^{1}

^{*}

^{1}

Multiple response surface methodology (MRSM) most often involves the analysis of small sample size datasets which have associated inherent statistical modeling problems. Firstly, classical model selection criteria in use are very inefficient with small sample size datasets. Secondly, classical model selection criteria have an acknowledged selection uncertainty problem. Finally, there is a credibility problem associated with modeling small sample sizes of the order of most MRSM datasets. This work focuses on determination of a solution to these identified problems. The small sample model selection uncertainty problem is analysed using sixteen model selection criteria and a typical two-input MRSM dataset. Selection of candidate models, for the responses in consideration, is done based on response surface conformity to expectation to deliberately avoid selection of models using the problematic classical model selection criteria. A set of permutations of combinations of response models with conforming response surfaces is determined. Each combination is optimised and results are obtained using overlaying of data matrices. The permutation of results is then averaged to obtain credible results. Thus, a transparent multiple model approach is used to obtain the solution which gives some credibility to the small sample size results of the typical MRSM dataset. The conclusion is that, for a two-input process MRSM problem, conformity of response surfaces can be effectively used to select candidate models and thus the use of the problematic model selection criteria is avoidable.

Multiple response surface methodology (MRSM) is when an industrial process with more than one response variable is investigated and studied through the analysis of reliably generated response models and corresponding response surfaces at some region of operability. Processes with two inputs provide a special case of MRSM problems whose response surfaces can be constructed in three dimensions and therefore can be analysed for conformity.

Hill & Hunter [

Myers et al. [

The first of three problems with industrial MRSM work is the uncertainty inherent in the process of model selection (Wit et al., [

The second problem concerns the small sample size bias problem of classical model selection criteria. MRSM is largely a regression modeling and model selection problem. Model selection is done through model selection criteria. Seghouane and Bekara [

Attempts have been made to correct some of the classical model selection criteria for the small sample size context. Suguira [

In accordance with the information-theoretic approach, a “best model” for analysis of data depends on sample size as smaller effects can often only be revealed as sample sizes increase. The amount of information in large datasets greatly exceeds the information availed by small datasets.

The small sample size bias problem and the one of model selection uncertainty are related in that they all have to do with the use of model selection criteria in choosing the best model. If the use of model selection criteria is avoided for the use of an approach with more certainty, both problems will be solved.

The third problem with industrial MRSM work is that most studies fall within small data analytics since they are based on analysing response models generated from datasets emanating from running designed experiments. Such experiments are designed to be cost effective and at the same time are expected to provide optimum information. It is true that most MRSM studies in industrial work use experimental designs which are within the small sample size context, sometimes below the order of (10 + k), to minimize cost. For example, most of the MRSM industrial examples used in Myers et al. [

The three problems encountered by MRSM practitioners related to the small sample MRSM dataset sizes are shown in

This paper focuses on the problems encountered with the use of model selection criteria in selecting best models with particular interest in MRSM. In other words, the focus is on obtaining the best result instead of the best models to use in optimization. In this way,

1) Sets of candidate models are selected based on conformity of response surfaces for each response,

2) Permutations of response model combinations are made,

3) Simultaneous “optimisation” is performed of response model combinations, and

4) The best result is obtained even by averaging.

The use of conformity of response surfaces to select candidate models for optimisation instead of model selection criteria is investigated in this paper. The use of conformity of response surfaces enables avoidance of model selection criteria and thus indirectly solves the model selection criteria uncertainty problem. In the same manner, the avoidance of the use of model selection criteria in model selection indirectly solves the worry of the small sample size bias problem. The selection of response candidate models based on their conformity of response surfaces enables the determination of candidate model combinations for optimisation. The permutation of response model combinations converts to the permutation of results which can be averaged to obtain the best result. This multi-model approach maintains the rigour required in model selection to ensure convincing and credible solutions for the small sample size problems in MRSM. The proposed MRSM theoretical framework of the current study is shown in

The remainder of this paper is organized as follows. Section 2 introduces the typical small sample MRSM dataset, which will be subjected to analysis in this study. Section 3 analyses the typical small sample MRSM dataset with sixteen model selection criteria taken from literature exposing the selection uncertainty problem in MRSM small sample sizes. In Section 4, the MRSM small sample size problem is resolved by a multi-model approach and the results are analysed. Section 5 looks at the validation of the results in line with the original problem of obtaining cure times of rubber covered mining conveyor belts for a Southern African manufacturer. Section 6 discusses findings and Section 7 concludes and proposes direction of future research.

The MRSM experimental design and results used for the current study are shown in

This dataset is chosen for use in this investigation because, in addition to being a typical small sample MRSM dataset, two factor experiments produce response models that have response surfaces that can be analysed. Where factors are more than two, it is difficult to construct response surfaces of models in three dimensions. So two factor processes are a special case of MRSM where it is possible to check response model fitness to data, prediction performance and analyse response surface model conformity to expectations before selection of “best” model for consideration in multi-objective optimisation. More complex problems require more complex ways to solve them.

All possible regressions modeling is applied to the dataset in ^{p−1}, where p is the total number of regressors) models for each of the two responses, that is, adhesion and hardness. ANNEXURE A shows the thirty one models of each response in detail.

This section analyses the problem of uncertainty that characterises the model selection criteria of response models. The multi-selection criteria approach is

Run | Random | X_{1} (T = TIME) | X_{2} (RT = RUBBER THICKNESS) | Ave. Adhesion (N/mm) | Ave. Hardness (˚shore A) |
---|---|---|---|---|---|

1 | 7 | 16 | 7.2 | 10.60 | 60 |

2 | 9 | 30 | 7.2 | 13.34 | 63 |

3 | 11 | 16 | 22.8 | 6.20 | 53 |

4 | 8 | 30 | 22.8 | 12.10 | 61 |

5 | 1 | 23 | 15 | 11.80 | 58 |

6 | 5 | 23 | 15 | 12.10 | 58 |

7 | 4 | 13 | 15 | 6.5 | 44 |

8 | 2 | 33 | 15 | 13.30 | 63 |

9 | 10 | 23 | 4 | 13.30 | 63 |

10 | 3 | 23 | 26 | 3.50 | 56 |

11 | 6 | 23 | 15 | 12.20 | 58 |

12 | 12 | 23 | 15 | 12.30 | 57 |

13 | 13 | 23 | 15 | 12.10 | 58 |

able to bring out the between-criteria uncertainty problem into the limelight.

The thirty one all possible regressions response models which are generated from the results of the MRSM experiment are analysed using sixteen different model selection criteria. The sixteen model selection criteria are used to select the best model for adhesion and hardness between the thirty one all regression models. Ten of the sixteen model selection criteria are corrected for small sample size inefficiency; three are not corrected and choose the best model fitting the MRSM dataset, while three choose the best model for prediction. The formulae and relevant details of the selection criteria are given in ANNEXURE B.

