Before adding capacity, change the control logic: finding what really governs output

When an operation struggles to keep up, the familiar response is to ask for more capacity: another machine, another line, a bigger compressor, more storage, another shift. Sometimes that is exactly right. But new capacity is also one of the most expensive ways to discover that the real problem was not physical capacity at all.

Existing equipment can underperform because of poor sequencing, demand peaks when several machines start together, rigid control settings, unnecessary waiting, frequent changeovers, or rules that made sense once and no longer do. If those causes are not understood, new capital simply enlarges an inefficient system, and adds years of maintenance, energy, spares and staffing obligations along the way.

This article explains why capacity is partly a property of how equipment is controlled, how to identify the variable or rule that actually governs output, why more of a good thing eventually stops helping, how product sequence can drive both lost time and resource use, and a staged review that tests operating changes before committing to new equipment.

Four kinds of capacity

It helps to separate four ideas that are often blurred together:

Physical capacity: what the installed equipment could theoretically process.
Effective capacity: what it reliably delivers under real conditions, including changeovers, breakdowns and variation.
Economic capacity: what it should deliver once energy, quality, maintenance, labour and risk are considered.
Strategic capacity: what the business needs to support its future plans.

A business can have spare physical capacity and still lack effective capacity. It can have enough effective capacity today and not enough strategic capacity for the future. These are different problems needing different responses. Improving control and scheduling raises effective capacity toward physical capacity. Only when that gap is small does new equipment become the obvious answer.

What engineering research shows

Studies from very different technical fields point to the same principle. Research on large groups of air-conditioning units found that letting units delay switching, rather than all responding at once to a change in settings, could cut the combined power peak during transitions to less than half in simulation, without reducing comfort. Research on scheduling crude oil movements inside a refinery reduced energy use in its case study by changing when and how existing pumps were used, rather than changing the pipework. Research on wastewater treatment found that changing the aeration pattern of an existing process improved nutrient removal in a pilot while using less aeration energy. And manufacturing research by Gould and colleagues showed that the order in which products are made can change how much cleaning, and therefore time, water and energy, changeovers require.

The technologies differ. The pattern is consistent: performance can change materially when timing, sequence and control rules change, even when the equipment stays the same.

Common misreadings

Utilisation equals productivity. A pump running flat out may use disproportionately more energy. Machines all starting together may create a peak none would create alone. An asset running harder is not necessarily creating more value.
The symptom shows the cause. Queues, missed output, high energy use and overtime all look like a lack of capacity. The cause may be variation, batch sizes, scheduling, changeovers, maintenance timing or information delays.
Teams can optimise without being told what for. A schedule can be optimised for throughput, energy, inventory, changeovers or on-time delivery, and these are not the same. Unless leaders state which matters most, teams optimise whatever they are measured on.

Four kinds of governing constraint

The governing constraint is the variable, rule or condition that currently limits the outcome you care about. It usually takes one of four forms:

Physical: a hard limit set by equipment size, flow, heat transfer, strength, residence time or power.
Interface: a loss where two parts of the system meet, such as a handover between shifts, a poor fit between two machines or a mismatch between a supplier’s output and your input requirements.
Policy: a rule that restricts the system even though the equipment could do more, such as batch sizes, shift patterns, approval levels or scheduling conventions.
Information: a decision delayed or degraded because the right information is missing, wrong or late.

The remedies differ. Adding equipment does not fix an information constraint. Training does not remove a physical limit. Improving one component does not fix a weak interface. When every part looks capable but the whole still underperforms, look at the interfaces and policies before replacing parts.

Where more stops helping

Ask how the outcome responds as a key variable changes. The shape of that response is valuable information:

A steep rise means a small change has high leverage.
A flat curve means more effort or investment is wasted.
A rise then a fall means pushing too hard damages performance.
A shifting curve means the best setting depends on conditions, so adaptive control may beat a fixed target.

Research on removing arsenic from groundwater in a biological treatment process illustrates the second pattern: longer retention time improved removal over an initial range, then additional time produced little further benefit. The specific numbers belong to that study. The management lesson is general: find the point where the response flattens, and stop investing past it.

Sequence is a control variable

In many operations, switching from one product, recipe, colour or material to another consumes time and resources. Where the cost of a changeover depends on what came before and what comes next, the schedule becomes an economic and environmental control:

Food production may need a full clean between allergen groups but only a quick clean between similar products.
Paint and coating lines need more purging when moving from dark to light colours than the reverse.
Chemical and beverage processes need flushing between incompatible recipes.
Heat treatment uses energy to move between temperature settings.
Packaging lines generate set-up scrap at each changeover.

A schedule that ignores these effects can waste hours of capacity and large quantities of water, energy and materials each week, without anyone seeing it as a capacity problem. Grouping and ordering work to minimise expensive transitions can recover capacity without new equipment. Because order mix changes, the best sequence is not fixed. It needs rules that adapt, not a single optimal answer.

Watch the peaks

Large gaps between peak and average demand often signal a timing problem rather than a capacity problem. Machines that start together after breaks, cleaning cycles that run at the same time, or deliveries that all arrive at once can create peaks that trip equipment, starve other users or drive up costs. Many business electricity tariffs include charges based on peak demand, so staggering start-ups can save money as well as capacity. Check your tariff structure with your electricity retailer.

Information constraints are easy to miss

Some of the most common constraints in small businesses are informational. A machine waits because the operator does not know which job is next. A batch is made too early because the planner cannot see current stock. A clean is longer than necessary because nobody recorded which product ran last. Material arrives late because purchasing did not know the schedule had changed. None of these looks like an information problem from the outside. They look like slow equipment or poor capacity.

