Adaptive Manufacturing: Levels, Example, Limits

Contents

Adaptive manufacturing is production that adjusts itself while it runs, based on data from the running process. It responds to changes such as variation in material, tool wear, a machine breakdown or a rush order. Depending on the level, it adjusts a machine parameter, the order sequence or the allocation of work to machines and shifts.

There is no standard definition. Fraunhofer IPT described the principle in 2016 with a fully automated stem cell plant: a microscope checks the growth density at regular intervals and triggers the next process step when a critical density is reached, so the product determines the course of the process (Fraunhofer, 2016). NIST uses the related term smart manufacturing systems for fully integrated systems that respond in real time to changing demands and conditions in the factory, the supply network and customer needs (NIST program plan).

Adaptive, additive and flexible manufacturing

Term	Core idea	Example
Adaptive manufacturing	adjusts parameters, sequence or allocation to actual data while running	an injection molding machine corrects holding pressure when the measured part weight drifts
Additive manufacturing	builds three-dimensional parts by successively adding material (ISO/ASTM 52900)	a bracket printed from metal powder
Flexible manufacturing	numerically controlled machines linked by material handling and coordinated by a computer (University of Auckland)	a machining cell that runs twelve part numbers without manual changeover

Flexibility describes what a system can do. Adaptivity describes how it reacts to what is happening in operation. A flexible system is often the precondition, because work can only move to another machine if that machine can make the part.

Three levels of adaptation

Level	What is adjusted	Data	Response time	Who decides
Process	feed, temperature, pressure and other machine parameters	measurements on the part or machine	milliseconds to seconds	machine control
Operations	order sequence, machine allocation, staffing	downtime, quantities, order status	minutes to hours	shift lead, dispatcher
Planning	capacity, shift patterns, lot sizes	load and demand over weeks	days to weeks	production planning

This page focuses on the operations level. The process level belongs to machine control, the planning level to capacity planning.

Example: a rush order meets a breakdown

The numbers are invented for illustration. A machining cell has three CNC machines. At 10:00 machine 2 breaks down, and the repair will take four hours. A rush order of 120 parts is due at 16:00. On machine 2 each part takes 2.0 minutes. Machines 1 and 3 can make the part in 2.5 minutes after a 30-minute setup, but each is running an order of its own.

Response	Rush order finished	Delay to other orders
Wait for the repair, then run on machine 2	18:00, two hours late	none
Breakdown noticed after 30 minutes, order moved to machine 3	16:00, without any margin	machine 3 order: 5.5 hours
Alert after 5 minutes, order moved to machine 3	15:35	machine 3 order: 5.5 hours
Alert after 5 minutes, order split across machines 1 and 3	13:05	two orders: 3 hours each

Changing machine 3 back to its own order adds another setup, which the table leaves out. Each faster option has a price, because other orders slip. Adaptive operations therefore need rules agreed in advance: which orders may be delayed, from which expected repair time a move pays off, and who decides. With such a rule the dispatcher acts within minutes instead of starting a discussion.

Preconditions and limits

Data within minutes. Downtime, quantities and order status must be current. A report the next morning is too late for any rescheduling. See downtime monitoring.
Rules. It is clear who may reschedule and above which threshold. Otherwise every shift decides differently.
Approved ranges. Where a process is validated, adjustments stay within the approved limits.
Stable control. A loop that reacts to every small deviation overcorrects and starts to oscillate.
Measurement. If the measured value is wrong, the loop corrects in the wrong direction.
Root cause. Compensating tool wear over and over hides it. Recurring adjustments belong in a root cause analysis.

How SYMESTIC supports adaptive operations

SYMESTIC provides the data for the operations level and the tools to act on it. Downtime is detected automatically, even without a machine signal: if a machine sends none, SYMESTIC detects the stop because no more parts are counted. Notifications go out by push or e-mail when freely definable conditions are met. Status and current operation of every order are visible live, and orders can be released and prioritized. Machines are connected via digital signals with the IoT Box, without changes to the PLC, or via OPC UA.

More on the pages on alarm management and production control.