Building an IoT Platform for Hazardous Locations

Case Studies  |  Monday, 8th April

Not all industries have been able to jump aboard the wave of IoT (Internet of Things) adoption or take part in the trend of digital transformation. While some industries have been able to shift towards smart manufacturing methods or adopt Industry 4.0 practises, some sectors have been left behind.


What have we learnt from IoT adoption in the oil and gas sector?

I. Background

The list of sectors is long; oil and gas, energy, defence, chemical, utilities, pharma, food processing. Combustible atmospheres coupled with regulations and restrictions mean commercial electronic devices are prohibited. In those workplaces where combustibles are present, around 15% of the global workforce is unable to regularly access modern tools such as field force data. The challenge for these industries and combustible locations in particular is that mobile devices, the sensors they use and networks they operate on need to be certified ‘intrinsically safe’ and therefore not able to spark or ignite a volatile environment to be permissible. It is only with this ‘intrinsically safe’ classification that those sectors will also be able to participate in IoT adoption.

Electronic devices such as intrinsically safe tablets are required to meet the standards for ATEX/IECEx Zone 1 or Class I Division 1 and be tested and certified as such. The human element in areas of hazardous operation therefore finds themselves using the traditional paper and pencil to record problems or anomalies. These observations can only then be entered into a system at a later time – sometimes this is hours, days or weeks later. Without real time communication and data capability is inherently most costly and less efficient. It also increases risk which could lead to loss of life in the worst case scenario.

Being able to leverage an efficient IoT tool within this hazardous environment would therefore bring productivity and safety benefits. Operations could be connected. Mobile devices could integrate with areas such as IoT sensors and artificial intelligence applications to create a cohesive, positive solution for critical industries like oil and gas.

The issue is that designing a suitable IoT platform, being able to prototype that platform, test it via field trials in its intended environment and acquiring the needed certifications to do inhibits innovation.

II. Problem

Developing the required ‘intrinsically safe’ devices / intrinsically safe tablets required for hazardous environment use demands a product development timeframe of months or in some cases, years. It’s also nearly impossible to then carry out field testing in the intended environment. The cost is also prohibitive with design, development and certification a minimum of $250,000. To then create a solution which can be utilised in a range of sectors means that thousands of other variations are needed – a process which requires investment of hundreds of millions of dollars.

Establishing a process or method to accelerate the design, development and testing process is a more practical goal. This would replace outdated designs, security protocols and communication methods with modern technology – and the benefits that means for areas like safety, productivity and efficiency to large, globally significant industries.

III. Hypothesis

If it were possible to trial IoT technology in live and volatile environments or scenarios, effective solutions could well emerge. This would allow proof of concept to be trialled and demonstrated before production, giving cost savings and speeding up the overall process.

IV. Pre-Testing Assumptions and Design Criteria

Acknowledging the inherent difficulties that working in a range of volatile environments represents, a number of requirements must be addressed. These are:

1. Tests must demonstrate economies of scale.

This means the IoT for hazardous locations platform would meet the different needs of a range of industries. Tests should replicate conditions found in the following sectors:

  • Oil and Natural Gas – Offshore/Onshore, Midstream, Downstream
  • Chemicals
  • Pharmaceuticals
  • Public Safety
  • Defence
  • Plastics and Rubber Products
  • Textile Mills and Product Mills
  • CPG, Dry food/goods production
  • CPG, Distilleries
  • Printing and related products
  • Paint and industrial coatings
  • Utilities
  • Aerospace
2. The sensor must be a singular design.

This is needed to ensure that testing and certification is homogenised and permit highly localised and unique configurations.

