Introduction
A plant manager once asked his team to increase output on a line by 20 percent.
The team asked for a new machine. The manager asked them to prove they needed one.
Two weeks of measurement later they found something uncomfortable. The line was running at 61 percent OEE. The bottleneck station was idle for eleven minutes every changeover waiting for a tool that lived in another building. And one operation was running at a feed rate set five years earlier for a different material.
They got the 20 percent without buying anything.
That is process optimisation. Not new equipment, not more people, but finding what the process is already losing and taking it back.
This guide covers the techniques that actually work, in the order you would use them.
What Is Manufacturing Process Optimization?
Manufacturing process optimization is the systematic improvement of a process to increase output, quality or efficiency while reducing cost, time and waste.
The important word is systematic.
Anyone can change a parameter and hope. Optimisation means measuring the current state, identifying the real constraint, changing one thing deliberately, and confirming the result.
What you are usually optimising for
Throughput, meaning more output from the same resources.
Quality, meaning fewer defects and less rework.
Cost, meaning less material, energy, labour and tooling per part.
Lead time, meaning faster delivery to the customer.
Flexibility, meaning faster changeover between products.
The trade off you must understand
These objectives conflict. Running faster often increases defects. Reducing inventory increases sensitivity to disruption. Tighter tolerances raise cost.
Optimisation is not maximising everything. It is finding the best balance for what the business actually needs.
Students who miss that point end up proposing changes that improve one number and quietly damage two others.
Why Optimization Matters More Than New Equipment
The instinct when a line cannot meet demand is to ask for capital.
Why that is usually the wrong first move
Most plants are not running at their existing capacity. A typical OEE of 60 percent means 40 percent of available capacity is already being lost. Buying a second machine to run at 60 percent is expensive.
New equipment hides the problem. If changeovers take two hours, a faster machine still loses two hours.
Capital is slow. Approval, procurement, installation and commissioning take months. Optimisation projects deliver in weeks.
Optimisation is cheap. Most improvements come from method, sequence and parameter changes rather than purchases.
The argument that works in a real plant
“Before we buy a machine, let me show you what we are losing on the one we have.”
That sentence has funded more engineering careers than any technical skill.
Step by Step Approach to Process Optimization
Techniques matter less than sequence. Applying a good technique to the wrong part of the process wastes effort.
Step 1: Define the objective. Are you chasing throughput, quality, cost or lead time? Pick one primary target, because optimising for everything optimises for nothing.
Step 2: Measure the current state. Cycle times per station, downtime with reasons, scrap rate, changeover time and OEE. Without a baseline you cannot prove improvement.
Step 3: Map the process. Draw it as it actually runs, not as the procedure says. Walk it. The two are rarely identical.
Step 4: Find the constraint. Every process has one bottleneck that limits everything else. Improving anything else changes nothing.
Step 5: Analyse the losses at the constraint. Break the loss into categories. Downtime, slow running, changeover, quality, waiting.
Step 6: Generate improvement options. Involve the operators. They usually know what is wrong before you do.
Step 7: Trial one change at a time. Changing three things together tells you nothing about which one worked.
Step 8: Measure the result. Compare against the baseline honestly, including any side effects on quality or other stations.
Step 9: Standardise. Update the work instruction, control plan and training. Improvements that are not documented disappear within months.
Step 10: Move to the next constraint. Once you fix the bottleneck, it moves somewhere else. Optimisation is a loop, not a project with an end.
Bottleneck Analysis and the Theory of Constraints
This is the most important optimisation concept and the one students most often skip.
The core idea
Every process has one step that limits total output. That step is the bottleneck or constraint.
The consequence that surprises people
Improving any non bottleneck station does not increase output at all. It only creates more work in process piling up in front of the bottleneck.
How to find the bottleneck
- The station with the longest cycle time
- The station with material queued in front of it
- The station that is never starved of work
- The station where overtime is always required
The five focusing steps of the Theory of Constraints
1. Identify the constraint.
2. Exploit it, meaning get maximum output from it without spending money. Do not let it stop for lunch breaks, tool changes that could happen offline, or waiting for material.
3. Subordinate everything else to it. Other stations should run at the pace the constraint can absorb, not at their own maximum.
4. Elevate the constraint, meaning add capacity through investment if the first three steps are exhausted.
5. Repeat, because once fixed, the constraint moves elsewhere.
The practical insight worth remembering
An hour lost at the bottleneck is an hour lost for the entire plant. An hour lost at a non bottleneck costs nothing.
That single sentence changes how you prioritise maintenance, changeovers and operator attention.
