MTS Adhesives Project 5
Measurements For Optimizing Adhesives Processing
Task 2 Optimisation of Key Process Parameters
Report 8
Case Study: Steel Plate Bonding
S Hurley P.A. J Gibbs
Taywood Engineering
This report forms part of the deliverable for Task 2
Reports for Task 2
Report 2; Optimisation of Key Process Parameters - Summary Report Report 3; Case Study: Precision Mechanical Assembly in the Business Machines Industry Report 4; Case Study: Packaging Applications 1 &2: Summary Report Report 5; Case Study Packaging Application 1: Full Report Report 6; Case Study: Packaging Application 2: Full Report Report 7; Case Study: Access Flooring Application Report 8; Case Study: Construction Application - Steel Plate Bonding Report 9; Optimisation of Key Process Parameters - Draft Code of Best Practice
1
Technical Report
Title
& Date of Issue
Report No. 1303S/96/9318
Page 1 of 45
Measurements For Optimizing Adhesive
Processing
A Statistical Approach for Determining the
Factors Affecting Bond Strength
Dr A Olusanya, National Physical Laboratory (2 copies) S A Hurley Author (2 copies)
Abstract This report contributes to Project 5 of the programme on adhesives supported by the DTI through the
Measurement Technology and Standards (MTS) budget. This project is led by the National Physical Laboratory.
This report presents experimental data and a Taguchi approach of experimental design to determine important processing variables likely to influence the development of bond strength in construction related steel plate bonding applications.
This report is copyright. Reproduction of the whole or any part thereof must not be made without the express permission of Taywood Engineering Limited. This report and the results shown and any recommendations or advice made herein are based upon the information, drawings, samples and tests referred to in the report Taywood Engineering Limited accepts no liability for any damages, charges, costs (including, but not limited to, legal rests) or expanses in respect of or in relation to any damage to any property or other loss (save for death or personal injury occasioned by reason of any negligence on the part of Taywood Engineering Limited) whatsoever arising either directly or indirectly from the use of the report, the carrying out of any recommendations contained
Taywood Engineering Ltd Technology Division
—
Taywood House. 345 Ruislip Road . Southall . Middlesex. UB1 2QX Issue 4 / 4482 / December 1994
Taywood Engineering Limited Report No. 1303S/96/93 18
Page 2 of 45
CONTENTS
SUMMARY
1.
2.
3.
4.
5.
6.
7.
INTRODUCTION
OBJECTIVES
METHODOLOGY
3.1 Application
3.2 Site Study
3.3 Bonding Process
3.4 Process Variables
EXPERIMENTAL
4.1 Bond Quality Assessment
4.2 Adhesive
4.3 Process Variables
4.3.1 Steel Substrate
4.3.2 Concrete Substrate
4.4 Experiment Design
RESULTS
5.1 Steel Substrate
5.2 Concrete Substrate
CONFIRMATORY TRIAL
CONCLUSIONS
ACKNOWLEDGEMENTS
Figures 1-2
Tables 1-6
Plates 1-5
Appendix 1
Appendix 2
Taywood Engineering Limited Report No. 1303S/96/93 18
Page 3 of 45
SUMMARY
The work reported here forms part of a project concerned with the optimisation of adhesive
processing led by the National Physical Laboratory and, thereby, contributes to a wider ranging
research programme on adhesives funded by the DTI.
The main objective of this work was to assess the benefits of applying Taguchi statistical
methods to an evaluation of the key processing variables associated with a construction based
application of adhesives. The bonding of steel plate reinforcement to concrete and cast iron
structures with a two-part cold curing epoxy resin was selected as a representative application.
Following a site visit ten variables, each of which could influence bond strength were
identified for investigation. These included adhesive grade (fast/slow curing) and aspects of
substrate surface condition, adhesive application and cure. In conjunction with Xyratex
Limited, Taguchi methods were then used to define an array of 16 experiments which would
allow the significance of each variable to be evaluated at two levels for a given substrate (the
equivalent array given by a full factorial approach would require 1024 experiments).
