Maximize quality and certainty through robust design & strategy
Giving you the probabilistic skills that you really need
Predicting or quantifying the effects of random variability and uncertainty can be of value, but minimizing them is what you really want to do. Robustica has been designed to give you the skills and abilities to do just this. With Robustica you can perform various types of optimization in the face of uncertainty and random variability. This includes robustification, design for quality, business risk management and minimization, and Design for Six Sigma (DFSS). Robustica, helps you do more than simply determine how uncertainty and random variability will affect performance. With Robustica, you can now optimize your system to reduce the negative effect of uncertainty. This is sometime called parameter design, robustification or robust design, and Robustica was explicitly designed to help you with these tasks. Read on, and learn more about how Robustica will help you minimize the negative effects of random variability and uncertainty. You can also download an animated slide presentation on Robustica here.
Using Robustica
Robustica can be used for various tasks that require the prediction and management of random variability. If you would like to see how it is actually used then download the free trial version here, and try it yourself. The trial version comes at no costs and has no expiration date, it simply limits the number of input variables that you can specify. Alternatively, read on to learn more about how Robustica can help you.
As design for quality and robust design software
Robustica was originally developed to provide design engineers with robustification software. It would enable an engineer to robustify the systems that they design (products or manufacturing procedures for example) before implementation, and minimize those quality costs that are caused by random variability. This is what essentially makes Robustica robustification software or design for robustness software, but it could just as easily be called design for quality software.
With Robustica, you can optimize your designs to limit the effects of uncertainty (robust design) and determine the sources of random variability that require most attention and control during manufacture. Robustica enables you to achieve the highest levels of quality through design for quality, and Robustica can also be thought of as design for quality software.
Examples showing the use of Robustica as robustification software and design for quality software can be seen in the cases page.
As Design for Six Sigma software
Design For Six Sigma (DFSS) is similar to robust design (see the articles page), and Robustica can be used as DFSS software. Design for Six Sigma applies many of the tools associated with Six Sigma before a system is implemented. By using Robustica in the design stage, you can optimize a system to limit the effects of random variability on performance and identify where control efforts must be applied post implementation. This covers some of the major DMAIC steps in Six Sigma, and Robustica can be a powerful piece of Design for Six Sigma software for the Six Sigma and quality professional. You can read more about using Robustica for DFSS here.
As business risk management software
The principles of DFSS and robust design can be applied to commercial ventures, and Robustica can be used as business risk management software. By applying Robustica to a venture’s strategy you can optimise the strategy to reduce the chance of failure, and to lower the risk. In fact, it can be found that the optimized strategy will reduce risk and demonstrate higher returns. This is because Robustica’s optimized strategies will often exhibit a healthier cash flow, which both reduces risk and improves returns.
Robustica also enables you to identify the significant sources of uncertainty. You can then implement only the essential and effective controls during the venture to limit the negative effects of this uncertainty. Alternatively, further improvements can be made to the strategy to make it more robust against that same uncertainty. Typically both approaches are needed, but the end result is more efficient business risk management than would otherwise be gained.
Calculating business risk and optimizing a business strategy against uncertainty with Robustica is of value to various people in the commercial world. Entrepreneurs can use Robustica to develop the most efficient form of financing, venture capitalists and venture brokers can use Robustica to perform technical due diligence for themselves and their investors respectively, and established firms can quantify and minimize the business risks associated with a business concept innovation.
Examples of the how Robustica can be used as business risk management software are on the cases page.
Robustica: Unlike the rest
There are a number of applications available today that allow one to predict the effects of uncertainty and random variability on the performance of a product, business strategy or any system that can be modeled in Microsoft Excel. They are typically based on Monte Carlo, and they are more suited to prediction than optimization. However, as competition increases, prediction is simply not enough. Now, anything and everything that will minimize the negative effects of uncertainty must be done. If this is indeed to happen, then the application of probabilistic methods more suited to optimization is required.
The problem is that many professionals and students, who are very capable in their own areas of expertise, have difficulty with the concepts that underlie probabilistic methods and robust design (the minimization of the effects of random variability). The solution to this problem was the development of a probabilistic software application that provided advanced probabilistic techniques (especially robust design or robustification) to the Microsoft Excel user through a new probabilistic addin: Robustica. You will find that with Robustica you can do much more to robustify, Design for Six Sigma, design for quality and manage business risk than you can with other applications.
