A noise wall does most of its work at one specific height. FHWA’s abatement guidance puts the threshold plainly: a barrier can achieve a 5 dB noise level reduction when it is tall enough to break the line of sight from the highway to the receiver. Every foot below that height buys very little, and every foot above it buys about half a decibel, since the guidance credits roughly 1 dB of additional reduction for each 2 feet of height once line of sight is broken, up to a maximum theoretical total of 20 dB(A).
The federal regulation governing this work sets its own bar at the same number. Under 23 CFR 772.13, a noise abatement measure is feasible only if it achieves at least a 5 dB(A) reduction at impacted receptors. The acoustics and the rulebook meet at the point where a wall first blocks sight of the traffic, which is a rare piece of tidiness in highway regulation and the reason barrier heights cluster where they do.
The decibel scale is why seven is the working target
Decibels are logarithmic, and the practical translation FHWA uses is that effective noise barriers can reduce noise levels by 10 decibels, cutting the loudness of traffic noise in half. That single sentence explains the shape of the whole regulation. A 5 dB reduction is audible and useful. A 10 dB reduction is the point at which a resident describes the road as half as loud. There is no realistic wall that delivers 30.
So the regulation brackets the ambition. Each highway agency must define, with FHWA approval, a noise reduction design goal of at least 7 dB(A) but not more than 10 dB(A), and must state how many benefited receptors have to achieve it. A benefited receptor is itself defined by those two numbers: one that receives a reduction at or above the minimum threshold of 5 dB(A), but not exceeding the agency’s design goal.
Against that frame, the national result is modest and unsurprising. Barriers built between 2020 and 2022 achieved an average noise reduction of 7 dB(A), which is the bottom of the permitted design goal range.
The criteria are impact thresholds, and the table says so in a footnote
Table 1 of Part 772 lists the noise abatement criteria by activity category, in hourly A-weighted sound levels. Category A, for lands where serenity and quiet are of extraordinary significance, sits at 57 dB(A) Leq(h). Residential land, Category B, sits at 67. Interior criteria for Category D facilities sit at 52. Hotels, motels, offices and restaurants, in Category E, sit at 72. Category F, which covers agriculture, industrial land, rail yards, retail and similar uses, has no impact criterion at all and requires no noise analysis.
The fine print around that table settles more arguments than its numbers do. A project may use either Leq(h) or L10(h) but not both. The criteria values, in the table’s own footnote, are “for impact determination only, and are not design standards for noise abatement measures,” which is the sentence to reach for whenever a wall is described as bringing a neighborhood down to 67. And a predicted level that merely approaches the criterion counts as an impact, with each agency required to establish an approach level of at least 1 dB(A) for that purpose.
The second route to an impact has nothing to do with the table. A substantial increase over existing levels also qualifies, expressed in the regulation as design year build levels running 5 to 15 dB(A) above the existing level. A quiet road that gets noticeably louder can trigger abatement while a loud road that stays equally loud does not.
Reasonableness is where most walls are decided
Feasibility is acoustics and engineering. Reasonableness is policy, and 23 CFR 772.13 makes three factors mandatory rather than optional, requiring that they be achieved collectively so that failing any one of them makes the measure unreasonable.
The first is the viewpoints of the property owners and residents of the benefited receptors. The agency must solicit all of them and obtain enough responses to document a decision either way, and must define, with FHWA approval, how many responses constitute a decision. The second is cost effectiveness, where each agency sets a baseline cost reasonableness value approved by FHWA, which may be built from construction cost, cost per square foot, maximum square footage per benefited receptor, cost per benefited receptor, or cost per benefited receptor per dB(A) of reduction, and which has to be re-analyzed at least every five years. The third is the design goal described above.
Read that structure back and the practical test for a wall is a cost per household per decibel, compared against a threshold each state wrote for itself. Two identical neighborhoods on either side of a state line can receive different answers, lawfully, and the environmental documents described in how highway projects are planned are where that determination is recorded.