Laboratory-based tensile pull-off tests were used, with a replication level of 5, to determine the
effects of the 16 selected conditions on bond strength. Both concrete and steel substrates were
investigated. The results of these tests were analysed with Xyratex Limited using Taguchi
statistical procedures.
Due to the scatter in results and the limitation on failure stress resulting from the inherently
low tensile strength of concrete, it was not possible to rank the variables for this substrate.
However, for steel, bond strength was found to be significantly affected by 5 variables;
additionally, 3 significant interactions were revealed, each of which involved the adhesive
grade.
The Taguchi analysis also predicted
the optimum processing conditions.
satisfactorily with this prediction.
the bond strength which should be achieved on steel under I
A further set of 10 pull-off tests gave results which agreed
It was concluded that, for this type of application, valuable information can be obtained from
minimal testing using the Taguchi approach to the design of experiments.
Taywood Engineering Limited Report No. 1303S/96/9318
Page 4 of 45
1. INTRODUCTION
This report contributes to Project 5 of the 1993-95 programme on adhesives supported by the
DTI through the Measurement Technology and Standards (MTS) budget. This project is led
by the National Physical Laboratory (Division of Materials Metrology) and is concerned with
measurements for optimizing adhesive processing.
The rationale for this project is based on the view, supported by an earlier and wide-ranging
survey, that the confident use of adhesives in many industrial sectors may be increased and
made more profitable by addressing two particular areas (amongst others) viz:
(i) the validation of quantitative methods which can be used in the manufacturing
environment to measure and rank properties which are associated with processing, for
example: viscosity, slump, tack, open time and cure.
(ii) the development of a methodology which can be used to define and evaluate the key
aspects of a process, thus enabling methods of control to be implemented effectively.
The second area is important because many seemingly simple industrial processes which utilise
adhesives do, in fact, depend upon complex and possibly synergistic interactions.
Consequently, reliable and relatively rapid methods of quantitatively distinguishing the
influential parameters could lead to greater cost-effectiveness.
As sub-contractors to NPL, part of Taywood Engineering’s role in Project 5 was to address
this second area in the context of the construction industry (Project Tasks (T)3c.5-(T)3c.6).
More specifically, TEL were requested to investigate the processing of typical adhesives used
in civil engineering applications. This work was carried out in conjunction with Xyratex Ltd
(formerly IBM, Havant) where a Design of Experiments philosophy that incorporates
‘Taguchi’ experimental methods has been widely used for quality improvements and process
optimisation of adhesive bonding operations. I
The well-established technique of bonding steel plate reinforcement to concrete and cast iron
structures was used as a model application for this work. This process relies upon the
structural performance of an adhesive which is commonly applied in external conditions. In the
present study, therefore, it may be considered also to represent other applications which place
similar demands upon adhesive use.
Taywood Engineering Limited Report No. 1303S/96/9318
Page 5 of 45
2. OBJECTIVES
A large number of process variables and their possible synergistic interactions can affect the quality of
adhesive bonding in civil engineering applications. consequently, through an understanding of the
important variables and their relative interactions, methodologies could be developed to improve the
quality of adhesive bonding performed on-site. the aim of the work described in this report was to
assess taguchi methodologies of experimental design for dete rmining important processing variables
affecting the quality of adhesive joints
structures for experimentation, small
processing variables.
used in civil engineering. due to the obvious unavailability of real
scale laboratory tests were utilised to determine the important
3. METHODOLOGY
3.1 APPLICATION
The bonding of external steel plate reinforcement to concrete and cast iron structures was
considered to be a suitable civil engineering application for investigation with the Taguchi
methodology of experimental design. Due to the obvious inability to experiment with real
strengthening operations, small scale laboratory specimens were used to investigate a number
of variables likely to affect bond quality. To aid identification of these processing variables, a
site visit was carried out prior to the Taguchi experimental design.
