measuring how deep

the stone sits

Problem statement

A chip seal — a thin layer of asphalt emulsion sprayed on a road and covered with a single layer of stone chips, then rolled — is one of the cheapest and most widely used pavement-preservation treatments in the United States, and its life is decided by one geometric quantity: how deep each stone is embedded in the binder. Too shallow and the stones ravel out under traffic, exposing bare asphalt and throwing loose aggregate; too deep and the binder bleeds up through the stones, creating a slick surface. Yet according to the chair of AASHTO's Emulsion Task Force there is "categorically" no field test that can measure percent embedment quickly and accurately on the finished mat, so agencies accept chip seals by application rates and visual judgment and learn about embedment problems only when the surface fails months later — after the contractor has been paid.

Why this matters

Chip seals are the workhorse of low-cost road preservation; Texas alone runs an approximately $300 million annual chip seal program, and the problem statement calculates that adding a single year of life to only 20% of sealed lane-miles would save that state more than $9 million per year. Because embedment cannot be measured on the spot, it cannot be written into contracts as a pay item, so there is no incentive/disincentive lever on the workmanship variable that most determines performance. FHWA's own preservation reviewer called correct embedment "key to the success or failure of the treatment, and directly related to public acceptance or rejection of this technique" — chip seals have a reputation problem with the public precisely because raveling and bleeding are common and visible.

What’s been tried and why it hasn’t worked

The standard field texture test, ASTM E965 (the sand-patch method), measures pavement macrotexture depth, not how far aggregate is embedded in binder — it is a proxy at best and cannot distinguish a well-embedded coarse chip from a poorly embedded fine one. Ozdemir et al. compared three approaches and got usable percent-embedment numbers only from image processing of chip-seal cross-sections, which requires cutting cores — destructive, slow, and useless for same-day quality control. The Ames laser texture device gives a quasi-embedment reading but is a laboratory instrument, and Komaragiri et al. (2018) demonstrated a cart-mounted 3D laser scanner producing a point cloud from which embedment, bleeding, raveling, and void ratio can be computed — a research prototype rather than an inspector's tool. Meanwhile an FHWA reviewer noted, fairly, that "there has been a lot of chip seal research lately" and best-practice specifications have already improved performance, so the marginal value of one more test is uncertain — which is exactly why any solution must be cheap and fast enough to be run routinely, not a lab exercise. NCHRP has since funded Project 10-124 (Auburn/NCAT, $400,000, March 2024–March 2027) to "identify, adapt, or develop a rapid field test method(s)" — evidence that the gap remained open in 2024, and the field remains open to a low-cost approach.

What would unlock progress

The unlock is a non-destructive, near-real-time measurement of embedment on a fresh mat that an inspector can run right after sweeping — the problem statement's definition of "rapid" is results in time to correct the contractor's application rates on the same job. Two routes look plausible: optical (smartphone photogrammetry or structure-from-light of the surface, inferring embedment from visible chip height above the binder plane, calibrated against cores) and physical (a fast pull-off or displacement gauge that relates chip retention force to embedment). The adjacent solved problem is agricultural and food imaging, where percent-coverage and particle-height metrics are routinely extracted from photos of granular surfaces; the chip-seal case adds a binder plane that must be located optically.

Entry points for student teams

A team could build lab chip-seal panels at known embedment percentages (30–70%), photograph them with a phone under a simple shading hood, and train/validate an image model that predicts embedment against sectioned cores — then test the device on a real chip-seal job with a state DOT or county. A mechanical team could prototype a hand-held gauge (e.g., a spring-loaded probe or profilometer comb) and quantify repeatability against laser-scan ground truth. A specifications team could draft how a rapid embedment number would enter an acceptance/incentive clause, drawing on the AASHTO Emulsion Task Force's chip-seal best-practice documents. Relevant skills: pavement materials, computer vision, instrumentation, and construction quality assurance.