Cycle Time Reduction Techniques
Break the cycle into elements first. Loading, positioning, cutting, unloading, measuring, waiting. You cannot reduce what you have not separated.
The insight that surprises students
In most machining operations, the actual cutting is a minority of the cycle. Loading, unloading, tool changes and waiting usually dominate.
Techniques that work
Reduce non cutting time. Faster clamping, better fixture design, pre positioning of tools and material.
Run operations in parallel where possible. Load one part while machining another using a pallet changer or a two station fixture.
Optimise cutting parameters. Feed and speed set five years ago for a different material or tool grade are extremely common.
Reduce tool changes through longer lasting tool grades and better coolant.
Combine operations. Doing two features in one setup removes a handling cycle entirely.
Improve tool paths in CAM, particularly rapid moves and air cutting, which consume time and remove nothing.
Reduce inspection time by moving from full measurement to sampling once capability is proven.
The rule of thumb
Attack the largest element first, and it is usually not the cutting.
Setup and Changeover Reduction with SMED
Why changeover matters more than students expect
A line that loses 90 minutes per changeover with three changeovers a day loses over four hours of production. That is often larger than every other loss combined.
SMED means Single Minute Exchange of Die, a structured method developed by Shigeo Shingo.
The core distinction
Internal setup is work that can only be done while the machine is stopped.
External setup is work that can be done while the machine is still running.
The SMED method in four steps
Step 1: Observe and record the current changeover. Film it. Almost every team is surprised by what they see.
Step 2: Separate internal from external. Fetching tools, preparing fixtures, gathering paperwork and pre heating dies are all external activities that are usually done internally.
Step 3: Convert internal to external. Prepare and stage everything before the machine stops.
Step 4: Streamline what remains. Quick release clamps instead of bolts, standardised die heights, colour coded settings, and eliminating adjustment through positive location.
The classic result
Changeovers routinely drop from hours to minutes without any capital spending, simply by reorganising the sequence.
Why it enables everything else
Short changeovers make small batches economical, which reduces inventory, shortens lead time and improves flexibility. SMED is often the single highest leverage optimisation technique available.
Line Balancing and Layout Optimization
The problem line balancing solves
If one station takes 90 seconds and the others take 40, everyone waits for that station and output is set by the slowest one.
Key terms you must know
Takt time is the pace required to meet customer demand, calculated as available time divided by demand.
Cycle time is how long each station actually takes.
Line balance efficiency is the total work content divided by the number of stations multiplied by the longest station time.
How to balance a line
List every task and its time. Determine takt time from demand. Assign tasks to stations so each is as close to takt time as possible without exceeding it, respecting the sequence that tasks must follow.
Practical balancing techniques
Move a task from the overloaded station to a lighter neighbouring one.
Split a long task into two if it can be divided.
Add a parallel station only for the bottleneck operation rather than duplicating the whole line.
Change the method so the task takes less time.
Layout optimisation
Cellular layout groups machines by product family rather than by machine type, which cuts transport and waiting dramatically.
U shaped cells allow one operator to handle several machines and shorten walking distance.
Point of use storage puts material where it is consumed rather than in a central store.
Spaghetti diagrams, drawing the actual path of a part or operator on the layout, reveal transport waste that is invisible from a desk.
Automation and Technology Based Optimization
Technology is a valid optimisation route, but it belongs after the method improvements, not before.
When automation genuinely helps
Repetitive, high volume tasks with consistent input.
Ergonomically difficult or unsafe operations.
Operations requiring precision beyond human consistency.
Tasks where labour is genuinely unavailable.
When automation disappoints
When it is applied to a process that has not been simplified first. Automating a wasteful process produces waste at higher speed with better documentation.
Technology options in rough order of investment
Sensors and data collection, which cost little and reveal where the losses actually are.
Poka yoke devices, simple mechanical or sensor based error prevention.
Semi automation, such as automatic loading with manual operation.
Robotics for handling, palletising and welding.
Machine monitoring and predictive maintenance to reduce unplanned downtime.
Digital twins and simulation for testing layout and sequence changes before implementing them.
The sequence that works
Simplify, then standardise, then automate.
Automating before simplifying locks the waste permanently into an expensive machine.
Quality Focused Optimization
Optimising for speed while ignoring quality moves the problem rather than solving it.
Statistical process control to detect drift before defects are produced, and to distinguish normal variation from a real signal that needs action.
Process capability studies to confirm the process can actually hold the tolerance. If Cpk is below 1.33, no amount of inspection will make the output reliable.
Poka yoke to make errors impossible rather than detectable.
Root cause analysis on the top defect from a Pareto chart, rather than reacting to whichever defect occurred most recently.