3.2 SITE STUDY
The bonding of steel plate reinforcement to the soffits of three canal bridges running under
Grosvenor Road, Embankment, London, was chosen as a suitably complex operation for
determining g the variables likely to influence bond quality. The site visit was conducted with
representatives from Sika Ltd (adhesive manufacturer), Taywood Engineering, NPL and I
Xyratex, thus enabling the important processing variables to be discussed in detail.
The Grosvenor Road operation was particularly complex due to the tidal rise and fall of the
River Thames, (Plates 1 and 2). Bonding works were only possible during low tide when the
soffit became accessible. At high tide, the scaffolding platform became covered, leaving a gap
between the bridge soffit and water level of approximately lm. Due to this low clearance,
water would occasionally splash onto prepared surfaces (Plate 3). This was likely to effect
bond quality significantly and was to form an important process variable in the experimental
work.
Taywood Engineering Limited Report No. 1303S/96/9318
Page 6 of 45
3.3 BONDING PROCESS
The bonding process, as summarised in Figure 1, involved the following three important
stages:
. Preparation of the steel plate reinforcement.
. Preparation of the bridge soffit.
. Mixing and application of the adhesive.
To reduce on-site preparation problems, the steel plates were pre-treated prior to delivery.
The plates were pre-treated on all faces in the following manner:
● degreased
. dry blasted
. vacuum cleaned
● primed with a micaceous iron oxide based paint
. packed with a desiccant.
On site, the bridge soffit was treated in the following manner:
. degreased
. dry blasted
. vacuum cleaned
Additionally, where excessive roughness and pitting of the soffit existed, a thin coat of the
adhesive was applied to smooth the bond surface.
The two part epoxy adhesive, Sikadur 31 PBA (plate bonding adhesive) was supplied to site in
premeasured packs. To minimise mixing problems, the base resin component was supplied in
the colour white, with the amine hardener supplied in the colour black. These colours
provided a simple visual indication of any mixing inconsistencies.
Mixing of the Sikadur 31PBA was achieved by scrapping the contents of the premeasured
hardener pack into the preweighed resin pack and mixing for approximately 2-3 minutes with a
drill. The mixing process was continued until a uniform colour was achieved, indicating
adequate mixing. Following mixing, the adhesive was allowed to stand for 10 minutes whilst
the steel plate reinforcement was unpacked and given a final decreasing. The adhesive was
trowel applied to the soffit of the bridge deck to give a uniform layer lmm in thickness (Plate
4), and to the plates with a profiled thickness ranging from 3mm at the centre to lmm at the
edges.
Taywood Engineering Limited Report No. 1303S/96/93 18
Page 7 of 45
Bonding was achieved by jacking the plates upwards and locating predrilled holes in the steel
plate reinforcement on anti-peel bolts fixed to the bridge soffit. The bonded joint was
completed by further jacking and tightening of the anti-peel bolts (which held spacer washers
to achieve a consistent 2mm bond-line thickness).
Following cure of the adhesive, the bond was checked for voids by hammer tapping. Any
voids were filled by drilling through the steel plate and injecting additional adhesive.
Finally, the bonded plates and edges of the adhesive joint were treated with a high build paint system to provide long-term environmental protection (Plate 5).
3.4 PROCESS VARIABLES
Discussions were held immediately after the site visit to determine the process variables likely
to affect bond strength. As a result, the following important variables were identified for
investigation:
●
●
●
●
●
●
●
●
●
●
Adhesive type.
Mixing ratio of the resin and hardener.
Workability and compaction of the adhesive.
Moisture contamination of the substrates prior to bonding.
Effectiveness of solvent cleaning.
Effectiveness of vacuum cleaning.
Humidity during mixing and adhesive application. ,
Open time of the joint before closure.
Cure temperature.
Cure time.
Taywood Engineering Limited Report No. 1303S/96/9318
Page 8 of 45
4. EXPERIMENTAL
4.1 BOND QUALITY ASSESSMENT
Tensile pull-off tests were conducted in the laboratory using 20mm diameter dollies to
investigate the effects of the identified variables on bond strength. Two separate experiments
were devised to examine adhesion on concrete and mild steel substrates.
