How Capacity Evidence Answers a New Machine Request

The "do we need another machine" argument is usually decided by conviction, seniority, and whoever presents last, because nobody in the room can state what the station's real utilization is. EDGEBIC by User Solutions settles it a different way: measure the station against actual shift calendars, then model each option and reschedule the same jobs so every choice is judged on the same three numbers. The capital decision stops being a negotiation and becomes a comparison.

This post covers the capital-decision outcome. It sits under the EDGEBIC results guide. For the step-by-step version, see comparing two what-if scenarios for a capacity decision.

Why the Argument Never Resolves

Both sides are arguing from the same missing number.

Production says the mill is slammed, and they are describing something real: the mill is where the work piles up and where every expedite lands. Finance says utilization was 62 percent last year, and they are quoting a figure computed from a coarse formula that ignored shift calendars, weekends, and holidays. Neither number is a lie and neither is a measurement.

So the argument runs on anecdote. Production brings the week the mill ran three weekends. Finance brings the month it sat idle. Both weeks happened, and neither describes the year.

A Number You Can Defend

The Work Center Utilization report exists to end that. It computes available hours from the real shift calendars and per-day capacity overrides rather than approximating them, and it rates stations on fixed bands:

Four tabs carry the detail: a per-work-center view with a job backlog drill-down, a 14-day daily heatmap, top jobs by load, and a plan versus actual variance tab.

Read together they answer the question the bands cannot answer alone: is this station overloaded, or is it well loaded and badly sequenced?

• The heatmap shows whether the overload is spread evenly or concentrated in three weeks.
• Top jobs by load shows whether two large orders are responsible.
• The backlog drill-down names the jobs actually queued behind it.

A station at 128 percent because three big jobs landed in the same week is a sequencing decision. A station at 128 percent evenly across a quarter, with a broad backlog, is a capacity decision. Only the second one has to be solved with money.

One caution before you quote a number in a capital meeting: two reports here can disagree about utilization, because Work Center Performance approximates available hours with a coarse formula while Work Center Utilization computes them from real calendars. In a spending conversation, quote the second one.

Model It Before You Buy It

Once the overload is real, the next question is which fix closes it. That is a controlled experiment, and it has four steps.

Step 0, capture the baseline. Write down three things: the late job count today, the constraint station's rating, and the completion dates of the most at-risk orders. These are the scoreboard, and every option gets judged on the same three.

Step 1, model the cheap option. Add a night shift to the station and reschedule the same jobs. Reread the scoreboard.

Step 2, revert, then model the expensive option. Create the second machine, assign it a shift, and add it to the routing steps as an independent parallel or true-alternative work center so jobs can actually reach it. That last part is the step people forget: a machine nobody's routing points at absorbs no work. Reschedule and reread.

Step 3, compare.

Both options clear the lateness, which is the finding that would never emerge from an argument. What separates them is headroom: a night shift leaves the mill busy with little room for growth, two mills at 64 percent leave real room. Now the room is choosing between headroom and capital, which is a decision finance is equipped to make. The numbers above are illustrative; the method is exactly how you run it on your own jobs.

The Four Cheaper Answers the Data Usually Finds

Before the purchase, the same measurement usually surfaces at least one option that costs a fraction of it:

An alternate work center that already exists. Shops routinely own a second machine capable of the work and never configure it on the routing. The BOR Optimization Suggestions report flags alternates configured but never used, among seven rules it ranks by recoverable hours. Counting how often work already gets bailed from one station onto another gives you the same evidence from the floor side.

A setup sequence that is fighting itself. On a changeover-heavy station, a large share of the load can be setup rather than run. Grouping like work recovers hours without capital. See how sequence-dependent setup shapes a schedule.

A routing standard that is overstated. Load is calculated from routing hours. If those hours are wrong in the pessimistic direction, part of the overload is fiction, and the plan versus actual variance tab will show it.

A shift you already pay for and do not schedule into. Adding a second shift roughly doubles a station's available hours, and the model tells you whether that is enough before you staff it.

What the Evidence Cannot Settle

It cannot decide your risk appetite. Running the constraint at HIGH is a legitimate choice for a shop with steady demand and a bad one for a shop chasing growth. The model tells you where each option lands. Which landing you want is strategy.

It does not price anything. The reports deal in hours, jobs, and dates. Capital cost, labor premium, and the return on either come from your finance model, and the scheduling data is an input to it rather than a substitute.

It cannot forecast demand you have not entered. The model reschedules the jobs you have. If next year's volume is 30 percent higher, the answer changes, and testing that means loading the forecast rather than assuming the current backlog is representative. See seasonal capacity planning for that framing.

Department roll-ups are not a substitute. The department view carries planner-maintained capacity and planned-hours figures rather than live calculations, which makes it a rough-cut lens. For a spending decision, work at the work center level. See department capacity analysis for exactly which figures are maintained and which are derived.

It will not stop the argument being political. Some machine requests are about a department wanting its own equipment rather than about hours. Evidence changes what gets said out loud; it does not change what people want. What it does reliably do is make the cheap option impossible to ignore, which is usually where the saving is. See the capacity you already have but cannot see and the generic capacity utilization KPI.

Want to test a capital request against your own backlog? Bring your jobs and routings to a demo and we will model both options and read the scoreboard with you.