Dishwasher improvement at ASKO 
Developing a simplified test method to determine the 

influence of spray arm speed and pressure 

Master of Science Thesis in Product Development 
 

 

PETER TSOUKNIDAS 

XIANG ZHANG 
 

 

 

Department of Product and Production Development 

Division of Product Development 

CHALMERS UNIVERSITY OF TECHNOLOGY 

Göteborg, Sweden, 2010  



  



MASTER OF SCIENCE THESIS IN PRODUCT DEVELOPMENT 

 

 

DISHWASHER IMPROVEMENT AT ASKO 

Developing a simplified test method to determine the 

influence of spray arm speed and pressure 

 

 

 

PETER TSOUKNIDAS 

XIANG ZHANG 

 

 

 

 

 

 

 

 

 

 

 

 

Department of Product and Production Development 

 

Division of Product Development 

CHALMERS UNIVERSITY OF TECHNOLOGY 

Gothenburg, Sweden, 2010 

  



 

 

 

 

Dishwasher improvement at ASKO 

Developing a simplified test method to determine the influence of spray arm speed 

and pressure 

PETER T. TSOUKNIDAS 

XIANG ZHANG 

 

© PETER T. TSOUKNIDAS, XIANG ZHANG, 2010. 

 

 

Examiner: Lars Almefelt, Chalmers University of Technology 

Supervisor: Marcel Michaelis, Chalmers University of Technology 

Industrial supervisor: Patrik Ivarsson, ASKO Appliances AB 

 

 

 

 

 

Department of Product and Production Development 

Chalmers University of Technology 

SE-412 96 Göteborg 

Sweden 

Telephone + 46 (0)31-772 1000 

 

 

 

 

 

 

 

 

 

 

 

 

Cover: 

[Standardized test method used at ASKO to measure performance, 

Page no. 12 for further reading in the report.] 

 

Chalmers Reproservice 

Göteborg, Sweden 2010  



FOREWORD 

This master’s thesis was conducted at the request of ASKO Appliances AB. The 

expected outcome was considered to be of such value to the company that a 

non-disclosure agreement was established to protect proprietary information. 

Therefore, some information has been masked in the public version of the report. This 

includes the exact experiment levels and values, which have been replaced by relative 

terms in order for the reader to be able to follow the working process, as well as some 

specific details regarding the construction of the test method. 

  



  



ABSTRACT 

This thesis report describes how a dishwasher was modified and used as a test rig 

together with a new test method to find the relation between cleaning performance 

and the rotation speed of the spray arms as well as the water pressure. All tests were 

conducted using an ASKO D5900 dishwasher. 

 

First off, the tests were planned in detail. To ensure that the tests would be reliable and 

not overly redundant proper Design of Experiments methods were used. A simplified 

test method was developed that would be easier and faster to use than the standard 

tests currently used by dishwasher developers. 

 

When the test rig was to be built several concepts for changing rotation speed and 

water pressure were generated through brainstorming sessions and the use of 

morphological matrices based on system design thinking. Kesselring matrices were 

used during the final selection. The selected concepts were further developed to suit 

an actual implementation in the dishwasher. 

 

The test rig was built and used to implement the cleaning performance test. Several 

pre-tests were conducted to evaluate the test method and data analysis method. The 

first test was implemented and showed a trend of relation between cleaning 

performance and rotation speed and pressure. Performance increased as the rotation 

speed decreased and the pressure increased, but due to large deviations the results 

were not found to be reliable. The deviation of the results was too large to be 

conclusive by statistical standards. This was the reason for a second test 

implementation where influencing variables like time and temperature were more 

accurately controlled. Stable results were acquired during the second test, which 

supported the results from the first test. The prototype was then tested using European 

standard dishwasher test methods to determine the external validity of the test method 

at ASKOs facilities in Vara. Sound tests were also conducted. Finally, some tests were 

done at a higher pressure level using a stronger pump in order to see if the trend was 

continuing at even higher pressure levels. 

 

The test results showed that performance increases with higher water pressure and 

lower rotation speed. The cost of higher pressure is an increasing sound level. The test 

method proved to be reliable. 

 

 

Keywords: dishwasher, improvement, pressure, rotation speed, DOE, test method. 

 

 

 

 

  



TABLE OF CONTENTS 

 

1 – INTRODUCTION ................................................................................................... 1 

1.1 Background........................................................................................................ 1 

1.2 Project purpose and goal ................................................................................ 1 

1.3 Scope and limitations ....................................................................................... 2 

2 – PRODUCT AND EXPERIMENT DEVELOPMENT ............................................. 3 

2.1 Design of Experiments ..................................................................................... 3 

2.1.1 Planning the experiment ............................................................................... 3 

2.1.2 Validity .......................................................................................................... 4 

2.1.3 Regarding the design selection ..................................................................... 5 

2.2 Product Development methods ...................................................................... 7 

2.2.1 System design ............................................................................................... 7 

2.2.2 Product development procedure ................................................................... 8 

2.2.3 Morphological matrix ................................................................................... 9 

2.2.4 Brainstorming ............................................................................................... 9 

2.2.5 Kesselring matrix ........................................................................................ 10 

3 – TEST RIG AND EXPERIMENT DEVELOPMENT ............................................ 11 

3.1 Designing the experiment.............................................................................. 11 

3.2 Development of Simplified test method ...................................................... 12 

3.2.1 Design parameters ....................................................................................... 13 

3.3 Development of test rig .................................................................................. 19 

3.3.1 Development procedure .............................................................................. 19 

3.3.2 Rotation speed actuator ............................................................................... 21 

3.3.3 Rotation speed sensor ................................................................................. 27 

3.3.4 Water pressure actuator ............................................................................... 30 

3.3.5 Water pressure sensor .................................................................................. 33 

3.4 Test environment documentation ................................................................. 36 

3.5 Experiment Execution .................................................................................... 38 

4 – INTERMEDIATE RESULTS ................................................................................ 39 

4.1 Performance .................................................................................................... 39 

4.2 Sound level tests ............................................................................................ 41 



4.3 Energy consumption ...................................................................................... 41 

5 – FURTHER IMPROVED EXPERIMENT ............................................................. 43 

5.1 Research for new experiment ....................................................................... 44 

5.1.1 Test of new plates ........................................................................................ 44 

5.1.2 Test of old and new marker ......................................................................... 44 

5.1.3 Stabilizing time and temperature ................................................................ 45 

5.1.4 New pressure measurements ....................................................................... 45 

5.1.5 New Experiment Execution ........................................................................ 45 

5.2 Cleaning performance results ....................................................................... 46 

5.3 Validation tests at ASKO ................................................................................ 48 

5.4 Sound tests at ASKO ..................................................................................... 49 

5.5 High pressure tests ........................................................................................ 51 

6 - DISCUSSION ......................................................................................................... 53 

6.1 The simplified test method ............................................................................ 53 

6.2 Design of Experiments ................................................................................... 54 

6.3 Regarding the test rig ..................................................................................... 54 

6.4 Results from the first experiment ................................................................. 54 

6.5 Results from the improved experiment ....................................................... 55 

7 - CONCLUSIONS .................................................................................................... 57 

REFERENCES ............................................................................................................ 58 

APPENDIX .................................................................................................................. 60 

 

   



TABLE OF FIGURES 

Figure 1.1: Factors influencing dishwasher performance 1 

Figure 2.1: Example of central composite designs 6 

Figure 2.2: Example of response surface plot 6 

Figure 2.3: Example of response surface plotted as contour lines 7 

Figure 2.4: Basic system design workflow according to Stevens et al. 8 

Figure 2.5: Product development procedure according to Ulrich & Eppinger 8 

Figure 2.6: Example of morphological matrix 9 

Figure 2.7: Example of Kesselring matrix 10 

Figure 3.1: Preliminary levels of experiment 12 

Figure 3.2: Painting patterns tested on plates 14 

Figure 3.3: Final painting pattern 15 

Figure 3.4: Painting of test plates 15 

Figure 3.5: Plate placement in lower and upper baskets 16 

Figure 3.6: Program flowchart 17 

Figure 3.7: Indicator plate before and after threshold filter was applied 19 

Figure 3.8: Description of analysis method 19 

Figure 3.9: Concept Frictioner 21 

Figure 3.10: Concept heavyArm 22 

Figure 3.11: Concept Turbine 22 

Figure 3.12: Concept ExMotor 22 

Figure 3.13: Concept InMotor 22 

Figure 3.14: Concept Turbine 23 

Figure 3.15: Concept Internal axis version 1 24 

Figure 3.16: Concept external axis 24 

Figure 3.17: Concept Internal axis version 2 24 

Figure 3.18: Upper bearing holder and motor holder 25 

Figure 3.19: Lower bearing holder 25 

Figure 3.20: Timing belt and pulleys mounted in machine 26 

Figure 3.21: Concept click-Counter 27 

Figure 3.22: Concept Bike computer 27 

Figure 3.23: Concept Listen 28 

Figure 3.24: Concept Light 28 

Figure 3.25: Concept Bike Computer 29 

Figure 3.26: Bike computer mounted 30 

Figure 3.27: Concept Restrictor 31 

Figure 3.28: Concept Holy 31 

Figure 3.29: Concept PowerMotor 31 

Figure 3.30: Concept Restrictor 32 

Figure 3.31: Different temporary pressure sensor constructions 33 

Figure 3.32: The fourth working water pressure sensor 33 

Figure 3.33: Results from preliminary water pressure tests 34 

Figure 3.34: New experiment levels 35 



Figure 3.35: Pressure measurements 35 

Figure 3.36: Spray arm holes used for pressure tests 36 

Figure 3.37: Sensor rig circuit 36 

Figure 3.38: Sensor rig 37 

Figure 3.39: Sound meter placement 38 

Figure 4.1: Performance scores of first experiment 39 

Figure 4.2: Response surface plot of results from the first experiment 40 

Figure 4.3: Response surface contour plot of results from the first experiment 40 

Figure 4.4: Results from sound test at Chalmers laboratory 41 

Figure 4.5: Power usage contra input water temperature 42 

Figure 5.1: New experiment levels 45 

Figure 5.2: Performance scores of improved experiment 46 

Figure 5.3: Response surface plot of final results 47 

Figure 5.4: Contour plot of final cleaning performance results 48 

Figure 5.5: Standardized tests at ASKO 48 

Figure 5.6: Sound test setup at ASKO 50 

Figure 5.7: Results of sound tests 51 

Figure 5.8: Results of higher pressure test 52 

Figure 6.1: Contour plot of performance level (red) and sound level (blue) 55 

Figure 6.2: Suggestion for lower sound level 56 

Figure 6.3: Suggestion for better performance 56 

 

INDEX OF TABLES 

Table 3.1: Substances tested for usability in simplified test 14 

Table 3.2: Camera specifications 18 

Table 3.3: Rotation actuator Kesselring matrix results 23 

Table 3.4: Rotation sensor Kesselring matrix results 28 

Table 3.5: Water pressure actuator Kesselring matrix results 31 

Table 4.1: Detailed results from first test 39 

Table 5.1: Evaluation of possible sources of error 43 

Table 5.2: Detailed performance results from improved experiment 47 

Table 5.3: Results from standardized tests at ASKO 49 

Table 5.4: Dishwasher sound level test – very low pressure level 50 

Table 5.5: Dishwasher sound level test – low pressure level 50 

Table 5.6: Dishwasher sound level test – middle pressure level 50 

Table 5.7: Dishwasher sound level test– background noise test 51 

Table 5.8: Results of higher pressure test 52 

 



1 

 

1 – INTRODUCTION 

Dishwashers are used for cleaning dishes and other eating utensils in restaurants and 

private residences. A dishwasher is a complex cleaning system that integrates many 

components both mechanical and electrical. Dishwasher manufacturers are 

continuously improving their products to satisfy consumer demands. Characteristics 

that consumers find attractive in a dishwasher are high cleaning performance, low 

sound level, short cleaning time and low energy consumption among others
1
. 