The pull-off test was considered to be a suitable technique for examining bond quality by
measuring variations in bond strength. The particular advantage of this test over other
potential methods is that it gives a relatively simple procedure that minimises the inclusion of
additional variables during joint manufacture. A standard portable pull-off tester was used in
the Taguchi test programme. This equipment (the ‘Limpet’), enables a maximum tensile load
of 10kN to be applied. Due to this limit, a 20mm diameter dolly was used for the tests. This
enables testing to be performed with an adhesive achieving an ultimate tensile/bond strength in
excess of 30 MPa on a steel substrate.
Both steel and concrete substrates were used to simulate the bridge soffit. A steel thickness of
10mm was chosen to minimise deflection during testing. In addition, the steel substrate was
grit blasted to an SA2½ finish (as specified in I SO 8501-01) controlled to give a peak to
trough surface roughness of 50- 10Opm. Concrete substrates were prepared in accordance with
the draft CEN standard EN 104-801-1 to the mix grade MC (0,40). The draft standard was
followed with the exception of clause 5.1 requiring aggregates with a water absorption of less
than 2%. The aggregate used was Thames Valley 5- 10mm with a water absorption of 3 .5°A
which was considered to have a negligible effect on the quality of the concrete. This clause
was not followed due to the difficulty in sourcing UK aggregates which comply.
4.2 ADHESIVE
Sikadur 31 Plate Bonding Adhesive (PBA), manufactured by SIKA Ltd., was chosen as a
suitable material for the Taguchi task. This material is produced in two grades, ‘Normal’ and
‘Rapid’, both of which are filled cold-curing two-part epoxy resin adhesives. Both products
have similar properties, with the rapid grade formulated for lower temperature applications
(down to a minimum of 5C).
Sikadur 3lPBA was specifically formulated for the bonding of external plate reinforcement for
the structural strengthening/stiffening of concrete and steel structures. This adhesive system
has a proven history within the construction industry, as does the methodology adopted for the
bonding process.
Taywood Engineering Limited Report No. 1303S/96/93 18
Page 9 of 45
4.3
4.3.1
Since the early 1980s, extensive testing has been carried out on Sikadur 3lPBA and some
typical properties for the fully cured ‘Normal Grade’ tested at 23°C are shown below:
Density (g/cm3) 1.5
Young’s Modulus (GPa) 7
Tensile Strength (MPa) 22
Compressive Strength (MPa) 70
Fracture Energy (kJm-2) 0.4
Poisson’s Ratio 0.3
Strain to Failure (%) 0.7
PROCESS VARIABLES
Steel Substrate
The following ten process variables were investigated at the two settings shown below:
Variables Setting 1 Setting 2
Adhesive Type Normal Grade Low Temperature Grade
Adhesive Application Levelled over Surface Applied as Blob
Resin to Hardener Mix Ratio Specified Ratio Minus 10% of Hardener
Mix Humidity 10% Humidity 70% Humidity
Open Time to Joint Closure Minimum Possible Recommended Maximum +1O%
Cure Temperature 23°C Recommended Minimum -5°C
Decreasing of Steel No Solvent Clean Solvent Clean
Vacuum Cleaning of Steel No Vacuum Clean Vacuum Cleaned
Cure Time to Test 1 day 2 days
Steel Surface Condition Wet Dry
Note: Actual test conditions combine elements of setting 1 and 2.
Taywood Engineering Limited Report No. 1303S196/9318
Page 10 of 45
4.3.2
4.4
Concrete Substrate
The concrete substrate experiment was designed to be as close to that with the steel substrate
as possible, hence a 20mm diameter dolly was used on both substrates; for concrete testing, a
50mm dolly is preferable and commonly but not universally used. The identical variables were
used except that the concrete surface condition varied from dry to damp.