Simple fixes often help: a visible schedule at the machine, a short daily planning conversation, a whiteboard showing the next three jobs and their materials, or a record of the last product run. Before investing in equipment or software, check whether the people running the operation have the information they need, when they need it.

Make operating rules part of governance

Control settings, sequencing rules, batch sizes and start-up routines are often set once by an engineer or an experienced operator and never reviewed. Treat them as business decisions with owners. Record the important rules, why they exist and when they were last reviewed, and review them when products, equipment or demand change. A rule that made sense for last year’s product mix may now be costing hours each week.

A control-before-capacity review

Before approving significant capacity, work through six stages:

Define the outcome in business terms, such as required output, service level, quality or an energy limit. Do not start with the proposed equipment.
Map the constraint using real data, not averages: peaks, idle time, blocking, starving, waiting, simultaneous starts, changeovers, control settings and maintenance windows.
Separate controllable from structural limits. Some constraints can change through scheduling or rules. Others are set by physics, safety, regulation or material properties.
Test operating alternatives: sequencing, dispatch rules, settings, batch sizes, priority rules, run windows, staggered starts. Judge them on output, energy, quality, wear, safety and customer effect.
Find the saturation point where further operating effort stops paying.
Rebuild the capital case, comparing no investment, control and scheduling changes, minor debottlenecking, modular additions, major expansion and different technology, on whole-of-life value.

A strong case for expansion will survive this review. A weak one will not, and the business will have saved its capital.

A worked example

This is an illustration. A small food manufacturer runs one mixing and filling line for five products in two allergen groups. It is about eight hours a week short of capacity, and the owner is considering a second line costing about $350,000.

The team maps a typical week. Products are run in order of customer due date, causing ten changeovers. Six are long cleans of about 90 minutes, needed when switching between allergen groups, and four are short cleans of about 20 minutes. Cleaning takes about 620 minutes a week, roughly 10.3 hours. Long cleans use about 1,200 litres of water each and short cleans about 300 litres, around 8,400 litres a week.

The team regroups production by allergen group, and orders products within each group from lightest to darkest, on a weekly cycle that still meets customer dates. There are still ten changeovers, but only two are long cleans. Cleaning time falls to about 340 minutes, about 5.7 hours, recovering about 4.7 hours a week. Water for cleaning falls to about 4,800 litres, a reduction of about 43%.

The team also finds that the mixer and air compressor start together after breaks, tripping the supply about twice a week and losing about 45 minutes each time. Staggering start-ups recovers about 1.5 hours a week.

Together, these changes recover about 6.2 of the 8 hours needed. A trial of slightly larger batches on two high-volume products recovers the rest, after checking that mixing quality is unaffected. The second line is deferred, with a trigger to revisit the decision if demand grows by more than about 15%.

How this applies to a small Australian business

Small businesses often have one or two key pieces of equipment and limited capital. Practical steps:

Measure where time goes on the constrained equipment for a few weeks.
List changeovers and what each costs in time, water, energy and scrap.
Look for sequence effects and group work to reduce expensive transitions.
Check for simultaneous starts and peaks, and review your electricity tariff.
Question rules such as batch sizes and shift patterns.
Find saturation points before investing to push a variable further.
Compare capital options on whole-of-life value once operating options are exhausted.

The articles on putting a dollar value on equipment losses and lean as an operating system cover related methods.

Signals worth watching

Capacity requests triggered by peaks rather than sustained load.
High utilisation on one machine with waiting elsewhere.
Frequent simultaneous starts and trips.
Energy or water use rising with product mix rather than volume.
Large differences between similar sites or shifts.
Schedules that depend on a few experienced planners.
Control settings unchanged since the process changed.
Expansion proposed before the constraint has been measured.

Common mistakes

Buying capacity before measuring effective capacity.
Treating utilisation as success.
Ignoring sequence-dependent changeovers.
Optimising without a stated objective.
Pushing a variable past its saturation point.
Replacing components when the problem is an interface or a rule.

Frequently asked questions

When is new capacity the right answer? When the operation is already well controlled, the governing constraint is physical, and demand genuinely exceeds what the existing equipment can deliver. The review in this article is designed to confirm that, not to prevent investment.

Do we need scheduling software? Not necessarily. Many sequencing gains come from simple rules, such as grouping by allergen or colour, agreed with the people who run the line. Software helps when product mix is large and changes often.

How do we find the saturation point? Change the variable in small steps and measure the result, or look at past data where the variable naturally varied. Where the outcome stops improving, further investment in that variable is unlikely to pay.

Questions to ask

What evidence shows our constraint is physical rather than scheduling, variation or control?
Which objective are we optimising: throughput, cost, energy, service or a defined combination?
What capacity could sequencing, settings or staggered starts release?
Where does more of a variable stop producing value?
What lifecycle obligations would new equipment add?
If we approve expansion, what would later tell us we invested too early?

Bringing it together

Capacity is partly a property of equipment and partly a property of how that equipment is scheduled, sequenced and controlled. Separate physical, effective, economic and strategic capacity, identify whether the governing constraint is physical, interface, policy or information, find where responses flatten, and treat sequence as a control variable for both time and resources. Then build the capital case. Understand the system, change the controllable logic, identify the remaining constraint, and invest with evidence.

Source: KEVOS notes, drawing on published engineering research on load switching, refinery scheduling, wastewater aeration, groundwater treatment and resource-efficient manufacturing scheduling. Examples and figures in this article are illustrations.