stavex

Statistical Design of Experiments (DoE) Instead Of Trial And Error

Effortless Process and Product Optimization with Smart Design of Experiments (DoE)

Our new software STAVEX 6.0 is the first web-based Design of Experiments (DoE) tool for the intranet and it is here to facilitate your work! Whether you are an expert or just getting started with Statistical Design of Experiments, STAVEX will enable you to take advantage of this methodology very easily – for instance for obtaining the Design Space in the context of Quality by Design (QbD).

Making DoE Easier than Ever.

The intuitive interface of STAVEX 6.0 and the streamlined workflow guarantee high usability from the initial project specification to the final optimization of your system.
The expert system STAVEX 6.0 features excellent user guidance that assists you through your whole DoE journey, in particular for the design choice and the interpretation of the results.
STAVEX 6.0 provides a clear and understandable report as well as modern and readable representations which are easy to interpret.
Thanks to totally new reporting functions, users can select which pieces of information (from the problem definition to the analysis and its plots) should be part of a PDF or HTML report.
STAVEX 6.0 is available in English, French, German and Spanish. This makes it easier for you to start using Design of Experiments.

Result Visualization in the STAVEX Workflow

Features

Concept

Features

Easy Design and Analysis of Experiments With STAVEX

Flexible User Specifications

Response variables: quantitative and qualitative, input of specification limits possible
Factors: quantitative, qualitative, alternative, and mixture factors
Alternative factor classes for defining those that mutually exclude each other, for process factors (e.g. for choosing a solvent) and mixture components (e.g. for selecting an excipient)
Multiple mixture-factor classes possible to handle formulation problems that include subformulations (e.g. 70% detergents + 30% perfumes = 100%)
Various factor restriction types available: linear or specific to mixture ratios
Weighted desirability function for calculating the best compromise factor setting when optimizing several response variables at once, e.g. to obtain the Design Space under the Quality by Design (QbD) paradigm
Fast and easy change of specification limits and response variable weights
Specification of transformations possible (for quantitative response variables and factors)
Selective specification of interactions
Blocking of known uncontrollable noise factors
Easy adaption of designs possible to account for hard-to-change factors, a temporal trend, violated factor settings or impossible experiments
Automatic completion of incomplete designs
Interactive specification of confirmatory runs possible
Interactive selection of kept factors and of their ranges upon moving to a new experimental cycle

User Guidance

Integrated Expert System
Support for the specification of response variables and factors
Restrictions: consistency check
Automatic rating of the feasible experimental designs
Analysis report in easily comprehensible wording
Hints to potential problems (outliers, disturbing factors, insufficient model fit, confounding of effects)
Statistics details linked within the report
Support in the case of incomplete results
Support in the transition to the next experimental cycle
Suggestions for confirmatory

A Multitude of Experimental Designs

STAVEX 6.0 supports a wide range of screening, modelling, and optimization designs, including various factorial and response surface designs, specialized mixture designs, and the ability to incorporate custom or external designs.
Screening, modelling and optimization designs:
Lin, Lin-Plackett-Burman, Plackett-Burman, Desperado, 2-level (full and fractional) factorial, 3-level factorial, central composite (CCD), Box-Behnken, hexagon, pentagon, Doehlert, D-optimal
Mixture designs:
D-optimal, simplex-vertex, simplex-lattice, simplex-centroid, axial, and projected designs, as well as designs enabling the simultaneous treatment of mixture and process factors
User-specified designs:
your own design specification, e.g. incorporating already existing experiments and results, designs from literature

Easily Understandable Analysis

Easily understandable analysis report with fully formulated sentences in clear language
Statistics details flexibly accessible by opening report sections in a mouse click
Separate models per response variable
Regression models including model diagnostics summary and plot recommendations
Automatic analysis of transformations for model improvement
Discriminant analysis (cross-validated) for the analysis of qualitative response variables
Best compromise calculation using a desirability function to combine several response variables
Information on the confounding of effects
Confidence intervals for predictions

