Calculating takt time before buying a machine for serial auto parts

Why you should calculate takt before requesting a price

A machine price alone tells almost nothing about future output. From a price list you can’t tell whether one machine will make 180 parts per shift or won’t reach even 140. For serial auto parts this affects everything: lead time, number of operators, shift buffer and the decision whether one machine is enough or you need two.

The most common mistake is simple. People look only at pure machining time, for example 2 minutes 40 seconds, and multiply the shift by that number. On paper everything looks fine. In reality each part gains loading, blowing out, measurement, applying correction, short stops and tool change. Separately these are seconds. Together they easily add 30–60 seconds per piece.

One extra minute changes output much more than it seems. If a part takes not 3 minutes but 4, in 8 hours you get not 160 parts but 120. That’s 40 parts per shift. Over a month it becomes hundreds of parts, underdelivery on a contract or an urgent search for a second machine.

You also need this calculation to compare similar models. Two CNC lathes can cost almost the same but give different results on the same part. One changes tools faster, another has easier loading, a third requires more frequent measurement. Without the calculation you won’t see that.

Practically, it’s better to raise these questions before asking for price. You need not just a machine, but a clear answer: what output will it give on your part and what buffer will remain in a shift. If the actual takt is already close to the limit, better know this in advance. Then you can immediately consider a second machine, automatic loading or a different routing, instead of fixing the mistake after purchase.

What’s included in the takt of one part

Takt is almost never equal to pure cutting time. For a serial part you count the whole interval from when the operator picks up the blank to when the finished part moves on. If you take only the program time, the machine looks faster on paper than it does in a real shift.

Usually takt consists of four parts: loading and unloading the part, machining time, measurement with correction, and losses from tool changes and short stops.

Loading seems trivial, but this is where seconds are often lost on each piece. The operator opens the door, removes the part, puts a new blank, clamps the chuck, checks seating and starts the cycle. If the blank is heavy, hot or requires precise orientation, this stage grows noticeably.

Main machining is not only cutting. It includes tool approach and retract, spindle acceleration and deceleration, axis positioning, turret indexing, clamp operation and program pauses. That is why calculated time and real machine time often differ.

Measurement must also be included. Even if the operator checks every fifth or tenth part, those minutes are still distributed across the entire batch. The same applies to entering corrections after measurement.

Tool changes and short stops add another layer. An insert wears out, the operator cleans chips from the clamping area, waits for stable dimension after the first pieces. To choose a machine, count not the ideal takt but the one the cell can sustain every day.

What data to collect in advance

Accurate calculation starts not from the catalogue but from numbers about the part and flow. If you estimate by eye, you may buy a machine with excessive margin or, conversely, create a bottleneck in the first month.

First fix the production plan in three cuts: per shift, per day and per month. One monthly volume says little. For machine selection it’s more important to know how many parts must be produced in a real shift considering weekends, demand peaks and available hours.

Then clarify batch size and frequency of changeovers. The same monthly volume can be produced in large weekly batches or small daily batches. In the second case changeover time noticeably eats the available capacity and takt requirements become stricter.

Next collect data about the part itself: material, blank type and required tolerances. Bar steel, forging and cast blanks behave differently. Allowance, clamping rigidity, cutting time and measurement frequency change. If tolerances are tight, the operator will measure more often and that becomes part of the cycle.

Manual actions are better measured than remembered. You need real seconds for loading, locating, clamping, opening and closing the door, removing the part, blowing and recording the result. Do several repeats and take the average. These small items often take more time than expected.

Don’t forget adjacent operations. It’s important to know what comes before and after this machine: blank cutting, washing, heat treatment, grinding, inspection. If the previous cell feeds parts in bursts and the next doesn’t accept them immediately, this affects both the in-process buffer and the required pace.

For a serial part like a bushing or housing this is especially visible. A 15–20 second difference in loading and measurement per piece quickly becomes hours per shift.

How to calculate takt step by step

First convert the production plan into parts per hour. If the shop needs 480 parts in 8 hours, that doesn’t mean the required pace is automatically 60 parts per hour. First subtract breaks, handover and time when the machine isn’t producing.

Suppose the shift is 8 hours but actually available is 7.2 hours, or 25,920 seconds. Then required takt is: 25,920 / 480 = 54 seconds per part. This is the limit the future cycle must meet.

Then calculate the cycle line by line:

1. Take the pure machine processing time for one part.
2. Add manual actions in seconds: loading, clamping, start, removing part, blowing the area.
3. Put measurement on a separate line and convert it to seconds per part. If the operator measures every 10th part for 30 seconds, that gives 3 seconds per part.
4. Also include correction and tool change. If a correction is needed every 50 parts and takes 60 seconds, that’s 1.2 seconds per part. If a blade change occurs every 200 parts and lasts 8 minutes, that’s another 2.4 seconds per part.
5. Sum all lines and compare the total with the required takt.