Model Selection Uncertainty 1: Model selection criteria do not necessarily agree on the best model

• Adhesion

There are two adhesion response models with the most selections in ^{2}], and the other with five, [T, T*RT, T^{2}, RT^{2}]. Another three response models have lower numbers of selections. In total the sixteen model selection criteria chose five different response models as “best”. The mere dispersion of the “best” selections all over

T, RT, T*RT, T^{2}, RT^{2} | T, T*RT, T^{2}, RT^{2} | T, RT, RT^{2} | T*RT, RT^{2} | T, T^{2} | |
---|---|---|---|---|---|

R^{2} | X | ||||

R^{2} (adj.) | X | ||||

RMSE | X | ||||

R^{2} (Pred.) | X | ||||

Adeq (prec.) | X | ||||

AICc | X | ||||

APCp | X | ||||

BIC | X | ||||

HQc | X | ||||

KICc | X | ||||

HQ | X | ||||

MAIC | X | ||||

MKIC | X | ||||

KIC (chm) | X | ||||

KIC (csb) | X | ||||

KIC (cc) | X | ||||

Count | 1 | 6 | 1 | 5 | 3 |

• Hardness

The model selection uncertainty problem is again evident with hardness as shown in ^{2}, RT^{2}], is selected by eleven different criteria whilst another two are selected for the remaining five times. Even the ten small sample size corrected criteria do not agree with five choosing the full model, three choosing [T, T^{2}] and two [T, T^{2}, RT^{2}]. This uncertainty is characteristic of the model selection process which selects one model as best.

Model Selection Uncertainty 2: Good fitness to data and/or good prediction performance does not imply conforming response surface

• Analysis of Response Surfaces

Response surfaces for each of the thirty one models are obtained and analysed as to whether they conform to expectations. Data matrices are constructed to confirm.

Only four adhesion response models have conforming response surfaces which are: [T, RT, T*RT]; [T, RT, T*RT, T^{2}]; [T, RT, T*RT, RT^{2}], and [T, RT, T*RT, T^{2}, RT^{2}]. Only one hardness model has a conforming response surface which is [T, RT, T*RT, T^{2}].The response surfaces are shown in ANNEXURE C.

From

T, RT, T*RT, T^{2}, RT^{2} | T, T^{2}, RT^{2} | T, T^{2} | |
---|---|---|---|

R. Sq. | X | ||

R. Sq. (adj.) | X | ||

RMSE | X | ||

R. Sq. (Pred.) | X | ||

Adeq. (prec.) | X | ||

AICc | X | ||

APCp | X | ||

BIC | X | ||

HQc | X | ||

KICc | X | ||

HQ | X | ||

MAIC | X | ||

MKIC | X | ||

KIC (chm) | X | ||

KIC (csb) | X | ||

KIC (cc) | X | ||

11 | 2 | 3 |

Adhesion Table | Hardness Table | ||||||
---|---|---|---|---|---|---|---|

MODEL | FIT | Prediction | RS | MODEL | FIT | Prediction | RS |

T, T^{2} | X | T, T^{2} | X | ||||

T*RT, T^{2} | X | T*RT, T^{2} | |||||

T*RT, RT^{2} | X | T*RT, RT^{2} | |||||

T^{2}, RT^{2} | T^{2}, RT^{2} | ||||||

T, RT, T*RT | X | T, RT, T*RT | |||||

T, RT, RT^{2} | X | T, RT, RT^{2} | |||||

T, T^{2}, RT^{2} | T, T^{2}, RT^{2} | X | X | ||||

T, RT, T*RT, T^{2} | X | T, RT, T*RT, T^{2} | X | ||||

T, RT, T*RT, RT^{2} | X | T, RT, T*RT, RT^{2} | |||||

T, T*RT, T^{2}, RT^{2} | X | T, T*RT, T^{2}, RT^{2} | |||||

T, RT, T*RT, T^{2}, RT^{2} | X | X | T, RT, T*RT, T^{2}, RT^{2} | X | X | ||

COUNT | 4 | 2 | 4 | 3 | 2 | 1 |

response surface at the same time. Best fit and/or best prediction does not ensure conforming response surface. This is the second uncertainty problem.

Model Selection Uncertainty 3: Good fitness to data does not necessarily imply good prediction performance

The other uncertainty problem is that a model with good fitness to data is not necessarily good at prediction performance. This is evident from the table. This is the third model selection uncertainty problem.

Model Selection Uncertainty 4: The uncertainty of positioning/ranking by the individual criteria

^{2} only.

In addition to the uncertainty of selection as “best” model, as previously noted, there is also the uncertainty of positioning/ranking by the individual criteria. One cannot predict the positioning/ranking of models by the individual criteria.

The best selection position of the only hardness model with the conforming response surface is number three as seen in

criteria put this model on position three in their rankings. However, the same response model is ranked twenty five by four other criteria. There is, therefore, no known formula to predict the ranking of the models with conforming response surfaces, as different selection criteria rank them differently. The only way that models with conforming response surfaces can be recognised is by inspecting response surfaces. This is now the fourth model selection uncertainty problem.

Twenty eight model selection criteria are used to select the best model and the frequency of selection per number of regressor variables (p) is determined. This is done for each of the response variables of adhesion and hardness.

Adhesion

A visual picture of

The mere fact that there is a selection for each p-value when all the twenty eight model selection criteria are used indicates model selection uncertainty.

REGRESSORS (p) | 1 | 2 | 3 | 4 | 5 |
---|---|---|---|---|---|

ALL 28 | 2 | 17 | 1 | 8 | 3 |

Corrected | 0 | 7 | 0 | 2 | 1 |

Regressors (p) | 1 + 2 | 3 | 4 + 5 |
---|---|---|---|

Total Frequency | 7 | 0 | 3 |

position and the one position, and 2) there are seven selections to the left of three regressors (median term) and only three to the right.

Hardness

Uncertainty in selection is obvious in both the adhesion and hardness cases where twenty eight model selection criteria are used and when only small sample size corrected criteria are used.

This could be posing a query to the achievement of parsimony in balancing bias (lack of fit) and penalty especially where small sample size corrected criteria is concerned as there could be over-correction which results in criteria selecting models with fewer regressors or underfitting in trying to correct small sample size inefficiency.

REGRESSORS (p) | 1 | 2 | 3 | 4 | 5 |
---|---|---|---|---|---|

ALL 28 | 1 | 5 | 2 | 4 | 16 |

Corrected | 1 | 3 | 0 | 3 | 3 |

Regressors (p) | 1 + 2 | 3 | 4 + 5 |
---|---|---|---|

Frequency | 4 | 0 | 6 |

MRSM is based on the use of response surfaces of a selected combination of response models to simultaneously determine the region with the desired results. There being no certain method of using a classical model selection criterion to predict models with conforming response surfaces implies the simplest way to deal with the problems of model selection uncertainty and small sample size bias is to avoid choosing candidate “best” models with model selection criteria. In two-input MRSM processes this is possible as response surfaces can be used to select candidate models. Therefore, in this section the four possible results from the permutations of the four adhesion response surfaces and one hardness response surface are analysed. The four results are obtained by simultaneously solving the tabled four pairs of models. The mean square errors (MSE) of the conforming models are also determined.

The permutations of adhesion-hardness response model pairs with conforming response surfaces is shown summarized in

The desired results are obtained by constructing data matrices for each response in a candidate pair using Microsoft Excel and overlaying them to determine the region where the two response models simultaneously achieve customer expected results of adhesion ≥ 12 N/mm and hardness ≥ 60 ˚Shore A. The detailed process of determination of desired results for each combination is shown in ANNEXURE D. Pair 2 is used here as an example to demonstrate the process.

