Work Center Capacity Planning: See the Overloaded Week Before You Promise the Date

It is Thursday, 4pm. Sales is on the phone with a customer who wants 300 units by the 12th, and they need an answer before the call ends.

You look at the shop. The welding bay looks busy. The machining cells look busy. Everything always looks busy. So you do what everyone does: you add a week of padding to be safe, you say yes, and you find out three weeks later that the welding bay was already carrying twice what it could run.

That is the gap work center capacity planning closes. Not with a crystal ball, and not with a planning consultant. With two numbers per work center, per week: the hours already committed, and the hours you actually have.

The 4pm question nobody can answer

Ask a plant manager how loaded the shop is and you will usually get one of three answers.

• The gut answer. "We're slammed." True, unquantified, and useless for deciding whether one more order fits.
• The spreadsheet answer. A load sheet that was correct on the Tuesday somebody last updated it, living in one person's file, with three conflicting copies in circulation.
• The floor answer. You walk the shop and see where the carts are stacked. Accurate, and weeks too late.

None of these are laziness. They are what happens when the plan and the shop's real hours were never connected. The routing knows how long each operation takes. The calendar knows how many hours the cell runs. Nothing put the two together, so the promise date was a guess the moment it left your mouth.

The overload was always in the data. It just got discovered on the shop floor instead of in planning.

The consequences are boringly predictable and expensive. Optimistic dates become late shipments. Padded dates lose quotes to faster competitors. And every corrective option left once the overload arrives, overtime, rush subcontracting, resequencing, is the expensive version of a decision you could have made calmly six weeks earlier.

What work center capacity planning actually means

Strip the jargon and it is a single comparison, repeated for every work center and every time bucket:

Hours of work already booked, against hours the work center actually has.

Booked over available gives you a utilization percentage. Under 80% and there is room. Between 80 and 100% it is tight. Over 100% and you have promised more work than the cell can physically run, which means something is going to be late. You just do not know what yet.

The whole discipline is getting those two numbers right, and getting them in front of a human being early enough to matter.

In Qualis the chain runs left to right. A calendar says when a work center runs. A routing says how long each operation takes. Releasing a manufacturing order turns that routing into dated work orders with real hours attached. The load board adds those hours up per bucket and paints the result.

Nobody types a load figure anywhere. It is a by-product of work you already do.

Where your real available hours come from

Here is where most capacity numbers go wrong before they start. "The cell has 40 hours a week" ignores everything that makes those hours real: the second shift, the overnight shift, how many people are on each one, how the cell actually performs against standard, and the plant shutdown in week 34.

In Qualis, availability lives on a work calendar attached to the work center. A calendar carries its working days, one or more shifts with start and end times, a crew size and an efficiency percentage per shift, and its holidays.

Read the summary at the bottom of that page and the arithmetic is right there. Five working days, three shifts, 24 hours a day of shift coverage, so 120 gross hours a week. Twelve people across the three shifts at an average of 88.67% efficiency, so 430.4 effective hours a week of labour capacity.

Those are two genuinely different numbers, and which one counts depends on what constrains the cell.

Every work center declares a capacity basis:

• Machine basis. The constraint is the equipment. Available hours are the gross shift hours, because a second operator does not give you a second spindle.
• Labor basis. The constraint is the people. Available hours scale with crew size and efficiency, because two welders really do give you two welders' worth of hours.

A shift that runs past midnight is handled as an overnight shift rather than as a negative number, and a work center with no calendar of its own falls back to the organization default. That is a small detail with a large consequence: it is the difference between planning against hours you have and hours you wish you had.

What lands on the board when you release an order

A manufacturing order sitting in planning does not load anything. The moment you release it, Qualis freezes its bill of material and routing, spawns one work order per operation, and schedules them in sequence against the work calendars.

Each work order comes out with a planned start, a planned end, and the capacity minutes it will consume. The arithmetic follows the way shops actually quote work:

• Setup time is per batch. Running 300 pieces does not set the machine up 300 times.
• Run and finish time are per unit, scaled by the quantity and stretched by the operation's efficiency.
• Queue and move time consume calendar time but never capacity. A part waiting in front of a machine does not occupy the machine. Getting this wrong is how manual load sheets inflate a bottleneck that was never there.
• Outside processing consumes elapsed time and zero internal capacity. Plating and heat treat take two weeks off your delivery date without taking a single hour off your own shop. On a spreadsheet that operation is almost always booked as one or the other, never correctly as both.