Versatile Graphics

Very readable modern-looking graphics with informative dynamic tooltips to facilitate their interpretation
Large graphics library for response variables: two- and three-factor contour plots (also for qualitative response variables), response surface (RSM) plots, curve plots, half-normal plots
Graphics for formulation problems: ternary contour plots
Specific colour marking for out-of-specification (OOS) areas
Multiple-response contour plots: overlay plots for several response variables, plots for the desirability function – both suitable for represent the Design Space in the context of Quality by Design (QbD)
Model diagnostic plots: normal quantile plots, residual plots, and raw data plots (response variables vs. factor levels)
Plot matrix representations enabling to represent several graphics in a same tab (with a common scale for response variables): the impact of up to 4 factors (up to 5 in the case of ternary contour plots) can be visualized and easily analyzed at once
Interactive features: informative dynamic tooltips with the value of factors and response variables at the mouse position, definition of additional confirmatory runs in a mouse click
Plot export in different sizes possible, moreover in PNG or SVG format, for effortless integration in reports or presentations without subsequent resizing
Easy management of the generated plots for their effortless new visualization at a later time or for deleting those not needed anymore

Many Result Visualization Options in STAVEX

Further Features

Flexible reporting functions to generate DoE project reports (in PDF or HTML format) that contain exactly the information you wish (incl. graphics) and in the order desired
Clear interface for the management of multiple projects, shared directories for easily exchanging information with colleagues
Friendly and competent helpdesk (in English, German and French) for quick advice in case of questions – per phone, e-mail or web meeting

Installation, Licensing and Support

Single-user local installation and multiple-user server installation possible (named user concept)
Light version without mixture factors available
Easy online license file request, direct activation over the STAVEX user interface once received
Maintenance (updates, access to helpdesk for installation and support) included during 12 months after purchase date

Concept

So Easy Is Statistical Design of Experiments (DoE) Using STAVEX

The Concept of STAVEX
The integrated expert system guides you through all stages of sequential Design of Experiments (DoE).

DoE Project Definition
Entering:
the quality criteria which are to be optimized ("response variables"),
the potential influence factors, and
restrictions, if necessary.

Learn more

Most Effort Before Starting the Experiments

At the beginning of each DoE (Design of Experiments) project, the objectives have to be specified exactly. This in particular means deciding which quality criteria (response variables or quality attributes) and which potentially influential factors are to be investigated.

STAVEX focuses on the project, i.e., when using the software, you logically also start by entering the response variables and the factors. The high flexibility of STAVEX makes a practical specification easy. Where possible, the integrated expert system supports you through short messages, for instance when a larger number of levels of a qualitative response variable or of a qualitative factor would enlarge the experimental design substantially.

Response Variables

Defining the Response Variables

The aim of DoE (Design of Experiments) is to optimize one or more quality criteria.
STAVEX allows entering

Quantitative response variables (yield, byproducts, viscosity...), as well as
Qualitative response variables (colour: white / yellowish, thermosealing tight: yes / no...).

If several response variables have to be combined, a desirability function can be used for finding the best compromise setting. In the specific context of the pharmaceutical and biotechnological industry, this allows obtaining the Design Space very easily, as recommended under the Quality by Design (QbD) paradigm (FDA PAT and ICH Q8 guidelines). A typical application is the optimization of a dissolution profile – often described using several response variables which represent each the proportion of dissolved product at a different timepoint.

Setting the Optimization Objectives

You can easily enter the desired optimization direction:

Maximum (yield, purity...),
Minimum (impurities, quality deviation...), or
Target value (coating thickness, particle size, dissolved proportion after 2, 4, 6 or 8 h...),

together with their lower or upper specification limits, if applicable.

A last note: a precise analysis requires comparatively more experiments for qualitative response variables than for quantitative response variables. Therefore, STAVEX suggests choosing a quantitative scale if possible (rating on a scale of 0 to 100 instead of good / mediocre / bad). This allows for a higher flexibility in the judgement of intermediate states: for the software, as well as for yourself.

