Service Life of Exterior Wood Coatings: A review of measurement and performance factors Jon Graystone The Paint Research Association, UK •CTBA Paris 2004 26-27 April Relevance of Service Life Issues to the Exterior Wood Coatings Industry - Supply Chain Participants • Architects & Specifiers • Retailers (Trade & Retail) • Building Constructors • Paint Manufacturers • Building Owners • Raw Material Suppliers • Maintenance Officers • Wood Providers and • Painting Contractors Converters • Trainers and Training • Wood Fabricators and Organisations • Research and • Specialist Contractors Technological • Householders Organisations • Standards Organisations • Industrial Applicators Relevance of Service Life Issues to the Exterior Wood Coatings Industry - Benefits • Shorten Development Cycle – reduce risk • Enhanced Performance – Warranty • Increase Customer Satisfaction and Confidence • Aid New Technology Development • Increase Confidence in New Product Launch • Test new technology or chemistry – VOC and SHE issues • Understand performance in different climates – relative rankings • Gain Competitive Advantage Service Life in Context Reliability Service Life Quality Durability Service Life Definitions • Durability – The period of exterior exposure during which the coating will retain an acceptable appearance, protect the substrate and not require any maintenance other than cleaning. – The degree to which paints withstand the destructive effects of weather (BSI) – The capability to reliably perform an intended function over a long service life under reasonable conditions of use. • Statistical Definitions – Reliability is the probability that an item will perform its intended function for a specified time interval under stated conditions • User focussed – A reliable product is one that does what the user wants it to do, when the user wants it so do so. • Quality – “on-target with minimum variance” • Capability Ratios • Statistical Process Control • Assumed link with performance Service Life Practical Usage of Terminology • Service life might be defined as a time period before replacement or repair is required. – Hence for practical purposes • Durability = Service Life Reliability – qualitative vs quantitative • Quality control is associated with production Service Life • Reliability is associated with design or development Quality Durability • Service Life Prediction and Durability Studies focus on – measuring and predicting failure and failure rates • probabilistic or deterministic – methods to accommodate failure • maintenance, tolerances, redundancy, warranty – preventing failure • improved service life – deterministic causes of failure analysis Difficulties related to the Service Life Prediction of Wood Coatings • Wood as a natural material is a difficult substrate, – species, grain orientation, heartwood sapwood ratio etc. • Many other causal factors can influence performance – design, the nature of previous coatings and the exposure history. • An overarching problem is that of the weather itself. – As an uncontrolled variable, climatic influences are difficult to separate from other sources of experimental variance. • Poor test capability, in terms of both repeatability and reproducibility. • An unresolved problem that has bedevilled the development of European performance standards. • Coatings are supplied as an Intermediate – Upstream and downstream influences on performance • Factory vs Maintenance • New work vs restoration • Fault Tree Analysis is useful for showing relationships Fault Tree Analysis Coated Window Failure Predictive Tests - Mechanical Properties Phenomenological Tests - Simple panel tests Coating Formulation - Modified panels - Jointed test pieces - Constructions (such as window - Design Elements frames) - Test houses Wooden Substrate Visual Inspection - Real buildings Environmental Conditions Application Factors Damage Factors Fault Tree - Expanded C oa te d Wi nd ow Fa ilu re (Multiple causes) Design Aspects Coat ing Syst em Wood Dist res s fact ors Applic at ion Climate and Formulation Subst rat e (Damage) Cons ideration Weather Joint s Profil es Permeability Mechanical Properties Wood Species Ex ternal Indust ri al And A ngles Liqui d & Vapour St rength and m odulus (Or deriv ed produc ts) Fact ors Applic at ion Ex ternal Int ernal Climate Climate Glazing Poly mer ty pe Pigm ent at ion Heartwood Moisture Fact ory And seal ant s (Binder) Phase volume Sapwood Movem ent Condit ions UV Hum idty Radiat ion Condensation Weathers eal Cure Early and Mechanical Weather And DPC Mwt & Tg Latewood Dam age (Ex ternal) Free Therm al Radicals Conduct ivit y End Grain Physical Form Grai n Orientat ion Therm al Temperat ure Sealing Soluti on or Dispersion Knots and Defect s Movem ent & Wind Temperat ure Temperat ure Cycle Cycling Quality Ex pos ure Adhesion Penetration Porosit y Biologic al Hum idity Work manship And Orientation Int ercoat and subst rat e And Fl ow Dens ity At tac k & Rain Freeze Thaw Surface Tension Rheology Ex tract ives Quality Surface Energy Col loi dal Forces Wett ing Visc osity Decay resist anc e Blue s