88

95Faculty of Engineering, Kasetsart University

Design of Experiment for New Product Development from Polyisocyanurate Foam Residues

Sirang Klankamsorn Faculty of Engineering Sri Racha, Kasetsart University at Sri Racha Campus

Abstract

The problem in this research is to study a newly-developed product, an interlocking block and ventilation block, from PIR foam residue. This residue is left in large amounts during the PIR foam production process and can harm human health and affect the working place environment. The objective of this research is to determine suitable components for producing the interlocking block and ventilation block from the PIR foam residue by using the Design of Experiment technique (ANOVA: Analysis of Variance). The product weight is investigated in this research. A lightweight product is required from this experiment. The results show that the PIR foam percentage that is mixed in the block raw material components will provide a lower weight of the product.

Keywords : Polyisocyanurate Foam, Interlocking Block, Ventilation Block, ANOVA

Introduction

Energy control insulation and foam insulation with a type of polyisocyanurate (or PIR foam) are extensively used in various applications. This is because PIR foams are regarded as energy-efficient and cost-effective insulation products. The PIR foam is always used in heating, ventilation and air conditioning systems which are installed in roofing, vertical walls, warehouses, office buildings, health facilities and manufacturing facilities. Generally, the PIR foam is considered as a rigid thermal insulator with a typical value of thermal conductivity of approximately 0.023 W/mK.

• EffectsofPolyisocyanurateFoamProductionProcess

A process for manufacturing the PIR foam at the case study company starts by mixing the monomers of diphenylmethane diisocyanate, elastophor and dichlorofluoroethane with the polyol components. This mixture is coated between two

96 The Best Faculty in Production of e and i Engineers

major flexible surfaces and the three-layered laminate or facing material is further inserted between those two immovable surfaces. These processes result in the foam-forming mixture, which is later placed inside a laminator. In the laminator, there is a gap for foam expansion in which the foam mixture will expand until the gap is filled completely. The processing temperature is controlled at around 48.8°C to 65.5°C. Generally, the reaction of the mixture occurs for 10 seconds after the mixing process and is completed within 45 seconds, whereas the foam is expanded approximately 40 times compared to its initial dimensions. After that, the finished foam is taken out and kept at normal room temperature for 10 days before the cutting process. The finished product’s properties are tested according to the customer’s product specifications. Various properties are tested, for example, 1) the physical or structural properties (dimensional stability, compressive strength, water absorption, moisture vapor transmission, product density and impact test) 2) the thermal properties, and 3) the flammability properties. The testing methods used in the case study company follow the standards of the American Society for Testing and Materials. Cutting programming software is used to design the cutting processes that will provide the desired products and the least residue. Figure 1 shows the PIR foam before the cutting process in the desired shape from the customer order and Figure 2 shows the finished goods of the PIR foam after the cutting process.

Figure1: Examples of the PIR foam before the cutting process into blocks

Figure2: Examples of the finished PIR insulation after the cutting process

Since various types and sizes of PIR products are ordered by customers, the remaining residues left in the processes are substantial amount. The company has to concern the suitable method to handle these residues. The main reason is that the residues of the PIR foam can affect human health and pollute the environment. In this research, the data of the residues left over from the production process for only a single lot size is collected to investigate the percentage of the residue. The results are shown in Table 1. It can be seen that the percentage of the PIR foam residue that remains from the production process is larger than 50% per year (56.41 %) compared with the product weight. The

Design of Experiment for New Product Development from Polyisocyanurate Foam Residues

97Faculty of Engineering, Kasetsart University

company loses a large amount of money from these residues each year. The cost is mainly paid through the residue disposal. Generally, the company pays a PIR residue disposal cost of 480,000 baht per year (for approximately 33,000 kg per year of residues).

