Examples

• 1: Flame cutting & blasting
• 2: Brewery
• 3: Semiconductor
• 4: Cycle time monitoring in an assembly cell
• 5: Quality monitoring in a bottling line
• 6: Pharma packaging
• 7: Weaving
• 8: CNC Milling

Metalworking industry

Flame cutting & blasting

Retrofitting of 11 flame cutting machines and blasting systems at two locations using sensors, barcode scanners and button bars to extract and analyze operating data.

See also the detailed documentation.

Identification of the optimization potential of two CNC milling machines

Two-month analysis of CNC milling machines and identification of optimization potentials. Automatic data acquisition coupled with interviews of machine operators and shift supervisors revealed various optimization potentials.

See also the detailed documentation.

Textile industry

Cycle time monitoring

See also the detailed documentation.

Retrofitting of weaving machines for OEE calculation

Retrofitting of weaving machines that do not provide data via the PLC to extract operating data. Subsequent determination of the OEE and detailed breakdown of the individual key figures

See also the detailed documentation

Filling & packaging industry

Performance management in a brewery

Retrofit of a bottling line for different beer types. Focus on the identification of microstops causes and exact delimitation of the bottleneck machine.

See also the detailed documentation.

Retrofit of a Japanese pharmaceutical packaging line

Retrofit of a Japanese pharmaceutical packaging line for automatic analysis of microstop causes as well as to relief the machine operators of data recording.

See also the detailed documentation.

Quality monitoring in a filling line

TODO: #69 add short description for DCC quality check

See also the detailed documentation.

Semiconductor industry

Identification of optimization potential in the context of the COVID-19 crisis

Use of the factorycube for rapid analysis of bottleneck stations. The customer was thus able to increase the throughput of critical components for ventilators within the scope of COVID-19.

See also the detailed documentation.

1 - Flame cutting & blasting

Profile

Photos

Challenges

Lack of transparency about production processes

• Lead times are unknown
• No comparison between target and actual times
• Duration and causes of machine downtimes are unclear

Heterogeneous machinery and machine controls from different manufacturers

• Only minimal connection of the machines to the ERP system
• Manual input of production data into the ERP system
• Machine controls are partially locked by manufacturers
• Machine controls use different protocols

Reliable production planning and quotation generation not possible

• No data on past and current machine utilization available
• Quotation preparation is done with theoretical target times, as no information about actual times is available

Solution

Integration

TODO: #68 add integration for flame cutting

Installed hardware

factorycube

factorycube sends the collected production data to the server. See also [factorycube].

Gateways

Gateways connect the sensors to the factorycube.

Models:

• ifm AL1352
• ifm AL1350

Light barriers

Light barriers are installed on cutting machines and are activated when the cutting head is lowered and in use. Used to measure machine conditions, cycle times and piece counts.

Models:

• ifm O5D100 (Optical distance sensor)
• ifm O1D108 (Optical distance sensor)

Vibration sensors

Vibration sensors are installed on the beam attachments and detect the machine condition via vibration. Used to measure machine conditions.

Model:

• ifm VTV122 (vibration transmitter)

Button bar

Button bar is operated by the machine operator in case of a machine standstill. Each button is assigned a reason for standstill. Used to identify the causes of machine downtime.

Model:

• Self-made, based on Siemens SIRIUS ACT

Barcode scanner

Barcode scanners are used to scan production orders, which contain the target times. Used to scan target times for target/actual comparison.

Model:

• Datalogic PowerScan PD9531
• Datalogic USB Cable Straight 2m (CAB-438)

Implemented dashboards

The customer opted for our SaaS offering. We created the following dashboards for the client.

Navigation

Default navigation options from Grafana, which we modified to allow custom menus.

1. Customizable menu lets you quickly navigate between dashboards
2. In the time selection you can adjust the times for the current dashboard

Plant-manager dashboard

1. Dashboard for the plant manager / shift supervisor, which gives an overview of the production in the factory
2. For each machine the current machine status
3. For each machine, the overall equipment effectiveness / OEE for the selected time period
4. For each machine, a timeline showing the machine statuses in color
5. Overview of all orders, including target/actual deviation and which stop reasons, including setup times, occurred during the order

Machine deep dive

1. Dashboard for the machine operator / shift supervisor, which displays the details for a machine
2. The current machine status with time stamp
3. The overall equipment effectiveness / OEE for the selected time period, including trend over time
4. An overview of the accumulated duration of each stop reason
5. A timeline where the machine states are color coded
6. A timeline where the shifts become visible
7. A timeline where the orders are displayed
8. Overview of all orders, including target/actual deviation and which stop reasons, including setup times, occurred during the order
9. Overview of the number of individual stop reasons

Cross-factory dashboard

1. Dashboard for the cross-factory manager, who can use this to obtain an overview of the sites
2. The overall equipment effectiveness / OEE for the selected time period for all machines.
3. The minimum overall equipment effectiveness / OEE for the selected time period for machine type A.
4. The average overall equipment effectiveness / OEE for the selected time period for machine type A
5. The maximum overall equipment effectiveness / OEE for the selected period for machine type A
6. Overview of all orders, including target/actual deviation and which stop reasons, including setup times, occurred during the order
7. Export function as .csv