tain Decay Fungi Film thicknes s Precipit aion Molecular Forces Surface mould And condensat ion Im pregnat iom Chem ical Water repellency Levelling Hai l Resistance Nut rient s Im pac t dam age Preservat ive Treat ments Tim e of Wet nes s Funct ionalit y Spore dist ributi on Aeration & pores Chem ical Type Pollut ion Crosslink densit y Insect Ac id rain Net work st ructure At tac k Penetration Oxygen Opacit y Col our Int ernal Carbon dioxide & UV absorbtion & Gloss Fact ors Monomers Or Polymers Fungal Res ist ance Coat ing Shrinkage St res s Machining And cutt ing Surface compression Previous Ex pos ure Nat ural Weathering Chem ical/ Phys ical Plas ma. Flame, Ac ety lation Previous Coat ings Maintenanc e & Preparation Ancilliary P roduct s Fillers, S ealers Surface Treat ments Sil icone etc Fault Tree Analysis Coated Window Failure Coating Formulation Design Elements Coating System Wooden Substrate Formulation Environmental Conditions Application Factors Permeability Mechanical Properties Liquid & Vapour Damage Factors Strength and modulus Polymer type Pigmentation (Binder) Phase volume Cure Physical Form Mwt & Tg Solution or dispersion Adhesion Chemical Intercoat and substrate Resistance Surface Energy Colloidal Forces Functionality Crosslink density Molecular Forces Network structure Penetration Water repellency & Flow Surface tension Rheology Wetting Viscosity Opacity Colour & UV absorbtion & Gloss Fungal Resistance Biocidal Properties Wood Substrate Wood Species Impregnation (Or derived products) Heartwood Early and Preservative Monomers Sapwood Latewood Treatment Or polymers Grain Orientation Chemical Penetration Knots and Defects Type Porosity Extractives Density Decay resistance Machining Previous And cutting Coatings Surface compression Maintenance Surface treatments & Preparation Silicone etc Sanding Ancilliary Products & finishing Fillers, sealants Previous Exposure Natural Chemical/Physical Weathering Plasma, Flame, Acetylation Fault Tree - Design Aspects Design Aspects Joints Profiles And angles Glazing Weatherseal And sealants And DPC End Grain Sealing Quality Exposure Workmanship And Orientation Fault Tree - Distress Factors Distress factors (Damage) External Internal Factors Factors Moisture Mechanical Coating Movement Damage Shrinkage Stress Thermal Biological Movement Attack Blue stain Surface mould Insect Attack Decay Fungi Nutrients Spore distribution Fault Tree- Application Application Considerations Industrial Weather Application (External) Factory Temperature Humidity Conditions & Wind & Rain Quality Film thickness Levelling Aeration & pores Fault Tree - Climate & Weather Climate and Weather External Internal Climate Climate UV Temperature Humidity Temperature Radiation Cycle Condensation Cycle Free Freeze Thermal Radicals Thaw Conductivity Precipitaion Pollution And condensation Acid rain Hail Time of Wetness Impact damage Oxygen Carbon dioxide Durability & Service Life Alternative Objectives • To quantify Service Life • To Predict Service Life • To Investigate Mechanistic Aspects of Service Life – The different objective will define the type of investigation • Probabilistic • Deterministic Reliability Concepts Probabilistic Reliability Deterministic Reliability Studies overall failure rates Studies individual failure independently of causes mechanisms as a function of time. Used to predict failure for e.g. Investigates failure at component, or maintenance scheduling causal level. Studies system reliability, e.g. Tries to eliminate failure at the through redundancy, fault causal level elimination etc Concerned with multiple failure rate Concerned with single failure distributions distributions and the impact of random events. Probabilistic reliability methods are widely used in military, aerospace and telecommunication specification. In deterministic studies it is essential that there is a conceptual model around which the failure mode can be analysed. Types of Reliability model • STRESS-STRENGTH: An item fails if the applied stress exceeds its strength. Below this level there is no damage and failure may be ascribed to random events. The failure distribution is exponential. (Hail Damage) • DAMAGE-ENDURANCE: This relates to accumulation of damage and also covers fatigue. Failure is sometime described as an endurance limit. In coatings the build up of residual stress is known to lead to failure under some circumstances. (Moisture Movement) • CHALLENGE-RESPONSE: Here a defect remains dormant until a specific event, or combination of events ‘challenge’ the coating and cause failure. (Adhesion Failure) • TOLERANCE-REQUIREMENT: This term is used to describe degraded performance that does not immediately lead to failure, such that the user accepts inferior performance rather than incurring correction costs. (Maintenance) Physical Ageing and Degradation • Physical ageing affects many substances including polymers, and is described as the reversible change in a property as a function of storage time under no influence from external conditions – Non -equilibrium (T<Tg> ambient) • may lead to contraction, change in