 

1.1 Background 

ASKO Appliances AB is a Swedish company located in Vara. They develop and 

manufacture household appliances such as refrigerators, washing machine and 

dishwashers of high quality and high performance. ASKO strives to be a leader in 

dishwasher technology and they are constantly working on improving their products
1
. 

A lot of effort is put into improving the washing performance. The main parameters 

influencing the dishwasher performance are: use of detergents/chemicals, water 

temperature, washing time and mechanical impact on the dishes
1
. The parameters and 

their relative importance can be seen in Figure 1.1. 

 

 

Figure 1.1: Factors influencing dishwasher performance 

 

Keeping an environmental friendly profile has a high priority at ASKO. Decreasing 

water and energy consumption is beneficial since it lowers the usage costs for the 

customer and makes the product more environmental friendly. 

 

1.2 Project purpose and goal 

The cleaning performance is the most important competitive factor. One way of 

influencing the performance is by making changes in the mechanical impact on the 

dishes (as can be seen in Figure 1.1). The two main factors influencing the mechanical 

impact on the dishes are the rotation speed of the spray arms and the water pressure in 

the spray arms. Structured tests to determine how these factors influence the 

performance have not been done at ASKO. The rule of thumb, generally adopted at the 

company, is that higher pressure and lower rotation speed increase the performance. 

                                                 
1
 Patrik Ivarsson, ASKO Appliances AB 



2 

 

ASKO would like to confirm their theories regarding these factors. 

 

This project was to focus on exploring the effect of mechanical impact on dishes, more 

specifically how the rotation speed and water pressure in the dishwasher spray arms 

influence the washing performance. The two parameters are naturally coupled since the 

spray arm is driven by the water pressure alone. The goal of the project was to explore 

the interrelation between these parameters through testing. A simplified test method to 

measure dishwasher performance was to be developed to be used at Chalmers while 

conducting the tests. Methods for altering rotation speed and pressure were to be 

developed and implemented. All variables that could have an influence on the washing 

performance were to be documented. Sound measurements were to be made in order to 

see how changes in pressure and speed would influence the overall sound level of the 

machine. The results were to be such that they could be used at ASKO to improve their 

products.  

 

1.3 Scope and limitations 

The experiment was limited to one specific dishwasher model, the ASKO D5900. The 

focus was on improving the performance through mechanical impact on the dishes; 

none of the other influencing factors seen in Figure 1.1 should be altered. The only 

parameters that were to be altered were the spray arm rotation speed and the water 

pressure. Sound level measurements were to be conducted within the limitations of the 

laboratory environment. 

  



3 

 

2 – PRODUCT AND EXPERIMENT DEVELOPMENT 

In the following, a description of development methods used to develop the test rig and 

designing the experiment are provided. 

 

2.1 Design of Experiments 

This section gives a brief overview of the field of Design of Experiments (DOE). The 

methods described can be used in experiments ranging from very simple to highly 

complex. 

 

2.1.1 Planning the experiment 

In order to be reliable an experiment needs to be both well planned and well 

documented. Good planning also saves time and resources. The basic theory when 

designing an experiment is to follow these steps (Weber & Skillings, 2000): 

 

Clarify the goal 

Why is the experiment conducted? What is the phenomenon of interest? This needs to 

be clearly defined before any further work can be done. 

 

Select dependent variables 

Dependent variables, or response variables, are the main variables of interest. These are 

affected by the independent variables. The effect of the independent variables on the 

dependent ones is what is studied in the experiment, so one could say that the dependent 

variables are the 'result variables'.  

 

Identify independent variables 

Independent variables are all variables that affect the system and the dependent 

variables. These need to be found and evaluated. One or more of these will be used to 

induce an effect on the dependent variables. They are called the controlled variables. 

The controlled variables are varied in such a manner as to enable studies of their effect 

on the dependent variables. 

 

Other variables must be evaluated as to decide if they are to be kept constant or, if their 

influence on the dependent variables is insignificant enough to be disregarded, to be left 

uncontrolled. In some cases it can be difficult to identify all independent variables. 

Some might not be found when planning the experiment and thus be left uncontrolled. 

This will affect the validity of the experiment negatively. It is important to continuously 

document all variables in order to be able to determine unexpected influences and 

identify sources of error during the analysis. 

 

Some claim that there should be only one controlled independent variable so as to 



4 

 

isolate the effects of this as much as possible (Shuttleworth, 2008). On the one hand, 

this makes it easier to determine the results, and simplifies the test procedure. On the 

other hand, there might be interactions between independent variables that cannot be 

found if they are studied separately in different experiments. A so-called factorial 

experiment (see section 2.1.3 Factorial experiment) might then be preferable where an 

arbitrary number of controlled independent variables are allowed (Wikipedia, Design of 

Experiments). Isolating independent variables, thus missing the interaction between 

them which occur in the actual process studied, can lead to lower external validity. The 

decision of which variables to control, keep constant and leave uncontrolled is of 

highest importance to the outcome of the experiment. 

 

Set levels 

Each independent variable should have predefined levels between which they are to be 

varied. These are experiment-specific and should be derived from the experiment goal. 

 

Test design 

Selecting how the independent variables should be varied is also an important issue. 

There are a lot of different methods that have been developed to optimize the testing to 

get as good results as possible with as few tests as possible. Some of the most common 

ones are presented in section 2.1.3. 

 

The experiment can then be conducted and the result analyzed. The result of the 

experiment is an indication or description of the causal link between the independent 

and dependent variables. 

 

2.1.2 Validity 

It is important to address the question about experiment validity. Is the test efficiently 

isolating the phenomenon studied? Is it a good representation of the real-life 

phenomenon? 

 

Validity is usually divided into internal and external validity. Internal validity is a 

measurement of how well the experiment isolates the variables of interest and makes it 

possible to determine cause and effect. External validity concerns how well the 

experiment generalizes, i.e. if the test results correspond to the real-life phenomenon 

(Shuttleworth, 2008). External validity should not be confused with ecological validity. 

Ecological validity is a measurement for how well the experiment resembles the 

real-life situation (Wikipedia, Ecological validity). Ecological validity is not 

considered to affect the external or internal experiment validity. 

 

When designing an experiment one wants a certain level of validity, but in many cases it 

is difficult to get both high internal validity and high external validity. One might 

increase the level of internal validity by keeping some independent variables constant 

which are not controlled in real-life, but in doing so you decrease the external validity. 

This is an effect the test designer should keep in mind when designing the experiment. 



5 

 

 

2.1.3 Regarding the design selection 

When conducting an experiment the choice of design is important. One wants to do as 

few tests as possible, but still get statistically reliable results. This section describes 

some of the most common test setups. 

 

Factorial experiment 

A factorial experiment (also referred to as a full factorial experiment or fully-crossed 

experiment (Wikipedia, Factorial experiment)) is an experiment of one or more 

variables which vary in an equal number of steps (called levels) from a minimum to a 

maximum. The tests are performed with every possible combination of these levels of 

variables (Milton & Arnold, 2006). An experiment with four variables which each can 

have two different values would be called a 2
4
 factorial experiment. To complete this 

experiment one would have to do 16 tests. The general form is L
V
, where L is the 

number of levels and V the number of variables that are to be altered. 

 

Fractional factorial design 

Using a factorial design of an experiment often results in a very large number of 

experiments. This might not be time- or cost-efficient. One method to handle this 

problem is to use a fractional factorial design. The idea is to only conduct a part of all 

experiments of a factorial experiment (Milton & Arnold, 2006). The experimenter 

could choose which parts of the complete matrix of test combinations to perform based 

on a priori knowledge, or reduce the number of variables by merging two or more 

variables into one (Wikipedia, Fractional factorial experiment). The selection of which 

variables to merge should be done with the importance of the different variables in 

mind. Variables of high importance should not be merged (Milton & Arnold, 2006). A 

factorial design of, for example, 2
7
 can by this method be reduced to, for instance, 2

5
, 

thus reducing the number of tests from 128 to 32. 

 

When using a fractional factorial design, there is a risk of the effects of variables 

becoming indistinguishable from one another if the variable selection is not done 

properly (Milton & Arnold, 2006).
 

 

Central composite design 

Central composite design can be used instead of a full factorial design. The difference is 

the placement of testing points, as can be seen in Figure 2.1, where three levels are 

used. Figure 2.1 a, b and c describe different placement of the testing points around the 

variable span of interest. This design method is beneficial since it reduces the number 

of variable combination testing points (Wikipedia, Central composite design). In the 

case of three levels it reduces the number of points from 9 to 8 (compared to a full 

factorial design). 

 



6 

 

 

Figure 2.1: Example of central composite designs 

 

Response surface 

A response surface is a graphical way of presenting the response influenced by two or 

more variables (Montgomery, 2005). In Figure 2.2 is an example of two variables and 

their response plotted as a response surface. 

 

 

 Figure 2.2: Example of response surface plot 

 

The plot can be used to find a maximum (or minimum) in a relatively quick way with 

few experiments. The more levels used in the experiment, the more accurate the 

response surface becomes. If too few levels are used, there is a risk of missing a local 

minimum or maximum and thereby getting inaccurate results. 

