Whether you’re designing an engine or perfecting a process, efficiency is key—and you rely on your Design for Six Sigma practice to get you there. But if you’re only basing your analysis on hard data—not on the variation inherent in any manufacturing, engineering, or service process—you’re only getting halfway there. In your DFSS projects, how do you:

• Manage the risk around new product development?
• Simulating system performance for many different operating conditions?
• Predict design success with limited, estimated or no data?
• Predicting performance quality (capability) of design options?
• Balancing resource allocation within a time constraint and final product quality?
• Predicting manufacturing cost variation?
• Predicting total life cycle cost?

Oracle’s Crystal Ball takes you the rest of the way to success by allowing you to use simulation, modeling, optimization, and forecasting to predict and reduce the effects of variation.

Crystal Ball is a Microsoft® Excel®-based suite of analytical tools that includes Monte Carlo simulation, optimization, capability metrics, and forecasting. The most popular use of Crystal Ball is to define variable inputs as probability distributions and use simulation to view the effects of this variation on one or more outputs.

With little effort, you can apply these advanced analytical techniques to your new or existing spreadsheets to help understand and reduce the effects of variation on new or existing designs and processes. (For information on how Crystal Ball is applied in Six Sigma, Lean Six Sigma, and Procee Improvement, click here.)

Because testing on physical models can be prohibitively expensive, Crystal Ball is particularly valuable in Design for Six Sigma (DFSS) practices, providing designers with easy access to simulation and optimization techniques that help them predict capability, pinpoint critical-to-quality factors, and explore design alternatives.

Engineers use “design by analysis” and simulation to estimate data, improve designs and uncover defects before products are built—a process Crystal Ball facilitates by helping them identify, test, and control how the input (X) variation affects the output (Y). The result is better designs, which lead to overall savings. In the end, customers receive robust products and processes, and get to market fast while avoiding the costly consequences of bad design.

Key features of interest to your application include sensitivity analysis, capability metrics, correlation, and scatter charts. The sensitivity analysis helps you to understand which of the uncertain input variables are most critical and drive the uncertainty in your models. Capability metrics help you understand the quality of your virtual design or process prior to implementation. Scatter charts help you to better visualize the correlations between variable inputs and outputs.

Correlation lets you link uncertain inputs and account for their positive or negative dependencies. If historical data does exist, the data fitting feature will compare the data to the distribution algorithms and calculate the best possible fit and parameters for your data.

> See our Six Sigma datasheet for details on how Crystal Ball fits into DMADV structure

HOW DOES CRYSTAL BALL DIFFER FROM STATISTICAL SOFTWARE TOOLS?

Statistical analysis and data visualization software programs serve a different purpose than Crystal Ball. Statistical tools (e.g., MINITAB, JMP, QI Macros) help to plan experiments, analyze data, and visualize the meaning within the data. Most teams use statistical software to create a mathematical model (often just a transfer function) that represents a system. Crystal Ball works as a complementary tool with these packages and shares only a small number of features, mainly distribution fitting and regression analysis.

Once designers have developed a mathematical equation or system, they then use Crystal Ball to simulate the impact of the component tolerances on the system performance, helping the user to rapidly understand and optimize the system. This virtual testing helps developers to refine products and processes prior to implementation.

For example, a product development team needs to design a fuel injector and wants to understand the sensitivity of the system to certain critical part values. The team wants to select the best part values to create a robust product that meets its tolerances with minimum variation. To study this empirically, the team uses a statistics tool to design an experiment with three factors, at two levels for each factor. The team runs the experiment by substituting various parts into an injector body and measuring the resulting fuel volume. After collecting the data from the 24 trials in the experiment, the team uses the statistical tool to analyze the data and derive models that represent the system behavior.

Next, the team wants to simulate the effect of component tolerances so they covert the model to an Excel spreadsheet, usually represented by one or more formulas (Y’s) with multiple variable inputs (X’s), represented by probability distributions. The team can continue on to optimize the injector design with OptQuest and may work between both the statistical tool and Crystal Ball.

> View white papers with more examples on how Crystal Ball compliments statistical packages

LEARN MORE ABOUT CRYSTAL BALL FOR DESIGN FOR SIX SIGMA

This page offers links to a growing number of resources, including recorded Web seminars, articles, white papers, case studies, and example models. Additionally, you can view a list of common uses and examples reported directly from customers using Crystal Ball. You can also download a free trial version of Crystal Ball to see how it can help improve your business forecasts and decisions!

