Zipper merge Français

Use both lanes. Take turns.

A zipper merge means using both lanes up to where one ends, then taking turns. Below, the same road runs twice: with your setting (everyone zippers to start with) and with typical drivers, most of whom move over early and don’t let the few zipper mergers in. Zippering barely changes how many cars get through; it fits the waiting cars into both lanes, so the line is shorter.

Your setting 0 cars past the merge
no queue · ending lane used –
Typical drivers: most merge earlyTypical drivers 0 cars past the merge
no queue · ending lane used –
zipper driver: in slow traffic, stays in lane to the merge point and takes turns; while traffic flows, merges early early merger: leaves the ending lane soon after the first sign and rarely lets others in braking traffic speed along the road: fast → stopped

Tap a road to drop a car into a lane. It appears on both roads, so the comparison stays fair.

Queue length over time Your settingTypical drivers: most merge earlyTypical drivers
Show the numbers
MinuteYour settingTypical drivers

Fig. 1 The counters show how many cars got past the merge point since you started watching. Drag the sliders, add a burst of cars, or tap the road to drop one in. In light traffic (below about 1,300 cars an hour), no queue forms on either road: the zipper only matters once traffic backs up. Real roads with zipper signs look like about 75 to 100% here, since not everyone follows them.

What a zipper merge is

A zipper merge, also called a late merge, is how two lanes become one in heavy traffic. Drivers use both lanes right up to the point where one of them ends, then take turns: one car from the lane that continues, one from the lane that ends, then the next. It works like the two halves of a zipper meeting at the slider.

In U.S. work zones, it is signed with Stay in lane to merge point and Merge here take turns Federal Highway Administration (2025).Manual on Uniform Traffic Control Devices11th ed. with Revision 1 (Dec 2025), §§2C.47, 6B (Tables 6B-1, 6B-3, 6B-4), 6G.07, 6H.07, 6H.08, 6N.19, Figure 6N-1Technical standardread in fullCurrent edition. Late-merge text is Guidance and Option for agencies; it creates no driver obligation.Open source ↗ · Sources. Quebec’s transport ministry calls it convergence tardive (en « fermeture éclair ») and tells drivers to favour it whenever a lane ends or is closed for roadwork Gouvernement du Québec (MTMD) (2026).À propos des voies de circulationQuébec.ca, section « Convergence tardive (en « fermeture éclair ») lors d'une fusion de voies »; updated 13 Jul 2026Agency guidanceread in fullHeading read « Insertion obligatoire … » from Dec 2025 to Mar 2026, then softened; do not cite it as a rule.Open source ↗ · Sources. Germany wrote it into its traffic code: drivers in the continuing lane must let merging cars in, one at a time, “immediately before the start of the narrowing” Germany (2013).Straßenverkehrs-Ordnung (StVO 2013)§ 7 Abs. 4Law or regulationread in fullDate the "unmittelbar vor Beginn der Verengung" wording was added is not verified.Open source ↗ · Sources.

Why it feels wrong

Because most of us were taught the opposite. Quebec’s driver handbook still says to move into the open lane as soon as the narrowing is signed, and not to wait until the last second SAAQ (2024).Guide de la route4e édition (PDF posted 20 Jun 2025), pp. 159, 170, 177–178, 254–255Official driver handbookread in fullAdvises merging early in work zones (p. 255), contrary to Québec.ca.Open source ↗ · Sources. Driver manuals in Kansas, Nebraska and Pennsylvania give similar advice, even where the state’s own transportation department campaigns for the zipper Kansas (2022).Kansas Driving Handbookrev. Feb 2022, p. 92Official driver handbookread in fullSays "Merge as soon as possible."Open source ↗ · Sources Nebraska (2025).Nebraska Driver's ManualJan 2025, pp. 39, 43Official driver handbookread in fullSources Pennsylvania Department of Transportation.PUB 95PUB 95 (4-21), p. 19Official driver handbookread in fullDriver manual; says "Begin your merge well in advance of the sign."Sources.

