provides
Linear Program Validations

The linear programming model should be the primary optimization tool for the manufacturing plant. As such it is important that this optimization tool be a good representation of the plant’s capabilities. Qualities of a good linear programming model include: 

·        Accurate representation of the plant's manufacturing equipment.
·        Close representation of the plant’s actual yield performance.
·        Reliable execution with output capable of being interpreted.
·        The organization accepts results and has confidence in the model.

Linear programming models have grown in complexity as organizations demand more rigorous representations of operations. Validation of a linear programming model by an independent review is, in many ways, analogous to proofreading a large document. Typical models have row and column counts well over a thousand and many thousands of coefficients.  A single wrong coefficient can significantly affect optimization results, and it is highly unlikely that one person will catch every oversight. Just like proofreading by persons other than the author focuses on grammar and meaning, validation reviews by outside experts focus on the techniques and formulation of the problem statement. In both cases the contents are still the author’s prerogative. Good validation reviews by others should include comments on data that seem suspicious and any correlation that is out of the normal range.

Linear program validation is a three step procedure. The steps are:

·        A review of the mathematical structures implemented.
·        A consistency and completeness check of the model’s data.
·        Evaluation of the effectiveness of modeled representations for optimization.

The first step checks for proper implementation of sophisticated forms of mathematical structure such as the delta-base modeling technique. This determines if the model can work properly as intended.

The second step looks at the data. Missing or inconsistent data often cause problems in optimization models and, even if there is a mathematical solution, the results may be questionable operationally.

Finally, an optimizer will take advantage of any “loop hole” as it seeks the optimum solution for a least cost or maximum profit. Validation reviews identify and recommend methods for correcting any flaws in model formulation.

We are uniquely suited to provide linear programming model validation services. We have a great deal of experience building and using these systems in operating environments. Our proprietary modeling and optimization software provide us with a unique tool for use in this work. It allows us to do many things with an efficiency that other tools cannot match. We can help raise the quality of a client’s linear programming model to meet current or future expectations, or provide assurance that it is already there.