2 - Brewery

Profile

Photos

Challenges

Lack of transparency about production processes

• Duration and causes of machine downtimes are unclear
• High proportion of smaller microstops with unknown cause
• Exclusively reactive maintenance, as data on the condition of the components is lacking

Moving bottleneck

• Since the production process is highly interlinked, a stoppage of a single machine can lead to a stoppage of the entire line
• The problem machine “bottleneck machine” is difficult to identify, as it can shift during a shift and is difficult to see with the eye

High effort to collect data as part of the introduction of a continuous improvement process

• Changeover times must be recorded manually with a stop watch and are still not very standardized
• No data on past and current machine utilization available
• Maintenance actions recorded manually, no automatic system to log, store and visualize error codes from machine

Solution

Integration

At the beginning, a “BDE entry program” was carried out together with a lean consulting to identify optimization potentials and to present our solution. For this purpose, the [factorycube] was installed at the filler within a few hours in combination with the tapping of electrical signals from the control system and button strips. A connection of the PLC interfaces was initially out of the question due to time and cost reasons. After the customer decided on a permanent solution, the factorycube was dismounted.

All machines have been equipped with the “Weihenstephaner Standards”, a protocol commonly used in the German brewery industry and were already connected within a machines network. Therefore, the installation was pretty straightforward using our enterprise plugin for that protocol and one central server.

Installed hardware

Server

Implemented dashboards

The customer opted for our SaaS offering. We created the following dashboards for the client.

Navigation

Default navigation options from Grafana, which we modified to allow custom menus.

1. Customizable menu lets you quickly navigate between dashboards
2. In the time selection you can adjust the times for the current dashboard

Plant-manager dashboard

Dashboard for the plant manager / shift supervisor, which gives an overview of the production in the factory

1. For each machine the current machine status
2. For each machine, the overall equipment effectiveness / OEE for the selected time period
3. For each machine, a timeline showing the machine statuses in color

Performance cockpit

Dashboard for the supervisor to get an overview of the machine

1. The current machine status
2. The overall equipment effectiveness / OEE for the selected time period, including trend over time
3. The average changeover time
4. The average cleaning time
5. A timeline where the machine states are color coded
6. A timeline where the shifts become visible
7. A timeline with the machine speed
8. Overview of the number of individual stop reasons, excluding technical defects as they are not relevant for the shift

Maintenance cockpit

Dashboard for the head of maintenance to get an overview of the machine

1. The current machine status
2. The overall equipment effectiveness / OEE for the selected time period, including trend over time
3. The MTTR (mean time to repeair), an important key figure for maintenance
4. The MTBF (mean time between failures), an important key figure for maintenance
5. A timeline where the machine states are color coded
6. A timeline where the process value “bottle lock open/close” is visualized. This helps the manager of the maintenance to isolate the cause of a problem more precisely.
7. A timeline with the machine speed
8. An overview of the accumulated duration of each stop reason, that is relevant for maintenance
9. Overview of the number of individual stop reasons, that is relevant for maintenance

3 - Semiconductor

Profile

Photos

Challenges

Increasing demand could not be fulfilled

• the demand for the product, which was required for ventilators, was increasing over 1000% due to the COVID-19 crisis
• the production was struggling to keep up with the ramp up

Production downtime needed to be avoided at all costs

• production downtime would have meant not fulfilling the demand

A quick solution was needed

• to meet the demand, the company needed a quick solution and could not accept months of project time

Solution

Integration

We were given a 2h time slot by the company to install the sensors, from the time we entered the factory until the time we left (including safety briefings and ESD compliance checks). With the help of videos, we got an overview beforehand and created a sensor plan. During this time slot, we used the machine operator’s break to install all the sensors and verified the data during the subsequent machine run. Through VPN we were able to access the device and fine-tune the configuration.

Installed hardware

factorycube

factorycube sends the collected production data to the server. See also [factorycube].

Gateways

Gateways connect the sensors to the factorycube.

Models:

• ifm AL1352

Ultrasonic sensor

picture TODO

The ultrasonic sensor was used to measure whether the robot was currently moving and thus whether the machine was running.

Models:

• TODO

Proximity sensor

Proximity sensors were used to detect if the product was ready for operator removal. Together with the ultrasonic sensors, we were able to measure whether the machine was standing because the machine operator had not removed the product and was therefore not available.

Models:

• ifm KQ6005

Button bar

Button bar is operated by the machine operator in case of a machine standstill. Each button is assigned a reason for standstill. Used to identify the causes of machine downtime.

Model:

• Self-made, based on Siemens TODO

Implemented dashboards

The customer opted for our SaaS offering and a additional analysis of the data.

