EPCL leverages Computer Vision AI to enhance Health and Safety monitoring

Client: Engro Polymer & Chemicals Ltd (EPCL)

Business overview:  EPCL, a subsidiary of Engro Corporation, is involved in the manufacturing, marketing and distribution of PVC and other Chlor-Vinyl allied products.

Challenges: Like many organisations around the world, EPCL faced significant challenges during the COVID-19 pandemic. The nature of its work required core staff – control room engineers - to be present on site round the clock to manage, monitor and maintain the plant and to ensure the company remained operational. Naturally, the organisation had to implement immediate and effective measures to keep employees safe at work, mitigating the spread of the virus and containing any potential outbreaks.

With engineers working across day and night shifts the need to monitor their adherence to social distancing, mask wearing, and occupancy levels was paramount; the organisation needed to identify with pinpoint accuracy any breaches on a 24 x 7 basis. Any breach had the potential to spread the virus across both shifts, creating the need for both shifts to isolate and therefore shutting down the control room and the plant for that duration. The ability to identify behavioural patterns and adjust SOPs accordingly was also critical to ensuring the company and its staff always remained protected and productive throughout the pandemic.

Solution: EPCL implemented EmpiricAI’s WorkSafe Analytics (WSA) to help them to meet the challenges posed by COVID-19. WSA is a Computer Vision (CV) based, Artificial Intelligence (AI) application that detects incidences that match or violate agreed parameters. It leverages EPCL’s existing CCTV systems to recognise and interpret said incidences more reliably and accurately than manual spot checking conducted by health & safety personnel.

EPCL primary objective was to reduce the risk of COVID-19 infection in the most critical areas of the plant.  WSA helped EPCL monitor occupancy levels and adherence to social distancing and mask wearing within EPCL’s plant control rooms and engineering areas and provided insights on areas for improvement on safety.

Results: Following the implementation of WorkSafe Analytics, EPCL’s monitoring capability (manual spot checks vs Computer Vision) improved multi-fold, providing them a much more accurate view of risk.   In addition, they witnessed more than 80% reduction in social distancing violations and up to 90% reduction in mask noncompliance within the first six weeks of deployment.

The rapid and targeted insights and the ability to constantly monitor and detect non-compliance in real-time has allowed EPCL to identify areas that needed improvement in real-time. EPCL was able to identify trends (overcrowding during shift change over), isolate specific issues (individuals not correctly wearing masks) and improve SOPs to keep workers protected and the plant operational.

Monitoring Health & Safety beyond COVID: The versatility of Computer Vision AI is one of its major benefits to organisations. The same technology used to identify mask wearing, occupancy levels and social distancing violations can be trained to look for other health & safety aspects of process safety programs. For a large-scale industrial plant that has numerous potential hazards, CV AI has the capability to improve the safety of workers across the entire site and to improve operational efficiency.

In addition, it can detect when potential hazards are arising to alert staff to act. It’s able to identify smoke and fire, and to look for chemical spills and leakages. Spotting these hazards in time will allow decisive action to be taken and will prevent them from harming workers and causing extensive damage to the site and would reduce or halt operational capacity. Beyond personnel safety, EPCL is also exploring the use of Computer Vision AI to monitor process conditions, such as flame pattern of furnaces installed at site to detect anomalies and generate early warning alerts.

EPCL has already tested the efficacy of Computer Vision AI on their quality assurance process and safety headcount system to count the number of product bags being loaded into the container and to count the number of people entering/exiting their plant premises respectively.

Platform for the future: WSA has given EPCL a more comprehensive view of its workplace safety and a platform from which to make data-backed decisions that not only protects its employees and assets, but is being considered for many other uses.

Speaking about the success of the project, Jahangir Piracha, CEO at EPCL said: “WorkSafe Analytics has proven to be a very effective tool for EPCL. A useful feature of WSA is that it provides us with high quality pictorial evidence of infringements, which is a very effective method to reinforce COVID guidelines and carry out targeted training to avoid future lapses and to ultimately keep our workforce safe.”

Mahmood Siddique, VP Manufacturing at EPCL, added: “We are delighted with the results from deploying the WorkSafe Analytics software. The solution has provided prompt analytics for managing COVID SOPs effectively in the workplace and greater awareness of maintaining social distancing.”

Salman Chaudhary of EmpiricAI concluded: “EPCL has demonstrated how Computer Vision AI and Advanced Analytics can help keep its staff safe and avoid business disruption.

“Computer Vision AI has the potential to digitise and significantly improve health and safety inspections, quality control processes, and overall productivity in manufacturing plants, in areas such as monitoring PPE compliance, monitoring safe operation of machines and vehicles, detecting potential hazardous emissions, monitoring machinery health, inventory counting, monitoring quality control and many more.

“Organisations that do so, will have the opportunity to improve health and safety, drive higher quality, reduce costs, and ultimately transform their operations.“

Find out more about WorkSafe Analytics
Find out more about Engro Polymer & Chemical Ltd

Leave a Reply

Your email address will not be published.