Because it looks like cutting in line. Texas researchers describe drivers trying to keep “queue-jumpers” out, and trucks pairing up side by side to block the ending lane Ullman, G.L.; Barricklow, P.A.; et al. (2002).Traffic Management and Enforcement Tools to Improve Work Zone SafetyTexas Transportation Institute, Report FHWA/TX-03/2137-3Government research reportread in fullOpen source ↗ · Sources. At one Iowa closure, a video study counted 51 drivers straddling the lane line; those incidents led to 14 forced, last-second merges Hallmark, S.; Mudgal, A.; Stout, T.; Wang, B. (2011).Behavior Study of Merge Practices for Drivers at Work Zone ClosuresInTrans Project 09-359, Iowa State University / SWZDIGovernment research reportabstract onlyAbstract and executive summary read.Open source ↗ · Sources.

And because few people know it. In a 2024 survey of 1,050 drivers in nine Midwest states, 34% were familiar with the zipper merge and 21% were comfortable doing it. Only a quarter thought it was safer than merging early Savolainen, P.T.; Gates, T.J.; et al. (2024).Improving the Effectiveness of Zipper Merge Lane Control in Freeway Work ZonesSmart Work Zone Deployment Initiative, TPF-5(438), InTrans Project 20-733Government research reportread in fullOpen source ↗ · Sources.

What it actually does

In heavy traffic, the zipper merge makes the backup shorter, not faster. Here is what field studies have measured, and what our simulation adds.