creep etc. • In contrast with physical ageing, polymers may also undergo degradation and in some cases further cross- linking in response to external conditions. (Irreversible chemical ageing). • The effects of UV, water and temperature have all been widely studied in the context of durability – CORRELATION ISSUES • A useful approach in coating failure studies is to consider the chain of events which ultimately leads to failure. – TIME SERIES ANALYSIS Accelerating Feed Back • Artificial Weathering • Intensified Natural Weathering – angle of exposure – ‘Emmaqua’ • Extrapolation from Early Results – Time series – Analytical Techniques • Mechanistic Studies – photodegradation studies Accelerated Testing • “About the only non-controversial thing that can be said about accelerated testing is that it can be controversial” • L W Condra • Accelerated testing (weathering) is based on the premise that test panels will exhibit the same behavior after a short period at high stresses, as they would exhibit after a longer time at lower stresses. – Reciprocity Issues • Much work amounts to a quest for a ‘black box’ that will give a good correlation with natural weathering exposure. • Many regard the Coating Industry as behind other industries in predicting the service life – ”Attempts at refining the current durability methodology have been made over the last 80 years without any significant improvement in its predictive capability” • FCST Mongraph - Methodologies for Predicting Service Life (1996) – Martin, Saunders, Floyd, Wineburg [ref.6] Reciprocity Issues • There is an underlying assumption in much accelerated testing that raising the level of a ‘stress’ such as UV radiation will simply increase the rate at which changes occur such that the damage depends on the total absorbed energy, and is therefore independent of radiant intensity and exposure time. • This is known as a reciprocity law. • A recent review [ref.8] outlines the many exceptions to the simple assumption. At various radiant flux thresholds different reciprocity relationships have been observed, indicating that alternative, system dependent degradation paths have been followed. • A model based on band theory is been proposed to explain this behaviour, to complement photochemical mechanisms. Service Life Prediction Issues (SLP) • Practical coating failures invariably show a wide, and often non-normal distribution. • Incidents of early failure may have more practical significance than average service life. • Outdoor exposure conditions are not reproducible. • Quantification of relevant exposure environments (natural or artificial) is poor. • Durability data is multivariate. • There is no a priori reason to expect correlation between different coating chemistries under different service conditions whether natural or artificial. – A dosage based damage model is required • this represents a fundamental change in tactics for many coating manufacturers ! Quantification of Exposure Conditions • UV Dosage – Integrate • spectral irradiance • spectral absorption coefficient • spectral efficiency • duration • Temperature – Energy balance equations • absorptivity • emissivity • Water – time of wetness – relative humidity – precipitation Some Mathematical Aspects of Service Life Prediction (“Mathematics is a subject in which we do not know what we are talking about, nor care whether what we say is true” - Betrand Russel ) • Statistics and Experimental Design – variance and test capability • Regression • Non parametric statistics • Time Series Forecasting • Reliability Methodologies • Failure Modeling • Bayesian Methods – meta-analysis and the use of prior knowledge Descriptive Statistics Data Analysis Tools • Distribution Functions – Measures of location - mean, mode – Measures of spread - standard deviation, range • Correlations – linear association • practical implications of magnitude • Multivariate data analysis – do not depend on a formal probability model – information rich summaries to generate hypotheses • principal component analysis • multidimensional scaling • cluster analysis • correspondence analysis • Significance Tests Common Distributions ( “It is a capital mistake to theorize before one has data - Conan Doyle” ) A random variable may be discrete or continuous. Discrete distributions are described by a probability function, while continuous distributions have a probability density function. Each is applicable to particular problems. • Discrete • Continuous Distributions Distributions – Point probability – Cumulative or functions probability distribution • Bernoulli functions • Binomial • Uniform • Poisson • Normal • Geometric • Exponential • Gamma • Weibull The Normal Distribution THE NORMAL LAW OF ERROR STANDS OUT IN MANKIND’S EXPERIENCE AS ONE OF THE BROADEST GENERALISATIONS OF NATURAL PHILOSOPHY ~ IT SERVES AS THE GUIDING INSTRUMENT IN RESEARCHES IN THE PHYSICAL AND SOCIAL SCIENCES AND IN MEDICINE AGRICULTURE AND ENGINEERING ~ IT IS AN INDISPENSABLE TOOL FOR THE ANALYSIS AND THE INTERPRETATION OF THE BASIC DATA OBTAINED BY OBSERVATION AND EXPERIMENT W J Youden Important