 

There is a mathematical method for finding the maximum (or minimum) which could 

be used to reduce the number of experiments. The method is called the method of 

steepest ascent (steepest descent) and is carried through by plotting the response for an 

experiment of two variables, and then finding the direction of steepest ascent. The 

limits for the experiment are then moved in that direction and the procedure is repeated. 

This will eventually lead to a point where no direction can be determined, which will be 

assumed to be the maximum (or minimum) (Montgomery, 2005).
 
This method is 

suitable for tests where one has very accurate response data. 
 

Another way of presenting a response surface is as a contour plot (displayed in Figure 



7 

 

2.3). It is used to present a function or a curve based on two variables as a number of 

lines at which the response is at an equal value level. These lines are called isolines 

(Courant, Robbins & Stewart, 1996). Generally, there is a set particular value between 

each contour line (Wikipedia, Contour line). That means if the lines are close together 

the magnitude of the change is large: the variation is steep. Whereas a sparse line shows 

that the change is slight. 

 

 

 Figure 2.3: Example of response surface plotted as contour lines 

 

2.2 Product Development methods 

In this section some methods for product development and decision support methods 

that were used while developing the test method and test rig are described. 

 

2.2.1 System design 

System design starts with requirements definition and division into different 

subsystems. The division into subsystems makes project workflow more practical, 

efficient, flexible and secure. For each subsystem every component should be defined 

that contains units, interfaces, time and cost. The design should follow top-down 

hierarchy as shown in Figure 2.4 Thereby the independent interfaces can be tested 

(Stevens et al., 1998). 

 

All the components are integrated after they have passed the testing in the subsystem 

integration process. This is an opposite design direction, as shown in Figure 2.4. All the 

subsystem should be assembled together as a new system. 

 



8 

 

 

Figure 2.4: Basic system design workflow according to Stevens et al. 

 

2.2.2 Product development procedure 

When designing a new system, product development methodology should be 

implemented. This section describes product development methodology as written by 

Ulrich and Eppinger, 2008. Product development methods are commonly used tools for 

concept development, system level design and components design. The process can be 

seen in Figure 2.5. 

 

 
Figure 2.5: Product development procedure according to Ulrich & Eppinger 

 

Problems can be formulated by the experimenter or collected from customers and users
 

(Ulrich & Eppinger, 2008) (University of Leeds, Problem Identification). 

 

The problems are then reformulated as target specifications which are to represent goals 

and aspirations for the project of solutions for the problems defined. The target 

specifications can be derived from the developers themselves, users or from 

regulations. 

 

The concept generation process begins with a set of needs and target specifications and 

results in a set of concepts from which a final selection will be made. A good concept 

generation leaves the team with confidence that the full space of alternatives has been 

explored. In this phase, the use of morphological matrices and brainstorming, which are 

based on external and internal searching, are helpful when generating the initial 

concepts.
 

 

The next step is concept selection. In this phase the Kesselring matrix could be used to 

evaluate the concepts. Some concepts might be combined together to generate a new 

and improved concept. Finally, a winning concept is selected. 

 

The target specifications will then be revised to become final specifications after a 



9 

 

product concept has been chosen. 

 

2.2.3 Morphological matrix 

In the product design process the morphological matrix is a very common design 

method to hinge ideas when one has an glimpse of “what you wish to do” and “how you 

might go about doing it” (Fargnoli, Rovida & Troisi, 2006). It can be used to combine 

different part-solutions into a complete solution. 

 

The form of a matrix is comprised by a single left-hand column and several right 

columns. The left-hand column is used in listing the parameter essential to design as 

shown in Figure 2.6. To the right of each element in the column is a row containing the 

possible ways of achieving that particular parameter. Illustrations can be used in the 

matrix to get direct impression. 

 

 

Figure 2.6: Example of a morphological matrix 

 

2.2.4 Brainstorming 

Brainstorming is a popular creativity tool (Baiduzhidao, Brainstorming) that helps 

designers generate creative solutions to a problem (Ulrich & Eppinger, 2008). It is 

particularly useful when the designers want to develop new ways of looking at things. 

In a brainstorming group, it helps designers to utilize the diverse experience of all group 

members to solve the problem (Mindtools, Brainstorming).  

 

Individual brainstorming can sometimes be more effective than brainstorming in a 

group since it often results in ideas of better quality (Mindtools, Brainstorming). 

 



10 

 

2.2.5 Kesselring matrix 

The Kesselring matrix is a design tool which connects each specification with different 

concepts. The Kesselring matrix is applied to evaluate different concepts in a form. To 

use the matrix for this purpose, designers have to carefully establish the specification 

and assign the weight value in the left-hand column. To the right of each element in the 

column is a row containing the weight value of specific concepts as, Figure 2.7 shows. 

If every element has received its relevant weight value, the designer can get the final 

scores by summing up all elements in the same column. 

 

 

Figure 2.7: Example of Kesselring matrix 

  



11 

 

3 – TEST RIG AND EXPERIMENT DEVELOPMENT 

The entire project was divided into subsystems and subtasks in order for it to be easier 

to develop and find optimal solutions for each part. After system functional tests, the 

new dishwasher system could be used in the dishwasher cleaning performance test. The 

main subsystems and subtasks are: 

 

 Designing the experiment 

 Creating a simplified test method 

 Test rig construction 

o Rotational speed actuator 

o Rotational speed sensor 

o Pressure actuator 

o Pressure sensor 

 Test environment documentation 

 

These subsystems and subtasks were developed independently but with the coming 

integration in mind.  

 

3.1 Designing the experiment 

The goal of the experiment was to explore the influence of the rotation speed of the 

spray-arms and the water pressure on the dishwasher cleaning performance. The 

purpose was to answer the following question: Can a higher cleaning performance be 

achieved by changing the water pressure or the spray-arm rotation speed? 

 

The cleaning performance was selected to be the dependant variable. This was the 

obvious choice considering the goal of the experiment. How this was measured is 

described in section 3.2. The experiment goal also gives that the controlled independent 

variables were to be the water pressure in the spray-arms, and the rotation speed of 

these. It was decided that both variables should be controlled in the same experiment, 

since it was thought that this would increase the external validity. All other independent 

variables, such as temperature and humidity, were kept constant and closely monitored 

(see Chapter 3.2 Development of simplified test method). 

 

The levels of interest in the experiment were set, according to instructions from ASKO, 

from low rotation speed to high rotation speed with very low pressure to very high 

pressure. This is a span set around the original machine settings. 

 

It was decided to use a modified form of factorial design with 3 levels as depicted in 

Figure 3.1. (This was later changed, see 3.3.4 Water pressure actuator) 

 



12 

 

 

Figure 3.1: Preliminary levels of experiment 

 

The tests were made in arbitrary order. The results were then analyzed using JMP 

Statistical Discovery Software v. 8.0 (JMP) which automatically calculates accuracy in 

different factor spans and presents the result in a 3D response surface plot. 

 

The response surface steepest ascent method was not applicable in the experiment 

because of the relatively low accuracy. It was decided that it was more important to get 

a larger general picture of the interrelations of the rotational speed and pressure than 

being able to pinpoint the exact maximum. It was also decided to do all testing points 

since not enough dependable a priori knowledge was available to be able to exclude any 

points (as in a fractional factorial design). 

 

3.2 Development of Simplified test method 

The standardized test methods used at ASKO (and at other appliances developers) to 

measure the dishwasher cleaning performance are intricate and strictly defined. 

Specified amounts of foodstuff are placed on different types of plates and glasses. They 

are then subjected to heating and drying in microwave ovens before they are placed in 

the dishwasher. The dishes are washed using a specific detergent and a specific 

program, and are then examined and graded by an expert. Performing the planned 

experiment using these methods requires more time and resources than what was 

available for the master’s project. Therefore, a simplified test method was developed to 

be used at the lab facility at Chalmers. The purpose of the simplified test method was to 

save time and resources and still deliver an accurate and reliable test result. The test 

method had to be able to detect changes in input and indicate this with notable changes 

in output, but not so large changes so that the output becomes saturated. 

 

A table of requirements was set up listing all the needs of a functional test in a 

structured way. These were graded in order to verify which parameters were most 



13 

 

important. The entire target specification can be found in Appendix A. The target 

specifications was reviewed and consulted several times during the development of the 

simplified test method. 

 

3.2.1 Design parameters 

By dividing the simplified test method into its main components or functions the design 

parameters available for alteration were revealed. A performance indicator that would 

be able to reflect results of changing water pressure and rotational speed of the 

cleansing arms was determined to be dependent on several parameters. Through 

empirical studies and tests the influence of the different parameters on the final result 

could be determined. Solutions for how to practically alter these parameters were then 

acquired through research and brainstorming sessions. These were then put in a 

morphological matrix in order to be able see the alternatives in a structured and easy 

way and make a selection (see Appendix B). The final selections are presented below. 

 

Indicator 

The indicator is the object that is placed in the dishwasher with the soiling agent applied 

to it. The spontaneous choice of indicator was glasses, dishes or other kitchenware, 

which would have a high degree of ecological validity. Custom made indicators with 

different shapes, materials and textures were considered as well. The simplicity and 

high ecological validity of regular plates were the main reasons to why white plates 

with a diameter of 25 cm were chosen. These are standard plates used in the 

standardized tests at ASKO (Artzberg, European standard plates). They are easy to 

come by, naturally fit in the dishwasher and provide an easy way of applying a soiling 

agent as well as placing them in the machine in a consequent way. Both European 

standard plates and Australian standard plates were tested, but the European proved to 

have the right surface properties. Cutlery and other kitchenware of more complex forms 

were disregarded to keep the test as simple as possible. Plates were thought to be a 

satisfying representative for all kitchenware used in a dishwasher. 

 

Soiling agent 

The soiling agent is the substance placed on the indicator plate. It should be resistant 

enough not to be removed completely in the washing process, but not as tough as not to 

be removed at all. A lot of research in this project was made to find either food or 

synthetic paint that would fit the requirements of the test. Table 3.1 displays the 

different substances that were regarded and the factors influencing their applicability. 

Through preliminary simplified washing tests and research the decision was made to 

use paint from artistic markers, applied using the marker itself. 