A small example. Machining time — 41 seconds. Loading and removal — 8 seconds. Measurement gives 3 seconds, correction 1.2 seconds, tool change 2.4 seconds. Total cycle is 55.6 seconds.

With a required takt of 54 seconds such a process won’t meet the plan. Even if trials show 52–53 seconds, the margin is too small. In series it is eaten by chips, re-measurement, small stops and operator fatigue.

If the final cycle is at least 10–15% below the required takt, the picture looks safer. For serial auto components this is more useful than choosing a machine by the specification or an appealing cutting time.

Table template

It’s most convenient to calculate takt in a simple table where each action is visible separately. This format quickly shows where seconds go and removes guesswork.

In the first column write the action. Include not only cutting, but everything around the part: loading, clamping, measurement, tool change, cleaning the area, short stop. In the second column put time in seconds. In the third indicate how often the action repeats. In the fourth convert everything to a single number — time per part.

The logic is simple. If an action happens on every part, copy the second column. If it happens less often, divide the time by the frequency. A 30-second measurement every 15 parts gives 2 seconds per piece.

Keep a separate line for small stops. If you mix it with machining or loading, the calculation becomes too optimistic and will quickly diverge from reality.

If the routing is long with several operations, make such a table for each operation. Then look for bottlenecks and compare which machine will sustain the required output without running at its limit.

How to account for loading, measurement and tool change

The most common mistake is lumping all losses into a single number. Don’t do that. Loading, measurement and tool change follow different rules so they are better counted separately.

If an operator feeds the blank manually, measure the real time to open the chuck, remove the part, place the new blank, close the chuck and start the cycle. If feeding is automatic, count the loader or robot time. These scenarios shouldn’t be mixed in one line.

The same for measurement. Don’t put the full measurement time on every part. If the operator spends 90 seconds inspecting one part every 15 pieces, the takt gets 6 seconds per part, not 90.

Four simple formulas are handy in the calculation:

• loading per part = loading time / 1
• measurement per part = single measurement time / measurement period
• tool change per part = change time / tool life in pieces
• small stops per part = total stop time / output over the period

Tool change: if an insert is changed in 4 minutes and lasts 240 parts, add 1 second to the cycle. Small on one piece, but relevant over a long run.

Keep batch setup separate. First tool fixation, trial part, dimension correction and restart after changeover are not single-part cycle items. These costs affect shift time and batch size but should not spoil machine-to-machine comparisons.

It’s useful to separate two lines: “pure machine cycle” and “add-ons per part”. Then you immediately see where time is lost.

How to set in-process buffer

In-process buffer is not just “for peace of mind”, it’s to survive short disruptions between operations. If it’s too small the line waits. If it’s too large you hide a slow area and freeze money in WIP.

Count buffer in parts between two specific operations. Look at the time to the first good part after startup, regular losses on that transition and divide the sum by the takt of the next operation; then round up to whole parts and check whether the buffer became too big for your batch.

Simple scheme:

• take the time to the first good part after startup;
• add usual repeat losses on this transition during the shift;
• divide the sum by the takt of the next operation;
• round up to whole parts;
• check that the buffer isn’t excessively large for your batch.

Example: after changeover the lathe produces the first good part in 18 minutes. The next operation takes a part every 90 seconds. Just for startup you need 12 parts of buffer. If you keep 40–50 parts, that buffer no longer protects startup but hides a slow or unstable area.

A rule of thumb: buffer should cover short breaks, not a whole shift. If the buffer equals half the shift output, it’s usually a sign to check the process rather than increase buffer.

Always check buffer against batch size. For a batch of 60 parts a buffer of 20 may make sense. For a batch of 25 it’s excessive.

Example for a serial auto part

Take a steel bushing in a monthly batch of 6,000 parts. The part is simple: OD turning, boring, face cut and chamfer. On this example the calculation is clear without extra theory.

Input data

• Schedule: 2 shifts of 8 hours, 22 days per month
• Machine availability: 85% of calendar time
• Loading and removal: 18 seconds
• Program cutting: 118 seconds
• Measurement: 20 seconds every 5th part
• Tool change: 6 minutes every 120 parts

Calendar time per month: 2 × 8 × 22 × 60 = 21,120 minutes. With availability that’s 17,952 minutes. Divide by plan 6,000 parts and get required takt 2.99 minutes per part.