Potential Influence Factors

Entering Factors

STAVEX offers a great flexibility for entering the factors. All factor types can be investigated within a same experimental design:

Quantitative (temperature, concentration, throughput...)
Qualitative (supplier, type of catalyst, enzyme...)
Alternative (either acid A at a concentration between 5% and 10% or the stronger acid B at a concentration between 2% and 6%)
Mixture components belonging to one of possibly several mixture factor classes, in each of which the respective factors add up to a fixed value (e.g. 100% or 220 mg).

Moreover, STAVEX allows taking various real-life restrictions into account:

Alternative factors (process factors or factors within a same mixture factor class): either 3 to 6 mg of catalyst A or 1 to 5 mg of catalyst B
Ratios: the mole surplus of some educt must be at most 1.4; strawberry and raspberry aromas are to be used in a ratio that must lie between 1:2 and 2:1.
Linear restrictions:
- In order to be sure that the reaction is over, the waiting time after the reaction must be at least 30 times the reaction time.
- Some components of a formulation together should not exceed a certain proportion, e.g. there should not be more than 5% of preservatives.
Accounting for increments: for instance, the throughput can only be increased in steps of 1000 units/h.

The Right DoE Design for Any Situation
STAVEX implements the approach of sequential Design of Experiments (DoE) consequently. Depending on the number of potential influence factors, STAVEX suggests suitable experimental designs for the specific type of question.

Learn more

Sequential Design of Experiments (DoE)

Using the DoE tool STAVEX, you do not need any specific knowledge in statistics for deciding on an appropriate experimental design. The software generates a list of suitable experimental designs depending on the input you have entered. Among them, the experimental designs are presented from best to worst, and a rating is displayed to indicate their quality for the problem addressed. If you do not have any preferences, you can just select the topmost design of the list, and you will achieve good results.

Obviously, you are free to choose another experimental design (e.g. with more experiments for a higher precision) or to further adapt the experimental design selected according to your needs.

Sequential Design of Experiments (DoE)

The DoE tool STAVEX implements the methodology of so-called sequential Design of Experiments. Depending on the number of potentially influential factors, STAVEX proposes different suitable experimental designs.

DoE Screening Stage (Recommended for More than 8 Factors):
Here, only a relatively small number of experiments is performed, compared to the number of factors considered. This also means that relatively few information will be obtained. However, the only aim at this DoE stage is to reduce the number of factors so as to be able, at the next stage, to analyze the important ones in more detail.

The following main designs are available: Lin, Lin-Plackett-Burman, Plackett-Burman, 2‑level (full and fractional) factorial, D‑optimal, and Desperado designs.

DoE Modelling Stage (Recommended for Approximately 4‑8 Factors):
At this stage, an intermediate number of experiments, compared to the number of factors considered, is used. A linear model with interactions is fitted. The aim of this DoE stage is to identify the most essential factors, so as to be able, at the next stage, to analyze only them in detail for finding their optimal setting.

The following main designs are available: 2‑level (full and fractional) factorial and D‑optimal designs.

DoE Optimization Stage (Recommended for Approximately 1‑3 Factors):
Here, a quadratic model is fitted. These models allow identifying minima (lowest point of a "bowl") and maxima ("peaks"). Consequently, at this stage, the optimum is finally found. This cannot be achieved at the modelling stage, as there the model is not quadratic, but linear.

The following main designs are available: 2‑level (full and fractional) factorial, 3‑level factorial, central composite (CCD), Box‑Behnken, hexagon, pentagon, Doehlert, and D‑optimal designs.

Flexible Experimental Designs for Special Real-World Cases:

Within the above framework of sequential Design of Experiments, several design variations are possible.

Mixture Designs:
In a formulation or mixture problem, all components must add up to a fixed sum, e.g. 100% or 250 mg. This implies that the factors cannot be varied independently anymore. The statistical experimental design (and the later statistical analysis) needs to take this into account.

The following designs are then available: D‑optimal, simplex-vertex, simplex-lattice, simplex-centroid, axial, and projected designs, as well as designs enabling the simultaneous treatment of mixture and process factors.