Table1: presents the residue data of the polyisocyanurate production process

PIR Total/Year

Foam volume 1,429.41 (m3)

Part volume 623.06 (m3)

Residue volume 806.35 (m3)

% Residue 56.41 %

Residue weight 33,454 (kg)

Only large residue sizes can be recycled, but most of the residues in the case study company are of small sizes in which a gathering process is required. The collection of small size residues has to be ground before the disposal process. The residue of PIR foam can lead to human health hazards because the excessive residues always remain in the workplace area. Many effects on workers occurred in the case study company; for example, irritants to human skin, eyes and the upper respiratory system during fabrication, in the short term. Corneal injuries to operators always occurred in the same manner. From the “Material Safety Data Sheet” described in http://www.owwco.com, the hazardous decomposition products from polyisocyanurate foam are composed of carbon dioxide, carbon monoxide, hydrogen cyanide, halogen acids and nitrogen oxides during combustion, which affect the eyes, skin, ingestion system and inhalation system. Since the company is preparing to develop the adoption of ISO14000 standards, the activities of environmental management are strongly concerned in the company. The ISO14000 standards are the environmental management systems (EMS) that provide requirements for an EMS and provide the general guidelines for EMS implementation. The standards focus on the identification and controlling of environmental impacts. The improvement and implementation of the environmental performances are further systematically achieved. In addition, the assurance of environmental management is provided through the organizational activities. From the benefits of ISO14000 standards, the company then aims to develop a new product from PIR foam residues, which is expected to reduce the disposal cost and increase company profits by selling a new product type.

In this research, the knowledge of the new product development from the PIR foam residue will be studied in which the environmental improvements are expected from the residue reduction in the work area. The new products that will be considered are the interlocking block and ventilation block. The characteristics of the interlocking block and ventilation block from the previous form are described as follows.

98 The Best Faculty in Production of e and i Engineers

• Interlockingblock

Generally, the interlocking block is produced from masonry cement (normally ASTM C91 Types) and used for construction purposes with structural systems. The conventional interlocking block is used in the construction of buildings and houses by stacking the blocks, supported by mortar to hold them in place. The mixture of the block production is composed of soil, sand, cement and crushed dust. The mixture is pressed in the interlocking block machine to finish the process.

• Ventilationblock

The ventilation block is widely used in construction activities. This type of block is produced by mixing cement and fine sand with water. The block is set up by using a block milling machine. This type of block makes the air flow through the block for cooling purposes.

These new products are expected by the case study company to reduce the PIR foam residue, to increase profits for the company, to add value to the residue and finally to improve the environmental impact of the company. There are several past studies describing mainly the experimental methods to improve the properties of the PIR foams. For example, the halogen-free flame retardants were investigated to reduce the rate of heat release and to improve the fire protection behavior in [1] and the physical or mechanical viewpoints of PIR foam were analyzed in [2]. However, rare papers are found mentioning the recycling methods and new product development from PIR foam residue. In [3], the recycling process of PIR foam using the heating process with the glycol mixtures was presented. In Thailand, a study of using the recycling process to produce the lightweight construction blocks has been presented by [4] in which the recycled steel was used. When the construction blocks are produced, various properties of these construction blocks are tested whereas the mechanical properties are always tested. Generally, the main tested properties comprise 1) compressive strength, 2) weight, 3) density, 4) moisture absorption and 5) failures. The interlocking mechanism always considers the deformed shapes under the load compression. In addition, the impact test is used as a failure test mechanism to examine the crushing or splitting of a part of the block [5-8]. In [9], the destruction-specific energy under pressure and vibration is tested to study the failure behavior. The tested properties of the ventilation blocks are mentioned in [10], in which the indoor temperature, air change rate and daylight contribution are tested.

Since the case study company needs to use the PIR foam residue to produce a lightweight construction block. The company can use the block to construct its workers’ accommodations. In addition, the company can sell these products in the future. Therefore, the objective of this research is to use the Design of Experiment method with

Design of Experiment for New Product Development from Polyisocyanurate Foam Residues

99Faculty of Engineering, Kasetsart University

Analysis of Variance (or ANOVA) to determine suitable combinations between the PIR foam residue and product components with various levels of parameters to produce the lightweight interlocking block and ventilation block.

The next section mentions the research methodologies of the ANOVA.

Research Methodologies

The Randomized Block Design or Two-Way ANOVA (Analysis of Variance) is used in this research, whereby the technique is considered as a basic statistical method to analyze the experimental data [11, 12]. The variance values of the treatment, block and error are calculated by sum square treatment, sum square group and sum square error to determine the sum square total. The assumptions are made as follows:

(i) H0 : τi = o; i (The treatment factor does not have any effect on the weight)

H0 : τi ≠ o; i (The treatment factor has an effect on the weight)

The test statistic is F-value in which the H0 is rejected when p-value < α or F-value > Fa,u (of the treatment factor), whereas u is degree of freedom and α is the significance level.