3. Diverse operating climates must be built into the test.

These should account for the following:

  • Intrinsically Safe standards in accordance with ATEX/IECEx Zone 1 or Zone 0, UL 913 5th and/or 8th Edition Class 1 Division 1/Zone 1 and select country specific standards.
  • High and low humidity
  • Salt air, salt spray
  • IP67 or greater device ruggedness requirement
  • An operating range spanning – 40C to + 60C
  • Extended-life battery (2+ years)
  • Sensors must be low cost to replace and not require calibration
  • Highly corrosion resistant and chemical degradation resistant equipment
4. Variable communication protocol options must be included
5. It must be possible to install and use equipment in volatile conditions
6. Third-party legacy sensors and solutions must be able to be integrated
V. Test platform hypothesis

The following major components must be included within the test platform in order for the Hypothesis to be tested:

1. Mobility to Connect Personnel:

The introduction of the humble desktop computer four decades ago initiated the process of digital transformation for business. Following this, many industries moved with the times, gradually incorporating technology such as LANs, WLANs and cloud computing as it became available. This isn’t the case for hazardous environments. They didn’t progress beyond the fixed HMI terminals. The first step should be to equip test participants with intrinsically safe tablets so that they can capture and distribute data to personnel working in hazardous locations (Hazloc). A gas leak can occur with no prior warning which in a disaster scenario could lead to a large-scale catastrophe.

Mobile devices included in the test plan should comply with the following criteria:

  • Certified intrinsically safe and comply with all criteria defined in Section IV. The only product to achieve this requirement is Aegex 10 Intrinsically Safe Tablet for field force deployment in any environment at any time.

IMPORTANT NOTE ON DEVICES EXCLUDED:

  • Commercial mobile devices not designed to be intrinsically safe do not meet the criteria since, (a) there is only one (1) point of failure for safety protection that relies solely on the integrity of a third-party-provided external case and; (b) commercial products not designed as intrinsically safe typically have stored capacitance of energy of five (5) to ten (10) times the energy required to ignite hydrogen.
  • Portable gas detectors that alert users to cease the use of non-certified products do not meet the criteria since, (a) there is only one (1) point of failure as a form of safety protection when relying solely on the performance of the gas sensor, and (b) in a disaster scenario, new risk factors are created where there is a reliance on users to save work and shut down their mobile device when volatile gas is present.
2. Configurable Sensors:

Sensors which monitor ambient conditions or anomaly conditions are use-case specific. Sensors must be configurable to the field to so that the condition requirements of that locality can be met in a disaster scenario. IoT sensors in hazardous environments are used to monitor toxic or volatile gases, the ambient environmental conditions and the operating condition of equipment. High customisation is required for all variants , diverse environments and monitoring objectives, all while configured in a volatile location.

It should be possible to combine any single sensor with any other sensor so that custom configurations and highly contextualised analytics can be created.

Image demonstrating typical configuration options of hazardous area sensors (click to enlarge)
Typical Hazardous Area Sensor Configuration

3. Meta-Scale Operations:

Often hazardous operating environments are found in very large locations, where there is the potential for micro-climates to form and air flow to alter, the presence of which may then impact the dispersal of chemical plume, the calibration of sensors and corrosion
The introduction and integration of machine learning and data management in these large, complex operations is known as “MetaScale IoT.” Meta-Scale HazLoc areas including locations like refineries, chemical plants, pharmaceutical plants, and urban/Smart City environments are usually not suited to testing the Hypothesis because they are either in use or have volatile atmospheres that are too dangerous in which to run tests.

For the following test case, a disaster training test facility at The Guardian Centres near Atlanta, Georgia USA was selected. This facility meant we were able to very closely replicate actual operating environments. The Guardian Centres facility includes over 100 buildings, spans 3 square KMS, has a 1-kilometre tunnel, a power sub-station plus fuel storage and pumping. These features were built to enable first responder and disaster scenario training with the release of toxic and volatile gases.

4. Communications:

A range of locations including remote and complex in addition to dense environments such as you may find at a refinery must be accommodated by the IoT solution. A number of wireless stands for mobile devices must be supported and should be configurable for scenarios such as operating in low-power and adverse morphologies.

Mobile devices must support at least one of the following:

  • 4G LTE (over 20 bands of support)
  • 4G NB-LTE / Cat M LTE (over 20 bands of support)
  • Wi-Fi
  • NFC
  • Bluetooth
  • LoRa
  • Satellite communications / Remote back-haul
  • Li-Fi (including commercial solutions, private networks and Li-Fi)
5. Interoperability:

Devices to be used in field service, including intrinsically safe tablets, sensors, remote monitoring, app integration and communications standards should integrate with legacy systems while also achieving the test criteria and assumptions laid out in this document. To ensure this is the cases, IoT sensors need to support either physical or wireless connectivity with the legacy equipment. In the experiment to be detailed next, this legacy connectivity was accomplished with the use of a Wild Card sensor node plus drones, robots, wearable devices and other devices using common API’s to achieve interoperability of information on a common cloud. IBM’s Watson was selected for AI along with Microsoft’s Azure.

