End-of-Shift Report · Summit Grading & Excavation · org_8f2c1a90

Rosecrans Ave Widening — Phase 2

Worksite▾Machines▾Time period▾

Pages

stacked (z-axis)

site_44b1 · Rosecrans Ave & Aviation Blvd, Hawthorne, CA · America/Los_Angeles

Shift · July 8, 2026 · 06:00 → 18:30 PDTGenerated · 6:30 PM PDTDevices 31/33 · LOW_POWER · LTE

Viewing as Jessica Alvarez · ORG_ADMIN · last GNSS fix 33.9164, -118.3526 · 9/11 sats

Executive review14 findings ranked by deviation · 2 need action now

Derived · not curated

01
Dump Truck 11 ran much longer than it usually does
6 h this shift vs 1.9 h last week
+216% against its own baseline··basis: Engine hours compared to the same machine's trailing week
act now
02
A person came within 1.2 m of Skid Steer 04 while it was moving
12:45:42 · HUMAN_BODY_PROXIMITY · PERSON at REAR · score 91
Hard incident · 1.2 m against a 2.6 m shift mean··basis: Any event graded INCIDENT is surfaced regardless of rank
act now
03
Dump Truck 14 ran much longer than it usually does
5.2 h this shift vs 3.4 h last week
+53% against its own baseline··basis: Engine hours compared to the same machine's trailing week
look today
04
The fleet spent close to half the shift not working
54% working time · 21.4 h burned idle
16 points under the 70% expectation··basis: Working hours over total on-time against the standing utilization floor
look today
05
Several machines went quiet at the same moment
1.3 h block from 11.3:00 · Delivery block · IDLE (5 assets)
6.3 machine-hours lost in one synchronized stall··basis: Overlapping IDLE windows bracketed by STARTED / STOPPED events
look today

DRAFT

—

Machines working (active footprint)

8assets

Total time engines ran (cumulative runtime)

46.5hours

Time sitting still (idle share)

46%

Change vs last week (Δ)

+6.3hours

What broke away from normal3 of this book’s findings · ranked by deviation

act nowthreshold

Dump Truck 11 ran much longer than it usually does

this shift against the machine's own trailing week

+216% against its own baseline

Engine hours compared to the same machine's trailing week

look todaythreshold

Dump Truck 14 ran much longer than it usually does

this shift against the machine's own trailing week

+53% against its own baseline

Engine hours compared to the same machine's trailing week

look todaythreshold

3 machines reported through an unsettled sensor

3 unsettled · 9 calibrated

3 of 12 streams not CALIBRATED

Calibration status of every reporting stream

Busiest two machines (top-2 share)

30%

Unevenness score (concentration index)

0.09

Average per machine (selection mean)

5.8h

The dashed diagonal is perfectly even scheduling; the orange curve is the actual cumulative runtime distribution. The further it bows below the diagonal, the more the workload concentrates on a handful of machines while others sit under-scheduled.

Fleet presence held constant at eight assets on the primary corridor. Steady utilization early in the shift gave way to a sharp drop-off in the late afternoon — the street surface has been broken and cleared, so Excavator 12 (asset_7c3d) and Excavator 07 (asset_1f88) now report MotionStatus = OFF on the street margin under the 72-hour staging protocol. Open the scope panel to stack several worksites or machine kinds at once; every figure, row, and curve recomputes against the current selection.

Ready-made questions · type to see more

›

DRAFT

—

Time doing real work (productive fraction)

54.2%

Time wasted with engine on (wasted engine hours)

21.3hours

Most machines stopped at once (peak co-idle)

6assets

How fast waste grew late (late-shift waste slope)

1.78h / h

What broke away from normal3 of this book’s findings · ranked by deviation

look todaythreshold

The fleet spent close to half the shift not working

54% working · floor at 70%

16 points under the 70% expectation

Working hours over total on-time against the standing utilization floor

look todaythreshold

Several machines went quiet at the same moment

1.3 h stall from 11:00 · 5 machines quiet together

6.3 machine-hours lost in one synchronized stall

Overlapping IDLE windows bracketed by STARTED / STOPPED events

look today2.8σ

Forklift 05 sat still far more than the rest of the fleet

population of 12 · mean 33% · this reading 72%

2.8σ above fleet mean of 33%

Idle share z-scored across all machines reporting this shift

Machine (asset)Working vs sitting still (productive / idle)Sitting h (idle)Sitting % (idle)Stops