Now compute the real cycle. To loading and removal 18 s add 118 s cutting. Measurement isn’t on every part, so average it: 20 / 5 = 4 s per part. Tool change averaged: 360 / 120 = 3 s per part. If the operator spends another 7 s on clamping and start, total cycle = 18 + 118 + 4 + 3 + 7 = 150 seconds, or 2.5 minutes.

There is a margin but not big. The machine must sustain a takt of 2.99 minutes, while the calculated cycle is 2.5 minutes. The 29-second difference per part is consumed by small pauses, chip removal, tool correction and rare stops. For machine selection these numbers are more useful than spindle speed on the datasheet.

Suppose the next operation is washing and inspection with a cycle of 2.7 minutes. It’s reasonable to size the in-process buffer based on the longest regular pause on the lathe. If a tool change and first-part check take about 8 minutes, the downstream station needs at least 8 / 2.7 ≈ 3 parts of buffer. I would set 5 parts. This protects the next operation from short stops without inflating WIP.

This example shows where time hides: not only in cutting but in loading, measurement and tool change.

Where mistakes happen most often

First mistake — use only the datasheet cutting time. The spindle does not make the part alone. The operator loads the blank, removes the finished part, blows the chuck, sometimes sets a stop and measures. If cutting is 38 seconds and manual actions add 22, the real takt is 60 seconds, not 38.

Second mistake — overstate losses. Measurement and tool change are sometimes simply added to each part in full, although these costs should be divided by their periodicity. If an operator inspects one part every 20 and spends 30 seconds, that’s 1.5 seconds per part, not 30. Same with a tool change: 6 minutes every 120 parts adds 3 seconds per part.

Third mistake — assume a full 8-hour shift of pure production. In reality there are blank deliveries, chip removal, size checks, small failures and conversations. Even 20–30 minutes of such losses per shift noticeably change output. If you don’t include them, the machine meets the plan only on paper.

Another error is about the production plan. Sometimes equipment is chosen for a rare peak, e.g., one large order per quarter. Later the machine is underloaded most of the time. Usually it’s wiser to size for normal output and address peaks separately: a second shift, time buffer or redistributing batches.

One more often-forgotten point: the first parts after changeover are checked longer. Sometimes a trial run, repeated measurement and tool correction are done. If changeovers are frequent, these minutes cannot be hidden in the general background.

Quick check before choosing a machine

Before requesting a price it’s useful to put all numbers on one sheet. If any block is “approximate”, the result is almost always too optimistic.

First check the production plan. You need two numbers: parts per shift and parts per month. Shift numbers show whether the machine will hold the current pace. Monthly numbers reveal how much is eaten by downtime, changeovers and days off.

Then ensure the cycle includes not only cutting. For serial parts people often underestimate simple manual actions: remove part, place blank, close chuck, press start. If this takes 18 seconds but your estimate uses 8, the error quickly becomes tens of parts per shift.

A short pre-selection checklist:

• you have the production plan per shift and per month;
• you have measured loading and removal time at least for several cycles;
• you know inspection frequency;
• you know tool change frequency with associated stop and correction time;
• you have an in-process buffer and understand where buffer is needed and where it just inflates WIP.

A small test helps. Take one part and calculate the shift not from 480 minutes but from available time. If after subtracting breaks, measurements and tool changes you have 390 minutes left, compare the cycle to that time.

If any item is not confirmed, don’t ask the supplier to guess the model “by a photo of the part”. Much better to send the table with notes where numbers are exact and where they are estimated. Then the conversation is productive from the start.

What to do after the calculation

A calculation alone doesn’t answer everything. You need to validate it against reality. Next step is simple: check whether one machine can meet the plan and where time is lost.

First compare the calculated takt with the cycle on a similar part you already make or a partner makes. If the difference is large, the reason is usually one of three: too optimistic loading, forgotten measurement time or underestimated tool change losses.

Then convert takt into shift and monthly output using available time. Don’t count ideal 8 hours — subtract breaks, setup, cleaning and small stops. Often on paper one machine meets the plan, but in a real shift you are short by 10–15%.

A quick validation answers four questions:

• how many parts come out per shift at your takt;
• how many shifts are needed for the monthly volume;
• is there margin for scrap, re-measurement and unplanned tool change;
• is one machine enough or should you immediately consider two?

After that prepare a table for the supplier. The less vague wording, the better. Usually four blocks are enough: material and part dimensions, monthly volume, calculated cycle with operation breakdown, accuracy and measurement requirements. If you have a drawing, attach it. Photos of the current part and data on an existing similar cycle also help.

If you select equipment in Kazakhstan, it’s useful to discuss the calculation with a supplier who will also handle commissioning and service. EAST CNC typically looks at the task this way: not just the machine price but the whole cycle for the part, output and shift buffer. This helps quickly decide whether one machine is enough or a different cell format is better.