Pair 2: [T, RT, T*RT, T^{2}] vs. [T, RT, T*RT, T^{2}]

The response surfaces and corresponding data matrices of Pair 2 are presented as an example. The data matrices are overlaid to obtain the desired region from which results of cure time per rubber thickness are obtained. The boxed figure 9 in the data matrix in

PAIR | Adhesion Response Surface Model | Hardness Response Surface Model |
---|---|---|

1 | T, RT, T*RT | |

2 | T, RT, T*RT, T^{2} | T, RT, T*RT, T^{2} |

3 | T, RT, T*RT, RT^{2} | |

4 | T, RT, T*RT, T^{2}, RT^{2} |

CT | 21 | 22 | 23 | 24 | 25 | 26 | 27 | 28 | 29 | 30 |
---|---|---|---|---|---|---|---|---|---|---|

RT | ||||||||||

20 | 9 | 9 | 9 | 10 | 10 | 11 | 11 | 11 | 11 | 12 |

19 | 9 | 9 | 10 | 10 | 11 | 11 | 11 | 11 | 12 | 12 |

18 | 9 | 10 | 10 | 10 | 11 | 11 | 11 | 12 | 12 | 12 |

17 | 10 | 10 | 10 | 11 | 11 | 11 | 12 | 12 | 12 | 12 |

16 | 10 | 10 | 11 | 11 | 11 | 12 | 12 | 12 | 12 | 13 |

15 | 10 | 11 | 11 | 11 | 12 | 12 | 12 | 12 | 13 | 13 |

14 | 11 | 11 | 11 | 12 | 12 | 12 | 12 | 13 | 13 | 13 |

13 | 11 | 11 | 12 | 12 | 12 | 12 | 13 | 13 | 13 | 13 |

12 | 11 | 12 | 12 | 12 | 13 | 13 | 13 | 13 | 13 | 13 |

11 | 12 | 12 | 12 | 13 | 13 | 13 | 13 | 13 | 14 | 14 |

10 | 12 | 12 | 13 | 13 | 13 | 13 | 13 | 14 | 14 | 14 |

9 | 12 | 13 | 13 | 13 | 13 | 14 | 14 | 14 | 14 | 14 |

8 | 13 | 13 | 13 | 13 | 14 | 14 | 14 | 14 | 14 | 14 |

7 | 13 | 13 | 14 | 14 | 14 | 14 | 14 | 14 | 14 | 14 |

• Adhesion [T, RT, T*RT, T^{2}] Data Matrix

The data matrix for the adhesion model [T, RT, T*RT, T^{2}] if overlaid with a hardness model data matrix from a conforming response surface gives the desired region with optimum cure time per rubber thickness.

• Hardness [T, RT, T*RT, T^{2}] Data Matrix

The hardness data matrix of ^{2}] is overlaid with the data matrix for the adhesion model [T, RT, T*RT, T^{2}] to give the desired region with cure time per rubber thickness meeting customer expectations.

Cure times per belt rubber thickness are selected to ensure customer expectations are met and right levels of productivity maintained. The region in red has both adhesion and hardness results in levels acceptable to the customer expectations. The boxed figures in the desired region indicate the cure time per belt rubber thickness combinations considered for work instructions.

Mean Sum of Squares (MSE’s) for the models with conforming response surfaces are determined using Equation (9) for comparing of accuracy of response models. The formula for MSE is given below for a sample size n.

MSE = ∑ i = 1 n ( Y i − Y ^ i ) 2 n (9)

where Y i is the measured response, Y ^ i is the predicted response.

T | 20 | 21 | 22 | 23 | 24 | 25 | 26 | 27 | 28 | 29 | 30 | 31 | 32 | 33 |
---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|

RT | ||||||||||||||

20 | 56 | 57 | 58 | 59 | 59 | 60 | 60 | 60 | 60 | 60 | 60 | 60 | 60 | 59 |

19 | 56 | 57 | 58 | 59 | 59 | 60 | 60 | 60 | 61 | 60 | 60 | 60 | 60 | 59 |

18 | 57 | 58 | 58 | 59 | 60 | 60 | 60 | 60 | 61 | 60 | 60 | 60 | 59 | 59 |

17 | 57 | 58 | 59 | 59 | 60 | 60 | 60 | 61 | 61 | 60 | 60 | 60 | 59 | 59 |

16 | 57 | 58 | 59 | 59 | 60 | 60 | 60 | 61 | 61 | 60 | 60 | 60 | 59 | 59 |

15 | 57 | 58 | 59 | 60 | 60 | 60 | 61 | 61 | 61 | 60 | 60 | 60 | 59 | 58 |

14 | 58 | 59 | 59 | 60 | 60 | 60 | 61 | 61 | 61 | 60 | 60 | 60 | 59 | 58 |

13 | 58 | 59 | 59 | 60 | 60 | 61 | 61 | 61 | 61 | 60 | 60 | 60 | 59 | 58 |

12 | 58 | 59 | 60 | 60 | 60 | 61 | 61 | 61 | 61 | 60 | 60 | 59 | 59 | 58 |

11 | 59 | 59 | 60 | 60 | 61 | 61 | 61 | 61 | 61 | 60 | 60 | 59 | 59 | 58 |

10 | 59 | 60 | 60 | 60 | 61 | 61 | 61 | 61 | 61 | 60 | 60 | 59 | 59 | 58 |

9 | 59 | 60 | 60 | 61 | 61 | 61 | 61 | 61 | 61 | 60 | 60 | 59 | 58 | 58 |

8 | 59 | 60 | 60 | 61 | 61 | 61 | 61 | 61 | 61 | 60 | 60 | 59 | 58 | 57 |

7 | 60 | 60 | 61 | 61 | 61 | 61 | 61 | 61 | 61 | 60 | 60 | 59 | 58 | 57 |

CT RT | 21 | 22 | 23 | 24 | 25 | 26 | 27 | 28 | 29 | 30 |
---|---|---|---|---|---|---|---|---|---|---|

20 | 60 | |||||||||

19 | 60 | 60 | ||||||||

18 | 61 | 60 | 60 | |||||||

17 | 61 | 61 | 60 | 60 | ||||||

16 | 60 | 61 | 61 | 60 | 60 | |||||

15 | 60 | 61 | 61 | 61 | 60 | 60 | ||||

14 | 60 | 60 | 61 | 61 | 61 | 60 | 60 | |||

13 | 60 | 60 | 61 | 61 | 61 | 61 | 60 | 60 | ||

12 | 60 | 60 | 60 | 61 | 61 | 61 | 61 | 60 | 60 | |

11 | 60 | 60 | 61 | 61 | 61 | 61 | 61 | 60 | 60 | |

10 | 60 | 60 | 60 | 61 | 61 | 61 | 61 | 61 | 60 | 60 |

9 | 60 | 60 | 61 | 61 | 61 | 61 | 61 | 61 | 60 | 60 |

8 | 60 | 60 | 61 | 61 | 61 | 61 | 61 | 61 | 60 | 60 |

7 | 60 | 61 | 61 | 61 | 61 | 61 | 61 | 61 | 60 | 60 |

This section shows the results obtained by optimizing each of the four pairs with the methodology outlined in Section 5. The computed MSE results of the individual models with conforming response surfaces are shown.

Using overlaying of data matrices, tables of rubber thickness-cure time combinations are determined for the four pairs of models. The tables are shown as Tables 13-16.

If all the result tables are averaged, a result equivalent to

Response model accuracy is checked by the size of the MSE.

The MSE values for adhesion models show that the full model, [T, RT, T*RT, T^{2}, RT^{2}], has the best fit-to-data accuracy compared to the other three models with conforming response surfaces.

