The board created a task force to deal with the recurring problems with the elevators. They created a report which is attached at the end of this post.

There are a few problems with this analysis, as follows:

• The referenced article clearly states that when the number of callbacks for any individual unit exceeds 4 per year, there is an issue that needs to be addressed. Averaging callbacks across all our units is an incorrect statistical procedure and is misrepresenting what the article is saying.
• The article did not distinguish between “user error” callbacks and mechanical error issues, so choosing to exclude “user error” issues is not justified based on the article.
• Simply using number of callbacks does not adequately address the quality of service. Reliability is generally measured by a set of factors, including
  • Mean Time to Failure (MTTF), which is roughly what this memo is trying to address (again with incorrect statistics);
  • Mean Time to Repair (MTTR), which is not addressed at all but is critical;
  • and Operational Availability (A_o), also not addressed. An extreme example would be having only a single callback in a year, but if it required a year to repair, then the availability of the elevator is 0 for the year.
• The chart provided shows the time each outage began but does not show when the elevator was returned to service. That information is essential for properly assessing the elevators and would allow computation of the Mean Time to Repair and the Operational Availability. As an example, Elevator 1 went out of service sometime before 1 PM on Tuesday September 27 2022. It was returned to service approximately 48 hours later, meaning (for a single data point) the MTTR = 48 hours. These are the kinds of statistics that need to be gathered to meaningfully evaluate the elevator reliability. For example, when Elevator 1 goes out of service again, we can begin to calculate MTTF, which is NOT what is being recorded in the report – MTTF is the time from when the elevator goes back into service to the time it goes out of service. We can also begin to calculate operational availability at that point for Elevator 1 by dividing the time the elevator was in service by the time from one failure to the next.
• The elevators seems very fragile, particularly the door mechanisms. Is there a way to reduce the fragility and hence the amount of time the elevators are down due to “user error”? Would it make sense to add more signage to the elevators to prevent “user error”?