RECORDED WEB SEMINARS

Monte Carlo Simulation as Process Control Aid Statistically designed experiments (DOEs) have become an essential tool in many fields of research because they can lead to rapid learning and optimization in less time and with less cost. Learn how Monte Carlo simulations helped uncover a “hidden” process factor and how a process was improved through sequential DOE. Presented by Dirk Jordan, Ph.D., Six Sigma Black Belt at Motorola Recorded July 18, 2007	View recording Download files
Selecting Distribution Models Distribution models are important tools for all statistical tasks, including estimation, prediction, simulation, and communication. This seminar presents essential tools for selecting and applying distribution models. Presented by Andy Sleeper, President of Successful Statistics LLC Recorded March 21, 2007	View recording Download files
Monte Carlo Simulation in Chemical Process Design This seminar presents a case study demonstrating the use of Monte Carlo simulation in assessing and managing project financial risk (NPV), capital engineering budget risk and design technical risk. Presented by Randy Perry, Master Consultant with Sigma Breakthrough Technologies, Inc. (SBTI) Recorded February 22, 2007	View recording Download files
Teaching Simulation for Six Sigma Learn how simulation works with Design for Six Sigma (DFSS) to reduce cycle costs, improve cycle time, and increase customer satisfaction while eliminating rework or scrap and reducing end-of-line testing. Presented by Crystal Campbell, Trainer and Product Consultant for Decisioneering and Motorola University Black Belt Recorded October 5, 2006	View recording Download files
Minitab and Crystal Ball Synergy for Multiple Response Optimization Learn how Minitab and Crystal Ball can be used in combination to explore design tradeoffs (i.e. improving one response might unfavorably impact another response) and find the optimum in the presence of variability. Presented by Eric Maass, Motorola's Director for SSPD / DFSS Methods and Technology Recorded July 6, 2006	View recording Download files
DFSS Approach: Using Monte Carlo Simulation for Probabilistic Design Learn how to integrate probabilistic modeling, Monte Carlo analysis and Filtered Monte Carlo optimization into a process map of sequential analytic experimentation and design optimization. Presented by Martha Gardner, Ph.D., Global Quality Leader, GE Global Research Recorded June 28, 2006	View recording Download files
Using Simulation and Optimization to Improve Your DFSS Projects Learn, through a series of case studies, how to use simulation and optimization to create a successful new product or process when it has little or no available design data. Presented by Karl Luce, Master Black Belt with Decisioneering, Inc. Recorded May 9, 2006	View recording Download files
Roadmap for Success: Monte Carlo Simulation in Design for Six Sigma Presents an overview of the DFSS product commercialization process. Shows when to apply Monte Carlo Simulation in financial analysis, multiple response optimization and statistical tolerancing. Presented by Randy Perry, Senior Consultant with Sigma Breakthrough Technologies, Inc. (SBTI) Recorded April 11, 2006	View recording Download files

Top

WHITE PAPERS & ARTICLES

Crystal Ball as a Complimentary Tool to Statistical Packages

Crystal Ball and Minitab: Complementary Tools for Statistical Automation By Andy Sleeper, Successful Statistics LLC	Download
Implementing Design for Six Sigma (DFSS) with Crystal Ball and JMP By Bob Launsby, Launsby Consulting	Download
Using Crystal Ball and MINITAB Together in Six Sigma Projects By Andy Sleeper, Master Black Belt, Successful Statistics LLC	Download

Applications in DFSS

Applications of Monte Carlo Simulation in Design For Six Sigma Randy Perry, Consultant, SBTI	Download
Crystal Ball and Design for Six Sigma By Lawrence I. Goldman and Crystal Campbell, Decisioneering, Inc. (Winter Sim '04)	Download
Crystal Ball® Implementation in Engineering: Engineering Design Under Uncertainty By Yosef Amir, General Electric	Download
Development of Simulator to Improve Process Efficiency (in Steel Industry) 2007 Jong Hag Jeon, Master Black Belt, POSCO	Download
Missile System Design Integrated with Production Planning using Predictive Modeling Techniques By Rick Wood, Lockheed Martin Missiles and Fire Control, Jorge Pica, Lockheed Martin Missles and Fire Control / Brigham Young University	Download
Modeling Medical Device Performance Using Crystal Ball Marty Stout, Principal Engineer/Project Leader, BD Medical	Download
Monte Carlo Simulation as Process Control Aid Dirk Jordan, Six Sigma Black Belt, Motorola	Download
Monte Carlo Simulations for Risk Analysis in Pharmaceutical Product Design Bir Gujral, PAT Coordinator, DSM Pharmaceuticals Inc.	Download
Role of Monte Carlo Simulations in Design For Six Sigma By Narendra Soman, General Electric (Click here for abstract)	Download
The Use of Monte Carlo Simulation in Production Modeling Jerry Hamilton, Lean Six Sigma Coordinator - Black Belt, Lockheed Martin Dallas MFC Jorge Pica, Research Engineer, Lockheed Martin Dallas MFC Sean Elliot, Graduate Student, Florida A&M University Randy Burch, Sr. Manager Production Contracts PAC-3 Program, Lockheed Martin Dallas MFC	Download
Using Monte Carlo Simulations for Probabilistic Design By Martha M. Gardner, Ph.D., Global Quality Leader, GE Global Research	Download

