Question

Preamble:

Nico Case is an industrial engineer working in Vancouver, BC. Ms. Case is responsible for the design and development of a new warehousing operation for a national manufacturer of yoga clothing. Nico has several design alternatives available, each with differing types of equipment, production strategy, cost, and potential throughput. Accordingly, Ms. Case has decided to build a simulation model to guide her analysis. The following questions relate to this study.  

Q 1a18. Ms. Case is interested in setting the warm up period for her model. She has run a pilot run of her simulation and output hourly throughput for her system. She receives the following data:

Time	Rep 1	Rep 2	Rep 3
1	3	5	4
2	49	40	37
3	20	19	19
4	22	24	24
5	24	20	20
6	22	25	25
7	25	20	22
8	25	21	24
9	21	20	24
10	21	24	25

a) Define what is meant by "transient" as opposed to “steady-state" for a simulation model.

b) Why is it important to identify a warm up period for a simulation model?

c) Using Welch's technique and a moving window with w = 1, estimate the warm up period for this model.

                                                               

Answer 1

a.)

The difference between a steady state simulation and marching a transient solution to steady state is that the SS simulation ignores many of the cross terms and higher order terms dealing with time. These terms all go to zero in steady state so they don't affect the steady state result. The transient simulation includes all these terms. Usually this means the steady state model has an easier convergence as there are less terms to model and some transient non-linearities are removed, but in a few models these non-linearities help convergence (but this is infrequent).

A fully converged simulation, run to steady state by either steady state or transient approaches should be the same. The only exception is when "local timescale factor" is used in a steady state run as it can accelerate convergence nicely but as different timescales are used across the domain can cause accuracy problems. As long as a steady state run is run to final convergence with a physical timescale (including auto timescale) then it should be fine.

The timescale is a steady state simulation is like under-relaxation from SIMPLE based solvers. Too high a URF and the simulation diverges, too low and convergence is slow, so you try to fiddle until you get the optimum in the middle somewhere.

b.)

The warm up time is the time that the simulation will run before starting to collect results. This allows the queues (and other aspects in the simulation) to get into conditions that are typical of normal running conditions in the system you are simulating.

In the Warm-Up Period dialog box untick the “Include in Displayed Clock Time” check Box. The clock will run until the end of the Warm-Up period and then reset to the start time of the clock.

Sometimes you do not need a warm up period. An examples of such a situation would be when you are simulating a shop that starts and ends the day empty. A warm up time would be required when simulating a week in a factory, when the factory starts on Monday morning in the exact same state as it finished on Friday evening.

You can sometime reduce the length of warm-up time required by “priming” your simulation with some Work Items in Queues around the simulation. See Queues, initial contents.

c.)