 

  



14 

 

Table 3.1: Substances tested for usability in simplified test 

Substance Reasoning Result 

Egg + High ecological validity 

- Difficult to apply and measure accurately 

- Difficult to analyze 

- Relatively costly to use in large quantities 

- Not tough enough without treatment (heating) 

Not selected 

Balsamic 

vinegar glace 

+ High ecological validity 

- Time-consuming to apply 

- Difficult to analyze 

Not selected 

UV-pen + Quick and easy to apply 

+ Easy to standardize 

+ Relatively cheap 

+ UV-light could be used to detect even the tiniest spots 

- Not tough enough to withstand washing 

- Low ecological validity 

Not selected 

Spray-paint + Quick to apply 

+ Easy to standardize 

+ Relatively cheap 

- Too tough to get off during washing 

- Difficult to apply 

- Low ecological validity 

Not selected 

Permanent 

markers 

+ Quick and easy to apply 

+ Easy to standardize 

+ Relatively cheap 

- Too tough to get off during washing 

- Low ecological validity 

Not selected 

Artistic 

markers 

+ Quick and easy to apply 

+ Easy to standardize 

+ Relatively cheap 

- Low ecological validity 

Selected for tests 

 

Another factor of interest is the application method. Given the decision of using 

markers painting was seen to be the most obvious choice. The painting process was 

made as standardized as possible. Templates were made to cover the plates while 

painting. Some different patterns and sizes were tested, which can be seen in Figure 3.2. 

The pattern depicted in Figure 3.2a. was tried on the small plates in the upper basket 

and figure 3.2b. and 3.2c. were tested on the larger plates in the lower basket. 

 

 

Figure 3.2: Painting patterns tested on plates 



15 

 

The final choice of pattern, a simple square, can be seen in Figure 3.3. 3.3a. depicts the 

pattern for the large plates and 3.3b. for the small ones. This pattern was decided to be 

the best choice since it would cover a larger area of the plate, making differences in the 

surface less important. It would also be easier and faster to analyze the result using the 

computer vision method (see Analysis method below). 

 

 

Figure 3.3: Final painting pattern 

 

The templates were made from thick building plastics and masking tape, as can be seen 

in Figure 3.4. 

 

During the initial tests it was noticed that variances in the plate surface influenced the 

result a lot. Plates that were more worn were harder to clean than less worn ones. 

Observations were also made of differences on the same plate. Therefore it was decided 

to use the same area of the plates every time. To ensure that the same area was painted 

every time permanent markings were made (see Figure 3.4) within which the indicator 

paint was applied. The masking tape had to be replaced approximately every 10
th

 test 

run since its adhesive ability decreased due to the repeated painting. The permanent 

markings had to be repainted roughly every 8
th

 test run. 

 

 
Figure 3.4: Painting of test plates 



16 

 

There is also the influence of the preparation method. The soiling agent could be 

prepared for instance by drying, heating, burning or freezing (see morphological matrix 

in Appendix B). This would affect the properties of different soling agents, especially 

foodstuff
2
. 

 

Using the markers, heating the paint using a hairdryer resulted in the paint being much 

harder to remove. This method of regulating the adhesiveness of the paint was very 

difficult to standardize, and was therefore rejected. Another method that would be 

applicable was letting the paint dry for different amounts of time. Tests were made in 

order to determine if the amount of time the paint was allowed to dry had an effect on 

the result. Paint was allowed to dry for 10, 5, 2 and 1 minutes before the machine was 

started. No differences in the result could be observed by visual inspection, but further 

investigations were needed. A more structured test was done using the computer vision 

analysis method (see Analysis method below) and the standard test setup (see Chapter 

3.5 Experiment Execution). The plates were cleaned with detergent using the 

Quickwash-program, then rinsed clean from detergent using the Quickwash-program 

again, without detergent. The tests differed in the amount of time the paint was allowed 

to dry. With 20 minutes drying time 83% of the paint was removed, and with no drying 

time 92% was removed. Apparently, the waiting time was an important independent 

variable, so this was kept constant throughout the experiment. 

 

Dishwasher environment 

When considering the dishwasher environment, ecological validity was thought to have 

a substantial influence on the external validity. Placement, spray angles and such would 

most likely affect the result. Therefore, the placement of the plates was set as can be 

seen in Figure 3.5. 

 

 
Figure 3.5: Plate placement in lower (left) and upper (right) baskets 

                                                 
2
 Patrik Ivarsson, ASKO Appliances AB 



17 

 

It was decided to place plates A3, A4, A5, B1, B2 and B3 behind another plate to better 

mimic the real life situation of a fully loaded dishwasher. The placement was not altered 

during the experiment; all plates had their specific placement. The rest of the machine 

was left empty. It was properly cleaned several times before the experiment 

commenced. The filters were cleaned at a regular basis (once per day, approx. every 

fifth test run). 

 

The detergent used to clean the plates between each test run was Sun Professional 

Maskindisk (Supplier: DiverseyLever). ASKO standard testing detergent (Ref. 

detergent Type B, Batch: GSMB 177-280, 07.12.2009) was tested but it was not 

capable of removing the paint. 

 

Program 

The program used during the experiment was the Quickwash program which takes 

about 12 minutes to run. It first takes in water heating it up to 30 degrees C while 

spraying the dishes. It then empties this water. After this it takes in water (without 

heating it) and sprays the dishes for a couple of minutes and empties the water again. 

This is done twice. (see flowchart in Figure 3.6) 

 

 

Figure 3.6: Program flowchart 

 

The Quickwash program time varies somewhat depending on differences in input water 

temperature. This was considered to be a part of the inherit system behavior and to 

contribute to a higher ecological validity, even though this would leave the time 

variable uncontrolled. 

 

Analysis method 

Different methods for analyzing the result were tested. One method that was tested was 

visual inspection by an analyst, where the analyst estimates how large a percentage of 

the marked area was cleaned during the test. The indicator plate was inspected spot by 



18 

 

spot, where each of the 28 painted squares in the grid (as can be seen in Figure 3.2b.) 

was graded between 0 and 1 in steps of 0.05. The grades for all squares were added and 

then divided by 28 to acquire a relative score. After some initial tests it was apparent 

that the grading could vary somewhat each time the analyst estimated an indicator plate. 

The grading also varied depending on different analysts. The accuracy of the analysis 

method was approximated to ± 10%. 

 

To get better accuracy, tests were made with a computer analyzing the indicator plates. 

This was done by photographing the plates with a digital camera from a set angle and in 

a specific lighting, and by importing the data to a computer. A digital single-lens reflex 

(DSLR) camera was used to take all photos of the plates (see Table 3.2). The cameras 

capability was sufficient for the test. All camera settings were kept constant during the 

tests. A tripod was used to hold the camera, keeping it stable and at the right distance, 

which was approximately 1.5 meters between lens and plate surface. 

 

Table 3.2: Camera specifications 

Camera Model Canon 400D 

Lens Model Sigma 24-70mm 1:2.8 EXDG 

Program Manual 

Focus AI focus  70mm 

Shutter speed, aperture and ISO 1/40 second, F6.3, iso400 

Distance between lens and plate surface 1.5 meter 

 

The photos were imported to a computer and photo editing software (GIMP 2.6.8) was 

then used to manually cut out the painted area. In order to find the edges of the painted 

area they were marked with a permanent marker that was not removed during the test. 

These permanent markings also guaranteed that the same area of the plate was used 

every time, thus eliminating the problem with the surface of the plates influencing the 

results (see Soiling agent). 

 

A threshold-filter was then applied to the cropped image (can be seen in Figure 3.7). 

The filter sets the color to either black or white depending on the threshold level, pixel 

by pixel. The threshold was set manually by visually inspecting the result and selecting 

a best fit. These manual tasks were recognized as a source of inaccuracy. In order to 

determine how much these influenced the result a series of tests were made. Two 

analysts got a set of five images of plates which they were to analyze three times each 

and then compare the results. The average standard deviation was less than 1 %. 

 



19 

 

  

Figure 3.7: Indicator plate before and after threshold filter was applied 

 

The black-and-white image resulting from the threshold filter was then saved as a raw 

data file. In a raw data file the image data is stored pixel by pixel in three bytes, one for 

each color ingredient, red, green and blue. A 5 by 5 pixel image would thus result in a 

75 byte large raw-file. This differs somewhat between software, but this is how GIMP 

stores the data. MATLAB was then used to import the data as a text string and count the 

number of white and black pixels and deliver the result in the form of a percentage of 

white. The whole process is depicted in Figure 3.8. 

 

 

Figure 3.8: Description of analysis method 

 

3.3 Development of test rig 

The following part describes how the test rig was developed and how product 

development methods were used. 

 

3.3.1 Development procedure 

The development process was done in accordance with the methods described in 

section 2.2. 

 

Problem identification 

In the beginning of the project the goals were identified and most of the information and 

needs were retrieved from ASKO Appliance AB. The following tasks were identified: 

 

 Rotational speed adjustment 

 Water pressure adjustment 

Then the tasks were considered based on system design principles described in 2.2.1, 



20 

 

thus the tasks were divided into subtasks or subsystems. 

  

 Rotation speed actuator 

 Rotation speed sensor 

 Water pressure actuator 

 Water pressure sensor 

 

From the literature study and company visits, important values were derived: less 

energy consumption, little water usage, low sound level and so forth. Furthermore, 

there were some other important values that should be dealt with during the whole 

project, such as safety. 

 

In addition, the project workflow should be considered as well to determine every step 

in the project to avoid extra steps or missing some steps, the workflow is easy to follow 

and useful for estimating the workload. 

 

Target requirement specifications 

Most of the requirements for the adjustment of water pressure and rotation speed were 

from ASKO and other safety requirements from Chalmers. But whatever the origin 

they were set in a way so that they were possible to validate and verify. 

 

In order to achieve the goal, some parameters of dishwasher which had the largest 

potential in adjusting pressure and rotation speed within the timeframe and did not 

hinder other functions, were put in focus. Some of the parameters of dishwasher were 

excluded early, because of their strong effect on whole washing system or other 

problems. 

 

Concept generation and screening 

The brainstorming and morphological matrix methods were used when generating the 

initial concepts (see Appendix B). To be able to select the best concepts a Kesselring 

matrix was used. The requirements were listed and rated on a scale from 1 to 10 on their 

relative importance to the overall goal, from trivial factors to highly prioritized factors. 

Concepts were then rated on how they fulfilled the requirements, also on a scale from 1 

to 10. This rating was based on engineering analysis and logical evaluation. The two 

best concepts for each subsystem were selected for further development. Some of the 

key requirements were explored further for each concept. For details on the selection 

see Appendix C. 

 

Selection of final concept 

The final selection was made based on further studies of the primarily chosen concepts. 

