Page 258 -

240 8 Mining Additional Perspectives

To understand the importance of short-term simulation, we elaborate on the
difference between transient analysis and steady-state analysis. The key idea
of simulation is to execute a model repeatedly. The reason for doing the ex-
periments repeatedly, is to not come up with just a single value (e.g., “the
average response time is 10.36 minutes”) but to provide confidence intervals
(e.g., “the average response time is with 90 percent certainty between 10 and
11 minutes”). For transient analysis, the focus is on the initial part of future
behavior, i.e., starting from the initial state the “near future” is explored. For
transient analysis, the initial state is very important. If the simulation starts in
a state with long queues of work, then in the near future flow times will be
long and it may take some time to get rid of the backlog. For steady-state anal-
ysis, the initial state is irrelevant. Typically, the simulation is started “empty”
(i.e., without any cases in progress) and only when the system is filled with
cases the measurements start.
Steady-state analysis is most relevant for answering strategic and tactical
questions. Transient analysis is most relevant for operational decision making.
Lion’s share of contemporary simulation support aims at steady-state analy-
sis and, hence, is limited to strategic and tactical decision making. Short-term
simulation focuses on operational decision making; starting from the current
state—loaded from the BPM system—the “near future” is explored repeat-
edly [95]. This shows what will happen if no corrective actions are taken.
Moreover, “what if” analysis can be used to explore the effects of different
actions (e.g., adding resources and reconfiguring the process).
In [84], it is shown how this approach can be realized using the BPM system
YAWL, the process mining tool ProM, and the simulation tool CPN Tools.This
illustrates the potentially spectacular synergetic effects that can be achieved
by combining workflow automation, process mining, and simulation.