BLEND OPTIMIZATION

THE BLEND PROBLEM

When testing products to find the best blend product configuration, there are particular problems encountered for certain product parameters. It is difficult to find the best blend product parameters when the product is a "blend". For example, steel is an alloy (blend) of metals including iron, carbon and other constituents. In a blend the parameters must add up to a constant total, usually 100%. For example, consider a product which has only three (3) parameters expressed in percentages with the following current standards:

 

Parameter

% of Product

 X1

50%

X2

30%

X3

20%

Total Blend %

100

To find the best product blend, products with varying amounts of X1, X2 and X3 could be tested. However, when the amount of X1 is increased in a test product, what is done with X2? And X3? Presumably these constituents would be decreased, but by how much? An important problem in testing products is that:

WITH PRODUCT BLENDS, WE CANNOT VARY ONE (1) PORTION ONLY!

When one portion is varied, at least one other portion must be varied to keep a constant total (usually 100%). Thus analysis of the data may not show which constituents had effects on performance variables (responses). To compound the problem, there are usually more ingredient interactions in blend problems.

What is needed to solve this problem is a technology to handle multiple changes in product ingredients. This technology can be called "blend optimization". Through using blend optimization one can understand which parameter changes result in improved products.

HOW DOES BLEND OPTIMIZATION WORK?

1. A statistically designed test plan is developed showing the necessary test data to collect.

2. The test data is used to generate statistical models.

3. The models are then used to search through all possible products to find the best product.

A BIG advantage of using Blend Optimization is that: one can evaluate a large number of possible products without having to actually build these products.

TIME and MONEY are SAVED in finding the best product.

DOES Institute, Inc. has developed methods which show vital information on what is the best product blend – by conducting designed tests and using Blend Optimization technology. More importantly, DOES has developed proprietary software which finds the optimum product blend that meets all product specifications.

The software searches through all products to find the product design which maximizes consumer rating. In searching through possible products, the software actually simulates the building of hundreds of products without the cost of building and testing more products.

BLEND OPTIMIZATION THUS GREATLY REDUCES THE COST OF TESTING

The following graph shows how, through a search algorithm, a product with a consumer rating of .82 can be found.

 

Illustration of Product Search for Higher Yield

 

Blend Optimization is easy to use but requires planning and an organized, rigorous approach to product testing usually requiring the following steps:
Steps in Applying Product Optimization

1. Design product tests – statistically

2. Gather product test data
3. Define product constraints
4. Analyze data using software
5. Interpret results and select optimum product

In step #1 above, DOES, Inc. should be consulted with in choosing a valid test design. Step #2 is very important because the test personnel’s knowledge of the product, test equipment, test conditions, etc. Are invaluable in this process.

Using Blend Optimization results in better products at lower costs.

 

Back to Top

Home Clients Experience Staff References DOE Knowledge Services Contact

DOES, Inc. · 2531 Woodbine Road. · Winston-Salem, NC 27104 · Phone:(336) 624-2294
Fax
(336)-721-2441 email: Bill@Doesinc.com