Simulation Design Metrics

Based on our experience over more than 30 years and following the design of fifty plus simulations we have developed a series of design metrics that are indicative of the time taken to develop a new simulation. The page provides several simulation design metrics and case studies of actual designs.

Metrics

Simulation Complexity	Simple	Intermediate	Complex	Note
Duration	2 - 4 Hours	1 Day	1.5 - 2.5 Days	1
Number of Decisions	3 to 5	10	15 to 25	2
Model Proportion	4%	6% - 9%	9%-10%	3
Shell Proportion	94%	94% - 91%	91% - 90%	4
Model Development	20 - 40 days	60 - 100 days	120 - 300 days	5

Types of Simulation

For convinience we separate our simulations in to four classes:

Simple: lasting from two to four hours. Typically, these simulations involve participants using their own microcomputer and printer and are based on our Direct Use, Planning, Analysis and Negotiation Shells
Intermediate: lasting approximately a day. Normally, these simulations are interactive with the decisions of one team affecting anothers and so use our Tutor Mediated shell and the trainer processing the decisions through a single microcomputer and printer. Less often, simulations addressing functional issues will involved each team with their own microcomputer and printer and use our Direct Use shell.
Complex: lasting from a day and a half to two and a half days. Like the intermediate simulations these are usually interactive and utilise our Tutor Mediated shell. However, the complexity of these may mean that it is necessary to incorporate decision support software and this will use our Decision Support Shell. (However, our architecture means that the models are common between the two simulators.)
Very Complex: lasting more than three days. (Because, rarely, business people can afford the time, we have not included this in the metric.

2. Number of Decisions

Our award winning research presented at the 1994 ABSEL Conference showed an empirical link between duration and simulation complexity expressed in terms of the average number of decisions made each period. The formula we derived is as follows:

Duration (hours) = 0.83 * Number of Decisions - 0.355

This asumes that eight periods are simulated and all the decisions are independent.

3. Model Size

This shows the size of the models relative to the total size of the simulator program. The actual number of statement correlate well with the number of decisions but are affected by the independancy of the decisions. The model size shown is based on a simulation that incorporates a full tutor/participant support system. If this is eliminated it halves the model size.

4. Shell Size

This represents the rest of the simulator. (I.E. Model Size and Shell Size adds to 100%.)

5. Model Development Time

This based on the development of recent simulations is indicative of the time actually taken by use to develop the models. If a client is using their own models for incorporation in our shells, the time is likely to be longer (to compensate for lack of experience and the non-availability of a library of models).

Case Studies

The design metrics (above) assume that a new simulation is developed from scratch. But besides this, we are often able to customise existing simulations, modify or extend existing simulations.

Customisation involves taking an existing simulation and changing the terminology and/or the parameters driving the simulation.
Modification involves taking an existing simulation and replacing a few of the models.
Extensions involves taking a simulation and adding models to address additional issues and provide new learning.

Most recent update: 21/07/03
Hall Marketing, Studio 11, Colman's Wharf, 45 Morris Road, London E14 6PA, ENGLAND
Phone & Fax +44 (0)20 7537 2982 E-mail jeremyhall@simulations.co.uk