(ii) H0 : βi = o; i (The group factor does not have any effect on the weight)

H0 : βi ≠ o; i (The group factor has an effect on the weight)

The test statistic is F-value in which the H0 is rejected when p-value < α or F value > Fa,u (of the group factor), whereas u is degree of freedom and α is the significance level.

The normal probabilities, the effects of factors, errors and the interaction among factors will be tested. βTable 2 shows the ANOVA test statistics that are used in the research.

A

A

E

E

Table2: the ANOVA test statistics [11]

Source d.f. SS MS F

Treatment

Block

Error

a-1

b-1

M-a-b+1

SSA

SSB

SSE

MSA

MSB

MSE

Total N-1 SST

FA= MSA

MSE

FB= MSA

MSE

100 The Best Faculty in Production of e and i Engineers

Where,

SST = Sum Square Total =

SSA = Sum Square Treatment =

SSB = Sum Square Block =

SSE = Sum Square Error = SST-SSA-SSB

yijk = Experimental data of ith treatment, jth group and kth repetition

N = Number of data

a = Number of treatments

b = Number of blocks

r = Number of repetitions for each of ith treatment and jth group

T... = Total of data values

Ti.. = Total of data for ith treatment

T.j. = Total of data for jth group

In Section 3, the experimentation methods and results from ANOVA are presented.

Experimentation Methods and Results

There are 2 experiments in developing new products from PIR foam residue in this research with various levels of parameters, which are concluded as follows:

Experimentation1: Interlocking block production

The production process of interlocking block starts by mixing the soil, sand, crushed dust and cement. The ratios of the mixers are shown in Table 3. The PIR foam residue is then added into the mixer. The block machines and equipment that were used to produce the interlocking block in this research is a block milling machine and a hydraulics processing machine. Generally, the weight of the interlocking block is approximately 4.79 kilograms per block. In the experiment, the interlocking block components are mixed with the PIR residue using the percentages in Table 3. The response of the experiment is the weight of the finished block. A lightweight block is preferable in this research. The product weight should be lower than 4.79 kilograms.

!

!

!

Design of Experiment for New Product Development from Polyisocyanurate Foam Residues

101Faculty of Engineering, Kasetsart University

Table3: shows the level of the experimentations of the interlocking block

Ratioofsoil:sand:crusheddust:cement(Treatment:A)

PercentageofPIRfoamresiduevolume(Block:B)

1:9:2:1 30% and 50%

1:9:2:3 30% and 50%

1:9:2:5 30% and 50%

The assumptions of the ANOVA are made as follows:

1) For ratios of product components (the treatment, A);

H0: The ratio of product components does not have any effect on the interlocking block’s weight

H1: The ratios of product components have an effect on the interlocking block’s weight

2) For PIR foam residue percentage (the block, B)

H0: the PIR foam residue percentage does not have any effect on the interlocking block’s weight

H1: The PIR foam residue percentages have an effect on the interlocking block’s weight

3) For interaction between the ratios of product components and the PIR foam residue percentage

H0: the interaction between the ratios of product components and PIR foam residue percentage does not have any effect on the interlocking block’s weight

H1: The interactions between the ratios of product components and PIR foam residue percentage have an effect on the interlocking block’s weight

Experimentation2:Ventilation block production

The production process of ventilation block starts by mixing the cement and fine sand into the block machines and equipment that were used to produce the ventilation block. The percentage of PIR foam volume is then added into the mixer as shown in Table 4. The regular specifications of ventilation block weight are approximately 5.63 kilograms per block. Table 4 shows the experimentation factors and parameters of the ventilation block; in which the raw material components of the ventilation block are mixed with the PIR foam residue. The response of the experiment is the weight of the finished block. A lightweight block is expected from the experiment. The product weight should be lower than 5.63 kilograms.

102 The Best Faculty in Production of e and i Engineers

The assumptions of the ANOVA are made as follows:

1) For ratios of product components (the treatment, A)

H0: The ratio of product components does not have any effect on the ventilation block’s weight

H1: The ratios of product components have an effect on the ventilation block’s weight

2) For PIR foam residue percentage (the block, B)

H0: the PIR foam residue percentage does not have any effect on the ventilation block’s weight

H1: The PIR foam residue percentages have an effect on the ventilation block’s weight

3) For interaction between ratios of product components and PIR foam residue percentage

H0: the interaction between the ratios of product components and PIR foam residue percentage does not have any effect on the ventilation block’s weight

H1: The interactions between the ratios of product components and PIR foam residue percentage have any effect on the ventilation block’s weight

• Theexperimentationresults

The Analysis of Variance by error testings at α = 0.05 was conducted for experiments as shown here.

