Magnolia crossing simulator
What happens when a crossing closes—and how much does added traffic change the outcome? Follow the same day, on the same streets, with both directions of travel.
Both inbound and outbound. The map always shows all traffic; this filter changes the trip metrics.
Illustrative buildout sizes, independent of city growth forecasts.
The first three rows close only their named crossing; the fourth keeps every crossing open. Left and right use the same closure and matched existing trips. The difference measures the selected added traffic within these model assumptions; it does not establish the real zoning effect.
Fort Street and Commodore Way remain available in every row.
Who moves in, and how do they travel?
Seven profiles vary residents, daily trips and car use. Car trips are calculated from these inputs.
No added car tripsMore residents and car use
See the exact inputs below6 residents/home · all drive alone
4 one-way trips/person/day
Ranges combine household occupancy, trip frequency and car use. Road operations are held fixed. The maps show your selected slider setting. Bounds are assumptions, not probabilities.
Three repair scenarios + all crossings open
Trip window: · all Magnolia trips
Trip outcomes observed through 02:00 next day; map and cumulative totals follow the clock.
○ Representative home areas, not exits ◆ External road boundaries ✕ Closed span · Gray streets provide context; colored streets are simulated. Hover a colored road for its directed segment and 10-second measurements. Maps use the same wider extent.
Confidence and evidence What supports these planning estimates, and how to strengthen them
Field validation: incomplete. Historical counts for the crossings and 15th were collected in different years; there is no common complete weekday in the supplied surveys. Current signal plans, observed queues and travel times, and verified buildout/geography for the exact proposal are still missing. The displayed vehicle-demand scenarios are independently specified.
| Evidence | Finding | Meaning for these maps |
|---|
SUMO engine: tests and release record
Data that would improve the estimates
Identify the exact proposal version, geography and occupied-home buildout; simultaneous directional and turning counts including Fort/Commodore and 15th; route travel times and queue onset/recovery on those dates; signal phases, offsets, turn-pocket lengths and lane permissions; and held-out days for validation. Use the same study boundaries and measurement intervals for observed and simulated data. These inputs would turn conditional planning estimates into a locally validated forecast.
Earlier city studies did analyze Magnolia Bridge closures. This project does not claim no prior closure study exists; the unresolved task is a verified comparison tied to the current proposal across individual crossings.
Operating assumptions, explained How vehicles, junctions and rerouting work
SUMO is an established traffic simulation package; these results use its mesoscopic queue engine and remain uncalibrated. Engine pedigree does not verify the imported roads, generated signals or assumed demand. This is native SUMO, separate from the earlier custom Python queue engine. The rejected microscopic stress runs are not used as results.
Can throughput drop to zero on green? Yes: a full receiving road can stop upstream discharge. A controlled native test verifies this and recovery after release. However, this model has no vehicles occupying the interior of a junction. It cannot establish the particular effect of an opposing car blocking a left turn onto Magnolia Bridge. Shared-approach lane use is also simplified.