PFMEA to identify risks before production begins.
Reduce inspection through capability. Once a process is proven capable and stable, inspection frequency can be safely reduced, which itself removes cycle time and cost.
The relationship worth understanding
Quality improvement is often throughput improvement in disguise. Eliminating a 4 percent scrap rate increases good output by roughly 4 percent without changing the cycle time at all.
Common Mistakes in Process Optimization
Optimising a non bottleneck. The most common and most wasteful error. It produces impressive local numbers and no change in output.
Changing several things at once. You cannot attribute the result, and if it fails you do not know why.
No baseline. Without a before measurement, you cannot prove improvement, and someone will eventually question whether anything changed.
Ignoring the operators. They know where the process actually loses time, and they will quietly work around a change they were not consulted about.
Optimising for one metric only. Faster cycle time with higher scrap is not an improvement.
Buying equipment first. Capital before method locks in the existing inefficiency.
Not standardising. An improvement that is not documented and trained fades within months.
Assuming the standard is correct. Cycle times, feed rates and standard times are frequently years out of date.
Skipping the measurement system check. If the gauge is unreliable, your data is fiction.
What Students Should Learn About Optimization
Concepts to understand properly
Bottleneck and Theory of Constraints, since it determines where effort should go.
Takt time versus cycle time, and line balance efficiency.
Internal versus external setup, which is the whole basis of SMED.
DOE basics, particularly why one factor at a time misses interactions.
Skills to build
Time study. Break an operation into elements and time them. This is the foundation of every optimisation project and requires nothing but a stopwatch.
Bottleneck identification. Given cycle times for five stations, say which is the constraint and why.
Basic DOE, at least a two factor two level design run in Minitab or a spreadsheet.
Payback calculation, because every proposal needs a financial case.
A project worth doing before you graduate
Take any multi step process you can observe, even a college workshop operation or a canteen queue. Time each step. Identify the bottleneck. Propose one change. Estimate the improvement.
That exercise teaches bottleneck thinking better than any chapter, and it gives you a concrete example to describe when an interviewer asks how you approach improvement.
Frequently Asked Questions (FAQs)
1. What is manufacturing process optimization?
It is the systematic improvement of a process to increase output, quality or efficiency while reducing cost, time and waste.
It involves measuring the current state, finding the constraint, making deliberate changes and verifying the result.
2. What is the first step in optimising a process?
Defining the objective and measuring the current state.
Without a baseline you cannot identify the real problem or prove that anything improved.
3. What is a bottleneck and why does it matter?
It is the step that limits the output of the whole process.
Improving any other step does not increase output, which is why identifying the bottleneck first is essential.
4. What is SMED?
Single Minute Exchange of Die, a method for reducing changeover time.
It works by separating internal setup done while the machine is stopped from external setup done while it is running, then converting internal activities to external.
5. What is the difference between takt time and cycle time?
Takt time is the pace required to meet customer demand.
Cycle time is how long the process actually takes.
If cycle time exceeds takt time, demand cannot be met.
6. What is Design of Experiments?
A structured method of varying several factors at once in a planned pattern, so you can determine the effect of each factor and their interactions from a small number of trials.
7. What is the difference between DOE and Taguchi methods?
DOE is the general approach to planned experimentation.
Taguchi methods are a specific approach focused on robustness, using orthogonal arrays and signal to noise ratios to find settings that perform consistently despite uncontrollable variation.
8. Should I automate to optimise a process?
Only after simplifying and standardising it.
Automating a wasteful process produces waste faster and locks the inefficiency into expensive equipment.
9. What is the most common mistake in process optimisation?
Improving a station that is not the bottleneck.
It produces better local numbers and no increase in overall output.
10. How do I prove an optimisation project worked?
Measure the same metrics before and after, check for side effects on quality and other stations, and express the result in both operational and financial terms.
Conclusion
Process optimisation is less about clever techniques and more about looking in the right place.
Three principles that matter most
Find the bottleneck first, because improving anything else changes nothing.
Measure before you change, because without a baseline nothing can be proven.
Change one thing at a time, because otherwise you learn nothing from the result.
Two techniques worth learning properly
SMED, because changeover time is often the largest single loss and it costs nothing to reduce.
Time study, because every optimisation project starts with breaking an operation into elements.
One thing worth carrying into your career
The instinct in most plants is to ask for a new machine. The engineer who instead says “let me show you what we are already losing on the machine we have” becomes valuable very quickly.
Most factories are running well below their existing capacity. Finding that lost capacity costs almost nothing, and it is the most reliable way an early career engineer can prove their worth.