Injuries and fatalities associated with conveyor systems at sand and gravel mining operations continue to be a problem in the mining industry. Research performed by the National Institute for Occupational Safety and Health (NIOSH) showed that from 2000 to 2007, 14% of all mining-related accidents involved conveyors, with most accidents occurring at surface operations [1]. The Mine Safety and Health Administration (MSHA) has also recognized this problem, stating in a recent request for information (RFI): “Since 2007, there have been 17 fatalities related to working near or around belt conveyors, of which 76 percent were related to miners becoming entangled in belt drives, belt rollers, and discharge points. Factors that contribute to entanglement hazards include inadequate or missing guards, inadequate or an insufficient number of crossovers in strategic locations, and/or inappropriate lock out/tag out procedures. Systems that can sense a miner’s presence in hazardous locations; ensure that machine guards are properly secured in place; and/or ensure machines are properly locked out and tagged out during maintenance would reduce fatalities” [2].

To address the problem of entanglement-related accidents, NIOSH researchers continue to investigate emerging technologies that may be beneficial in improving conveyor safety, specifically through enhanced situational awareness. In 2013, NIOSH researchers reported on a wireless miner tracking system, which was repurposed to provide intelligent machine guard monitoring to improve miner safety [3]. More recently, NIOSH researchers developed and installed a proof-of-concept wireless Internet of Things (IoT)-based system to provide real-time monitoring of machinery and conveyors during operation and maintenance [4, 5].

While the above-mentioned systems showed great potential for improving conveyor safety, they required an immense amount of technical expertise to set up and deploy. However, small mines with 10 or fewer employees had the highest incident rate for machine-related severe accidents, and these operations may not have the technical resources or finances to deploy such systems [1]. For these reasons, NIOSH researchers believed a more cost-effective solution requiring less technical expertise to deploy was needed. With IoT making inroads into the mining sector and continually evolving to meet industry needs, low-cost commercially available IoT systems that are fully integrated (turnkey) solutions were explored [6, 7]. This paper describes one example of a commercial off-the-shelf (COTS) wireless IoT system and how it was implemented on a purpose-built conveyor test bed to assess its feasibility to improve conveyor safety. Included are discussions of the successes and challenges of the implementation.

The wireless IoT system selected for assessment has many features and options which lends itself to a straightforward deployment. The option of a non-cloud-based server with a fully developed Windows-based Human-Machine Interface (HMI) pre-installed, coupled with the availability of a wide variety of sensor nodes and a control node, allows the solution to be easily optimized to enhance conveyor safety for mining operations. The server with HMI software pre-installed is priced at $2500 and sensors, control nodes, and gateways range from $150 to $350, making the solution cost effective for even the smallest deployment.

The system was configured as a non-cloud-based solution utilizing a local server. This addressed two possible concerns. Foremost, some mining operations are located in remote geographical areas that may not have reliable, or any, access to cloud storage service providers. Second, a local off-network server could ease any security concerns operators may have regarding cloud-based storage of proprietary data [8, 9].

A closed platform Windows-based HMI is used to configure sensors, display sensor data, and provides an interface to historical sensor data. If connected to a network, the HMI is capable of being set up to send email and text alert notifications based on sensor threshold data. Figure 1 shows the HMI’s sensor status screen.

The system’s wireless components operate in the 915 MHz Industrial, Scientific, and Medical (ISM) frequency band using the frequency-hopping spread spectrum (FHSS) non-time synchronized protocol method (non-TSPM) for data transmission [10]. This combination is especially well suited for low-data rate, large-scale IoT applications [11, 12]. The system employs a star network topology where all sensors communicate directly through a dedicated gateway to the system server [13].

The system sensors were fully programmable and powered by 3.6-V 2400-mA/h batteries. The level of programmability varied by sensor, but all had basic settable parameters such as data acquisition rate, minimum and maximum data thresholds, data transmission rate during the period while thresholds were exceeded, and network connection status check-in interval. The sensors, described in detail later, transmitted data immediately when thresholds were exceeded but otherwise only transmitted on the set interval for system check-in. The control nodes were powered from 110-V mains and had the ability to open or close a pair of 30-Amp relays based on predefined sensor data. A sensor, control node, and gateway (left to right) are shown in Fig. 2.

