Steel Fab Parts Forecast - Masaba Inc

Loading Parts Forecast

Fetching cache from CDN...

Parts Cut vs On-Time Completion

Bars = total fabricated parts due each week (anchored to pre-fab goal = PSD − 7d). Solid burgundy = actual BOMS counts; striped purple = forecast estimates from per-model / per-type averages. Green line = OTC% computed per-workstation and mirrored from the OTC dashboard (|days±|≤30, MANU+PROGRAM excluded). Click any bar to see the projects driving that week.

Historical parts (actual) Forecast parts (estimated) OTC % Today

Top model averages (n ≥ 5)

Model	n	Avg parts	p50	p90

Equipment type averages

Type	n	Avg parts	p50	p90

Parts Cut vs Average Days Late

Same weekly parts bars. Amber line = average days late across all non-on-time workstation completions that week (mirrors otc-dashboard avg-days-late summary cell). Click any historical bar for the per-project breakdown of late jobs.

Historical parts (actual) Forecast parts (estimated) Avg days late Today

Forecast estimate source mix

Source	Projects	Est parts

Upcoming forecast weeks

Week starting	Projects	Est parts

BOMS Capture Verification

Per-year fabricated part counts captured by the BOMS reader vs. the engineer-curated BARS+PLATES Total Parts summary cells inside each yearly Excel file. The 2023 over-capture is because that year's PLATES SORTED summary cell was incomplete (5k vs 130k+ actual) so the file's own summary undercounts; the ORIGANOL/main-sheet F-row sum is the real number.

Week drill-down

How is this dashboard calculated?

Two time-series charts that combine historical actuals from the steel fab BOMS spreadsheets with a forward forecast based on upcoming PSDs. Click any historical bar to drill into the projects, any forecast bar for the upcoming projects.

Historical parts (left-axis solid bars)

Sum of fabricated BOM lines per job from the four yearly files at S:\MFG ENG\STEEL ANNUAL DATA\YYYY YEARLY BOMS.xlsx.
Rows where the M/P (or F/P) column is F or M. The 2022 reader auto-detects column-swap corruption (Mass and indicator swapped in ~21k rows) and recovers them — without that fix 2022 totals were 65% short.
Each job's parts are anchored to its pre-fab goal week (PSD − 7 days) so the bars align with when cutting was scheduled, not when parts were ordered or shipped.
Per-year fill-rate vs the engineer-curated BARS+PLATES summary cells: 2022 102%, 2024 98%, 2025 101%. (2023 is over because that year's PLATES summary cell is incomplete.)

Forecast parts (right side, striped purple bars)

For each upcoming Production Queue item with a PSD in the next 60 days, we anchor the parts to its pre-fab goal week and estimate part count using the first available source:

released_actual — the job already has BOM lines in the yearly files, so we use the actual count.
model_avg — description parses to a known model (e.g. 36x60, 1316 IMPACTOR) with n≥3 historical samples, so we use that model's average.
type_avg — falls back to the equipment-type average (Plant / Conveyor / Magnum / Screen / Other).
global_avg — type is Unknown, so we use the all-time average across every job (least precise).

OTC % (green line)

Computed per-workstation using the same exact filter as /otc-dashboard:

Drop rows where the workstation is MANU or PROGRAM.
Drop rows where |days±| > 30 (data quality clip — extreme outliers usually indicate a wrong date).
OTC% = on-time rows / total surviving rows for the week.
Aggregated by completedAt week — i.e. the week the workstation actually clocked done, not the week the part was scheduled to cut. This is identical to how the OTC dashboard reports OTC.

Average days late (amber line)

Mean daysDiff across all non-on-time filtered rows for the week. Mirrors generateSummary() in otc-dashboard.

Type filter

When you pick an equipment type, parts/jobs and the OTC/late lines are all re-aggregated client-side using the slim otcRows array shipped in the cache and the per-job equipType field. So picking "Plant" gives you the same Plant-only OTC numbers you'd see if the OTC dashboard had a type filter.

Pre-fab cut window assumption

For each project, the cut window is assumed to start 14 days before pre-fab goal and span 7 days (PSD − 21 to PSD − 14). Parts are anchored to the pre-fab goal week for both historical and forecast bars so the timing aligns.