Experimental Designs Specified by The User:
Independently of the problem setting considered, experimental designs can also be specified by the user. This enables using STAVEX for the analysis of already existing data, e.g. for experimental results obtained without using an experimental design or taken from earlier reports or literature references.

A common further use case is to let the DoE tool complete a few preliminary experiments – e.g. performed to identify good factor ranges – to a sound design. In this way, the initial experiments get reused, which enables keeping the total experimental effort low even more.

Easy Experimental Data Handling in STAVEX

Easy Data Handling
The simplest possibility of entering results is to import the data directly from Excel into the STAVEX results table. Even if problems have arisen when performing the experiments, causing e.g. incomplete results, this of course does not mean that the whole work was useless.

Learn more

Flexible Entering of Data

Entering the Experimental Results of the DoE

Data available in Excel

It is of course possible to enter all experimental results of the DoE directly into the Results table. However, you can also import them very easily from Excel. If STAVEX is configured to work together with automated laboratory devices, this step may even become useless.

Data import to STAVEX

Deviations in the Factor Settings

You accidentally used 65°C instead of 70°C in some experiment? If you neglect this fact, the analysis outcome will be biased. But an easy solution exists: a few clicks are sufficient to correct the respective experimental setting in the design, before finally entering or importing the DoE results.

Results imported into STAVEX

Incomplete Results

Various reasons can lead to a preliminary stop of the experimentation: some experimental settings surprisingly do not yield any measurable results, or the analytical instrument breaks down at the last analysis. Depending on the situation, it is possible to adapt the design by adding appropriate restrictions to the factor specification. However, STAVEX also allows stepping forward to the analysis directly. The DoE tool then checks whether the available results suffice for executing the statistical analysis. If yes, it gets performed; in the other case, STAVEX suggests suitable additional experiments. Another option is to perform the analysis at an earlier stage of the sequential DoE approach (e.g. screening instead of modelling), thus renouncing to some details.

Easy and Intuitive DoE Analysis
STAVEX performs the entire statistical analysis of the experimental results and supports you in its interpretation. The analysis report is written in an easily understandable language with fully formulated sentences; the desired statistical details can be displayed by accessing the corresponding sections.

Learn more

We Speak Your Language!

In particular for the analysis of experimental results, it is very important to present the outcome in an easily understandable way. Specialists in other fields should not need to become statistics experts in order to use Design of Experiments (DoE) successfully.

STAVEX performs the full statistical analysis of your experimental results for you and assists you for its interpretation. The analysis reports are written in clear natural language; the details of the statistical analyses are available by opening specific report sections if needed. The various graphics allow for an intuitive understanding of the results.

Analysis Summary

Selecting the summary provokes the display of a table providing an overview of the main analysis results for each response variable. In this way, you can easily recognize synergy effects and conflict potential. For each response variable, the reliability of the respective model is indicated.

Analysis Report (Short Form)

Top of the analysis report (here for modelling)

The short form of the analysis report displays all the most relevant results in a structured and easily understandable way. The corresponding graphics can be created simply using links located at the appropriate place; they can also be produced using the green generic "Create plot" button.

Screening:

Influence factors, presented in decreasing order of importance
Warnings in presence of outliers or of disturbing uncontrolled factors

Modelling:

List of the factors identified as (probably) important by the statistical analysis
Additionally, less clear candidates are shown as "possibly important"
Model equation
Model diagnostics summary table, including a transformation suggestion if appropriate
Model diagnostics plot recommendations to easily examine model quality problems if needed
Best level combination (to show the tendency)

Optimization:

Model equation
Model diagnostics summary table, including a transformation suggestion if appropriate
Model diagnostics plot recommendations to easily examine model quality problems if needed
Optimum (within the factor region and globally), with confidence interval
Analysis of all confirmatory runs performed

Detailed statistics

You can access any necessary additional statistical results by opening individually more sections of the report than in the short form. Also here, links to create the respective graphics are placed in appropriate section, so that they are all easily accessible.