Rubber Thickness (mm) | 7 | 8 | 9 | 10 | 11 | 12 | 13 | 14 | 15 | 16 | 17 | 18 | 19 | 20 |
---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|

Cure Time (min.) | 21 | 22 | 23 | 24 | 24 | 25 | 25 | 26 | 26 | 27 | 28 | 28 | 29 | 30 |

Rubber Thickness (mm) | 7 | 8 | 9 | 10 | 11 | 12 | 13 | 14 | 15 | 16 | 17 | 18 | 19 | 20 |
---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|

Cure Time (min.) | 21 | 22 | 23 | 24 | 24 | 25 | 25 | 26 | 26 | 27 | 27 | 28 | 29 | 30 |

Rubber Thickness (mm) | 7 | 8 | 9 | 10 | 11 | 12 | 13 | 14 | 15 | 16 | 17 | 18 | 19 | 20 |
---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|

Cure Time (min.) | 21 | 22 | 23 | 24 | 24 | 25 | 25 | 26 | 26 | 27 | 27 | 28 | 29 | 30 |

Rubber Thickness (mm) | 7 | 8 | 9 | 10 | 11 | 12 | 13 | 14 | 15 | 16 | 17 | 18 | 19 | 20 |
---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|

Cure Time (min.) | 21 | 22 | 23 | 24 | 24 | 25 | 25 | 26 | 26 | 27 | 27 | 28 | 28 | 29 |

ADHESION | HARDNESS | ||
---|---|---|---|

MSE | MSE | ||

T, RT, T*RT | 2.3441 | ||

T, RT, T*RT, T^{2} | 2.1777 | T, RT, T*RT, T^{2} | 5.3806 |

T, RT, T*RT, RT^{2} | 1.2789 | ||

T, RT, T*RT, T^{2}, RT^{2} | 0.9762 |

The results from the first seven conveyor belts to be run from the process, shown in

Validation by T-Tests

The first validation test proves that there is no significant statistical difference between the forecasted results and the results obtained from belts run from the normal belt production process. The results from the first seven conveyor belts to be run from the process are compared from model forecasted results using a paired T-Test as shown in

Validation by MSFE

In this section, validation of results is done by checking the effectiveness to obtain better or the same rubber thickness-cure time combinations as obtained in the first MRSM experiment.

SFE = ∑ i = 1 n ( Y i − Y F i ) 2 n (10)

where Y i is i^{th} the measured response, Y F i is the i^{th} forecasted response.

The MSFE values for adhesion models shown in ^{2}, RT^{2}], has the best prediction accuracy compared to the

# | Rubber Thickness (mm) | Cure Time (min) | Actual Adhesion (N/mm) | Predicted Adhesion (N/mm) | Adhesion Error (N/mm) | Actual Hardness | Predicted Hardness | Hardness Error |
---|---|---|---|---|---|---|---|---|

1 | 7 | 21 | 11.1 | 13.0 | −1.9 | 60 | 60 | 0 |

2 | 8 | 22 | 13.4 | 13.0 | 0.4 | 60 | 60 | 0 |

3 | 8 | 22 | 13.0 | 13.0 | 0.0 | 61 | 60 | +1 |

4 | 9 | 23 | 13.3 | 13.0 | 0.3 | 60 | 61 | −1 |

5 | 8 | 22 | 12.1 | 13.0 | −0.9 | 60 | 60 | 0 |

6 | 8 | 22 | 12.8 | 13.0 | −0.2 | 61 | 60 | +1 |

7 | 7 | 21 | 12.5 | 13.0 | 0.5 | 61 | 60 | +1 |

ADHESION | HARDNESS | ||
---|---|---|---|

MSFE | MSFE | ||

T, RT, T*RT | 2.1760 | ||

T, RT, T*RT, T^{2} | 2.4926 | T, RT, T*RT, T^{2} | 0.3047 |

T, RT, T*RT, RT^{2} | 2.0456 | ||

T, RT, T*RT, T^{2}, RT^{2} | 1.8455 |

other three models with conforming response surfaces. The best combination of adhesion versus hardness models for the best accuracy results is therefore the pair: Pair 4 [T, RT, T*RT, T^{2}, RT^{2}] vs. [T, RT, T*RT, T^{2}].

Averaging all the results tables gives a table similar to

This section discusses the small sample size MRSM datasets problems of using classical model selection criteria to select “best” models for use in determining desired solutions and how they are solved.

The Small Sample Size Bias Problem

The small sample size model selection criteria bias problem is solved by avoiding the use of model selection criteria for selecting “best” models in this study. In fact where response surfaces can be used to select candidate models, classical model selection criteria become irrelevant with all their problems.

The Model Selection Uncertainty Problem

The model selection uncertainty problem is related to the use of classical model selection criteria for selecting best models. Even the use of small sample size corrected criteria is shown to have the problem of model selection uncertainty. It is difficult to predict a model with a conforming response surface with classical best model selection criteria whether small sample size corrected or not. MRSM depends a lot on the use of conforming response surfaces in simultaneously determining the desired results. Research on design of experiments for MRSM should focus more on determining conforming response surfaces than just models with good fit to available data or prediction. The conformance of the response surface within the region of interest should be the focus of MRSM experimental designing research.

Using Response Surfaces to Select Response Models

According to Moral-Benito [

The use of response surfaces to select conforming models is very possible with two-factor multiple regression models although it is not that easy with processes of more than two factors. However, whilst still dealing with two factor processes, selecting models based on conformity of response surfaces avoids use of classical model selection criteria in choosing the best models and hence the problems of model selection criteria uncertainty and small sample size datasets bias. In MRSM that is very important. Using response models with conforming response surfaces within the region of interest is the fundamental focus of MRSM. The selection of candidate models with conforming response surfaces within the region of interest also opens the door for multiple model solution methodology which introduces rigour, transparency and therefore credibility into the MRSM results. The focus of MRSM, therefore, shifts from obtaining the “best” model to obtaining the best results within the region of interest.

Selecting candidate response models based on conformity of response surfaces avoids the uncertainty and small sample size bias problems that are related to using classical model selection criteria in selecting best models. So in two-input process problems, where response surfaces can easily be constructed and analysed, it is better for practitioners to use response surfaces to select candidate models for determining the permutations of response model sets for onward simultaneous optimisation.

Multiple model MRSM approach ensures credibility as rigour is maintained up to the final result. The problem of model selection uncertainty is kept clear of affecting the final result in a very transparent way.

However, for best results, the proposed multiple model MRSM approach based on using candidate models with conforming response surfaces requires prior knowledge on the expected ideal response surface.

Future Research

1) More datasets from two-factor processes need to be studied to ensure generalizability of findings to all other two-factor processes.

2) A usability study of the multivariate approach (Fujikoshi and Satoh [

3) Model averaging (Yuan and Yang [

4) Applicability of envelope models and methods to MRSM.

5) Research on generalising beyond the two-factor process to three and higher factor processes is necessary.

The authors declare no conflicts of interest regarding the publication of this paper.

Pavolo, D. and Chikobvu, D. (2019) A Practical Solution to the Small Sample Size Bias and Uncertainty Problems of Model Selection Criteria in Two-Input Process Multiple Response Surface Methodology Datasets. Open Journal of Statistics, 9, 109-142. https://doi.org/10.4236/ojs.2019.91010

In this section, the MRSM dataset is modeled to produce adhesion and hardness response models studied in this univariate investigation. Response modeling is done using Minitab 17. The model selection uncertainty of this MRSM small sample dataset is analysed. The general multivariate multiple regression model is

y u i = f i ( X u , β ) + ε u i , u = 1 , 2 , ⋯ , n _{}and i = 1 , 2 , ⋯ , r (1)

where X u is the vector of settings of k design variables at the u^{th}_{ }experimental run, β is a vector of unknown parameters, f i is a function of known form for the i^{th} responses, and ε u i is a random error associated with the i^{th} response for the experimental run u. It is assumed that the ε u i ‘s are normally distributed such that E ( ε u i ) = 0 , E ( ε u i × ε v j ) = 0 , for all i ≠ j , u ≠ v , V a r ( ε u i ) = σ u i ._{ }

There are two important mathematical models that are commonly used in multi-response surface methodology which are special cases of model (1).

