Experiments are conducted in order to observe how changing a single factor or multiple factors affect the response variable(s). The purpose of experimental design or DOE is to create an effective procedure to create efficient experiments to make it much more easier to reach objective conclusions. Experimental design has three basic principles and they are briefly described below.
Randomization
This is the first principle of experimental design. It is a simple process of randomly allocating changes to the factors. The purpose of “randomizing” it is to ensure that the probability of changing the factors remains the same. The main reason why “randomization” is a very important part of experimental design is because it makes the entire process unbiased and removes any external variation to affect the experiment. Example : Rolling dice , Picking a card at random from a deck , Flipping a coin etc.
Save your time!
We can take care of your essay
- Proper editing and formatting
- Free revision, title page, and bibliography
- Flexible prices and money-back guarantee
Repetition
This is the second principle of experimental design. It is basically repeating the basic experiment over and over again. Most of the time the experimental unit or factor does not remain identical. Hence repetition is used to remove the “variable” or said deviation among the factors. Replication is mainly used for two reasons:
- Secure an accurate estimate of the experimental error
- Increase precision
Local control
All extraneous sources of variation are not completely removed by repetition and randomization. This creates a need to make a more refined process to make the experiment more accurate and precise. In other words , a design should be created to ensure that all sources of variance are brought under control. There are two aspects of local control:
- Balancing: This merely states that the changes applied to the factors must be done in an orderly fashion in order to ensure that the results are a balanced arrangements of the changes.
- Blocking: This means that all the factors must be collected to form a rather identical or similar group.
Examples of blocking are as follows:
An experiment in a chemical process requires two batches or raw material for conducting the experiment runs. In order to minimize the effect of batch to batch variability , each batch of raw material would form a block.
The strength of a steel product needs to be increased. Four factors are being taken into consideration. Each factor is taken at two levels. An eight trial experiment is chosen but only four can be done in a day. Here each day can be treated as a block.
Design of experiments refers to the process of designing and analyzing the experiment so that precise and accurate conclusions can be drawn efficiently.
Applications of Design of Experiments:
Designing a metalworking coolant:
Houghton International Inc. used Design of Experiments (DoE) to develop a new metal cutting fluid that is capable of handling steel, cast iron, aluminum and cast aluminum in half the time compared to the previous models.
Houghton has improved their abilities to develop cutting fluids by applying DoE to the product development process. In moving to DoE based development , the chemists could focus on the actual chemistry and not bother about the statistical design. The company accomplished this goal by selecting statistical software that is designed for use by people without any statistical background yet still provides all of the analytical power needed for designing specialty chemical products. Design-Expert software from Stat-Easy ( a DOE software) made it so much easier for the chemists to come up with objective conclusions for their experiment which greatly helped them with creating their new product.
Producing ethanol from corn:
A supplier of industrial equipment wanted to market an existing product to companies producing ethanol from corn. There was evidence indicating that the device increased ethanol yield. However, the company wanted to find all the favorable factors would increase ethanol yield. The company contacted Stites & Associates to measure the performance of the device so that they could inform their customers about the device.
They first had to find an accurate measurement method . After an accurate measurement method had been found, they had to run tests to figure out what factors actually affected the response variables . Finally, researchers performed a design of experiments (DOE) and came to an objective conclusion. The conclusion that they had come across had greatly improved ethanol yield from corn farms all over the midwest.
Resin Manufacturing
Resin manufacturer Interplastic ,a major gel coat producer, recently developed a new gel coat for open molding. Early testing showed problems with the formation of holes in the gel coat’s surface. Craters can form because the gel coat lacks the right surface tension balance, which causes the coating to be repelled.
Interplastic’s technical director, Scott Crump, was tasked to create an experimental design to find out the right proportion of the ingredients while ensuring cratering does not take place and the good qualities of the gel are maintained.
They first started out with OFAT( one factor at a time) design and found out that they would not reach anywhere using that design as it would take almost an eternity to come to the right conclusion.“This approach is too slow,” says Crump, “particularly when we are faced with the challenge of developing a new product quickly to meet an urgent market need.”
Crump then decided to use the DOE system to come up to an objective conclusion. He opted to use the tool Design-Expert software to assist him and his team to come to an answer. Design-Expert created a 31-run experiment with 28 unique gel coat blends. By using the DOE system, it reduced the number of runs required to achieve the required level of accuracy by 50 percent compared to OFAT. After experimental test runs were over and they were able to find the right amount of ingredients required to create a fantastic and functioning product.
Fine-tuning flotation cells
FLSmidth recently installed two turnkey SuperCell flotation machines(the world’s largest flotation cells)at Rio Tinto’s Kennecott Utah Copper concentrator near Salt Lake City, Utah, USA. Before intensive testing had to be conducted in order to fine-tune the machines and test the performance of the cells. In this case FLSmidth used statistical Design of Experiments (DoE) software to test the response variables and the effectiveness of the cells under various conditions. “We heard that DOE might have the potential to reduce the time and cost involved in cell validation while also providing the potential to improve cell performance,” Darius Lelinski( product development manager on the project) said. “We looked at several different statistical software packages with DoE capabilities but they seemed to be targeted more at statisticians than engineers. Design-Expert software from Stat-Ease, on the other hand, is designed for people like us—engineers that want to solve a problem but don’t want to get neck-deep in statistical methods.”
The first supercell was constructed, commissioned and tested in a short period of just 110 days. According to Lelinski, they had never finished a project of this scale at this rate ever before.
Stuck on your essay?