Experimentation1: Interlocking block

The production process of the interlocking block and the finished product of the interlocking block in this study are shown in Figure 3 and Figure 4.

Table4: shows the level of the experimentations of the ventilation block

Ratioofcement:finesand(Treatment:A)

PercentageofPIRfoamresiduevolume(Block:B)

1:14 30% and 50%

2:14 30% and 50%

3:14 30% and 50%

Design of Experiment for New Product Development from Polyisocyanurate Foam Residues

103Faculty of Engineering, Kasetsart University

Using the ANOVA with α=0.05 [14, 15], the normal probability plot of the residuals is constructed and is found to possess linear characteristics. The analysis of residuals and the fitted values is further conducted to test the independence of the data. Since the data variance is found to be constant, there is no significant sign of the patterns of the data. The results are shown in Figure 5.

Figure3: Production process of interlocking block

Figure4: Finished product of interlocking block

So the data are considered to be independent of each other. The raw data of the experiment is shown in Table 5. The results of the analysis of the variance of the response (the interlocking block’s weight) by 2 factors of ratios of the product components (the treatment, A) and the PIR foam residue percentage (the block, B) are shown in Table 6. Each experiment is conducted from 15 samples.

Figure5: the normal probability plot, the analysis of residuals and fitted values of experimentation 1

104 The Best Faculty in Production of e and i Engineers

From Table 6, the p-value of the interactions between the ratios of product components and the PIR foam residue percentage (0.598) is more than α (0.05). It can be concluded that the interaction does not have a significant effect on the product weight. The p-value of the ratios of product components (0.697) is also more than α (0.05). Thus; the product weight is not significantly affected by the ratios of product components. However, the p-value of PIR foam residue percentages is less than α (0.05). The result shows that the PIR foam residue percentages have an effect on the interlocking block’s weight. Thus; the average weight of the product is calculated in order to determine the suitable level of the PIR foam residue percentage that provides a lightweight product. The average weight of the interlocking block is shown in Table 7. Each experiment is averaged from 15 samples.

Table5: shows the level of the experimentations of the ventilation block

Ratioofsoil:sand:crusheddust:cement

PercentageofPIRfoamresiduevolume 30% 50%

1:9:2:1

3.74 4.01 3.89 3.43 3.26 3.08 3.41 3.13 3.61 3.98 4.02 3.66 2.79 3.17 2.95 3.21 3.57 3.97 3.87 3.86 3.19 3.25 3.04 3.17 3.85 3.98 4.02 3.27 3.15 3.06

1:9:2:3

3.55 3.74 4.01 3.76 3.16 2.93 3.28 3.39 4.16 4.31 3.36 3.33 2.97 3.34 3.11 3.06 3.89 3.95 3.82 3.76 2.98 3.15 3.13 3.11 3.98 3.95 4.13 3.42 3.18 3.29

1:9:2:5

4.33 4.24 3.76 3.28 3.26 3.12 3.44 2.94 3.33 3.58 3.62 3.86 3.66 2.72 3.19 2.93 4.12 3.95 4.32 3.74 3.05 3.46 3.12 3.17 3.88 3.96 3.92 3.74 3.24 3.28 3.11 3.17

Table6: shows the results of the analysis of the variance for the interlocking block

Source DF Seq SS Adj SS Adj MS F p_value

ratios of product components (treatment: A)

2 0.0759 0.0759 0.0380 0.36 0.697

PIR foam residue percentage (block: B)

1 4.9730 4.9730 4.9730 47.63 0.001

Interaction between A and B (A*B)

2 0.1085 0.1085 0.0543 0.52 0.598

Error 42 4.3855 4.3855 0.1044

Total 47 9.5429

Design of Experiment for New Product Development from Polyisocyanurate Foam Residues

105Faculty of Engineering, Kasetsart University

Table7: shows the experimentation results of average interlocking block weight

Ratioofsoil:sand:crusheddust:cement

Averageweightofinterlockingblock(kg.)