EXPERIMENT DESIGN
A full factorial experiment covering ten variables set at two levels would require 1024 separate
runs (i.e. 210). This type of design is therefore impractical on a cost and time basis when
considering the even larger number of variables that may need to be addressed in some
construction related applications. Using the Taguchi approach to experimental design, it was
possible to reduce the total number of experiments to 16, as shown in Figure 2. This design of
experiment approach enabled each of the main variables to be assessed. Experimental arrays
for the Taguchi analysis were determined by Xyratex. Two separate arrays for the steel and
concrete substrates were developed and are shown in Tables 1 and 2 respectively.
However, due to the large number of variables, each main factor had a possible alias, or
interaction of other variables that could shadow its effect (Table 3). A larger number of
experiments could have been carried out to avoid the possibility of first order interactions;
however, it was considered impractical to increase the number of experiments beyond 16. A
possible alternative to increasing the number of experiments would have been to reduce the
total number of variables to below 8. However, this approach was not favoured due to the
importance of each chosen variable and its potential for having a large effect on bond strength.
It was therefore decided to examine the ten variables, with a 16 experiment array, and use
engineering judgement to determine whether further experimentation should be carried out to
deconvolute problems with aliases.
For each of the 16 experiments, five repeat runs were carried out. Five runs were considered
to be sufficient to detect changes of less than 10%, assuming a variability in runs of
approximately ± 15%.
Taywood Engineering Limited Report No. 1303S/96/9318
Page 11 of 45
5. RESULTS
5.1 STEEL SUBSTRATE
Results for the steel pull-off tests are shown in Table 4. This data was generally consistent,
with scatter between tests averaging around 15%. The box whisker plot (Figure 2) shows
variability in a clearer graphical form described with the following key: small circle
representing the mean whiskers representing maximum and minimum scatter; top horizontal
line of the box representing the 75th percentile; bottom horizontal line representing the 25th
percentile; and the middle line representing the 50th percentile.
Details of the Taguchi analysis for the steel substrate are given in Appendix 2. The Taguchi
software was able to reasonably explain 68% of all variability in results with the calculated
factors, as indicated by the R-squared value of 0.68 shown in the table of coefficients.
Analysis of the coefficients revealed that the most significant variables influencing bond quality
are:
● Adhesive type.
● Humidity during mixing.
. Time to test after joint closure (i.e., cure time).
. Vacuum cleaning the adherends prior to bonding.
Graphical representations of the effects of each variable on bond strength are shown in the
variable response plot given in Appendix 2. This plot represents the response of the mean
bond strength for each variable and its setting. For example, for variable A (adhesive type)
point Al represents the low temperature adhesive and A2 represents the normal grade; as
shown, the low temperature grade results in a greater bond strength for the experimental
conditions investigated. The upper and lower horizontal lines of the graph represent
boundaries within which there are no significant differences. Significant factors that result in
an increase in bond strength were found to be:
. The low temperature grade provides a higher bond strength than the normal adhesive.
● Storing and mixing the adhesive at a high humidity results in a lower bond strength.
. Solvent cleaning the steel prior to bonding results in a lower bond strength.
. Vacuum cleaning the steel adherends prior to bonding results in a higher bond strength.
● A cure time of 1 day resulted in a significantly higher bond strength than for a 2 day cure
period.
Taywood Engineering Limited Report No. 1303S/96/93 18
Page 12 of 45
As previously discussed, each imposed variable has a number of possible aliases which could
represent the main variable. A table of coefficients representing each of the possible first order
interactions and aliases is given in Appendix 2. It was decided that first order interactions
representing each of the main process variables were unlikely and further experimentation to
confirm this assumption could not be justified.
From the table of coefficients for aliases and first order interactions, three significant
interactions were identified:
● Adhesive type and humidity.
● Adhesive type and cure temperature.
. Adhesive type and joint open time.
AU significant interactions involved adhesive type. Furthermore, the interacting variables are
known to affect the performance of two-part cold-curing epoxy adhesives. Consequently, the
significant first order interactions were not unexpected.
5.2 CONCRETE SUBSTRATE
Results for the concrete pull-off tests are shown in Table 5. A large scatter was observed
which was attributed to the variability of the concrete substrate, the unsuitable ratio of
aggregate to dolly diameter, and the decision not to core the concrete surrounding the dolly
prior to testing.
