End-to-end process · Manufacturing in special machine building

Plan-to-Produce

From production order to delivered machine: how process intelligence compares planned and actual dates per work center, finds waiting times between work centers and makes the capital tied up along the way visible.

Jump to the findings

Work in progress: This page is a work in progress: screenshots from the analysis in Noreja will follow shortly. The figures are example values from a synthetic scenario.

What is Plan-to-Produce?

Plan-to-produce is the end-to-end process from production planning to the finished end product. In special machine building it covers production planning and costing, production orders for parts manufacturing and sub-assemblies, and final assembly and delivery. Because every machine is a customer project with a batch size of one, many production orders run in parallel across shared work centers and have to come together again for final assembly. Noreja measures this flow at the level of the individual operation: planned and actual dates, waiting times between work centers and the capital tied up along the way.

Note: All figures are example values from a synthetic demonstration scenario. They illustrate analysis paths and methods and are not production figures of any customer.

At a glance

41%

Operations starting late

start more than one working day after the planned start in the routing

78%

Idle time in the lead time

of parts manufacturing and sub-assembly lead time is spent waiting between work centers

≈ EUR 3.1m

Tied-up capital

average material stock at work centers and buffer areas (work in progress)

46%

Final assembly waits for the control cabinet

of customer projects in which the control cabinet is the last sub-assembly to be finished

The process in three phases

After planning, a customer project breaks down into many production orders that run in parallel across shared work centers and meet again for final assembly. Measurement takes place at every level: operation, work center, production order and customer project.

Phase 1

Production planning & costing

The released customer project yields the manufacturing bill of materials and the routing. Preliminary costing values material and standard times; scheduling sets a planned start and planned end for every operation at its work center. Releasing the production orders makes the plan binding.

Business objects

  • Customer project
  • Bill of materials
  • Routing
  • Operation
  • Preliminary costing
  • Production order
  1. Customer project released
  2. Manufacturing BOM created
  3. Routing created
  4. Preliminary costing completed
  5. Scheduling & capacity planning
  6. Production order released

Guiding question

Are the standard times in the routing realistic – and where are they systematically underestimated?

Phase 2

Production order, parts manufacturing & sub-assemblies

Once the production orders are released, material is staged at the work centers. Mechanical parts, the control cabinet and sub-assemblies are produced in parallel strands that share the same work centers. Every operation is confirmed with an actual start and actual end, and every material movement changes the stock value at the work center.

Business objects

  • Production order
  • Operation
  • Work center
  • Confirmation
  • Material
  • Stock
  • Sub-assembly
  • Paint batch

Mechanical parts manufacturing

  1. 1Material staging
  2. 2Cutting / sawing
  3. 3Milling on the machining center
  4. 4Turning
  5. 5Welding
  6. 6Painting

Electrical & control cabinet

  1. 1Electrical design released
  2. 2Picking of electrical material
  3. 3Control cabinet assembly
  4. 4Wiring & testing

Sub-assemblies

  1. 1Picking of parts & purchased parts
  2. 2Sub-assembly
  3. 3Sub-assembly inspection
  4. 4Storage in buffer area

Guiding question

Where do parts wait between two work centers – and how much capital does that tie up?

Phase 3

Final assembly & delivery

Final assembly only starts once all sub-assemblies are available. It is followed by commissioning and the factory acceptance test with the customer; the machine is then prepared for transport and delivered. Final costing compares the actual costs with the preliminary costing.

Business objects

  • Final assembly order
  • Sub-assembly
  • Acceptance report
  • Delivery
  • Final costing
  1. Sub-assemblies complete
  2. Final assembly start
  3. Final assembly end
  4. Commissioning
  5. Factory acceptance test (FAT)
  6. Packing & shipping
  7. Final costing

Guiding question

Which sub-assembly holds up the start of final assembly – and what does that mean for the delivery date?

A customer project has many production orders, a production order many operations, a work center processes parts from many projects, and a paint batch bundles parts from different orders. These relationships are preserved in the event knowledge graph – which is the only way to trace every waiting time back to the affected part and customer project.

Context that explains the manufacturing data

Not every piece of information that explains a deviation is part of the production order. These context objects enrich the analysis without being steps of the actual flow.

Shift calendar
Shows whether a work center was staffed at all at the planned start – and separates waiting caused by missing capacity from waiting caused by the flow.
Machine breakdown
Downtimes and maintenance windows from plant maintenance explain duration deviations that are not caused by the standard time.
Engineering change
Change orders from the PLM system show which production orders were adjusted again after release.
Quality notification
Inspection findings and rework orders explain why parts jump back to a previous work center.
Purchased-part order
The link to the purchase-to-pay process shows whether an assembly is waiting for a late purchased part.
Delivery date
The delivery date promised for the customer project turns a late start at a work center into a schedule risk for the customer.