6. System Architectures:

The system architecture must consider options for data aggregation, data analytics and data distribution. Gartner’s five architectures must be assessed in functional tests. They are:

  • Thing-Centric Architecture – machines/things are “smart” on their own and store their own data; only communicate with the Internet for coordination;
  • Gateway-Centric Architecture – gateway houses application logic, stores data and communicates with the Internet for things that are connected to it; things do not have to be as smart;
  • Smartphone-Centric Architecture – a mobile device houses application logic, stores data, and communicates with the Internet for things that are connected to it; things do not have to be as smart;
  • Cloud-Centric Architecture – cloud acts as connection hub, performs analytics and stores data; things do not have to be as smart;
  • Enterprise-Centric Architecture – things are behind firewall and located together; little need for external Internet
7. Security:

The highest available security protocols for critical operations must be used to protect data, as required by US Military and the Department of Homeland Security. They should comply with International Transfer Arms Restrictions (“ITAR”) as used in private sector oil and gas production, chemical manufacturing, and other highly sensitive industrial processes. AES 128 or AES 256 Encryption is typically required or other standards or methods which are similar in scope – this could include Blockchain IoT solutions.

8. Artificial Intelligence & Analytics in a Meta-Scale Project:

The design of the system architecture and sensors shouldn’t impact efficiency or safety standards in any hazardous location without ensuring that information can be auctioned first. In the event that 30,000 sensors are used to capture data on a standard offshore oil rig, just 1% of the data captured by those sensors is transformed into actionable data for decision making purposes. This is also true of other industries.

It seems incredibly futile to have the ability to capture billions of data points and then fail to translate them into useable information that could allow workers to make necessary or life-saving adjustments. To turn this volume of data into useable, actionable information, data analytics and AI is required.

Any test platform for a new IOT platform intended for hazardous operations requires a range of affordable sensors covering large areas, real-time communication capability, interoperability, machine learning, system architecture and security protocols to function in conditions that are changeable and unpredictable.

VI. The Experiment

To test this hypothesis, Aegex Technologies, Verizon Enterprise Solutions and Nokia collaborated on a 3-day test and conference. During this collaboration, staged disaster scenarios would be demonstrated. Each scenario would be worst case so that technologies could be assessed against the test criteria. The use of the worst case scenario means that normal operating conditions are exceeded.

Case Study: Operation Convergent Response (OCR) 2017

Operation Convergent Response (#OCR2017) was a live demonstration of technologies for first responders, the U.S. Department of Homeland Security, the U.S. Department of Defence and representatives from the Oil and Gas, Chemical, Coatings, Aerospace and Insurance/Risk Management industries.

Three crisis situations were created to test the Hypothesis:

Test Scenario #1: Neighbourhood Flood 37 million litres of water flooded six city blocks at 2 meters deep
Test Scenario #2: Chemical Plant Explosion Realistic chemical plant explosion and 3-story building collapse
Test Scenario #3: Subway Terrorist Attack Terrorist attack/explosion in 0.5 km subway tunnel

Notes on Infrastructure for the Tests Conducted
Aegex IoT sensors connected to an Aegex Gateway using LoRa. The Aegex Gateway backhauled via KLAS Telecom portable Wi-Fi and IoT sensors, communicating via Wi-Fi. Both networks were also backhauled via satellite communications. This was done to test latency and provide back-up redundancy. All sensor data was distributed and presented by the VCORE fourDscape platform in command and control and on Aegex 10 Intrinsically Safe Tablets in the field. Microsoft Azure or IBM Watson environments were used to store data.