If you wish to improve your ability to deal with the risks and the effects of uncertainty and random variability, you can purchase Robustica by clicking here. Alternatively, to first see how Robustica can help, you can download the trial version here.
Questions and Answers
What prompted the development of Robustica given the presence of other applications on the market?
Initially, it was thought that the combination of Monte Carlo and genetic algorithms, which is common in other competing applications, would provide engineers with robustification software. However, it was found that Monte Carlo was unable to provide the precision needed to calculate the expressions often used to measure robustness.
Why error propagation and not another method like Monte Carlo?
Error propagation offers two key advantages over other methods. The first is speed, error propagation requires relatively few calculations to find the mean and the standard deviation, especially when compared to Monte Carlo. This advantage is of most significance when robustifying, which is the primary purpose of Robustica. The second advantage is repeatability. The expression used to create the objective function for robustification includes both the deviation of the mean from target and the variance. This expression is very sensitive to random fluctuations (which are typical of Monte Carlo for example) in the calculated moments. This sensitivity has been found to inhibit the optimization process. So in summary error propagation is used because of its speed and repeatability, both of which are essential for robustification.
How will Robustica help with quality engineering efforts?
Quality engineering focuses upon efforts to ensure that the customer receives exactly what they want. This requires both a product that is insensitive to the effects of random variability and control systems that prevent random variability from becoming excessive. Robustica helps you with both. You can first use Robustica to ensure that your design will exhibit the highest quality possible through Robustification. This is the first quality engineering step that should be taken when trying to develop a quality design. You can then use Robustica to find those design features (input variables) that will have the greatest effect upon quality (through random variability). This will provide the most efficient and effective application of your statistical quality control efforts.
Is Robustica suitable for reliability in engineering design?
Reliability efforts within in engineering design typically focus upon ensuring that there is a low probability of failure. Usually the failure modes are ‘hard’. That is, there is no continual degradation in quality but rather a sudden change in operating conditions that prevent proper operation of the system. Such a failure mode is the breakage of a component: it is either broken or it is not. While Robustica focuses mostly on quality problems that show a continual degradation or ‘soft’ failure modes, it can be used to predict (and thus minimize) the probability of a variable exceeding a limit. This limit can be equivalent to a failure mode, and therefore Robustica can be used to optimize a design for higher reliability.
What limitations does Robustica have?
The major limitation displayed by Robustica stems from the fact that error propagation is an approximation method. At times the calculations will be exact; however, when the model being considered displays significant non-linearities the calculations of the mean and standard deviation will be incorrect. An example of the systems that display such non-linearities is the payoff for a financial option, which is not only non-linear but is also discontinuous at the strike price.
How does robustification actually work?
The primary principle that underlies robustification is the location of a point on the hyper-plane (defined by the relationship between the inputs and the outputs of the respective model) that corresponds to the target output value and is as flat as possible. A more detailed, and probably clearer, explanation can be read in the respective article on the articles page.
Does this mean I no longer need my Monte Carlo based programs?
No! Error propagation is an approximation method and might at times produce erroneous results. For this reason Robustica has been designed such that if erroneous results are detected, a warning is given. In such cases it is recommended that a Monte Carlo based program be used to verify the accuracy of the moments that Robustica has calculated. It is still possible that Monte Carlo will have issues in these situations.
What do I need before I can run Robustica?
Robustica is an addin to Microsoft Excel, and requires Microsoft Excel to be installed first. Robustica runs on Microsoft Excel 2000, XP and 2003. For the 2000 and 2003 versions owc10.exe will need to be installed first. This can be downloaded free from the Microsoft website. Some people have found that they also need to download owc11.exe.
Why Excel?
Microsoft Excel is a ubiquitous piece of software that is well utilized by numerous professionals in many fields that are faced with risk and uncertainty. These fields range from the commercial to the technical, and there is really no other suitable application that is familiar to all of them. Therefore, Excel was the ideal choice for an environment that would allow as many people as possible to take advantage of the process of robustification. This is the reason for Excel. If you feel Excel is unsuitable for your complex systems, which require dedicated simulation packages, you might want to read the surrogate-models articles on the articles page. It is possible that in the future a version for Mathematica will be developed.
CJSteele Uncertainty Management
Predicting and mitigating the effects of uncertainty on technical and commercial systems