The combination of the star network topology and non-TSMP method for data transmission allowed for a lower sensor transmit duty cycle and enhanced battery life over mesh-based networks using traditional TSMP methods [14].

Independent of the IoT system, six other components were selected to enhance the functionality of the solution. A pair of point-to-multipoint wireless Ethernet bridges operating in the 2.4 GHz ISM band were selected to extend the range between the sensor gateway and the system server. To allow handheld wireless devices, such as tablets and smart phones, to access the system server’s HMI remotely, 2 Gb wireless access points (WAPs), a network router, and network switch were also selected.

A purpose-built conveyor test bed was developed to assess the selected IoT system’s potential to improve conveyor safety. The core of the test bed was a pair of small “troughing” conveyors, one 12 ft. in length and the other 16 ft. (Fig. 3.)

Since the conveyors came standard with 110-V motors controlled using simple toggle switches, motor controllers were designed and fabricated in-house to simulate those typically used in the sand and gravel mining sector. Separate motor controllers for each conveyor included a mains voltage disconnect, a three-pole motor contactor, momentary switches for start and stop, an emergency stop (E-Stop), and front panel lamps to indicate the presence of mains voltage, the status of the motor contactor, and the existence of a hazardous situation.

The conveyors were also modified to allow for monitoring of the motor guard placement, motor mains voltage, and bearing temperature of the idler and return rollers. The feed-discharge L-shape layout of the two conveyors is allowed for proximity detection at the transfer point.

The purpose of integrating the IoT system with the test bed was to assess its potential to enhance situational awareness associated with conveyor system operation and maintenance. As implemented, should sensor data (status) indicate an abnormal (fault) condition that results in a hazardous situation, such as a guard not in place, personnel would be notified, but the conveyor(s) would not be automatically shut down. However, personnel would have the option of shutting down the conveyors using E-Stops on the motor controller enclosures. Conversely, should a fault condition exist before starting a conveyor, the system would automatically prevent the conveyors from being started. Further, handheld devices, such as smart phones or tablets, could be used to access the server’s HMI via the WAPs. This would allow for remote verification of guard, mains voltage, bearing temperature, and E-Stop status, or for checking conveyor startup attempts or worker proximity to hazards. Remote access to the HMI would be especially useful as part of a lockout/tagout (LOTO) procedure before beginning conveyor maintenance or restarting when maintenance is complete.

A block diagram of the system as implemented is shown in Fig. 4. One controller and conveyor sensor set is omitted for clarity.

The research goal of the project was to assess the feasibility of a fully integrated (turnkey), cost-effective, COTS IoT system to improve conveyor safety. With this in mind, the assessment focused on three areas—technical expertise required for system deployment, system reliability, and mostly importantly, the ability of the system to provide timely notifications of potential safety hazards.

While technical expertise was not essential to deploy or maintain the system, there were challenges and setbacks that required perseverance as well as basic troubleshooting skills to resolve. Documentation for the HMI and system components was primarily available online and was often not up-to-date with current product versions. And, while specifically described as an off-network solution, the HMI and server were optimized to have internet connectivity. These two factors resulted in system settings and sensors being misconfigured on numerous occasions. Subsequently, an excessive amount of time was spent troubleshooting issues and determining the correct settings and configurations.

Initial component reliability was also of concern. Out of 25 sensors and control nodes deployed, one would not power up on arrival, two failed within a week, and a fourth became intermittent within a month. The failed sensors and node were returned under warranty along with a request for failure analysis reports. As of this writing, these reports were not yet available.

