Why a fast laser sits idle: CAM, nesting, and planning as a hidden bottleneck ART-014

The laser is ready to cut, but there is nothing ready to run — that is not one problem

An operator is standing beside a perfectly functional laser. The previous sheet has already been removed, the machine is free, and there are no alarms. Yet the next cycle does not start.

The reason may always sound the same: “there is no work.” In reality, that phrase can hide very different situations. The drawing has not yet gone through CAM preparation into an NC program for the specific machine. The program exists, but the parts have not been grouped into a nest. The nest is ready, but the required sheet exists only in the inventory system and is physically somewhere else. The material is available, but nobody has decided which of five urgent jobs should run first. Or everything may be ready for cutting while bending is already overloaded, so accelerating laser output would only increase work in process.

To the machine, all of these causes look identical: it is not cutting. To a production manager, they are five different bottlenecks and five different classes of solution.

That is why a fast laser can have relatively low actual utilization even without technical faults. The shorter its own cutting cycle becomes, the more visible the speed of the surrounding organization becomes — especially how quickly it can make decisions and prepare the next job.

A job should become “ready to cut” before the machine becomes free

In many shops, the status “in production” is too broad. It does not answer a simple question: can the operator load the sheet and press Start right now without extra searching, phone calls, or data rework?

For a laser, a genuinely ready job usually means that several conditions are true at the same time:

• the current part geometry is approved;
• the technology program is prepared for the specific machine;
• the parts are combined into a usable nest;
• material, grade, thickness, and sheet format are known;
• the required material is actually available and reserved for this job;
• the dispatch priority is defined;
• it is clear where the parts go after cutting and whether the next operation can accept them.

If even one of these items is still open, the job is not fully ready, even if the customer order was accepted long ago.

This distinction becomes critical with a productive machine. A slower laser may spend long enough on one sheet for a programmer to prepare the next one in parallel. If the machine becomes twice as fast, the same office process can suddenly stop keeping up. The problem is not a shortage of orders but a shortage of jobs that are ready to start.

The first bottleneck: the program is not ready yet

A CAM bottleneck occurs before the operator ever sees the upcoming job on the machine.

Incoming geometry may need cleanup, contour checks, technology assignment, lead-ins and pierce points, micro-joints, marking, or other preparation. With non-standard product mixes, each new order may require substantial manual work. If source files arrive in different formats or contain errors, part of the programmer's time is spent not on CAM itself but on repairing input data.

Two different measures should not be confused here. A company may have a fast programmer and still regularly leave the laser without an NC program if jobs reach programming too late. Conversely, a program may take a long time to prepare without causing machine downtime if there is a sufficient queue of already prepared work.

For diagnosis, therefore, it is useful to measure not only “how many minutes the programmer spends on one part,” but the time from when a job is released for preparation to when it is actually ready to start. That interval includes waiting for drawings, clarifications, the queue in front of the programmer, repeat corrections, and programming itself.

If the laser systematically waits for this stage, the solution may be cleaner input-data standards, preparing programs earlier, technology templates, offline programming, balancing programmer workload, or removing repeated approval loops. Buying an MES by itself will not fix a drawing that is still not ready.

Nesting is not only about saving sheet — it is a decision about what should run now

Nesting is often treated as a mathematical task: place more parts on a sheet and improve material utilization. Production needs a broader decision.

To build an efficient combined nest, it can make sense to wait until more parts of the same grade and thickness accumulate. But the machine and the customer both operate in time. If an urgent order waits for hours merely to gain a few more percentage points of sheet fill, the material saving may be purchased at the cost of machine waiting or a missed due date.

The opposite extreme is also costly. If every order is released on a separate sheet as soon as it arrives, the laser may remain busy while the plant creates excess remnants, more material changeovers, and poor overall sheet yield.

Nesting therefore balances at least three objectives:

material utilization → order timeliness → continuity of machine operation.

That is why “there is no ready nest” does not always mean the nesting algorithm is weak. Sometimes the missing element is a rule for when to stop waiting for additional parts and release a plan that is already good enough for production.

Manufacturing-control systems explicitly connect nesting with due date, priority, and the moment a job is released, not only with geometric sheet utilization. The broader lesson applies regardless of software brand: the release rule is part of the production-flow technology.

The second kind of “no work”: the program exists, but the material is not at the machine

The inventory screen may show ten sheets of the required steel. That still does not mean the next sheet is ready to cut.

Some of that stock may already be reserved for other orders. A sheet may be in another warehouse, under another stack, have a different actual format or surface condition, be a remnant of the wrong geometry, or still be waiting for receiving. In the simplest case, the material physically exists, but the forklift has not yet received a task to bring it to the laser.

For the operator, all of these cases become waiting time. Meanwhile, the programmer can be fully productive and the production plan can look correct on paper.

It is therefore useful to separate “material available” into at least three states:

1. recorded in inventory; 2. allocated specifically to this job; 3. physically available at the required time in the loading area.

TRUMPF solutions link order planning separately to stock levels, material locations, and transport orders. AMADA likewise separates material management, standard sheets, and remnants from CAM itself. The production logic is important: having an NC file and having ready material are different readiness loops.

If material is the reason for waiting, CAM automation will not solve it. The plant needs reliable stock data, reservations, controlled remnant handling, known physical locations, and a rule for staging material before the machine needs it.

The third bottleneck: every job is “urgent,” so the queue keeps being rebuilt

Even when programs and material are ready, the laser can lose time because there is no stable decision about sequence.

The planner builds a queue for the shift. Twenty minutes later, a salesperson brings an urgent order. Then welding asks for two missing parts from another kit to be cut immediately. An hour later, the team discovers that the first urgent job does not have the required sheet. The nest is rebuilt, priorities are changed, and the operator waits for a new decision.