Lifetime Distributions The Four most important distributions used in lifetime studies are:- • The Normal Distribution – Very important in Statistics and Quality control, • slightly less common in reliability work • The Lognormal Distribution – very common in reliability work, used particularly for failure distributions due to accumulated damage • The Exponential Distribution – applicable to failures which are totally random (independent of past history) e.g. electrical fuse • The Weibull Distribution – Widely used in reliability work and failure modeling due to its versatility • has a shaping factor allowing it to be used for all parts of the bathtub curve, or as an exponential distribution. Impact of Variance on Measurment • It has been common in the past to describe variance as ‘error’, and to divide this into systematic and random errors. This is generally derogated, and the term ‘common cause’, and ‘special cause’ variation are much to be preferred in order to make a distinction between ‘mistakes’, and the variance that is often part of the normal process of measurement. • Common cause variance influences the accuracy of results, while special cause errors will affect the precision of the results • Accuracy is the difference between the average measurement of a sample and its ‘true’ value, while precision is the variation in measurements observed with a test method while measuring the same sample multiple times on the same instrument by the same operator Impact of distribution on Test Capability Test Capability specification control limits Reliability Based Methodologies Lifetime Distributions Especially useful where observed times to failure have a high variability • Specimens are exposed until they reach an absolute, or arbitrary relative failure point – non -failures are said to be ‘censored’ • Lifetime Representations – Failure Density (Probability density function) • number of failures/unit time as % of original – decreasing - old is best (infant mortality) – constant - old is as good as new (youth and middle life) – increasing -new is best (wear out, old age) – Cumulative Distribution Function • cumulative number of failures up to a given time – Survival Function • cumulative number of survivors t a given time – taken as a measure of reliability – Failure Rate • number of failures per unit time expressed as fraction of the number of survivors – Characteristic Life – Failures per unit time – Mean time between failure 5 4 Cracking Score 3 (b) (a) 2 1 0 10 20 30 40 50 60 70 80 90 100 Time in weeks Differences between traditional and reliability methodologies for a hypothetical example. In the traditional approach (a) the cracking scores are averaged at two years. In the reliability approach (b), the times to reach a pre-denominated failure level (here 3) are recorded. Typically less replicates are used in the traditional approach. Failure Modeling - Time Transformation Functions • Assumes that time to failure distributions are a common set for which a function exists to relate distribution parameters to the intensity of intensity of weathering f weathering ng d mechansims exponential se Power Law al to stress most common ss modelling lling re dependent elationships ndau, Ferry) Some Relatively Recent Research Observations (1) e.g. AIR project, Arwood, DOE Multiclient Studies, PRA Multiclient • “The type of paint has no effect on the average moisture content in the window frames in contrast to the type of wood” – Centrum voor Onderzoek (20 Paint ystems in 13 housing complexes) • “In general coating performance on panels and L joints did not correlate with that on window components. It appears from this trial therefore that these test specimens cannot be used to reliably predici performance on windows” – AIR Project CT94/2463 - Task 1 • “The results show a multitude of factors with numerous interactions influencing the wood moisture content. In absolute terms the most important factor was seasonal climatic fluctuations. The coating system played only a minor role compared to the influence of the sample design” – AIR Project CT94/2463 - Task 2 • “The rank order of performance (integrity) correlates with the initial water absorption value of the coating” – DOE Multiclient • “Some coatings give rise to high localised moisture concentraion at the wood coating interface” – PRA DTI Project Some Relatively Recent Research Observations (2) e.g. AIR project, Arwood, DOE Multiclient Studies, PRA Multiclient • “It has been shown that the type of wood has no effect on failures in the paint system,” – Centrum voor Onderzoek (20 Paint ystems in 13 housing complexes) • “Confirmation that the substrate can affect the performance of all coatings is one of the major findings from this trial” – DOE Multiclient • “After 12 months exposure it was clear that for many coatings this had been insufficient to unequicocally discriminate their performance characteristics…” • “…analysis confirmed there was no correlation between performance after 12 months and after 24 months” – -DOE Multoclient • “…correlations of the assessment results between 12 & 24 