What the inventory records
FHWA maintains a barrier inventory covering construction from 1963 to 2022, built from data supplied by 52 state agencies including all 50 states, the District of Columbia and Puerto Rico. Through 2022 it accounts for 300,101,207 square feet of barrier totaling 3,866 linear miles, at approximately $10.8 billion and an average unit cost of $41.93 per square foot in 2022 dollars. Average barrier height nationally is 14 feet, ranging by state from 7 feet to 18 feet. Concrete makes up 59 percent of single-material barriers, block 16 percent and wood 5 percent.
The most recent reporting period is more interesting than the cumulative total. Between 2020 and 2022 states built 22,534,945 square feet across 222 linear miles, for approximately $925 million. The inventory’s own summary describes that as 18 fewer linear miles than the previous period, though the period figures it publishes for 2017 to 2019 (21,166,687 square feet across 257 linear miles) put the actual difference at 35 miles, not 18. Unit cost fell from $53.49 per square foot in 2017 to 2019 to $48.76 in 2020 to 2022. Dividing those published pairs shows what moved: square footage per linear mile rose from roughly 82,000 (21,166,687 divided by 257) to roughly 102,000 (22,534,945 divided by 222), so agencies bought taller walls over shorter distances and paid less per square foot for them. Absorptive materials were used on at least 40 percent of the barriers built in the period, and at least 95.8 percent protected Category B outdoor residential land.
The distribution is severe. Of the 52 reporting agencies, 49 have built at least one barrier and three have built none; 37 have built at least 10 miles and 16 have built at least 100 miles; and just seven account for over half of total barrier length, at 51.3 percent.
What a wall cannot reach
A barrier works by geometry, which puts anything above it or beside it outside its reach. FHWA’s guidance states that noise barriers “do very little good for homes on a hillside overlooking a road or for buildings, which rise above the barrier,” and that openings for driveway connections or intersecting streets reduce the effectiveness of barriers. Both limits follow from the same line of sight that produced the first 5 dB.
Vegetation deserves its own correction, because it is the abatement measure the public most often asks for. FHWA’s own figure is that a 200-foot width of dense vegetation can reduce noise by 10 decibels, and its own conclusion is that it is usually impossible to plant enough vegetation along a road to achieve such reductions, and that because substantial reduction does not occur until the vegetation matures, FHWA does not consider planting to be a highway traffic noise abatement measure. Trees screen the road from view, which changes how people report the noise, and that is a different effect from attenuation. Reported benefit and measured benefit come apart repeatedly across more on roads and the environment, and planting is the plainest case.
Federal participation follows the same logic. Funds are available for abatement where impacts have been identified and measures determined feasible and reasonable, and the eligible list includes insulating the interiors of Category D facilities, though maintenance and operating costs after installation are not eligible. Buying property as a buffer zone is permitted on Type I projects only. Construction noise sits under a separate section entirely, requiring identification of affected land uses and abatement written into the plans and specifications, which is the mechanism behind the restrictions discussed in night highway construction. Barriers meant to stop a vehicle rather than a sound wave answer to a wholly separate body of criteria, set out in guardrails and barriers.
The retrofit question, in one 1973 barrier
The first Type II barrier in the inventory, meaning one built to protect an existing neighborhood rather than as part of a highway project, was a berm and wood combination in Michigan that cost $308,554 in 2022 dollars. Type II construction has since reached 42,093,691 square feet and 560 linear miles at $1.7 billion, roughly 14 percent of all barrier square footage, at an average $40.62 per square foot. During 2020 to 2022 it ran at $28.81 per square foot, well under the $48.76 average for the period, which places retrofit walls at the cheaper end of the inventory.
Twenty-eight agencies have reported building at least one Type II barrier, leaving 24 of the 52 reporting agencies that never have. The reason is written into the funding rule rather than into acoustics. Under 23 CFR 772.15(b), no money from the Highway Trust Fund may be used to construct a Type II barrier that was not part of a project FHWA approved before 28 November 1995, and the remaining federal eligibility reaches only land that was developed or under substantial construction before the existing highway’s right of way was acquired. A state that had no retrofit program in 1995 now builds these walls, if it builds them, out of its own revenue, which is visible in the inventory line reporting that 16 of the 222 miles built between 2020 and 2022 were state or toll funded. That is the quiet architecture behind an apparently local decision. A cutoff date set in 1995 still decides which neighborhoods a state protects out of its own revenue, and which sit in the 24 agencies that have never built one.