This was done by re-looping the Kesselring matrix and adjusting the values according 

to the latest findings. Integrating with ASKO’s requirements on the solution, the final 

requirement specifications were accomplished which included weighting depending on 

their significance. The different requirements have to the greatest possible extent been 



21 

 

connected to metrics. 

  

Final specifications and system construction 

The final specifications were made specific enough to be usable during the part 

selection and detailed construction. In the final stage the prototype was built based on 

the final specifications. 

 

3.3.2 Rotation speed actuator 

Established target specification for rotation actuator: 

 

FR1 Possible to achieve target value 

FR2 Quick to prepare for use 

FR3 Measure velocity with high accuracy 

FR4 Not affect other parameter 

FR5 Affect other function as little as possible 

FR6 Robust enough to withstand washing procedure 

FR7 Durable 

FR8 Be able to work in different temperatures 

FR9 Be water proof 

FR10 Physically possible to produce prototype within time frame 

FR11 Low cost 

FR12 Be safe to use 

D1 Should be easy to set the value 

D2 Modify the dishwasher as little as possible 
 

 

Initial concepts 

Several design parameters were set in the left columns in the morphological matrix, 

such as adjustment system and influence method. Then, more possible methods were 

filled in the matrix form. All the details are shown in Appendix B. The following 

concepts were generated: 

 

Concept: Frictioner 

Friction is created between the 

rotating spray-arm and some 

part attached to the machine 

base, which causes the rotation 

to decrease. The friction could 

be regulated by increasing the 

force applied to the spray-arm. 

 

 

 

Concept: HeavyArm 

Figure 3.9: Concept Frictioner 



22 

 

Weights are added to the tips of the spray-arm in order to make it rotate slower. The 

weights can be varied to attain different speeds. 

 

 Figure 3.10: Concept HeavyArm 

 

Concept: Turbine 

Air-pressure is used in a completely separated and closed system to propel turbine-like 

propellers that are directly connected to the spray-arms. The pressure in the system can 

be regulated by using a potentiometer. 

 

 

 Figure 3.11: Concept Turbine 

 

Concept: ExMotor 

An external electrical motor is used to 

regulate the speed. The motor is connected 

to an axis inside or outside the machine 

which is connected to the spray-arms 

through timing belts. The motor is regulated 

either by using a potentiometer or pulse 

width modulation through a microcontroller. 

 

 

 

Concept: InMotor 

A waterproof motor is placed in the 

machine in direct connection to the 

spray-arm axis. A hole is cut out in the 

bottom in order for it to fit. The motor is 

regulated by using a potentiometer. 

 

 

 

 

Concept selection 

A Kesselring matrix was used in selecting the concepts for further development; all 

details are shown in Appendix C. The winning concepts were the “External motor” 

and “Turbine” as can be seen in Table 3.3. Although the “Heavy arm” concept got a 

Figure 3.12: Concept ExMotor 

Figure 3.13: Concept InMotor 



23 

 

higher score, it was disregarded due to its inability to increase the rotation speed. 

 

 Table 3.3: Rotation actuator Kesselring matrix results 

Concepts Frictioner Ex-motor In-Motor Turbine Heavy arm 

Scores 564 715 603 614 643 

 

Concept: Turbine 

Air-pressure is used in a completely 

separated and closed system to propel 

turbine-like propellers that are directly 

connected to the spray-arms. The system is 

connected to an existing air pressure 

source and a mechanical valve is added to 

be able to regulate the air flow. Positive air 

flow would make the blades rotate faster 

and negative air flow would slow them 

down. The positive or negative air flow 

could be controlled by switching inlet. 

 

Construction time: This system would be quite difficult to construct and take several 

days in the prototype lab. 

 

Modification level: A lot of space would be required by the air system, which could 

influence the spray pattern in the machine. It might also lead to larger modifications in 

the basket system and the indicator plate placement. 

 

Robustness: It is hard to estimate how such a complex system would hold out during the 

experiment. 

 

Usability: It would probably be quite easy to set the desired levels of speed, and all 

desired levels could be set. 

 

Cost: A rough estimation of the cost of the parts that could not be acquired in-house is 

500 SEK. 

 

Concept: ExMotor 

The concept has been divided into three versions of the original idea, depending on the 

different placement of the axis. The basic idea is still the same with an external motor 

that can be used to set the wanted speed. The motor is controlled using a potentiometer 

to regulate the effect of the motor. 

 

Figure 3.14: Concept Turbine 



24 

 

 

Figure 3.15: Concept Internal axis version 1 

 

    

Figure 3.16: Concept external axis   Figure 3.17: Concept Internal axis version 2 

 

Construction time: Approximately 3 days 

 

Modification level: It would probably not influence the performance of the dishwasher. 

It would lead to some modifications of the basket system. 

 

Robustness: If it is made from stainless components it would be robust. 

 

Usability: It would probably be quite easy to set the desired levels of speed, and all 

desired levels could be set. 

 

Cost: Motor ~300 SEK + timing belt and pulleys ~600 SEK + bearings ~150 

= ~1050 SEK 

 

 

Final selection 

The “Internal axis version 2” was selected for the rotation actuator subsystem. This 



25 

 

was considered the best mainly because of its less complex construction and 

robustness. A hole in the top of the machine would leak less water than a hole in the 

machine wall. It would also be easier to connect the motor to the axis as specified in 

the “Internal axis version 2” compared to the other two “ExMotor” derived concepts. 

 

Detailed construction - ExMotor 

One drawback of the construction was that both spray arms would be rotated in the 

same direction. This was accepted as a loss of ecological validity in favor of the 

simplicity of the construction. 

 

A hole was drilled into the top of the dishwasher in the back left corner. A 

construction made of wood and a 1 mm steel plate (as can be seen in Figure 3.20) was 

fastened on top of the hole. This allowed for an easy connection between the motor 

and the axis. A steel axis with a diameter of 8 mm was used, which was supported by 

two bearings, one in the upper wooden holder (Figure 3.18) and one in the lower 

wooden holder (Figure 3.19). The lower wooden holder was fastened to the back of 

the machine using a bent steel plate. 

 

      
Figure 3.18: Upper bearing holder and motor holder  Figure 3.19: Lower bearing holder 

 

The bearings had an inner diameter of 8 mm and an outer diameter of 22 mm. They 

were slightly lowered into tapered holes in the wooden holders to keep in place. The 

bearings took up the radial loads. The axial load caused by the weight of the axis and 

the timing belt pulleys was handled by the motor, which had a maximum limit of 5 N 

of axial load. 

 

Two T5/18-2 (Pitch: 5mm, Number of teeth: 18, Flanges: 2) timing belt pulleys were 

fastened on the axis at the height of the spray arms. These were connected to larger 

T5/36-2 pulleys, which had been fastened on the spray arms, by a 5 mm wide timing 

belt. This gave the pulley system a gear ratio of 2:1. The pulleys were made 10 mm 

thick in order to fit around the spray arm holders without hindering the rotation. The 

construction can be seen in Figure 3.20. 

 



26 

 

 

 

Figure 3.20: Timing belt and pulleys mounted in machine 

 

When selecting a motor it was decided that a DC-motor was the best choice since the 

rotation speed could then easily be regulated by changing the input voltage. To select 

a motor that could be regulated within the desired rotation speed range with sufficient 

torque, some rough calculations were made. The motor would have to be strong 

enough to rotate the spray arms even at the highest water pressure level. The torque 

induced by the water pressure was measured using a force gauge. This was fastened 

by a thread at a distance of 0.2 meter from the spray arm center. The gauge indicated 

that the force was less than 2 N on both the upper and lower spray arm. This gives that 

the torque needed to counter both spray arms would be 2 * 2 * 0.2 = 0.8 Nm. Since 

the gear ratio was 2:1 only half the torque would be needed, 0.4 Nm. This was 

considered the maximum torque needed, not including extra torque needed to counter 

the friction. 

 

The spray arm rotation speed needed was masked rpm. Since the gear ration between 

the timing belt pulleys was 2:1 the motor would have to be able to rotate at least twice 

the highest speed, i.e. masked rpm. 

 

The motor selected was a Micro Motors s.r.l. model HL149, with a maximum torque 

of 0.15 Nm, a gear ratio of 43.3:1 and a rotation speed of 55 rpm at 24 V DC. 

The motor was fitted atop of the axis. This can be seen in Figure 3.20. 

 



27 

 

3.3.3 Rotation speed sensor 

Established target specification for rotation sensor: 

 

FR1 Be able to measure rotational speed 

FR2 Quick to prepare for use 

FR3 Measure velocity with high accuracy 

FR4 Not affect other parameter 

FR5 Affect other function as little as possible 

FR6 Robust enough to withstand washing procedure 

FR7 Durable 

FR8 Be able to work in different temperatures 

FR9 Be water proof 

FR10 Physically possible to produce prototype within time frame 

FR11 Low cost 

FR12 Be safe to use 

FR13 Easy to replace 

D1 Should be easy to get the value 

D2 Modify the dishwasher as little as possible 

 

Initial concepts: 

Several design parameters were set in the left columns in the morphological matrix, 

such as measuring method, placement and so forth. Then more possible methods were 

filled in the matrix form. All the details are shown in Appendix B. Finally some 

concepts have been generated as shown below: 

 

Concept: Click-Counter  

A device is constructed that has a feather-recoiling 

lever sticking out of its side. The lever is activated 

each time a small outshoot on the spray-arm axis 

passes it.  

 

 

 

 

 

 

Concept: Bike Computer 

A regular bike computer is attached to 

the rotating axis of the spray-arm. The 

output from the sensor is recalculated to 

get the rpm value. 

 

 

Figure 3.21: Concept click-Counter 

Figure 3.22: Concept Bike computer 



28 

 

Concept: Listen  

A person is trained to hear the rotation of the 

washer arm while the machine is running. A 

timer is used and the number of laps is 

counted during a specific time-span.  

 

 

 

 

 

 

Concept: Lookie lookie 

A replacement hatch is made from plexiglas through which a person can see the spray 

arms rotate and count the number of laps during a specific time-span. 

 

Concept: Light  

A black line is painted along one side of 

the spray-arm axis. A light-sensitive 

sensor is used to detect if the painted 

section is placed in front of it. A 

microcontroller is used to acquire data 

from the sensor and count the number 

of laps. 

 

Concept selection 

Kesselring matrix was used to select the concepts for further development, all details 

can be found in Appendix C. As can be seen in Table 3.4 the concepts “Bike 

computer” and “Listen” were selected. 