The tensile pull-off strength of the standard MC (0,40) concrete ranged from 3.1 to 7.7 MPa
with an average strength of 5.3 MPa. The draft CEN standard states that an average pull-off
strength of 3.5 MPa should be achieved, with a minimum of 3.0 MPa, when tests are carried
out in accordance with the draft CEN Standard EN 104-846. Test data obtained here showed
a higher strength than that described in the draft standard, but this may be attributed to
differences in the test procedure. Pull-off testing for the Taguchi task was deliberately carried
out without coring the concrete surrounding the test dolly. Without coring, higher strength is
expected due to the different failure mode initiated in the concrete. Coring was omitted since
the higher failure loads would be more likely to show processing problems due to variations in
the surface preparation.
Taywood Engineering Limited Report No. 1303S/96/9318
Page 13 of 45
Results for the Taguchi analysis on the concrete substrate are given in Appendix 1. The
analysis failed to show significant effects for any of the experimental variables due to the large
scatter in results. This is shown in the variable response plot where significant effects fall
inside the upper and lower scatter limits marked on the graph. Additionally, the table of
coefficients for the Taguchi analysis, (Page 31), shows an R-squared value of only 0.32,
indicating that only 32°/0 of all variability is explained by the calculated analysis factors.
Due to the scatter in results, the Taguchi experiment was unable to detect significant effects
likely to influence bond quality for a concrete substrate. The Taguchi methodology was
therefore considered to be unsuitable for the concrete substrate test conditions investigated.
.- Taywood Engineering Limited Report No. 1303S/96/9318
Page 14 of 45
6. CONFIRMATORY TRIAL
The Taguchi analysis identified the significant process variables to be:
. Adhesive type.
. Vacuum cleaning of steel substrate.
. Solvent cleaning of steel substrate.
. Relative humidity.
. Cure time.
The optimum process conditions for a maximum bond strength were identified as:
. Low temperature grade adhesive.
. 10% relative humidity.
. Vacuum cleaning of steel substrate.
. 24h cure time.
. No solvent cleaning of the steel.
Ten test specimens were prepared using the above conditions and the following prediction of
the results was calculated with the Taguchi software: a bond strength of 23.49 MPa ±
1.69 MPa to a confidence level of 90% assuming a run of 7 test samples.
Results from the confirmatory trial are shown in Table 6. Sample numbers 7, 10 and 11 were
omitted from the calculation of mean bond strength due to large defects at the edge of the
obtained.
The Taguchi calculation of bond strength agreed with the experimental data although the latter
was at the lower limit of the predicted value. This might be explained by a slight change in the
material used in the confirmatory trial which was supplied as Sikadur 31 and not Sikadur 31
PBA - although identical properties would be expected (a’ slight difference in coloration was
noted between the materials). However, despite this difference, the predicted result agreed
closely with that given by the confirmatory trial.
Taywood Engineering Limited Report No. 1303S/96/9318
Page 15 of 45
7 CONCLUSIONS
The Taguchi analysis for the concrete substrate was inconclusive. Problems arose due to
scatter in the test data which maybe attributed to the following factors:
. Variability y of the concrete substrate.
. Unsuitable aggregate to dolly diameter ratio.
. Omitting coring of the concrete at the base of the test dolly prior to carrying out the pull-off
test.
A larger diameter test dolly combined with coring of the concrete would possibly have reduced
scatter in the pull-off test data. However, it is unlikely that additional information would have
been obtained from the experimental variables investigated due to the inherently low tensile
strength of the concrete substrate. All of the test conditions investigated resulted
predominantly in concrete cohesive failure and, therefore, in similar tensile strengths being
obtained. It was therefore concluded that, as changes in surface treatment did not give bond
strengths below the tensile strength of the concrete, little advantage would have been obtained
with a more refined test procedure to reduce scatter. The inherently low tensile strength of
concrete presents a general problem in adhesion testing as, for other than very low bond
strengths, substrate cohesive failure commonly dominates. However, until bond testing is
performed, it is usually not possible to predict confidently that this will occur.
