Many Multiframe users ask us how we know that Multiframe produces the correct results.  This following explains how FORMSYS has verified that Multiframe gives accurate results and what steps we take to make sure that each version of the software we ship is as reliable as possible.

Quality Principles

While it is impossible to ensure that any software product is completely free of bugs, we follow a series of engineering and testing principles and procedures to ensure that Multiframe will produce results which are consistent with the level of accuracy and thoroughness a professional engineer applies to design work.  To this end we follow a development and testing path which includes use of structured programming techniques, verification of the underlying algorithms, testing of the computer implementation of those algorithms, testing of real world problems in-house and beta testing in the field at Multiframe user sites.

Structured Programming

The best defence against bugs in software is to use structured programming techniques that have been proven to improve software reliability.  Without going into the technical details of our software development methodology, we summarize by saying that we utilize structured code, object oriented design, data hiding and encapsulation and fault tolerant programming practices to enhance our software's reliability.  Multiframe is a complex software system of over 400,000 lines of code and we believe our history of reliability reflects the effort we have put into using reliable coding practices.

Verification of Algorithms

When new design or analysis algorithms are introduced into Multiframe, we first carry out testing on the algorithms on proven test cases with known analytical solutions.  These generally come from engineering texts such as Refs 1 to 3.  These test cases will include samples which independently examine the various degrees of freedom (Mx bending, My bending, Axial tension etc. ) followed by examples which superimpose the effects of multiple degrees of freedom.  These simple test cases are performed for structures aligned with the principal axes as well as those rotated to arbitrary angles.

Testing of Implementation

Once the basic algorithms have been proven correct, testing is then carried out on more complex sample problems to which a solution has already been established using a proven analysis program.  These results may either come from structural engineering texts such as Refs 1 to 3 as well as from other results carried out by Formation Design Systems or other engineers using other software products such as SAP, Nastran etc.

Testing of Upgrades

As each new version of Multiframe is released we perform a series of tests to ensure it functions correctly.  Among these tests is an analysis of a frame which exercises every different feature of Multiframe i.e. every possible member release, section type and orientation,  load type, combination etc.  At each release the results from this frame are compared with the results from the previous release to ensure conformance with answers which have been established as being correct.

Beta Testing

Immediately prior to the release of each new version of Multiframe, we conduct a beta test of the software.  This involves sending the software to practicing engineers and having them use it on design work in progress to determine its reliability for actual design use.  These beta testers provide us with feedback on the reliability and accuracy of the program as well as its useability and suitability for everyday work.  Once the beta test program is completed and all testers are happy with the program, we begin shipping the commercial version.

Version Control

Each new version of Multiframe displays a version number indicating the version and the date the software was first shipped.  If the version is a development, alpha test or beta test release, the version number may also include a letter and number suffix indicating the type and number of the release.  A development version is usually only for internal use and is a very early demonstration of a possible new product or feature.  It is highly experimental and not reliable. An alpha release is a first public release of a program for initial testing and comment, it is not reliable.  A beta release is a final test version of the program released for field testing prior to commercial release. It is mostly reliable but may contain some bugs.  A commercial release is a completed, debugged program reliable and ready for professional use.
For example
1.0d1        The first development release of version 1.0
1.5a2        The second alpha test release of version 1.5
1.6b2        The second beta test release of version 1.6
1.64          A commercial release of version 1.64

But we're not Perfect

We make every effort to ensure that our software will meet our users' needs and perform accurately.  However, as with all complex software systems, it is possible for errors to occur.  If you suspect a problem with Multiframe, please contact our technical support staff by email at support@formsys.com and explain what you believe the problem to be.  In the unlikely event of a problem being found, we will correct it as soon as practicable, and send you a new corrected version of the program.

 To get accurate results from Multiframe, it is necessary for you to model the problem correctly and to correctly interpret the results produced.  This requires structural engineering experience combined with an understanding of matrix structural analysis.  It is the users' responsibility to correctly model the structure and assume responsibility for the results.