2002 Sep 01
Process Simulation
David M. Raab
DM News
September, 2002
.

Process simulation uses step-by-step computer models to describe activities such as moving parts through a factory or paperwork through an office. Each step is assigned resources, such as workers, machines and materials, and steps are connected by paths. Models often also include queues where jobs wait between steps when resources are unavailable. Different jobs may take different paths based on job characteristics, random selections, or specific business rules.

The value of process simulation is that it gives detailed information on the costs, time, resource requirements and other features of an activity. This allows users to spot bottlenecks or underused resources in existing processes, and to assess the impact of planned changes without implementing them in the real world.

Direct marketers most often use process simulation to help manage operational functions such as call centers or fulfillment. They also sometimes use simulation for forecasting or lifetime value calculations. Today’s simulation systems nearly always display the process on a graphical flow chart, often with animation that lets the user watch the process run at greatly accelerated speed. After the simulation has executed a specified number of cycles–usually measured in time although sometimes measured in number of jobs–the system generates detailed statistical reports on the costs, resource utilization, wait times, and other results.

Setting up a simulation usually begins with drawing a flow chart. Users create one box for each step in the process and connect the boxes with arrows. This is easy for any user familiar with even simple graphics software like Microsoft Powerpoint. In fact, many process simulation software packages can use diagrams created in Microsoft’s Visio drawing product. The hard part comes after the flow chart is created, when users must define the attributes associated with each step and the rules for moving from one step to another. Attributes generally include the costs and time accrued at each step, and often extend to detailed descriptions of resource capacity, availability and substitutes. Nearly all simulation systems let users define custom attributes and attach notes and documents to the steps.

But accurate modeling requires more than custom attributes. Manufacturing simulation requires nuances such as spacing on conveyor belts, movement of batches in and out of ovens, handling of random equipment breakdowns, and jobs with components that are processed separately and later combined. Call center simulation needs detailed treatment of work schedules, peaks and valleys in job arrival rates, different rules to select calls from hold queues, and variations in the time to complete a job. General-purpose refinements include complex rules for distributing jobs among branches, independent subprocesses shared across multiple simulations, resources shared across multiple steps, and assigning steps and resources to departments.

All simulation system produce standard and custom reports on process results. Most store the results of different model runs so these can be compared with each other. Graphical as well as tabular data displays are also fairly common. Automated sensitivity analysis and optimization are less common but can be useful when trying to fine-tune process changes. Some systems can call external programs or data sources from within the simulation. A few expose their own functions so users can build custom interfaces.

Dozens of process simulation packages are available, with prices starting around $1,000 for a single user. Buyers should also plan on purchasing a day or more of training. Modules that let multiple users share models and results cost extra, although some systems offer browser-based viewers that can run prebuilt simulations without having installed the software itself.

Here is a quick look at some representative products.

AllCLEAR (Proquis, Inc., 877-876-3430, www.proquis.net, $695) is a flow charting package that provides limited simulation capabilities. The software is particularly easy to use–setting up a basic simulation took only a few minutes with no training. Unlike other systems, AllCLEAR does not provide any predefined step attributes; paradoxically, this simplifies set-up because users need only work with the attributes they needs. The system also lets users define process flows in a text outline as well as the usual graphical format. Both formats are displayed simultaneously and users can switch between them at will. But AllCLEAR provides only the simplest possible simulations, with no options to identify different job types, no control over the time between steps, and no capacity limits on resources.

iGrafx Process 2000 (Micrografx, www.igrafx.com, 888-744-1210, $995) is also built on a process charting platform, but provides a much more complete set of simulation functions. The interface is generally easy to use, except for an annoying difficulty with relating branch attributes to lines on the flow chart. The system provides a number of advanced features, including split jobs, a spreadsheet view of the process, and sophisticated statistical functions to generate new jobs and distribute jobs across branches. It’s particularly strong at capturing information about the simulation results. Users can accumulate cost and time data at transaction, department, process and scenario levels, and take snapshots at regular intervals during the similation. Process 2000 also has convenient functions to assign steps to different departments, using horizontal bands commonly called swim lanes. Steps can also be assigned to phases represented by vertical divisions. The optional Process Central module allows Web-based access to a central model repository.

Sciforma Process (Sciforma Corporation, www.sciforma.com, 800-533-9876, $2,500) comes from a firm with a background in project and process management. It too supports split tasks, complicated input calendars, swim lanes for departments, and a spreadsheet view of the underlying data. It provides particularly sophisticated functions for manufacturing simulation, including resource sharing, prioritization and substitution; special options for materials; and functions for batches, overtime and random equipement failures. Users can set up tabs to display independent subprocesses or segments of a large flow chart. They can also integrate Visual Basic scripts for precise control of the simulation flow, and can distribute a completed simulation as an ActiveX object that users can run but not change. A multi-user repository is available at extra cost.

* * *

David M. Raab is a Principal at Raab Associates Inc., a consultancy specializing in marketing technology and analytics. He can be reached at draab@raabassociates.com.

Leave a Reply

You must be logged in to post a comment.