Five findings: measure, understand, optimise

The use cases build on each other: the plan vs. actual comparison shows where the plan does not hold, the waiting time analysis where parts sit, and the stock values what that costs. The levers for final assembly and costing follow from there.

Finding 1 · Schedule adherence

Starting late is not the same as taking too long

For every operation, the routing holds a planned start and planned end; the confirmations from shop floor data collection provide the actual start and actual end. Instead of only asking whether an order finishes late, the analysis splits the deviation into two measures: the start delay (actual start versus planned start) and the duration deviation (actual duration versus planned duration).

The two call for different measures. An operation that starts late is waiting for predecessors, material or capacity – the cause lies before the work center. An operation that starts on time and takes too long points to tight standard times, set-up effort or rework at the work center itself.

The example shows exactly these two profiles: the 5-axis machining center starts 58% of its operations late but is only 6% over plan in processing time. The welding shop mostly starts on time but takes a median 31% longer than planned.

Business question to Minerva
“At which work centers do operations start late, and where does processing take longer than the routing plans?”
Plan vs. actual per work center (selection, example values; WD = working days)
Work centerOperationsStart delay > 1 WDDuration vs. planFinished on time
Cutting / sawing2,91022%+4%81%
5-axis machining center3,42058%+6%44%
Turning2,18035%+9%63%
Welding shop1,64018%+31%49%
Paint shop1,21064%+3%38%
Control cabinet assembly96047%+22%36%
Sub-assembly2,65039%+12%57%

41%

of operations start more than one working day after the planned start

Instead of “Which order is late?”

the analysis asks: “Does the operation start too late or take too long – and does the cause lie before the work center or at it?”

Finding 2 · Waiting times

Parts wait longer than they are worked on

The waiting time between two work centers is the gap between the actual end of one operation and the actual start of the next. Across all transitions of a production order this gives a complete picture of where parts wait in the process – and how much of the lead time is actually processing.

In the example, the processing share of parts manufacturing is 22%. The most conspicuous transition runs from the welding shop to the paint shop: welded parts wait a median 4.6 working days because the paint line runs in batches and only starts once it is sufficiently loaded.

Not every waiting time has the same reason. The analysis distinguishes whether a part is waiting for capacity at the next work center, for missing material or purchased parts, for a batch or for an engineering change.

Business question to Minerva
“Between which work centers do parts wait longest, and what are they waiting for?”
Longest transitions between work centers (example values; WD = working days)
TransitionCasesMedian waiting timeMain reason
Welding → painting1,1804.6 WDBatching
Parts manufacturing → sub-assembly2,3403.8 WDMissing purchased parts
Machining center → turning9602.1 WDCapacity
Painting → sub-assembly1,0901.9 WDDrying & transport
Cutting → machining center2,4801.7 WDCapacity

78%

of the lead time in parts manufacturing and sub-assembly is idle time between two work centers

Instead of “How long does manufacturing take?”

the analysis asks: “Where does a part wait, what is it waiting for – and which transition extends the lead time the most?”

Its own flow or a shared resource?

The paint line is a shared resource for parts from many production orders and customer projects. Whether a part waits therefore does not depend on its own order but on which other parts occupy the same paint run.

In the event knowledge graph, the relationship between part, operation and paint batch is preserved. A flat event log per production order only sees that the part is waiting – not what it is waiting for.

  • Which colours or part sizes prolong batch formation?
  • Would a fixed paint-run cycle work better than a minimum load?
  • Which customer projects lose their buffer to final assembly in the paint shop?
  • How does waiting time change if urgent parts get runs of their own?

Finding 3 · Tied-up capital

Waiting parts tie up capital – above all just before final assembly

With the stock values of the materials at the work centers, waiting time turns into money. For every point in time it is known which material of which value sits at which work center. This yields the capital tied up in work in progress – per work center, per production order and per customer project.

What matters is the combination of value and duration: a EUR 200 raw part that sits for ten days ties up less capital than a fully assembled EUR 60,000 module that waits for two. In the example, 43% of the tied-up capital sits in the buffer area before final assembly – in finished sub-assemblies whose value, from material and work already done, is highest.

At an imputed interest rate of 8%, the average work in progress costs around EUR 250,000 a year – before floor space, handling, damage risk and parts made obsolete by later engineering changes.