Screening:

Table of effect sizes
Test for outliers and for uncontrolled factors
Confounding table

Modelling / Optimization:

Table showing, for each effect considered in the model, the calculated coefficient (both in the original factor scale and after factor range standardization), including its standard deviation, and the p-value
Correlation matrices of the model parameters
Coefficient of determination (R²) and adjusted coefficient of determination (R²_c)
Experimental results, model predictions and resulting model deviations
Test for non-normality of the model deviations (Shapiro-Wilk test, graphical analysis)
Analysis of variance for the model deviations (including Bartlett test for differences in the means per factor level and F-test for variance inequality)

A Multitude of Graphics
An appropriate graphical representation facilitates the interpretation of the obtained results significantly. STAVEX offers a wide choice of graphics.

Learn more

Visualization Made Easy

Contour Plots

Contour plots best enable to visualize the impact of several factors on a response variable. STAVEX proposes them as two- and three-dimensional graphics. In both cases, a convenient dynamic display tooltip allows accessing easily the values of the represented factors and of all response variables at any mouse position.

Two-Factor Contour Plot
A usual two-dimensional contour plot facilitates the definition of the Design Space at process development or characterization. the process validation. Zones in which a specification limit (if defined) is violated are coloured in dark red.

Three-Factor Contour Plot
Here as well, the response variable is plotted using a colour scale, but depending on three factors. The interactive plot allows turning the cube, so as to present the most interesting region in the foreground.

Plot Matrix Representations

For any plot type, several similar plots can be combined to form a plot matrix representation by varying the value of up to two additional factors. The common colour scale facilitates very much the interpretation, as the colour differences from plot to plot clearly show which factor settings are inappropriate or rather promising.

Here are examples for the contour plots. Dynamic tooltips are also available for such plot matrices.

Two-Factor Contour Plots
A matrix of two-factor contour plots enables an easy analysis of the impact of up to four factors at once.

Three-Factor Contour Plots
A matrix of three-factor contour plots enables an easy analysis of the impact of up to five factors at once. Each interactive plot can be turned individually.

Response Surface (RSM) Plots

Various plot types in which the response variable is not only represented by the colour scale (as in the above contour plots), but also assigned to an axis, are available as well. Especially the surface plots can be considered advantageous for more spectactular representations, e.g. in reports or presentations, or by anybody disliking the flat look of two-factor contour plots.

Surface Plot
Plotting the response surface on the vertical axis allows an intuitive understanding of the response variable behaviour depending on the values taken by two factors.

Curve Plot
Each such plot visualizes the impact of a single factor on a response variable for fixed values of the other factors. Thus, in presence of interactions, curve plots must be interpreted with caution.

Ternary Contour Plot for Formulation Problems

Mixture factors are often needed to account for the constant-sum restriction that the components of a formulation have to fulfil. In such cases, if the values of all factors but one are known, the last one must amount to the complement needed to reach the intended sum. Also the contour plot shape must account for this.

Contour Plot for Three Mixture Factors
Such a plot represents the impact of three formulation components on a response variable. The value it takes depending on the formulation is indicated by the colouring.

Qualitative-Response Contour Plots

Also qualitative response variables can be represented graphically as contour plots. The colours correspond to the possible values of the response variable. From the pictures below, it can easily be seen that the information that can be obtained is less precise for qualitative responses than for quantitative ones.

Two-Factor Contour Plot
As in the quantitative case, the plot visualizes the impact of two factors on a response variable. To facilitate the interpretation, green is always assigned to the optimal response level.

Three-Factor Contour Plot
The response variable is plotted here to show its dependency on three factors. Like all other three-dimensional plots, the visualization can be turned freely.

Multiple-Response Contour Plots

Real-world applications often include several response variables. In such cases, both the optimal and the acceptable factor settings most often depend on the variable considered. Features allowing the joint visualization of the factor effects on several response variables at once then provide a very valuable help for interpreting the DoE analysis results globally.

Overlay Contour Plot (Multiple-Response Design Space)
Several response variables are visualized simultaneously. This enables an easy and clear assessment of the factor region – on a cream background – in which all response variables are within their specification limits.