Stops per machine (idle episodes · MotionStatus = IDLE)

60 stops (episodes)

Excavator 07

50%

Excavator 12

43%

Skid Steer 04

51%

Dump Truck 08

47%

Dump Truck 09

46%

Dump Truck 11

45%

Compactor 02

42%

Trencher 01

41%

Machines stopped at the same time (co-idle density · peak 6)

06070809101112131415161718

How long the stops were (episode shape · stuttering vs dead-time blocks)

0–10 min

2.4h

10–20 min

7.2h

20–30 min

30–60 min

60+ min

8.7h

◼ Short stops (micro-stoppages) · 53 · 12.6h◼ Long stops (continuous blocks) · 7 · 8.7hTypical stop (median) 15 min · typical wait between stops (mean gap) 0.2h

Wasted time piling up vs work finished (waste accumulation vs production burn-up)

growth speed (slope) 1.77 → 1.78 h/h

◼ Total time sitting still (cumulative idle) · 21.3h◼ Work finished (production burn-up) · 25.2h

54.2% of powered-on time executed active work; the remaining 21.3 engine hours burned with zero task. The waste is not evenly scattered — co-idle peaks at 6 assets between 11:30 and 11:40, which points at a downstream dependency failure rather than individual operator behaviour: the 11:15 window matches commercial deliveries on Rosecrans Ave that force theDUMP_TRUCK units into IDLE and block the FRONT_LOADER clearing paths, while the 15:00 hold is surveyor sign-off on hole depths with engines left running. 53 micro-stoppages (12.6h) sit alongside 7 continuous blocks (8.7h), so the site is losing time to both logistical stuttering and forgotten engines. Waste accrues at 1.77 h/h before midday and 1.78 h/h after, accelerating exactly as the production burn-up plateaus.

Ready-made questions · type to see more

›

DRAFT

—

Task eating the most time (dominant cost code)

CC-40234%

Share on real building work (production codes)

41%

How jumpy the effort was (step-variation index)

1.01bursty

Tightest attack on one task (synchronization)

100%CC-105

What broke away from normal2 of this book’s findings · ranked by deviation

keep an eyethreshold

A slice of the shift is not attached to any task

6% not attached to a task · tolerance 5%

Unbooked effort above the 5% tolerance

Hours landing on CC-000 versus total allocated hours

keep an eye1.3σ

CC-402 absorbed more of the shift than any other task

population of 4 · mean 11.0 h · this reading 14.5 h

1.3σ above the mean task load of 11 h

Booked hours per cost code, z-scored across booked codes

Tasks in the logs (cost codes) · 5 codes · 31 tagged sessions

Task (cost code)

Share of effort (allocation)

Hours

Machines

Window · tightness

Kind of work the hours went to (cost-code group mix)25.8 h tagged

Real building work (production)41.1%

Helping work (support)52.0%

Paperwork & waiting (admin)7.0%

How the effort piled up (cumulative tagged hours)burst-driven · stepiness 1.01

The dashed diagonal is perfectly steady execution. Each riser is one clock hour of booked work: mean 1.99 h/h, steepest 6.03 h in the hour ending 10:00 (shaded). Long flat treads between risers mean the tag stayed open while nothing was booked.

When each task was worked (cost-code attack windows)Tightness = share of the window actually booked

06:0008:0010:0012:0014:0016:0018:00

CC-402

63% · ≤3 at once

CC-801

48% · ≤2 at once

CC-309

26% · ≤3 at once

CC-105

100% · ≤1 at once

CC-000

4% · ≤3 at once

Tightest pass: CC-105 — 1.9 h inside 1.9 h. Most dragged out: CC-000 — 1.8 h smeared across 5.5 h.