| Assumption | What it means | Practical limit |
|---|---|---|
| Lane and junction geometry | SUMO imports the pinned OpenStreetMap streets: 2,941 directed road edges and 720 eligible residential access roads. Magnolia, Dravus, Emerson Place, Fort Street and 15th are represented individually. | Connections, control directions and lane permissions are unverified. Turn pockets and bus-only lanes can be incomplete. This is not a surveyed lane inventory. |
| Signals | Generated fixed phases have a 90-second cycle, three seconds yellow and two seconds all-red. Junction control checks the actual signal phase rather than an average delay multiplier. | The 32 generated controllers are not SDOT timing plans. Offsets, pedestrian demand and bus priority are not calibrated. |
| Following and turning | Finite road queues use native free-flow and jammed headways, approximately 25 m segments, separate lane/turn queues and generated right-of-way rules. | Individual lane changes and vehicle bodies inside intersections are not modeled. A shared approach can delay through traffic even when another lane could be physically usable. Controls and discharge rates need verification. |
| Congestion rerouting | All drivers use travel-time estimates updated every minute, with a five-minute opportunity to reroute. New departures refresh their route estimates every minute. A closure triggers a local detour. | This assumes broad route information. It does not simulate habitual routes, individual navigation apps, or perfect knowledge of future queues. |
| 10,000 through trips/day | An assumed 5,000 trips each way travel between the 15th/Elliott external boundaries. They use capacity and appear on maps. Magnolia commute and cumulative-loss figures count only trips to or from Magnolia. | This is a background-traffic assumption, not an SDOT estimate. Other Queen Anne, port, freight and local-only demand is incomplete. |
| Closure operation | Only the named crossing closes, in both directions. Repair closures last the full day and drain; event/maintenance blockages last one hour. Drivers already on the crossing may finish traversing it; new entries are forbidden until reopening. | There is no vehicle disappearance or teleporting to clear congestion. Native route timestamps are checked for forbidden new admissions. |
| Maps and trip minutes | Colors show mean blocked vehicles over ten seconds divided by estimated road storage. Commute comparisons group the same existing trips into five-minute departure windows. | The display bins are not the simulation timestep. Colors measure blocked road queues, not instantaneous vehicle speeds or every form of slow traffic. Unfinished trip times are withheld, with elapsed burden retained. |
Method references: FHWA microsimulation guidance · SUMO intersections · SUMO safety checks.
How to read the planning range
Default range: everyone uses the bus through six residents per home, all driving alone, with four one-way trips each day. A second view focuses on the three profiles where everyone drives. The formula is homes × residents/home × one-way trips/person/day × car share ÷ people/car. These vary vehicle demand; signal, turning and other operating assumptions remain fixed. This does not bound every source of uncertainty.
For each five-minute departure window, compare the same existing trips on both sides. Outbound means leaving Magnolia; inbound means arriving. The headline difference is the change in average elapsed minutes from scheduled departure to the modeled road boundary (or the reverse), including origin waiting.
If either side has an unfinished trip in the matched window at 02:00, both trip-minute headlines and the added-minutes figure are withheld. The CSV keeps observed elapsed minutes in separately labeled columns. For these windows the range panel still reports the difference in elapsed minutes accrued by 02:00, including every unfinished trip. It is not the eventual delay range or a lower bound on it. Empty windows display a dash.
Cumulative time lost adds the time that all selected drivers have accrued beyond their own open-network free-flow travel time. It includes extra drivers, detours, queues and waiting to enter the network. It is shown in driver-hours: 60 driver-minutes = 1 driver-hour. Its delta is a different measure from the matched-trip minutes above. The “have not entered roads” count identifies requested trips still waiting outside the map. CSV origin-wait hours are already included in trip time; do not add them to the loss total.
Assumptions to keep in view
Added traffic is calculated from household assumptions, independently of city zoning trip estimates or assumed bus uptake. At 1,000 new homes, profiles range from zero to 24,000 added vehicle trips/day. The upper end is an outer stress scenario: six residents per home, four one-way trips daily, each in a separate car.
Homes are additional occupied units, not zoned capacity or a predicted construction total. Daily trips are separate from cars owned. These modeled trips enter or leave Magnolia; local-only errands are not included. Added trips follow the reference geographic shares and daily timing, split equally inbound and outbound in expectation. The bus endpoint assumes no added private-car trips; it does not establish that transit service can accommodate those riders.
Reference demand is provisional; signal timings are assumed. The model includes 10,000 illustrative independent through trips on 15th/Elliott per day; bus capacity, freight and other external demand remain incomplete. The day starts empty at midnight. Each slider stop is a separate run with the same random seed; there is no interpolation or uncertainty interval.
One-hour outages at 08:00, 12:00 or 17:00 are illustrative, not announced works. The Emerson row closes Emerson Place bridge, leaving Fort Street–Commodore Way and Emerson Street open. No “north exit” exists inside Magnolia: the northern external boundary is on 15th Ave W toward Ballard.