 

 Table 3.4: Rotation sensor Kesselring matrix results 

Concepts Click 

counter 

Bike 

computer 

Listen Light sensor Lookie 

Scores 657 822 907 730 814 

 

Concept: Listen 

By placing the ear against the machine you can hear the water splashing on the walls of 

the dishwasher and count the laps. A watch is used to time a set number of laps. The 

rotation speed can then be calculated using the following formula: 

 

𝜔 =  
𝑛

𝑡
  

ω: Rotation speed of spray arm [rpm] 

n: Number of laps 

t: Time [minutes] 

Figure 3.23: Concept Listen 

Figure 3.24: Concept Light 



29 

 

 

Construction time: No construction time. 

 

Modification level: No modification needed. 

 

Robustness: Quite robust. 

 

Usability: Very simple to use, but it might take a few minutes to set the desired speed 

level. 

 

Cost: 0 SEK 

 

Concept: Bike Computer 

A regular wireless bike computer is 

used to transmit the speed to a 

receiver outside the dishwasher. The 

value indicated by the computer 

could be multiplied with a constant 

to obtain the actual rotation speed. 

 

Construction time: This device 

could be set up in less than an hour. 

 

Modification level: The sensor could be placed so that it doesn’t influence the system at 

all. 

Robustness: The device was made to withstand outdoor use, so it should be quite 

robust. 

 

Usability: Very simple to use. 

 

Cost: ~150 SEK 

 

Final selection 

The “Bike computer” concept was selected as the best solution for the rotation speed 

sensor, mainly because it would be much easier and quicker to acquire the 

measurement. 

 

Detailed construction – Bike Computer 

A bike computer (Velleman Model: BC15S) was used to measure the rotation speed. It 

was attached to the axis on the outside of the machine, thus elimination the risk of 

getting damaged by water, as can be seen in Figure 3.26. 

 

Figure 3.25: Concept Bike Computer 



30 

 

 
Figure 3.26: Bike computer mounted 

 

3.3.4 Water pressure actuator 

Established target specification for water pressure actuator: 

 

FR1 Possible to achieve target value 

FR2 Quick to prepare for use 

FR3 Measure velocity with high accuracy 

FR4 Not affect other parameter 

FR5 Affect other functions as little as possible 

FR6 Robust enough to withstand washing procedure 

FR7 Durable 

FR8 Be able to work in different temperatures 

FR9 Be water proof 

FR10 Physically possible to produce prototype within time frame 

FR11 Low cost 

FR12 Be safe to use 

D1 Should be easy to set the value 

D2 Modify the dishwasher as little as possible 

 

Initial concepts: 

Several design parameters were set in the left columns in the morphological matrix, 

such as power source and data communication method. Then more possible methods 

were filled in the matrix form. All the details are shown in Appendix B. Some 

concepts have been generated as follows: 

 

Concept: PowerMotor 

To be able to increase the water pressure the motor is replaced with a more powerful 

one. Due to its size it may not fit underneath the dishwasher and could be placed behind 

instead. 



31 

 

Concept: Restrictor  

The water is partly blocked by a metal restrictor 

plate with a small hole in it. By varying the size 

of the hole, different pressures can be acquired. 

 

 

 

 

Concept: ArmVolume 

Different sets of sprayer arms are constructed with different inner volume. The volume 

is regulated to get different pressure. 

 

 

Concept: Holy 

Holes are made in the pipe that leads to the 

spray arm in order to decrease the 

pressure. 

 

 

 

 

Concepts selection 

A Kesselring matrix was used to select the concepts for further development; all 

details are shown in Appendix C. The “power motor” and the “Restrictor” were 

selected, as can be seen in Table 3.5. 

 

 Table 3.5: Water pressure actuator Kesselring matrix results 

Concepts Power motor Restrictor Arm volume Holy 

Scores 755 709 595 566 

 

Concept: PowerMotor 

The original motor should be replaced with a more 

powerful one which can be controlled to run at different 

power levels. 

 

Construction time: Help would probably be needed from 

specialists at ASKO which would take some time to 

organize and would be costly for ASKO. 

 

Modification level: The larger motor would not fit in the 

regular compartment and would have to be placed 

behind the dishwasher. This could possibly increase the sound level of the machine. 

 

Robustness: Since the motor is a standard part made specifically for a dishwasher it 

Figure 3.27: Concept Restrictor 

Figure 3.28: Concept Holy 

Figure 3.29: Concept PowerMotor 

 



32 

 

would probably be quite robust. 

 

Usability: It would be easy to regulate the desired pressure level, and it could be set to 

any level in the span of interest. 

 

Cost: Parts could be obtained from ASKO. No external costs. 

 

Concept: Restrictor 

Thin circular plates with different sized holes should be 

manufactured. These are to be placed between the spray arm 

water input and the water pipe. The hole partly shuts off the 

water supply, thus decreasing the pressure in the water exiting 

the spray arm. 

 

Construction time: These could easily be constructed from 

metal or plastic in a day or two. 

 

Modification level: It is a bit uncertain how the change in 

pressure will affect the rest of the machine. It might cause the 

motor to run slower. Apart from that the influence on the rest 

of the system would be minimal. 

 

Robustness: The wear on the restrictors would be minimal, they would be very robust. 

 

Usability: It would take some time to get them set at the right pressure level. Once this 

is done they will be very simple to change and use. It is not certain that all pressure 

levels could be attained. 

 

Cost: Plastic or metal plates could be obtained for free from Chalmers. 

 

Final selection 

For the pressure actuator subsystem the Restrictor concept was selected due to its 

simplicity and low influence on the overall sound level. 

 

Detailed construction - Restrictor 

The restrictor plates were constructed using 2 mm thick plastic that was cut into 

circular plates that would fit inside the spray arm holder (diameter of 21 mm). 

Different sized holes were drilled in the center of the plates. Different sizes were 

tested and finally 7.1 mm and 8.2 mm holes were used which gave a pressure level of 

“very low” and “low” respectively in the lower spray arm. Due to the extra space 

required by the plates a thin layer had to be rasped off of the bottom of the spray arm 

holder. Tests were made with very thin aluminum plates, but they were too weak to 

withstand the pressure without bending. 

Figure 3.30: Concept 

Restrictor 



33 

 

3.3.5 Water pressure sensor 

It was thought that the easiest and most accurate way of measuring the pressure would 

be to use a standard pressure meter. Therefore, no concepts were developed for a 

pressure sensor.  

 

A pressure sensor was required to be able to set the spray arm pressure at the desired 

levels. A temporary sensor was constructed in order to verify the initial theory 

supporting the Restrictor concept. Using this temporary sensor a relative measurement 

could be attained, which would then be supported by a more accurate pressure sensor 

measurement later on. The reason why this temporary sensor was made was that 

constructing and fine-tuning the restrictors would take some time, and the more 

precise pressure sensor could not be borrowed for that amount of time. The temporary 

sensor was constructed using an air pressure sensor placed in a chamber separated 

from the water by a membrane. A 500 ml PET-bottle was used as pressure cell and 

tough plastic bags as membrane. The construction was sealed using glue. See Figure 

3.32. Several models were constructed before a usable one was obtained. See Figure 

3.31. 

 

Figure 3.31: Different temporary pressure sensor constructions 

 

 
Figure 3.32: The fourth working water pressure sensor 



34 

 

First sensor (Figure 3.31a.) 

Pressure cell volume: 0.5*10
-3

 m
3
 

Membrane area: 8.5*10
-3

 m
2
 

Sensor: ASKO DW20 

Results: Not usable. The signal was saturated due to a too large membrane. It could be 

improved by using a smaller membrane or a pressure sensor with a larger measuring 

span. 

 

Second sensor (Figure 3.31b.) 

Pressure cell volume: 0.58*10
-3

 m
3
 

Membrane area: 0.35*10
-3

 m
2
 

Sensor: ASKO DW20 

Results: Not usable. Too small changes in signal due to a too small membrane. It 

could be improved by using a larger membrane or a more sensitive sensor. 

 

Third sensor (Figure 3.31c.) 

Pressure cell volume: 0.4*10
-3

 m
3
 

Membrane area: 2.8*10
-3

 m
2
 

Sensor: ASKO DW20 

Results: Not usable. The sensor values showed signs of drifting and did not return to a 

zero level between tests. This was probably caused by a leak or a damaged sensor. 

  

Fourth sensor (Figure 3.31d.) 

Pressure cell volume: 0.5*10
-3

 m
3
 

Membrane area: 8.5*10
-3

 m
2
 

Sensor: ELFA pressure sensor SPD-015-G-2 

Results: The results can be seen in Figure 3.33. 

 

The initial tests resulted in an implied relationship between the hole diameter and the 

pressure as can be seen in Figure 3.33. 

  

 
 Figure 3.33: Results from preliminary water pressure tests 



35 

 

It was detected that the restrictors could not be used to increase the pressure, only to 

decrease it. This meant that either the experiment boundaries had to be changed or 

another concept would have to be selected for development. Reasoning that the 

Restrictor concept would be easier and faster to implement and that the competing 

concept, using a more powerful motor, would make sound tests unusable (which were 

thought to be very important at ASKO), it was decided to change the experiment 

boundaries. The new pressure levels were set as in Figure 3.34. 

 

 

Figure 3.34: New experiment levels 

 

Some different restrictors were prepared for the 

more accurate pressure test. The pressure sensor 

used was an Elite Digital Manometer HM3500, 

DLG300 (serial no. 1020203) used at ASKO 

(see Figure 3.35). It was set to measure the 10 

seconds average pressure in kPa. The standard 

deviation of the readings was 0.21 kPa. 

 

The sensor was placed over the holes indicated 

in Figure 3.36. A simplification was made; the 

same size of restrictors was used in the upper 

and lower spray arm. This meant that the 

pressure levels for the upper spray arm would 

not be evenly distributed across the pressure 

span. It would be very difficult and time-consuming to match the restrictors to get the 

exact wanted levels in both the upper and lower spray arms, so this simplification was 

accepted considering the time gain. The results from the upper test could thus not be 

evaluated in the same way as for the lower one. 

 

The two extra spray holes in the bottom of the machine that provide extra water on the 

dishes from below were covered using glue. 

 

  Figure 3.35: Pressure measurements 



36 

 

 
Figure 3.36: Spray arm holes used for pressure tests 

 

3.4 Test environment documentation 

Independent variables which were to be kept constant were strictly monitored and 

documented. 

