What is resilience?

Resilience describes how well a system keeps its essential function when something goes wrong. A supply chain that continues delivering after a port closes is resilient. It is different from efficiency, and often at odds with it, because slack and backups cost money in normal times.

How it works

Resilience comes from several features: buffers such as extra stock or cash, redundancy such as alternate suppliers, diversity so one failure does not hit everything, and fast feedback so problems are noticed early. Systems tuned to remove all slack are efficient until something unexpected occurs, and then fail quickly. Increasing resilience therefore trades a small, steady cost for protection against a rare, large loss. A model can show how big the buffer must be for the system to survive shocks of different sizes.

Resilience should be measured, not assumed. Useful measures are the time to recover after a shock, the size of the shock the system can take without losing function, and how many failures in a row it can absorb.

Example

Say a factory uses 100 parts per day and holds 200 parts in stock, with one supplier that sometimes fails for up to 3 days. Stock lasts 2 days, so a 3-day failure stops production for 1 day. Holding 300 parts costs more in storage but covers 3 days. If failures cost 50,000 per stoppage day and occur twice a year, the extra buffer pays for itself. This is an illustration with round numbers.

How to model it in Machinations

Model the stock as a Pool, the supplier as a Source with a probability of failing using a Gate, and consumption as a Drain. Add an End condition for when the Pool reaches zero. Play many Plays at once and compare how often the factory stops for different buffer sizes set by a Lever.