Top

CASE STUDIES

Six Sigma DMADV Case Study #2 - Compressor Design By Decisioneering, Inc.	Download
Six Sigma Tolerance Design Case Study: Optimizing an Analog Circuit Using Monte Carlo Analysis By Andy Sleeper, Successful Statistics LLC	Download
Six Sigma Consulting Crystal Ball Provides Quality Insights to Six Sigma Consultant Andell Associates	Download
Designing Portable Products Motorola Labs Engineers Apply Crystal Ball Pro to Improve the Design of Portable Products	Download

Top

EXAMPLE MODELS

Catapult Model (for Design of Experiments) From: John J. O'Neill, Jr., Compass Quality Management, jjoneill@compuserve.com (more contact information can be found on John O'Neill's Consultants' Corner Listing). Detail: The Catapult model provides a simple, yet practical example of how Monte Carlo simulation can be used to predict the capability of a design as a function of the "X's." Users can vary Spring Constant, Pull Distance, Mass, and Launch Angle to help understand the benefits of simulation for Design of Experiments. The model can be used as a stand-alone demonstration of Monte Carlo Simulation, or can be used to support a presentation/lecture on tolerance analysis.	Download For: Crystal Ball Level: Simple
Design for Six Sigma Model - Piston Displacement Detail: This model demonstrates how Crystal Ball can be used in the design phase of a project to determine the optimal specifications for production. In this case, the product is a piston assembly. The piston displacement needs to be within a certain range to meet customer requirements. The values that impact the piston displacement are defined as assumptions with the appropriate probability distributions. As a result, you can determine the likelihood of producing assemblies outside of the specification. NOTE: This model uses macros, so choose to accept macros when you open the file in Excel.	Download For: Crystal Ball Level: Moderate
Design for Six Sigma Model - Pressure Vessel Detail: The DFSS example model was designed to demonstrate the use of simulation to obtain early visibility into design performance, resulting in overall reduced costs and improved quality during the design process. NOTE: This model uses macros, so choose to accept macros when you open the file in Excel.	Download For: Crystal Ball Level: Moderate
Using Simulation with Design of Experiments Detail: This example model studies how three variables (factors) in an injection molding process affect part length (response) and demonstrates the application of combining simulation with a designed experiment. The model uses linear regression to determine the impact of each of the factors on the response and displays the results of a designed experiment in a table on the Model worksheet. The three factors are defined as assumptions with appropriate probability distributions, and simulation of the model results in a distribution of the response and the likelihood of defective parts. You can adjust the controllable factors to minimize the production of defective parts. NOTE: This model uses macros, so choose to accept macros when you open the file in Excel.	Download For: Crystal Ball Level: Moderate

Top

COMMON USES & EXAMPLES

The following examples were provided by our customers and represent only some of the potential corporate finance applications for Crystal Ball.

• Business process simulations
• Cost estimating
• Design analysis
• Estimating potential exposures of tolerances and non conformance
• Financial modeling for commercialization of new products
• Material selection
• New Product Development
• Process optimization and financial modeling
• Project planning and forecasting
• Project selection based on constraints
• Reliability analysis
• Robust design and performance predictions
• Simulating Process Improvement solutions
• Simulation and estimation on design
• Statistical tolerance analysis
• Tolerance analysis and robustness plus reliability modelling
• Use for Quality analysis such as DOE, Process Capability, Process Optimization, and Design For Six Sigma
  

Top

TEXTBOOKS

Commercializing Great Products with Design for Six Sigma

Design for Six Sigma in Technology and Product Development

Design for Six Sigma Statistics

Six Sigma and Minitab: A complete toolbox guide for all Six Sigma practitioners, 2e

Six Sigma Distribution Modeling

Top