months were good, and this gives a basis on the suggested duration of one year for an exposure time in EN 927-3 – AIR Project CT94/2463 - Task 1 • “’opaque’ films can vary significantly in their opacity to light” – PRA DTI Project • “Performance (integrity) after 12 months correlates with dry film thickness – DOE Multiclient Service Life Studies - Current Trends  • Damage Indices – photohydrolytic mechanisims • UV, Temperature , Time of Wetness, Humidity • Spectral Power Distribution Functions (SPD) – effect of colour • Dosage models and verification – integrated sphere • Degradation Rate models • Damage precursor studies • Non traditional metrics – fracture energy, UVA spectra, HALS analysis, photo-oxidation analysis, DMA, DSC • Newer Techniques – FTIR, photoacoustic TOF-SIMS, Atomic Force Microscopy • Arrhenius-Weibull Models – influence of Tg • Time Series Analysis • Fundamental Molecular Orbital Predictions Recent (2003) NIST Publications - all fully available as PDF on web pages • Advanced Techniques for Nanocharacterization of Polymeric Coating Surfaces. • Effects of Spectral UV on Degradation of Acrylic-Urethane Coatings. • Enhancing Sensitivity of Atomic Force Microscopy for Characterisation of Surface Chemical Heterogeneity. • Integrating Sphere Sources for UV Exposure: A Novel Approach to the Artificial UV Weathering of Coatings, Plastics and Composites. • Laboratory Apparatus and Cumulative Damage Model for Linking Field and Laboratory Exposure Results. • Nanoscale Characterisation of Coatings Surface Degradation With Tapping Atomic Force Microscopy. • Reciprocity Law Experiments in Polymeric Photodegradation: A Critical Review. • Relating Laboratory and Outdoor Exposure of Coatings. Part IV: Mode and Mechanism for Hydrolytic Degradation of Acrylic-Melamine Coatings Exposed to Water Vapour in the Absence of UV Light. • Repeatability and Reproducibility of Field Exposure Results. • Use of Laser Scanning Confocal Microscopy for Quantitative Characterisation of Physical Changes of Polymer Coatings After UV Exposure. Temperature Controlled Weathering Rack (PRA - Teddinngton) Service Life Prediction Methodology Quantification of Exposure Environmental Natural Weathering Environment. Exposure -distributed sites. - ambient and test panel Analysis of existing - UV, temperature, wetness etc. coatings Laboratury Exposures Quantification and systematic - 'artificial weathering variation of environment - static and dynamic Dosage Metrics Response Techniques Monitoring -SEM, AFM Initial Data Analysis - FTIR - distribution statistics - Thermal Analysis - time series - Mechanical Analysis Degradation - reliability methods - Gloss, Colour, Chalking etc. - Biological Mechanisms Delivered Model Application Model Validation Time transformation Model Objectives - correlation studies Data = Fit + Residual - Arrhenius Durable Long-life - Power Law Coatings - WLF etc. State of the Art - General Conclusions (1) Scope for further work • Despite the vast amount of published work there remain many unanswered questions and contradictions relating to the performance of coated wood. • Investigating all the many interactions (fault tree) would be hugely expensive • There is a need to focus on basic issues which define a framework within which specific issues might be prioritised – of clear value if answers could be found • however funding has proved almost impossible to obtain • Further critical meta-analysis of past work could yield valuable clues (Knowledge Management Issues) Scope for further work - some examples • Influence of exposure conditions on rank order of performance – TC139/WG2 - Round Robin • Optimum mechanical properties for long term performance • Moisture permeability limits – influence on performance • Next Presentation • Effect of parameters which influence both performance and Test Capability – Colour, Film Thickness, Wood Species and microstructue • Dosage based methods for service life prediction – lessons from other disciplines • reliability theories Summary and Conclusions for Service Life Studies • Development of long lived coating systems requires improved methods to measure and validate Durability or Service Life Prediction. • There are a wide variety of mathematical and statistical tools that aid the prediction of durability and service life. • Choice of the right method requires a clear understanding of the objectives and the limitations and advantages of each technique. • Traditional Methods have failed to deliver the confidence found in other industries. • Reliability Methods are more appropriate to some types of progressive failure and in multi-component situations. • Dosage models are needed to interpret Natural Weathering, and the bridge the gap between Natural and Artificial Weathering. • Service life can in principle be predicted from damage precursors before they become or cause catastrophic failure. • Time Series Analysis should be considered for sequential observations that are not independent.
Pages to are hidden for
"Service Life of Exterior Wood Coatings VTT Virtual project"Please download to view full document