it sends a daytime signal to every person and animal within range, at exactly the time their bodies should be winding down for sleep.

Melatonin suppression from blue-rich light is not a marginal effect. Studies reviewed by the AMA show that exposure to blue-rich LED light at night suppresses melatonin production for significantly longer than the light exposure itself. A street that bathes bedroom windows in 5000K light from 11pm is affecting the sleep quality of residents hours after lights-out. For children, the elderly, and people with sleep disorders, the effect is more pronounced.

The glare problem is separate from the health problem

Blue light scatters more in the human eye than longer wavelengths. This means that a 5000K LED street light causes more discomfort glare, and more disability glare, than a 3000K LED at the same lumen output. Disability glare is the kind that makes it harder to see objects adjacent to the light source, which is directly relevant to pedestrian and cyclist visibility when a driver is approaching. The AMA flagged this as a road safety issue, not just a comfort issue.

The visual acuity benefit of "whiter" light, often used to justify higher CCT in road safety contexts, is real but smaller than commonly claimed. IPWEA's (Institute of Public Works Engineering Australasia) analysis found approximately a 20% improvement in driver hazard detection on main roads when comparing 4000K to 3000K. Whether that difference justifies the health and community cost on residential roads, where speeds are lower and hazard detection is less critical, is a design question. The Australian guidance answers it by recommending 3000K for residential roads and reserving 4000K for main roads specifically.

What Australia's own guidance says

IPWEA, the peak body for public works engineering in Australia and New Zealand, has published guidance on LED street lighting colour temperature that draws directly on the AMA findings and international experience. Their recommendation is straightforward:

• 3000K or lower for approximately 70% of all public lighting, including residential streets, local roads, parks, shared paths and car parks.
• 4000K for main roads, arterials and highway lighting where driver visual performance is the primary design criterion.
• 2700K or lower for environmentally sensitive areas near wildlife corridors, coastal zones, or dark sky reserves.
• Smart dimming controls to reduce output in the early hours of the morning, which further reduces circadian disruption and extends LED lifespan.

AS/NZS 1158, the Australian standard for road lighting, does not prescribe CCT. It specifies illuminance levels, uniformity ratios and glare limits. CCT selection sits within the discretion of the designer and specifier. That discretion matters, because the default offering from many LED street lighting suppliers is still 4000K, partly for historic reasons and partly because it is slightly cheaper to manufacture at scale. Unless you specify otherwise, you may get a specification that is fine for a highway and inappropriate for a suburban cul-de-sac.

Wildlife: the underappreciated impact

Australia's biodiversity context makes the blue light issue more significant here than in most countries the US experience is drawn from. A 5000K LED on a suburban street near a bushland corridor, a wetland or a coastal nesting area does measurable ecological damage that a 3000K LED at the same lumen output does not.

Blue-rich artificial light at night disrupts migratory cues in birds, interferes with sea turtle nesting and hatchling navigation, affects insect populations (which cascade through food chains), and disrupts the nocturnal behaviour of a wide range of native mammals. The International Dark-Sky Association and ornithological researchers both recommend sub-3000K lighting near any area with wildlife value. White LED light is estimated to be five times more effective at suppressing melatonin in animals than HPS, given the same light output.

Australian councils in coastal areas, near national parks, or in outer-suburban growth corridors where bushland is adjacent to development are making decisions right now that will affect wildlife corridors for the 25-year lifespan of the LED fittings they install. Retrofitting to a warmer CCT later is possible but costs money twice.

The specification checklist

Whether you are a council facilities manager evaluating a street lighting tender, a commercial property owner replacing car park HPS fittings, or a developer specifying outdoor lighting for a new precinct, these are the things to put in writing before you accept a quote.

• 🌡️
  Correlated Colour Temperature (CCT): 3000K maximum for residential and pedestrian areas For main roads, 4000K is acceptable. For parks, shared paths, car parks and residential streets, require 3000K or lower. Put the number in the specification, not just "warm white."
• 📊
  Blue light percentage: request spectral data, not just CCT Two 3000K LEDs can differ in blue light content depending on phosphor design. Ask for the spectral power distribution (SPD) or the percentage of output in the 380–500nm range. A lower percentage means less circadian and wildlife impact.
• 🔆
  Glare rating: require full cutoff or full-cutoff equivalent optics AS/NZS 1158 specifies maximum glare limits. Ensure the luminaire has a glare control rating appropriate to the application. Flat-glass optics with downward-directed distribution reduce sky glow and neighbour spill.
• ⏱️
  Smart dimming: specify adaptive controls for late-night dimming Dimming street lights to 50% or lower after midnight (when pedestrian and traffic volumes drop) reduces energy consumption, extends LED life, and further cuts circadian disruption. Many LED street lights support 0-10V or DALI dimming as standard.
• 📋
  Lumen output: match to AS/NZS 1158 category, not to HPS wattage A 250W HPS lamp nominally produces 27,500 lumens, but actual maintained output is lower. Design to the lux level and uniformity required by the road category, not by substituting watts. Over-lighting wastes money and worsens all the problems above.
• 🏆
  Warranty and lifespan: L70 rated life of 50,000 hours minimum L70 means the fitting still produces 70% of its initial lumen output at the rated hour. Street lights operate in harsh outdoor conditions. Require IP66 ingress protection, IK08 impact resistance, and a minimum 5-year on-site warranty.

HPS to LED: the comparison at a glance

The bottom line for Australian buyers

The energy and maintenance savings from replacing HPS with LED are real and well-documented. Ausgrid's network-wide data shows around 60% energy reduction. The financial case does not depend on colour temperature. A 3000K LED saves exactly as much energy as a 5000K LED of the same wattage. There is no cost trade-off between doing the right thing for residents and wildlife and doing the financially optimal thing. The 3000K option costs marginally more in some catalogues and marginally less in others depending on supplier and volume.

The US experience is instructive precisely because it happened at scale and in public. Hundreds of council meetings, resident petitions, media stories and in some cases expensive retrofits played out over a decade. The AMA's 2016 warning was not a fringe opinion: it was a formal policy position from the largest physician organisation in the United States, adopted after reviewing peer-reviewed evidence on melatonin suppression, glare and wildlife disruption.

Australia is still making these decisions. Ausgrid's NSW rollout is 82% complete. Other distributors and councils around the country are mid-programme or yet to start. The choices being locked in now will run for 25 years. Specifying 3000K instead of 4000K on a residential road costs nothing extra and avoids a problem that is entirely avoidable.