Desirability Contour Plot (Multiple-Response Design Space)
The best option to identify an ideal common factor setting for several response variables is to use a desirability function. It can be visualized just as the other response variables: in the red zone, at least a response variable is out-of-specification (OOS).

Model Diagnostic Plots

The quality of a regression model can be assessed using two types of graphical representations: normal plot of model deviations (quantile plot) and model deviations vs. factor levels (residual plot). They allow for a quick discovery of outliers or undesired structures among the model deviations.

Normal Plot of Model Deviations
This plot enables to easily judge whether the hypothesis that the model deviations are normally distributed is fulfilled. If yes, the red points shall more or less build a straight line. The plot also includes the p-value of the related Shapiro-Wilk test.

Model Deviations vs. Factor Levels
The residual plot (here, a matrix combining several) in particular enables to detect whether a nonlinear model (or, equivalently, a transformation of the response variable) would be more appropriate.

Half-Normal Plot

In the screening stage, whenever applicable, the half-normal plot is the tool of choice to visualize and interpret the analysis results. It enables to identify important factors and also to detect whether the experimental results include outliers or have been influenced by some uncontrolled factor.

Half-Normal Plot
In the example presented, the red points to the right of the green line indicate the seven factors – among eleven – found to be important. Other graphical arrangements would indicate the existence of outliers or an uncontrolled factor.

Raw Data Plot

If the statistical analysis of the experimental data does not provide any clear view of the factor impact, looking at the results themselves may be very helpful. The raw data plots just visualize them to find out how they are influenced by each single factor.

Raw Data Plot
Each plot of the matrix presented here contains a small blue triangle for each experiment – with its number indicated – and the mean result obtained for each factor level appears as an empty grey diamond.

Assistance for Planning the Next Steps
STAVEX is the only DoE tool which guides you through the entire process of sequential Design of Experiments (DoE). After completing an experimental cycle, STAVEX supports you in the decision of how to proceed further. Eventually, the entire project is documented within a single file.

Sequential DoE: Factor Reduction in STAVEX

Applications

Technical Details

Applications

Who Is STAVEX Intended For and When Should It Be Used?

Users

Scientists trying to improve a process or product
Chemists
Chemical engineers
Process engineers

Typical Situations

Process and product development
Process scale-up and optimization
Analytics
Formulation development

Some References within the STAVEX Community

Life Sciences

Aarti Industries
Acino Pharma
CSL Behring
Hikma Pharmaceuticals
Idorsia Pharmaceuticals
Novartis
R-Biopharm
Sanofi
Siegfried
Zoetis

Fine Chemicals

Alcan Airex
BASF
Clariant
DSM
Ems-Chemie
Givaudan
Huntsman
Merck
Mifa
NOCIL
Syngenta

Construction & Coating

BASF Construction Chemicals
Baumit
Emil Frei (Frei Lacke)
Hesse-Lignal
Isofloc
Röfix
URSA Insulation

Wood

Danzer Deutschland
Gutex

Food

Danisco
Gelita
Nestlé

Equipment Suppliers

Frewitt
Glatt
Synventive Molding

Other Sectors

Bystronic Laser
Compact Dynamics
ThyssenKrupp
Weidmüller Interface

Research Institutes

Fraunhofer Institute for Building Physics
Fraunhofer Institute for Chemical Technology

Universities

HES-SO Valais-Wallis
Ludwig Maximilian University of Munich
Martin Luther University Halle-Wittenberg
Ostwestfalen-Lippe University of Applied Sciences and Arts
University of Applied Sciences and Arts Northwestern Switzerland
University of Basel

Technical Details

Technical Requirements

Single-User

On your network:
- The TCP/IP protocol and DNS must be installed (it does not work with Novell or anything similar).
On the user's computer:
- Windows 7, 8, 10 oder 11
- At least 16 GB RAM and 500 MB disk space
- Modern web browser (version of e.g. Firefox, Edge, or Chrome which is not older than 2020)
- An Internet connection is only necessary for accessing the STAVEX help and for ordering the license file.