CC-402 (Utility Trenching / Light Foundations) carries 8.7 of 25.8 tagged hours — 34% — booked by 3 machines across 7 sessions between 07:14 and 11:52. Production codes hold 41% of effort against 59% on support and unassigned work — the price of staging inside a live public corridor. Effort accrues at 1.99 h per clock hour with a step-variation index of 1.01, so execution reads as burst-driven: vertical risers separated by flat treads, peaking at 6.03 h in the hour ending 10:00. CC-105 was attacked in one tight pass while CC-000 was picked up and dropped across 5.5 h — the code to challenge in tomorrow’s sequencing.

Ready-made questions · type to see more

›

DRAFT

—

Close calls (clearance breaches)

How often clearances break (breach rate)

2.95per active hour

How deep threats got (mean penetration)

2.03 m

Share in the danger core (critical zone)

18%inside 1.5 m

What broke away from normal3 of this book’s findings · ranked by deviation

act nowthreshold

A person came within 1.2 m of Skid Steer 04 while it was moving

1.2 m closest pass · shift mean 2.6 m

Hard incident · 1.2 m against a 2.6 m shift mean

Any event graded INCIDENT is surfaced regardless of rank

look todaythreshold

A quarter of close calls were unlike anything the site normally sees

28% unlike normal · tolerance 20%

28% anomalous · closest approach 1.2 m

Events past the P5 distance threshold

keep an eye2.0σ

Close calls spiked in the 12:00 hour

hour by hour · peak 11 at 12:00

2σ above the hourly mean of 5.7

Hourly proximity-event counts z-scored across the shift

Machines by breach rate (events / active hour) · 8 machines · 74 events · 8 surprises (anomalies)

Machine (asset)

Rate vs fleet (breaches / active h)

Events

Rate

Expected

In core

Closest · mean depth

When risk surges (breaches per clock hour)Shift mean 5.7/h · 73% inside milestones

Mobilization & crew check-in

Commercial delivery block

Surveyor sign-off hold

Demobilization & wash-down

Hottest milestone: Mobilization & crew check-in — 12.7 breaches per hour against 2.7/h outside milestone windows. Shaded bands are the scheduled moments the site changes shape.

Does risk scale with runtime (breaches vs active hours)Scaling exponent 2.28 · compounding

0.0 h active · dots = machines, labels = top 3 by rateDashed = linear risk at 2.95/h4.4 h active

Skid Steer 04 is the worst offender at 3.70 breaches per active hour — 1.3× the fleet rate — against a spread of 2.17 across the selection.

How deep the bubble was penetrated (clearance penetration depth)Mean 2.03 m of 4.6 m · deepest 3.99 m

0 m — boundary clip3.1 m — enters 1.5 m core4.6 m — contact

13 of 74 breaches (18%) crossed the 1.5 m critical core; mean standoff at the closest point of approach was 2.57 m, closest 0.61 m.

What gets close, by how close (threat composition per zone)Fleet mix · people 47% · vehicles 49% · furniture 4%

3–4.6 m (outer shell · 15 ft)

people 0% · vehicles 86% · furniture 14%

30%

1.5–3 m (inner buffer)

people 56% · vehicles 44% · furniture 0%

53%

0–1.5 m (critical core · 5 ft)

people 100% · vehicles 0% · furniture 0%

18%

Inside the 1.5 m core people are 100% of the population; in the outer shell vehicles and street furniture hold 100%. Volume is set by traffic, consequence by bodies.

74 clearance breaches across 25.1 active engine hours — 2.95 per working hour, one every 20 minutes of real motion. Risk is scheduled, not random: 73% of breaches land inside a milestone window and the worst hour is 06:00 with 14. Fitted against runtime the scaling exponent is 2.28, so exposure compounds — the longest-running machines carry more than their share of risk. Threats penetrate a mean 2.03 m into the clearance bubble, with 13 events inside the 1.5 m core where people are 100% of the population — the zone to brief tomorrow's crew on.

Ready-made questions · type to see more

›

empty book

Click any underlined name in the report to open its own pad, then file a page here — or into books 01–04 — from the dossier header.