The first-degree model (d = 1),

y = β 0 + ∑ j = 1 k β i x i + ε (2)

And, the second-degree model (d = 2)

y = β 0 + ∑ i = 0 k β i x i + ∑ i = 0 n ∑ j = 0 n β i j x i x j + ∑ i = 0 n β i i x i 2 + ε (3)

In this study, y is either adhesion or hardness, and x i is cure time (T) or total rubber thickness (RT). The parameters (β’s), are estimated using statistical software.

All regression methodology is employed to produce thirty one response models for each of the two responses, that is, adhesion and hardness from the MRSM dataset.

The general second degree model of Equation (3) is expanded into the following belt curing model:

Adhesion / Hardness = β 0 + β 1 ∗ T + β 2 ∗ RT + β 12 ∗ T ∗ RT + β 11 ∗ T 2 + β 22 ∗ RT 2 + error (4)

where T is cure time in minutes, RT is rubber thickness in millimeters, β 0 is the intercept and β 1 , β 1 , β 12 , β 11 , and β 22 are estimates of parameters.

1) Adhesion Response Models

Adhesion = 3.26 + 0.3244 ∗ T (5)

and

Adhesion = 15.41 − 0.3127 ∗ RT (6)

The rest of the twenty nine models are expanded in a similar manner.

Using the classical model selection process anyone of these thirty one models shown in

MODEL | β_{0} | β_{1} | β_{2} | β_{12} | β_{11} | β_{22} |
---|---|---|---|---|---|---|

T | 3.26000 | 0.32440 | ||||

RT | 15.41000 | −0.31270 | ||||

T*RT | 12.26000 | −0.00389 | ||||

T^{2} | 6.58000 | 0.00668 | ||||

RT^{2} | 13.64000 | −0.01114 | ||||

T, RT | 7.95000 | 0.32440 | −0.31270 | |||

T, T*RT | 3.26000 | 0.51000 | −0.01235 | |||

T, T^{2} | −2.30000 | 0.83500 | −0.01110 | |||

T, RT^{2} | 6.18000 | 0.32440 | −0.01114 | |||

RT, T*RT | 15.41000 | −0.79100 | 0.02078 | |||

RT, T^{2} | 11.67000 | −0.31270 | 0.00668 | |||

RT, RT^{2} | 11.08000 | 0.38000 | −0.02309 | |||

T*RT, T^{2} | 8.96000 | −0.01189 | 0.01048 | |||

T*RT, RT^{2} | 10.49700 | 0.02025 | −0.02579 | |||

T^{2}, RT^{2} | 9.91000 | 0.00664 | −0.01109 | |||

T, RT, T*RT | 12.94000 | 0.10700 | −0.64600 | 0.01450 | ||

T, RT, T^{2} | 2.41000 | 0.83500 | −0.31270 | −0.01110 | ||

T, RT, RT^{2} | 3.61000 | 0.32440 | 0.38000 | −0.02309 | ||

T, T*RT, T^{2} | −2.28000 | 1.02000 | −0.01235 | −0.01110 | ||

T, T*RT, RT^{2} | 9.14000 | 0.09100 | 0.01559 | −0.02242 | ||

T, T^{2}, RT^{2} | −0.25000 | 0.91900 | −0.01290 | −0.01122 | ||

RT, T*RT, T^{2} | 15.24000 | −0.77100 | 0.01990 | 0.00031 | ||

RT, T*RT, RT^{2} | 11.08000 | −0.09800 | 0.02078 | −0.02309 | ||

RT, T^{2}, RT^{2} | 7.52000 | 0.35800 | 0.00661 | −0.02240 | ||

T*RT, T^{2}, RT^{2} | 10.39000 | 0.01890 | 0.00054 | −0.02485 | ||

T, RT, T*RT, T^{2} | 7.40000 | 0.61800 | −0.64600 | 0.01450 | −0.01110 | |

T, RT, T*RT, RT^{2} | 8.61000 | 0.10700 | 0.04700 | 0.01450 | −0.02309 | |

T, T*RT, T^{2}, RT^{2} | −4.25000 | 1.02100 | 0.43000 | −0.01510 | −0.02476 | |

T, RT, T^{2}, RT^{2} | 1.95000 | 0.75900 | 0.01676 | −0.01491 | −0.02336 | |

RT, T*RT, T^{2}, RT^{2} | 11.21000 | −0.11300 | 0.02150 | −0.00026 | −0.02317 | |

T, RT, T*RT, T^{2}, RT^{2} | 0.74000 | 0.80400 | 0.09700 | 0.01450 | −0.01510 | −0.02476 |

2) Hardness Response Models

MODEL | β_{0} | β_{1} | β_{2} | β_{12} | β_{11} | β_{22} |
---|---|---|---|---|---|---|

T | 45.75000 | 0.51300 | ||||

RT | 60.25000 | −0.18000 | ||||

T*RT | 56.18000 | 0.00395 | ||||

T^{2} | 52.13000 | 0.00966 | ||||

RT^{2} | 58.32000 | −0.00297 | ||||

T, RT | 48.46000 | 0.51300 | −0.18000 | |||

T, T*RT | 45.75000 | 0.60400 | 0.00608 | |||

T, T^{2} | 15.30000 | 3.32000 | −0.06100 | |||

T, RT^{2} | 46.53000 | 0.51300 | −0.00297 | |||

RT, T*RT | 60.25000 | −0.96100 | 0.03390 | |||

RT, T^{2} | 54.84000 | −0.18000 | −0.00966 | |||

RT, RT^{2} | 69.31000 | −1.62900 | 0.04830 | |||

T*RT, T^{2} | 52.89000 | 0.00454 | −0.01111 | |||

T*RT, RT^{2} | 55.08000 | 0.02090 | −0.01810 | |||

T^{2}, RT^{2} | 52.90000 | 0.00965 | −0.00291 | |||

T, RT, T*RT | 57.50000 | 0.03200 | −0.71800 | 0.03210 | ||

T, RT, T^{2} | 18.00000 | 3.32000 | −0.18000 | −0.06100 | ||

T, RT, RT^{2} | 57.51000 | 0.51300 | −1.62900 | 0.04830 | ||

T, T*RT, T^{2} | 15.30000 | 3.41000 | −0.00608 | −0.06100 | ||

T, T*RT, RT^{2} | 41.93000 | 0.87600 | −0.02420 | 0.01460 | ||

T, T^{2}, RT^{2} | 15.90000 | 3.34000 | −0.06160 | −0.00335 | ||

RT, T*RT, T^{2} | 65.06000 | −1.49600 | 0.05720 | −0.00860 | ||

RT, T*RT, RT^{2} | 69.31000 | −2.40900 | 0.03392 | 0.04830 | ||

RT, T^{2}, RT^{2} | 64.01000 | −1.66100 | 0.00983 | 0.04940 | ||

T*RT, T^{2}, RT^{2} | 52.66000 | −0.00940 | 0.01270 | 0.00390 | ||

T, RT, T*RT, T^{2} | 29.10000 | 2.84000 | −0.91800 | 0.03210 | −0.06100 | |

T, RT, T*RT, RT^{2} | 68.60000 | 0.03200 | −2.36600 | 0.03210 | 0.04830 | |

T, T*RT, T^{2}, RT^{2} | 13.40000 | 3.53000 | −0.01960 | −0.05920 | 0.01080 | |

T, RT, T^{2}, RT^{2} | 29.42000 | 3.00200 | −1.45000 | −0.05410 | 0.04230 | |

RT, T*RT, T^{2}, RT^{2} | 73.31000 | −2.84700 | 0.05400 | −0.00744 | 0.00475 | |

T, RT, T*RT, T^{2}, RT^{2} | 40.50000 | 2.52100 | −2.18700 | 0.03210 | −0.05410 | 0.04230 |

Adhesion = 45.75 + 0.513 ∗ T (7)

and

Adhesion = 60.25 − 0.18 ∗ RT (8)

The rest of the twenty nine models can be similarly expanded.