A. Test Scenario #1: Neighbourhood Flood
1. Scenario Background

In this scenario, residents have failed to take hurricane precautions. When the hurricane makes landfall, a 2m water surge traps residents and damages electricity, gas and communications infrastructure. Hazards are submerged under water which makes rescue difficult for first responders. First responders are also at risk as victims demand to be rescued and create distractions.

2. Test Plan and Objectives

This test aimed to create a common disaster scenario (the hurricane) in order to determine if new technologies could mitigate the impact of the disaster on the community and improve rescue efforts even with unplanned complications to contend with. A number of technologies meeting the Hypothesis test criteria were assessed. Organisers were keen to test productions and solutions deployed over a number of days to understand how they could be used in a crisis situation in order to speed up product development and deliver daily operational improvements.

Objectives:

  • Identify and test technologies to reduce the level of risk for those trapped by a hurricane
  • Identify and test technologies that give first responders searching for hurricane victims a “forward view” of locations
  • Identify and test technologies able to be quickly deployed in the event that communication infrastructure and systems are damaged or destroyed
  • Identify and test methods of aggregating information

Scenario Results
Network connectivity was a major priority for both response and recovery efforts in this hurricane scenario. First responders unable to communicate with each other, with hurricane victims and real-time sensors shows that quickly deployable communication networks were vitally important.

Mobile devices and sensors powered by batteries helped to facilitate the flow of information. Sensor performance was as expected and in this scenario, they identified and shut down power grid components which could have been dangerous for first responders. The use of the cloud to distribute information was also successful with a chemical plant able to initiate early shut down to ensure that there was no repeat of the fire which occurred in Texas at the Arkema plant in the wake of Hurricane Harvey.

3. Findings and Lessons

The use of a flood scenario allowed hazardous duty technology to be tested. This scenario showed challenges resulting from extreme stresses on infrastructure and meant a fast process of improvement was enabled, ultimately leading to a robust solution.

  • First responders were not prepared for the presence of the chemical plant. Aegex’s intrinsically safe tablets gave first responders access to maps, sensor feeds and aerial views from drones and allowed them into areas they would have been prohibited from with standard commercial devices.
  • Battery-powered and wireless IoT sensors detected water and water levels which translated into advance alerts for the chemical plans. LAN and AC powered sensors failed the tests. Had chemicals escaped, the sensors provided redundant advance warning, protecting the first responders in case the situation further deteriorated.
  • The video feed from aerial drones helped identify locations of victims for first responders.
  • A deployed aerostat balloon gave 4G LTE private network capabilities to allow secure communication at the disaster scene.
  • Multiple parties could collect and distribute information securely in real time. The use of Azure to aggregate data meant intelligent alerts could be used to help with rescue efforts.
B. Test Scenario #2: Chemical Plant Explosion
1. Scenario Background

A chemical processing facility is fitted with sensors ahead of a planned IoT deployment to gather data which has detected pipe and flange vibration anomalies along with variations in temperature. The data suggests that a leak could be imminent but no action is taken. The compromised facility is placed in further peril when a tornado hits. Hydrocarbon vapours are detected and the building explodes, trapping a number of people inside. First responders require a breathing apparatus as the atmosphere is considered immediately dangerous to life and health. Because flammable gases are present, only intrinsically safe electronics can be used.

2. Test Plan and Objectives

This test was intended to create the worst case scenario in this environment where information from an AI technology wasn’t acted on or properly monitored. The test attempted to identify technology and solutions which could be used to detect an impending disaster and aid rescue efforts in the aftermath.

Objectives:

  • Identify and test disaster-warning technology
  • Identify and test technologies to help first responders rescue victims and identify evacuation zones
  • Identify and test methods of information gathering within a common operating environment
3. Scenario Results

A chemical company had initiated full IoT including new intrinsically safe sensors to improve legacy processes but, the operations and maintenance team had continued with their old processes such as using paper forms and inputting details to the system at the end of the shift. Fail points were not being monitored by older explosion-proof sensors, and rounds could not be digitally validated in real-time. Newer senses could have been used to monitor more inputs and predict the leak. AI could have averted the disaster but the data which would have ensured this was dismissed as its value wasn’t understood.