 

Temperature 

The room temperature and in-machine air temperature were measured, but could not 

be kept constant. This was not considered a problem since it varied very little and was 

thought to have little influence on the dependent variable. 

 

Input water temperature at the water tap was also measured. This varied somewhat 

during the day due to variations in water usage in the building. The influence of the 

water temperature on the machine was considered to have minor effects on the results, 

but it would affect the washing-time which in turns would affect the results. 

 

Humidity 

Room and in-machine humidity was measured and documented. These variables 

could not be controlled but were quite stable during the tests. 

 

The temperature and humidity measurements were done using a simple measuring 

station built as depicted in Figure 3.37. This allowed for quick and accurate reading of 

sensor values. 

 
Figure 3.37: Sensor rig circuit 



37 

 

 

The circuit is depicting three National Semiconductor temperature sensors model 

LM35 TO-92 (accuracy ±0.5°C) and two Sencera humidity sensors model H25K5A 

(accuracy ±5%RH). A photo of the rig can be seen in Figure 3.38. 

 

 
Figure 3.38: Sensor rig 

 

Water properties 

Water properties, such as chemical composition and water hardness, could not be 

controlled. This was recognized as a source of inaccuracy. The water hardness was 

considered to be at a fairly constant level throughout the experiment, and was 

measured to 3 ºdH. 

 

Time 

Due to the construction of the dishwasher program washing time varied depending on 

the input water temperature. Colder input water resulted in the machine taking longer 

time to heat it to the set level, thus prolonging the washing procedure. The time was 

measured using the machines built-in time indicator. The small changes in washing 

time were considered to have a minor effect on the test results. This was considered an 

acceptable source of inaccuracy since it was thought that using a pre-set program 

would contribute to the ecological validity, as mentioned earlier. 

 

Sound level 

Sound tests were made during the experiment to see how a change of spray arm 

rotation speed or water pressure would influence the general machine sound. For this 

purpose a Reed sound level meter model ST-805 was used to measure sound levels in 

dBA and dBC. It was run in slow mode which presented the 1 second averages with 

an accuracy of ±1.5 dB. The measurement was taken during the washing period of the 

program at a distance of 200 mm from the machine front, as indicated in Figure 3.39. 

The results are presented in chapter 4.2 Sound level tests. 

 



38 

 

 

Figure 3.39: Sound meter placement 

 

Energy and water usage 

For every test-run the approximate energy usage was measured. The amount of water 

used by the machine was also measured. Energy and water usage were seen as system 

indicators rather than independent variables related to the washing performance. They 

were collected to verify that the system was running as it should. 

 

3.5 Experiment Execution 

A defined order of when and how to do each step in the test was set before the 

experiment commenced. This was done to minimize the risk of inaccuracies in the 

result due to variances in how the test was performed. One entire test run is described 

step by step below. 

 

First, the plates were washed clean using approximately 10 ml Sun detergent and 

dried using paper. They were then painted within the specified area using the template 

and left to dry for 10 minutes. 

 

While the plates were drying the air, in-machine and input water temperatures were 

measured as well as the humidity. The correct restrictors were put in place in the 

machine and the energy measurement instrument was set to zero. 

 

When 10 minutes had elapsed the speed of the spray arms were set by fine tuning the 

output voltage from the power supply unit while checking the rotation speed on the 

bike computer. The plates were set in place and the Quick program started. After the 

first water intake sound measurements were made when the spray arms were rotating. 

When the machine had finished the energy usage was read and the output water was 

measured, which had been collected in a bucket. Photos of the plates were taken and 

the plates were put in the machine again while the results were analyzed. 

  



39 

 

4 – INTERMEDIATE RESULTS 

Tests were only done on plates in the lower basket due to initially bad results. It was 

decided to continue with the lower ones and add the upper in a second experiment. 

 

4.1 Performance 

The performance is presented as a cleanness percentage in Figure 4.1. If the score is 0 

it means that the plate was not cleaned at all, if the score is 100 it means that the plate 

was totally clean. 

 

 

Figure 4.1: Performance scores of first experiment 

 

 Table 4.1: Detailed results from first test 

Pressure 

[kPa] 

Rotation 

speed [rpm] 

1
st
 

Test 

2
nd

 

Test  

3
rd

 

Test 

4
th

 

Test 

5
th

 

Test 

6
th

 

Test 

Avg. σ 

Very low Low 34 47 59 -- -- -- 46.7 12.5 

Middle 37 39 54 -- -- -- 43.3 9.3 

High 36 53 52 -- -- -- 47.0 9.5 

Low Low 36 60 61 -- -- -- 52.3 14.2 

Middle 53 42 44 59 56 64 53.0 8.6 

High 37 57 62 -- -- -- 52.0 13.2 

Middle Low 54 64 76 -- -- -- 64.7 11.0 

Middle 37 40 59 -- -- -- 45.3 11.9 

High 37 63 63 -- -- -- 54.3 15.0 

 

As Table 4.1 shows, the lower basket cleaning performance seemed to be not so stable. 



40 

 

The standard deviation is quite large at almost every point. E.g. the standard deviation 

is nearly 15 % at low pressure level with low rotation speed. 

 

The test result proved inconclusive, so some improvements to the test method would 

have to be made in order to get more stable data. During this test, one can see a trend 

that middle pressure with low rotation speed could enhance the cleaning performance. 

The cleaning performance is improving when the water pressure was increased. In 

addition the rotation speed only affected the cleaning performance to some extent in 

this experiment. The response surface figure and its sectional view are represented in 

Figure 4.2 and 4.3. 

 

 

Figure 4.2: Response surface plot of results from the first experiment 

 

 

Figure 4.3: Response surface contour plot of results from the first experiment 

 



41 

 

In the Figure 4.3, the contour lines show the level of cleaning performance at two 

variables is a curve with a constant value. As can be seen, at a rotation speed between 

high speed level and middle speed level cleaning performance changes very little with 

increasing pressure, whereas at a low rotation speed the pressure has a lot more 

influence on the performance. Still, this is only an indicator since the variation was 

very high. 

 

4.2 Sound level tests 

The results from the sound tests are presented in Figure 4.4. 

 

 

Figure 4.4: Results from sound test at Chalmers laboratory 

 

The differences between testing points are not very large. In fact most changes are 

smaller than the measurement equipment accuracy of ±1.5 dB. This indicates that the 

sound level does not change much depending on pressure or rotation speed within the 

span that was tested. The measured sound levels cannot be seen as absolute values, 

since the machine could not be isolated from all other sound sources in the laboratory. 

The ambient sound level in the laboratory was perceived to be fairly constant during 

the tests. As can be seen in the difference between the dB(A) and dB(C) levels, there 

was a large contribution of sound below the range of human hearing. 

 

Taking both equipment inaccuracies and unpredictable laboratory ambient sound into 

account it was decided to make more accurate sound tests at ASKO. The result from 

these can be seen in chapter 5.4. 

 

4.3 Energy consumption 

The energy consumption varied between 0.13 and 0.17 kWh. This was due to different 



42 

 

test run time, which depended on the input water temperature. The power 

consumption/input water temperature relation is plotted in Figure 4.5. 

 

 
Figure 4.5: Power usage contra input water temperature 

 

The graph suggests that the power consumption is proportional to the inverse of the 

temperature of the input water. 

  



43 

 

5 – FURTHER IMPROVED EXPERIMENT 

Since the measurements from the first experiment were subject to large inaccuracies, 

it was decided to try to analyze the sources of these inaccuracies and try to control 

them. In order to do this, a table of potential sources of errors was made and the 

estimated influence of each of these was analyzed (see Table 5.1). Suggested solutions 

to the potential problems were generated through brainstorming. 

 

 Table 5.1: Evaluation of possible sources of error 

Potential source of error Evaluation Solution 

Painting Paint Paint properties might change when 

markers have been used a lot. 

Do a test with a new and an old 

marker. If this influences: only use 

new markers. 

  Plates Worn and scratched plates might 

give an uneven result. 

Do tests with new plates. 

  Drying Drying time does influence the 

result. 

Is already kept constant. 

Time In machine Time is not constant during tests due 

to the dishwasher program. This 

must be kept constant. 

Use special settings in machine and 

use an external timer to keep time in 

machine constant. 

  Drying This has been done using paper 

which might influence the surface of 

the plate. 

Let plates dry without touching the 

surface. 

Temperature Room Has little effect on dishwasher 

performance. 

No changes needed. 

  Input water Does affect the time variable. Does not need to be controlled if the 

temperature in the machine is 

controlled as described below. 

  In machine Affects the performance a lot, and 

should be better controlled. 

The machine should be set to run until 

it reaches 30° C before the dishes are 

put in. 

Humidity   Uncertain how the humidity affects 

the results. Humidity has been 

constant during tests. 

No changes needed.  

Detergent Build-up This might be the reason for the drift 

in the previous results. Should be 

avoided.  

Wash plates one time extra without 

detergent between every run. 

  Wear on 

plates 

Surface properties might change 

during test due to wear caused by 

repeated washing with detergent. 

New plates should be washed several 

times before use, to get them stable. 

Water Properties Chemical composition of water 

changes constantly. It is difficult to 

evaluate the effects of this. 

If none of the other solutions work, all 

tests could be done at ASKO where 

strictly controlled water is used. 

 



44 

 

The outcome of this analysis was that time and temperature should be fixed, new 

plates should be used and the machine should be cleaned more properly between test 

runs. 

 

5.1 Research for new experiment 

Some research was done on how to stabilize the results according to the solutions 

from Table 5.1. Tests were done with new plates and paint. All changes to the 

experiment are described in detail below. 

 

5.1.1 Test of new plates 

In order to avoid problems with differences in plate surface influencing the results 

new plates were tested. The plates tested were: 

 

Färgrik Tallrik 27 cm, Color: white, IKEA ID: 301.462.75 

Lugn Tallrik 23 cm, Color: beige, IKEA ID: 133.123 

Dinera Tallrik 32 cm, Color: beige, IKEA ID: 100.570.67 

 

Before the plates were tested they were washed with detergent several times. The first 

tests proved fruitless; the plates were totally cleaned from all paint already after 30 

seconds of washing. Apparently the surface was too smooth for any paint to stick. In 

an attempt to make the surface more rough, the plates were scraped using a dish 

sponge. The sponge was too soft to affect the surface and the plates were still cleaned 

from all paint. 

 

Steel wool was used to very lightly scrape the plates. This resulted in more paint 

being left on the plates after the washing procedure, but the results were very 

unstable. To scrape the plates evenly across the surface proved difficult. Several 

attempts were made to stabilize the results but without success. 