Multi-User

On your network:
- The TCP/IP protocol and DNS must be installed (it does not work with Novell or anything similar).
On the server:
- Windows Server 2016 (or more recent version), Windows 10 or Windows virtual machine
- At least 1 GB RAM and 500 MB disk space (for up to 10 users; otherwise 25 MB more per additional user)
- Permanently running hardware, so that STAVEX 6.0 can run continuously as service (the necessary web server is already part of STAVEX and no web certificate is required).
- An Internet access is only necessary for ordering the license file (see the corresponding section below).
- The management of the STAVEX user accounts occurs over a file which is accessed by the software and can be modified (by STAVEX administrators) over its user interface. The data which is input by users or generated by STAVEX is also managed internally in form of files. Therefore, no database is necessary.
On the client computers:
- Modern web browser (version of e.g. Firefox, Edge, or Chrome which is not older than 2020)
- An Internet connection is only necessary for accessing the STAVEX help.
- STAVEX does not require the Java runtime. Only version ES6 of JavaScript is used, which should be implemented in all usual web browsers since 2015.

STAVEX Releases

Current Release

Release 6.0 is available since June 2024. It is compatible with Windows 7, 8, 10, 11 and Windows Server starting 2019 and offers many enhancements of the functionality.
New Features:
- STAVEX 6.0 is the first fully web-based DoE software
- Complete redesign of the already user-friendly interface
- Enhanced design rating tools and project management features
- Modern and readable visualizations
- New flexible reporting functions
Courses:
- 4-day and 2-day courses are available (in English, German and French)

Current Subrelease

Release 6.0.348 is available since July 2025 and offers the following improvements:
- More readable informative dynamic tooltips in all 3D graphics
- Better management of the experimental results during the transition to the cycle with confirmatory experiments
- Improved representation of some graphics
- Improved usability of the table containing the experimental results
- Various minor improvements

Previous Releases

5.2 (October 2013)
- Compatible to current Windows 7, 8, 10 versions (32 bit and 64 bit)
- Improved graphics export: now also an automatic positioning of the legend is possible, as well as high resolution output
- Improved optimization algorithm (increased precision)
- Facilitated licensing scheme, now also via Internet
- Various small adaptations
5.1.1 (October 2010)
- Compatibility to Windows 7
- Various small improvements
5.1 (January 2008)
- Alternative process factors
- Tetrahedron plot: response variable in dependence of 4 mixture parameters
- Graphics improvement in the 4-D plots
- Back transformation: mathematically transformed response variables are now shown in the original scale
- Including previously performed experiments: this allows "recycling" of earlier experiments within the sequential experimental approach
- Automatic adaptation of the mixture class sum after factor reduction
5.0 (April 2006)
- 4-D contour plot for a response in dependence of 3 factors: allows screening through the factor space and localizing the optimum within a few seconds
- Enlarged library of experimental designs
- Analysis of incomplete experiments
- Assessing the blocking effect
- Calculation of Costs

Looking To Explore STAVEX Or DoE?

Training

Consulting

Knowledge

DOWNLOAD

Flyers

Case Studies

Tutorials

Flyers

STAVEX 6.0 Flyer

Read more about how STAVEX 6.0, the first web-based DoE tool for the Intranet, supports you in making your experiments run more efficiently.

News STAVEX 6.0

Discover the improvements in our latest release, STAVEX 6.0. New features include more modern and readable representations, expanded reporting and project management features and a more intuitive user experience.

Usability STAVEX 6.0

Explore how STAVEX 6.0 revolutionizes DoE usability. Backed by university research, this newest version was developed to create a smooth process and intuitive user experience.

