The Utilization Report Says CRITICAL: Finding and Fixing the Overload

A CRITICAL rating on the work center utilization report means the plan booked more hours on a station than its calendar can hold, and the fix is almost never the first thing a planner reaches for. This walkthrough takes one overloaded grinder in EDGEBIC by User Solutions from a red rating through diagnosis to a fix, and along the way rejects three tempting answers that would have cost money and solved nothing. It is one of the worked examples where the arithmetic is on the page.

The report finds the problem in about 90 seconds. Diagnosing it correctly takes another 20 minutes, and that is where the money is.

Step 1: Run the Report and Read the Strip

Open Reports and click Work Center Utilization in the Diagnostics group. The date range dialog appears with a default window of 30 days back to 60 days forward. For a capacity question, narrow it to the next four weeks: October 6 to October 31.

The KPI strip above the tabs reads:

Plant utilization at 71 percent says the shop as a whole is not full. Two overloaded stations alongside five underloaded ones says the load is badly distributed, which is a very different problem from being out of capacity. Read those three numbers and move on.

Step 2: Work the Per Work Center Tab

Switch to the Per Work Center tab. Each row is one station over the window, with a rating band:

Two rows are red:

Each row also carries a plain-English load profile line stating the booked hours against capacity hours, the size of the gap, and what to check. Read it before you start theorizing, because it usually points at the right place.

GRIND-1 is 63 hours over in four weeks. That is roughly eight working days of work with nowhere to go.

Step 3: Reject the Expensive Answer First

The instinct is to add a shift on GRIND-1, or to quote a second grinder. Both are real levers and both are wrong until three cheaper causes are ruled out.

Cause one: the instance count. Open the work center and check how many instances GRIND-1 is configured with. If the shop physically has two identical grinders and the record says one, the report is arithmetically correct and factually wrong: it is charging two machines' load against one machine's capacity.

In this case GRIND-1 genuinely has one machine. Cause one is ruled out, but it is worth checking every single time because it is the single most common explanation for a station that reads permanently overloaded while the operators say they are fine. The setting is covered in how to configure instances and utilization.

Cause two: schedule-at-utilization. If the station is deliberately planned at less than 100 percent of its clock capacity to leave room for variability, the available hours in the report reflect that dial, not the raw shift hours. GRIND-1 is set to 100 percent here, so this is not it either.

Cause three: the load itself is wrong. That takes a drill-down.

Step 4: Drill Into the Station's Backlog

Select the GRIND-1 row. Its job backlog appears underneath: every operation booked on the station inside the window, with job number, product, sequence, planned and actual dates, scheduled and actual hours, status, quantity, due date, and the routing's own timing fields (setup, queue, flow, transit, move, teardown, transfer batch size, transfer delay).

Sort by scheduled hours. The top of the list:

Nothing anomalous in the timing fields. The routing parameters are visible even when zero, which is exactly so you can tell "configured as zero" apart from "never configured", and here there is no hidden queue time inflating anything.

But J-4471 stands out for a different reason: 62 hours on a single grinder is nearly two full weeks of one station's life for one job. Open its routing. The grind step is bound to GRIND-1 specifically, and GRIND-2 exists as an idle sibling.

Step 5: Check the Idle Gap Column

Before making the routing change, look at two columns on the per-work-center tab that are worth more than they appear:

• Capacity waste: available hours the plan never booked at all.
• Idle gap hours: the gaps between different jobs on the same machine, instance, and day.

GRIND-1 shows 11.5 idle gap hours over the window. That is recoverable time created by sequencing, not by missing capacity, and it is the cheapest capacity in the building. It will not solve a 63 hour overload on its own, but it is worth knowing that roughly a day and a half of the station's time is being lost between jobs.

Meanwhile SAW-1 at 36 percent and three other stations in the LOW band are the other half of the picture, the half most capacity meetings skip entirely. The plant is not out of capacity. It is out of capacity in one place.

Step 6: Fix It With the Cheapest Lever That Works

Three fixes are available, in ascending cost:

Fix A, route the big job to the idle sibling. GRIND-2 exists and is barely loaded. Bind the grind step on J-4471's routing to the machine pool that contains both grinders, or add GRIND-2 as an alternate work center on the step. Re-plan.

Result: 62 hours move off GRIND-1. Its load drops from 231 to 169 hours against 168 available: 101 percent. Still critical, barely.

Fix B, a capacity override for the peak. Narrow the report window and find where the remaining hour actually bites. It is one Thursday in week three. Add a capacity override of 8 hours on GRIND-1 for the Saturday of that week. The engine creates a synthetic override-day slot and places work there.

Result: available hours rise to 176 against 169 booked. 96 percent, HIGH. Tight, not broken.

Fix C, the permanent answer. If this load profile is the new normal rather than a one-quarter spike, HIGH at 96 percent is not a place to live: it has no room for a breakdown. That is the conversation where a second shift or another machine is genuinely the right answer, and it is a conversation you can now have with a number instead of an argument. Adding a second shift to break a bottleneck walks through that decision.

Step 7: Re-Run and Confirm

Run the report again over the same window.

One down, one to go. DEBURR-1 gets the same treatment on the next pass, which is the right rhythm: fix one station, confirm, move to the next, rather than changing six things at once and losing track of which one helped.

What This Walkthrough Proves

1. CRITICAL is about the plan, not the machine. It says the schedule committed hours the calendar cannot hold.
2. The underloaded count matters as much as the overloaded one. Five LOW stations beside two CRITICAL ones is a distribution problem.
3. Check the instance count first, every time. It is the most common false overload and it costs nothing to verify.
4. The backlog drill-down is where the cause lives. One 62 hour job pinned to one machine explained most of a 63 hour overload.
5. Buy capacity last. Routing changes and a one-day override took a station from 137 percent to 96 percent without a purchase order.

For the monthly rhythm this fits into, month-end capacity review is the companion journey, and production bottleneck identification covers the general theory of finding the constraint before you spend money on it.