Once fully deployed, the system predominantly functioned as designed. Under fault conditions, the ability to start conveyors was disabled, and fault indicators on the motor controllers were illuminated. Sensor and gateway data were logged to the system server, and historical trends were easily viewed using the HMI. The HMI could also be successfully accessed using wireless devices via a WAP. While the system server did have the ability to send email and text notifications—an excellent feature to improve situational awareness—this functionality was not assessed due to the off-network nature of the system.

Sensor network reliability was, for the most part, reasonable with sensors and control nodes checking in at their predefined intervals as well as reporting and taking action on fault conditions promptly. In most circumstances, latency between the occurrence of a fault condition and a resultant system notification and action was typically less than 5 s. However, at one point during the assessment, latencies degraded from seconds to minutes. After extensive troubleshooting, it was discovered that a transceiver (radio) operating on a channel in the center of the 915 MHz ISM band had been placed in close proximity to the IoT system’s gateway. This resulted in wireless channel congestion [15, 16]. Reconfiguring the radio to a channel on the lower end of the ISM band resolved the issue. This behavior did, however, validate a concern over data transmission latency for wireless systems operating in ISM bands [17]. It also highlighted the fact that considerable technical expertise in troubleshooting radio frequency (RF) issues could be required to deploy and maintain a wireless IoT system [18].

Another area of concern was race conditions between sensors. It was observed that occasionally a motor controller’s fault indicator would not be illuminated when a fault condition existed. Working closely with the system’s manufacturer, it was determined that if two sensors indicated fault conditions simultaneously and one sensor then returned to a nonfault condition, the system would re-enable the conveyor’s ability to start and turn off the motor controller’s fault indicator. This would occur even though the other sensor remained in a fault condition. While the system’s manufacturer stated the race conditions could be resolved by reconfiguring monitoring and control algorithms in the HMI, as of this writing the correct algorithm configurations had not yet been identified.

Sensor battery life exceeded expectations. Over the 6-month period of system fabrication, configuration, and assessment, no sensor battery needed replacing, and the HMI status screen showed that all sensors still had healthy battery statuses at the end of the system assessment (Fig. 1.) It must be taken into consideration that this assessment was with sensors configured for hourly check-ins. Should the sensors have been configured to check in more frequently for increased confidence in system reliability, it would have had a direct impact on battery life.

The evaluation of this particular IoT system revealed several challenges in implementing this technology for safety-critical applications, i.e., applications where failure could result in loss of life [19, 20]. While the system has potential uses for increasing situational awareness during operation and maintenance, improvements must be made to increase reliability.

As hoped, an in-depth level of technical expertise was not required to deploy the system. However, the addition of the WAPs and a network router required a basic knowledge of how to assign IP addresses. The increased latency due to channel congestion was not totally unexpected and is a potential issue with any wireless IoT system operating in an ISM band. The two limiting factors of the system were the poor initial reliability of system components and the sensor race conditions. These factors, along with potential latency issues associated with operating in an ISM band, indicate that the particular system evaluated should probably not be considered for safety-critical applications. However, should component reliability improve and the sensor race conditions be resolved, the system may be acceptable as a means to enhance existing safety policies, procedures, and safety controls through improved situational awareness. But even if used for situational awareness, a system must reliably indicate the status of critical sensors, or it will not be trusted.

Through the deployment and assessment, it was found that many factors must be considered when selecting wireless IoT systems for safety applications:

This study suggests that low-cost wireless IoT technologies may not always be suitable for safety critical applications. However, with some improvements and if used in conjunction with existing policies, procedures, and safety controls, a wireless IoT system shows potential to improve mine worker safety around conveyors through enhanced situational awareness [4, 5]. An IoT system can be configured to generate visual warnings—and potentially text and email notifications—to indicate guard removal, status of mains voltage disconnects, worker proximity to hazardous locations, and other unsafe conditions. A system can also be used to implement engineering controls such as preventing conveyor startup until an unsafe condition has been resolved.

This study also suggests the current state of IoT technologies is not “plug and play.” Some technical knowledge of computer networking and RF environments is needed to deploy and maintain even the simplest “turnkey” wireless IoT systems.