Case Studies

Methodological information:

Subject		Title		Download
Design of Experiments - Data Mining: which method for which purpose?		Process optimization: with data or without?		DE EN FR ES
Increasing quality using Design of Experiments in the PAT context		Quality by Design in pharmaceutical development - the contribution of Design of Experiments		DE EN FR ES

Pharmaceutical industry:

Subject	Title	Download
Definition of the Design Space (University of Applied Sciences and Arts of Ostwestfalen-Lippe)	Defining the Design Space	DE EN FR ES
Optimizing a crystallization (Novartis Pharma)	Crystal clear results	DE EN FR ES
Producing child-resistant blister packages	Away from the red zone	DE EN FR ES
Optimizing a tablet formulation (Novartis Pharma)	Straight from lab to market	DE EN FR ES
Design of Experiments for the development of emulsions (University of Applied Sciences and Arts of Ostwestfalen-Lippe)	Two in one sweep	DE EN FR ES
Ideal film coatings for tablets (University of Applied Sciences and Arts of Ostwestfalen-Lippe, DIOSNA Dierks & Sons)	Statistical Design of Experiments for determining process parameters in film coating processes in the Vertical Centrifugal Coater 3-15	DE EN FR ES
Compliance with particle size specifications in milling (Siegfried)	Searching for the optimum	DE EN FR ES

Biotechnology:

Subject		Title		Download
Optimal composition of culture media in biosynthesis		Gourmet mix for bacteria		DE EN FR ES

Scale-up:

Subject		Title		Download
From laboratory to production scale (Novartis Pharma)		Good planning: the key to success!		DE EN FR ES
Quick determination of the process parameters ensuring an optimal dissolution profile for modified-release pellets (Glatt)		Quality by Design		DE EN FR ES

Chemical industry:

Subject		Title		Download
Optimizing polyamides for offshore tubes (Ems Chemie)		Flexible development		DE EN FR ES
Optimizing an additive combination for rigid PVC (Alcan Airex)		Successful mixture		DE EN FR ES

Biogas:

Subject		Title		Download
Optimizing the gas yield through process settings and the selection of enzymes		Biogas: statistics instead of lottery		DE EN FR ES

Insulating materials:

Subject		Title		Download
Design of Experiments in the production of insulating materials (URSA)		DoE in the production of insulating materials		DE EN FR ES

Coatings:

Subject		Title		Download
Optimizing coating formulations		The taming of the shrew		DE EN FR ES

Food industry:

Subject		Title		Download
Design of Experiments for optimizing Tilsit cheese (Technical Universität Berlin)		Efficient experiments		DE EN FR ES

Car industry:

Subject		Title		Download
Optimizing hub-shaft connections (ThyssenKrupp Presta)		Less effort - more strength		DE EN FR ES

Tutorials

Design of Experiments (DoE)

The small video series gives an introduction to the most important concepts of Design of Experiments (DoE). This eliminates the most usual uncertainties.

Play Video

What is DoE?

Play Video

Good Bye OFAT

Play Video

Sequential Design of Experiments (Structure of STAVEX)

Play Video

Sequential Design of Experiments (Structure of STAVEX)

Play Video

Design Selection

Play Video

Plan Selection

DoE example: from the first screening to the final confirmatory experiment in about 30 runs

An example from process optimization, but which is easily transferrable to product optimization issues.

Play Video

Screening

Play Video

Modelling

Play Video

Optimization (Response Surface Modelling - RSM)

Play Video

Confirmatory Experiments

Lorem ipsum dolor

Lorem ipsum dolor sit ametasd cnur adipi olor sit amet dolor consectetur sollicitudin lacus.

Play Video

Lorem Ipsum Dolor

Play Video

Lorem Ipsum Dolor

Play Video

Lorem Ipsum Dolor

Play Video

stavex

Making DoE Easier than Ever.

Easy Design and Analysis of Experiments With STAVEX

Flexible User Specifications

User Guidance

A Multitude of Experimental Designs

Easily Understandable Analysis

Versatile Graphics

Further Features

Installation, Licensing and Support

So Easy Is Statistical Design of Experiments (DoE) Using STAVEX

The Concept of STAVEX

DoE Project Definition

The Right DoE Design for Any Situation

Easy Data Handling

Easy and Intuitive DoE Analysis

A Multitude of Graphics

Assistance for Planning the Next Steps

Who Is STAVEX Intended For and When Should It Be Used?

Users

Typical Situations

Some References within the STAVEX Community

Life Sciences

Fine Chemicals

Construction & Coating

Wood

Food

Equipment Suppliers

Other Sectors

Research Institutes