PIRfoamresiduepercentage

30% 50%

1:9:2:1 3.83 3.14

1:9:2:3 3.85 3.17

1:9:2:5 3.86 3.18

From Table 7, it is found that PIR foam residue at 50% provides a lower weight of interlocking block compared with 30% of the PIR foam residue. The recommendation for production is made of a ratio of soil: sand: crushed dust: cement of 1:9:2:1 with 50% of PIR foam residue, which is preferable because the cost of cement is lower. Normally, the weight of the interlocking block is approximately 4.79 kilograms per block. If the PIR foam is used in the block components, the weight of the block will be reduced, and will be lighter than a block selling in the market. The lightweight product is suitable for building small size accommodation, such as workers’ accommodation as per the requirement of the case study company. The PIR foam percentage around 30% of the total volume of the product components will provide approximately 3.8 kilograms of the product’s weight, whereas the PIR foam percentage of around 50% of the total volume will provide approximately 3.1-3.2 kilograms of the product’s weight.

Experimentation2: Ventilation block

The production process of the ventilation block and the finished product of the ventilation block in this study are shown in Figure 6 and Figure 7.

Figure6: Production process of the ventilation block

Figure7: Finished product of the ventilation block

106 The Best Faculty in Production of e and i Engineers

The product weight from the experiment is shown in Table 8.The results of the analysis of the variance of the response (the ventilation block’s weight) by 2 factors of the ratios of product components (the treatment, A) and the PIR foam residue percentage (the block, B) are shown in Table 9. Each experiment is conducted from 15 samples. The response is the weight of the ventilation block. A lightweight product is required from the experiment.

Figure8: the normal probability plot, the analysis of residuals and fitted values of experimentation 2

Using the ANOVA with α=0.05, the normality and data independence testing are conducted in a similar manner in the interlocking block experimentation. The normal probability plot, the analysis of residuals and fitted values of experimentation 2 is shown in Figure 8.

Table8: shows the product weight by ratios of the ventilation block mixer

Ratioofsoil:sand:crusheddust:cement

PercentageofPIRfoamresiduevolume 30% 50%

1:14

4.98 5.21 5.2 5.16 4.44 4.32 4.48 5.06 4.57 5.02 4.58 5.25 4.51 4.48 4.5 4.38 5.26 5.13 5.07 4.98 4.48 4.59 4.55 4.46 5.18 5.21 5.11 4.47 4.59 4.55

2:14

5.38 4.89 5.35 5.03 5.41 4.22 4.77 4.51 5.4 4.74 5.17 4.91 4.38 4.55 5.02 4.62 5.19 5.16 5.12 5.23 4.48 4.44 4.61 4.53 5.16 5.25 5.08 4.56 4.38 4.71

3:14

5.6 5.85 5.22 5.53 4.71 4.43 4.75 5.13 5.38 4.79 4.47 4.7 4.37 4.64 5.05 4.26 5.05 5.14 5.24 5.19 4.82 4.63 4.61 4.66 5.12 5.22 5.26 4.59 4.74 4.77

Design of Experiment for New Product Development from Polyisocyanurate Foam Residues

107Faculty of Engineering, Kasetsart University

Table9: shows the results of the analysis of the variance for the ventilation block

Source DF Seq SS Adj SS Adj MS F p_value

ratios of product components (treatment: A)

2 0.2609 0.2609 0.1304 1.20 0.312

PIR foam residue percentage (block: B)

1 1.2160 1.2160 1.2160 11.18 0.002

Interaction between A and B (A*B)

2 0.0028 0.0028 0.0014 0.01 0.987

Error 42 4.5680 4.5680 0.1088

Total 47 6.0478

From Table 9, the p-value of the ratio of product components (0.312) is more than α. The result shows that ratio of product component does not have a significant effect on the product weight. If the p-value of the interaction between the ratio of product components and the PIR foam percentage is 0.987, which is larger than α, the result shows that the interaction also does not have significant effect on the product weight. When the p-value of the PIR foam percentage (0.002) is less than α, it can be concluded that the PIR foam percentage has a significant effect on the ventilation block weight. The experimentation is further conducted to determine the weight of the ventilation block by focusing mainly on the PIR foam percentage. A lower weight of the block is expected from the experiment. Table 10 shows the weight results from 15 samples of the experiments.

