Small and medium manufacturers (SMMs) have been largely hesitant to make the transition to smart manufacturing due to the following widely held beliefs:
1. Smart manufacturing principles are not useful for SMMs.
2. Smart manufacturing is costly, and the return-on-investment is uncertain.
3. Smart manufacturing cannot be implemented with the in-house legacy infrastructure and manpower resources available to SMMs.
The SMMs mentioned in this article are mainly from the auto industry. Only a small minority of the SMMs encountered had a sophisticated network infrastructure and/or computer/programmable logic controller (PLC) controlled machine—whereas, a large majority had very poor infrastructure, lack of in-house information technology skills and legacy machines that did not have a PLC or computer control.
Simple hardware/software and wireless infrastructure
A simple edge processing hardware device (called “Hub”) comprising Wi-Fi capability, a set of current sensors, an Internet of Things (IoT) signal to convert/transmit data, associated firmware, and application software (smart manufacturing system) formed the minimal infrastructure required. The current sensors used could be wrapped-around the power line cables in the electrical cabinet of the machines. Therefore, the installation of these sensors within the electrical cabinet required just a few minutes for each machine. The installed current sensors were connected to a small IoT node placed inside the electrical cabinet. The sensor produced an induced electromotive force (EMF) in response to the current flowing in the power cable. Using this signal, the power consumption of the machine and the various run time and down time states of the machine could be sensed.
This IoT node converted the induced EMF at any instant into contextual data, buffered the data and wirelessly transmitted the data to the remote edge processing Hub. A Wi-Fi network was established between the IoT nodes and the Hub—which housed the firmware and the application software. The Hub was configured to receive the sensor data from each of the machines at a specified sampling interval. Typically, full configuration of each machine required about an hour and therefore the overall installation time was about four hours.
Smart manufacturing systems
The following smart manufacturing systems were implemented in the SMMs’ facilities:
Production performance improvement to enhance throughput, availability, quality, and reduce waste. In the case of legacy machines, the signatures of the current sensors were captured at high frequency. Using appropriate signal processing algorithms, the machine states could be distinctly identified.
Quality performance improvement to eliminate errors and omissions in the measurements, and reduce cost of quality. In this case, dimensional measuring instruments were integrated with the Hub via Bluetooth to automatically read the measurements of produced parts and eliminate errors and omissions by quality inspectors.
Asset performance improvement to reduce unplanned downtimes and maintain optimal health of assets via proactive interventions before asset failures. From the identified machine states to the current sensor data, as stated above, a statistical model for the occurrence of unplanned downtimes was developed to help predict failures and consequently drive proactive interventions.
Energy performance improvement to reduce the energy intensity. The current sensors calculated the power used at any given instant by each machine which could be integrated over time as energy consumption values.
Dispelling SMM concerns
The implementation of smart manufacturing systems at SMM facilities dispelled the following concerns:
Too costly to implement: Total cost for connecting five critical machines and for applying the four smart systems was very affordable even for an SMM whose annual revenues were only $10-15 million.
Too time consuming to implement: Each implementation could be completed in less than half a day.
Need sophisticated infrastructure: In many facilities, a Wi-Fi LAN was available. But in some cases, even internet connections were not available. In those circumstances, the system was provided with the capability to establish a local Wi-Fi network. In the event remote access was required, 4G connectivity was used.
Legacy machine connectability: Using current signature analysis, legacy machines could be converted into smart machines.
Extensive training: The systems were deployed in a familiar tablet and smart phone app environment.
Lessons learned
Based on the more than 100 implementations of smart manufacturing systems for SMMs that we’ve been involved with, the following can be summarized as the key lessons learned:
1. Buy-in of both the owners and workers is necessary.
2. A minimalistic, bottom-up, incremental and evolutionary approach to transformation has the best chance of success.
3. Apart from low-cost concerns, frictionless implementation is a must.
4. Real time actionable intelligence collected must be useful—not just more data.
5. Identify the various ways to connect new technologies to legacy machines to enable their transformation into smart assets.
6. Include and empower factory workers as part of the transformation process.
7. Integration of plant with corporate information technology systems to provide one version of the truth to executive management.