Business question to Minerva
“How much capital is tied up at which work centers, and how much of it is just waiting?”
Tied-up capital by area (example values; WD = working days)
AreaAvg. stock valueShareMedian idle time
Raw material at work centersEUR 0.31m10%2.4 WD
Parts in parts manufacturingEUR 0.62m20%6.8 WD
Parts before and after paintingEUR 0.38m12%5.9 WD
Sub-assemblies in assemblyEUR 0.46m15%4.2 WD
Buffer area before final assemblyEUR 1.33m43%9.7 WD

≈ EUR 3.1m

average capital tied up in material at work centers and buffer areas

Instead of “How high is the stock?”

the analysis asks: “Which stock is just waiting – where, for how long and with what value?”

Finding 4 · Final assembly

Final assembly waits for the last sub-assembly

A special machine consists of several sub-assemblies – base frame, drive and handling units, safety enclosure, control cabinet – each with its own production orders. Final assembly can only start when all of them are available. Its start is therefore not determined by the average but by the sub-assembly that is finished last.

In the example, that is the control cabinet in 46% of customer projects and a sub-assembly with missing purchased parts in another 21%. By then, the other sub-assemblies have been finished for a median 8 working days and are waiting in the buffer area – exactly where most capital is tied up.

The lever is therefore not to build all sub-assemblies faster but to synchronise them: schedule the critical sub-assembly earlier and start non-critical ones later so they do not wait for weeks for the last part.

Business question to Minerva
“Which sub-assembly is most often the last one when final assembly starts, and how long do the others wait?”

46%

of final assemblies only start once the control cabinet is finished as the last sub-assembly

Instead of “Why does final assembly start late?”

the analysis asks: “Which sub-assembly determines the start – and how long do all the others wait for it?”

Why the control cabinet becomes the bottleneck

Business question to Minerva
“Does control cabinet assembly start later in projects with late engineering changes than in projects without a change?”

Control cabinet assembly can only start once the electrical design is released, and that depends on changes from the mechanical design. Linking the production orders with the change orders from the PLM system shows that projects with at least one engineering change after production release start control cabinet assembly a median 6 working days later.

This is a hypothesis, not yet proof of a cause. But it can be tested: if late changes are the driver, projects without a change must show a markedly more punctual cabinet start – regardless of machine type and utilisation.

Finding 5 · Costing

Actual times show where preliminary costing is too optimistic

The standard times in the routing are the basis not only of scheduling but also of preliminary costing. If actual durations at individual work centers are systematically above plan, delivery dates and costs are underestimated alike.

Aggregated across all operations, manufacturing costs per machine in the example are around 7% above preliminary costing. Two thirds of the deviation arise in the welding shop, control cabinet assembly and sub-assembly – exactly where the plan vs. actual comparison shows the largest duration deviations.

If the measured actual times per work center and machine type are fed back into the routings, quotes for new special machines are based on realistic times and capacity planning puts buffers where they are actually needed.

Business question to Minerva
“At which work centers do actual times deviate systematically from standard times, and what does that mean for costing new machines?”

+7%

higher manufacturing costs per machine than costed, two thirds of it from three work centers

Instead of “How large was the deviation in the last project?”

the analysis asks: “Which standard times are systematically too tight – and for which machine types and work centers?”

From finding to measure

Each finding suggests a direction for improvement. Which one is implemented first is decided by the business based on impact, effort and risk – the analysis provides the basis for that decision.

Prioritised action plan
FindingPossible improvement directionImpact to assess
Schedule adherenceManage start delay and duration deviation separately: check material and predecessors before the work center starts, revise standard times at work centers with duration deviationsMore operations finished on time, fewer rush orders and rescheduling
Waiting timesRun paint batches on a fixed cycle instead of a minimum load, check missing parts before releasing sub-assemblyShorter lead time in parts manufacturing, higher processing share
Tied-up capitalIntroduce stock limits per buffer area and anchor tied-up capital per work center as a KPI on the shop floor boardLess tied-up capital, lower capital costs, less floor space
Final assemblySchedule sub-assemblies backwards from the final assembly start, release and prioritise control cabinet assembly earlierEarlier final assembly start, shorter idle time of finished sub-assemblies, better delivery reliability
CostingRegularly feed measured actual times per work center and machine type back into routings and preliminary costingMore realistic quotes and delivery dates, lower margin risk

Prioritisation is a business decision. Minerva can prepare analysis paths, context and options; control and decision-making authority stay with production management and the business.

The loop: Insight → Action → Impact → Feedback

After implementation, the effect of a measure is measured again with the same manufacturing data. Minerva Frontier Agents can accompany this loop continuously: spotting new deviations, developing hypotheses further and observing the effect of implemented measures.

  1. 1

    Insight

    Finished sub-assemblies wait a median 8 working days for the control cabinet and tie up most of the capital while doing so.