 

Finally, it was decided to use the same set of plates as in the previous experiment, 

since they had proved stable enough independently, even though the surface 

properties varied within the set. 

 

5.1.2 Test of old and new marker 

Tests were made to determine if the properties of the marker changed during use. A 

whole test run was done with a new marker at machine standard speed and pressure. 

The test was re-done using an old, almost finished, marker. There was a difference of 

7% in the average plate score. The old paint was slightly more difficult to remove than 

the new one. It was decided that markers should be used until it was about 10 % full, 

then change to a new one to avoid this drift. 

 



45 

 

5.1.3 Stabilizing time and temperature 

To get the desired temperature level a temperature sensor was placed in the machine, 

the Quick-wash program was started and then stopped when the water temperature 

reached 30 ° C. The program was aborted, while keeping the hatch closed, and the 

service mode was accessed. The prepared plates were then inserted and the machine 

water pump was started. Time was controlled using an external analogue clock. This 

procedure was predicted to decrease inaccuracies due to changes in time and 

temperature. 

 

5.1.4 New pressure measurements 

To ensure that the pressure levels were still correct, the pressure was measured again 

using the same equipment as before. Pressure levels had dropped a little bit. The 

levels were adjusted accordingly, and the restrictors slightly modified to get the 

desired levels, but the name of pressure level were named as same. The new pressure 

levels can be seen in Figure 5.1. 

 

 

 Figure 5.1: New experiment levels 

 

5.1.5 New Experiment Execution 

The plates were washed clean using 5 ml Sun detergent, and then washed again 

without detergent to remove all residual detergent. They were put in a stand, at a 90° 

angle, to dry for 20 minutes. They were then painted within the specified area using 

the template and left to dry for 10 minutes. One template for each plate was used to be 

able to do the painting as simultaneously as possible. 

 

While the plates were drying the correct restrictors were put in place in the machine 

and the regular Quick-wash program was started. Input water and room temperature 

was documented. The start of the program was timed so that the temperature in the 

machine would reach 30° C when the 10 minutes were up. The machine was stopped 

and set into service mode which allowed for access to specific machine functions. 

Spray arm speed was set using the power supply unit and the bike computer. The 



46 

 

plates were put in place and the machine was then set to run the water pump. The 

plates were cleaned for three minutes, which was timed using an analogue clock. 

Photos of the plates were taken and the plates were put in the machine again while the 

results were analyzed. 

 

5.2 Cleaning performance results 

Due to the uneven pressure levels in the upper spray-arm these results could not be 

used. Only the lower basket results were analyzed. The evaluation methods were the 

same as for the first experiment. If the plate was not cleaned at all, it got a score of 0, 

and if the plate was totally cleaned a score of 100. The new test result can be seen in 

Figure 5.2. 

 

 

Figure 5.2: Performance scores of improved experiment 

  



47 

 

 Table 5.2: Detailed performance results from improved experiment 

Pressure [kPa] Rotation speed 

[rpm] 

1
st
 test 2

nd
 test 3

rd
 test 4

th
 test Avg. σ 

Very low Low 30.6 33.9 -- -- 32.3 2.4 

Middle 26.9 28.1 -- -- 27.5 0.9 

High 20.6 27.3 -- -- 23.9 4.8 

Low Low 40.7 34.9 -- -- 37.8 4.1 

Middle 27.2 33.4 30.5 32.7 30.9 2.8 

High 30.7 34.4 -- -- 32.5 2.6 

Middle Low 46.7 44.7 -- -- 45.7 1.4 

Middle 34.2 35.8 -- -- 35.0 1.2 

High 38.3 35.3 -- -- 36.8 2.2 

 

In the improved experiment, the standard deviation of the cleaning performance was 

much lower than in the intermediate result. The standard deviation between two tests 

at the same point is between 0.9 and 4.8 as Table 5.2 shows. 

 

During the new test, it was confirmed that high pressure at low rotation speed could 

enhance the cleaning performance, as suggested in the intermediate results. In 

opposite, the cleaning performance was decreased a lot at low pressure and high 

rotation speed. This trend is obvious considering Figure 5.2 and 5.3. 

 

 

Figure 5.3: Response surface plot of final results 

 

As can be seen in Figure 5.4 the contour lines indicate that the worst rotation speed is 

between middle speed and high speed. In another hand, the performance is more 

sensitive to changes in water pressure at lower rotation speeds. This is indicated by 

the density of contour lines around lower speeds in Figure 5.4. 



48 

 

 

Figure 5.4: Contour plot of final cleaning performance results 

 

5.3 Validation tests at ASKO 

The external validity of the test method was evaluated by performing some of the test 

points using the standardized test method at ASKO (see Figure 5.5). This was done to 

see how well the simplified test method generalizes, if the simplifications using fewer 

plates and paint concur with the standardized test methods. Due to the cumbersome 

procedure of the standardized test the only points tested were very low pressure with 

high speed and middle pressure with low speed. Two tests were made on each point. 

 

  

Figure 5.5: Standardized tests at ASKO 



49 

 

The expected outcome was that the lower pressure setting would acquire the worst 

results while the higher pressure setting would acquire better results. 

 

This was confirmed by the validation tests. Results are measured relative a standard 

machine, 1 would mean equal to standard machine and 1.05 would mean 5% better 

than the standard. The test results are presented in Table 5.3. 

 

 Table 5.3: Results from standardized tests at ASKO 

Pressure [kPa] Rotation speed [rpm] 1
st
 Test 2

nd
 Test 

Very low High 0.76 0.78 

Middle Low 1.07 1.12 

 

The higher pressure and lower rotation speed enhanced the cleaning performance by 7 

to 12% compared to the reference dishwasher. On the other hand, the cleaning 

performance went down 22 to 24% at the lower pressure level. This means a span of 

approximately 33%. These figures can be compared to the simplified test, which 

indicates an improved performance of approximately 15% at high pressure and low 

speed and a loss of performance by around 35% at low pressure and high speed, i.e. a 

span of 50 %. The simplified test method appears to be valid, but more sensitive than 

the standard method. 

 

5.4 Sound tests at ASKO 

More accurate sound tests were made at ASKO to document how changes in pressure 

and rotation speed affect the machine sound level. The tests were made according to 

ISO 3743 in a reverberation room built for this purpose. The machine was placed in a 

wooden cabinet used when testing according to standard IEC60704-2-3, as can be 

seen in Figure 5.6, to mimic the kitchen environment. An insulated plastic can was put 

on top of the motor to block out as much of the motor sound as possible. The machine 

was fully loaded with clean dishes. 

 

The test was run two times at every preset level of speed and pressure. Three tests 

were then run at different speed levels without any water in order to determine how 

much of the sound that was coming from the motor and spray arm pulley system. The 

results are presented in tables 5.4-5.7. 



50 

 

 

Figure 5.6: Sound test setup at ASKO 

 

In the third test at very low pressure and at the second tests at low pressure and middle 

pressure (all marked in red) there was a change in sound due to that a part of the spray 

arm holder broke down and had to be replaced. The accuracy at the same level was 

quite high and can be seen if the first and second tests in Table 5.4 are compared. 

Results in A-weighted sound power level relative to 1 pW. 

 

Table 5.4: Dishwasher sound power level test – very low pressure level 

Pressure level Rotation speed [rpm] First Test [dB(A)] Second Test [dB(A)] Third Test [dB(A)] 

Very low  Low 44.86 44.85 46.00 

Very low Middle 45.33 45.15 46.07 

Very low High 46.63 46.71 46.95 

 

Table 5.5: Dishwasher sound power level test – low pressure level 

Pressure level Rotation speed [rpm] First Test [dB(A)] Second Test [dB(A)] 

Low Low 46.32 46.34 

Low Middle 47.10 46.53 

Low High 47.86 47.34 

 

Table 5.6: Dishwasher sound power level test – middle pressure level 

Pressure level Rotation speed [rpm] First Test [dB(A)] Second Test [dB(A)] 

Middle Low 48.67 48.02 

Middle Middle 48.97 48.40 

Middle High 49.83 48.72 

 

 

 

 



51 

 

Table 5.7: Dishwasher sound power level test – background noise test 

Pressure level Rotation speed [rpm] Sound Test [dB(A)] 

0 Low 33.31 

0 Middle 37.90 

0 High 42.00 

 

Studying the frequency plot of the tests of Table 5.7 it was found that the pulley 

system contributed a lot to the background noise. This means that the tests cannot be 

seen as exact levels for the D5900, but the influence of rotation speed and water 

pressure can still be seen. This is depicted as a contour plot in Figure 5.7. 

 

 
Figure 5.7: Results of sound tests 

 

As the plot describes, the sound level is rising more rapidly at higher pressure levels. 

It appears that the rotational speed has less influence as the pressure increases. The 

water pressure obviously has the largest influence on sound power level. 

 

5.5 High pressure tests 

Some further tests were done to see if the trend of better cleaning performance would 

continue at even higher pressure levels. A more powerful water pump was put in the 

machine which could increase the water pressure in the lower spray arm. This 

pressure level is marked as high pressure level. Because of a difference in pump 

structure, the cleaning performance was not directly comparable to the previous 

experiment result. Therefore two of the same points as in the previous tests were done 

again so that the relative score could be compared. The new test was implemented at 

the middle pressure level and high pressure level and rotation speed as low speed 

level and high speed level. The test procedure was exactly the same as for the 

improved test. 



52 

 

The results are shown in Table 5.8. As can be seen in Figure 5.8 it seems that the 

initial hypothesis is valid. The best cleaning performance was acquired at highest 

water pressure and lowest rotation speed. 

 

Table 5.8: Results of higher pressure test 

Pressure [kPa] Rotation speed [rpm] 1st test [%] 2nd Test [%] Average [%] σ 

Middle  Low 19.0 19.1 19.1 0.1 

High 15.4 15.5 15.5 0.1 

High  Low 26.6 26.6 26.6 0.0 

High 20.8 20.7 20.8 0.1 

 

 

Figure 5.8: Results of higher pressure test 

 

  



53 

 

6 - DISCUSSION 

6.1 The simplified test method 

The test method, i.e. the use of plates, paint and camera, proved adequate for 

performing the tests. The small changes made in the independent variables could be 

detected as changes in the dependent variable without it getting saturated. The method 

was successfully used to map the influence of pressure and rotation speed on the 

dishwasher performance. It is hard to stabilize all influencing variables, a dishwasher 

system is quite complex and many factors influence the results. 

 

The internal validity was consid