Slow it down before the geometry forces the issue.
Curve Speed Warning uses V2X communications to notify connected vehicles approaching a curve about the recommended speed for that geometry. The infrastructure detects excessive speeds upstream and broadcasts advisory limits, letting CVs decelerate before they reach the curve itself.
Each test curve is divided into an upstream speed-detection region and a downstream warning-and-adjustment region. The RSU collects speed data — from infrastructure sensors or BSMs — and broadcasts advisory speed limits via RSMs to nearby connected vehicles. CVs comply 100% and resume normal behavior past the curve.
Connected vehicles within the detection region transmit Basic Safety Messages every 0.1 seconds (10 Hz), carrying their position and speed to the RSU.
The infrastructure analyzes BSM data each second and flags vehicles traveling above the recommended curve speed. Loop detectors and roadside sensors can also feed in.
When excessive speeds are detected, the RSU broadcasts Road Safety Messages at 1 Hz containing the advisory speed and curve geometry, blanketing the curve with a 1640 ft (500 m) range.
Connected vehicles in coverage adjust their speed to the broadcast limit while traversing the curve, then resume normal speeds on exit. Modeled with a 100% compliance rate.
20 mph speed reduction at the curve. RSU placed at the static advance warning sign, 150 ft from the start of the speed detection region per MUTCD Table 2C-3.
Tighter geometry, 20 mph reduction. Curve radius selected per AASHTO 2018 (7th ed.) Table 3-7 for the target design speed.
Multi-lane freeway analog. Same advance-placement logic, scaled traffic demands of 6,300 / 5,670 / 5,040 / 4,410 / 3,780 vph.
Lower-speed multi-lane curve. Network capacity assumption: 2,100 vph per lane; demands of 4,200 / 3,780 / 3,360 / 2,940 / 2,520 vph for 2-lane analog.
Across thousands of replications, a clear demand-driven sweet spot emerged for the connected-vehicle penetration rate.
Activating Curve Speed Warning increases curve-area delay by design — vehicles slow down. At 100% CV penetration, average vehicle delay can double or more.
There's a sweet spot that lowers average curve speed while keeping the delay increase 'acceptable.' Below it, too few vehicles comply; above it, too many slow down for no extra safety gain.
The sweet spot has little correlation with lane count or speed limit. It's driven by traffic demand: 10% CVs at capacity, 15% at 90%, 20% at 80%, 25% at 70%, 30% at 60% of capacity.
With 0% CVs, simulated vehicles travel faster than the posted limit — consistent with real-world field observation. The lighter the traffic, the faster they go.
To bring average curve speeds down to the level achieved naturally under heavy traffic, lighter conditions require a higher CV penetration rate to compensate.
Curve Speed Warning isn't latency-sensitive — drivers don't need to react in milliseconds. Setting latency and packet loss to zero in simulation reflects this tolerance.
Reaching 100% CV penetration is neither necessary nor desirable. The infrastructure investment scales not with road geometry but with how loaded the corridor is.
It isn't surprising that delay rises when the warning fires — the application's whole purpose is to convince fast drivers to slow down. What's exciting is that the trade-off has a clean, identifiable optimum, and that optimum depends on demand rather than on the curve itself.
Practically, this means a planner sizing a Curve Speed Warning deployment can look at corridor demand profiles rather than re-running scenarios per curve. The same 10–30% CV target translates across two-lane and three-lane curves at both 55 and 65 mph approach speeds.
The simulator's biggest gap is on the safety side: the Surrogate Safety Assessment Model can't read Aimsun's output reliably, so safety here is inferred from average curve speed rather than measured in near-misses. Future work needs a SSAM-compatible export or a purpose-built crash-surrogate metric.
Size the deployment by demand: aim for 10% CV penetration on corridors near capacity, scaling up to 30% on corridors averaging 60% of capacity.
Ensure the RSU's communications range covers the entire curve — not just the advance warning location — so CVs receive the broadcast limit while still inside the curve.
Don't budget for low-latency hardware. Curve Speed Warning is tolerant to communications delay and packet loss; standard-grade DSRC/C-V2X equipment is sufficient.
Trigger on BSM-derived excessive speeds. Only connected vehicles respond to the advisory anyway, so detecting them directly avoids unnecessary sensor infrastructure.
Pair deployment with a more rigorous safety-evaluation tool. Average curve speed is a proxy; a near-miss or potential-crash metric is needed to quantify the real safety gain.