Two settings decide how the chain gets placed on the calendar.

Scheduling direction decides the anchor. Forward walks the operations onward from the planned start date, which answers "if I start now, when is it done?". Backward walks them back from the promised end date, which answers "when must I start to hit the 12th?". Job shops usually live forward. Make-to-order shops working to a customer date usually live backward.

Finite capacity scheduling decides whether the shop's existing commitments push back. Leave it off and each order is scheduled as if the cell were empty, which is fast, simple, and lets you see the overload. Turn it on and each day is capped by the capacity already committed to other open work orders, so a job that does not fit today spills into tomorrow instead of pretending both fit.

Reading the load board

This is the screen the whole chain exists to produce.

Rows are work centers. Columns are time buckets. Every cell reads scheduled / available h, with the utilization percentage on hover, and the colour tells you the verdict before you have read a single number: green under 80%, amber from 80 to 100%, red above 100%.

A few deliberate choices make it usable rather than merely correct:

• The worst rows sort to the top. The busiest work centers are the ones you came to find. In the board above, Deburring and Finishing is carrying 114 hours into a 45 hour week, and it is the first thing you see.
• You pick the bucket size. Day, week or month, over a horizon you choose: up to 60 days, 26 weeks or 18 months. Day buckets answer "can we run this Tuesday?". Month buckets answer "do we need to hire?".
• Months step on real calendar boundaries, so a horizon that crosses February does not quietly drift.
• An order spanning three weeks is spread across those three weeks, in proportion to how much of it falls in each, rather than dumped entirely into the week it starts.
• A totals row gives you the same picture for the whole shop, which is the number to look at before you agree to a rush order at all.
• Hours booked onto a work center with no capacity behind it show red, not blank. A cell nobody can plan is the worst case a planner can face, not a missing value, so it is never allowed to look harmless.

Clicking the red cell

A red cell tells you there is a problem. The next question is always the same: which jobs?

Click the cell and you get the work orders behind the number, each with its manufacturing order, its operation and its hours, ranked so the biggest contributor is obvious. Two orders out of three are carrying 92 of those 114 hours. That is the conversation: move one of them a week, and the cell goes from impossible to tight.

This is the step that turns a report into a decision. Level the load, add a shift, subcontract the operation, or call the customer while there is still time for the answer to be "the 19th" instead of an apology.

Why this is worth having

The honest case for capacity visibility is not that it magically creates hours. It is that it moves every decision earlier, when the cheap options are still on the table.

• Promise dates stop being guesses. You answer the 4pm question by looking at the week in question instead of padding by instinct.
• The bottleneck stops being a rumour. "Welding is our constraint" becomes "welding is 52 hours over in week 38", which is the sentence that justifies a second welder or a second shift.
• Overtime and subcontracting get planned, not panicked. Rush rates negotiated six weeks out are simply cheaper than rush rates negotiated on Friday afternoon.
• The schedule keeps its credibility. A plan that quietly triple-books one machine gets ignored by the floor, and once supervisors go back to the hot list, planning is overhead that produces nothing anyone uses.

It is worth knowing how the rest of the market handles this, because the answers differ more than you would expect:

• Standard NetSuite manufacturing schedules at infinite capacity. Work center calendars set the available hours and scheduling uses them, but the resulting load is shown for visibility rather than enforced. Finite scheduling and the ready-made work center capacity and utilization reports arrive with the Advanced Manufacturing add-on.
• Infor CloudSuite Industrial ships MRP and infinite planning as standard. The constraint-based finite mode, and the resource load profiles that read utilization per period, are separately licensed.
• Epicor Kinetic handles finite capacity in its core scheduling engine, resource by resource, with load-versus-capacity visibility. The visual scheduling boards and multi-constraint scheduling come with its Advanced Planning and Scheduling add-on.
• Standard SAP Business One puts capacity on resources rather than work centers, as a quantity per day built from numeric factors, and shows load as a day-by-day grid of internal, committed, consumed and available quantities. Overload is not blocked: it surfaces as a negative available quantity, and you fix it by editing production order dates by hand.
• Odoo's standard work center carries working hours, a time efficiency multiplier and a load figure, and its planning views give you a Gantt of scheduled work orders. Shifts and crew size are not attributes of the work center: the usual pattern is a working-hours schedule per shift, and a duplicate work center for each shift to maintain alongside it.
• Katana models production capacity as a single company-wide weekly throughput figure, spread evenly across the weekdays, so a bottleneck machine and an idle one draw from the same shared pool.
• MRPeasy, to its credit, schedules against finite workstation capacity on every plan, including the entry tier, with per-workstation loading charts.
• Fishbowl schedules work orders by picking a date on a shared calendar, so clashes are something you spot by eye.