The four adhesion and one hardness conforming response surfaces constructed using excel are shown.

1) Adhesion [T, RT, T*RT] (

2) Adhesion [T, RT, T*RT, T^{2}] Response Surface (

3) Adhesion [T, RT, T*RT, RT^{2}] Response Surface (

4) Adhesion [T, RT, T*RT, T^{2}, RT^{2}] Response Surface (

The four adhesion response surfaces show the expected continuously rising modulus behaviour of rubber skim compounds whose specification is synthetic rubber based.

5) Hardness [T, RT, T*RT, T^{2}]

The response surface for [T, RT, T*RT, T^{2}] is the only conforming response surface for hardness among the thirty one hardness models and is shown in

The hardness behaviour of the cover compound is as expected from a natural rubber based compound designed for hardness.

1) Response surfaces, data matrices and determination of results

a) Data Matrices

i) Adhesion [T, RT, T*RT] (

CT RT | 21 | 22 | 23 | 24 | 25 | 26 | 27 | 28 | 29 | 30 |
---|---|---|---|---|---|---|---|---|---|---|

20 | 8 | 9 | 9 | 10 | 10 | 10 | 11 | 11 | 12 | 12 |

19 | 9 | 9 | 9 | 10 | 10 | 11 | 11 | 11 | 12 | 12 |

18 | 9 | 9 | 10 | 10 | 11 | 11 | 11 | 12 | 12 | 12 |

17 | 9 | 10 | 10 | 10 | 11 | 11 | 11 | 12 | 12 | 13 |

16 | 10 | 10 | 10 | 11 | 11 | 11 | 12 | 12 | 12 | 13 |

15 | 10 | 10 | 11 | 11 | 11 | 12 | 12 | 12 | 13 | 13 |

14 | 10 | 11 | 11 | 11 | 12 | 12 | 12 | 13 | 13 | 13 |

13 | 11 | 11 | 11 | 12 | 12 | 12 | 13 | 13 | 13 | 13 |

12 | 11 | 11 | 12 | 12 | 12 | 13 | 13 | 13 | 13 | 14 |

11 | 11 | 12 | 12 | 12 | 13 | 13 | 13 | 13 | 14 | 14 |

10 | 12 | 12 | 12 | 13 | 13 | 13 | 13 | 14 | 14 | 14 |

9 | 12 | 12 | 13 | 13 | 13 | 13 | 14 | 14 | 14 | 14 |

8 | 12 | 13 | 13 | 13 | 13 | 14 | 14 | 14 | 14 | 14 |

7 | 13 | 13 | 13 | 13 | 14 | 14 | 14 | 14 | 14 | 15 |

The data matrix for the adhesion model [T, RT, T*RT] if overlapped with a hardness model data matrix from a conforming response surface will give a wanted region with optimum cure time per rubber thickness results.

ii) Hardness [T, RT, T*RT, T^{2}] (

CT RT | 21 | 22 | 23 | 24 | 25 | 26 | 27 | 28 | 29 | 30 |
---|---|---|---|---|---|---|---|---|---|---|

20 | 57 | 58 | 59 | 59 | 60 | 60 | 60 | 60 | 60 | 60 |

19 | 57 | 58 | 59 | 59 | 60 | 60 | 60 | 61 | 60 | 60 |

18 | 58 | 58 | 59 | 60 | 60 | 60 | 60 | 61 | 60 | 60 |

17 | 58 | 59 | 59 | 60 | 60 | 60 | 61 | 61 | 60 | 60 |

16 | 58 | 59 | 59 | 60 | 60 | 60 | 61 | 61 | 60 | 60 |

15 | 58 | 59 | 60 | 60 | 60 | 61 | 61 | 61 | 60 | 60 |

14 | 59 | 59 | 60 | 60 | 60 | 61 | 61 | 61 | 60 | 60 |

13 | 59 | 59 | 60 | 60 | 61 | 61 | 61 | 61 | 60 | 60 |

12 | 59 | 60 | 60 | 60 | 61 | 61 | 61 | 61 | 60 | 60 |

11 | 59 | 60 | 60 | 61 | 61 | 61 | 61 | 61 | 60 | 60 |

10 | 60 | 60 | 60 | 61 | 61 | 61 | 61 | 61 | 60 | 60 |

9 | 60 | 60 | 61 | 61 | 61 | 61 | 61 | 61 | 60 | 60 |

8 | 60 | 60 | 61 | 61 | 61 | 61 | 61 | 61 | 60 | 60 |

7 | 60 | 61 | 61 | 61 | 61 | 61 | 61 | 61 | 60 | 60 |

The hardness data matrix for model [T, RT, T*RT, T^{2}] will overlap with any conforming response model to give the optimum region with cure time per rubber thickness results.

CT RT | 21 | 22 | 23 | 24 | 25 | 26 | 27 | 28 | 29 | 30 |
---|---|---|---|---|---|---|---|---|---|---|

20 | 60 | 60 | ||||||||

19 | 60 | 60 | ||||||||

18 | 61 | 60 | 60 | |||||||

17 | 61 | 60 | 60 | |||||||

16 | 61 | 61 | 60 | 60 | ||||||

15 | 61 | 61 | 61 | 60 | 60 | |||||

14 | 60 | 61 | 61 | 61 | 60 | 60 | ||||

13 | 60 | 61 | 61 | 61 | 61 | 60 | 60 | |||

12 | 60 | 60 | 61 | 61 | 61 | 61 | 60 | 60 | ||

11 | 60 | 60 | 61 | 61 | 61 | 61 | 61 | 60 | 60 | |

10 | 60 | 60 | 60 | 61 | 61 | 61 | 61 | 61 | 60 | 60 |

9 | 60 | 60 | 61 | 61 | 61 | 61 | 61 | 61 | 60 | 60 |

8 | 60 | 60 | 61 | 61 | 61 | 61 | 61 | 61 | 60 | 60 |

7 | 60 | 61 | 61 | 61 | 61 | 61 | 61 | 61 | 60 | 60 |

2) Pair 2: [T, RT, T*RT, T^{2}] vs. [T, RT, T*RT, T^{2}]

