Using AI to improve HSE compliance programs

Compliance with health, safety and environmental (HSE) regulations requires a lot of resources, and it is time-consuming. Non-compliance can have a significant impact on safety, productivity, profitability, liability and company reputation.

The situation is made worse by the fact that HSE departments are often understaffed and not properly funded. All of this increases both non-compliance risks and the severity of impacts.

How can artificial intelligence (AI) alleviate the pain associated with HSE compliance efforts, and reduce risks?

The compliance burden

Organisations with operations in multiple countries have to comply with many regulations that cover vastly different areas, such as workplace safety, occupational health, air emissions, water discharges and hazardous waste. But this is only the beginning of the complexity faced by companies. The texts of regulations and permits are voluminous and difficult to digest. It’s a challenge to identify specific obligations scattered throughout hundreds of pages. For example, environmental permits include conditions that specify what is allowed and what is forbidden, but the conditions must first be located within the lengthy permit documents. In addition, most permits are site-specific, not generic to all locations, which means each permit must be carefully looked at.

HSE staff carry a heavy compliance burden. They must:

• Identify the regulations and permits that apply to each site.
• Identify and extract specific requirements or conditions from each individual permit or regulation.
• Determine the applicability of requirements and conditions all the way down to the level of each individual item of equipment or asset at each site.
   

Moreover, the regulatory landscape is constantly changing, with regulations being amended or new ones added, thus further increasing complexity.

A perfect storm

Regulatory complexity and a lack of HSE resources create a perfect storm that increases non-compliance risks. And the consequences can be severe, including:

• Penalties and fines
• Revocation of permits
• Worker injuries and compensation claims
• Environmental harm
• Reputational damage
• Equipment damage
• Production delays
   

The situation is even more worrisome for global organisations, because these pain points grow exponentially. For example, a large multinational oil and gas corporation shared with Enablon the example of a permit at one site. That single permit contains 168 conditions and over 1600 site- and equipment-level tasks that need to be completed to maintain compliance. This is just one HSE permit at one site!

Each site can have dozens of permits, and this oil and gas firm has hundreds of sites around the world, which means that they have thousands of permits to comply with globally.

How artificial intelligence helps

AI addresses many challenges through different technologies. One of them consists of natural language processing (NLP). Through NLP, machines can read and understand human language. For example, a computer or software program can ‘understand’ the contents of documents, including contextual nuances of the language within them. NLP can then accurately extract information and insights contained in the documents.

NLP offers many opportunities for HSE managers to save time during compliance tasks. Suddenly, instead of taking hours to read a long regulatory document in order to extract specific obligations (or paying high fees to a consultant), the task is performed automatically in mere minutes.

A typical process looks like this:

• Users submit compliance documents (eg, permits) into HSE software. This triggers an automated processing of the documents.
• NLP capability embedded in HSE software analyses each document and detects relevant HSE compliance language entities (eg, constraint signals and quantitative values).
• Conditions are recognised and extracted from detected language entities.
• The user is guided through the condition review and approval process (to ensure that the conditions were properly identified by AI). Once approved, the conditions are stored in HSE software.
   

For example, in the case of a permit, HSE software looks for wording that may relate to compliance actions, in order to find conditions. The types of wording can include:

• Constraint signals (eg, “shall”, “must”)
• Requirements (eg, “monitor and record”, “inspect”, “provide training”)
• Equipment (eg, “boiler”, “compressor”)
• Quantitative values
• Periods of time
   

The conditions extracted by the HSE software system are reviewed by the user to ensure that they agree with the extracted conditions. They are then approved, stored in the system and broken down into compliance tasks for workers to complete.

The benefits (and business case) for HSE managers

The benefits of leveraging AI for HSE compliance include:

• Automated processing of regulatory documents: from extracting to associating assets and conditions in a few hours instead of days.
• Considerable cost-savings: there is no longer a need to hire consultants for condition extraction.
• Significant time savings: HSE departments can spend more time on activities requiring their expertise.
• More accurate and complete view of compliance requirements: all actionable language is detected so that conditions requiring action are not missed.
• Augmented capabilities for HSE departments: HSE managers understand the breakdown of compliance documents and the extraction of conditions.
• Reduced non-compliance risks: there is a lesser likelihood that regulatory obligations are missed because of human error.
• Data integration: a company’s compliance data is immediately available within their HSE software and ready to be acted upon.
• Easier permit management: the permit deconstruction process is streamlined because conditions are highlighted and reviewed by users. Also, the sources of permit-based compliance obligations are retained.
   

Ultimately, AI makes the job of an HSE manager easier, without replacing their role. AI becomes a valuable tool to maintain compliance and improve the productivity of HSE departments, thus further enhancing the value that they bring to the rest of the organisation.

However, as with all emerging technologies, it is important to ensure any AI solutions are applied responsibly. The following should be considered:

• Overall benefit to the community and whether AI is the appropriate solution for the challenge at hand.
• The best AI solution will depend on the quality and relevance of the data being fed in and as such requires careful data management.
• Any AI project must consider data privacy and security. Stakeholders must have confidence that AI projects have this as a primary consideration.
• Data must be used safely and securely and with a level of transparency to engender trust in any AI project.
• Decision-making should remain the responsibility of organisations and individuals.
   

Image credit: ©stock.adobe.com/au/Shutter B