The Capacity You Already Have But Cannot See

The cheapest capacity in your plant is the capacity you already own and cannot see. It is hours consumed by avoidable setups, lost to jobs queued behind a nominal machine while an equal one sits idle, and spent building parts you already had in stock. None of it needs a purchase order. EDGEBIC by User Solutions finds those hours by scheduling better, which is why the gain shows up without a capital line.

This post names the three places existing capacity hides, shows the arithmetic of recovering it, and is honest about when found capacity is real throughput and when it is only slack. For the related question of what you are losing, see how much capacity are you already losing. This post sits under the EDGEBIC results guide and focuses on the capacity you own but do not see.

Why the Cheapest Capacity Is Invisible

When demand exceeds what a plant seems able to make, the instinct is to add: a machine, a shift, a hire. Those are real options, and sometimes the right ones. But they are also the most expensive answers, and shops reach for them before checking whether the hours they need are already inside the building, spent on things that produce nothing.

Existing capacity hides because the waste is distributed and unlabeled. No report line says "hours lost to bad sequencing." The setup time looks like normal machine time. The job queued behind a busy machine looks like a scheduling accident, not a systemic leak. The part built to stock that was already on the shelf looks like normal production. Add them up across a quarter and they are often the size of the shift you were about to add, but you cannot add them up until something makes the waste visible, which is what a finite capacity schedule does.

Leak One: Hours Spent Changing Over Instead of Running

The largest hidden reservoir on many machines is setup. A machine that is never idle is not necessarily always producing: on a resource with heavy changeovers, a third of its busy hours can be setup, not run time. And much of that setup is avoidable, because it is sequence-dependent (the time depends on what ran last), so running the jobs in a smarter order shrinks it.

EDGEBIC charges the true changeover for each transition and can sequence to avoid the expensive ones. The paint-booth worked example is the clean demonstration: the same three jobs carried 330 minutes of setup in due-date order and 90 minutes sequenced like-to-like, a 73 percent reduction, freeing about four hours on a single booth in a single day. See the throughput gain from cutting setups for the full arithmetic and how EDGEBIC cuts changeover hours for the mechanism. Those four hours were always there, inside the machine, disguised as setup. Sequencing turns them back into run time.

Leak Two: Jobs Queued Behind a Nominal Machine

The second reservoir is contention that is not real. A job stacks up in front of one work center while an interchangeable one sits idle, because the plan pinned the work to a specific machine rather than to any machine that can do it. The queue looks like a capacity shortage; it is actually a placement mistake.

EDGEBIC can treat interchangeable machines as a pool and balance work across them, so a job flows to whichever equivalent resource is free instead of waiting behind a busy one. See how finite capacity scheduling handles a shared bottleneck. The capacity recovered here is the idle time on the equal machine that the old plan left stranded. It was paid for and available; the schedule simply was not routing to it.

Leak Three: Building What You Already Have

The third reservoir is the job that never needed to run. Where a finished item is stocked, netting demand against on-hand balance means you build only the shortfall, and the hours the covered order would have consumed stay available for work that actually needs them.

The documented netting case: 80 units on hand, a 60-unit order, and a 30-unit order behind it. The 60-unit order is covered entirely from stock and never becomes a scheduled job, and the 30-unit order manufactures only the 10-unit balance. See how EDGEBIC nets demand against stock. The recovered capacity is blunt: an order that produced zero hours of work because it produced zero parts, freeing all the machine time it would have consumed.

The Heritage Scale

Put the leaks together on the constraint and the aggregate is documented. In the User Solutions track record, Technical Glass Products gained about 4 percent capacity and cut lead time by two weeks by scheduling this way, without adding equipment. Four percent may sound modest until you place it on a bottleneck running flat out: it is throughput the plant did not have the day before, found inside machines that already looked fully loaded. The two-week lead time cut is the same recovered hours seen from the customer's side, as jobs stopped waiting behind avoidable waste.

The Homestead Furniture record adds the other half of the picture: the visibility that finds the leaks in the first place. Their gains came not only from the labor saving but from finally seeing the bottleneck and the wasted hours around it, which is the precondition for recovering them. You cannot reclaim capacity you cannot see, and making it visible is the first thing a finite schedule does.

What the Software Cannot Do Alone

Three honest limits keep found capacity honest.

Found off the constraint is slack, not output. This is the limit that trips up most shops. Recover four hours on a machine that is not your bottleneck and you get idle time, not shipments, because the real constraint still caps output. The capacity gain becomes throughput only on the resource that gates the plant, which means the first job is knowing which resource that is. See production bottleneck identification.

No demand means no gain this quarter. Recovered hours become product only if there is demand to fill them. Free capacity on a constraint with an empty queue and you have flexibility for the future, which is worth something, but not this quarter's throughput. The gain assumes work is waiting to use the hours.

Found capacity is bounded, not infinite. Scheduling recovers the hours that waste was consuming; it does not conjure hours that never existed. If your genuine demand exceeds your real capacity even after the leaks are closed, the honest answer is still the expensive one: add a shift, add a machine, or requote. What the schedule guarantees is that you make that decision after you have recovered the free hours, not before, so you buy capacity only when you have proven you need it. That sequence, recover first and buy second, is often the difference between a capital request and a scheduling change.

The machine you were about to buy may be the one you already have, run in a better order and routed more intelligently. The only way to know is to make the hidden hours visible and add them up. To see the recovery mechanisms in full, read the throughput gain from cutting setups and how EDGEBIC raises bottleneck utilization; to see the whole set of results, start from the EDGEBIC results guide; and to see EDGEBIC itself, visit the product page.