4. Findings and Lessons Learned

This scenario studied how connected workers, backed by configurable sensors could improve safety and efficiency. Although warnings generated by the system were ignored, this test shows that the technology could have a significant impact. The use of intrinsically safe tablets and sensors meant first responders were better able to carry out their search and rescue duties. The equipment gave them access to vital information while on site, which meant response could be accelerated and other explosions avoided.

  • A single or dual gas sensor in an explosion-proof housing can cost upwards of USD $25,000. Battery-powered and wireless intrinsically safe sensors can capture more inputs and fuel greater insights.
  • Digital rounds management reduced duplicated effort by 30%
  • GPS and NFC readers could be used to track location and calculate time spent on tasks such as making rounds. Time reductions of around 66% could be achieved on some tasks.
  • Aerial footage provided by drones meant additional threats could be spotted. Real-time thermal imaging helped first responders to focus their response.
  • Various parties were able to access and securely distribute information in real time using Azure. Sensors connected to a Verizon LTE network with LoRa redundancy created a predictive signature in advance of a leak. This could have been used to prevent the disaster from happening.

The equipment and analytics cost one third of the price of installing a single standard gas detector. AI configuration included some of the following data points on a single device:

  • Vibration
  • Light flash
  • Temperature, Humidity, Air pressure
  • Carbon dioxide
  • Oxygen
  • Hydrogen sulphide
  • Others
C. Test Scenario #3: Subway Terrorist Attack
1. Scenario Background

A terrorist attack takes place in the subway. A number of explosions destroy train carriages, which are sitting in a tunnel 0.5km long. Access to the subway is blocked and information is sparse as the situation develops, hampering rescue efforts. As the victims are in a confined space, exposure to chemical toxins is a concern for them and first responders. There is also the added danger of additional explosions.

2. Test Plan and Objectives

This test was designed to create a very extreme scenario in the challenging environment of a confined space with possible exposure to either volatile or toxic chemicals.

The test was carried out to identify solutions and technologies that could aid rescue efforts in the event of a real life terrorist attack as well as identify how different solutions could impact in order to speed up product development.

Objectives:

  • Identify and test low-cost technologies and configurable confined space monitoring
  • Identify and test technologies to manage typical rounds and lower risk
  • Identify and test technologies that could be quickly deployed in enclosed or confined spaces such as those of a subway station or other environments
  • Identify and test methods of bringing together data in a common information environment
3. Scenario Results

Access control and unauthorised access to the disaster space were immediately apparent issues. This is due to the fact that a subway station is a public space by design, with small scale access control in the form of turnstiles or ticket barriers. Once in the station and on the platform, motion detectors could have been used to detect terrorist movement across train tracks. The sensors would need to be designed to ignore the noise and vibrations caused by trains using contextual data. In the event that sensors had been in place to detect unauthorised access, the disaster could have been averted.

Unauthorised access also raises red flags with respect to a similar situation in other public spaces. Current technology costs mean tunnel monitoring is prohibitively expensive. New sensors, which can be configured and are easy to install without relying on power or communications from the building itself gave a detailed insight into operations before and after the simulated explosion. A connected response was made possible by a Wi-Fi, WiFi, LTE and Li-Fi network. This network also gave first responders real-time updates on conditions within the disaster site including the presence of chemicals or toxins in the confined space.

4. Findings and Lessons Learned

Due to the location selected for the tests, real subway trains were used within the location’s tunnel. This meant the scenario and confined space were realistic. A key finding was that preventive technologies warrant further review.

  • Due to the presence of volatile gases in the tunnel, only intrinsically safe tablets could be used. This enabled first responders to stay connected and access information which wouldn’t have been possible with standard commercial tablets.
  • Aegex IoT battery powered and wireless sensors were easy to place in the tunnel, either on the ground or mounted on a wall. The sensors were used to provide important information about hazardous gases. The IoT solution was also very effective at managing unauthorised access with contextual data that isn’t yet available with current methods.
  • The use of Azure for data aggregation enabled stakeholders to securely distribute information in real-time. It was also useful for alerts to aid in rescue and recovery including those of unauthorised access.