a) Response Surfaces

i) Adhesion [T, RT, T*RT, T^{2}] (

ii) Hardness [T, RT, T*RT, T^{2}] (

b) Data Matrices

i) Adhesion [T, RT, T*RT, T^{2}] (

CT RT | 21 | 22 | 23 | 24 | 25 | 26 | 27 | 28 | 29 | 30 |
---|---|---|---|---|---|---|---|---|---|---|

20 | 9 | 9 | 9 | 10 | 10 | 11 | 11 | 11 | 11 | 12 |

19 | 9 | 9 | 10 | 10 | 11 | 11 | 11 | 11 | 12 | 12 |

18 | 9 | 10 | 10 | 10 | 11 | 11 | 11 | 12 | 12 | 12 |

17 | 10 | 10 | 10 | 11 | 11 | 11 | 12 | 12 | 12 | 12 |

16 | 10 | 10 | 11 | 11 | 11 | 12 | 12 | 12 | 12 | 13 |

15 | 10 | 11 | 11 | 11 | 12 | 12 | 12 | 12 | 13 | 13 |

14 | 11 | 11 | 11 | 12 | 12 | 12 | 12 | 13 | 13 | 13 |

13 | 11 | 11 | 12 | 12 | 12 | 12 | 13 | 13 | 13 | 13 |

12 | 11 | 12 | 12 | 12 | 13 | 13 | 13 | 13 | 13 | 13 |

11 | 12 | 12 | 12 | 13 | 13 | 13 | 13 | 13 | 14 | 14 |

10 | 12 | 12 | 13 | 13 | 13 | 13 | 13 | 14 | 14 | 14 |

9 | 12 | 13 | 13 | 13 | 13 | 14 | 14 | 14 | 14 | 14 |

8 | 13 | 13 | 13 | 13 | 14 | 14 | 14 | 14 | 14 | 14 |

7 | 13 | 13 | 14 | 14 | 14 | 14 | 14 | 14 | 14 | 14 |

ii) Hardness [T, RT, T*RT, T^{2}] (

CT RT | 21 | 22 | 23 | 24 | 25 | 26 | 27 | 28 | 29 | 30 |
---|---|---|---|---|---|---|---|---|---|---|

20 | 57 | 58 | 59 | 59 | 60 | 60 | 60 | 60 | 60 | 60 |

19 | 57 | 58 | 59 | 59 | 60 | 60 | 60 | 61 | 60 | 60 |

18 | 58 | 58 | 59 | 60 | 60 | 60 | 60 | 61 | 60 | 60 |

17 | 58 | 59 | 59 | 60 | 60 | 60 | 61 | 61 | 60 | 60 |

16 | 58 | 59 | 59 | 60 | 60 | 60 | 61 | 61 | 60 | 60 |

15 | 58 | 59 | 60 | 60 | 60 | 61 | 61 | 61 | 60 | 60 |
---|---|---|---|---|---|---|---|---|---|---|

14 | 59 | 59 | 60 | 60 | 60 | 61 | 61 | 61 | 60 | 60 |

13 | 59 | 59 | 60 | 60 | 61 | 61 | 61 | 61 | 60 | 60 |

12 | 59 | 60 | 60 | 60 | 61 | 61 | 61 | 61 | 60 | 60 |

11 | 59 | 60 | 60 | 61 | 61 | 61 | 61 | 61 | 60 | 60 |

10 | 60 | 60 | 60 | 61 | 61 | 61 | 61 | 61 | 60 | 60 |

9 | 60 | 60 | 61 | 61 | 61 | 61 | 61 | 61 | 60 | 60 |

8 | 60 | 60 | 61 | 61 | 61 | 61 | 61 | 61 | 60 | 60 |

7 | 60 | 61 | 61 | 61 | 61 | 61 | 61 | 61 | 60 | 60 |

CT RT | 21 | 22 | 23 | 24 | 25 | 26 | 27 | 28 | 29 | 30 |
---|---|---|---|---|---|---|---|---|---|---|

20 | 60 | |||||||||

19 | 60 | 60 | ||||||||

18 | 61 | 60 | 60 | |||||||

17 | 61 | 61 | 60 | 60 | ||||||

16 | 60 | 61 | 61 | 60 | 60 | |||||

15 | 60 | 61 | 61 | 61 | 60 | 60 | ||||

14 | 60 | 60 | 61 | 61 | 61 | 60 | 60 | |||

13 | 60 | 60 | 61 | 61 | 61 | 61 | 60 | 60 | ||

12 | 60 | 60 | 60 | 61 | 61 | 61 | 61 | 60 | 60 | |

11 | 60 | 60 | 61 | 61 | 61 | 61 | 61 | 60 | 60 | |

10 | 60 | 60 | 60 | 61 | 61 | 61 | 61 | 61 | 60 | 60 |

9 | 60 | 60 | 61 | 61 | 61 | 61 | 61 | 61 | 60 | 60 |

8 | 60 | 60 | 61 | 61 | 61 | 61 | 61 | 61 | 60 | 60 |

7 | 60 | 61 | 61 | 61 | 61 | 61 | 61 | 61 | 60 | 60 |

3) Pair 3: [T, RT, T*RT, RT^{2}] vs. [T, RT, T*RT, T^{2}]

a) Response Surfaces

i) Adhesion [T, RT, T*RT, RT^{2}] (

ii) Hardness [T, RT, T*RT, T^{2}]