  1. Shorter backups. With both lanes holding the queue, the queue is shorter. Field studies have measured reductions of roughly 8 to 35% Kang, K.-P.; Chang, G.-L.; Paracha, J. (2006).Dynamic late merge control at highway work zones: evaluations, observations, and suggestionsTransportation Research Record 1948, 86–95Peer-reviewed studyread in fullRead as the authors' preprint.Open source ↗ · Sources Minnesota Department of Transportation (2004).Evaluation of 2004 Dynamic Late Merge SystemITS JPO Knowledge Resources record 2005-L00041Government research reportvia another sourcePrimary not found; known via the FHWA comparative analysis and Radwan et al. 2009. The ITS JPO record returned 403.Open source ↗ · Sources; half is the theoretical best case, two lanes of storage instead of one McCoy, P.T.; Pesti, G. (2001).Dynamic late merge-control concept for work zones on rural interstate highwaysTransportation Research Record 1745, 20–26Peer-reviewed studyread in fullRead as the TRB preprint. Its "about 50 percent" queue reduction is a conceptual argument, not a measurement.Open source ↗ · Sources. The often-quoted “up to 40%” comes from a 2008 Minnesota DOT memo based on what staff had seen, not from a published study Minnesota Department of Transportation (2008).LATE MERGE … the ZIPPER SYSTEM2 pp., Oct 2008; contact K. Johnson, State Work Zone EngineerAgency guidanceread in fullSource of "up to 50% (40% is common)", based on field experience; no study cited.Open source ↗ · Sources. In our simulation the advantage is biggest early on: with 2,400 cars an hour arriving, the queue was 49% shorter with everyone zippering after 10 minutes, 32% after 20 and 21% after 30. It shrinks because once the queue reaches back past the warning signs, early mergers end up using both lanes too. It also depends on traffic (53% at 1,800 cars an hour, 23% at 3,200) and, most of all, on where early mergers actually move over, which we had to assume. If they all move over at the first sign, the zipper queue is 41% shorter; if they wait until between the second and third signs, 14% shorter. Waiting until the last few hundred metres would leave far more cars in the ending lane than field counts show, so we ruled that out. (Figures over time average 5 random draws; the others, 3.)
  2. Both lanes get used. With conventional signing, about 5 to 18% of drivers are still in the ending lane near the taper. With zipper signing in congestion, it is 38 to 49% Vaughan, C.L.; Cline, A.; et al. (2026).Dynamic Zipper Merge for Work Zones: Safety and Operational EffectsNCDOT Project 2024-05, FHWA/NC/2024-05, ITRE/NC StateGovernment research reportread in fullObservational, not before/after; the authors note ADT differences could favour the dynamic zipper.Open source ↗ · Sources Wolshon, B.; Wilmot, C.G. (2012).Design of Lane Merges at Rural Freeway Construction Work ZonesLTRC Report 484, Technical Summary (Oct 2012)Government research reportread in fullTechnical summary read.Open source ↗ · Sources Savolainen, P.T.; Gates, T.J.; et al. (2024).Improving the Effectiveness of Zipper Merge Lane Control in Freeway Work ZonesSmart Work Zone Deployment Initiative, TPF-5(438), InTrans Project 20-733Government research reportread in fullOpen source ↗ · Sources. Our model gives 12% with typical drivers and 50% with everybody zippering, but that mostly follows from what we set: one in ten typical drivers tries to zipper, and with perfect zippering half the cars are in each lane by definition. Field counts also find fewer drivers in the ending lane as queues grow, about 14% with a short queue and 5% with a long one; our model stays at 12% in both, so it misses that trend.
  3. Not more cars through. This is the part most explainers get wrong. Careful studies of two-to-one closures found no statistically significant gain in throughput Beacher, A.G.; Fontaine, M.D.; Garber, N.J. (2004).Evaluation of the Late Merge Work Zone Traffic Control StrategyVirginia Transportation Research Council, VTRC 05-R6Government research reportread in fullOpen source ↗ · Sources Radwan, E.; Harb, R.; Ramasamy, S. (2009).Evaluation of Safety and Operational Effectiveness of Dynamic Lane Merge System in FloridaFDOT Contract BD548-24, University of Central FloridaGovernment research reportread in fullOpen source ↗ · Sources Wolshon, B.; Wilmot, C.G. (2012).Design of Lane Merges at Rural Freeway Construction Work ZonesLTRC Report 484, Technical Summary (Oct 2012)Government research reportread in fullTechnical summary read.Open source ↗ · Sources. Minnesota’s own memo says the zipper “has no effect on travel time through the work zone” Minnesota Department of Transportation (2008).LATE MERGE … the ZIPPER SYSTEM2 pp., Oct 2008; contact K. Johnson, State Work Zone EngineerAgency guidanceread in fullSource of "up to 50% (40% is common)", based on field experience; no study cited.Open source ↗ · Sources. Once traffic has queued, a typical two-to-one closure passes roughly 1,300 to 1,650 vehicles an hour Krammes, R.A.; Lopez, G.O. (1994).Updated capacity values for short-term freeway work zone lane closuresTransportation Research Record 1442, 49–56Peer-reviewed studyread in fullRead by a research sub-agent.Open source ↗ · Sources Codjoe, J.; Thapa, R.; Osafo, Y.A. (2021).Evaluation of DOTD's Existing Queue Estimation ProceduresLTRC Final Report 642Government research reportread in fullOpen source ↗ · Sources. Our model gives 1,450 cars an hour with typical drivers and 1,420 with everybody zippering. That difference is small next to the spread across assumptions we can’t pin down, such as how much room drivers leave in the work zone or whether early mergers close up to keep others out: changing them moves zipper merging anywhere from −16% to +2%.
  4. Fewer forced merges. In Nebraska’s field test, the late merge had about 75% fewer forced merges than the standard layout McCoy, P.T.; Pesti, G. (2001).Dynamic late merge-control concept for work zones on rural interstate highwaysTransportation Research Record 1745, 20–26Peer-reviewed studyread in fullRead as the TRB preprint. Its "about 50 percent" queue reduction is a conceptual argument, not a measurement.Open source ↗ · Sources. Our model can’t count quite the same thing: its forced merges are defined by our own rules (a driver takes a smaller gap than drivers normally accept, or pushes in after waiting 40 seconds at a crawl). With typical drivers, 9% of merges were forced (6–11% across runs), against 2% with everybody zippering. Part of that comes from the push-in rule: without it, typical drivers would be at 6%. Near capacity, before a queue forms (1,500 cars an hour), the two are much closer (5% against 6%), because zipper drivers merge early while traffic still flows, as agencies advise.
  5. Possibly fewer crashes. A 2026 North Carolina study of rural freeway work zones estimated about 44% fewer crashes where sensor-triggered zipper systems were used (95% confidence interval: 29 to 57%). It compared different sites rather than the same site before and after, so this is promising, not proven Vaughan, C.L.; Cline, A.; et al. (2026).Dynamic Zipper Merge for Work Zones: Safety and Operational EffectsNCDOT Project 2024-05, FHWA/NC/2024-05, ITRE/NC StateGovernment research reportread in fullObservational, not before/after; the authors note ADT differences could favour the dynamic zipper.Open source ↗ · Sources.