During the manufacture of specimens for the Taguchi experiment, it was found difficult to
produce the standard concrete specified in the draft CEN standard EN 104-801-1. Difficulties
arose in the sourcing of UK aggregates able to comply with this standard. Since the present
work commenced, this problem has been communicated to and addressed by the appropriate
BSI committee.
The Taguchi analysis on the steel substrate successfully indicated important factors likely to
influence bond strength when using Sikadur31 PBA.
Furthermore, the analysis also showed that certain factors, expected to influence bond strength
significantly, had little effect. These factors were:
●
●
●
Dampness of the adherends prior to bonding (note: in the laboratory experiment water was
added shortly before closing the joint).
Small differences in the mix ratio of hardener to resin
Open time before joint closure
Taywood Engineering Limited Report No. 1303S/96/9318
Page 16 of 45
It must be emphasised that the significance of variables found in this
particular product; other epoxy based adhesives may perform quite
qualification does not detract from the value of the work as the main intent
Taguchi methodology.
work applies to a
differently. This
was to examine the
A surprising result from the Taguchi experiment was that the shorter cure time resulted in a
higher bond strength. This suggests a possible change from ductile to brittle failure with a
higher level of cross-link density. Such a change could lead to failures being initiated at
defects, for example entrapped air bubbles in the adhesive bond-line.
Vacuum cleaning the steel substrate prior to bonding resulted in an increase in bond strength.
However, further surface preparation using a decreasing solvent resulted in a lowering of bond
strength.
spreading
This result was unexpected and may indicate solvent retention at the surface or the
(rather than removal) of small traces of contamination.
Taywood Engineering Limited Report No. 1303S/96/9318
Page 17 of 45
ACKNOWLEDGEMENTS
The work in this report was funded by the Department of Trade and Industry as part of the
Measurements Technology and Standards (MTS) Adhesives Project 5, ‘Measurements for
Optimizing Adhesive Processing’. This work was sub contracted to Taywood Engineering by
the National Physical Laboratory (NPL), the managers for Project 5.
Taywood Engineering Limited Report No. 1303S/96/9318
Page 18 of 45
Figure 1 Flow Diagram of the Steel Plate Bonding Process
I Fix anti-peel bolts and spacers
Join and tighten bolts
Cure
Remove loading jacks
Apply surface coatings
.
Taywood Engineering Limited Report No. 1303S/9619318
Page 19 of 45 .
Figure 2 Box Whisker Plots for the Concrete and Steel Substrate Taguchi Analysis Data
TAYWOOD ENGINEERING
●
CONCRETE
RUN NUMBER
.
Taywood Engineering Limited Report No. 1303S/96/9318
Page 20 of 45
*,
RUN
6 7
BLOB AREA BLOB BLOB
Taywood Engineering Limited Report No. 1303S/96/9318
Page 21 of 45
-.. .
Table 2
RUN
STAN STAN STAN STAN STAN STAN STAN
Experimental Array for the Concrete Substrate.
Taywood Engineering Limited Report No. 1303S/96/9318
Page 22 of 45
A B`
Taywood Engineering Limited Report No. 1303S/96/9318
Page 23 of 45
Table 4 Taguchi Analysis Pull-Off Test Results for Steel.
Taywood Engineering Limited Report No. 1303S/96/9318
Page 24 of 45
Table 5 Taguchi Analysis Pull-Off Test Results for Concrete.
Taywood Engineering Limited Report No. 1303S/96/9318
Page 25 of 45
Table 6 Pull-Off Test Results for the Confirmatory Trial on the Steel Substrate.
Dolly Number Force to Failure
(kN)
1 6.65
2 7.16
3 6.76
4 6.04
5 7.44
6 6.53
7 4.76
8 6.70
9 7.48
10 4.99
11 4.83
Failure Stress
(MPa)
21.17
22.79
21.52
19.23
23.68
20.79
15.15
21.33
23.81
15.88
15.37
Failure Mode Comments
COH
COH
COH
COH
COH
COH
ADH Poor sample
ADH
COH
ADH Poor sample
ADH Poor sample
Tests 7, 10, 11 are ignored due to poor test samples containing large defects.