Table10 : shows the experimentation results of ventilation block weight

Ratioofsoil:sand:crusheddust:cement

Averageweightofinterlockingblock(kg.)

PIRfoamresiduepercentage

30% 50%

1:14 5.12 4.52

2:14 5.14 4.60

3"14 5.18 4.68

From Table 10, it is found that the PIR foam residue at 30% provides approximately 5.1-5.2 kilograms of the product, whereas the PIR foam residue at 50% provides approximately 4.5-4.7 kilograms of the product. The ventilation block that is mixed with the PIR foam has a lower weight compared with the block selling in the market (5.63 kilograms).

108 The Best Faculty in Production of e and i Engineers

Conclusion and discussion

In this research, the new product developments from PIR foam residue are studied. These are an interlocking block and ventilation block for construction purposes. The experiments in the research consider the set of factors and parameters that have an impact on the response, which is the product weight. The treatment considered in the research is the product’s raw material components by ratios and the block is the PIR foam percentage that is mixed with the product’s raw material components. The conclusion can be drawn that the PIR foam can provide a lightweight product as required by the case study company. However, other product properties, for example, the compressive strength test, impact test or the fire protection test, have to be further analyzed. These are not included in the scope of this paper.

References

[1] M. Modesti and A. Lorenzehi., 2001, “An Experimental Method for Evaluating Isocyanate Conversion and Trimer Formation in Polyisocyanate-Polyurethane Foams”, European Polymer Journal, Vol. 37, pp. 949-954.

[2] M. Modesti., A. Lorenzetti., F. Simioni and M. Checchin., 2001, “Influence of Different Flame Retardants on Fire Behaviour of Modified PIR/PUR Polymers”, Polymer Degradation and Stability, Vol. 74, pp. 475-479.

[3] H. Ulrich., A. Odinak., B. Tucker and A.A.R. Sayigh., 1978 “Recycling of Polyurethane and Polyisocyanurate Foam”, Polymer Engineering & Science, Vol.18, No.11, pp. 844-848.

[4] C. Chaijuk., S. Songpirijakij and K. Wattanakul., 2003, “The Lightweight Concrete Block Production from Steeled Recycle Process”, the 1st National Concrete Proceedings, pp. 125-131, May 2003.

[5] W.A. Thanoon., M.S. Jaafar., M.R.A. Kadir., A.A.A. Ali., D.N. Trikha and A.M.S. Najm., 2004, “Development of an Innovative Interlocking Load-Bearing Hollow Block System in Malaysia”, Construction and Building Materials, Vol. 18, pp. 445-454.

[6] M.S. Jaafar., W.A. Thanoon., A.M.S. Najm., M.R. Abdulkadir and A.A.A. Ali., 2006, “Strength Correlation between Individual Block, Prism and Basic Wall Panels for Load Bearing Interlocking Mortarless Hollow Block Masonry”, Construction and Building Materials, Vol. 20, pp. 492-498.

Design of Experiment for New Product Development from Polyisocyanurate Foam Residues

109Faculty of Engineering, Kasetsart University

[7] J. Brožovský., O. Matějka and P. Martinec., “Concrete Interlocking Paving Blocks Compression Strength Determination using Non-Destructive Methods”, The 8th International Conference of the Slovenian Society for Non-Destructive Testing, September 1-3, Slovenia, pp. 91-97.

[8] Ministry of Industry Thailand, TISI-58-2530, Industrial Standards of Hollow-Non- Load-Bearing Concrete Masonry Unit.

[9] U. Atici and A. Ersoy., 2008, “Evaluation of Destruction Specific Energy of Fly Ash and Slag Admixed Concrete Interlocking Paving Blocks (CIPB)”, Construction and Building Materials, Vol. 22, pp. 1507-1514.

[10] J. Khedari., M. Rungsiyopas., R. Sarachitti and J. Hirunlabh., 2004, “A New Type of Vented Concrete Block for Zero Cooling Energy”, Building and Environment, Vol. 39,pp. 1193 - 1197.

[11] D.C.Montgomery, Design and Analysis of Experiments, 7th ed., John Wiley & Sons, 2009.

[12] A.M. Dean and D.Voss, Design and Analysis of Experiments, Springer Publishing, 1999.