Dashboard screenshot

The customer opted for SaaS solution and required only a very simple dashboard as most insights were gained from a detailed analysis. The dashboard includes the functionality to export data as .csv.

Additional analysis

The data was exported into .csv and then analyzed in Python & Excel. Together with interviews of the operators and supervisors we could extract multiple insights including optimization potential through alignment of the work processes and improvement of changeovers through Single-minute exchange of die (SMED).

4 - Cycle time monitoring in an assembly cell

Profile

An assembly cell was retrofitted to measure and optimize cycle times. Customizable textile wristbands are produced in the assembly cell.

Photos of the machines

Challenges

Lack of information about production performance

• Cycle times are unknown
• Bottleneck of the assembly cell cannot be identified
• No information about productivity of individual employees
• Piece counts are not documented
• No comparison between target and actual performance

Lack of transparency about downtimes

• Frequency and duration of downtimes of the assembly cell are not recorded
• Causes of downtime are often unknown and not documented

Connection of assembly cell to conventional systems not possible

• Sewing machines do not have machine controls that could be connected

Solution

Integration

TODO: #66 Add integration for assembly analytics

Installed hardware

factorycube

factorycube sends the collected production data to the server. See also [factorycube].

Gateways

Gateways connect the sensors to the factorycube.

Models:

• ifm AL1352

Light barriers

Light barriers are installed on the removal bins and are activated when the employee removes material. Used to measure cycle time and material consumption.

Models:

• ifm O5D100 (Optical distance sensor).

Proximity sensor

Proximity sensors on the foot switches of sewing machines detect activity of the process. Used to measure cycle time.

Models:

• ifm KQ6005

Barcode scanner

The barcode scanner is used to scan the wristband at the beginning of the assembly process. Process start and product identification.

Model:

• Datalogic PowerScan PD9531
• Datalogic USB Cable Straight 2m (CAB-438)

Implemented dashboards

The customer opted for a combination of our SaaS offering with the building kit (and thus an on-premise option). The customer decided to go for PowerBI as a dashboard and connected it via the REST API with factoryinsight.

Used node-red flows

With the help of Assembly Analytics Nodes, it is possible to measure the cycle time of assembly cells in order to measure and continuously improve their efficiency in a similar way to machines.

Here is an exemplary implementation of those nodes:

There are 2 stations with a total of 4 cycles under consideration

Station 1 (AssemblyCell1):

1a: Starts with scanned barcode and ends when 1b starts

1b: Starts with a trigger at the pick to light station and ends when station 1a starts

Station 2 (AssemblyCell2):

2a: Starts when the foot switch at the 2nd station is pressed and ends when 2b starts

2b: Starts when the quality check button is pressed and ends when 2a starts.

Assumptions:

• Unrealistically long cycle times are filtered out (cycle times over 20 seconds).
• There is a button bar between the stations to end the current cycle and mark that product as scrap. The upper 2 buttons terminate the cycle of AssemblyCell1 and the lower ones of AssemblyCell2. The aborted cycle creates a product that is marked as a scrap.

Nodes explained:

• Assembly Analytics Trigger: Cycles can be started with the help of the “Assembly Analytics Trigger” software module.

• Assembly Analytics Scrap: With the help of the software module “Assembly Analytics Scrap”, existing cycles can be aborted and that produced good can be marked as “scrap”.

• With the help of the software module “Assembly Analytics Middleware”, the software modules described above are processed into “unique products”.

Here you can download the flow described above

5 - Quality monitoring in a bottling line

Profile

A bottling line for filling water bottles was retrofitted with an artificial intelligence quality inspection system. With the help of a camera connected to an ia: factorycube, the bottles are checked for quality defects and sorted out by a pneumatic device in the event of a defect.

Photos of the machines

Challenges

Manual visual inspection causes high costs

• Each individual bottle is checked for quality defects by an employee
• One employee is assigned to each shift exclusively for quality inspection

Customer complaints and claims due to undetected quality defects

• Various quality defects are difficult to detect with the naked eye and are occasionally overlooked

No data on quality defects that occur for product and process improvement

• Type and frequency of quality defects are not recorded and documented
• No data exists that can be analyzed to derive improvement measures for product and process optimization

Solution

Integration

TODO: #67 Add integration for DCC quality check

Installed hardware

factorycube

A machine learning model runs on the factorycube, which evaluates and classifies the images. See also [factorycube].

Gateways

Gateways connect the sensors to the factorycube.

Models:

• ifm AL1352

Light barriers

A light barrier identifies the bottle and sends a signal to the factorycube to trigger the camera.

Models:

• ifm O5D100 (Optical distance sensor)

Camera

A camera takes a picture of the bottle and sends it to the factorycube.

Models:

• Allied Vision (Mako G-223)

Detectable quality defects

Automated action

As soon as a quality defect is detected the defect bottle is automatically kicked out by the machine.

6 - Pharma packaging

Profile

TODO: #70 add pharma packaging case

7 - Weaving

Profile

TODO: #71 add weaving case

8 - CNC Milling

TODO #65