Mean results for the confirmatory trial (neglecting points 7, 10, and 11) are:
Tay wood Engineering Limited Report No 1303S
Page
:/96/9318
28 of 45
Plate 5: Completed Joint coated with a High Build PVC/Acrylic Paint System to Provide Long Term Durability (photograph courtesy of Sika Ltd)
Taywood Engineering Limited Report No. 1303S/96/93 18
Page 29 of 45
Appendix 1
Taguchi Results for the Concrete Substrate
Taywood Engineering Limited Report No. 1303S/96/93 18 .
Page 30 of 45
ANALYSIS OF FACTORIAL
------
Taywood Engineering Limited Report No. 1303S/96/9318
Page 31 of 45
I
TABLE OF COEFFICIENTS 1
0.11485 0.066673 0.091125
Total ERROR was used for F tests.
Taywood Engineering Limited Report No. 1303S/96/9318
Page 32 of 45
95
90
75
25
10
5
ANALYSIS OF FACTORIAL EXPERIMENT: TAYWOODC
.
..’ .
Taywood Engineering Limited Report No. 1303 S/96193 18
Page 34 of 45 ‘.
TABLE OF COEFFICIENTS
.
Taywood Engineering Limited Report No. 1303S/96/9318
Page 35 of 45
ANALYSIS OF FACTORIAL EXPERIMENT: TAYWOODC
95
90
75
5
. . .
. . .
. .
. . . . . . .
. . . . . . . . . . . .
.
. . . . . . . . .
.
. . . . . .
------
. .
:
0.4 1
*
. . . . . . . .
. . .
. . . . . .
.
. . . . . .
. . . . . . .
.
. . . . . . . . . . . .
.
.-
.
Taywood Engineering Limited Report No. 1303 S/96/93 18 ●
Page 36 of 45
ANALYSIS OF FACTORIAL EXPERIMENT: TAYWOODC
1 ,
—
.
Taywood Engineering Limited Report No. 1303S/96/9318
Page 37 of 45
Appendix 2
Taguchi Results for the Steel Substrate
Taywood Engineering Limited Report No. 1303S/96/9318
page 38 of 45
o.
ANALYSIS
0.44402 46.33
OF VARIANCE
0.44402 46.33
119.9 119.9 97.511 1.4132 67.718 13.544 29.793 0.46552
304.5 3.8544
F
29.093
Total ERROR was used for F tests.
Taywood Engineering Limited Report No. 1303S/96/9318
Page 39 of 45
i
. --
. . . . . .
--
Taywood Engineering Limited Report No. 1303S/96/9318
Page 40 of 45
95
90
75
10
Taywood Engineering Limited Report No. 1303S/96/9318
Page 41 of 45
ANALYSIS OF FACTORIAL DENSITY FUNCTION
EXPERIMENT: TAYWOODS
RESIDUAL
Taywood Engineering Limited Report No. 1303S/96/93 18
Page 42 of 45
. . . TABLE OF COEFFICIENTS
16 OBSERVATIONS R-SQUARED = 0.9239 11 VARIABLES
STANDARD ERROR = 0.36714 ADJ R-SQUARED = 0.77169
`0.18357 0.18357 0.18357 0.18357 0.18357 0.18357 0.18357
SIG LEVEL 0.00032835 0.023192 0.56038 0.060087 0.25607 0.021818 0.023296 0.96826 0.056145 0.13587 0.018341
0.
0.12042 1.0642
ITS
7
-. ,
Taywood Engineering Limited Report No. 1303 S/96/93 18
Page 43 of 45
90
75
25
0
5
ANALYSIS OF FACTORIAL EXPERIMENT TAYWOODS
–1.0 –0.5 o 0.5 1.0
FACTORIAL EFFECTS 1
.
.—
#
—
Taywood Engineering Limited Report No. 1303S/96/93 18
Page 45 of 45
t
1
.