Discover how Bow-Ties and AI will improve your safety

Bow-tie diagrams are used widely to improve safety in high hazard industries, but they can provide a better understanding of how to prevent harm in any industry. In particular, if you are looking for insights on ways to use new technology – such as AI-powered computer vision (CV) – to reduce risk, the bow-tie could be the tool you need. To illustrate this, we’ll use an example from a warehouse. 

Step 1: Define a hazardous event

Your incident records will help you to identify hazardous events. Imagine you’ve had several incidents where fork lift trucks (FLTs) have hit racking, causing objects to become unstable. Perhaps some have almost fallen. Others have fallen, but injury has been avoided. Or someone has been injured. Define this hazardous event in the centre of the bow-tie (the yellow circle in our example).

Step 2: List all the hazards

For the event you’ve described, list all the hazards you can think of related to that event. Don’t worry about whether hazards cause the hazardous event on their own, or if several are needed in combination. In the orange section of our bow-tie we’ve listed hazardous activities that might contribute to a FLT hitting the racking – driving too fast, using the wrong route, or a pedestrian causing the FLT to manoeuvre too close to the racking to avoid a collision.

Step 3: List the outcomes

You can leave out incredibly unlikely outcomes – for example, unless you are storing flammable items, it is unlikely that a falling box will cause a fire. However, you should include a range of possible outcomes. In this case we’ve listed an object falling off the shelf when there is no one there, causing only financial losss, the load falling in front of a vehicle and causing it to swerve or brake, and something falling onto a worker and injuring them directly. These are described in the red section of the bow-tie. 

Step 4: Consider the barriers

There are two types of barriers: things that stop the hazardous event occurring, and things that, once the hazardous event occurs, stop the outcome from being serious. If a fire was the central event, controlling ignition sources and combustible materials would be on the left (the blue area) and fire alarms, emergency lighting and evacuation procedures would be on the right (the green zone). The recovery and mitigation don’t stop the hazardous event, but they do reduce the severity of the outcome. Existing barriers to prevent an FLT hitting the racking could include speed limits (enforced by vehicle controls, training, observation and signage) and separation of routes for vehicles and pedestrians. These are shown as blocking the arrow from the hazard to the hazardous event. Once an FLT has hit the racking, your key barriers might be reporting and inspection of the racking, correcting any stacking problems. This is shown in the green part of the bow-tie.

‍

Step 5: Review the barriers 

The key to controlling risk is to have multiple lines of defence. Speed delimiters fail or are disabled, preferred routes become blocked, people make wrong decisions. People might forget to report a hazardous event. With traditional approaches it might not be practical to do much more – vehicles can be inspected more often, and workers can be reminded of the importance of reporting. But CV allows you to add extra lines of defence. If you are familiar with Reason’s Swiss Cheese model, this is like adding an extra slice of cheese, with fewer holes. The bow-tie model supports us in looking for additional controls either side of the hazardous event – what can we do to prevent the FLT hitting the racking, and if that occurs, how can we recover from the event, and mitigate any harmful outcomes?

Computer vision (CV) can provide extra layers of prevention and mitigation:

Prevention: CV can detect when FLTs travel too quickly, when they are in the wrong location, and even when they travel the wrong way along a route. It can detect pedestrians using a vehicle-only route. Being able to detect and measure these hazardous activities improves your chance of preventing the hazardous event.

Mitigation: Rather than relying on individual drivers to report when they hit racking, CV can detect close proximity events and feed short video clips to a manager to assess whether a collision has occurred. If it has, the manager can arrange for the racking to be inspected, correcting any stacking issues if necessary.

Conclusion

Bow-tie diagrams are a powerful tool for improving safety. By visualising the sequence of hazards, prevention, event, recovery and outcome you can review existing barriers and identify opportunities to use new solutions, such as computer vision, to reduce the risk of harm.

‍

‍

Protex AI uses computer vision to analyze real-time video feeds and detect potential safety hazards before they occur. By alerting employees to potential dangers, Protex.ai helps prevent accidents and injuries. Additionally, our software provides detailed reports, allowing businesses to identify trends and take proactive steps to improve safety. To see Protex.ai in action, watch our demo video and discover how it can help your business.

‍