VII. Addressing the Hypothesis – Findings Hypothesis:

The ability to test and trial IoT concepts in live or volatile environments would facilitate the development of effective solutions. This would mean that proof-of-concept could be offered before a product was commercialised, giving obvious cost and time savings.
The three tests carried out were done on a large scale in a carefully chosen location to mirror real life challenges. Solely lab testing technology and solutions couldn’t have replicated the challenges faced in disaster scenarios. Due to the presence of those challenges, a number of solutions were identified which may well have taken years in a lab environment reliant on certifications, product design and commercialisation.

Solutions that emerged from testing the Hypothesis during Operation Convergent Response 2017 include:

  • Intrinsically safe tablets from Aegex. These tablets can go anywhere and are a mobile solution when rugged performance is needed and suitable for use in both hazardous or non-hazardous environments. A key lesson was that it is dangerous to assume an area is not explosive. Aegex’s Windows 10 tablet helps to alleviate this with design and certification which meets international standards for industrial safety.
  • IoT/Industry 4.0 deployment has the potential to bring significant change to industry. A single-use sensor does not significantly generate contextual data. (Note: contextual information rather than false positives are generated by an air temperature sensor vs. surface temperature offset by a light sensor). Enabling personnel to assemble sensor arrays of up to 32 different sensing nodes in one device generates a different view of complex environments than is often expected. The Aegex IoT Sensors meant previously unavailable diagnostics were made available.
  • IoT Sensors must be extremely low-cost and quick/easy to install.
  • IoT Sensors, tablets and other devices must support multiple radios for communications redundancy.
  • Devices may be deployed to a broad range of operating environments, thereby requiring uniform certifications appropriate for all settings.
  • All device platforms ran Windows 10
  • Live video feeds from drones gave useful perspectives and powered situational awareness.
  • Real-time, cloud-based data may impact response time. It gives a bigger picture of the environment at hand to aid learning.
  • Digital tools can be used to coordinate rescue efforts.
  • Efficiency and safety can be enabled with a clear strategy.

VIII. Conclusion

While it is impossible to prevent every disaster, their impact can be lessened with access to better data and enhanced communications. Organisations can be better prepared in the event of a problem by monitoring critical assets and processes. Emergency response can be better coordinated, work more efficiently and take place safely when empowered with accurate, up-to-date information.

When building IoT solutions for hazardous locations, special conditions relating to machine learning in highly volatile operations plus disaster response operations must be carefully considered.

The IoT for Hazardous Locations must factor in components developed specifically for hazardous environments. This is vital to ensure that data is gathered and then used to prevent disasters.

The tests conducted for this research generated a lot of valuable information and insight. This has been channelled into creating products with proven capability. Tests have also shed light on areas where more study and understanding is needed. The tests also influenced the design of the Aegex IoT Platform for Hazardous Locations. This platform includes IoT Sensors evolved from the prototype product.

Aegex Technologies’ IoT Platform connects people with machines, processes and the Cloud in hazardous industries. It improves efficiency, productivity and safety.

Tests offered a realistic environment for research and OCR2017 provided the unusual opportunity of testing IoT under extreme conditions. Building on this research, a new series of tests is planned called Operation Convergent Response 2018 (#OCR2018) and will introduce new scenarios. These include a refinery collapse triggered by an earthquake, an active shooter crisis, a biological terror attack in a subway, a hurricane and flood, a motorway pile-up, helicopter crash and nuclear detonation.

IoT Platform Components

Field trials were conducted based on a typical IoT infrastructure stack5. Key components included data capture devices, wireless infrastructure, cloud computing, AI and actionable delivery to users.

Summary of each component (click to enlarge)
IoT Component Summary

IX. 247able & Aegex

ABLE’s e-commerce division, 247able, offers the choice between purchasing Aegex 10 intrinsically safe tablets with 4G / Wi-Fi, charger, carry case and stylus included, or alternatively purchasing each product / accessory individually. A docking station for further improved productivity is also available to buy. 247able deliver these products globally, and any in-stock UK mainland order placed before 16:00 will be delivered next day.

If you would like more information regarding this article, please contact us at [email protected] or +44 (0)118 916 9420.

Leave a Reply

Your email address will not be published.

22 minute read