The hardness response surface is shown already in

b) Data Matrices

i) Adhesion [T, RT, T*RT, RT^{2}] (

CT RT | 21 | 22 | 23 | 24 | 25 | 26 | 27 | 28 | 29 | 30 |
---|---|---|---|---|---|---|---|---|---|---|

20 | 9 | 9 | 9 | 10 | 10 | 11 | 11 | 11 | 11 | 12 |

19 | 9 | 9 | 10 | 10 | 11 | 11 | 11 | 11 | 12 | 12 |

18 | 9 | 10 | 10 | 10 | 11 | 11 | 11 | 12 | 12 | 12 |

17 | 10 | 10 | 10 | 11 | 11 | 11 | 12 | 12 | 12 | 12 |

16 | 10 | 10 | 11 | 11 | 11 | 12 | 12 | 12 | 12 | 13 |

15 | 10 | 11 | 11 | 11 | 12 | 12 | 12 | 12 | 13 | 13 |

14 | 11 | 11 | 11 | 12 | 12 | 12 | 12 | 13 | 13 | 13 |

13 | 11 | 11 | 12 | 12 | 12 | 12 | 13 | 13 | 13 | 13 |

12 | 11 | 12 | 12 | 12 | 13 | 13 | 13 | 13 | 13 | 13 |

11 | 12 | 12 | 12 | 13 | 13 | 13 | 13 | 13 | 14 | 14 |

10 | 12 | 12 | 13 | 13 | 13 | 13 | 13 | 14 | 14 | 14 |

9 | 12 | 13 | 13 | 13 | 13 | 14 | 14 | 14 | 14 | 14 |

8 | 13 | 13 | 13 | 13 | 14 | 14 | 14 | 14 | 14 | 14 |

7 | 13 | 13 | 14 | 14 | 14 | 14 | 14 | 14 | 14 | 14 |

ii) Hardness [T, RT, T*RT, T^{2}] (

CT RT | 21 | 22 | 23 | 24 | 25 | 26 | 27 | 28 | 29 | 30 |
---|---|---|---|---|---|---|---|---|---|---|

20 | 57 | 58 | 59 | 59 | 60 | 60 | 60 | 60 | 60 | 60 |

19 | 57 | 58 | 59 | 59 | 60 | 60 | 60 | 61 | 60 | 60 |

18 | 58 | 58 | 59 | 60 | 60 | 60 | 60 | 61 | 60 | 60 |

17 | 58 | 59 | 59 | 60 | 60 | 60 | 61 | 61 | 60 | 60 |

16 | 58 | 59 | 59 | 60 | 60 | 60 | 61 | 61 | 60 | 60 |

15 | 58 | 59 | 60 | 60 | 60 | 61 | 61 | 61 | 60 | 60 |

14 | 59 | 59 | 60 | 60 | 60 | 61 | 61 | 61 | 60 | 60 |

13 | 59 | 59 | 60 | 60 | 61 | 61 | 61 | 61 | 60 | 60 |

12 | 59 | 60 | 60 | 60 | 61 | 61 | 61 | 61 | 60 | 60 |

11 | 59 | 60 | 60 | 61 | 61 | 61 | 61 | 61 | 60 | 60 |

10 | 60 | 60 | 60 | 61 | 61 | 61 | 61 | 61 | 60 | 60 |

9 | 60 | 60 | 61 | 61 | 61 | 61 | 61 | 61 | 60 | 60 |

8 | 60 | 60 | 61 | 61 | 61 | 61 | 61 | 61 | 60 | 60 |

7 | 60 | 61 | 61 | 61 | 61 | 61 | 61 | 61 | 60 | 60 |

CT RT | 21 | 22 | 23 | 24 | 25 | 26 | 27 | 28 | 29 | 30 |
---|---|---|---|---|---|---|---|---|---|---|

20 | 60 | |||||||||

19 | 60 | 60 | ||||||||

18 | 61 | 60 | 60 | |||||||

17 | 61 | 61 | 60 | 60 | ||||||

16 | 60 | 61 | 61 | 60 | 60 | |||||

15 | 60 | 61 | 61 | 61 | 60 | 60 | ||||

14 | 60 | 60 | 61 | 61 | 61 | 60 | 60 | |||

13 | 60 | 60 | 61 | 61 | 61 | 61 | 60 | 60 | ||

12 | 60 | 60 | 60 | 61 | 61 | 61 | 61 | 60 | 60 | |

11 | 60 | 60 | 61 | 61 | 61 | 61 | 61 | 60 | 60 | |

10 | 60 | 60 | 60 | 61 | 61 | 61 | 61 | 61 | 60 | 60 |

9 | 60 | 60 | 61 | 61 | 61 | 61 | 61 | 61 | 60 | 60 |

8 | 60 | 60 | 61 | 61 | 61 | 61 | 61 | 61 | 60 | 60 |

7 | 60 | 61 | 61 | 61 | 61 | 61 | 61 | 61 | 60 | 60 |

4) Pair 4: [T, RT, T*RT, T^{2}, RT^{2}] vs. [T, RT, T*RT, T^{2}]

a) Response Surfaces

i) Adhesion [T, RT, T*RT, T^{2}, RT^{2}] (

ii) Hardness [T, RT, T*RT, T^{2}]

The hardness response surface is shown in D2 above.

b) Data Matrices

i) Adhesion [T, RT, T*RT, T^{2}, RT^{2}] (Tables D10-D12)

CT RT | 21 | 22 | 23 | 24 | 25 | 26 | 27 | 28 | 29 | 30 |
---|---|---|---|---|---|---|---|---|---|---|

20 | 9 | 10 | 10 | 10 | 11 | 11 | 11 | 12 | 12 | 12 |

19 | 10 | 10 | 10 | 11 | 11 | 11 | 12 | 12 | 12 | 12 |

18 | 10 | 11 | 11 | 11 | 12 | 12 | 12 | 12 | 13 | 13 |

17 | 11 | 11 | 11 | 12 | 12 | 12 | 13 | 13 | 13 | 13 |

16 | 11 | 11 | 12 | 12 | 12 | 13 | 13 | 13 | 13 | 13 |

15 | 11 | 12 | 12 | 12 | 13 | 13 | 13 | 13 | 14 | 14 |

14 | 12 | 12 | 12 | 13 | 13 | 13 | 13 | 14 | 14 | 14 |

13 | 12 | 12 | 13 | 13 | 13 | 13 | 14 | 14 | 14 | 14 |

12 | 12 | 13 | 13 | 13 | 13 | 14 | 14 | 14 | 14 | 14 |

11 | 12 | 13 | 13 | 13 | 13 | 14 | 14 | 14 | 14 | 14 |

10 | 12 | 13 | 13 | 13 | 13 | 14 | 14 | 14 | 14 | 14 |

9 | 13 | 13 | 13 | 13 | 14 | 14 | 14 | 14 | 14 | 14 |

8 | 13 | 13 | 13 | 13 | 13 | 14 | 14 | 14 | 14 | 14 |

7 | 13 | 13 | 13 | 13 | 13 | 14 | 14 | 14 | 14 | 14 |

CT RT | 21 | 22 | 23 | 24 | 25 | 26 | 27 | 28 | 29 | 30 |
---|---|---|---|---|---|---|---|---|---|---|

20 | 57 | 58 | 59 | 59 | 60 | 60 | 60 | 60 | 60 | 60 |

19 | 57 | 58 | 59 | 59 | 60 | 60 | 60 | 61 | 60 | 60 |

18 | 58 | 58 | 59 | 60 | 60 | 60 | 60 | 61 | 60 | 60 |

17 | 58 | 59 | 59 | 60 | 60 | 60 | 61 | 61 | 60 | 60 |

16 | 58 | 59 | 59 | 60 | 60 | 60 | 61 | 61 | 60 | 60 |

15 | 58 | 59 | 60 | 60 | 60 | 61 | 61 | 61 | 60 | 60 |

14 | 59 | 59 | 60 | 60 | 60 | 61 | 61 | 61 | 60 | 60 |

13 | 59 | 59 | 60 | 60 | 61 | 61 | 61 | 61 | 60 | 60 |

12 | 59 | 60 | 60 | 60 | 61 | 61 | 61 | 61 | 60 | 60 |

11 | 59 | 60 | 60 | 61 | 61 | 61 | 61 | 61 | 60 | 60 |

10 | 60 | 60 | 60 | 61 | 61 | 61 | 61 | 61 | 60 | 60 |

9 | 60 | 60 | 61 | 61 | 61 | 61 | 61 | 61 | 60 | 60 |

8 | 60 | 60 | 61 | 61 | 61 | 61 | 61 | 61 | 60 | 60 |

7 | 60 | 61 | 61 | 61 | 61 | 61 | 61 | 61 | 60 | 60 |

CT RT | 21 | 22 | 23 | 24 | 25 | 26 | 27 | 28 | 29 | 30 |
---|---|---|---|---|---|---|---|---|---|---|

20 | 60 | 60 | 60 | |||||||

19 | 60 | 61 | 60 | 60 | ||||||

18 | 60 | 60 | 60 | 61 | 60 | 60 | ||||

17 | 60 | 60 | 60 | 61 | 61 | 60 | 60 | |||

16 | 60 | 60 | 60 | 61 | 61 | 60 | 60 | |||

15 | 60 | 60 | 60 | 61 | 61 | 61 | 60 | 60 | ||

14 | 60 | 60 | 60 | 61 | 61 | 61 | 60 | 60 | ||

13 | 60 | 60 | 61 | 61 | 61 | 61 | 60 | 60 | ||

12 | 60 | 60 | 60 | 61 | 61 | 61 | 61 | 60 | 60 | |

11 | 60 | 60 | 61 | 61 | 61 | 61 | 61 | 60 | 60 | |

10 | 60 | 60 | 60 | 61 | 61 | 61 | 61 | 61 | 60 | 60 |

9 | 60 | 60 | 61 | 61 | 61 | 61 | 61 | 61 | 60 | 60 |

8 | 60 | 60 | 61 | 61 | 61 | 61 | 61 | 61 | 60 | 60 |

7 | 60 | 61 | 61 | 61 | 61 | 61 | 61 | 61 | 60 | 60 |

2) Results

See

^{2}]; (b) Pair 2: [T, RT, T*RT, T^{2}] vs. [T, RT, T*RT, T^{2}]; (c) Pair 3: [T, RT, T*RT, RT^{2}] vs. [T, RT, T*RT, T^{2}]; (d) Pair 4: [T, RT, T*RT, T^{2}, RT^{2}] vs. [T, RT, T*RT, T^{2}].