To get started this article provides some guidance on how best to chain Bowties together. First, start with building normal Bowties. Chaining BowTies is a solution if you need more detail, but should not be the default. Mainly because there’s a trade-off between gaining complexity while losing how easy the diagram is to communicate. Once you’ve identified Threats or Consequences in a normal Bowtie that require additional investigation, create a new Bowtie that has as it’s Top Event that Threat or Consequence. Any Threat that you want to create a separate Bowtie for should only contain previous Threats (essentially creating a left-sided Bowtie). A Bowtie to elaborate on a Consequence should only have further Consequences (creating a right-sided Bowtie). That will most likely be sufficient for most purposes.
After the Bowties have been extended, we also need to look at the Barriers. Some Barriers in the original Bowtie will be focused on eliminating the Threat. Logically, these barriers take effect before the Threat to stop it, but in Bowtie we place them to the right to create a single diagram. But since we’re creating extra diagrams before, we can take all the elimination barriers out of the original Bowtie, and place them on the correct lines in the previous Bowtie.
The same is true for Barriers on the right-hand side. Some Barriers focus on mitigating the Consequences, which means they logically take their effect after the Consequence has occurred to minimize further Consequences. If we’re creating an extension of the Bowtie, we can place these mitigation Barriers in the next diagram.
In summary, chaining Bowties can greatly increase the level of detail, but always at a cost of more complexity, so you need to make sure the situation you’re trying to model is sufficiently complex to warrant this kind of diagram.