The pattern worth naming is that capacity-aware scheduling is routinely the upgrade rather than the baseline. The shop that most needs to see an overloaded work center is often the one that has not bought the ability to see it.

Frequently Asked Questions

How do you calculate a work center's available capacity in hours?

Start with the shift pattern: the length of each shift multiplied by the number of working days in the period. That is your gross, or machine, capacity. For a work center constrained by people rather than equipment, multiply each shift by its crew size and its efficiency percentage before adding the shifts together. Then subtract the holidays. In Qualis all of this comes off the work calendar attached to the work center, so the figure updates itself when the shift pattern changes.

What is the difference between finite and infinite capacity scheduling?

Infinite capacity scheduling places each order as if the work center were empty, so three jobs can be scheduled on the same machine in the same hour and nothing objects. Finite capacity scheduling checks what is already committed and pushes work out when a period is full. Infinite is faster and shows you the overload honestly on a load board; finite produces dates that respect the shop's real limits. Qualis defaults to infinite and lets you switch finite capacity on per organization, so each day is capped by the capacity already committed to other open work orders.

Should I use forward or backward scheduling for my work orders?

Forward scheduling starts from the planned start date and walks the operations onward, answering "if I start now, when will it be finished?". Backward scheduling starts from the promised end date and walks the operations back, answering "when do I have to start to hit that date?". Shops that run to a customer delivery date generally prefer backward, because it exposes an order that needed to start last Tuesday. Shops loading a queue generally prefer forward. In Qualis it is a single organization setting, and an order with no planned end date is scheduled forward regardless, because there is nothing to work back from.

Does crew size double a work center's capacity, or only its cost?

Both, but not in the same place, and this trips people up constantly. Crew size raises the available hours on a labour-basis work center: four welders on a shift genuinely give you four welders' worth of hours in the week. What crew size does not do is halve the duration of a single operation, because putting a second person on a job does not usually make that job finish in half the time. On a machine-basis work center crew size does not change available hours at all, since the constraint is the equipment.

How do subcontracted operations count against internal capacity?

They should consume elapsed time and zero internal capacity, which is exactly how Qualis treats an operation flagged as outside processing. Plating that takes ten days pushes every downstream operation ten days later, but it does not book a single hour against one of your own work centers. Booking it as internal capacity blocks a cell that is actually free; ignoring the elapsed time makes the job look ready two weeks before the parts come back. Both errors are routine on a manual load sheet, and both produce dates you cannot hit.

Is a work center running at 95% utilization a good thing?

Not usually, and this is the most counter-intuitive part of capacity planning. Queue time rises steeply as a work center approaches full utilization, which is a standard result in queueing theory: the fuller the cell, the longer everything waits in front of it, and the relationship is not linear. Pushing a constraint towards 100% looks like excellent asset management on a utilization report while quietly stretching your lead times. A load board that shows the percentage per bucket at least lets you make that trade deliberately rather than discovering it in your delivery performance.

The bottom line

Capacity planning is not a forecasting problem. The hours are already sitting in your routings and your shift patterns. They just need to be added up per work center, per week, and shown to someone before the promise is made rather than after it is broken.

That is what the Qualis work center load board does. Calendars carry the real hours, including shifts, crew and efficiency. Releasing an order turns its routing into dated work orders with real capacity minutes. The board compares the two per day, week or month, sorts your worst problems to the top, and lets you click any cell to see exactly which jobs are causing it.

If you want to see it against your own shop, start a trial of Qualis and set up one work center with its real shift pattern. The first week you look at that board, you will find something you did not know.

Related reading: Standard Cost Rollup: Know What a Product Really Costs Before You Quote It, which uses the same routing times to build a cost rather than a schedule, and Available to Promise: How to Stop Overselling Stock You Have Already Committed, the material-side twin of this problem.