The problem is not necessarily that priorities change. In contract manufacturing, they often have to. The problem begins when every priority change destroys an already prepared queue without a visible cost and without one place that shows the current decision.

For diagnosis, it is useful to count:

• how many times per shift the sequence of already prepared jobs changes;
• how many prepared nests have to be recalculated;
• how much time passes between the end of one program and the actual choice of the next;
• what share of urgent insertions were genuinely necessary versus caused by late upstream planning.

If this loss is significant, the plant needs dispatching rules rather than “an even better CAM”: who may change the queue, under what criteria, and how shipment date, material readiness, and the status of downstream operations are considered.

A fast laser needs a buffer of ready decisions, not merely a backlog of orders

One of the simplest ways to see the problem is to look not at “all open orders” but only at jobs that can be launched without additional preparation.

If the operator calls the programmer or supervisor after every sheet to ask “what is next?”, the machine is effectively running without a buffer of ready decisions. Any office delay immediately becomes machine downtime.

A small queue of ready-to-cut jobs creates a time buffer. While the laser runs one program, the next jobs already have NC, material, and priority. If one job is unexpectedly blocked, the operator can switch to another without stopping the resource completely.

But the opposite extreme is also easy to reach. If too many parts are released “just in case,” the shop fills with work in process and priorities lose meaning. It is especially risky to cut parts long before bending, welding, or assembly is ready to receive them.

The target is therefore not the maximum possible queue in front of the laser. The target is a sufficient, controlled buffer of ready work that protects the machine from short upstream delays without creating uncontrolled downstream WIP.

For estimating how many machine hours are actually required and how much calendar reserve should remain, a separate real-world laser utilization model is useful.

Sometimes the correct decision is deliberately not to maximize laser utilization

For a local OEE or machine-utilization metric, idle time almost always looks bad. For end-to-end production flow, it is not always bad.

If the laser can cut twice as many parts per shift as bending can accept, constant maximum cutting utilization creates a mountain of work in process. Parts occupy floor space, have to be sorted, stored, found, and moved. An urgent order can disappear among parts that the next operation is not yet ready to process.

In such a flow, planning has to answer not only “is there an open slot on the laser?” but also “should this particular kit be released into the system right now?”

Systems in this class may show part availability and priorities at downstream operations; some also plan against work-center capacity and analyze circulating stock. That is not proof that every plant needs MES. It supports the underlying principle: cutting, logistics, and downstream operations should be planned as one flow.

If downstream is already overloaded, another ready nest is not automatically useful. It may be more productive to switch the laser to another material group, run work whose downstream route is open, or intentionally leave a controlled gap rather than create another buffer in front of the real bottleneck.

Start with the reasons for waiting, not with the choice of software

The worst way to address this problem is to buy a “planning system” first without knowing what the machine is actually waiting for.

Over a representative period, it is useful to classify each meaningful idle period of a healthy machine with simple reason codes, for example:

• no ready NC program;
• waiting for a nest or replanning;
• material not confirmed;
• material exists but has not been delivered;
• next-job priority not defined;
• waiting for operator/loading/unloading;
• job intentionally not released because of downstream conditions;
• another organizational reason.

Then look not only at total minutes but also at recurrence. One three-hour warehouse mistake may be an exception. Ten ten-minute pauses every day while deciding on the next nest are already a systemic process.

Several additional measures can also help:

• share of working time when at least one fully ready-to-cut job is available in front of the machine;
• average age of a job from release to NC readiness;
• planned versus actual start time;
• number of priority changes after program preparation;
• number of starts blocked by material;
• amount of WIP accumulating after cutting.

TRUMPF also provides for non-productive-time tracking and plan/actual comparison at work centers. The method matters more here than the specific system: without reason codes, a single number such as “the laser was idle for 18% of the shift” says very little about what should be improved.

Different losses require different improvements

Once the waiting time is measured, it usually becomes clear that the word “planning” was hiding several independent problems.

If the main loss is the program is not ready, work-preparation changes are needed: cleaner input data, earlier release into programming, technology templates, offline CAM, and balanced programmer workload.

If the weak point is nesting, the plant needs better grouping rules: when to wait for more parts of the same thickness, when to release a partially filled sheet, and how to handle urgent items and remnants.

If the machine waits for material, the center of the solution is warehouse and logistics: reliable stock balances, reservations, addressable storage, remnant control, and advance delivery.

If the problem is a chaotic queue, dispatching rules and one current priority are needed — not another geometric nesting algorithm.

If the laser produces too many parts into an overloaded downstream operation, the answer is release control and flow balance, not maximizing local machine utilization. The broader logic of how a bottleneck moves between operations is covered in the article on laser-shop throughput.

Only after separating these loss classes does it make sense to decide which tools are required: a simple ERP status, a readiness board, reservation discipline, CAM integration, automatic nesting, APS/MES, or another solution. Software should address a loss that is already understood, not define the problem on behalf of production.

Test the system on a real week of orders, not on a demo screen

If a plant plans to upgrade CAM or production planning, a supplier's feature demonstration says very little about future throughput.

A more useful test is to take one typical past week: real orders, materials, due dates, urgent insertions, sheet remnants, and the capacity of downstream operations. Then replay how every decision would move through the new process from drawing to laser start.

You need to see:

• when the job becomes available to CAM;
• when the ready program appears;
• how it enters a nest;
• when the sheet is reserved;
• who changes priority and when;
• what the operator sees after the previous program finishes;
• how the process reacts if material cannot be found;
• whether bending or another downstream operation is ready.

If, after this walkthrough, the laser almost always has a ready next job and people spend less time clarifying and replanning, the improvement is doing real work. If the screen has more attractive charts but the operator still calls the supervisor to ask “what should I run?”, the bottleneck remains.