  2. 2

    Action

    Release control cabinet assembly earlier, schedule non-critical sub-assemblies backwards from the final assembly start.

  3. 3

    Impact

    Measure idle time in the buffer area, tied-up capital and delivery reliability again.

  4. 4

    Feedback

    If the effect is not confirmed, the hypothesis is adjusted – for example towards late engineering changes.

Check after every measure

  • Does the idle time of finished sub-assemblies before final assembly fall?
  • Does the capital tied up in the buffer area fall?
  • Does final assembly start closer to the planned start?
  • Does utilisation in parts manufacturing stay stable?
  • Does delivery reliability towards the customer improve?
  • Is the hypothesis confirmed – or does it need adjusting?

More questions for the manufacturing data

The same data – routing, confirmations and stock values – answers further business questions:

  • Which work centers are real bottlenecks, and which only look like one because they wait for predecessors?
  • How often do parts jump back to a previous work center because of rework?
  • Which customer projects are already foreseeably at risk of missing their date?
  • How do rush orders affect the schedule adherence of all other orders?
  • Which machine types have the most stable lead times, and which the least predictable?
  • How much capital would one week less idle time before final assembly release?

Four principles behind the analysis

Measure at operation level

Measurement happens at the individual operation and work center – not only at the finished order, where delays have long since overlapped.

Keep relationships

Part, production order, sub-assembly, customer project and batch stay connected objects instead of being condensed early into a flat event log.

See time and money together

Only the stock value makes waiting time manageable: tied-up capital shows which idle time is really expensive.

Separate patterns from causes

A deviation is a starting point for analysis, not yet a cause. Hypotheses are tested against comparison groups.

Conclusion: lead time is lost between the work centers

In special machine building, the delivery date is rarely decided at a single work center. It results from many small start delays, from idle time between work centers and from sub-assemblies waiting for each other.

The plan vs. actual comparison at operation level shows where the plan does not hold. The waiting time analysis shows where parts sit. The stock values show what that costs. Together they form a control logic that does not demand “work faster” but “synchronise better”.

Compare plan and actual. Make waiting visible. Free up capital.

Frequently asked questions about Plan-to-Produce

What is the plan-to-produce process?

Plan-to-produce is the end-to-end process from production planning to the finished end product. In special machine building it covers production planning and costing, production orders for parts manufacturing and sub-assemblies, and final assembly and delivery. In the operational data it consists of customer projects, production orders, operations, confirmations and material stocks.

How do you measure schedule adherence in manufacturing at work center level?

For every operation, the planned start and planned end from the routing are compared with the actual start and actual end from the shop floor confirmations. This yields, per work center, the share of operations started late, the start delay, the duration deviation and the share of operations finished on time.

What is the difference between start delay and duration deviation?

The start delay measures how much later an operation starts than planned – the cause usually lies before the work center, for example in predecessors, material or capacity. The duration deviation measures how much longer processing takes than planned – the cause usually lies at the work center itself, for example in standard times, set-up effort or rework.

How is the waiting time between work centers calculated?

The waiting time is the time between the actual end of one operation and the actual start of the following operation at the next work center. Summed across all transitions, it shows what share of the lead time is idle time. With context such as shift calendars, batches and purchased parts, the analysis can tell what a part is waiting for.

How can tied-up capital in manufacturing be measured?

Through the stock values of the materials at the work centers: for every point in time it is known which material of which value sits where. This yields the capital tied up in work in progress per work center, production order and customer project. With an imputed interest rate, the capital cost of waiting can be quantified.

What data does Noreja need to analyse a manufacturing process?

At its core, production orders with routings and operations (work center, planned start, planned end), confirmations with actual start and actual end, and material movements or stocks with values per work center. Context such as shift calendars, breakdowns, engineering changes or quality notifications makes the analysis more explanatory but is not a prerequisite. A traditional flat event log does not have to be built first.

Why is an event knowledge graph suited to special machine building?

Because the decisive effects arise between objects: many production orders run across shared work centers, a paint batch bundles parts from different orders, and final assembly depends on several sub-assemblies at once. The event knowledge graph preserves these relationships; a flat event log per case could not explain them.

Are the figures from a real company?

No. All figures are example values from a synthetic demonstration scenario. They illustrate the analysis method and are not production figures of any customer.

See the case on your own manufacturing data

The detailed case description with screenshots from Noreja is in progress. Until then, we are happy to show you the analysis directly: plan vs. actual, waiting times and tied-up capital – based on your routings and confirmation data.

Book a demo

Want to analyse your own process? Talk to us.

More end-to-end processes

Ready to Start

Understand your end-to-end process causally — talk to us

Join companies that are redefining growth and efficiency with intelligent process analysis.

Book a Demo