One honest caveat. How drivers behave at the merge (who lets whom in, when an early merger gives up waiting and pushes in) is our assumption: no study measures it, and the numbers above depend on it. The traffic physics is checked against field data; that behaviour is not. How the model works lists every assumption and shows how much each one moves the results.

When not to zipper

The zipper is a heavy-traffic tool. When traffic is moving at highway speed, agencies advise moving over early, as soon as there is a safe gap Minnesota Department of Transportation.Zipper mergeAgency guidanceread in fullIts "as much as 40 percent" backup reduction cites no study; agency rule of thumb.Open source ↗ · Sources. Britain’s Highway Code recommends merging in turn only “when vehicles are travelling at a very low speed” United Kingdom.The Highway CodeRule 134Official driver handbookread in full"Should" rules are not offences in themselves but can be used in evidence. First sentence spot-checked.Open source ↗ · Sources, and Belgium requires it only in heavily slowed traffic Belgium (1975).Koninklijk besluit van 1 december 1975 houdende algemeen reglement op de politie van het wegverkeer (Wegcode)art. 12bis "Ritsen"Law or regulationread in fullRule in force since 1 Mar 2014; the amending decree number was not verified.Open source ↗ · Sources. Sensor-triggered systems switch their zipper messages on only once speeds drop below about 35 to 45 mph Datta, T.K.; Hartner, C.; Grillo, L.F. (2007).Evaluation of the Dynamic Late Lane Merge System at Freeway Construction Work ZonesMichigan DOT Report RC-1500, Wayne State UniversityGovernment research reportread in fullOpen source ↗ · Sources Kang, K.-P.; Chang, G.-L.; Paracha, J. (2006).Dynamic late merge control at highway work zones: evaluations, observations, and suggestionsTransportation Research Record 1948, 86–95Peer-reviewed studyread in fullRead as the authors' preprint.Open source ↗ · Sources.

A few places officially want the opposite. Oklahoma’s law says a merge at a lane closure must be completed “as soon as practicable” after the sign Oklahoma.Maintenance and construction zones47 O.S. §11-1302 (Title 47 complete-title file, last modified 31 Dec 2025)Law or regulationread in fullSubsection (D) requires merging as soon as practicable after the sign; (G) makes violations a misdemeanor.Open source ↗ · Sources, and Tennessee’s driver manual says “Merge early!” Tennessee DOS (2022).Comprehensive Driver License ManualJul 2022 ed., pp. 18, 39Official driver handbookread in fullRead via a Wayback Machine capture.Open source ↗ · Sources. Posted signs and local rules come first.

Common claims, checked

How we know

Every number on this page links to its source. Tap a citation to see what kind of evidence it is and whether we read it in full. The simulation’s parameters, its assumptions and its calibration against field data are on How the model works. The full reading list is under Sources.